Sei sulla pagina 1di 282

VOLUME 10 VOLUME 10 VOLUME 10 VOLUME 10

JULY 10 JULY 10 JULY 10 JULY 10


2009 2009 2009 2009
Philosophy Philosophy Philosophy Philosophy
of Economics of Economics of Economics of Economics

EDITED BY
MAURO ROSSI & LAURA BERITELLI
QUARTERLY JOURNAL
-
ISSN: 1972-1293
Journal of Philosophical Studies
Biblioteca Filosofica 2007 - Humana.Mente,
Periodico trimestrale di Filosofia, edito dalla Biblioteca Filosofica -
Sezione Fiorentina della Societ Filosofica Italiana,
con sede in via del Parione 7, 50123 Firenze
(c/o la Facolt di Scienze della Formazione dell'Universit degli Studi di Firenze) -
Pubblicazione regolarmente iscritta al Registro Stampa Periodica del Tribunale di Firenze
con numero 5585 dal 18/6/2007.

REDAZIONE - Via del Parione 7, Firenze, presso Biblioteca Filosofica - Facolt di Scienze della
Formazione, Universit degli Studi di Firenze


Direttore editoriale: Alberto Peruzzi
Direttore responsabile: Duccio Manetti
Vice Direttore: Silvano Zipoli
Supervisore scientifico: Marco Salucci


Redattori e Area di Ricerca:

Elena Acuti - Filosofia Politica
Scilla Bellucci - Filosofia del Linguaggio / Antropologia Filosofica
Laura Beritelli - Filosofia del Linguaggio / Ermeneutica Filosofica
Alberto Binazzi - Filosofia della Mente / Scienze Cognitive
Matteo Borri - Filosofia e Storia della Scienza
Roberto Ciuni - Ontologia / Logica
Giovanni Casini - Filosofia della Scienza / Logica
Chiara Erbosi - Bioetica / Filosofia della Mente
Marco Fenici - Logica / Scienze Cognitive
Riccardo Furi - Filosofia della Mente
Tommaso Geri - Filosofia del Linguaggio / Ermeneutica Filosofica
Matteo Leoni - Antropologia Filosofica
Stefano Liccioli - Filosofia Morale
Umberto Maionchi - Filosofia della Scienza / Logica / Filosofia del Linguaggio
Francesco Mariotti - Storia e Filosofia delle Neuroscienze / Filosofia della Mente
Giovanni Pancani - Filosofia Politica / Antropologia Filosofica
Daniele Romano - Filosofia e Storia della Scienza
Silvano Zipoli - Filosofia e Storia della Scienza
Segretario di Redazione: Matteo Leoni

I contenuti di Humana.Mente sono sottoposti a refereeing



Il comitato scientifico della rivista composto dai membri del Consiglio Direttivo
della Biblioteca Filosofica




JOURNAL OF PHILOSOPHICAL STUDIES JOURNAL OF PHILOSOPHICAL STUDIES JOURNAL OF PHILOSOPHICAL STUDIES JOURNAL OF PHILOSOPHICAL STUDIES
Progetto Grafico: Duccio Manetti
Sito web: www.humana-mente.it
La struttura del sito realizzata da
Emiliano Mazzetti e Simone Gallorini
Per informazioni e collaborazioni:
redazione@humana-mente.it

Humana.Mente,Issue10July2009

INDEX

p.I
INTRODUCTION PhilosophyofEconomicsMauroRossi p.III
PREFACE EditorialLauraBeritelli p.VII
COLUMN PhilosophicallyIncorrect:CrisiEconomica.Unamodestariflessione p.XI
EVENTS Neuroeconomics:HypeorHopeRotterdam,2022November,2008
(reviewedbyFedericaRusso)
p.XV
Decision,GamesandLogicHECLausanne,1517June,2009
(reviewedbyUmbertoGrandi)
p.XVII
LaboratoriodiOntologiaApplicataFlorence,July,2009
(papereditedbyRobertaLanfredini&AlessandroBemporad)
p.XIX


PAPERS
Esperimentiparadigmatici:ilgiocodellUltimatumFrancescoGuala p.1

ScientificUnificationinEconomicsCaterinaMarchionni p.11

CausalArrowsinEconometricModelsFedericaRusso p.25

WhatCanNeuroscienceOffertoEconomics?MatteoColombo p.41

ANeuroeconomicPerspectiveonCharitableGivingDavidYokum,
FilippoRossi
p.59

WhyCognitivistsAboutPracticalReasonCanOnlybeSemi
CognitivistsGiacomoMollo
p.73

ComparingPreferencesMauroRossi p.93

DoesFreedomofChoiceCauseSatisfaction?MargheritaBottero p.111

UnderstandingMarxsNormativeCritiqueofCapitalism:Reification
RevisitedVolkandam
p.123
BOOKREVIEWS
FilosofiadellEconomia,byDanielM.Hausman
(reviewedbyAlbertoBinazzi)
p.141
TheNewSpiritofCapitalism,byLucBoltanski&veChiapello
(reviewedbyGeorgHull)
p.143
Criticadellaragioneeconomica,editedbyMatteoMotterlini&
MassimoPiattelliPalmarini(reviewedbyAlbertoBinazzi)
p.147
TheCultureofNewCapitalism,byRichardSennett
(reviewedbyMarcoSolinas)
p.151
EquilibriuninEconomics:ScopeandLimits,byValeriaMosini
(reviewedbyGiovanniCasini)
p.155

Eticaedeconomia:ilrapportopossibile,editedbyB.GiovanolaeF.
Totaro(reviewedbyChiaraErbosi)

p.159
Economiaeconcezionedell'uomo,editedbyP.Grasselli
(reviewedbyScillaBellucci)
p.163
Freakonomics,byStevenLevitt&StephenJ.Dubner
(reviewedbyChiaraErbosi)
p.167
CapitalismoTotale,byJeanPeyerelevade(reviewedbyRiccardoFuri) p.171
Oikonomia,editedbyEdoardoFerrario(reviewedbyRiccardoFuri) p.173

COMMENTARIES
Ilcapitale,KarlMarx(commentedbyM.Sattler) p.177
Lagrandetrasformazione,KarlPolanij(commentedbyG.Pancani) p.181
Problemiepistemologicidell'economia,L.V.Mises
(commentedbyA.Binazzi)
p.185
Sullanaturaesull'importanzadellascienzaeconomica,L.Robbins
(commentedbyS.Bellucci)
p.191
Ilgrandecrollo/Storiadell'economia,J.K.Galbraith
(commentedbyG.Pancani)
p.195
MicromotivesandMacrobehavior,T.Schellling
(commentedbyP.Maffezioli)
p.199
Eticaedeconomia,A.Sen(commentedbytheEditorialStaff) p.207
Ilbanchieredeipoveri,M.Yunus(commentedbytheEditorialStaff) p.209
LaGlobalizzazioneeisuoioppositori/LaGlobalizzazionechefunziona,
J.E.Stiglitz(commentedbyG.Pancani)
p.211
Economiesenzadenaro,M.Pittau(commentedbyS.Liccioli) p.215
NeuroEconomics:anIntroductoryReview,editedbyMatteoBaccarini&
SilvanoZipoliCaiani
p.219
INTERVIEWS
InterviewwithRICCARDODUGINI(DirigenteBancaEticaFirenze)
editedbyScillaBellucci&LauraBeritelli
p.233
III

Introduction
PhilosophyofEconomicsAnIntroduction

MauroRossi
mauro.rossi@umontreal.ca


Theeconomicsciencehasrarelybeenimmunefromcontrastsandcontradictions.
Originally a branch of moral philosophy, economics emerged as an autonomous field of
social scientific enquiry little more than 200 years ago. In the wake of the natural sciences
dramatic success, economists soon embraced a methodology that treated economic
phenomena simply as another class of natural phenomena. This methodological stance
progressivelyledtoamoreandmoreformalandrigorousanalysis,characterisedbyagrowing
mathematization of economic models. Since then, economics has not ceased to receive the
admiration of other, more recent, social sciences and its methodology has often been
consideredasanexampletoimitatebyallthosewhohavebeenimpressedbytheprecisionof
itsformalresults.
Nevertheless, the history of economics is far from being a collection of triumphs. Indeed,
the economic sciences sophisticated formal analysis and its reputation for methodological
accuracy often contrasted with poor predictions, explanations and interpretations of real
economic events. The systemic crises that periodically shake the world economy cast a dark
light on the practical efficacy of the economic science. It appeared as if economics was
destined to be a dismal science not because of the unwelcome content of its longterm
predictions,asMalthusoneofthefounderofmoderneconomicsbelieved,butbecauseof
itsincapacitytogetthingsright.
Further, these failures offered an easy empirical argument to all those eager to express
theirdissentagainstthenaturalisticmethodologyembracedbymainstreameconomics.While
orthodox economists defended their approach by appealing to the yet young state of
economic research, it seemed to more heterodox economists that the contrast between
theoreticalandpracticalresultsjustifiedamoreseriouscomplaintabouttheusefulnessofthe
naturalisticapproachtoeconomicphenomena.Morespecifically,itseemedtothemthat,qua
science of human and social behaviour, economics needed a distinctive method of analysis,
differentfromtheoneemployedforstudyingnaturalphenomena.
These tendencies are still alive nowadays. However, they somehow take different forms
and expressions in the light of a new phenomenon, which has characterised the last two
decades, namely, the growing exchange between economics and other disciplines (e.g.
psychology, anthropology, neuroscience, ethics), which followed years of reciprocal isolation
andmutualdisinterest.Indeed,thisinterdisciplinaryexchangehasdrawnrenovatedattention
onsomeofthecontrastingfeaturesthathavecharacterisedthehistoryofeconomics.
On the one hand, the influence that economics exercises on other research areas has
assumed a more pronounced and direct form. The appeal of economics is no longer confined
to its methodology, but it derives also from the flexibility of its theories and models, which
makes them more and more frequently chosen tools for explaining phenomena pertaining to
differentdomainsofresearch.Aclearexampleofthistendencyisthedevelopment,inrecent
years, of various accounts trying to explain, for instance, the evolution of moral norms (e.g.
Binmore 2005, Bicchieri 2006, McKenzie 2007) and the behaviour of the human brain (e.g.
Platt and Glimcher 1999) by means of the extensive employment of game and decision
theoreticresources.

Humana.MenteIssue10July2009

IV

On the other hand, the results obtained in research areas outside economics (e.g. social
psychology, cognitive science and experimental philosophy) have provided further support to
the doubts raised by the predictive failures of the economic science. Indeed, the
unprecedentedavailabilityofempiricalstudiesconcerninghumanbrainandbehaviourhasput
into question the core assumptions on which mainstream economics is based, namely, that
individuals are selfinterested agents and that they are instrumentally rational. A significant
example is offered by recent works in psychology and neuroscience (e.g. Camerer and Fehr,
2006)thatchallengethetraditionalassumptionaccordingtowhichthemaineconomicagents,
i.e.consumersandfirms,actonthebasisofselfishmotivesalone.
Thecombinationofthesetendenciesproducedaseeminglycontradictoryresult:economics
seems, at the same time, to conquer and to be conquered by other disciplines. In turn, this
interestingphenomenonopensthespaceforphilosophicalreflection.Thereby,thephilosophy
of economics has been emerging as a, more than ever, fascinating and important area of
philosophical enquiry. This is the reason why Humana.mente has decided to dedicate the
currentvolumetothisdisciplineandtotheexplorationofsomeofitsliveliestdebates.Before
illustrating them in more detail, it is worth to briefly illustrate the main areas of analysis with
whichphilosophyofeconomicsisconcerned.
First and foremost, the philosophy of economics is concerned with questions about the
method adopted for the prediction, explanation and interpretation of economic phenomena.
Morespecifically,thisresearchareainvestigatesthenatureofeconomictheoriesandmodels
and the role that they play within the analysis of economic phenomena. Second, the
philosophy of economics is concerned with the core assumptions and postulates on which
economic theory is based. In particular, a great deal of analysis focuses on rational choice
theory, broadly considered, that is, as a theory of individual, strategic and collective decision
making. Third, the philosophy of economics examines the links existing, and the mutual
contributions, between economics and more normative domains on inquiry, such as political
economy and ethics. For instance, key debates in this area are those relative to notions like
wellbeing,freedom,equalityanddistributivejustice.
The current volume of Humana.mente explores the most recent tendencies characterising
philosophyofeconomicsresearchbymeansofaseriesofarticleswrittenbyItalianprofessors,
postdoctoratefellowsandPhDstudents.Allthesearticlestouchonthemesbelongingtoone
oftheareasofanalysispresentedabove.
Caterina Marchionni explores the first tendency illustrated above. She argues that a great
dealofcontemporaryeconomicresearchcanbeunderstoodintermsoftheroleplayedbythe
ideal of scientific unification, which she characterises as the application of the same
principles, explanatory schemata, models and tools to the study of phenomena in different
domains. Marchionni distinguishes four level at which unification can take place:
explanations, theories, fields and disciplines. She then shows how this conceptual framework
can help us understand the emergence of new trends in economics research, like the New
EconomicGeography,andtheirimperialisticeffectsonneighbouringdisciplines.
Toalargeextent,theexplanatorypowerofeconomicsdependsonitsabilitytotellreliable
causalstoriesabouteconomicphenomena.FedericaRussoexploresthecrucialmethodological
question of how econometric models come to establish that some correlations involving
economicvariablesareindeedcausal.Rejectingastrictdichotomybetweenpurelyinductivist
and purely deductivist approaches, according to which causality is either inferred from
statisticalpropertiesofdataaloneordirectlygivenbytheeconomictheory,Russodefends
a midway view, according to which the attribution of causality results from a more dynamic
interplay of inductive and deductive elements. These elements form a rather rich apparatus,
Introduction

V

which is constituted by background knowledge, statistical, extrastatistical and causal


assumptionsandahypotheticodeductivemethodology.
As Russo shows, the test of welldefined causal hypotheses is an important step within
economic theorizing. Often, but not always, testing involves the use of experiments. In his
paper, Francesco Guala explores the reasons why the Ultimatum Game (UG) has become a
paradigmaticexperiment,employednotjustineconomics,butinothersocialsciencesaswell.
Guala emphasises two features. First, UGbased experiments possess some virtues
(replicability, robustness and disciplinary cohesion), which make them standardised
experimental designs. Second, the UG can be used as a measurement device that helps us
highlight the existence of significant realworld causal relations to which further analysis can
thenbedirected.Itisthankstothesefeaturesthat,accordingtoGuala,theUGemergedbya
process of social selection in experimental game theory, as a robust design that tps on
somethingthatseemstomatterforus.
Matteo Colombo explores the question how neuroscience can contribute to economics.
Moving from Gualas reflection on paradigmatic experiments, Colombo uses the UG as the
benchmark for his subsequent analysis. One of the main features of the UG is the contrast
betweenthestandardgametheoreticpredictionsandtheobservedexperimentalresults.This
feature casts more than one doubt on one of the core assumptions underlying game theory,
namely,theassumptionthatindividualsaremovedsimplybyselfinterest.Colomboexamines
Cristina Bicchieris sophisticated attempt to explain the agents actual behaviour in terms of
the notion of social norms. He argues that Bicchieris account presents both theoretical and
practical limitations, which compromise its usefulness as a predictive tool. Colombo then
argues that one way to obviate these limitations consists in enriching Bicchieris model with
neurobiologicallyinformed evidence. This analysis leads Colombo to conclude that
neurosciencecancontributetoeconomicsbyimprovingitspredictions.
Like Colombo, David Yokum and Filippo Rossi are interested in how disciplines like
psychology and neuroscience can combine with economics in the effort of explaining certain
problematicfeaturesofindividualbehaviour.YokumandRossifocusspecificallyoncharitable
donations. By examining public good games, they show that the standard analysis, based on
the assumption of selfinterested behaviour, leads to predictions that are at odds with the
agents observed behaviour. They propose four different models accounting for the observed
anomalies, which leave room for the possession of nonselfish considerations in addition to
selfish ones as grounds for behaviour. Such considerations may include feelings of personal
satisfaction,orwarmglow,andpreferencesforfairness,reciprocityanddeservedness.Inthe
last part of their paper, they consider how these models fare with respect of neuroscientific
evidence.

While the previous authors focus on the role of selfinterest in modelling and explaining
individual behaviour, Giacomo Mollo examines the other core assumption underlying
economic theory, namely, the assumption that individuals are practically (and instrumentally)
rational. His goal is not to reject this assumption, but rather to elucidate how we can
understand practical rationality. Mollo examines a position in the literature, i.e. cognitivism,
accordingtowhichwecanmakesenseoftherequirementsofpracticalrationalityintermsof
the requirements of theoretical rationality. After defending cognitivism against several
objections, Mollo argues that there is nonetheless one case where cognitivism is unable to
explain the failure of practical rationality in terms of a corresponding failure of theoretical
rationality. This analysis leads Mollo to conclude that one can be, at best, only a semi
cognitivistaboutpracticalrationality.

Humana.MenteIssue10July2009

VI

The last three papers deals with issues of relevance for more normative domains of
analysis. Mauro Rossi considers the problem of interpersonal utility comparisons (IUCs).
According to a long tradition in economics, comparing the utilities of different individuals
posesinsurmountableepistemologicaldifficulties.Thisisbadnewsforseveralareasofanalysis
(e.g. normative economics, social choice theory, ethics), which are in serious trouble without
the possibility of making meaningful IUCs. In his paper, Rossi assesses one promising strategy
forsolvingtheproblemofIUCs,whichfocusesonhowordinarypeopleattributeandcompare
otherindividualsmentalstates.Hearguesthatalltheargumentsattemptingtovindicatethis
strategyfail.Rossiconcludesthatthecurrentstateofresearchlegitimates,althoughdoesnot
entail,amoderateformofscepticismtowardsIUCs.
One area of normative debate to which both philosophy and economic analysis have
recently brought a significant contribution is the one concerning individual wellbeing.
Margherita Botteros paper offers an example of how it is possible to fruitfully combine both
disciplines in relation to this topic. Botteros goal is to investigate whether freedom of choice
mattersforwellbeing,whenthelatterisconceivedaslifesatisfactionorhappiness.With
the help of a meticulous econometric analysis, she shows that freedom of choice is positively
correlated with individuals life satisfaction. Bottero notes that the robustness of the results
obtained suggests the existence of a causal relation between the two variables. At the same
time, she is careful in invoking the need for further research in order to confirm the causality
interpretationandclarifysomeofitsfeatures,likethedirectionofcausality.
Volkan dam offers an interesting example of more heterodox research in philosophy of
economics. dam focuses on the question of how we can understand Marxs normative
critique of capitalism. The question arises because the traditional readings run into a sort of
dilemma.EithertheyadoptEngelsportraitofMarxasastrictpositivescientist,inwhichcase
Marxs normative critique appears either unintelligible or ideological, or they ground his
normative critique in commonsense moral intuitions, which Marx explicitly rejected in all his
writings. dam offers an original way out of this dilemma. Revisiting and modifying Luckass
conceptofreification,hearguesthatwecanbetterunderstandMarxsargumentsintermsof
the normative ideal of recognition, which sets the standard of his normative critique of
capitalismandwhichtherebymotivatestherejectionofallsocialrelationsthatsuccumbtoa
deficientmodeofintersubjectivity.
VII

Editoriale

LauraBeritelli
laura.beritelli@humanamente.it

Mercatomondiale

Quelcheadessocioccorreunafresa,
uncilindrocimatore,unaschedad'interfaccia,
portatiinvolodalGiappone.Eccoqui
tutt'auntrattosouvenirdiTimbuctu,
icone,neonatirapiti.Ovunque
cisonolestesselamettedabarba,
congressistieapikiller.
Circolanoautobombe,donne
atterranodaicataloghiinvendita,
contibancarisispostanoviasatellite.
Giungonoinvolovirusinusitati.
Soloognitantoalmarginedellastrada
c'unmendicante,immobile.

HansMagnusEnzensberger,Pileggeridell'aria.Poesiemorali,Einaudi,Torino,2001.

Humana.Mente ha sempre avuto interesse alla possibilit di far dialogare la tradizione di


pensieroanalitica,alcuirigorelogicoeformales'ispira,conquellacontinentale,alcuisguardo
critico non pu e non vuole sottrarsi. Nell'intenzione di presentare i pi recenti lavori di
filosofia dell'economia in Italia, si infine realizzata la fortunata occorrenza attraverso cui
mostrare le comuni attitudini, le rassomiglianze di famiglia tra le due diverse prassi di
pensiero come anche, di contro, le difformit che esistono tra di esse. C', in effetti, una
critical mass concettuale su cui l'approccio analitico e quello continentale si scontrano o,
meglio, non si incontrano; ma, in definitiva, sembra dipendere dai diversi ordini di valori che
orientano le rispettive prospettive di lavoro ed intervento, piuttosto che per una qualche
effettivaincompatibilit.

Comeampiamentespiegatonell'editorialeacuradiMauroRossi,proponiamoquindiunaserie
diarticoliscientifici,opapers,afirmadigiovanifilosofidell'economia,chestannodefinendoe
rendendograndeconillorolavoroquestainnovativabrancadelsaperee,conessi,unavasta
panoramica delle ricerche in itinere nelle pi prestigiose universit, in particolare quelle del
mondo accademico anglosassone; ospitiamo, con grande interesse, un articolo di Francesco
Guala,affermatopubblicistaefilosofodell'economia,articolochel'autorehatradottopernoi
eche,inlinguainglese,comparsosuPhilosophyofScience(75,2008,pp.658669).
A questi lavori associamo una interessante collaborazione con l'Istituto di Studi Sociali di
Francoforte. Ringraziamo per aver affrontato assieme a noi questa avventura i giovani filosofi
eredi di quella prestigiosissima tradizione, elaborata nel cuore dell'Europa continentale e che
tanto ha influenzato la cultura mondiale. Ci ha interessato presentare questi lavori proprio

Humana.MenteIssue10July2009

VIII

perch riconosciamo l'importanza che quelle idee rivestono ancora oggi, laddove alcune
agisconoancoraepipotentementesottotraccia
Dalla Scuola di Francoforte non ci arrivano dunque lavori di filosofia della scienza
economica tout court, cui il numero dedicato; ma un simile accostamento risulta per noi
altrettanto fondamentale, perch mostra quanto sia ineludibile l'approccio complessivo, etico
e sociale all'economia, indipendentemente che essa sia intesa come scienza sociale o scienza
naturale dibattito questo che si vanifica nel momento in cui l'economia a dettare il senso
del mondo; dacch, come scrive Giuseppe Limoni nella sua introduzione a A. Micocci The
metaphisic of Capitalism (Lexington Books, Lanham, Maryland, 2008) l'economia ha oggi
acquisito una centralit regia, paragonabile a quella posizione architettonica che un tempo
sembravaspettareallateologia.
Se quindi vero, con lo stesso autore, che la scienza economica oggi [...] scienza della
struttura capitalistica, che la formazione storicamente determinata propria dell'et
moderna, allora di questo che la tradizione continentale si occupa; e se vero, come
prosegue, che essa non pu sottrarsi a tutti i quesiti epistemologici che ne definiscano e
comprovino la scientificit, allora di questo che si occupa la tradizione analitica. Concludo,
citando ancora Limone, che la scienza economica in quanto descrizione e spiegazione del
mondo, non solo descrive e spiega il mondo, perch invece, in quanto creduta scienza, ed in
quanto applicata come tecnica, si fa mondo diventando di fatto un prodotto oggettivo ed
indipendente del suo modo di pensare. cos che ancora una volta la lucidit della filosofia
continentale si manifesta, invitando al contempo all'analisi endogena delle proposizioni della
scienza economica cui attendono, per l'appunto, i lavori dei filosofi della scienza e, nel
particolare,ifilosofidell'economia.
Suggeriamo quindi come sia possibile assolvere insieme al compito complessivo e pi
propriodellafilosofia,ovveroquellodicustodireevigilaresullosviluppodeisaperi,unendoe
nontenendobenseparatelerisorsedientrambigliorientamentidiindagine,sottol'egidadi
un agire del pensiero che gi etico e che, essendo intrinseco alla filosofia tutta, appartiene
all'operatodientrambi.

inquestosensochenellamonografiaPhilosophyofEconomicssiparladietica,diecologia,
didirittoedituttiglialtriaspetticuipuessereorientatoildiscorsoeconomico.Seneparlasia
nellacolonnadedicataaglieconomisti,quelladelleriletture,ocommentaries,cheproponeuna
selezionedilavoricapitali,quellichehannofattolastoriadell'epistemologiadell'economia;sia
nellacolonnadedicataaifilosofidell'economia,quelladellerecensioni,obookreviews,dacui
speriamosiapossibilecominciareafarsiun'ideadelperchdell'affermazionediquestanuova
disciplina.

ANALOGIETEMATICHE

Analiticiecontinentali,nonostanteladiversitdelledeclinazioni,passanoentrambialvagliodi
problemi fondamentali in filosofia come quello della libert dell'uomo, sempre sotteso alle
discussionimache,esplicitamente,ricorretantonell'articolofrancofortesedelgiovaneVolkan
Cidam, in cui si sottolinea quanto la libert in Marx fosse una libert radicale la riconduce
quindi all'idealismo tedesco e perch sia necessario recuperarla in quell'accezione
misconosciutaancheesoprattuttodall'ortodossiamarxista;quantonell'articolodiMargherita
Bottero, della Stockholm School of Economics, che si propone di analizzare per la prima volta
quanto la libert condizioni il benessere e quindi abbia peso nella omonima teoria
economica.
Editoriale

IX

Dalla giustapposizione dei due lavori si noter per esempio quanto l'econometria della
Bottero debba scendere a compromessi con quello che lei argutamente rileva essere uno dei
due fondamentali problemi cui si trovata davanti: che la libert, non ben chiaro come
misurarla.ScriveinveceCidam:

BysocialmediationIrefertoaformofsocialrelationthatplaysaconstitutiveroleforsubject
positions. A constituted subject position denotes a particular attitude towards other subjects
and the outside world. This attitude in return depends upon a particular conceptualization of
freedom.

LariletturadeIlCapitalediKarlMarx,acuradiMartinSattler,professoreemerito(Mannheim)
e direttore della Scuola di Heidelberg, ha invece dalla sua parte una eccezionale sagacit:
metteinfattiilMarxeconomistafuorimercatoenesottolineaillinguaggiovagoemetaforico,
attintodalleSacreScritture.Alcontempo,Sattlerevidenziaanchel'attualitdiquellacriticaal
capitalismo e la sua profeticit: una profeticit sbalorditiva dovuta, probabilmente, al tragico
fatto che il capitalismo di oggi non ha attuato alcuna maturazione reale e mostra ancora le
stessecontraddizionioperativeedepistemologichediquellopresoinesamenell'Ottocentoda
Marx. Scrive infatti Sattler: [...] but the infancy of the capitalist system lasted until our days
and the elements Marx saw in the child period are still with us and they are growing into a
supergigantic scale in our days. Il capitalismo, se osservato nella sua complessit storico
filosofica,risulterebbequindicresciutoinmodoelefantiaco,masenzaunarealematurazione;
e non solo nel senso inteso allora da Marx, ovvero lasciando spazio al comunismo ma, detto
conunaltrolinguaggiopartoritodall'idealismotedesco,quellodiSchelling,approfondendosiin
vorticosa spirale, a partire dai suoi stessi principii approssimativi, quegli stessi che oggi sono
statiapertamenteabbandonatitantodaglieconomistiquantodaifilosofidellascienza.
L'approccio pi proprio della filosofia continentale ci offre dunque ancora quella sua
presenzacriticairrinunciabileelofasiainsistendodivoltainvoltasuisottesipipericolosidi
unadottrinaeconomicaesocialeche,consideratanell'attualeampiezzadellesuearticolazioni,
determina profondamente le nostre vite (The New Spirit of Capitalism, recensione di George
Hull); sia sottolineando la persistenza, nella comune considerazione del capitalismo e dei suoi
strumenti,tantodiunaconcezioneancorametafisica(talvoltaaddiritturadiunatteggiamento
teologicoversoilcapitalismo,diunafedefattadicredenzeedogmi)quantodiunaconcezione
darwiniana, di organismo vivente (pensiamo solo all'idea di autoregolamentazione del
mercato).
L'impiantoteologicoel'organicismosonoentrambiretaggidiquellemodernefilosofiedella
storia che, se da un lato hanno contribuito ad elevare la scienza economica a prassi di
pensiero, dall'altro le hanno consegnato quelle stesse criticit che, si capito poi, hanno
impeditolorodiesserestrumentidianalisidurevoli.
Concludendo,alcunecredenzesottesesimantengonoinsenoallanostraculturainascoltate
evannoacostituireilpotenzialedinuovederivedisenso,chemalsiconformanoall'ambizione
dellascienzaeconomicadiesserelanarrazioneprivilegiatadelreale.

LACOMPRENSIONEDELL'UOMO

Diversiapprocciquindimanontanto,anostroparere,nelmetodo,quantonell'individuazione
ecircoscrizionedelproprioambitodilavoro.Lafilosofiasiattuainentrambiicasiosservandoi
diversiaspettifondamentalideglistessifenomeni,fenomenichesembranoriguardarcisempre
pi da vicino (The Culture of the New Capitalism, recensito da Marco Solinas). Ci accorgiamo

Humana.MenteIssue10July2009

X

infatti di quanto i temi dibattuti nella filosofia dell'economia, per quanto ancora una volta
moltotecnici,sianoancheparticolarmentevicinianoitutti.
qui,nell'ambitodellacriticaall'economia,chesigiocaoggilacomprensionedelmondoe
dell'uomo. il luogo in cui i filosofi danno battaglia per fornire, da un lato, un modello
scientificounificanteedunorizzontecomunealledisciplinepidisparatechecollaboranoallo
sviluppodell'economia,comepossiamoleggerenell'articolodiCaterinaMarchionni;dall'altro,
permantenerevigililecoscienzedegliscienziatistessidavantiall'avanzaredinuoveevecchie
istanzedellacomunittutta:istanzeesistenziali,socialiedambientali.

LETTERATURASCIENTIFICAEINTERDISCIPLINARIET.ILPRIMATOGNOSEOLOGICODELL'ECONOMIA

Come filosofia della scienza, la filosofia dell'economia ha una letteratura mista, scritta dagli
stessi economisti da un lato e dai filosofi dall'altro. Ancora una volta risulter chiaro il suo
carattere interdisciplinare ed innovativo, direte voi, ma attenzione: essa eredita con una
siffatta eterogeneit e ricchezza anche un'impostazione finalistica che le impedisce, se non
integrata con una pi profonda riflessione, di svilupparsi liberamente. Nonostante infatti
l'economia ci convinca oggi di essere la candidata favorita al primato gnoseologico, tuttavia
questainterdisciplinarietchelacoadiuvasistaancoraelaborando.
L'economia ha una sua connotazione interpretativa pervasiva ed attinge a piene mani dai
modi e dai metodi delle scienza naturali e umane; i temi sono dunque tanto attuali quanto
atavicamentelegatiallafilosofiaquali,daunlato,lalibert,larelazioneel'intersoggettivit,il
desiderio, l'identit, il conflitto, la comunit che possono farle da premessa; dall'altro, il
valore, la libert di scelta, la preferenza, la definizione di razionalit, l'analisi delle credenze e
delle convinzioni e dei pregiudizi delle persone, su cui la filosofia della mente, per usare un
altroterminegenericoedobsoleto,haancoramoltodalavorare.

STATUTOONTOLOGICOINCERTODEISUOIOGGETTI

Gli oggetti della scienza economica hanno ancora uno statuto ontologico incerto, sopratutto
quando, usati come strumenti tecnici, la discussione sulla loro efficacia sembra rifluire
nell'atavico problema dell'affidabilit dell'introspezione. Ci accade poich essa stessa in
primis una teoresi borderline tra scienza e tecnica, da un lato osservazione invasiva dei
comportamenti umani e dall'altro affinamento dello strumento. Essa , nel suo modo pi
generale,legataallafinanza,allapoliticaedallastrategia.Edperquestoanchecheessapu
ambire oggi, attraverso il lavoro dei filosofi, a quell'indiscusso primato gnoseologico di cui
sopra cosa questa che comporta una certa assunzione di responsabilit: si tratta infatti di
coincidere con un orizzonte semantico esteso all'intera sfera d'influenza della cultura
occidentale,diunlinguaggiochetuttidevonoimparanoaparlare(equisipotrebbeaprireun
ulterioredibattito,enonmenoimportante,sulfattoreeducazionale).
In conclusione, attraverso la filosofia, l'economica assume un nuovo ruolo nel panorama
della ricerca scientifica denotandosi come un'ottima cornice gnoseologica ed una
strumentazioneeuristicaincisiva;tuttavia,nell'intentodiassurgeredavveroalruoloregiochei
filosofi gi le riconoscono, ci accorgiamo come essa manchi ancora sul piano epistemologico,
cosa da cui evidentemente inficiata. L'economia dunque non basta a se stessa e non basta
per impiantare una sistematica; tuttavia oggi capace, vista la sinergia che crea tra le tante
discipline, di proporre un modello scientifico unificante, atto non solo ad interpretare il reale
ma anche a veicolare nuove possibilit, direi uniche, di comprensione tanto della realt
naturalequantodiquelladell'uomo.
XI

PhilosophicallyIncorrect
Crisieconomica:unamodestariflessione

UmbertoMaionchi

Non so a voi, ma a me i meccanismi di funzionamento della Borsa hanno sempre


provocato una certa inquietudine. La semplice idea che un evento, magari anche
insignificanteeanonimo,accadutochissdove,possainfluirepesantementesulvalore
della moneta che ho in tasca, mi sempre apparsa francamente assurda. Com
possibile,misonochiestospesso,chepiccolifattiapparentementesenzaimportanzao
chesembranononaverealcunlegameconiprocessieconomicisirivelinoinvececos
importantiperlinteraeconomiasoloperilfattodiaverinfluenzatolaBorsa?Oggipoi
che le transazioni finanziarie viaggiano velocissime attraverso oceani e continenti,
leconomiamondialeassomigliasempredipiadunapazzalotteriaincuileestrazioni
sisusseguonoadunritmoforsennatoesenzaalcunalogica.
Maallora,misonochiesto,selecosestannopropriocos,qualegradodiaffidabilit
possiamo attribuire alla cosiddetta scienza economica ed ai suoi principi, ai suoi
modelli, alle sue teorie? Quel che accaduto pochi mesi fa (e che sta tuttora
continuando ad accadere) in tutto il mondo ha mostrato chiaramente, anche ai pi
scettici,lafragilitdelsistemaeconomicoglobale.
La crisi finanziaria che ha dato lavvio al processo solo laspetto pi eclatante e
superficialedellacrisicomplessivadiunmodelloeconomicoormaiadottatodalmondo
intero. E a me, in particolare, appare sempre pi evidente che proprio la nozione
classica di economia di mercato a dimostrarsi, ogni giorno di pi, una pericolosa
utopia.Certo,ancheilmodellocomunistadeipianiquinquennali,disovieticamemoria,
sirivelquasisubitounaingenuautopia,maoggipossiamotranquillamenteaffermare
chequellaeramoltomenopericolosa:idannicheavrebbepotutoprodurresarebbero
statisicuramentepilievi.purvero,infatti,cheunsistemaeconomicocaratterizzato
dallabassaproduttivitdeterminaunmercatomenovivaceeunacompetitivitfiacca.
Ma altrettanto innegabile che la catastrofe che stiamo sfiorando in gran parte
dovuta ad un drammatico eccesso di produttivit. Per non parlare dello scandaloso
spettacolo della distribuzione della ricchezza mondiale e delle differenze, ormai
drammatiche ed esplosive, nello sfruttamento delle risorse, elementi che concorrono
entrambiallainstabilitgenerale.
Daltrondealtrettantoinnegabilechetuttiglistrumentifinoramessiincampoper
arginare questa deriva si sono dimostrati inefficaci o, addirittura, in qualche caso,
capaci di aggravarla. Come esempio tipico mi vengono in mente gli interventi del
FondoMonetarioInternazionale(FMI)che,comeinmoltiormaiconcordano,nonsolo
nonhannoridottoledistanzetraleeconomiedeiPaesiricchiepoverima,inmolticasi,
le hanno approfondite, portando intere nazioni sullorlo del collasso (ricordo, di
sfuggita,tuttelecampagneumanitarieperlazzeramentodeldebitodeiPaesiinViadi
Sviluppo).

Humana.MenteIssue10July2009

XII

Credopercicheormaisiachiaroatutti(oaquasitutti!)cheleCRISIECONOMICHE
non si possano trattare n al pari di eventi naturali come un terremoto o un ciclone
(sebbene, anche nel caso del ciclone, si possa invocare la mano delluomo, vedi l
effettoserra),ncomeeventidiviniedeccezionali(anchesecivienechiestospesso
unasortadiattodifede).
In realt, infatti, ci che mi ha colpito nelle analisi e nei commenti di molti
economisti (o presunti tali) la certezza pressoch assoluta che gli eventi economici
cuistiamoassistendononpossanochemanifestareununicacatenacausalee,dunque,
cheesistaunsolomododiinterpretarli.Inaltritermini,ancheselocalmentepossono
essersi presentate delle singolarit, il fenomeno globale viene descritto come un
passaggio contingente, magari difficile ma non strutturale: il modello generale non
vienepostoindiscussione.Edproprioquestasicumeralaspettopreoccupante.
Ad un osservatore appena un po distaccato, le clamorose smentite alle previsioni
sbandierate anche solo qualche mese fa non possono che apparire come un
importante indizio di fragilit teorica. Daltra parte, che leconomia non potesse
fregiarsiapienotitolodellappellativodiscienzaesattalosannoelosapevanounpo
tutti, eccetto una minima percentuale di fanatici in cerca di una copertura di
legittimit!
Se poi cerchiamo di argomentare pi liberamente lasciando in secondo piano
questioni che potremmo definire ideologiche, vorrei osservare che se dovessimo
prenderesulseriolascientificitdelleteorieeconomichesullapiazza,avremmoanche
lobbligo di sottoporle, com usuale, a test di tipo sperimentale secondo criteri e
protocolli moderni, ampiamente accettati in ambito scientifico. In realt, in questi
ultimi tempi stiamo assistendo alla nascita di nuove discipline che cercano di fare
proprio questo: dotandosi di strumenti pi raffinati presi in prestito dalle
neuroscienze,inparticolare,cercanodianalizzareeventi,modellieteorieeconomiche
in una prospettiva di tipo scientifico allo scopo di formulare previsioni sempre pi
attendibilieprecise.

Maseoratorniamoaconsiderazioniditipopigenerale,ciaccorgiamosubitoche
forseladifficoltpigrandenellaffrontareletematicheeconomichestanelfattochei
principi del liberalismo classico vengono tuttora assunti come verit di fede, come
postulati unici ed indiscutibili: una sorta di atteggiamento fideisticoreligioso che
contrastaplatealmenteconlevidenzaempiricadiretta.
E allora, se le cose stanno cos (ed abbiamo buone ragioni per crederlo!) mi
permetto di suggerire un modesto rimedio, frutto di familiarit con discipline un po
esotericheediesperienzeche,storicamente,hannodatobrillantirisultati.
Poich abbiamo bisogno di un profondo ripensamento dei concetti e dei metodi
fondamentali delle discipline economiche, perch non seguire lesempio della
cosiddetta rivoluzione geometrica del XIX secolo e cercare di costruire una sorta di
economia noneuclidea, sulla negazione di qualche assunto di base, di qualche
postulato critico (come, ad es. quello che assicura la totale autoregolazione del
mercato libero)? Oppure pensare ad una profonda rielaborazione dei principi
economicisullasciadellarelativiteinsteinianaeprovareadelaborareunateoriache
OurColumn:PhilosophicallyIncorrect

XIII

potremmo chiamare: macroeconomia relativistica? Oppure ancora perch non
pensare una microeconomia quantistica, che tenga conto dellindeterminismo dei
microeventilegatiallesceltesoggettive,odelleinterazioninonlocalidelleBorse,dei
ruolicorrelatiosservatoreevento?

Trascinatodallentusiasmomipermettoanchediaggiungereche,perconvinceregli
scettici pi riottosi, sarebbe auspicabile far precedere tutti questi tentativi dalla
dimostrazionediunospecialeteoremadiincompletezzaperleconomiadimercato,
cheriuscisseamettereinevidenzalasuaintrinsecaimpossibilitdiautofondazionee
autogiustificazione. Un tale risultato mostrerebbe anche (proprio come nel caso
dellaritmetica formale) che impossibile una assiomatizzazione finita delle teorie
economiche e che, quindi, per definirle e legittimarle, necessario ricorrere a
strumentieconcettiesterniadessee,ingenere,pipotenti(lanozionedirazionalit
limitataoquelladibenecomune,tantoperfaredueesempimoltosemplici).
Chiudonotandochelanozionedieconomiadimercatoassomigliasempredipia
quella di libera caccia: ambedue si possono praticare finch sussistono condizioni
naturali molto particolari e restrittive e tali da costituire, curiosamente,
un'astrazione vicina allutopia. Una cosa comunque certa: il liberismo economico
classico, nonostante gli aggiornamenti e le modifiche, resta oggi una delle teorie pi
vecchie e che hanno tratto meno profitto dal tumultuoso progresso delle scienze.
Volendo a sua volta definirsi teoria scientifica, questo fatto non rappresenta un
piccolodifetto.

Humana.MenteIssue10July2009

XIV


XV

Conference
Neuroeconomics:HypeorHope?
Rotterdam,2022November2008

FedericaRusso


The conference Neuroeconomics: Hype or Hope? was held in Rotterdam on 2022
November 2008. It was been hosted by the Erasmus Institute for Philosophy and Economics
(EIPE), coordinated by an excellent scientific committee (Jack Vromen, Caterina Marchionni,
JulianReiss,FransdeWinden)andanefficientorganisationassistant(JoshuaGraehl).
Giventhefairlyrecentadventandfastdevelopmentofneuroeconomics,EIPEsoughttoset
up a platform for discussion on the very foundations, aims, and empirical findings of
neuroeconomics.
Theorganisersinvitedleadingscholarsinthefieldandsolicitedcontributedpapersaround
questions such as: What standards of scientific respectability and progress are implied (or
invoked)intheclaimthatneuroeconomicswillfinallymoveeconomicsintoitsproperstanding
ofamodernscience?;Whatconsequencesdoesneuroeconomicshaveforthesubjectmatter,
scope and method of economics?; How do the different disciplines of economics and of
cognitive neuroscience relate to each other in neuroeconomics? Does the relationship
between economics on the one hand and cognitive (neuro)science on the other need to be
redefined?; Do we first need to know how different levels of analysis (e.g. of observable
choice behaviour, of its underlying computational algorithms and of the neural hardware in
which they are implemented) relate to each other before we can tell how neuroeconomic
evidence and findings bear on economics? If so, what levels are at stake and how are they
related?; What light can insights from contemporary philosophy of mind shed on the topics
raisedhere?;Howisneuralactivityinpeoplerelatedtothevariousinstitutionsinwhichthey
function? How can an improved understanding of neural processes inform institutional
analysis?;Whatistheroleandplaceofevolutionarytheoryinneuroeconomics?.
Answers to these and other questions varied greatly, and this comes as no wonder since
the very definition of neuroeconomics is currently under debate. The list of keynote speakers
was remarkable: Ariel Rubinstein (Tel Aviv University, New York University), Paul J. Zak
(Claremont Graduate University), Don Ross (University of Alabama Birmingham, University of
Cape Town), John Davis (University of Amsterdam, Marquette University), Uskali Mki
(University of Helsinki), Jack Vromen (Erasmus University Rotterdam), Francesco Guala
(UniversityofExeter,SanRaffaeleUniversity).
The invited speakers spanned very different themes. Francesco Guala presented the first
results the research project The Neuroscience of Conventions and Norms at Exeter; Ariel
Rubinsteinexpressedhisdeepscepticismaboutthepossiblesuccessofneuroeconomics;Jack
Vromen was concerned whether looking into the brain/mind would really open the black box
of decision making; John Davis discussed what neuroeconomics can contribute to
understanding the concepts of individual and of identity; Don Ross considered, inter alia,
whetherandhowgametheorycanenlightenneuroscience;UskaliMkiofferedamilderview
where neuroeconomics is at the same time a hope and a hype; finally, Paul Zak presented
experimentsusedtoidentifycausalrelationshipsthatcanbeusedtoimprovecurrentmodels.
Ten invited contributions touched on a variety of other topics, yet the dominating theme
certainly was the foundational issue: what is exactly neuroeconomics and is it really a

Humana.MenteIssue10July2009

XVI

methodological and conceptual revolution? For the full list of contributed talks see the
programmeonlinehere:

(http://www.eur.nl/fileadmin/ASSETS/fw/Neuroeconomics_Conference_Program.pdf).

Alltalkswereremarkablygood.Arelaxedandinformalatmospherecertainlycontributedto
having outstanding discussions. It has been a merit of the invited speakers to deliver
presentation highly specialised and nonetheless accessible to the nonspecialised audience.
This, in my view, allowed to set up a common ground for a critical appraisal of
neuroeconomics. The format of the conference has been a very well chosen one as well.
Contributed papers had a discussant to kick off the discussion. This very much facilitated
exchangesbetweentheparticipants.
Itseemsthat,asoftenhappensinphilosophy,nodefiniteanswerhasbeenreachedtothe
question of whether neuroeconomics be a hype or a hope. It has been suggested by various
peoplethatthefactthatitmaybewellbothatthesametimeisreasonenoughtopursuethe
neuroeconomicprojectaccompaniedbyfoundationalandmethodologicaldiscussions.


XVII

Conference
Decision,GamesandLogic2009
HECLausanne,1517June2009

UmbertoGrandi

The third edition of the workshop on Decision, Games and Logic was held in Switzerland, at
HEC in Lausanne, from the 15th to the 17th of June. It was coordinated by an excellent
organizing committee composed of Christian Bach and Alessandro Facchini from the local
institute, Conrad Heilmann from LSE (London School of Economics) and Olivier Roy from the
University of Groningen. The format of the workshop reflected the scope of the past two
editions: three high profile tutorials and very participated comment sessions aimed at
fostering interaction between graduate students and young researchers in the areas of
decision theory, game theory and logic. New features of this edition has been a panel on
What is Rationality? and the P. van Emde Boas Swap Session, during which various
researcherspresentedeachother'sworkinajointpresentation.
The first tutorial was given by Jaques Duparc, Professor of Formal Logic at HEC Lausanne,
and aimed at introducing the basis of calculus, an extension of modal logic that has a great
potentialinmodelingfixedpointrelatednotionsinepistemiclogicandingametheory.Onthe
seconddayoftheworkshopPierpaoloBattigallifromBocconiUniversitygaveadetailedlesson
on type structures. He presented in detail the mathematical assumptions underlying this
model, widely used in game theory to represent interactive beliefs, and outlined the formal
characterisation of backward and forward induction using these structures. The argument of
the third and last tutorial, given by Luc Bovens from LSE, was an introduction to Bayesian
epistemology.Themainscopeofthelessonwastoshowhowmuchprobabilisticmodelshave
to offer to formal philosophy, a scope successfully accomplished trough the use of many
insightfulexamples.
ThePVeBSwapSessionwasthemainactivityduringthesecondpartofthedays.Theidea
of this session, originated during the last edition of DGL in Amsterdam, was to have two
researcherspresentingthemainideasofeachother'swork,withcommentsandquestionsfor
a final reply of the author. Many talks gave rise to long discussions with the audience.
ExamplesincludeanexchangebetweenGroningenandParis(wherethenextDGLwillbeheld)
whereOlivierRoypresentedaworkondynamicepistemiclogicforvagueinformationbyPaul
Egre, who replied presenting Oliviers work on agreements theorems using weaker
assumptionsofcommonbelief.
The amount of contributed talks was very much reduced with respect to the previous
edition.BetweenthemitisworthnotingapresentationbyJulienDutantfromtheUniversityof
Geneva,whoproposedamoregeneralmodelforknowledgeasbeliefbasedonasafemethod,
and a thorough formal study in the area of peer disagreement by Giacomo Sillari from the
University of Pennsylvania. During the first day of the workshop a very successful poster
sessionprovidedvariousyoungresearcherwithusefulcomments.
For a list of all talks and posters consult the programme on the website of the workshop
(http://www.meansandends.com/workshop09/?body=programme).
The last event of the second day of the conference was a very successful panel on
rationality chaired by M. Tomassini. Three researchers, representing the full spectrum of
opinionsaboutrationality,developedaninterestingdiscussionwithgoodinteractionfromthe
audience. Richard Bradley from the LSE presented his middle point view of rationality as

Humana.MenteIssue10July2009

XVIII

coherence,andhasfoughthard("bitingthebullet"ashehaverepeatedlysaid)againsthistwo
opponents: Ulrich Hoffrage from HEC Lausanne, whose concept of rationality as adaptation
allows inconsistency as an adaptive strategy, and Pascal Engel from the University of Geneve,
moreentrenchedonmoreclassicalrationalistpositions,stressingtheimportanceoftruthside
bysidewithcoherence.Thedayclosedwithanexcellentdinnerenjoyingasuperbviewonthe
lakeofLausanne.
The workshop has been highly successful in fostering interaction between researchers of
different communities and institutions. The tutorials were of high quality and very technical,
whereas the main approach of the conference remains philosophical. Some areas, however,
werenotextensivelycovered,
i.e. foundational and epistemic aspects of decision and game theory, where logic plays a
more explicit role, and topics that include interactions from computational logic. The "PvEB
Swap Session" was the main focus of the workshop, and resulted in insightful comments for
thespeakersandinterestingdiscussionsfortheaudience.Onedefectisthatitwassometimes
hard to listen to a work not directly from the source. Moreover, this session also limited
sharply external contributions to four talks and a poster session. Nevertheless, a great
compliment is due to the organizing committee, who succeeded in preparing a high profile
programmeandanenjoyablesocialevent.Thenexteditionoftheworkshophasalreadybeen
announcedandwillbeheldinParisin2010.

XIX

Annuncio
TwistOntology

AlessandroBemporad

RobertaLanfredini

PRESENTAZIONE

Questolavorosinseriscenellambitodiunprogettocheharecentementepresoavviopressoil
Dipartimento di Filosofia dellUniversit di Firenze. Si tratta dellistituzione di ENTIALAB
1
, un
Laboratorio congiunto fra Universit e Impresa (Symmetric
2
, software company con sede a
Firenze). Il Laboratorio si occupa di Ontologia Applicata, cio della messa a punto di possibili
modelli conoscitivi in grado di rappresentare oggetti o, per meglio dire, particolari domini di
oggetti.
Il Laboratorio di ontologia applicata nasce dalla condivisione di due campi di studio e di
applicazione tra loro apparentemente eterogenei: da un lato lontologia filosofica e la teoria
dellaconoscenza,dallaltrolanalisi,laprogettazioneelimplementazionedisistemisoftware.I
membridelLaboratorioprovengonoquindidasettoriprofessionalimoltodiversi:innanzituttoi
settori maggiormente connessi alla filosofia teoretica e allingegneria del software,
immediatamentedoposettoridialtrotipo,quali,adesempio,lamedicinaelagenetica.
ENTIALAB ha fra i propri obiettivi fondamentali quello teoricofilosofico di individuare e
definire le categorie fondamentali della realt, le strutture profonde che si nascondono tra le
pieghe del mondo che ci circonda. Strutture che, si suppone, rimangono costanti al variare
dellesperienzachediquelmondonoituttifacciamo.Essohatuttavia,comesecondoobiettivo
fondamentale, quello di sottoporre le ipotesi cos definite al vaglio della loro concreta
applicabilit, oltrepassando e, per cos dire, mettendo alla prova la loro natura puramente
speculativa.IlcampodiapplicazionealqualeilLaboratoriofariferimentolimplementazione
di un insieme di sistemi software di nuova concezione, denominati knowledge organization
systems, i quali consentono di acquisire, mantenere e elaborare in modo automatico modelli
informativi estremamente complessi, attraverso uno schema di rappresentazione formale
dellaconoscenzabasatosuontologie.
Labaseinformatizzatadiconoscenzadiquestisistemisoftwarealimentataapartiredalle
informazionielementarichepossonoessereacquisitedalmondoesternoneimodipisvariati,
quali ad esempio flussi di dati, segnali provenienti da sensori o dispositivi elettronici, dati
linguistici e semantici generati da fulltext retrieval, insomma informazioni provenienti da
qualunquefontesensorialedigitale.
A questo proposito occorre segnalare come la stragrande maggioranza dellattuale
terminologiainformatizzatasibasisunozionieregolediclassificazionevagheeimprecise.Tale
disomogeneiteopacitponedueproblemipispecifici.Ilprimoproblemacheunanonben
definitadeterminazionedellarealtrappresentatafausciredallemaglieunagrandequantit
didatirilevanti.Ilsecondoproblemacheunatraduzioneinefficientefrasistemialternatividi
rappresentazione finisce inevitabilmente per rendere molti dati non traducibili e non
comunicabili. A questi due problemi corrispondono quelle che, allo stato attuale, possono
essere considerate le due maggiori esigenze dei processi dinformatizzazione: esatta

1
www.entialab.org.
2
www.symmetric.it.

Humana.MenteIssue10July2009

XX

determinazione dello schema rappresentativo utilizzato e traduzione degli schemi


rappresentativi(spessotacitieinconsapevoli)gidisponibili.
LericadutepratichedellattivitsvoltadalLaboratoriosonomolteplici.Oltreacontributidi
ricerca relativi allontologia e alla teoria della conoscenza, il Laboratorio intende apportare
elementi di innovazione tecnologica, con particolare riferimento a settori strategici quali la
medicina, la clinica e il dominio sociosanitario. Coadiuvando, sia pure indirettamente, il
miglioramento generale dei processi di cura. Il laboratorio di ontologia applicata intende
inoltre esplicare le proprie attivit, oltre che in ambito accademico, nel campo della
formazione e della consulenza, allinterno di una gamma potenzialmente molto ampia di enti
pubblicieprivati.
infineunadellefinalitdichiaratediENTIALABquelladifavorirelintroduzionenelmondo
del lavoro di giovani dotati di una preparazione mista, umanistica per vocazione ma
direttamente esposta allinnovazione tecnologica, in grado di affrontare i mutamenti e
levoluzionecheilmondostasemprepirapidamentesperimentando.

DiseguitopresentiamounabreveintroduzionealmodelloontologicodenominatoTwistmesso
apuntonellalveodelleattivitdellaboratorio.

PRIMAPARTEINTRODUZIONE

TWISTunsistemasoftwareprogettatoeimplementatoperconsentirelarappresentazionee
lelaborazione della conoscenza, in conformit a un modello ontologico formale sottostante.
Nellaparteintroduttivadiquestasezioneespliciteremoglielementiconcettualichestannoalla
base del sistema, mentre nella seconda parte verr invece descritto nel dettaglio il modello
ontologicosullabasedelqualeTWISTstatoimplementato.

ChecosalOntologiafilosofica

Lontologia quella branca della filosofia che si occupa in senso molto generale e astratto di
ci che vi . Il suo scopo quindi quello di isolare tipi e strutture fondamentali della realt
(oggetti, propriet, relazioni, eventi, processi), fornendone una chiara caratterizzazione. Luso
chequifacciamodellanozionediontologiasostanzialmentedescrittivoeintuitivo,nelsenso
chesisforzadianalizzarefedelmenteilpuntodivistadelsensocomune,aderendoquantopi
possibileallimmaginemanifestadelmondo.
Ladeterminatezzadellavoroontologiconondeveessereconfusaconlasuaunivocit.Non
esiste,infatti,unsolomododiconcettualizzarelarealt.
Nella recente filosofia analitica viene proposta, a questo proposito, una distinzione fra
metafisicaeontologia
3
Ladistinzionerisultacosconcepita:lontologialateoriachestabilisce
cichevi;lametafisicainvecelateoriachestabiliscecosacichevi.Dauncertopunto
di vista quindi la metafisica, intesa come caratterizzazione descrittiva, preliminare. Non
affatto chiaro in che misura si possa stabilire se esistano certe entit senza offrirne una
caratterizzazione.Questofattorisultaevidentenellescienzenaturali:nonsipu,adesempio,
scoprirelesistenzadiunanuovaspeciesenzasaperedichecosasitratta.Analogamente,non

3
Si veda A.C.Varzi, Ontologia, RomaBari, Editori Laterza, 2005 e E.Runggaldier e C.Kanzian,
Grundprobleme der analytischen Ontologie, Paderborn, Verlag, 1998; trad. it. Problemi fondamentali
dellontologiaanalitica,Milano,VitaePensiero,2002.
AnnuncioTwistOntology

XXI

cisipulimitareadirecheesistonooggettimaterialisenzadarneunaqualchedescrizione(ad
esempio,glioggettimaterialisonoentitconcrete,impenetrabili,estesenellospazio,dotatidi
certeproprietessenziali,ecosvia).
Da un altro punto di vista, tuttavia, lontologia ad essere prioritaria. Il carattere
prospettico delloggetto rimanda infatti in misura essenziale alla convergenza delle varie
prospettive in un unico polo. Metafisica e ontologia contribuiscono quindi di pari passo alla
descrizioneeallidentificazionedeglioggettirelativiaundeterminatodominio.
Uno dei problemi fondamentali che lindagine ontologica si trova ad affrontare quello di
riuscire a fornire una relazione virtuosa fra metafisica e ontologia, individuando le
caratterizzazioni che meglio riescono a modellare e concettualizzare un determinato dominio
oggettuale. Ci pu avvenire solo mediante unaccurata caratterizzazione e unopportuna
identificazione degli oggetti appartenenti al dominio che ci troviamo a descrivere. Le
descrizioni che non presuppongono una modellazione concettuale, quindi unanalisi
ontologica, hanno spesso carattere informale e risultano per questo ambigue, circolari, o
addiritturacontraddittorie.
La stragrande maggioranza dellattuale terminologia informatizzata, inoltre, basata su
nozioni e regole di classificazione formulate in modo impreciso, il che finisce per far uscire
dalle maglie della descrizione moltissimi dati, e per renderne molti altri non traducibili e non
comunicabili.

OntologiaFormaleeOntologiaMateriale

Per risolvere questa difficolt, lontologia filosofica propone la distinzione fra ontologia
formaleeontologiamateriale
4
.
Lontologia formale neutrale rispetto a qualsiasi dominio particolare; essa si occupa in
sensogeneralissimodelloggettointesocomeunvuotoqualcosaedellesuecaratterizzazioni
(ad esempio la relazione partetutto; la differenza fra astratto e concreto; lidentit).
Lontologiaformale,persuastessadefinizione,svuotatadiqualsiasicontenutodeterminato.
Essavalequindiperglioggettiingeneraleo,ilchelostesso,perqualsiasioggettodiqualsiasi
dominioconsiderato.
Lontologia materiale si occupa, al contrario, di oggetti che appartengono a domini
determinati. Essa richiede quindi una caratterizzazione e una modellazione che faccia
riferimento alla struttura contenutistica (o qualitativa) dellambito considerato (ad esempio
oggetti,proprieterelazionirelativeaundominioontologicoditipobiomedico).lontologia
materialechepermettediformulare,eventualmente,nonsolodescrizionimaancheprevisioni
scientifiche.Lanalisipuinfattiindividuarelemergereditendenze,dilineeontologiche,da
interpretare,adesempio,attraversolemetodologieeglistrumentidellamedicinapredittiva.
La struttura materiale pu essere inoltre calibrata su diverse modularit o granularit: si
parla di caratterizzazioni a granularit fine (ad es. il numero e lordine dei geni in un

4
SivedaE.Husserl,IdeenzueinerreinenPhnomenologieundphnomenologischePhilosophieErstes
Buch: Allgemeine Einfhrung in die reine Phnomenologie, Husserliana III/1, Nijhoff, Den Haag 1976;
trad.it.Ideeperunafenomenologiapuraeperunafilosofiafenomenologica,vol.I,Einaudi,Torino2002
e E. Husserl, Terza ricerca logica, in Logische Untersuchungen, Zweiter Teil/1: Untersuchungen zur
Phnomenologie und Theorie der Erkenntnis. Zweiter Teil/2: Elemente einer Phnomenologische
Aufklrung der Erkenntnis, Husserliana XIX/12, Nijhoff, Den Haag 1984; trad. it. Ricerche logiche, Il
Saggiatore,Milano1968.

Humana.MenteIssue10July2009

XXII

cromosoma)oppuredicaratterizzazioniagranagrossa(es.ladescrizionediunamalattiaedei
suoisintomi,delcomportamentoedellesensazioniassociate).
Ognuno di questi punti ontologia formale, ontologia materiale o di dominio, granularit
delladescrizionedevonoesseresceltiefissatinelmodopichiaroenettopossibile.Ognuno
di essi concorre infatti a delineare la cornice metodologica allinterno della quale si colloca il
compito dellontologia. Si tratta di un compito che possiamo definire antiriduzionista e
qualitativo. Antiriduzionista, perch lanalisi descrittiva dellontologia interessata a tutti i
livelli di granularit (scientifico e comune; microfisico, mesocosmico e cosmologico).
Qualitativo,perchletassonomieontologichenonsirivolgonosoloacichequantificabilee
misurabile, bens anche (e forse soprattutto) a ci che accessibile solo mediante descrizioni
ditipoqualitativo.

UniversalieParticolari

Unulteriore distinzione, che ci sar utile per comprendere il modello ontologico che
prenderemofrapocoinconsiderazioneladistinzionefrauniversalieparticolari.
Gli universali corrispondono a quanto vi di generale nella realt. Essi sono ripetibili nel
senso che possono essere esemplificati da pi di un oggetto e in pi momenti temporali. In
virt di questa loro caratteristica, gli universali non sono localizzati in modo determinato nel
tempoenellospazio.Essipossonoesistereintuttiitempieintuttiiluoghie,inquantotali,si
diconoirreali.
Iparticolari,alcontrario,nonsonoripetibiliinquantopossonoesisteresoloinunluogoein
untempodeterminati.Iparticolariesemplificano(oistanziano)universalienonpossonoaloro
voltaessereesemplificati.
in virt della relazione di esemplificazione di uno stesso universale che due particolari
possono essere considerati simili sotto qualche aspetto (ad esempio due individui essere
consideratipersone).Iparticolarisonoquindifugaci,contingenti,irripetibili;gliuniversalisono
invece quanto nella realt vi di stabile, permanente, ripetibile. Il particolare un fatto, un
accadimento.Luniversaleunmodello,unidea.
Il concetto di universale distinto da quello di classe. Ogni universale ha una classe
corrispondente, ma non vale il viceversa, cio non ogni classe corrisponde a un universale. La
classe infatti un insieme o una collezione di particolari. Vi sono termini generali la cui
funzione di riferirsi a particolari che esistono in un luogo e in un tempo determinati (ad
esempio,laclassedeinatiaFirenzedal1959al1963;oppurelaclassedeilaureatiinfilosofia
nellultimodecennio).Inquestocasosifariferimentononauniversalioatipigenerali,bensa
gruppiocollezionidientitparticolari.Larelazionechesussistefrailparticolareeluniversale
diesemplificazione;quellachesussistefrailparticolareelaclassediappartenenza.

Oggettualit,determinazioniedentit

Infine,lecategorieontologicherisultanoorganizzategerarchicamenteapartiredaunulteriore
distinzione fondamentale: quella fra oggettualit e determinazioni. Entrambe le categorie
rimandano inoltre a una categoria superiore: quella di entit. Le oggettualit sono entit
costituite da fasci di determinazioni. Le determinazioni sono, in modo complementare, entit
checonfluisconoinoggettualit.
AnnuncioTwistOntology

XXIII

Alfinedichiarirnelanatura,ladistinzionefraoggettualitedeterminazionipuessereresa
anche come distinzione fra parti indipendenti e parti dipendenti
5
: le parti indipendenti sono
quelle entit che godono della possibilit di una autonomia concettuale e rappresentativa; le
parti dipendenti sono quelle entit che non godono di tale autonomia, ma che necessitano
invecedialtrepartidipendentiperpoterdarluogoaoggettualitindipendenti.
Un foglio sar quindi unentit indipendente, mentre il colore sar unentit dipendente.
Mentre il primo pu sussistere da solo, il secondo necessita di integrazione (nel caso del
colore, di essere integrato con lestensione). Analogamente, un individuo sar una parte
indipendente; mentre il suo essere medico (un ruolo di quellindividuo) sar una parte
dipendente (non esistono medici che non siano anche individui). Si noti: ogni entit , per
definizione, anche una parte. Questo vale non solo le parti dipendenti ma anche per quelle
indipendenti. Un individuo sar, ad esempio, parte di un pi ampio sistema sanitario, o di un
organico,ecosvia.
Chiameremo la necessit di integrazione fra parti dipendenti relazione di fondazione. Le
determinazioni devono fondarsi con altre determinazioni per dare luogo a parti indipendenti
(oggetti,intesicomeinteriofrazionidiinteri).
Le determinazioni possono essere universali o particolari. Nel primo caso si parla di
propriet,nelsecondodimodi.Inmodospeculare,ancheleoggettualitpossonoesserealoro
voltauniversalieparticolari.Nelprimocasosiparladitipi,nelsecondodioggetti.
Se interpretate come universali, cio come propriet, le determinazioni possono inoltre
godere di livelli diversi di generalit: un primo ordine quello della specie (ad esempio la
propriet dellessere rosso, o della rossezza); un ordine superiore quello del genere (ad
esempiolaproprietdellessercolorato).Lerelazionifraordinisonorelazionidinclusione(ad
esempiolesserrossoincludelessercolorato).
La combinazione delle due distinzioni categoriali fondamentali quella fra universali e
particolari e quella fra oggettualit e determinazioni introduce il particolare modello
ontologicocheprenderemoorainconsiderazione:ThefourCategoryOntology.

FCO:TheFourCategoryOntology

Possiamomettereapuntounontologiacheprevedaduecategoriefondamentalidiparticolari
eduecategoriefondamentalidiuniversali.Ilrapportospecularefraleduecoppieoriginauna
sortadiquadratoontologico
6
.
Le due fondamentali categorie di particolari sono gli oggetti e i modi. Le due categorie
fondamentalidiuniversalisonoitipielepropriet(erelazioni).LePropriet(universali)stanno
aiTipi(universali)comeiModi(particolari)stannoagliOggetti(particolari).
ProprieteModihannocomegeneresuperioreilconcettodideterminazione.TipieOggetti
hannocomegeneresuperioreilconcettodientit.Ilcheunaltromodoperdirecheesistono
entit particolari ed entit generali e che le determinazioni di quelle entit possono essere a
lorovoltaparticolariogenerali.
In sintesi, possiamo affermare che in ciascun dominio applicativo, allinterno quindi di una
determinata comunit semantica, coloro che vi operano distinguono i fenomeni che si
presentano nel corso delle loro attivit tra oggetti (concettuali o fisici), propriet di questi

5
E.Husserl, Terza ricerca logica, cit. Si veda anche R.Lanfredini (a cura di), A priori materiale, Milano,
Guerini,2006.
6
Si veda anche E.J.Lowe, The Four Category Ontology. A Metaphysical Foundation for Natural Science,
Oxford,OxfordUniversityPress,2006.

Humana.MenteIssue10July2009

XXIV

oggetti, e relazioni tra questi oggetti. Viene inoltre riconosciuta una differenza tra gli oggetti
individualieditipi(igeneri,lespecie)cheliconcettualizzano.
La determinazione dei tipi, delle propriet e delle relazioni che contraddistinguono gli
oggetti presenti in un determinato dominio, una volta individuati, descritti e classificati,
formano quindi unontologia, la quale pu successivamente essere implementata nei sistemi
software per asserire fatti sul mondo, comprensibili ed utilizzabili sia da esseri umani che da
sistemiinformatici.
EsaminiamoadessopidavicinoilmodelloFCO,laFourCategoryOntology:

Comepossibileosservarenellaprecedentefigura,ilmodelloFCOruotacomeabbiamodetto
attornoaquattroprincipalicategorieontologiche,cheandremoadessoadintrodurre.

La prima categoria, in un certo senso quella fondamentale, costituita


dagli oggetti individuali, o particolari (OBJECT). Per oggetto intendiamo
una qualunque entit individuale (concreta o astratta, dislocata nello
spaziotempo o anche solo nel tempo), dotata di una propria identit
riconoscibile,laqualesiadotatadiunaseriediproprietchelacaratterizzanoediunaseriedi
relazioni che la connettono con altri oggetti. Esempi di oggetti quindi possono essere singole
entit come una certa persona, un determinato luogo, uno specifico manufatto o un
particolare oggetto naturale (come una pietra, o un fiume), cos come anche singoli eventi
qualiadesempiounacertavisitamedica,undeterminatomatrimonio,lospecificopagamento
di una multa, un certo prestito di un libro in una biblioteca, etc. In questa prospettiva, gli
eventisonoquindiconsideraticomeoggettiatuttiglieffetti,dotatidiproprietedirelazioni.

La seconda categoria quella dei tipi di oggetto (KIND). Un tipo la


rappresentazione concettuale di un insieme di oggetti individuali
connessi tra loro da una natura comune, o da un insieme di propriet
possedute da tutti i membri di quel tipo. Rispetto agli oggetti
individuali,particolari,itipidevonoessereconsideraticomeconcettiuniversali,ingradopoidi
essere esemplificati attraverso singoli oggetti particolari. Persona, luogo, Organizzazione,
Manufatto,VisitaMedica,sonotuttiesempivalididitipidioggetto.

Laterzacategoriaquelladellepropriet(PROPERTY)edellerelazioni
(RELATION). Una propriet una specifica caratteristica, o qualit,
posseduta da tutti gli oggetti di un certo tipo. In altre parole, una
proprietpuessereconsiderataanchecomeunmododiesserediun
AnnuncioTwistOntology

XXV

certotipodioggetti.Adesempio,ilcolorepuessereconsideratounaproprietdeitavoli,cos
comelet,oilnomeproprio,possonoessereconsiderateproprietvalidepergliesseriumani.
Le relazioni sono invece da considerarsi come unestensione del concetto di propriet,
poich non sono caratteristiche intrinseche di uno specifico individuo (relative, cio, a un
singolo tipo di oggetti), quanto piuttosto connessioni, o associazioni caratterizzanti tra due o
pitipidiversidioggetto.Larelazionedimpiego,adesempio,caratterizzaunassociazionetra
un qualunque essere umano ed una qualunque azienda. Altri tipi di relazione sono quella
generativa, la quale dispone una serie di concetti in ordine ascendente di generalizzazione,
oppurequellapartitiva,chedisponeglioggettiingerarchiediaggregazioneedicomposizione
tuttoparti.

Neldescrivereleproprietelerelazioni,abbiamofattoriferimentoalla
loro definizione in quanto concetti universali, ovvero come
caratteristiche o associazioni tra tipi di oggetto. Cos come per gli
oggetti nei confronti dei tipi, anche le propriet e le relazioni possono
essere esemplificate individualmente, nel momento in cui vengono assegnate ai (o meglio
possedute dai) singoli oggetti individuali. In questo caso, e questa la quarta categoria nella
nostra ontologia, parliamo pi propriamente di modi (MODE), o di valori delle singole
propriet.
Ad esempio, mentre il nome proprio una propriet (PROPERTY) di un generico tipo
(KIND) Essere Umano, Mario uno modo (MODE), ovvero un valore specifico per la
propriet nome proprio quando questa viene esemplificata da uno specifico oggetto
individuale(unOBJECT,inquestocasounasingolapersona).

Per riassumere quanto detto, e per concludere questa breve introduzione metodologica,
possiamoevidenziareleseguentiimmaginidelmodelloFCO:

UniversaliParticolari

Abbiamovistocome,nellaFCO,iTipi(KIND),lePropriet(PROPERTY)eleRelazioni(RELATION)
sianoentituniversali,mentregliOggetti(OBJECT)ediModi(MODE)deglioggettisonoentit
particolari.

Humana.MenteIssue10July2009

XXVI

Nellafiguraprecedente,possiamoosservarelerelazionichesussistonotralequattrocategorie
cheabbiamodescritto:
0) GliOggettisonoistanzedeiTipi(quindiiTipisonoistanziatidagliOggetti);
1) I Modi sono istanze delle Propriet/Relazioni (quindi le Propriet/Relazioni sono
istanziatedaiModi);
2) I Tipi sono caratterizzati attraverso le Propriet e le Relazioni (le quali a loro volta
caratterizzanoiTipi);
3) Gli Oggetti sono caratterizzati dai Modi specifici (i quali a loro volta caratterizzano i
singoliOggetti).

SECONDAPARTETWISTONTOLOGYMODEL

Unavoltadefinitoilquadratoontologiconellesuelineegenerali,possiamoadessoesaminarlo
un po pi da vicino, allo scopo di introdurre un insieme di estensioni e di precisazioni che ci
condurrannoacompletareloschemaontologicochestiamoillustrando.Ilmodelloontologico,
progettato dal Laboratorio di Ontologia Applicata, che andiamo adesso a presentare
costituisce la base del TWIST ONTOLOGY MODEL, a partire dal quale Symmetric ha
implementato,inunprodottosoftware,ilproprioOntologyEngine.

IlconcettodiCaratterizzazione

Osserviamo quindi per prima cosa la relazione che abbiamo definito sussistere tra le due
categorie di universali introdotte in precedenza. Come abbiamo visto, gli elementi KIND (i tipi
dioggetto)sonocaratterizzatidaunaseriePROPERTY,ovverodallepropriequalit.

Definiamo quindi il concetto stesso di CHARACTERIZATION come un elemento base della


nostra ontologia, che va ad aggiungersi alle quattro categorie del quadrato ontologico per
AnnuncioTwistOntology

XXVII

modellare proprio questidea di caratterizzazione. Possiamo definire la caratterizzazione
osservando come una propriet possa caratterizzare tipi di oggetti diversi. Ad esempio, la
proprietColorepotrebbecaratterizzareiltipoPersona,seintesocomecoloredellapelle,cos
comeiltipoAutomobile,seintendiamoinquestocasoilcoloredellacarrozzeria.
Possiamo rappresentare il concetto di caratterizzazione come una relazione che connette
tralorounaPROPERTYedunKIND,nellaqualelastessaPROPERTYpuessereassociataapi
elementi KIND diversi, ed uno stesso KIND pu essere caratterizzato da pi PROPERTY
differenti:

Lintroduzionedellacaratterizzazioneconsentealnostroquadratoontologicodirappresentare
correttamenteillegamecheunisceglielementiPROPERTYedKIND.
Facciamo per adesso un passo avanti, ed esaminiamo il concetto stesso di PROPERTY.
Introduciamo nella seguente tabella un TOKEN che esemplifica il KIND Persona, lipotetico
avvocato Giovanni Rossetti, e indichiamo alcune delle possibili PROPERTY che caratterizzano
proprioilKINDPersona,assiemeaiMODEconiqualiquestePROPERTYvengonoistanziatenel
casodelTOKENinesame:

PROPERTY MODE
Nome Giovanni
Cognome Rossetti
DatadiNascita 25/03/1958
Professione Avvocato
TitolodiStudio LaureainGiurisprudenza
GruppoSanguigno BRh+
Altezza 178cm
Assistitoda Dr.GiuseppeVerdi
StatoCivile Coniugato
Coniuge FrancescaGalli
ComunediResidenza Firenze

Humana.MenteIssue10July2009

XXVIII

PROPERTY MODE
ComunediDomicilio Empoli

Echiarochestiamosemplificando(lofaremocontinuamentenelseguitodelnostrodiscorso),
edchiaroanchechequestenonsonocertoleunichePROPERTYchepotremmoutilizzareper
caratterizzare una Persona, ma certamente queste sono caratteristiche, qualit che possiamo
associare in generale allidea di persona (perlomeno quando questa persona un cittadino
delloStatoItaliano).

LeProprietSemplici

Scorriamoadessounaperunaquestepropriet.Ciaccorgeremocheessenonsonoproprio
tuttedellostessotipo:essepossonoessereinrealtscompostealorovoltainsottocategorie
della categoria base PROPERTY. Concentriamoci, tanto per iniziare, sul seguente gruppo di
propriet:

PROPERTY MODE
Nome Giovanni
Cognome Rossetti
DatadiNascita 25/03/1958

Ora,senzadubbiopotremmoattribuireaiMODEdiquestetreproprietunvalorequalunque,
purchsirispettinoalcuneregoledibasecheadessoandremoadefinire.Adesempio,ladata
di nascita di una persona pu essere teoricamente espressa da un qualunque valore, purch
rappresenti una data correttamente espressa. Il nome e il cognome di una persona devono
esseresequenzedicaratteriprividicifreedisegnidipunteggiatura(anchesechiarochenon
tutte le combinazioni di caratteri sarebbero riconosciute come nomi propri validi nella lingua
italiana). Attribuiamo a questo tipo di PROPERTY, i cui MODE possono variare liberamente
purchallinternodiundeterminatotipodidati(stringhedicaratteri,date,numeriinteri,etc.)
ilnomediSIMPLEPROPERTY(proprietsemplici):

Questa la prima delle sottocategorie delle PROPERTY che introduciamo nella nostra
ontologia: quella delle propriet semplici, ovvero di quelle qualit degli oggetti che sono
espresse da un valore, e che possono essere rappresentate nel modo pi diretto nel nostro
quadratoontologico,comenelseguenteesempio:

AnnuncioTwistOntology

XXIX

LeRelazioni

Proviamoadessoinveceadosservareilseguentegruppodipropriet:

PROPERTY MODE
Assistitoda Dr.GiuseppeVerdi
Coniuge FrancescaGalli
ComunediResidenza Firenze
ComunediDomicilio Empoli

evidente gi a una prima analisi superficiale che nel caso di queste propriet non possiamo
parlare di propriet semplici, come nel caso del nome e del cognome. In effetti, i MODE di
questeproprietnonsonorappresentatidasemplicivalori,mapiuttostodaconnessioniverso
altriTOKEN,iqualialorovoltaquindisonoesemplificazionidiunproprioKIND.
Adesempio,ilConiugediunaPersonasempreunaltraPersona;ilComunediResidenzadi
unaPersonainvecesempreunComune(ciounoggettodiunaltroKIND):

Come possiamo vedere dai due diagrammi precedenti, questa nuova sottocategoria delle
PROPERTY rappresenta le relazioni (chiamiamola quindi RELATION PROPERTY) e stavolta deve

Humana.MenteIssue10July2009

XXX

essereespressacomeconnessionetradueelementiKIND,deiqualiunoilsubject(ilsoggetto
dellarelazione),elaltrolobject(loggettodellarelazione).
Le RELATION PROPERTY vengono istanziate da MODE che connettono sempre due TOKEN
(anche in questo caso un subject ed un object) come possibile osservare nello schema
precedente,nelqualesirappresentalarelazionediResidenzatraunaPersonaedunComune.

LeClassificazioni

Dopo aver definito le SIMPLE PROPERTY e le RELATION PROPERTY, passiamo adesso a


esaminareinveceilseguentegruppodipropriet:

PROPERTY MODE
Professione Avvocato
TitolodiStudio LaureainGiurisprudenza
GruppoSanguigno BRh+
StatoCivile Coniugato

Certamente queste non sono propriet semplici, in quanto i valori ammessi non sono
liberamenteesprimibili.Possiamopensareanchestavoltaditrovarciinpresenzadirelazionifra
tipidioggetto?Concentriamociinparticolaresuunadelleprecedentipropriet:loStatoCivile.
ChecosloStatoCivile?certamenteunaproprietdellePersonema,pipropriamente,lo
Stato Civile la classificazione di un determinato fenomeno, ovvero quello che rappresenta
unaparticolareformadirelazionetraduepersone.
In questo caso, quindi, i valori ammessi per questo genere di propriet non sono liberi,
come nelle propriet semplici, ma piuttosto vengono scelti da un elenco prefissato, la
classificazione appunto, che nel caso dello Stato Civile potrebbe prevedere, oltre che
Coniugato,ancheglielementiNubile,Celibe,Divorziato,ecosvia.
Datochesononumerosissimeleformediclassificazioneprevisteperdescrivereifenomeni,
introduciamo ancora nel nostro schema ontologico una particolare sottocategoria di
AnnuncioTwistOntology

XXXI

PROPERTY, la CLASSIFICATION PROPERTY, proprio per individuare e descrivere tutte le
proprietdeltipoappenavisto.
Sorgeperadessounproblema,chepossiamoevidenziarenellafiguraseguente:

Se il KIND Persona caratterizzato dalla PROPERTY Stato Civile, e lo Stato Civile una
classificazione (nel senso di un elenco predefinito di rappresentazioni codificate di un certo
fenomeno), come (e dove) dovremo rappresentare i singoli elementi previsti e ammessi dalla
classificazione? La soluzione che abbiamo adottato nella nostra ontologia stata quella di
considerare una differenza tra una propriet che rappresenta una classificazione e lo schema
di classificazione vero e proprio, quello che contiene i singoli elementi della classificazione, il
quale viene considerato per non come una PROPERTY, ma piuttosto come un KIND, cio
comeunparticolaretipodioggetti.

ObjectTypeeClassificationScheme

Abbiamo cos introdotto anche due sottocategorie di KINDs, che possiamo descrivere
attraversoiseguentidueesempi:

NelcasodeiKINDscherappresentanoconcettiicuiTOKENsonoentitindividualienumerabili
(in modo teoricamente illimitato), ciascuna con la propria identit individuale, conveniamo di
parlare di OBJECT TYPE, ovvero di tipi di oggetto nel senso pi pieno. E questo il caso del
concetto di Persona che stiamo esaminando, cos come anche il caso di un numero
elevatissimodialtriconcetti,continuantiedoccorrenti,qualiiluoghi,tuttiitipidieventoedi
processo,leorganizzazioni,glioggettimateriali,etc.
QuandoinvecevogliamorappresentareglischemidiclassificazionecomeKIND,utilizziamo
unaparticolaresottocategoriadiKIND,ilCLASSIFICATIONSCHEME.Inquestocaso,iTOKENdi
un certo KIND non sono individui qualunque, come nel caso delle Persone, ma sono proprio
tutti e soli gli elementi dello schema di classificazione rappresentato da quel KIND
(nellesempio precedente, tutti i TOKEN che sono elementi della classificazione KIND Stato
Civile).
interessante aggiungere due brevi considerazioni sugli schemi di classificazione, prima di
passareoltre:

Humana.MenteIssue10July2009

XXXII

1) OgniTOKENdiunoschemadiclassificazionepossiedetipicamentedueSIMPLE
PROPERTY che lo descrivono: un codice identificativo e una descrizione, cos
comeaccadesemprenelcasodeglischemidiclassificazione;
2) Molti schemi di classificazione sono caratterizzati da una tassonomia, quindi
con una forma di rappresentazione della conoscenza ad albero rovesciato, o
ancheagrafoorientato.Inquesticasi,cheestendonoecomplicanoilsemplice
schema classificatorio a dizionario, possibile utilizzare delle RELATION
PROPERTYperconnetteretraloroinvarimodiisingolielementideglischemidi
classificazione (tipicamente attraverso relazioni ISA, cio relazioni di
generalizzazione tra concetti, che formano, appunto, gerarchie di
generalizzazione).

Senzaapprofondireulteriormentequestiargomenti,possiamocomunqueevidenziarecome
lintroduzionediquestinuoviconcetti(gliOBJECTTYPEediCLASSIFICATIONSCHEME,assieme
alleproprieteallerelazioninecessarieallalorocorrettarappresentazione)sianostatiaggiunti
allinternodelquadratoontologico,senzaulterioricategoriedibase.
Torniamo adesso ad esaminare le nostre PROPERTY, in quanto manca ancora un
tassellodasistemarepercompletarelimmaginedellontologiachestiamopresentando.

MeasurementPropertyeMeasurementUnitType

PROPERTY MODE
Altezza 178cm

In effetti era rimasta ancora una propriet da analizzare, tra quelle che avevamo indicato
nelnostroesempiodicaratterizzazionedelKINDPersona:lAltezza.
LAltezza,possiamoaffermareconsicurezza,unaqualitdeglioggettichevieneespressa
come la misura di un fenomeno, ovvero come una dimensione quantitativa che si differenzia
dallealtreproprietvistefinora,cheeranoinvecequalitative.Ora,lamisuradiundeterminato
fenomenounaproprietcompostasempredadueelementi:
1) Unvalore,tipicamentenumerico,cheesprimelaquantificazionedellamisurazione;
2) Ununitdimisura,ovverounospecificoelementodiunoschemadiclassificazione
progettatoproprioperstandardizzarelamisurazionediquantitfisiche.
Se guardiamo alla propriet che abbiamo evidenziato nel nostro esempio, possiamo appunto
vedereattiveentrambelecaratteristichedellamisura:ilvalore(178)elunitdimisura(cm).
Nel definire il nostro modello ontologico, abbiamo ritenuto essenziale poter esprimere il
concettodimisurazionecomeconcettooriginario,introducendounulterioretipodipropriet:
laMEASUREMENTPROPERTY.Attraversoquestonuovoconcettopossibilequindiidentificare
e definire qualunque misurazione, (come laltezza, nel nostro esempio) considerandola come
proprietcaratterizzantediunoopiKIND.
Per completare la definizione delle propriet che rappresentano delle misure, per
necessario aggiungere anche lindicazione di una unit di misura. A questo scopo, dato che le
unit di misura sono elementi di particolari schemi di classificazione, abbiamo introdotto un
ulteriore tipo di KIND, il MEASUREMENT UNIT TYPE (come ad esempio lunghezza, massa,
AnnuncioTwistOntology

XXXIII

tempo, forza, energia, pressione, etc.), i cui TOKEN sono proprio gli elementi dello schema di
classificazione(cm,kg,mg,sec,V,dyn,ecosvia):

La valorizzazione di un MODE relativo a una MEASUREMENT PROPERTY prevede quindi sia


lindicazione di un valore, che lindicazione del TOKEN che rappresenta lunit di misura
associata a quel valore, oltre naturalmente allindicazione del TOKEN che rappresenta il
soggettodellamisurazione:

CONCLUSIONE

Abbiamo descritto a questo punto il quadrato ontologico, assieme alle sottocategorie che
abbiamo ritenuto necessario introdurre per rappresentare un modello compiuto di ontologia
applicata, ovvero un modello di descrizione della realt che possa essere compiutamente
implementato in un vero sistema software di rappresentazione e manipolazione automatica
dellaconoscenza.
Come spunto per una futura evoluzione del nostro lavoro, introduciamo per concludere il
concetto di ENTITY, che pu essere considerato come un genere superiore rispetto alle
categorieKINDePROPERTY.Secorrettamenteinterpretatoegestito,ilconcettodiENTITY,che
modellalideadientitingenerale,fornirglistrumentiperaffinareulteriormenteilquadrato
ontologico,consentendoadesempiodirappresentaregerarchiedigeneralizzazioneanchetra
PROPERTY,oltrechefraKIND.

Humana.MenteIssue10July2009

XXXIV


XXXV

Obituaries


La redazione si unisce al cordoglio del mondo accademico per la scomparsa di Aldo Giorgio
Gargani, filosofo e critico illuminante. Nato a Genova, laureatosi in filosofia presso la Scuola
NormaleSuperiorediPisaespecializzatosipressoilQueen'sCollegediOxford,Garganistato
professoreordinariodiEsteticapressol'UniversitdiPisa,GastProfessorpressol'Universitdi
Vienna, membro del Wissenschaftskolleg di Berlino, della austriaca sterreichische Ludwig
Wittgenstein Gesellschaft, del College International de Philosophie. Non potendo in questa
sede nemmeno accennare al suo percorso intellettuale, data la vastit delle sue ricerche ed i
numerosissimiinteressi,desideriamotuttaviaricordarnequell'intensaattivitd'insegnamento
eseminarialeinEuropaenegliStatiUniti,chelohannoresoentrambeindiscussoprotagonista
dellaculturafilosoficadell'ultimoquartodelsecolotrascorso.

LaredazioneesprimesentitocordoglioallanotiziadellascomparsadiRalphDahrendorf(1929
2009), filosofo, sociologo e politico liberale. Nato ad Amburgo, Darhendorf conobbe
personalmentelaprigioneelesperienzadelcampodiconcentramentoperlasuaopposizione
al regime nazista. Dopo aver conseguito un dottorato in filosofia presso lUniversit di
Amburgo nel 1952, Dahrendorf ottenne un secondo dottorato, questa volta in sociologia,
pressolaLondonSchoolofEconomics,nel1956.QuivenneincontattoconleideediSociet
Aperta proposte da Karl Popper, di cui fu studente, e che, a detta dello stesso Dahrendorf,
davano una risposta alle grandi domande della moderna societ industriale poste da Marx.
Dopo aver insegnato a Saarbrcken, Tubinga e Costanza nella Germania dellOvest e Stanford
in California, Dahrendorf decise di entrare in politica. Fu eletto al Parlamento Europeo nel
1970. Nel 1974, fu nominato Direttore della London School of Economics, carica che ricopr
finoal1984.Dal1987al1997,fuinfineWardendelSt.AnthonyCollegepressol'Universitdi
Oxford. Da sempre interessato al problema delle divisioni di classe nella societ moderna,
Dahrendorffustrenuodifensoredelpluralismoliberale,unsistemasocialenelqualevengono
riconosciutietutelatidalleistituzioniinteressieaspirazioniconflittuali.

LaRedazione

Humana.MenteIssue9April2009

XXXVI

Esperimentiparadigmatici:ilgiocodellUltimatum
+

FrancescoGuala
*

francesco.guala@unimi.it

ABSTRACT

The Ultimatum Game is one of the most successful experimental designs in the history of the social
sciences.InthispaperItrytoexplainthissuccesswhatmakesitaparadigmaticexperimentstressing
inparticularitsversatility.Despitetheintentionsofitsinventors,theUltimatumGamewasneveragood
design to test economic theory, and is now mostly used as a heuristic tool for the observation of non
standardpreferencesorasasocialthermometerfortheobservationofculturespecificnorms.

1.INTRODUZIONE

Immaginadiavereappenaricevuto10euro.Questasommadovresseredivisaconunpartner
anonimo e invisibile, e dovrai accordarti con lui (o lei) su come dividerla. Lo spazio per la
discussionequasiinesistente:potraisoltantooffrireunadivisionedellatorta,eiltuopartner
potrsoloaccettareorifiutare.Selarifiuta,perdereteentrambilopportunitdidividervii10
euro;selaccetta,viportereteacasalavostraporzione,comeindicatodalladivisionechehai
proposto.
Questa sostanzialmente la situazione strategica nota come gioco dellUltimatum (GU).
Probabilmente avrai gi incontrato il GU. Se, come Robert Axelrod ha notato, il Dilemma del
Prigioniero diventato lE. coli delle scienze sociali, allora il GU ne senzaltro la Drosophila
melanogaster. La battuta di Axelrod va presa sul serio. Suggerisce che la teoria dei giochi
sperimentale ha raggiunto un livello di maturit paragonabile a quello di scienze sperimentali
affermate quali la biologia molecolare. Suggerisce anche che questa maturit si manifesta in
particolare nellemergere e nel consolidamento di prototipi sperimentali o esperimenti
paradigmatici analoghi agli organismi modello della biologia sperimentale. Gli esperimenti
paradigmatici hanno varie caratteristiche epistemiche distintive. In primo luogo, anni di
sperimentazione fanno s cheli conosciamo molto bene, sicuramente meglio di qualsiasi altro
sistema nel dominio di applicazione della scienza. In secondo luogo, essi vengono usati per
formulare una variet di inferenze ad altri sistemi che sono meno accessibili allo studio
sperimentale. E infine, sono solitamente strumenti versatili che possono essere utilizzati per
diversi scopi in diversi momenti e in diversi contesti di investigazione scientifica. In questo
articolomiconcentrerprincipalmentesuquestacaratteristica,maanchelealtreavrannoun
ruolonellastoriadelGU.


+
Una versione in inglese di questo articolo apparsa sulla rivista Philosophy of Science 75 (2008) pp.
658669. Marcel Boumans, Margaret Schabas, Bob Sugden, e diversi partecipanti alla conferenza della
Philosophy of Science Association a Vancouver nel Novembre 2006 hanno fornito diversi commenti e
suggerimenti.Tuttiglierroririmanentisonoamiocarico.
*
DipartimentodiEconomiaUniversitdegliStudidiMilano

Humana.MenteIssue10July2009

2

2.CONTROLLARELETEORIE

Il GU fu concepito e utilizzato per la prima volta da un gruppo di economisti sperimentali


tedeschi guidati da Werner Gth nei primi anni 80 del Novecento (Gth, Schmittberger and
Schwarze1982).Gthecolleghiintendevanostudiarelacontrattazionesequenziale,escelsero
il GU principalmente perch il problema di contrattazione sequenziale pi semplice che si
possaimmaginare.
1
Lasempliciteravoltaaminimizzareicosticognitividicalcolo.risaputo
che gli individui faticano ad analizzare i giochi dinamici complessi, in particolare quando
devonooperarelacosiddettainduzioneallindietro(backwardinduction).NelGU,nonsipu
noncapirecheilgiocofiniscedopolamossadelsecondogiocatore,percuiilrumoreneidati
dovuto alla mancata comprensione della decisione sperimentale dovrebbe essere ridotto al
minimo.
Il team di Gth intendeva anche controllare alcuni risultati di esperimenti precedenti.
Fouraker e Siegel, due psicologi, avevano realizzato negli anni 50 una serie di studi della
contrattazione ormai divenuti classici. Gth e i suoi colleghi trovavano sorprendente che le
offerte inique venissero rifiutate raramente negli esperimenti di Fouraker e Siegel.
Ipotizzaronochequestopotesseesseredovutoadalcunidettaglidellorodisegno,edeciserodi
controllarlimodificandoleggermentelasituazionesperimentale.
Ineffetti,leofferteiniquesonorifiutatenelGU.Nelloroprimostudio,Gth,Schmittberger
eSchwartzscoprironocheinmediailprimogiocatoreoffreil35%dellatorta.Osservaronouna
risposta modale al 50% e pochi rifiuti da parte del secondo giocatore (circa il 10%). Quando il
gioco venne ripetuto una settimana pi tardi (condizione con giocatori esperti) lofferta
mediasceseal31%.Loffertamodaleal50%scomparve,conlamaggiorpartedelleofferteora
fra il 20 e il 30%. Cerano anche pi rifiuti con la ripetizione (circa il 30%). Da allora, questi
risultati sono stati replicati numerosissime volte, e costituiscono la cosiddetta anomalia del
GU.Maanomaliarispettoachecosa?

3.CONTROLLARELATEORIAEMANIPOLARELEPREFERENZE

Secondolateoriadeigiochiclassica,ilprimogiocatoredovrebbeoffrirequasinullaeilsecondo
dovrebbe accettare. Lidea che il secondo giocatore dovrebbe affrontare una decisione
apparentemente banale: non portarsi a casa niente oppure ci che stato offerto dal primo
giocatore.Supponiamocheilminimochesipuoffriresia1euro.Uneuromegliodiniente,
quindiilsecondogiocatoredovrebbeaccettare.Assumendolaconoscenzacomunedelgiocoe
dellarazionalit,ladivisione1/9lunicoequilibriodiNashdelGU.
Questa previsione standard per la verit derivata per mezzo di un meccanismo
abbastanza complesso: da una teoria del gioco strategico (il nucleo della teoria dei giochi)
pi un insieme di assunzioni riguardo alle preferenze e alle credenze dei giocatori. Secondo
lassunzionedellegoismo,igiocatoripreferisconoaverepidenarochemeno,enonbadanoai
guadagnialtrui.IlGUdunqueunanomaliarispettoaunaprevisionederivatadaunmodello
composito (il modello della razionalit egoista) e solleva il noto problema di DuhemQuine.
La teoria del gioco strategico (lassunzione di razionalit) responsabile dellanomalia,
lassunzione dellegoismo inadeguata a catturare il comportamento nel GU, le credenze

1
La contrattazione solitamente rappresentata nella teoria dei giochi come un problema di
suddivisione, dove il surplus generato dallo scambio fra due beni deve essere allocato fra le parti. I 10
euronelGUrappresentanoinqualchemodoquestosurplus.

FrancescoGualaEsperimentiparadigmatici

3

individuali divergono dalle assunzioni classiche, oppure qualche altro elemento del modello
noningradodispiegarecichesuccedeinunGUsperimentale?
La teoria della razionalit che costituisce il nucleo della teoria dei giochi una teoria del
tipo se allora: dice che se le loro preferenze e credenze sono cosecos, allora gli
individui si comporteranno cosecos. Se le preferenze non sono egoistiche come postulato
dal modello classico, quindi la teoria della razionalit non pu essere controllata in un gioco
singolo come il GU. La teoria dei giochi sperimentale kosher dovrebbe dunque cominciare
con una misurazione delle preferenze in circostanze nonstrategiche, e proseguire
controllando se la teoria del gioco strategico formula previsioni corrette utilizzando le
misurazionipreliminaricomedatiinizialinellaproceduradicontrollo.
Tuttavia la maggior parte della teoria dei giochi sperimentale non kosher da questo
punto di vista. La maggior parte degli esperimenti comincia postulando il contenuto delle
preferenzeindividuali,invecediindividuarleempiricamente.IlGU,comeabbiamovisto,nonfa
eccezione. Ma se il GU non controlla davvero la teoria del gioco strategico (il nucleo della
teoriadeigiochi)perchdiventatocosfamosoeampiamentereplicato?
Come il Dilemma del Prigioniero, il GU cos semplice che nessuno dubita davvero che i
soggetti sperimentali siano razionali, nel senso minimale che le loro azioni seguono dalle loro
preferenze e credenze. Piuttosto, il GU interessante perch rende osservabili le preferenze
nonstandard. Intuitivamente sembra plausibile che il secondo giocatore possa preferire di
rinunciare a un po di soldi per punire le offerte inique, e che il primo preveda che unofferta
iniqua possa urtare i sentimenti del secondo giocatore. I secondi giocatori hanno preferenze
rivolte agli altri, influenzate forse da una preoccupazione per lequit. Infatti, se
consideriamo gli sviluppi teorici indotti da anomalie come il GU, notiamo che limpegno
maggioreconfluitoneltentativodimodellizzarefunzionidiutilitnonstandard,nellequalile
preferenze individuali sono eterorivolte piuttosto che strettamente egoistiche come nel
modelloclassico.
2
Il GU ha il merito di rendere particolarmente evidenti le motivazioni rivolte agli altri che
influenzano il comportamento. Se queste motivazioni possano essere incorporate nel quadro
della teoria dei giochi oppure no una domanda diversa, che non affronter in questo
articolo.
3
Ma rispondere a questa domanda non era essenziale per il successo del GU, che fu
ottenutocomunqueinmodoindipendente.

4.ISTITUZIONI

Le prime varianti sul tema del GU erano mirate a controllare la robustezza dellanomalia, e
spesso erano esplicitamente disegnate per farla scomparire. Col senno di poi, possiamo dire
che questi tentativi non hanno avuto successo. Lo status del GU non stato scalfito, come
documentano tutte le rassegne pubblicate fin dai tardi anni 80. Nei primi anni 90 tuttavia la
storiadelGUprendeunanuovapiega,conlapubblicazionediunimportantearticolodaparte
diungruppointernazionalediteoricideigiochiguidatodaAlvinRoth.

2
I contributi pi noti sono Rabin (1993) Fehr e Schmidt (2000) Bolton e Ockenfels (2000). Levidenza
sperimentale indica in modo chiaro che i modelli puramente consequenzialisti come quello di Fehr
Schmidt e quello di BoltonOckenfels sono inadeguati, mentre modelli pi complessi come quello di
Rabin funzionano meglio, pagando per un costo di indeterminazione predittiva in un vasto insieme di
giochi.
3
VediGuala(2006).

Humana.MenteIssue10July2009

4

Roth, Prasnikar, OkunaFujiwara e Zamir (1991) intendevano controllare due ipotesi: (1) la
concorrenza importante?, e (2) ci sono differenze rilevanti nella contrattazione in culture
diverse? Per controllare la prima ipotesi essi disegnarono, accanto al GU classico, un mercato
nel quale numerosi compratori competono per acquisire un bene posseduto da un singolo
venditore. Sia il GU che il gioco di mercato hanno equilibri estremamente asimmetrici (nei
qualiungiocatoresiaggiudicaquasituttalatorta).Perlequilibrioegoistasiottienesoltanto
nelle condizioni di mercato. La concorrenza ha leffetto di spazzare via le considerazioni di
equit.
Lesperimento di Roth uno degli articoli pi citati nelleconomia sperimentale. Incorpora
in un solo articolo quelli che sono generalmente considerati i tre pi importanti risultati
delleconomiasperimentale:lascopertache(1)ilmodellodellhomooeconomicusegoistanon
riesce a spiegare un gran numero di osservazioni; che (2) il modello della razionalit egoista
pu nondimeno rendere conto del comportamento in un ampio raggio di situazioni
sperimentali;einfine,combinandoquestiduerisultati,che(3)leistituzionisonoimportanti.
Valelapenadistinguerefraduetipidiistituzioni:normeinformalicomequelledigiustizia,
uguaglianza, cooperazione, e reciprocit che governano il nostro comportamento in molte
circostanze (come il GU); e regole esplicite di scambio, aggregazione, trasmissione
dellinformazione, ecc. che sono tipiche di istituzioni di mercato relativamente formalizzate.
Limportanza delle regole formali di scambio emerge nellarticolo di Roth dal confronto fra la
contrattazionenelGUelaconcorrenzadimercato.Limportanzadellenormeculturaliinvece
messa in risalto dalle repliche dellesperimento condotte in quattro diversi paesi (USA,
Giappone, Slovenia, e Israele). Roth e colleghi trovarono delle differenze di comportamento
significative: i giapponesi e gli israeliani tendono a offrire di meno (lofferta modale al 40%)
degliamericaniedeglisloveni,maleofferteiniqueaTokioeGerusalemmesonorigettatecon
lastessaprobabilitdelleofferte50/50diPittsburgheLubiana.Neltestareunipotesiteorica,
perlaprimavoltailGUvenivaancheusatocomeunostrumentodimisurazione.

5.MISURAZIONE

Nel 2000 lantropologo Joe Henrich ha riportato una serie di osservazioni sorprendenti
ottenute usando il GU fra i Machiguenga, un gruppo di indigeni peruviani. I Machiguenga
offrivano divisioni della torta pi inique, e rifiutavano le offerte inique meno spesso dei
soggettioccidentali.Inaltritermini:sicomportavanopispesso(anchesenonsempre)come
degli uomini economici. I risultati dei Machiguenga potevano apparire anomali soltanto su
uno sfondo di conoscenza sperimentale data per acquisita, cio dopo che i risultati del GU
nellesocietoccidentalieranodiventatiregolaritstandardizzateegeneralmenteaccettate.
Nei primi anni 80 il comportamento dei Machiguenga sarebbe stato considerato
probabilmentepocosorprendenteemoltomenoanomalo(inquantopivicinoallaprevisione
teorica)diquantononappaiaora.
Henrichconclusechepopolidiversihannodiverseaspettativeculturalienormedigiustizia,
echedobbiamostudiarecomequestenormevengonocreateesostenuteinciascuncontesto
sociale.Conquestaidea,ungruppodieconomistieantropologisipropostodiconfrontareil
comportamento in alcuni esperimenti classici (specialmente il GU e il gioco dei beni pubblici)
fra quindici microsociet in Sud America, Africa, e Asia. Questo progetto, finanziato dalla
MacArthur Foundation, il tentativo pi ambizioso ed esaltante di usare leconomia
sperimentale a scopi di misurazione che sia mai stato tentato fino a oggi (vedi Henrich et al.
2004).
FrancescoGualaEsperimentiparadigmatici

5

Il progetto MacArthur ha diverse caratteristiche metodologiche interessanti. La


sperimentazionesicombinaconunprofondostudioetnograficodelcontestosociale(strutture
politiche ed economiche, credenze religiose, riti, pratiche di divisione del cibo, ecc.). Queste
informazioni illuminano alcuni risultati sorprendenti (deviazioni dalle previsioni standard) in
un modo che inusuale per le indagini sperimentali classiche. Ma soprattutto, esse
sottolineano limportanza della standardizzazione per il consolidamento dei disegni
sperimentaliparadigmatici,chepossonoessereutilizzaticomestrumentiportatilipermisurare
fenomenisperimentaliinsituazionidiverse.

6.STANDARDIZZAZIONE

La standardizzazione il processo di consolidamento di un disegno sperimentale


paradigmatico. La standardizzazione determinata da diversi fattori, che possono avere un
pesodiversoneivaristadidiunprogrammadiricerca.
(1) Imitazione: come ha enfatizzato Thomas Kuhn, le tecniche di una disciplina scientifica
vengono spesso apprese replicando gli esemplari della generazione di scienziati
precedente.
(2) Test di robustezza: la logica stessa della variazione controllata richiede che i nuovi
disegnisianoconfrontaticonunabase.Questultimapuessereunaprevisioneteorica,
ma spesso semplicemente un risultato sperimentale classico che di conseguenza
continuaaesserereplicatoindefinitamente.
(3) Coesione disciplinare: molte discipline a un certo punto accolgono e impongono delle
buone pratiche di ricerca sperimentale. Pratiche come luso degli incentivi monetari,
lanonimit, la ripetizione, e il divieto di ingannare i soggetti fanno parte di questa
categorianelcasodelleconomiasperimentale.
4
Attraverso questi processi, la standardizzazione porta alla stabilizzazione dei fenomeni, e alla
relativa stabilizzazione di un disegno sperimentale.
5
Una volta che un disegno stato
standardizzato,alcunideisuoidettagliperdonoloriginariagiustificazionemetodologicaesono
conservati principalmente per ragioni pragmatiche: diventano dei mezzi per ottenere la
confrontabilitfradiversepopolazionidisoggetti.

7.VALIDITESTERNA

SidicespessocheaffrontiamounGUognivoltacheentriamoinungrandemagazzino.Civiene
offerto un prezzo, senza lopportunit di contrattare. Eppure in uneconomia di mercato non
percepiamo quasi mai i prezzi come delle offerte del tipo prendere o lasciare. Abbiamo
solitamente altre opzioni, come per esempio entrare in un negozio che vende beni simili e
controllare se ci offre dei prezzi migliori. Quando accettiamo un prezzo senza arrabbiarci,
perch crediamo (o assumiamo) che il prezzo risulti da un processo equo di concorrenza di
mercato, e che sia pi o meno il meglio che ci possa essere offerto data la produzione e altri

4
Alcune di queste pratiche sono formalizzate nei cosiddetti precetti delleconomia sperimentale
(Smith 1982) mentre altri sono giustificati in modo informale. Guala (2005, Cap. 11) include
unillustrazioneediscussionepidettagliata.
5
Sugden(2005)catturaentrambiquestiaspettiinmodoelegantepermezzodelconcettodiesemplare
sperimentale (experimental exhibit): un fenomeno accoppiato a un disegno standard, come il
modellino di una nave in una bottiglia. Boumans (2005) sottolinea correttamente laltro aspetto della
medaglia:talvoltalarobustezzadiunfenomenoportaallastandardizzazionedeldisegnoildisegnoche
megliocatturailfenomenonellasuamanifestazionepura.

Humana.MenteIssue10July2009

6

costi. (Ovviamente questo non sempre vero, e, infatti, le associazioni dei consumatori, gli
osservatori dei prezzi, e i boicottaggi dei consumatori hanno un ruolo importante nelle
economiedimercato.)
Considerazioni di equit emergono in modo diverso in diversi contesti, e il GU pu al
massimoesserevistocomeunesemplarerappresentativodiunampioraggiodisituazioninelle
qualilenormediequitgiocanounruolopiomenodirettonelcomportamentoeconomicoe
sociale.QuestoparticolarmenterilevanteperilprogettodellaMacArthurFoundation.Alcune
delle piccole societ avevano scarsa esperienza delle transazioni di mercato, alle volte erano
perfinopocoabituateallusodeldenaro.Quandoidisegnisperimentalifacevanoscattaredelle
normelocalierapipercasocheperintenzionedellosperimentatore.
6
La validit esterna il problema di generalizzare i risultati sperimentali dalle condizioni di
laboratorio ad altre situazioni che ci interessano. solitamente considerato il tallone di
Achille della sperimentazione nelle scienze sociali, anche se la sua rilevanza per gli
esperimenti nelle scienze naturali una questione per lo pi inesplorata e forse
sottovalutata. Ho sostenuto altrove (Guala 2005, Cap. 711) che la questione della validit
esterna deve essere risolta caso per caso. Non ha senso dubitare della validit esterna di un
esperimentoingenerale.Bisognaprimachiedersiqualesialobiettivodellesperimentoquale
sistema nonsperimentale si intendeva studiare in origine. Spesso gli esperimenti sono fatti
senzaunobiettivoesternoinmente,edunquelavaliditesternanonilloroprincipalescopo
opreoccupazione.
Intuitivamente la standardizzazione sembra essere nemica della validit esterna. Se il
mondoeconomicoesocialevariegatoecomplesso,abbiamobisognodiflessibilitinveceche
di rigidit. Dobbiamo ritagliare i nostri disegni sperimentali intorno alle specifiche domande
poste dal problema che stiamo studiando, piuttosto che il contrario. Ma allora perch la
consolidazione di un formato rigido per il GU stata utile e ne spiega il successo? Perch,
come vedremo, i criteri metodologici che si applicano alla misurazione sono diversi da quelli
chesiapplicanoalcontrollodelleteorie.

8.OSSERVAREEINTERVENIRE

I filosofi neosperimentalisti hanno identificato lintervento materiale come la differentia


specifica fra la scienza sperimentale e nonsperimentale. Secondo una versione di questo
approccio, sviluppata da Jim Woodward (2003) lintervento funzionale alla scoperta di
relazionicausalirobuste.Unintervento,secondoWoodward,unamanipolazionechirurgica
di una variabile che lascia il resto del sistema sperimentale intatto. Consideriamo la Figura 1.
SupponiamodivoleresapereseXcausaY,YcausaX,oppureseleduevariabilisonocorrelate
inmodospurio.Inuninterventochirurgico(I)Xmanipolatainmodotaledatagliaretuttii
collegamenticonisuoiantecedenticausali,manessunodeisuoieffettiputatividirettamente
influenzatodallamanipolazione.QualsiasimutamentodiYdovutoaIdevepassareattraverso
X,inaltreparole.Selinterventohasuccesso,eXcausaY,alloralavariazioneinXsirifletterin
una variazione in Y. Se al contrario Y causa X (il caso di mezzo nella Figura 1) oppure le due
variabili sono associate soltanto in modo spurio (caso a destra) la correlazione fra X e Y sar
interrotta.


6
Perunesempioevidente,cfr.Ensminger(2004)sullistituzionedellharambeefrailpopoloOrmadel
Kenia.

FrancescoGualaEsperimentiparadigmatici

7

Figura1

Usando il GU come uno strumento di misurazione, non confrontiamo due (o pi) stati di una
variabile che possiamo fissare a piacimento. Sicuramente stiamo confrontando qualcosa il
numerodirifiutidiofferteiniquenellesocietAeB,peresempiomasiassumecheilvalore
della variabile sia determinato da qualche fattore nascosto che non possiamo controllare
direttamente. In effetti estremamente importante che le procedure sperimentali non
interferiscanoconilfunzionamentodiquestifattorinascosti,perchaltrimentilamisurazione
sarebbeconsiderataartificialeeinutile.
Quando il GU usato come strumento di misurazione, ovviamente, in un certo senso
stiamo controllando qualcosa. Stiamo controllando lipotesi che due insiemi di osservazioni
(raccolte negli Stati Uniti e fra i Machiguenga, per esempio) provengano dalla stessa
popolazioneoprocessodigenerazionedeidati.propriopercontrollarequestaproposizione
che dobbiamo mantenere il disegno sperimentale il pi stabile (o standardizzato) possibile.
Altrimenti, non sar possibile attribuire differenze nei comportamenti osservati a delle
variazioniinqualchecosadidiverso(qualunquecosaessosia)daldisegnosperimentalestesso.
Unanalogia con la termometria pu essere utile: immaginiamo di volere dimostrare che
lacquabolleatemperaturediverseallivellodelmareesuunaltipiano.Abbiamobisognodiun
termometro standardizzato e portatile, che sia robusto a variazioni nellambiente esterno
quandotrasportatodaunsitodimisurazioneaunaltro.Poichnonvogliamointerferirecon
la variabile che stiamo misurando (la temperatura) saranno permesse soltanto manipolazioni
chesonofunzionaliallottenimentodiunamisuraoggettivaestandardizzatachepossaessere
usataperlaprevisioneeilconfronto.Storicamente,latemperaturastatamisuratapermolto
tempoinmodoindipendentedaunacomprensioneanchelontanamentecorrettadellanatura
delcalore(Chang2004).Ilprimopassonellosviluppodellatermometrialacostruzionediun
termoscopio.Untermoscopiomisuralatemperaturasuunascalaordinale,epuessereusato
per determinare uno o pi punti fissi per esempio il punto di congelamento e di ebollizione
dellacqua. Una volta fissati questi punti, possiamo costruire un termometro che misuri la
temperatura su una scala cardinale. Storicamente, un buon termoscopio era valutato dalla
sua conformit con le sensazioni ordinarie del caldo e del freddo: si deve espandere quando
sentiamocaldoecontrarrequandosentiamofreddo.Questogarantiscechestiamomisurando
qualche cosa di importante (per noi) ma ovviamente non pu garantire che abbiamo
incontratounarelazionefisicarobustaesignificativa.
Lo stesso vale per gli esperimenti economici quando sono usati come strumenti di
misurazione.LecausedellevariazionicomportamentalinelGUsonoancoracontroverse,ed
del tutto possibile che diverse cause siano operanti in diversi contesti sociali. Nonostante
questo, il GU pu essere usato a scopi di misurazione, se si ottenuta una standardizzazione
appropriata e lesperimento stato fatto con cura. In questo caso possiamo concludere che
qualsiasidifferenzavengatrovata,essadeveesseredovutaadifferenzenellepopolazionionei

Humana.MenteIssue10July2009

8

processi di generazione dei dati. Linferenza, per usare lespressione introdotta da Deborah
Mayo(1996)giustificataperchstatacontrollataseveramente.
UncontrolloseveroproducedatidiuncertotipoD1selipotesisottocontrollovera,eun
altro tipo di dati D2 se lipotesi falsa. In termini statistici pi familiari, esso minimizza la
probabilit di commettere errori di tipo I e II. Il controllo severo, secondo Mayo, il tratto
distintivodellasperimentazione,perchibuoniesperimentitipicamentesupportanoinferenze
che sono state controllate severamente dallesperimento stesso.
7
La caratterizzazione di
Woodward degli esperimenti volti alla scoperta causale per lo pi coerente con questa
concezione. Torniamo alla Figura 1. Se lintervento ha successo, e X causa Y (caso a sinistra)
alloralevariazioniinXsarannoriflesseinvariazioniinY.SeinveceYcausaXoleduevariabili
sonoassociateinmodospurio(alcentroeadestra)lacorrelazionefraXeYscomparir.Inun
esperimento genuino e ben fatto, in altri termini, si osserver un certo tipo di dati se vero
cheXcausaY,eunaltrotipodidatisenonlo.
8
Qualsiasiconclusionecausalearriveremoa
formulare,essasarseveramenteinvestigataocontrollatasecondoilcriteriodiMayo.

9.DATISPERIMENTALIERACCOLTISULCAMPO

Undisegnostandardpermettealledifferenzefradiversesituazionidiemergere.Importapoco
quali siano queste differenze. Potrebbero essere dovute a fattori locali, non controllati
(abitudini, norme, disposizioni psicologiche); esse sarebbero nondimeno interessanti, perch
queste differenze sono proprio quello che stiamo cercando. Infatti una delle lezioni pi
importanti degli studi interculturali riguarda la difficolt di generalizzare su tutte le varie
popolazioni
Gli studi della MacArthur Foundation mostrano che un comportamento pi o meno
cooperativo correlato con diversi fattori in societ diverse (ricchezza, genere, et, strutture
politiche, esposizione ai mercati, ecc.). Ancora pi importante la dimostrazione che tali
relazioni non sarebbero mai state scoperte senza la conoscenza etnografica profonda fornita
dagliantropologisulcampo.NelsuostudiodegliHadza,ungruppodicacciatoriraccoglitoriin
Tanzania, Marlowe (2004) per esempio riporta un grado sorprendentemente basso di
condivisione e cooperazione, in una societ altrimenti profondamente egalitaria. La
spiegazione proposta si appella a una sottile distinzione fra ci che si vede (per esempio la
cacciagione di grossa taglia) e ci che pu essere facilmente nascosto (come il denaro). Le
norme degli Hadza impongono la condivisione di ci che si vede, ma non proibiscono
effettivamente il possesso e il consumo individuale di beni pi piccoli. Lavarizia tollerata
quando non pu essere sanzionata in modo efficace, mentre il controllo sociale si applica alla
cacciagrossa.
Patton (2004) riporta un livello di comportamento cooperativo pi alto fra gli Achuar che
fraiQuichua,duegruppichevivonomoltovicininellaforestaamazzonicaecuadoriana.Queste
relazioni si osservano sia nei dati sperimentali sia nelle pratiche di condivisione della carne
osservate nelle due comunit. Tuttavia, le differenze allinterno dei gruppi mostrano che la
maggioregenerositdegliAchuarquasiinteramentedovutaaunapiccolaelitedicapichela
usanoperscopipolitici,inparticolarepercrearecolazionifortiestabilifrafamiglieevillaggi.I
Quichua, che hanno una struttura politica meno stabile e fra i quali le coalizioni tendono a
essere pi effimere, semplicemente non hanno prodotto una simile classe di condivisori
politici. La conoscenza della struttura di una societ e delle sue norme cruciale, perch le

7
O,piprecisamente,inferenzeaipotesichesonostatecontrollateseveramente.
8
VediancheWoodward(2000).
FrancescoGualaEsperimentiparadigmatici

9

correlazionimacrosonospessoingannevolioinutili.Ognicasodiverso;ognispiegazione
locale.
9

10.TERMOMETRISOCIALI

Un termometro linizio di una diagnosi, non la fine della storia. Misura un sintomo (un
sintomo importante e ubiquo) non una causa. Questo non lo rende meno utile al contrario,
staremmomoltopeggiosenzaitermometri.Comeunbuontermometro,ilGUdiventatouno
strumentoportatilechepossiamotrasportarediculturainculturapermisurareuninsiemedi
fenomeni che riteniamo siano associati alle norme sociali. Quali siano questi fenomeni
ancora controverso. A differenza della temperatura, dove il termometro pu essere
confrontatoconlesensazionipuredelcaloreedelfreddo,nonabbiamodellebasisimiliperil
GU.SentiamocheilGUpuessereusatopermisurarecicheapparegiustooingiusto,mail
nostro concetto di giustizia non necessariamente una buona guida per comprendere quelli
che prevalgono nelle societ che stiamo studiando. La nostra concezione dellequit pu
funzionare come punto di paragone, tuttavia, e abbiamo proprio bisogno di qualcosa del
genere per la misurazione. Il GU emerso attraverso un processo di selezione sociale nella
teoria dei giochi sperimentale, come un disegno robusto che cattura qualche cosa di molto
importante per noi. Non abbiamo nessuna garanzia che sia lo strumento di misurazione
miglioreperstudiarecicheimportanteperchiviveinsocietdiversedallanostra.Maun
punto di partenza, e la scienza sociale ha molto bisogno di una qualche piattaforma stabile
perchpossadecollarelaricerca.

BIBLIOGRAFIA

Bolton G. e Ockenfels A.(2000) ERC: A Theory of Equity, Reciprocity and Cooperation,


AmericanEconomicReview90:16693.
BoumansM.(2005)HowEconomistsModeltheWorldintoNumbers.Londra:Routledge.
Chang, H. (2004) Inventing Temperature: Measurement and Scientific Progress. Oxford:
OxfordUniversityPress.
Ensminger J. (2004) Market Integration and Fairness: Evidence from Ultimatum, Dictator,
andPublicGoodsExperimentsinEastAfrica,inHenrichetal.(2004)35681.
FehrE.eSchmidt,K.(1999)ATheoryofFairness,CompetitionandCooperation,Quarterly
JournalofEconomics114:81768.
GilWhiteF.(2004)UltimatumGamewithanEthnicityManipulation:ResultsfromKhodovin
BulganSum,Mongolia,inHenrichetal.(2004)260304.
Guala F.(2005) The Methodology of Experimental Economics. New York: Cambridge
UniversityPress.
GualaF.(2006)HasGameTheoryBeenRefuted?,JournalofPhilosophy103:239263.
Gurven M.(2004) Does Market Exposure Affect Economic Behavior? The Ultimatum Game
and the Public Goods Game among the Tsimane of Bolivia, in Henrich et al. (eds.
2004)194231.

9
SivedanoancheGurven(2004)Tracer(2004)eGilWhite(2004)peraltriesempidelgenere.

Humana.MenteIssue10July2009

10

Gth W., Schmittberger R. e Schwarz, B. (1982) An Experimental Analysis of Ultimatum


Bargaining,JournalofEconomicBehaviorandOrganization3:367388.
Henrich J. (2000) Does Culture Matter in Economic Behavior? Ultimatum Game Bargaining
among the Machiguenga of the Peruvian Amazon, American Economic Review 90:
97379.
Henrich J., Boyd R., Bowles S., Camerer C., Fehr, E. e Gintis H. (eds. 2004) Foundations of
Human Sociality: Economic Experiments and Ethnographic Evidence from Fifteen
SmallScaleSocieties.Oxford:OxfordUniversityPress.
MarloweF.(2004)DictatorsandUltimatumsinanEgalitarianSocietyofHunterGatherers:
TheHadzaofTanzania,inHenrichetal.(2004)168193.
Mayo D. (1996) Error and the Growth of Experimental Knowledge. Chicago: University of
ChicagoPress.
Patton J. (2004) Coalitional Effects on Reciprocal Fairness in the Ultimatum Game: A Case
fromtheEcuadorianAmazon,inHenrichetal.(2004)96124.
Rabin M. (1993) Incorporating Fairness into Game Theory and Economics, American
EconomicReview83:1281302.
RothA.,PrasnikarV.,OkunoFujiwaraM.eZamir,S.(1991)BargainingandMarketBehavior
in Jerusalem, Lubljana, Pittsburgh and Tokyo: An Experimental Study, American
EconomicReview81:10681095.
Smith V. (1982) Microeconomic Systems as an Experimental Science, American Economic
Review72:92355.
Tracer D. (2004) Market Integration, Reciprocity, and Fairness in Rural Papua New Guinea:
ResultsfromaTwoVillageUltimatumGameExperiment,inHenrichetal.(2004)232
259.
Sugden R. (2005) Experiments as Exhibits and Experiments as Tests, Journal of Economic
Methodology12:291302.
WoodwardJ.(2000)Data,Phenomena,andReliability,PhilosophyofScience67:S163179.
WoodwardJ.(2003)MakingThingsHappen.NewYork:OxfordUniversityPress.



11

ScientificUnificationinEconomics.
ThecaseoftheNewEconomicGeography

CaterinaMarchionni
*

caterina.marchionni@helsinki.fi

ABSTRACT

Scientificunificationtheapplicationofthesameprinciplesandtoolstothestudyofphenomenafrom
different domains is a powerful ideal in economics that significantly shapes its internal dynamics and
howitrelatestoneighbouringfields.InthispaperIdiscussaspectsofscientificunificationinthecaseof
theNewEconomicGeography,anapproachtospatialissuesrecentlydevelopedwithineconomics.The
analysisshowsthatthedrivetounifyaffectsarangeofintertheoretic,interfieldandinterdisciplinary
relations,andthatitmeetswithdisunitiesofvarioussorts,whichaffectthekindofproductsunification
ultimatelydelivers.

1.INTRODUCTION

Scientific unification the application of the same principles, explanatory schemata, models
and tools to the study of phenomena in different domains has been and still is a powerful
scientific ideal. Celebrated examples include Newtons theory of universal gravitation, which
unifiedcelestialandterrestrialphenomena,Einsteinstheoryofspecialrelativity,whichunified
electric and magnetic phenomena, and Darwins theory of natural selection, which unified
phenomena in fields such as comparative anatomy, biogeography, and palaeontology. In
economicstooscientificunificationisapowerfulforcethatsignificantlyaffectsitsowninternal
dynamics as well as its relationship to neighbouring fields. Consider for example the
widespread adoption of economicsstyle models and explanations in neighbouring social
sciences like sociology and political science. The search for unity partly lies behind the
emergence of new fields such as neuroeconomics, which integrates economics, psychology
and neuroscience to study how the brain makes economic choices; network theory, which
studies the properties of networks throughout the social and the natural sciences, and in its
economics application it is used to explain phenomena as diverse as scientific collaborations,
variations in crime, and differences in social cooperation; or the New Economic Geography,
whose unificationist achievements have been celebrated in the awarding of the 2008 Nobel
Prizeineconomicstoitsfoundingfather,PaulKrugman.

Thegeneralintuitionbehindtheidealofscientificunificationisthatallsciencesareafterall
concernedwithoneandthesameworld.Theunificationoftheentirescientificdomaineither
by means of a grand single theory, or of a small set of fundamental laws is probably
unattainable.Nevertheless,dealingwithcomplexinterrelationsbetweenphenomenaseemsto
require at least some degree of unity and integration of scientific knowledge. Take social
phenomenaforexample.Thesearecloselyinterconnectedthroughacomplexthicketofcausal
relations. In order to understand them, social scientists might often be forced to transcend
disciplinary boundaries. And at least prima facie, it seems plausible that biological theories

*
DepartmentofSocialandMoralPhilosophy,UniversityofHelsinki

Humana.MenteIssue10July2009

12

should also somehow bear on social scientific theories, for the individuals that constitute the
socialworldareafterallalsobiologicalsystems.
In the philosophy of science unification has traditionally been a hot topic of debate (see
Bechtel and Hamilton 2008 and Cat 2007 for overviews). Modern discussion began with the
classic paper by Oppenheim and Putnam (1958), which envisaged the unity of science in the
form of a reduction of all the sciences to physics. More recent accounts have suggested that
scientific understanding itself is a product of the unification of disparate phenomena, and
accordinglytakesciencetoprogressbyreducingthenumberoftypesoffactsthatwehaveto
accept as ultimate (Friedman 1974; Kitcher 1981). Today many philosophers of science have
dismissedtheseandsimilarproposalsasinadequatebothdescriptivelyandnormatively.
In spite of the dissatisfaction among philosophers however, scientists continue to pursue
unification in some form, and when successful these achievements are widely acclaimed. The
philosophical project of finding a onesizefitsall model has failed not because unification is
unimportant, but because it is a local, varied and contextdependent phenomenon. The real
challenge for philosophers of science now is to understand how and why scientists engage in
various kinds of unifying research strategies; strategies that might prove fruitful in some
contexts and not in others, and only for some purposes (e.g., explanation, confirmation, and
discovery).Somehavealreadybeguntoanalyzereductionistandunifyingstrategiesfromthis
perspective(e.g.,Grantham2004,Wimsatt2006,Wylie1999),butthusfarverylittleattention
hasbeenpaidtounificationineconomics.If,asitseemsincreasinglyclear,itassumesdifferent
featuresindifferentscientificcontexts,ineconomicsitislikelytobesomehowdifferentfrom
unificationinsay,biology.Wemightdiscoverthattheunifyingstrategiesdeployedwithinand
aroundeconomicsaremisguided.Itsmorelikely,however,thatwellfindthattheyworkonly
sometimes and only for some purposes, and we might be able to understand why they work
whentheydo.
My focus here is on the features of scientific unification in the case of the New Economic
Geography. What emerges is that unification drives theoretical development and shapes a
variety of intertheoretic, interfield and interdisciplinary relations. At the same time, it co
exists with disunities of various sorts that affect the kind of products that are ultimately
delivered.Similarideas,Ibelieve,couldbeappliedbeyondthecaseunderscrutiny.Attheend
of the paper I briefly speculate on how the pursuit of scientific unification and the sorts of
friction it faces play out in the case of neuroeconomics. The objective is to make a case
supportingtheclaimthatunificationisapowerfuldrivingforceineconomicsandthatittakes
placeoftenenoughtodeservephilosophicalattention.

2.VARIETIESOFSCIENTIFICUNIFICATION

Scientificunificationencompassesasetofactivitiesthatyieldtheunificationorintegrationof
scientific knowledge. I propose to distinguish four levels at which unification can take place,
whichcorrespondtoitemsthateitherenterintheprocessofunificationorareaffectedbyit:
phenomena, theories, fields and disciplines. Although differentiating these levels in practice
involves some degree of arbitrariness, I believe that each level brings to light important
features of scientific unification (though obviously not all four levels will be involved in every
case).Mycasestudyalsoshowsthat,ateachlevel,distinctivefrictionsinfluenceandconstrain
thekindanddegreeofunificationestablishedatthatlevel.
First, explanatory unification concerns phenomena. It is a matter of explaining several
classes of explanandum phenomena with the same, or fewer, explanantia. According to
Friedman(1974)andKitcher(1981),scientificexplanationamountstounification,andhenceit
CaterinaMarchionniScientificUnificationinEconomics

13

is a matter of deriving many explananda from a small set of premises, or argument patterns.
Onthisview,theexplanatorypowerofatheoryisafunctionofitsunifyingpowerandscience
progressesbyreducingthenumberoftypesoffactswehavetoacceptasultimate.Inorderto
appreciate the value of unified explanations however we dont need to endorse Friedmans
and Kitchers proposals in their entirety. Its indeed possible to view unifying power and
explanatorypowerasdistinctandstillappreciateexplanatoryunification.Forinstance,wecan
holdthatunifiedexplanationsaretobepursuedwhenandonlyinsofarastheyreflectadeep
ontological unity among apparently different kinds of phenomena (Mki 2001): when two
phenomena, say optical processes and electromagnetic processes, are identified as being of
the same kind. More modestly, unified explanations sometimes make salient commonalities
and general patterns, which would be concealed by disunified explanations that focus on the
detailsspecifictoparticularoccurrences(e.g.Sober1999).
1

Second,theoreticalunificationisamatterofunifyingpreviouslyseparatetheoriesbymeans
of one that possesses all (or most of) their explanatory content.
2
The paradigmatic cases of
unification in the history of science have often accomplished unification of both phenomena
and theories. For instance, Maxwells theory unified electromagnetism and optics in showing
thatelectromagneticwavesandlightwaveswereoneandthesamething.Similarly,Newtons
theory unified Galileos laws of terrestrial mechanics and Keplers laws of celestial bodies by
showingthatthemotionsofcelestialandterrestrialbodiescouldbeexplainedthroughthelaw
ofuniversalgravitation.Whatcountsasaclassofphenomenaisdefinedbyatheory,sothata
theory that successfully unifies two classes of phenomena (henceforth the unifying theory)
alsounifiestherespectivetheories(henceforththeunifiedtheories).
When it is a question of more mundane cases the distinction between the unification of
phenomenaandoftheoriesnonethelessstillprovesuseful.Thereasonisthataunifyingtheory
might account for only a subset of the phenomena that the unified ones, taken together,
explain. In many cases, a unifying theory includes a set of idealizations and abstractions that
differfromthatoftheunifiedtheories.Eachtheory,whetherunifyingorunified,maybecause
ofitsdistinctiveidealizationsandabstractionsbeusefulfordifferentexplanatorypurposes.In
such cases, explanatory unification may not produce a complete theoretical unification.
Theoreticalunificationturnsouttobeamatterofdegree.
Third, scientific unification can also occur at the level of fields. Lets call this interfield
integration. Focusing on the interconnections between fields rather than exclusively on those
between theories will produce a more general framework for examining scientific unification.
Fieldstypicallyincludeseveraltheoriesandmuchmorebesides:

A central problem, a domain consisting of items taken to be facts related to that problem,
general explanatory facts and goals providing expectations as to how the problem is to be
solved, techniques and methods, and sometimes, but not always, laws and theories which are
related to the problem and which attempt to realize the explanatory goals (Darden and Maull
1977:4)

1
Consideranexplanationofanincreaseincrimerate,whichappealstoahigherlevelofunemployment
and abstracts away from the changes of opportunities and motivations of the individuals involved. To
seewhythisisaunifiedexplanation,supposewehavetworegionsbothexperiencingariseinthecrime
rate as the consequence of an increase in the level of unemployment. The unemployment explanation
provides a unified explanation of the two occurrences, whereas the explanation that invokes the
changes of opportunities and motivations of the individuals involved would be different in the two
cases.
2
AccordingtoWhewell(1847),theoriesthatunifyphenomenaarealsomorelikelytobetrue(seealso
Thagard2007).

Humana.MenteIssue10July2009

14

The unification of fields occurs via a number of routes, of which the unification of theories is
just one. For instance, it may happen through the development of an interfield theory, in
other words a theory that identifies and explains the relations between fields (Darden and
Maull 1977), or through the development of a theory that explains the presuppositions of
anotherwithoutreplacingit(Kitcher1984).Fieldscouldalsobecomeprogressivelyintegrated
through the establishment of nontheoretical connections, which Grantham (2004) calls
practical unification. This takes place when one field comes to rely on the methods,
heuristics,ordataofaneighboringfield(Grantham2004:143),suchaswhenthetheoriesor
methodsdevelopedinonefieldgeneratenewhypothesesinanother,orwhendatafromone
fieldareusedtoconfirmhypothesesdevelopedinaneighbouringfield.Forthedata,methods
orheuristicsthataredistinctiveofonefieldtobearonaneighbouringfieldthereneedstobe
at least some theoretical understanding of how the two domains are related, but practical
unification can occur also with a relatively low degree of theoretical unification. Various
practices can therefore contribute to the unification of fields (and thereby of science),
althoughtheycanhardly,ifever,effectacompleteinterfieldunification,thatis,oneinwhich
thenewfieldsupplantspreviouslyexistingones.

Finally, the pursuit of unification may also affect interdisciplinary relationships: a theory
may enter a domain of phenomena that fall within the purview of a discipline distinct from
thatinwhichitwasoriginallydeveloped.Thisphenomenonisknownasscientificimperialism.
Talk of disciplines rather than fields allows us to emphasize the sociological and institutional
aspects that contribute to define disciplinary identities and boundaries, and which are absent
from the definition of a field. Although the term scientific imperialism has negative
connotations, following an idea advanced by Uskali Mki (2009), we can distinguish forms of
scientific imperialism, some of which can be positively evaluated. If we think of scientists as
seeking to increase a theorys degree of unification by way of applying it to new types of
phenomena, it is largely a matter of social and historical contingency whether these
phenomena are studied within other disciplines. If we endorse the idea that the scope of
theories should be determined by the real order of things and not by preexisting disciplinary
boundaries, then it follows that in principle there is nothing wrong with that kind of
imperialism that follows from the expansion of a theorys scope. But this is not all there is to
scientific imperialism. Scientific imperialism can also affect other aspects of the scientific
endeavoursuchasthestandingofonedisciplineand/orconceptionsaboutthescientificstatus
of one style of research visvis another. Whereas expanding the scope of a theory in search
of real unity among phenomena is potentially progressive, entering another discipline in the
attempttochangeitstheoreticalconvictionsandstyleofinquiryisamoresuspiciousformof
scientificimperialism.Thelattercancreateorreinforcepositionsofdominancethathavelittle
or nothing to do with recognized scientific merits. Whether breaking disciplinary boundaries
turnintosuspiciousformsofimperialismcanonlybeascertainedonacasebycasebasis.

3.UNIFICATIONINTHENEWECONOMICGEOGRAPHY

The New Economic Geography (NEG henceforth) is an approach developed within economics
at the beginning of the 1990s that seeks to explain spatial agglomeration (and spatial
dispersion) within a unified framework (Krugman 1991, Fujita et al. 1999, Fujita and Thisse
2002, Brakman et al. 2001, Baldwin et al. 2003). The concept of agglomeration refers to
seemingly very distinct empirical phenomena: the existence of the coreperiphery structure
CaterinaMarchionniScientificUnificationinEconomics

15

corresponding to NorthSouth dualism; regional disparities within countries; the existence of


cities and systems of cities, which are sometimes specialized in a small number of industries;
industryclusterssuchasSiliconValley;andfinallythepresenceofcommercialdistrictswithin
cities, such as Soho in London. Although each type of agglomeration could be the result of
differenttypesofmechanisms,NEGhypothesesthattheseapparentlydistinctphenomenaare
at least partly the result of similar mechanisms, viz. economic mechanisms yielding
agglomeration by relying on the tradeoff between various forms of increasing returns and
differenttypesofmobilitycosts(FujitaandThisse2002:1).
The approach rests on two building blocks: the presence of increasing returns at the firm
level and transportation costs.
3
Increasing returns at the firm level requires dropping the
standard economic assumption of perfect competition and replacing it with that of imperfect
competition.
4
On the aggregate level, increasing returns and transportation costs give rise to
pecuniary externalities, which are transmitted through the market via effects on prices. In
essence, their presence implies that the more firms and workers there are in a locality, the
largerthemarket,themoreattractiveitbecomesasalocationforfurtherfirmsandworkers.
5

This generates a cumulative process that may result in the concentration of all economic
activity in one location. While pecuniary externalities push towards such concentration,
immobile factors, congestion and the like, push towards dispersion. The relative strength of
thetwosetsofforcesdetermineswhetherspatialagglomerationoccurs.
NEG is now a wellestablished field in economics. In 2008 Paul Krugman was awarded the
NobelPrizeinEconomics.TheNobelPrizepressreleasementionedtheintegrationofthefields
oftradeandgeographyasoneoftheremarkableachievementsofKrugmanstheory:
Patterns of trade and location have always been key issues in the economic debate. What are
the effects of free trade and globalization? What are the driving forces behind worldwide
urbanization? Paul Krugman has formulated a new theory to answer these questions. He has
therebyintegratedthepreviouslydisparateresearchfieldsofinternationaltradeandeconomic
geography.(Myemphasis)
6

For NEG theorists too, unification is one of its main accomplishments: it shows that similar
kinds of economic mechanisms are at work in bringing about a host of phenomena that have
been separately studied, thereby achieving explanatory unification. In doing so, it answers
Ohlinscall(1933)fortheunificationoftradetheoryandlocationtheory(Brakmanetal.2001).
And yet this enthusiasm for the NEGs achievements is not shared by all. Economic
geographers, who have been traditionally concerned with spatial issues, received the NEGs

3
The presence of space implies that the cost of exchanging goods and services across locations
increases as (both physical and cultural) distance increases due to physical transport, tariffs, cultural
barriersandsoon.Thesecostsaregenericallytreatedastransportationcosts.
4
Increasing returns are defined as a decrease in the average costs per unit of output for the individual
firmasthelevelofoutputincreases.IntheimperfectcompetitionsettingNEGassumes,firmstakethe
pricesettingbehaviorofotherfirmsasgiven,anddonottakeintoaccounttheeffectsofchangingtheir
ownpriceontothepriceindex.Theproductsaresymmetric,whichmeansthatconsumersdonotprefer
one variety to another. However, they prefer variety for its own sake, which means that they always
prefer to consume a unit of a new variety than an additional unit of a product they have already
consumed.Thisisoneofthereasonswhyfirmagglomerationisbeneficialforworkers.
5
Externalitiesaredefinedasadecreaseinaveragecostsasaresultofanincreaseatthelevelofoutput
of the whole industry. Pecuniary externalities are transmitted through the market via price effects for
eachfirm,which,asaconsequence,maydecidetochangeitsoutputdecisions.
6
TheNobelPrizepressreleasecanbefoundat
http://nobelprize.org/nobel_prizes/economics/laureates/2008/press.html.

Humana.MenteIssue10July2009

16

appearance with suspicion, if not with outright hostility. Among the accusations economic
geographershavelevelledagainstNEGisthatitrepresentsyetanotherinstanceofimperialism
on the part of economics. We can already begin to see that the NEGs pursuit of explanatory
unificationhasaffectedarangeofintertheoretic,interfieldandinterdisciplinaryrelations.It
istotheseissuesthattheremainderofthissectionisdevoted.

3.1.EXPLANATORYUNIFICATION

NEG seeks to explain a host of spatial phenomena previously studied by separate theories in
different fields by showing how these are brought about by the same kind of economic
mechanisms. The account of explanatory unification advanced by Philip Kitcher (1981, 1989)
largely captures the structure of explanatory unification in NEG (Mki and Marchionni 2009).
AccordingtoKitcher,toexplainistounify,andunification(andhenceexplanation)isamatter
of inference and derivation. Explanatory unification proceeds by reducing the number of
argument patterns while maximizing the number of explanandum sentences that can be
derived.
7
In economics, argument patterns are typically embodied in model types, so that
unificationthereinproceedsviatheapplicationofasmallnumberofsimilarmodeltypestoan
increasingnumberofeconomicaswellasnoneconomicphenomena.
In the case of NEG two similar model types have effected the unification: the core
peripherymodel(henceforththeCPmodel)andtheverticallinkagesmodel(henceforththeVL
model).BothoftheseadoptthemonopolisticcompetitionframeworksetoutintheDixitand
Stiglitz (1977) model with transportation costs, and derive the agglomeration of economic
activity between two a priori identical locations. The difference between them lies in the
specific foundations for the agglomeration mechanism they postulate; in other words, each
type specifies more specific versions of the generic mechanism described in the previous
section.IntheCPmodel,thesizeofthemarketineachlocationisdeterminedbythemigration
decisions of workers: a larger market is a more attractive location to firms and hence to
workers via a reduction in the price of the goods. In the VL model, on the other hand, the
workers do not move, and the size of the market is made endogenous by the presence of
inputoutput linkages between firms: the more firms there are in a location the larger is the
market for upstream firms and the lower are the costs for downstream firms. These model
typesarethenfilledinwithspecificvariablesinordertoderivediversetypesofagglomeration
phenomena.
Two features of NEG however diverge from Kitchers account of explanatory unification.
First, in Kitchers, unification is a matter of deriving large classes of explanandum sentences
fromaparsimonioussetofinferentialpatterns.Butexplanatoryunificationcanalsobeseenas
a matter of redescribing a large number of apparently independent phenomena as
manifestations of a common set of mechanisms, thereby revealing an underlying ontic unity
between the phenomena (Mki 2001). In NEG the two kinds of unification go together. The
kindofunityNEGpursueshastodowithdifferentphenomenabeinggovernedbysimilarkinds
of economic mechanisms. The CP and the VL model types are not merely similar patterns of

7
Argument patterns include three components: 1) a schematic argument: a sequence of schematic
sentences, which are expressions in which most if not all nonlogical expressions are replaced with
dummy letters; 2) a set of filling instructions that indicate how dummy letters are to be replaced in
specific applications; 3) a classification for the schematic argument that provides the inferential
characteristics of the argument, dividing the schematic sentences between premises, conclusions, and
rulesofinference(Kitcher1981).
CaterinaMarchionniScientificUnificationinEconomics

17

derivation, but they embody stylized causal mechanisms, which are fleshed out according to
thespecificsofeachexplanandumphenomenon.
The second, related feature is that in contrast to Kitcher, NEG theorists endorse the view,
widespreadamongeconomists,thataphenomenonisgenuinelyexplainedonlyifitisderived
fromwelldefinedmicroeconomicparameters(MkiandMarchionni2009,Marchionni2009).
NEGscontributionvisvisexistingtheoriesisheldtobetwofold.First,itunifiesphenomena
that were previously treated separately: the unificationist contribution. Second, it is the only
field within economics that provides a microfoundation in a general equilibrium framework
for the spatial distribution of economic activity: the explanatory contribution. If genuine
explanationinNEGhastodowiththepresenceofmicrofoundations,thederivationofdiverse
phenomena from a small number of model types is not what the theorists take to do the
explanatory work. This is so even though it is the search for microfoundations (the
explanatory contribution) that has revealed that different classes of agglomeration
phenomena are governed by the same kind of economic mechanisms (the unificationist
contribution).
In the case of NEG, explanation and unification of phenomena have proceeded in parallel,
but they trace back to distinct components of the theory. Furthermore, for certain kinds of
explananda, they may go in opposite directions: the unification of phenomena demands that
explanatory information be abstracted away. As the theorists admit, explanatory unification
could not be achieved without neglecting the details that are specific to different kinds of
agglomerationphenomena:

By using highly stylized models, which no doubt neglect a lot of specifics about
urban/regional/international phenomena, geographical economics is able to show that the
samemechanismsareatworkatdifferentlevelsofspatialaggregation.[]
Inordertolaythefoundationsforaunifiedapproach,thereisapricetobepaidintermsofa
neglectofinstitutionalandgeographicaldetails.(Brakmanetal.2001:323)

Asthisquotesuggests,unificationrequiredthatahostofinstitutionalandgeographicaldetails
specific to the phenomenon be left out. But these specific details are needed to fully account
forthespatialphenomenathatNEGunifies.Itfollowsthatwhenthetheoryisusedtoexplain
particular aspects of specific phenomena (cities for instance), the NEG unifying mechanism
alone cannot do all the explanatory work. It needs to be complemented with information
specific to the kind of phenomenon to be explained (such as the presence of land developers
or the presence of knowledge spillovers).
8
As we will see in the next section, some of the
specificinformationisprovidedbyothereconomictheories.

3.2.THEORETICALUNIFICATION

Back in 1933, Bertil Ohlin, a wellknown international trade theorist, claimed that the
separationbetweeninternationaltradetheoryandlocationtheorywasartificial:International
trade theory cannot be understood except in relation to and as part of the general location
theory,towhichthelackofmobilityofgoodsandfactorshasequalrelevance(p.142).Later
on he complained that no serious attempt to build a general location theory and introduce
national borders had been made (Ohlin 1979: 6). Only twelve years later, Paul Krugman
(1991) published the seminal NEG model. NEG promises finally to realize Ohlins dream of a

8
Mki and Marchionni (2009) examine this issue in detail by distinguishing kinds of information and
kindsofexplananda.

Humana.MenteIssue10July2009

18

general theory of the location of economic activity, as the scientific background report of the
NobelPrizecommitteenotes:

Traditionally, trade theory and economic geography evolved as separate subfields of


economics. More recently, however, they have become more and more united through new
theoretical insights, which emphasize that the same basic forces simultaneously determine
specialization across countries for a given international distribution of factors of production
(trade theory) and the longrun location of those factors across countries (economic
geography)(p.1,myemphasis).
9

Furthermore, efforts have recently been made to integrate economic growth in the spatial
models of NEG. Thus NEG holds out the promise to unify theories of location, international
tradeandgrowth.

According to the standard view of theoretical unification, NEG, the unifying theory, will
eventually replace international and location theory, and possibly also growth theory. If NEG
hadalltheexplanatorycontentoftheunifiedtheories,thentheycouldbedispensedwith.But
unificationinscientificpracticedoesnotalwayslooklikethis.Iftheexplanatorycontentofthe
unifying theory only overlaps with and does not fully cover that of the unified theories, then
dispensing with the latter amounts to leaving some explananda unexplained. And at least so
farNEGcannotaccountforanumberofstylizedfactsaboutlocation,tradeandgrowth.Thisis
mainly because it focuses exclusively on a kind of economic mechanism (pecuniary
externalities) that is believed to operate in bringing about diverse classes of phenomena. In
reality there are other mechanisms and forces specific to each class of phenomena that NEG
doesnotencompass.Theimportanceofthealternativemechanismsandforceswillvaryfrom
case to case, so that whereas for certain stylized facts the NEG mechanism will be more
important than the specific ones, in other cases the reverse will be the case. Economists, in
effect, generally perceive growth theory, trade theory, location theory and NEG as
complementary.
10
Within each field different theories in fact postulate different kinds of
economicmechanismstoaccountfortheirrespectivephenomena.Dispensingwithonetheory
basically means dispensing with one kind of mechanism and one possible explanation.
Depending on the phenomenon we want to explain, a different mechanism or a different
combinationofmechanismsactingtogetherwillberelevant.Inprinciplefurtherdevelopments
withinNEGandneighbouringfieldscouldproduceagrandtheorythattellsuswhen,howand
which combinations of mechanisms operate in bringing about certain phenomena. As things
stand now, however, there is little reason to believe so. What we now have is a plurality of
overlapping,interlockingtheoriesindifferentsubfields,whichNEGhashelpedtorendermore
integrated. Here there is clear tension between the opposing forces for and against unity: on

9
The report can be found at http://nobelprize.org/nobel_prizes/economics/laureates/2008/sci.html.
Note that the report talks of economic geography as one of the fields that NEG unifies. But this is
misleading because as Ill explain below the field of economic geography distantiates itself quite
considerably from both economics and NEG. It is more appropriate to talk of location theory, an
interdisciplinary field, involving regional science, urban and regional economics, concerned with
modeling the determinants and consequences of the locational decisions of economic agents (see
Marchionni2009).
10
Growth theory, trade theory, and location theory include a variety of theories and explanatory
models. They are more like fields than theories. In this section for simplicity I use these labels to stand
fortheories.ThepointImakeabouttheoreticalunificationhoweverappliestothetheoriesthatwithin
each field seek to account for the phenomena NEG unifies (see Brakman et al. 2001 and Marchionni
2009).Section3.3shiftsthefocusfromtheoriestofields.
CaterinaMarchionniScientificUnificationinEconomics

19

the one hand the unifying ambitions of NEG push towards the unification of phenomena and
theories,andontheotherthepresenceoftheorieswhosedomainsonlypartiallyoverlapwith
thatofNEGmaintainsacertaindegreeofdisunity.

3.3.INTERFIELDINTEGRATION

LookingatinterfieldconnectionsoffersadditionalinsightsintoNEGunification.Tradetheory,
location theory and growth theory are perhaps more appropriately treated as fields rather
than theories because they include several explanatory models that aim to account for a
number of stylized facts about, respectively, international trade, location and economic
growth. From this broader perspective, NEG can be looked upon as an interfield theory that
examines the relations between the phenomena studied by different fields and thereby
contributestorenderthesefieldsmoreintegrated.
The development of NEG, in fact, is closely tied to that of contemporary theories of trade
andgrowthinthecontextofwhatisknownastheincreasingreturnsrevolutionineconomics.
NEGissaidtobethefourthwaveoftheincreasingreturnsrevolutionineconomics(Fujitaet
al.1999:3).Therevolutionconsistsinshiftingawayfromtheconstantreturnstoscaleperfect
competition paradigm that dominated the discipline of economics until the 1970s and 1980s,
and adopt in its stead an increasing returnsimperfect competition framework. Economists
havebeenawareofthecrucialimportanceofincreasingreturnsandimperfectcompetitionfor
long,andyettheylackedthetoolstoadequatelyincorporatethemintheformalandrigorous
models economists favour. The DixitStiglitz model of monopolistic competition offered
precisely that tool. Thanks to its workability and analytical flexibility, the model could be
applied to different areas of inquiry. It was originally conceived as a contribution to the
literature on product differentiation, but it was later applied to the phenomena of
international trade, growth, and geography, all of which are thought to be the result of the
presenceofincreasingreturns.
The application of the DixitStiglitz model to phenomena of growth and trade follows a
similarpath.Inbothcases,thethendominantneoclassicalvariant(basedonconstantreturns
and perfect competition) was unable to address some stylized facts and the existence of
increasing returns was thought to be necessary to explain them. In commenting the situation
of trade theory around the 1970s, Paul Krugman (1995: 1244) describes it as a collection of
highly disparate and messy approaches, standing both in contrast and in opposition to the
impressive unity and clarity of the constantreturns, perfect competition theory. Thanks to
the DixitStiglitz model, theories of growth and trade based on increasing returns (new trade
theory and new growth theory respectively) became serious complements to the
neoclassical ones. Then, a sequence of extensions to Krugmans new trade theory models led
totheformulationofthefirstNEGsmodel,thatis,Krugman1991(Brakmanetal.2001).
TheroleoftheDixitStiglitzmodelofmonopolisticcompetitionbecomesapparentoncewe
shiftthefocusfromtheoriestofieldsandthereintowhatGranthamcallspracticalunification.
The model functioned as a vehicle towards the integration of several fields in economics: the
sameanalyticaltoolwasemployedwiththeappropriatemodificationsinvariousfields,andfor
precisely the same purpose, namely to deal with increasing returns at the firm level and
imperfect competition. Its application to the domain of spatial phenomena finally led to the
development of NEG, which in turn significantly contributed to the integration of these fields
atthetheoreticallevel.NEGcanthenbethoughtasaninterfieldtheorythatisbothaproduct
of this process of integration as well as a vehicle of further integration. Both practical and
theoreticalconnectionshavesimultaneouslyworkedtowardstheintegrationofseveralrelated
fields.Evenso,thefieldscontinuetoproceedlargelyindependentlyfromeachother.Thedrive

Humana.MenteIssue10July2009

20

towardsunificationcontinuestoplayaprominentrolewithineachfieldandacrossthem,
11
but
as mentioned earlier, so far NEG is unable (and does not aim) to account for a number of
stylizedfactsthattheoriesbelongingtothesefieldsseektoexplain.

3.4.ECONOMICSIMPERIALISM

The development of NEG has also had effects on fields outside economics: economic
geography, a subfield of human geography, lays claim to substantial parts of its domain.
Economic geography is diverse in both scope and methods. At its core, however, is an
emphasisonthecomplexityofempiricalreality,onplaceratherthanspace,onconceptssuch
ascontingencyandspecificityand,onthelevelofmethod,ontheextensiveuseofcasestudies
andadiscursivestyleoftheorizing.Giventhatthemainstreamoftheeconomicsdisciplinehas
traditionally neglected spatial and geographical issues, it is not surprising that economic
geographersperceivedNEGasanattempttoinvadetheirownterritory.
12

Mki and Marchionni (2007) deal with this aspect of the relationship between NEG and
economic geography. We propose to consider NEGs foray into the field of economic
geography a consequence of its pursuit of unification and suggest that there is nothing
inherently problematic in this pursuit. Disciplinary boundaries are not sacred, and it just so
happensthatthephenomenaNEGunifiesalsofallwithinthepurviewofanotherdiscipline.On
theotherhand,wealsowarnagainstthedangerofunconditionallycelebratingunificationand
theensuingimperialism.ItisindeedpossiblethatthespatialphenomenaNEGunifiesarenot
in reality so unified, and/or that the range and significance of the spatial questions NEG can
addressisverylimited(aseconomicgeographershavenotfailedtopointout).Allthisneedsto
be established empirically, however. More importantly, as mentioned above, scientific
imperialism also affects sociological and institutional aspects of interdisciplinary relations.
When what is exported is not only a theory, but a purportedly superior research style and/or
thehigherstandingofadiscipline,themechanismssustainingthesenonepistemicaspectsof
imperialism may tip the balance in favour of theories for which empirical support is poor at
best.
Asignificantcomponentoftheconcernexpressedbyeconomicgeographersindeedhasto
dowiththeseaspectsofimperialism.Thedanger,astheyperceiveit,isthatNEGmightendup
enjoying increasing policy influence just because of the higher standing of economics.
Similarly,theallegedscientificityofeconomicscouldgiveNEGanextraedgeintheacademic
competition and end up colonizing economic geography at the cost of the latters theoretical
andmethodologicalcommitment.Asthingsstandnow,itisnotatallclearthatNEGwarrantsa
superior position visvis the theories of economic geographers in the domain of spatial
phenomena. In sum, the NEG pursuit of unification generates pressures on disciplinary
boundaries but in spite of attempts at promoting crossfertilization, which if successful could
promoteintegration,thetwofieldslargelyproceedinmutualdisregardforeachotherswork.


11
For example trade theory models built to include both neoclassical and new trade theory
determinantsofinternationaltrade,orrecenteffortsinNEGaimedattheconstructionofmoregeneral
modelsfromwhichspecificmodelscanbederivedasspecialcases.
12
The choice of calling it new economic geography has probably contributed to reinforce the
geographers perception of imperialism. Some NEG theorists proposed to adopt geographical
economics instead to underscore the disciplinary origin of the approach, but the original name still
remainsprevalent.
CaterinaMarchionniScientificUnificationinEconomics

21

4.CONCLUSIONS
Philosophical interest in the practice of scientific unification in economics has been scarce. A
perspective acknowledging the potential benefits of economists unifying practices as well as
thelimitationshaslargelybeenmissing.Thisisagapthatneedstobefilledbecausethequest
for unity constitutes a powerful driver of theoretical development in economics that
significantlyaffectsitsinternaldevelopmentaswellasitsrelationshipwithneighbouringfields
and disciplines. We have seen that the NEGs efforts to provide a unified explanation of
agglomeration phenomena by showing how they are brought about by the same kind of
economic mechanisms has influenced its relationship to other theories, fields and disciplines.
From this case, it also emerges quite clearly that unification encounters resistances that
ultimately influence the kind and degree of unification that is established at each level. First,
unification of phenomena turns out to be a distinct achievement from explanation and for
certainexplanatorypurposes,unificationandexplanationpullagainsteachother.Second,the
stylized mechanism that NEG postulates and on which the unification is based upon does not
suffice to do all the explanatory job of existing theories dealing with the phenomena NEG
unifies.Third,asaninterfieldtheoryNEGhasfurtheradvancedtheintegrationofanumberof
relatedsubfieldsineconomics,butforthemostpartthesecontinuetoproceedautonomously.
Finally,insofarasinterdisciplinaryrelationsareconcerned,inspiteofafewattemptsatcross
fertilization, NEG and economic geography continue to proceed in mutual ignorance of each
otherswork.
I believe that other cases of unification in economics can be usefully examined by paying
attention to the levels at which unification occurs and the frictions it faces at each of these
levels. Consider for example neuroeconomics. It is often seen as an attempt to reduce
economic theories to theories about the brain, and neuroeconomists themselves have
sometimes expressed unifying ambitions.
13
The following quote from two major
representativesistellinginthisrespect:

Economics,psychology,andneuroscienceareconvergingtoday

intoasingle,unifieddisciplinewith
theultimateaimofproviding

asingle,generaltheoryofhumanbehaviour.Thisistheemerging

field
of neuroeconomics in which consilience, the accordance

of two or more inductions drawn from


differentgroupsofphenomena,

seemstobeoperating.Economistsandpsychologistsareproviding

rich conceptual tools for understanding and modelling behaviour,

while neurobiologists provide


toolsforthestudyofmechanism.(GlimcherandRustichini2004:447)

UnlikeNEG,however,themainvehiclesofunificationarethenewtechnologies(PETandfMRI
for example), which make it possible to generate data on the brain level, and a batch of
methods, including experimental templates (e.g., the Ultimatum Game), which have become
stable enough to travel across disciplinary boundaries (cf. Guala 2009). These have made
availableaplethoraofnewdatainasituationinwhichempiricalevidencehasbeenchronically
lacking. Neuroeconomics potentially increases the degree of integration among economics,
psychologyandneuroscience,whichinturncouldleadtoimprovedknowledgeaboutdecision
making. For instance, neuroscientific knowledge may generate new behavioural hypotheses
andintroduce(mechanistic)constraintsonmodelsofindividualchoice(Clitheroetal.2008).
Fromthedisunityside,manyhaveclaimedthateconomicsisandshouldremainunaffected
by neuroscientific findings (e.g., Gul and Pesendorfer 2008). I suspect that this dismissive
reaction is, to some extent, due to worries concerning aspects of neuroscience imperialism

13
Thereductionofhigherleveltheoriestotheoriesatthelowerlevelalsocontributestotheunification
ofscience.ThisisinfactOppenheimandPutnam(1958)smodelofscientificunitymentionedabove.

Humana.MenteIssue10July2009

22

thatcouldthreateneconomistsowndisciplinaryidentity.Theideathateconomicsshouldbe
insulated from whatever is going on in neuroscience is perhaps too extreme. Nevertheless,
although neuroeconomics is still in its infancy and it is too soon to tell how it is will develop,
there are good reasons to doubt that it will ever replace standard ways of doing economics.
For one thing, the contribution of neuroscience to economics has to pass through psychology
(e.g.,PadoaSchioppa2008),andawealthofarguments,whichIwontgointohere,hasmade
it clear that the global reduction of psychology to neuroscience is unfeasible and possibly
detrimental
14
even though successful local reductions occur. As in the NEG case, here, too
the quest for unity plays an important role, but the frictions exerted by disunities of various
sortscruciallyinfluencethekindofproductsthataredeliveredinpractice.
Infacttheproductsthatthepursuitofscientificunificationdeliversareoftenafarcryfrom
the splendid achievements that standard philosophical models presuppose. Still, I believe,
economists unifying practices, and their effects within and around economics, are worth
philosophicalattention.Ifwethinkofunificationasasetofactivitiesgoodforsomepurposes
and only in some contexts, analyzing the way in which economists actually practice them will
help us to identify their strengths and their limitations, and possibly to calibrate them as
tools(Wimsatt2006:446).Somefeaturesmaywellremainstableacrosscontextsandlocales
even though the specific activities and the frictions they face are different from case to case.
The analysis of the NEG unification should be seen as a contribution to this more ambitious
project.

Acknowledgments: Part of this paper is based on work done jointly with Uskali Mki. I thank
Emrah Aydinonat and Alessandro Lanteri for useful comments on a previous draft of this
paper.Allremainingmistakesareobviouslymine.


BIBLIOGRAPHY

Baldwin R., Forslid R., Martin P., Ottaviano G. and RobertNicoud F. (2003) Economic
GeographyandPublicPolicy.PrincetonUniversityPress.
BechtelW.andHamiltonA.(2008)Reduction,integrationandtheunityofscience:natural,
behavioral, and social sciences and the humanities. In T. Kuipers (ed.) Philosophy of
Science:FocalIssues.NewYork:Elsevier,377430.
Brakman S., Garretsen H. and van Marrewijk C. (2001) An Introduction to Geographical
Economics:Trade,LocationandGrowth.Cambridge:CambridgeUniversityPress.
Cat J. (2007) The unity of science. Stanford Encyclopedia of Philosophy.
http://plato.stanford.edu/entries/scientificunity.
Clithero J.A., Tankersley D. and Huettel S.A. (2008) Foundations of neuroeconomics: from
philosophytopractice.PL.
DardenLandMaullN.(1977)Interfieldtheories.PhilosophyofScience43,4464

14
The standard defence against the reducibility of psychology to neuroscience is the multiple
realisability argument (Fodor 1974). How exactly mental states and brain states are related and what
consequencesthisrelationhasforthesciencesinvolvedisstillahotlydebatedissue.
CaterinaMarchionniScientificUnificationinEconomics

23

Dixit A.K. and Stiglitz J.E. (1977) Monopolistic competition and optimum product diversity
AmericanEconomicReview67,297308.
Fodor J.A. (1974) Special sciences (or: the disunity of science as a working hypothesis.)
Synthese28,97115.
FriedmanM.(1974)Explanationandscientificunderstanding.JournalofPhilosophy71,519
FujitaM.andThisseJF.(2002)EconomicsofAgglomeration.Cities,IndustrialLocationand
RegionalGrowth.Cambridge:CambridgeUniversityPress.
Fujita M. Krugman P. Venables A. (1999) The Spatial Economy. Cities, Regions and
InternationalTrade.CambridgeMass:MITPress.
Glimcher P.W. and Rustichini A. (2004) Neuroeconomics: the consilience of brain and
decision.Science306,447452.
Grantham T.A. (2004) Conceptualizing the (dis)unity of science. Philosophy of Science 71,
133155.
Guala F. (2009) Paradigmatic experiments: the ultimatum game from testing to
measurementdevice.PhilosophyofScience75,658669
Gul F. and Pesendorfer, W. (2008) The case for mindless economics. In A. Caplin, A.
Schotter (Eds.) The Foundations of Positive and Normative Economics. A Handbook.
Oxford:OxfordUniversityPress,339
KitcherP.(1981)ExplanatoryUnification.PhilosophyofScience48:25181
Kitcher P. (1984) 1953 and all that: a tale of two sciences. Philosophical Review 93, 335
373
Kitcher P. (1989) Explanatory unification and causal structure. Minnesota Studies in the
PhilosophyofScienceVol.13.Minneapolis:UniversityofMinnesotaPress,410505
Mki U. (2001) Explanatory unification: double and doubtful. Philosophy of the Social
Sciences31(4),488506.
Mki U. (2009) Economics imperialism: concept and constraint. Philosophy of the Social
Sciences.
Mki U. and Marchionni, C. (2007) Is geographical economics a form of intellectual
imperialism?Unpublishedmanuscript.
Mki U. and Marchionni, C. (2009) On the structure of explanatory unification: the case of
geographical economics. International Studies in the History and Philosophy of
Science.
MarchionniC.(2009)Geographicaleconomics&itsneighborsforcestowardsandagainst
unification. In U. Mki (ed.) Elsevier Handbook of the Philosophy of Science, vol. 13:
PhilosophyofEconomics.NewYork:Elsevier.
OhlinB.(1933)InterregionalandInternationalTrade.Cambridge,Mass:HarvardUniversity
Press.
Ohlin B. (1979) Some Insufficiencies in the Theories of International Economic Relations.
Princeton:PrincetonUniversity.

Humana.MenteIssue10July2009

24

Oppenheim P. and Putnam, H. (1958) The unity of science as a working hypothesis. In H.


Feigl, M. Scriven, and G. Maxwell (eds.), Minnesota Studies in the Philosophy of
Science.Minneapolis:UniversityofMinnesotaPress,336
PadoaSchioppa,C.(2008)Thesyllogismofneuroeconomics.EconomicsandPhilosophy24,
449457.
Sober E. (1999) The multiple realizability argument against reductionism. Philosophy of
Science66(44)542564
Thagard P. (2007) Coherence, truth, and the development of scientific knowledge.
PhilosophyofScience74,2847
Whewell W. (1847) The Philosophy of the Inductive Sciences Founded on their History.
London:JohnW.Parker.
Wimsatt W. (2006) Reductionism and its heuristics: making methodological reductionism
honestSynthese151,445475.
Wylie A. (1999) Rethinking unity as a working hypothesis for philosophy of science: how
archaeologistsexploitthedisunitiesofscience.PerspectivesonScience7(3):293317.


25

CausalArrowsinEconometricModels

FedericaRusso
*

f.russo@kent.ac.uk

ABSTRACT

Econometrics applies statistical methods to study economic phenomena. Roughly, by means of


equations, econometricians typically account for the response variable in terms of a number of
explanatory variables. The question arises under what conditions econometric models can be given a
causal interpretation. By drawing the distinction between associational models and causal models, the
paper argues that a proper use of background knowledge, three distinct types of assumptions
(statistical, extrastatistical, and causal), and the hypotheticodeductive methodology provide sufficient
conditionsforacausalinterpretationofeconometricmodels.

1.INTRODUCTION

A vexata quaestio in philosophy of economics is the extent to which econometric models tell
causalstories.Thequestioniscertainlynotnewandoccupiesmuchofphilosophicallyminded
discussionsofthemethodologyofeconomicsand,moregenerally,ofthesocialsciences.
Thispaperaddstotheliteratureofferingitsownposition.ThepositionIdefendstemsfrom
the dissatisfaction with a certain class of argumentswhich I shall call metaphysical
argumentsand from the recognition that another class of argumentswhich I shall call
methodologicalargumentsareontherighttrackbutonlygohalfwaythrough.
The paper is organized as follows. In section 2, I explain the particular epistemological
viewpoint I adopt in looking at econometric models. The motivation for an epistemological
approach to causality comes from the dissatisfaction with metaphysical arguments that
assumecausalstructuresgeneratingtheobserveddatainsteadofexplaininghowwecometo
establish whether a given correlation is causal. In section 3, I review traditional
methodological arguments given by philosophers of economics and philosophicallyminded
econometricians. In those arguments two trends can be identified: those who stress the
importance of causal mechanisms and those who put the whole burden of the causal
interpretationintheassumptionsofthemodel.Ithenofferinsection4myownanswer.Ishall
do that in two steps. First, I present an account of the conditions under which econometric
models can be given a causal interpretation by defending the following interrelated ideas: (i)
there is an important distinction between associational models and causal models; (ii) the
difference lies in their respective features, notably whilst associational models just have
statistical assumptions, causal models also have extrastatistical and causal assumptions,
employahypotheticodeductivemethodology,andusebackgroundknowledgeinanessential
way at each stage of the model building and model testing process. Second, I discuss the
methodology,results,andcritiquesofacasestudyontherelationsbetweenhealthandwealth
inelderlyAmericans.


*
Honorary ResearchFellow,UniversityofKent

Humana.MenteIssue10July2009

26

2.ANEPISTEMOLOGICALINVESTIGATIONINTOECONOMETRICMODELS

Let me make clear from the outset what perspective I shall take in looking at econometric
models.Inthephilosophyofcausalitythreebroadareasofinvestigationmaybedistinguished.
The metaphysics of causality is interested in what causality in fact is, in what kind of entities
causes are, or in what a causal claim means. The epistemology of causality, instead,
investigates how we come to know about causal relations. There is a fleeting borderline
between epistemology and methodology, but a line between the two can be drawn
nonetheless. Whilst methodology is concerned with problems of scientific methods and aims
at developing successful methods for the discovery and confirmation of causal relationships,
epistemology is rather interested in the conceptual issues behind those methods.
Metaphysical, epistemological, and methodological issues ought not to be conflated and a
jointinvestigationfromthesethreeperspectiveswillhopefullyallowustogetabettergripon
causation.
This paper entirely locates within the domain of epistemology and methodology. Namely,
the focus is on how we come to know about causal relations, regardless of the position one
may take about the metaphysics of causation, i.e., about what causation in fact is. The
motivation for an epistemological perspective comes from difficulties in metaphysical
arguments, according to which, simply put, genuine probabilistic dependencies are causal
dependencies,i.e.genuinecorrelationsarepersecausal.AnexampleisHausman(1998)who
reiterates the idea that true correlations are causal and that if, eventually, the correlation
turns out to be spurious this means that we picked out the wrong correlation (see, e.g.,
Hausman 1998, p.33, 56). Differently put, correlation does prove causationaccidental
correlations,inHausmansview,simplyarenttruecorrelations.Tobesure,whatismeantby
these kinds of argument is that modulo certain conditionsmost typically the socalled
Markov conditioncorrelations are causal. However, it is highly a controversial matter
whether the Markov condition ensure causality (see for instance Hausman and Woodward
1999and2004andCartwright2002onthispoint).
There is also another problem with metaphysical arguments. A stock example is the
positivecorrelationbetweentheincreasingnumberofstorksandtheincreasingofbirthrates
in Alsace. According to arguments la Hausman, this correlation is accidental, not genuine,
hence not causal. Agreed, it might well be the case that genuine correlations be causal, but
whywedonotbelievethattheincreasingnumberofstorksiscausallyrelatedtotheincreasing
number of births? The answer, it seems to me, is that our background knowledge does not
contain any theory or piece of information that makes this correlation plausible, let alone
plausiblycausal,inanypossibleway.
Inasimilarvein,Cartwright(1989)believesthatcapacitiesareresponsiblefor,orgiveraise
to, stable regularities and therefore are the very ontological basis of any observed statistical
correlation. Capacities have the peculiar feature of being stable across different background
conditions,butthequestionis,inthefirstplace,howdoweknowthatsomethingisacapacity.
At times, Cartwright (1989, pp.148 ff) seems to introduce a circularity in the account because
themainpointofeconometricmodelistoteststabilityofrelationshipsbetweenvariables;but
those relations will be stable in case they are the effects of capacities, therefore stability
cannotbeatesttoknowwhethersomethingisacapacityinthefirstplace.Steel(2008,pp.82
85)risesasimilarworryconcerningextrapolation,becauseifacausalrelationhastohavethe
feature of being stable across different contexts, then capacities cannot be what grounds
extrapolation inferences exactly because what we are trying to establish is whether (stable)
capacitiesareatwork.
FedericaRussoCausalArrowsinEconometricModels

27

Thus, even if we assume that there are true causes that give rise to the observed
correlations,thequestionstillremainsastohowwecometoestablishthatsomecorrelations,
andnotothers,arecausal.EchoingGranger(2003),thequestioniswhethercausalityapplies
to the model or the data generating process. Id opt for the former option, in particular, my
interest then lies in the process of model building and model testing, and, within in this
process, I aim to highlight the conditions under which econometric models can be given a
causalinterpretation.ThistaskIshallundertakeinsection4.

3.STANDARDVIEWS

The literature in econometrics and social science is indeed vast, but two broad families of
arguments can be identified nonetheless. A first family of arguments is centred around the
notion of mechanism and the second around the assumptions of econometric models. To be
sure, those two families are not mutually exclusive and there in fact a version of the
mechanismargumentthateventuallycollapseintotheassumptionargument.Letmeexplain
further.
Mechanisms are usually central for those who advocate a structural approach in
econometrics. Simply put, and leaving aside the technicalities, structural models explain
economicphenomenabymeansof(setsof)equationsthatdescribecausalmechanisms.
In the philosophy of economics and social science, a noteworthy partisan of the
mechanistic approach is Little (1990), who claims that causal analysis in the social sciences is
legitimate insofar as models identify social mechanisms. Little believes that such social
mechanismsworkthroughtheactionsofindividualsapositionalsoknownasmethodological
individualism.Hoover(2001),instead,stressesthecausalimportofthestructuralapproachin
econometricsarguingforarealityofmacroeconomicstructuresthatdoesnotboildowntothe
reality of microeconomic relations. Hoover also tags along with mechanistic approaches
becauseacausalstructureis,inhisview,anetworkofcounterfactualrelationsthatmapsout
theunderlyingmechanismsthroughwhichonethingisusedtocontrolormanipulateanother
(2001, p.24). The idea behind approaches la Little (but not necessarily la Hoover) is that
there is a causal relation between A and B if and only if there is a mechanism that links A to
Ba position hold, in slightly different ways, also by other mechanical philosophers.
Nevertheless,theemphasisonmechanismsisfamouslycriticizedbyKinkaidandmorerecently
byReiss.Kinkaid(1996),forinstance,thinksitisfalsethatinordertoknowwhetherXcausesY
atleastamechanismlinkingXtoYhastobeidentified.Reiss(2007band2008,ch.6),alongthe
same lines, argues that mechanisms might not be the most useful strategy to achieve other
goals,forinstancemeasuringconceptssuchasinflationrateorunemploymentrate.
A different emphasis on mechanisms is given by economic theorists, who describe
economic mechanisms that work ceteris paribus. Such position is typical of classical
economistssuchasAdamSmith,DavidRicardo,ThomasMalthus,andJohnStuartMill,orthe
ChicagoSchoolofEconomics.Onewayoranother,itistheeconomictheorythatdictatesthe
mechanism, instead of conceiving the whole modelling procedure as the attempt to identify
whatthemechanismisandhowitworks.Itisinthisveinthat,accordingtoHeckman(2008),
structural econometric analysis has the following peculiar feature: it aims to model the
generation of the outcome (i.e., the dependent variable) taking into account the agents
decisions to undertake a treatment. Thus, the outcome is explicitly modelled in terms of its
determinantasspecifiedby[economic]theory(p.18,myemphasis).
Within literature of econometricians and statisticians, instead, another trend can be
recognized. The goal here is to model the stochastic mechanism that generates the datain
their jargon, the data generating process. The specification of the data generating process, in

Humana.MenteIssue10July2009

28

turn, depends on the statistical model, and, consequently, the assumptions of model play a
crucial role. Freedman (2004), for instance, distinguishes between statistical and causal
assumptionsandrequiresinterventionstograntcausalinferences.Thecrucialassumptions,in
hisaccount,arethecausalones,whicheventuallyconsistinassumingthatstructuralequations
unveil the causal mechanism that generate the observed data. This way, however, there isnt
much difference between this methodological argument and the metaphysical arguments
mentioned in the previous section. In fact, under this account the causal interpretation of
structuralequationconsistsinassumingthatthereismechanismbehindbutstaysilentonhow
we come to establish whether there is such mechanism. Holland (1986) goes a step further
and draws a distinction between associational models and causal models, where the former
simply make descriptive claims about conditional distributions and the latter also aim to
quantify the causal effect of a treatment or intervention. Stone (1993), finally, focuses
specifically on the causal assumptions, ranking them from the strongesti.e., covariate
sufficiencytotheweakesti.e.,ignorabletreatmentassignment.
Alltheseargumentscertainlygetrightpartofwhatisatstakebutnot,Icontend,thewhole
story.Mechanismscertainlyplayarolehere.Butwhy?Isitbecauseweassumetheexistence
of a given causal structure that we believe gives rise to the observed distributions? Or is it
because we aim to model a causal mechanism? Assumptions are certainly central too. But
what is exactly their import in justifying the causal interpretation? The story I want to tell
somehow embraces both those views. Notably, I will defend the idea that we have to model
mechanisms paying particular attention to the different types of assumptions made in the
model.
Somehow, the view defended here is midway between the deductivist and inductivist
approachesineconometrics(Moneta2007).Intheformer,causesaregivenbytheeconomic
theory;althoughthereissomedegreeoffreedomastowhateconomictheorytochoose,once
achoiceismade,thisimposestherestrictionsonthemodel.Econometricsisthusreducedto
measuring (statistical) relations between variables, rather then (dis)confirming causal
hypotheses. In the latter, causes are inferred from statistical properties of data alone, by
imposingtothemodelthesimplestcausalstructurethatallowsidentificationamethodology
that strongly resembles presentday graphical models. The view defended here locates in
between those two positions (i) because economic theory has to play a role in the model
buildingandmodeltestingprocessaspartofbackgroundknowledge,butdoesnothavetobe
the sole element to determine the choice of variables and the interpretation of results; (ii)
because statistical analyses, where we let the data speak as much as possibleto echo
Moneta(2007,p.119)independentlyofanyaprioryeconomictheorycanalsoplayarolein
themodelbuildingprocessbutdonotexhaustthecausalanalysisitself.

4.CAUSALECONOMETRICMODELS

The arguments offered next hinge upon the distinction and comparison of two classes of
models: associational models and causal models. It is commonly agreed that associational
modelsjustmakedescriptiveclaimsaboutconditionaldistributions,whereascausalmodels,in
addition, aim at evaluating statistical relevance relations to quantify the causal effect of the
explanatoryvariablesontheresponsevariable.However,thiscannotbethewholestory,since
itstillstaysunclearhow,incausalmodels,correlationssuddenlyturnintocausalrelationsand
probabilisticdependenciesintocausaldependencies.Inthefollowing,Iarguethatdifferences
between associational models and causal models can be identified at three levels: (i)
FedericaRussoCausalArrowsinEconometricModels

29

background knowledge, (ii) assumptions, and (iii) methodology. The difference between
associationalandcausalmodelsisschematicallyrepresentedintable1.

AssociationalModels CausalModels
StatisticalAssumptions Background
knowledge/Causalcontext
StatisticalAssumptions
ExtraStatisticalAssumptions
CausalAssumptions
Hypotheticodeductive
Methodology

Table1:AssociationalModelsvs.CausalModels

ASSOCIATIONALMODELS

The goal of associational models is to describe how a given variable (the dependent variable)
variesdependingonothervariables(theindependentones).Associationalmodelsaretypically
usedtomakeexploratoryanalysesofdatainordertoseewhatcorrelationsbetweenvariables
hold. Background knowledge does not play any particular role in associational models and
variables do not play specific causal roles. Associational models rest on a number of standard
statistical assumptions. Leaving technicalities aside, we suppose that the model have some
characteristics (usually, linearity and normality) such that it is easy to manipulate, easy to
estimate statistically, and the resulting estimates have nice properties. We also assume that
variablesaremeasuredwithouterror,thattheerrorsarenotcorrelatedwiththeindependent
variables. When these assumptions are satisfied, the conditional distribution correctly
describes how variables covary. But at this stage there is no necessary causal information
conveyed by the parameters, nor it is generally valid to give the regression coefficients the
causal interpretation. Also, dependencies described in associational models are symmetric,
whichdoesnttellmuchaboutcausation,infact.
Tointerpretthecoefficientscausallymeansthatthecoefficientsappearingintheequations
measure the effect on the dependent variable caused by a change in the independent
variable(s).Togobeyondthedescriptiveclaimsweneed(i)accurateknowledgeofthecausal
context, (ii) further assumptions, and (iii) a methodology to confirm/disconfirm causal
hypotheses.

CAUSALMODELS

Causalmodelsareequippedwithamuchricherapparatusthanassociationalmodelssimplydo
not have. This involves: (i) background knowledge, (ii) further assumptions, and (iii)
hypotheticodeductivemethodology.

Humana.MenteIssue10July2009

30

BACKGROUNDKNOWLEDGE

Background knowledge certainly includes the economic theory but also includes general
knowledge of the sociopolitical context, knowledge of demographic characteristics of the
populationunderinvestigation,orinstitutionalknowledge(i.e.,knowledgeofthefunctioning
andproceduresofaninstitutionsuchastheCentralBank).Thisiswhatmanysocialscientists
usuallycallfieldknowledge.Insomecases,notablywhendealingwithdisciplinesthatneedto
include biological variables, background knowledge may also include knowledge of the
physicalbiologicalphysiological mechanism. This is the case, for instance, in epidemiology,
where one of the objectives is to understand how health variables affect socioeconomic
variables, or the other way round. Well established scientific theories also belong to filed
knowledge.Nodoubtevidenceisimportantforbackgroundknowledge.Inparticular,evidence
of the same putative causal relations operating in different populations may justify further
research, or evidence about different causal relations operating in other populations may
justify a different modelling strategy. Thus, the use of different/similar data and/or models
alsobelongstobackgroundknowledge.

ASSUMPTIONS

Let us now turn to the assumptions needed in causal models. Beside standard statistical
assumptions also made in associational models, causal models have two extra sets of
assumptionsthatIshallcalltheextrastatisticalassumptionsandthecausalassumptions.
ByextrastatisticalassumptionsImeanallthoseassumptionsthateitherarenotrelatedto
the statistical properties of the distributions or have causal meaning but are not subject to
statistical test. Among these assumptions we find, for instance, the direction of time, the
directionofthecausalrelations,thecausalmechanism.
BycausalassumptionsImeanallthoseassumptionsthataresubjecttostatisticaltestand
thatcontributetointerpretcausallytherelationsbetweenvariablesinthemodel.Twocausal
assumptionsareparticularlyimportant:covariatesufficiencyandnoconfounding.Covariate
sufficiency states that the independent variables are all the variables needed in order to
explainthevarianceofthedependentvariable.Theassumptionofnoconfoundingthenplays
acomplementaryroleandmeans,simplyput,thatweruledoutotherfactorsliabletoscreen
off the variables we took into account. These assumptions are causal because the causal
interpretationcruciallydependsontheminthefollowingsense.
Causal modelling rests on the socalled closure of the system. This assumption says that
thesystemunderanalysisisnotsubjecttoexternalinfluencesandthuscanbeseparated,soto
speak, from the larger web of interrelations in which it is located. Thanks to this assumption
we can, at least in principle, detect the causes acting in the system under investigation. The
assumption of the closure is related to covariate sufficiency and noconfounding thus: if we
put those three pieces of the puzzle together(weak) closure, covariate sufficiency, and no
confoundingwe end up with a causal model that describes a (quasi) hermetic mechanism
whereallandonlythechosenexplanatoryvariablesplayarole.

HYPOTHETICODEDUCTIVEMETHODOLOGY

In order to tell causal stories, econometric models have to employ a hypotheticodeductive


methodology. Simply put, hypotheticodeductivism is the view according to which scientists
FedericaRussoCausalArrowsinEconometricModels

31

first formulate hypotheses and then test them by seeing whether or not the consequences
derived from the hypotheses obtain. Popper (1959), who developed the HD methodology,
wasmotivatedbytheneedofcreatingascientificmethodinanoninductiveway.However,in
causalanalysis,hypotheticodeductivismtakesaslightlydifferentfacetspecificallyconcerning
deduction and, mostly, is not strictly falsificationist in character. Yet, it does borrow from the
Popperianaccounttheprimaryroleofthehypothesisformulationstage.Ishallgetbacktothis
point later. According to the HD methodology, model building and model testing essentially
involvethreestages:

1. formulatethecausalhypothesis;
2. buildthestatisticalmodel;
3. draw consequences to conclude to the empirical validity or invalidity of the causal
hypothesis.

The causal hypothesis to put forward for empirical testing does not come from a tabula rasa,
but emerges from a causal context, namely from background knowledge. The hypothesis
formulation stage may also be informed by results of associational models as they suggest
interestingcorrelationstosubmittofurtherscrutiny.Therestofthemodelbuildingandmodel
testing process is meant to (dis)confirm the causal hypothesis. In practice, this requires
buildingthestatisticalmodel,andthendrawingconsequencesfromthehypothesisinorderto
testthehypothesisagainstempiricaldata.
Itisthroughtheestimationofthestatisticalmodelandthroughhypothesistestingthatwe
will (dis)confirm the causal hypothesis. If the model is correctly estimated and fits the data,
and if certain conditions are satisfied (notably, exogeneity and invariance) the hypothesized
causal link is accepted, rejected otherwise. The hypotheticodeductive structure of causal
modelling is thus apparent: a causal relation is first hypothesized and then put forward for
empiricaltesting.Thatittosay,thecausalhypothesisisnotdirectlyinferredfromthedata,as
isthecasewithinductivestrategies,butacceptedorrejecteddependingontheresultsoftests
andonbackgroundknowledge.
As anticipated above, hypotheticodeductivism in causal modelling does not involve
deductionstrictosensu,butinvolvesaweakerinferentialstepofdrawingconsequencesfrom
thehypothesis.Oncethecausalhypothesisisformulatedoutoftheobservationofmeaningful
covariations between the putative cause(s) and the putative effect and out of background
knowledge,wedonotrequiredatatobeimpliedbythehypothesisbutjustthatdataconform
toit.Here,conformmeansthattheselectedindicatorsandrelationsamongthemadequately
representtheconceptualvariablesappearinginthecausalhypothesisandtherelationsamong
them. Thus, this way of validating the causal hypothesis is not, strictly speaking, a matter of
deduction, but surely is, broadly speaking, a deductive procedure. More precisely, it is a
hypotheticodeductive procedure insofar as it goes the opposite direction of inductive
methodologies: not from rough data to theory, but from theories to data. For a discussion of
theHDmethodology,seealsoLittle(1998,ch.9),Cartwright(2007,ch.2)andRusso(2008,ch.
3and2009).
The two main tests of model parameters concern invariance and exogeneity. Invariance
tests check whether the relation between two variables is stable across different
environments, for instance under intervention or across different panels of the data set. The
invariance condition acts as a constraint on the causal relation. This means that invariance
doesnotconstitutecausation,butisafeaturewerequireinordertointerprettheparameters
ascausalparameters.Theideabehindexogeneityisthefollowing.Exogeneityisaconditionof
separabilityofinference.Givenadatasetcontaininganumberofvariablesofinterest,thegoal

Humana.MenteIssue10July2009

32

of modelling will be to decompose an initial joint probability distribution of all the variables
into a sequence of marginal and conditional probability distributions. This (statistically)
corresponds to identify and specify the variables playing a specific causal role and the way
those causal variables interrelate. Background knowledge is of primary importance in guiding
thechoiceofthemarginalconditionaldecomposition.

ACASESTUDY:HEALTHANDWEALTHINELDERLYAMERICANS

A study published in the Journal of Econometrics (Adams et al 2003) examined the possible
causal links between health and socioeconomic status (SES) in elderly retired Americas. The
links between health and SES have been the object of numerous studies. A survey of the
literature shows that a significant association holds for a variety of health variables and
alternative measures of SES. Also, much discussion has been devoted to the mechanisms
behindthisassociation,andyettherehavebeenrelativelyfewstudiesthatallowcausalpaths
tobedefinitivelyidentified.Thisstudyadoptsaparticularmethodologyandappliesstatistical
methods to test for the absence of direct causal links from SES to health, and from health
conditionstowealth.
Adamsetal.examineasamplefromthepopulationofelderlyAmericansaged70andolder,
inparticulartheyexaminewhetherinnovationsinhealthandwealthinapanelareinfluenced
by their past values and by the past values of other relevant variables. Data come from three
surveys performed between 1993 and 1998, providing information about health, wealth, and
demographicinformationofindividuals.Afirstanalysisofdatashowsasignificantassociation
between health and SES. On the basis of background knowledge, including previous studies
and knowledge of the sociodemographic conditions of elderly Americans, Adams et al.
hypothesize that the links between health and wealth conditions can be represented as
(roughly)sketchedinfig.1.

Fig.1Possiblecausallinksbetweenhealthandwealth

In the graph, arrows represent causal relations; when the line is dotted, the causal relation is
absent. Thus, Adams et al., based on background knowledge and preliminaries analyses of
data, expect that no significant causal link be found from health history of retired people to
theirpresentwealthstatus.Onthecontrary,theyexpectthatsocioeconomichistory,through
theircurrentwealthstatus,doesaffecttheircurrenthealth.
Following the approach of Granger (1969), two specific causal hypotheses are put forward
forempiricaltesting:(i)thereisnocausallinkfromhealthtoSES,and(ii)thereisnocausallink
FedericaRussoCausalArrowsinEconometricModels

33

from SES to health. Those two hypotheses are evaluated by means of noncausality test and
invariance tests. Very informally, the structure of noncausality tests, in the approach of
Granger(1969),isasfollows:thepresentstateofavariable(say,currenthealthofindividuals
inthesample),causallydependsonallitshistory(thatisallthehealthhistoryofindividualsin
the sample) and on the history of other relevant variables (say, SES of individuals in the
sample). Thus, if SES and health turn out to be independent, then we say that SES does not
Grangercausehealth.Invariancetests,instead,aimtocheckwhethermodelparametersstay
stableacrossdifferentpanels.
In testing the significance of SES on a number of health outcomes, the authors draw the
following conclusions: SES is, usually, strongly associated with health; however, the link is
weakerwhentheycontrolforprevioushealthproblems;also,theassociationbetweenSESand
mentalandchronicillnessisstrongerthantheassociationbetweenSESandacuteandsudden
outset health conditions. Adams et al. suggest that those findings reflect the differential
coverageofMedicare;infact,insurancestypicallydonotcovermentalillness.
This simple presentation of the case study already illustrates the hypotheticodeductive
methodologyused.Lookatwherecausalityis;causalityisnotatmerelytheoutput,pulledout
of the statistical model. Instead, the two causal hypotheses are put forward for empirical
testing within the model, and then accepted or rejected depending on the results of test and
onwhethertheresultsarecongruentwithbackgroundknowledge.Toillustratethefeaturesof
causalmodelsdiscussedabove,letusdiscusseachfeatureseparately.
Background knowledge and causal context. In Adams et al. work, it includes previous
studies on the relations between health and wealth. In particular, the association between
health and wealth holds for a variety of health variables and alternative measures of SES.
Therefore,acausalinterpretationofthosecorrelationsisplausibleandworthtesting.
Causal hypothesis. In the health and wealth case study the causal hypothesis is split into
two:(i)isnocausallinkfromhealthtoSES,and(ii)thereisnocausallinkfromSEStohealth.In
ordertoevaluatethecausalimpactofSESonhealthandviceversa,Adamsetal.buildamodel
following the approach of Granger and perform noncausality and invariance tests on the
causalhypothesesspecified.
Statistical assumptions. The statistical analysis fits the approach of Granger, which is
essentiallybasedonregressionmethodsandwherestandardstatisticalassumptionshold.
Extrastatistical assumptions. In this case study two extrastatistical assumptions are at
stake: causal priority and causal ordering. In fact, in Grangerapproach it is explicit that the
historyofvariablecausallydeterminesthecurrentvalueofthevariable.Accordingly,statistical
tests are about whether heath histories determine current wealth and whether wealth
historiesdeterminecurrenthealth.
Causal assumptions. For the most part, Adams et al. discuss the assumption of covariate
sufficiency, that is all the relevant explanatory variables have been included. Another
important assumption concerns and structure of the causal relation: by adopting Grangers
approach,instantaneouscausalityisruledout,atleastforsufficientlybrieftimeintervals.
Tests. Within a Grangercausality framework, Adams et al. test model parameters for
invariance across different panels. More explicitly, they test (i) whether the conditional
distribution current health given wealth history and (ii) whether the conditional distribution
current wealth given health history hold stable across different panels analyzed. The
invariance conditions (largely) fails in the first case, i.e. variations in health generally do not
correspond to variations in wealth, hence, the hypothesis of no causality is accepted. On the
otherhand,theconditionaldistributioncurrentwealthgivenhealthhistoryresults(relatively)
stableandthereforethehypothesisofnocausalityisrejected.

Humana.MenteIssue10July2009

34

Hypotheticodeductive methodology. Adams et al. are clearly using a hypothetico


deductive methodology in their analysis. They first formulate the causal hypotheses out of
background knowledge and then, iteratively, build the model and perform statistical tests to
confirmordisconfirmthecausalhypotheses.Backgroundknowledgeandcausalcontextplaya
primary role at each stage: from the choice of variable until the evaluation of results. A clear
identificationofthestepsintheprocessofmodelbuildingandmodeltestingallowstoidentify
possibleleaksorpointsofdisagreementbetweenscientists.Whilstthereisasharedconsensus
on the background of this study (see the commentaries in the same volume of Adams et al.s
paper), a number of commentators suggested that a structural approach would be better
suited to study the links between health and wealth (Hausman (2003), Heckman (2003),
Hoover (2003)). The particular form of the invariance tests is also debated. For instance,
Hoover (2003) agrees that invariance is an important indicator for the presence of a causal
relationships, but doubts on the cogency of the evidence used by Adams et al. Analogously,
remarks on the chosen definition and tests of noncausality also come from Florens (2003).
Adda et al (2003) object to their suggestion that the observed links between SES and health
aretheeffectoftheaccesspeoplehavetohealthinfrastructures.ThereasonisthatAddaetal.
apply the very same methodology to two different data sets, one from UK and one from
Sweden, and get very similar results. However, since medical systems are very different in
those countries, coverage of Medicare cannot be what produces these results. As a
consequence,policiesinterveninginthatdirectionwouldbe,inAddaetal.sview,misplaced.

EPISTEMOLOGICALMORALS

Let me know state more precisely the epistemological morals to draw from the arguments
given above. I undertook this path of investigation to overcome the dissatisfaction with
available arguments that either assume a causal structure instead of saying how we come to
establish it, or with methodological arguments that stop too early in praising the role of
assumptions in causal models. In order to make causal claims we need different types of
assumptions. The distinction I drew between statistical, extrastatistical, and causal
assumptions is meant to make clear the import of each of them towards establishing causal
relations. For instance, extrastatistical assumptions such as causal ordering, usually come
from knowledge of the causal contexts, are typically not tested in the model, and yet
participateincausalattributionbecausetheyserve,forinstance,tofixwhichvariablesarethe
causesandwhichvariablesaretheeffects.Nevertheless,themainpointofthisdistinctionisto
show that it is not the case that the stricter the statistical tests the greater our confidence in
the causal interpretation. Causal relations are not pulled out of statistical hats. The hat, so to
speak, is itself causal. The conceptual hypothesis is in fact a causal claim the evaluation of
which demands for empirical testing. In other words, causality is a matter of confirmation, or
borrowing the statistical vocabulary, it is a matter of accepting or rejecting a given causal
hypothesis.
Let me linger more on this causal hat. Model building and model testing take place in a
causal context. Background knowledge plays a major role in the preliminary analyses of data,
intheselectionofvariables,andintheformulationofcausalhypotheses.Itisalsoonthebasis
of background knowledge that the choice of testing for invariance some conditional
distributions, but not others, relies. It is, finally, on the basis of background knowledge that
results are evaluated. The importance of background knowledge, however, goes beyond the
choice of variables and for the formulation of the causal hypothesis. It is also important in
order to rule out highly implausible causal relations before starting the whole machinery of
FedericaRussoCausalArrowsinEconometricModels

35

model building and model testing. In manipulationist accounts la Woodward (2003), the
relation between the increasing number of storks and the increasing number of births is not
causal because it would turn out not to be invariant under intervention; namely, no
manipulationonthevariablenumberofstorkswouldleadtochangesinthevariablenumber
of births. Nevertheless, should we go as far as testing for invariance in cases where we can
excludethatthecorrelationiscausalonthebasisofbackgroundknowledge?
Intheliterature,thecrucialroleofthecausalcontextandoftheconceptualframeworkhas
beenstressedbymanyeminentscientistsandphilosophersjuttonameafew:Fisher(1925),
Haavelmo (1944), Kendall and Stuart (1961), Suppes (1970), Humphreys (1989). And yet, this
emphasishasntbeengiventheproperimportance.Ithasoftbeentakenasaplatitudeabout
causalmodellingcausalmethodsrelyonbackgroundknowledgeinanobvioussenseorasa
hidden challengewhat in fact background knowledge taught us that we did not already
know? In other words, if model building and model testing are largely based on background
knowledge,howdowegobeyonditandgaincausalknowledge?
What allows us to go beyond background knowledge is the hypotheticodeductive
methodology of causal models. This is a dynamic methodology that allows a va et vient
betweenestablishedtheoriesandestablishingtheories.Establishedscientifictheoriesare(and
oughttobe)usedtoformulatethecausalhypothesisandtoevaluatetheplausibilityofresults
on theoretical grounds. But causal models also participate in establishing new theories by
generalizing results of single studies. This reflects the idea that science is far from being
monolithic, discovering immutable and eternal truths. If the model fits the data, model
parameters are (sufficiently) invariant and the relations are congruent with background
knowledge,thenwecansay,tothebestourknowledge,thatwehituponacausalrelation.But
what if one of these conditions fails? A negative result may trigger further research by
improvingthemodellingstrategies,orbycollectingnewdata,thusleadingtonewdiscoveries
that,perhaps,discardbackgroundknowledge.
Alastcorollaryofthisepistemologicalregardoneconometricmodelsisthat,byandlarge,
causalmodelsmodelmechanisms.Asmentionedabove,hypotheticodeductivecausalmodels
aim to (dis)confirm a causal hypothesis; the causal hypothesis, however, states something
more than simply a causal relationits about a causal mechanism. Let me develop this idea
further.
What do causal models do? Causal models model the properties of a social system. In
particular, they model the relations between the properties or characteristics of the system,
whicharerepresentedbyvariables.Tomodelthepropertiesofasocialsystemmeanstogive
the scheme, or the skeleton, of how these properties relate to each other. However, this
causalmechanismisnotmodelledintermsofspatiotemporalprocessesandinteractionsla
Salmon(1984and1990)butisstatisticallymodelled.Conceptstypicalofstatisticalcausality
e.g.,statisticalrelevance,comparisonofconditionalprobabilities,andscreeningoffareused
inordertoidentifythetypesofrelationshipsthatholdamongthevariablesofinterest.Thisis
howthecausalmodelmodelsthecausalmechanismgoverningthesocialsystem.Inparticular,
causal models seek to uncover stable variational relations between the characteristics of the
system.
Thisviewofmechanismsisnotperfectlyinlinewithmorephysicalviewssuchasthoseof
Glennan (1996), Machamer et al. (2000), Bechtel and Abrahms (2005), or Craver (2007).
However, it is not in opposition either. Craver, for instance, echoing the definition given by
Machameretal(2000),conceivesofmechanismsasasetofentitiesandactivitiesorganized
such that they exhibit the phenomenon to be explained (Craver 2007, p.5), and calls it a
skeletal description. Such characterization of mechanisms, I believe, is broad enough as to
accountformechanismsinvariousdomains.Shouldwetaketheentitiestobeneuronsandthe

Humana.MenteIssue10July2009

36

activities neurotransmitters release, the above skeletal description will well fit neuro
mechanisms.Shouldwetakeentitiestobesociodemoeconomicvariables,andactivitiestobe
theirinfluenceonothersociodemoeconomicvariables,theaboveskeletaldescriptionwillfit
equallywellsocialmechanisms.Thedegreeofphysicalrealityonewishestogivetoentities
andactivitiesmayleadtodifferentaccountsnotably,toadifferentontologicalcommitment
to the existence of mechanisms. In the social contexts we do not necessarily need to
endorsetheviewthatelementsandrelationsshouldalwayshavephysicalcounterparts.This
would lead us into debates concerning social ontology and methodological individualism,
which,ofcourse,arefarbeyondthescopeofthepresentdiscussion.

5.CONCLUDINGREMARKS

Econometrics applies statistical methods to study economic relations. By means of structural


models, econometricians typically account for the response variable in terms of a number of
explanatory variables. Under what conditions can econometric models be given a causal
interpretation? This paper defended the view that (i) a proper use of background knowledge,
(ii) statistical, extrastatistical, and causal assumptions, and (iii) hypotheticodeductive
methodology provides sufficient conditions for the causal interpretation. On the contrary,
associational models do not allow any causal interpretation as they are just equipped with
statistical assumptions. This goes against the views that causal structures are simply assumed
and are that which is responsible for the observed correlations, and against the views that
bootstrap causal relations from stringent statistical tests alone. The causal interpretation is
instead a matter of confirmation. In hypotheticodeductive terms, the goal is to confirm (or
disconfirm) a hypothesis that is explicitly causal. Nancy Cartwright famously expressed this
idea with the motto No causes in, no causes out. In spite of appearance, the account is not
circular,andifthereisacircleatall,itisvirtuousandnotvicious.Toputcausalityinthemodel
means that the evaluation of causal relations is relative to some conceptual and formal
framework; in this framework we find background knowledge, causal assumptions, and
hypotheticodeductivemethodology.Theaccountofferedgivessufficientconditionsthatallow
to interpret econometric models causally. It is worth emphasize that such account deeply
depends on the epistemological perspective I adopted. By focusing on how we come to
establish causal relations rather than on what causation in fact is, I implicitly endorsed an
epistemic view of causality very much in line with the view defended by Williamson (2005).
According to this view, results of econometric models are not immutable and eternal causal
truths, but justified beliefs, which are fallible. This fact is indeed mirrored in the hypothetico
deductivemethodologyofcausalmodelswhich,beingdynamicandflexible,allowsustorevise
ourcausalbeliefsbybuildingandtestingnewcausaleconometricmodels.

Acknowledgements. Very helpful and stimulating comments came from the fellows at the
Center for Philosophy of Science (University of Pittsburgh)Claus Beisbart, Erik Curiel, Laura
Felline, Ulrich Krohs, Flavia Padovani, Chris Pincock, John Norton (director). I am indebted to
AlessioMonetaformakinganumberofvaluablesuggestions,especiallyconcerningtheroleof
economic theory in causal analysis. Financial support from the FSRFNRS is also gratefully
acknowledged.

FedericaRussoCausalArrowsinEconometricModels

37

BIBLIOGRAPHY

AdamsP.,HurdM.,McFaddenD.,MerrillA.,Ribeiro,T.(2003)Healthy,wealthy,andwise?
Tests for direct causal paths between health and socioeconomic status. Journal of
Econometrics,112,356.Withdiscussion.
Adda J., T. Chandola, and M. Marmot (2003) Socioeconomic status and health: causality
andpathways.JournalofEconometrics,112,5763.
BechtelW.andA.Abrahamsen(2005),Explanation:amechanistalternative,Studiesinthe
HistoryandPhilosophyoftheBiologicalandBiomedicalSciences,36,421441.
CartwrightN.(1989)Naturescapacitiesandtheirmeasurements.Oxford:ClarendonPress.
Cartwright N. (2002) Against modularity, the causal Markov condition, and any link
between the two: comments on Hausman and Woodward. British Journal for the
PhilosophyofScience,53(3),411453.
Cartwright N. (2007a) Hunting causes and using them: approaches in philosophy and
economics.Cambridge:CambridgeUniversityPress.
CraverC.F.(2007),Explainingthebrain.Mechanismsandthemosaicunityofneuroscience.
Oxford:OxfordUniversitypress.
EngleR.F.,HendryD.F.,RichardJ.F.(1983)Exogeneity.Econometrica,51(2),277304.
Fisher R. A. (1925) Statistical methods for research workers. London: Oliver and Boyd.
http://psychclassics.yorku.ca/Fisher/Methods/.Accessed14March2008.
FlorensJP(2003)Sometechicalissuesindefiningcausality.JournalofEconometrics,112,
127128.
Florens JP, M. Mouchart and J.M. Rolin (1980), Rductions dans les Expriences
Baysiennes Squentielles, paper presented at the Colloque Processus Alatoires et
Problmes de Prvision, held in Bruxelles 2425 April 1980, Cahiers du Centre
dEtudesdeRechercheOprationnelle,23(34),353362.
FreedmanD.A.(2004)Onspecifyinggraphicalmodelsforcausation,andtheidentification
problem.EvaluationReview,26,26793.
GlennanS.(1996)Mechanismsandthenatureofcausation.Erkenntinis,44,4971.
Granger C. W. J. (1969) Investigating causal relations by econometric models and cross
spectralmethods.Econometrica,37(3),424438.
Granger,C.W.J.(2003)Someaspectsofcausalrelationships.JournalofEconometrics,112,
6971.
HaavelmoT.(1944)Theprobabilityapproachineconometrics.Econometrica,12,iiivi+1
115.
HausmanD.(1998)Causalasymmetries.Cambridge:CambridgeUniversityPress.
Hausman D., Woodward, J. (1999) Independence, invariance, and the causal Markov
condition.BritishJournalforthePhilosophyofScience,50,521583.
Hausman D., Woodward, J. (2004) Modularity and the causal Markov condition: a
restatement.BritishJournalforthePhilosophyofScience,55,147161.

Humana.MenteIssue10July2009

38

Hausman J. A. (2003) Triangular structural model specification and estimation with


applicationtocausality.JournalofEconometrics,112,107113.
HeckmanJ.(2003)Conditioning,causality,andpolicyanalysis.JournalofEconometrics,112,
7378.
HeckmanJ.(2008)Econometriccausality.IZADPNo.3425.
http://www.nber.org/papers/w13934accessed11April2009.
Holland P. W. (1986) Statistics and causal inference. Journal of the American Statistical
Association,81,945970.
HooverK.D.(2001)Causalityinmacroeconomics.Cambridge:CambridgeUniversityPress.
Hoover K. D. (2003) Some causal lessons from macroeconomics. Journal of Econometrics,
112,121125.
Hoover K. D. (forthcoming) Probability and Structure in Econometric Models. In The
Proceedings of the 13th International Congress of Logic, Methodology and
PhilosophyofScience.London:King'sCollegePublications.
HumphreysP.(1989)Thechancesofexplanation.Princeton:PrincetonUniversityPress.
KendallM.G.,&Stuart,A.(1961)Theadvancedtheoryofstatistics.Vol.2,London:Griffin&
Co
Kinkaid, H. (1996), Philosophical Foundations of the Social Sciences, New York: Cambridge
UniversityPress.
Little D. (1990) Varieties of social explanations: an introduction to the philosophy of social
science.Boulder:WestviewPress.
Little D. (1998) Microfoundations, method and causation. On the philosophy of the social
sciences.NewBrunswick,NJ:TransactionPublishers.
MachamerP.L.DardenandC.F.Craver(2000),ThinkingaboutMechanisms,Philosophyof
Science,67,125.
MonetaA.(2007)Mediatingbetweencausesandprobabilities:theuseofgraphicalmodels
in econometrics (pp. 109130) In F. Russo and J. Williamson (eds), Causality and
ProbabilityintheSciences.London:CollegePublications.
Mouchart M. and Russo, F. (2010) Causal explanation: recursive decompositions and
mechanisms. In P. McKay Illari, F. Russo, and J. Williamson (eds) Causality in the
Sciences.OxfordUniversityPress.
MouchartM.,RussoF.,WunschG.(2008)Structuralmodelling,exogeneityandcausality.In
H. Engelhardt, H. P. Kohler, A. Prskwetz (Ed.), Causal analysis in population studies:
concepts,methods,applications(Chapter4)Dordrecht:Springer.
Pearl, J. (2000) Causality. Models, reasoning, and inference. Cambridge: Cambridge
UniversityPress.
Peto R., Darby, S., Deo, H., Silcocks, P., Whitley, E., Doll, R. (2000) Smoking, smoking
cessation,andlungcancerinUKsince1950.BritishMedicalJournal,321,323329.
Popper,K.(1959)TheLogicofScientificDiscovery.Hutchinson,London.
FedericaRussoCausalArrowsinEconometricModels

39

Reiss,J.(2007a)Timeseries,nonsensecorrelations,andtheprincipleofcommoncause(pp.
179196)InF.RussoandJ.Williamson(eds),CausalityandProbabilityintheSciences.
London:CollegePublications.
Reiss J. (2007b) Do we need mechanisms in the social sciences? Philosophy of the Social
Sciences,37(2),163184.
Reiss J. (2008) Error in economics: towards a more evidencebased methodology. London:
Routledge.
RussoF.(2008)CausalityandCausalModellingintheSocialSciences.MeasuringVariations.
NewYork:Springer.
RussoF.(2009)ExplainingCausalModelling.Or,WhataCausalModelOughttoExplain.In
ProocedingsoftheSILFSConference,Milan810October2007.
SpirtesP., Glymour, C., Scheines, R. (1993) Causation, prediction, and search. New York:
Springer.
Steel D. (2008) Across the boundaries. Extrapolation in biology and social science. New
York:OxfordUniversityPress.
Stone R. (1993) The assumptions on which causal inferences rest. Journal of the American
StatisticalAssociation,55(2),455466.
SuppesP.(1970)Aprobabilistictheoryofcausality.Amsterdam:NorthHolland.
Williamson J. (2005) Bayesian nets and causality. Philosophical and computational
foundations.London:OxfordUniversityPress.
WoodwardJ.(2003)Makingthingshappen:atheoryofcausalexplanation.Oxford:Oxford
UniversityPress.

Humana.MenteIssue10July2009

40




41

WhatcanNeuroscienceoffertoEconomics?

MatteoColombo
*

M.Colombo2@sms.ed.ac.uk

ABSTRACT

The specific regions in the brain that are active when some behaviour is observed is a kind of
information that may be interesting for neuroscientists, but how could it be fruitful for economic
theory? The thesis defended in the essay is that the brain matters to prediction.
By using the Ultimatum Game as a benchmark, it is argued that if the goal of a model of human
behaviouristoyieldgoodpredictionsaboutimportantclassesofchoices,thenmodelsthatincorporate
neurobiologicalvariablesmayhavesomeadvantagesoveralternativemodels.Theessaycomprisestwo
parts.PartIfirstanalysestheUltimatumGameandillustratessomeofitsexperimentalresults.Then,it
evaluates in detail the merits and shortcomings of Cristina Bicchieris model based on social norms. It
centres on the predictive power of the model, and articulates some challenges it faces. Part II begins
with a review of some neurobiological findings which suggest a different approach to construct
predictive models of human behaviour. Drawing upon these findings, it gives some reasons why the
predictions of a neurobiologicallyinformed model seems to have some advantage over those of
Bicchierismodel.Acriticaldiscussionofthethesisagainstpossibleobjectionsterminatestheessay.The
conclusion follows that one way in which the study of the neurobiological foundations of decision
makingmightbefruitfulforeconomicmodellingisinenhancingthepredictivequalityofitsmodels.

INTRODUCTION

Thespecificregionsinthebrainthatareactivewhensomebehaviourisobservedisakindof
informationthatmaybeveryinterestingforneuroscientists.Butwhatdoesitaddtoeconomic
theory?Theclaimdefendedhereisthatthetypeofknowledgeofbrainprocessesofferedby
neurosciences matters to prediction (see also Camerer 2007). To the extent I am right, the
attempt to integrate evidence, concepts and tools from the fields of economics, psychology
andneurosciencewithinthenewdomainofneuroeconomicswillturnouttobearealizationof
the methodological ideal described by Milton Friedman in The Methodology of Positive
Economics. There, Friedman advocates a requirement of predictive success for judging a
positive scientific theory: The ultimate goal of a positive science is the development of a
theory or hypothesis that yields valid and meaningful (i.e. non truistic) predictions about
phenomenanotyetobserved.(Friedman1953,p.7)
An important cautionary note right from the beginning: It may sound that I focus on
prediction in order to avoid the real problems e.g. issues about explanation, and
understanding. I would like to reply to this charge by giving it a different twist. One way to
readmyclaimisthathoweveryouwanttoexplain,orunderstandhumanbehaviour,amodel
that takes into account neurobiological parameters seems to have an advantage over
competitorsforpredictions.Ifweconsidertheimportancethatpredictionshaveinourlives,it
may be easier to acknowledge that the choice to limit a research on prediction is still
worthwhile. Prediction serves at least two crucial goals: one pragmatic, the other epistemic

*
DepartmentofPhilosophy,UniversityofEdinburgh,UK

Humana.MenteIssue10July2009

42

(see Rescher 1997). On the one hand, predictions are necessary to interact successfully with
our environment. E.g., predicting that if we jump off the top of the Cathedral of Milan, nasty
consequencesforourhealthwillbeextremelylikelytoensue,wemayprefertotakethestairs
togodown.Ontheotherhand,predictionisatestofscientifictheories.Itisacommonview
that in assessing the confirmation or evidential support of a hypothesis, we must take into
accountespecially(andperhapsexclusively)thepredictivesuccessorfailureofitspredictions
(Musgrave1974,p.2).
Before spelling out the two criteria that will orient my assessment of the comparative
merits and problems of the predictions given by concurrent models, it is worth making clear
whatitismeantbyprediction.FollowingForster(2008),[t]hetermpredictionisalwaysused
to refer to the diction of past, present, and future events, the diction of a claim that
previouslywehadnoreasontobelieve.Hence,agoodpredictionisnotnecessarilytoreferto
the future. I take a good prediction to be one that is both secure and informative. A secure
prediction is based on reliable, wellevidentiated grounds. Because of this, it is likely that the
predictionturnsouttobecorrect.E.g.,thepredictionthattherewillbeafullmoonwithinthe
next thirty days is secure. The more adjustable parameters a model has, the more secure its
predictions,butthegreatertheriskofaccommodation.Accommodationisoneoftherisksofa
secure prediction. A model accommodates the data when it is merely consistent with them.
The typical case of accommodation is when a set of data is deduced from the model (hence,
modelanddataareconsistent),andthesamesetofdatawasusedintheconstructionofthe
model.E.g.,amodeloftheform(H&Y)doesnotgenuinelypredictthatY:ItaccommodatesY.
A good prediction is not trivial, is informative. Informative predictions are not vague,
preferablytheyarequantitativelyaccurate.E.g.,thepredictionthatthecometHaleBopphad
itsclosestapproach(atadistanceof1.315AU)toEarthonMarch22,1997isinformative.
With these conceptual tools at hand I can now make clear the structure of my argument.
TheUltimatumGame(UG)ismybenchmarkforevaluatingthepredictivequalityofamodel.I
begin by detecting an anomaly
1
in the UG. Standard gametheoretic prediction is at loss in
thefaceofthebehavioursofpeopleplayingtheUG.Thus,ifwewanttosticktothepredictive
successrequirement,wemustreviseourmodel.Severalenrichedmodelshavebeenproposed.
Some of them, because of their flexibility, seem promising to predict well. However, I argue
thatthisflexibilitycomesatsomecost:Becauseitisproblematictoobtainareliablemeasure
of the adjustable parameters they contain, these models are particularly subjected to merely
accommodate the evidence. Thats the main reason why we should try to develop a model
with adjustable parameters flexible enough to predict the variety of anomalies observed in
the UG, and specific enough to figure out how to reliably measure them so that the model
enablesustogivepredictionsthatarebothinformativeandsecure.Iarguethatamodelthat
incorporates neurobiological variables seems to satisfy these requirements. However,
neurosciences offer no panacea. I discuss some of the limits of the type of neuroscientific
resultsthatbackmyargument,andIprovidepositivesuggestionsonhowtheseshortcomings
mightbetackled.
If my argument is correct, it follows that if we want to take Friedmans advice seriously,
thenwehavegoodreasonstotrytoincorporateneurobiologicalparametersinourmodelsof
decisionmakingineconomicenvironments.


1
BorrowingRichardThalerswords:ananomalyisanempiricalresultifimplausibleassumptionsare
necessarytoexplainitwithinthe[rationalchoice]paradigm(Thaler1988,p.195).

MatteoColomboWhatcanNeuroscienceoffertoEconomics?

43

Theessayisorganizedintwoparts.
The first part is divided into two sections. Section I explains why I believe that the UG is
interestingandworthstudying.IdescribetheUG,itsgametheoreticanalysis,andhowpeople
play the game. Section II introduces the concept of enriched models. These models try to
accountfortheactualbehaviourofthepeoplewhoplaytheUGbyappealingtosuchconcepts
as fairness, warm glow, envy, social norms. I restrict my analysis to one of these
proposals.Afterhavingmotivatedmychoice,Ianalyzeindetailhowthenormbasedapproach
defended by Bicchieri (2006) works when its called for accounting for the UG results. I argue
that this kind of account risks to merely accommodate evidence and predictions because of
thenatureofitsadjustableparameters.

Thesecondpartoftheessayisshorter.Thereasonissimple,anditisimportanttobeclear:
Currently,thereisnoneurobiologicallyinformedmodel.Drawingonboththedetailedanalysis
ofthefirstpartandonsomerecentneurobiologicalfindings,mygoalistogivesomereasons
why if we want good predictions, then we should try to point to quantifiable biological
variableswhichhavealargeinfluenceonbehaviourandareunderweightedorignoredbothin
game theoretic, and enriched models such as Bicchieris. After having described one study on
theneuralbasisofeconomicdecisionmakingintheUG,Icriticallydiscusspossibleobjections
tothepotentialpredictivesignificanceofneurobiologicalvariables.

I.THEULTIMATUMBARGAINING

Game theory is a collection of models attempting to understand situations in which decision


makers interact with one another. Game theoretic analyses predict that rational, self
interested players will make decisions to reach outcomes, known as Nash equilibria, from
whichnoplayercanincreaseherownpayoffunilaterally.Strategicbargainingbehaviourisone
oftheconcernsofgametheory.
Toseetheroleplayedbytheassumptionsofrationalityandselfinterestingametheory,let
usconsidertheUltimatum(ortakeitorleaveit)Gamewhichisoneofthesimplestformof
bargaining. This twostage, twoperson game is defined as follows. A sum of money m is
provided.Player1proposesthatxunitsofthemoney(xm)beofferedtoplayer2.Player1
wouldretain(mx).Player2respondsbyeitheracceptingorrejectingtheofferx.Ifplayer2
accepts, player 1 is paid (m x) and player 2 is paid x; if she rejects, each player receives
nothing(0,0).Ineithercasethegameisover.
Two features of the UG are worth emphasizing. First, the UG is a noncooperative game:
Players cannot make binding agreements about what to do. They have to form expectations
aboutotherplayersactionwithoutcommunicating.Second,theUGisverysimpleinitsgame
theoretic analysis: It requires only two assumptions to make a prediction. It is simple in its
instructions:Playersunderstandquicklyandwithoutefforttherulesofthegame.
The UG is one of the most successful experimental designs in the history of the social
sciences(Guala2008).AccordingtoGuala,thesuccessoftheUGcanbeexplainedifwefocus
onsomeoftheepistemicfeaturesthatqualifyitasaparadigmaticexperiment.Oneofthese
featuresisitsversatility:Aparadigmaticexperimentenablesustomakecomparisonsanddraw
a variety of inferences in different contexts of scientific inquiry. This versatility makes the UG
suitable for my argument, where I evaluate the comparative merits and problems of the
predictionsgivenbytwomodelsbuiltintwodifferentcontexts.
Thegametheoreticanalysisofthegameyieldsaprecisepredictionaboutwhatplayerswill
do.Thepredictionofarestrictiveconceptingametheory,thesubgameperfectequilibrium,is
that for any positive amount offered by Player 1 (the proposer), Player 2 (the responder)

Humana.MenteIssue10July2009

44

knowsthatshefacesachoicebetweengainingnothing(ifsherefusestheoffer)orsomething
(if she accepts). If the responder maximizes her own payoff, she will accept any positive
amount. If the proposer maximizes her own payoff and expects the responder to maximize,
she will offer the smallest amount possible. Hence, the proposer will offer the minimum
possiblesplittotheresponder,whowillaccept.
Theassumptionsofrationalityandselfinterestentailthisprediction.Accordingtotheself
interestassumption,playersprefermoremoneytoless,anddontcareabouttheoutcomesor
preferences of others. Notice that this is a narrow conception of selfinterest, concerned
with money alone. Rationality has to be understood as practical, instrumental, rationality.
Given the agents perfect knowledge of the outcomes of the alternatives open to her and to
the other player, and given that she can identify the best of them, she will be practically
rational in choosing action A if A is the action she believes will lead to the consequences she
prefers.
The experimental literature on the UG indicates a robust behavioural pattern at odds with
the game theoretic prediction. Since the first experiment which studied the UG (Gth,
Schmittberger, & Schwarze 1982), the UG has been studied in many diverse settings where
differentparametersofthegameweremodified.Supposingthatthetotalsumofmoneyis10,
thesplitofferedistypicallyaround(6,4).Andlowoffers,namelyoffersaround2orlessoutof
10,areverylikelytoberefused(Camerer2003).
2
TheUGistypicallyanonymousandoneshot:
Players dont know the identity of the other, and play only once. The rationale for these two
characteristics is to abstract away from the possibility of incentives for reciprocity and
cooperativenesswhichwouldbeinvolvedinrepeatedgameswiththesamepartnerandto
keep players behaviour insulated from such influences as the desire to please the
experimenter,orthefearofruiningafriendship,ortheincentivetobuildagoodreputation
whichwouldbeinvolvediftheidentityoftheplayerwasknown.
The behavioural pattern displayed by people playing the UG immediately raises the well
knownDuhemproblem:Whydopeopletendnottoplaythesubgameperfectequilibrium,and
insteadtendtocoordinateon5050or6040splits?Whatpartofthestandardgametheoretic
model has to be blamed for the anomaly? And, once we have identified the culprit, what
modificationhastobedonesoastoenablethemodeltopredictwell?

II.BICCHIERISSOCIALNORMMODEL.ANANALYSIS

ByvirtueofthesimplicityoftheUG,wehavereasontomaintainthattheplayersarerational
in the minimal sense specified above: their actions follow from their preferences and beliefs.
What seems to be revised is the assumption of strict selfinterest that individual preferences
are concerned with money alone. Accordingly, the target of most of the theoretical
developments in the gametheoretic literature of the last two decades has been the
assumption of strictly selfinterested preferences. The most common move has been to build
models with nonstandard utility functions, according to which individuals have other
regarding preferences. This kind of models allows a players utility function to take into
accounttheoutcomes,preferences,andexpectationsofotherplayers.Noticethatacommon
feature of these enriched models is the preservation of the logical framework of expected

2
There is however some variation in the findings. The most remarkable variations are found either
acrossculturesoracrosssubjectswithaneurologicalconditionsuchasautism.(seeRothetal1991,and
Henrichetal.,eds2004oncrossculturalvariations;Sally&Hill2006foraresearchwithautistics).
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

45

utilitytheory:Theydonotrejecttherationalityassumption,theypointtothemaximizationof
anonclassicalutilityfunctionwhoseempiricalsubstanceisprovidedbythenewparameters.
3

In this section, I critically analyze the enriched model developed by Cristina Bicchieri in
The Grammar of Society (2006) (henceforth, GS). I focus on one model to facilitate a more
detailed discussion than it would be possible if I were to consider a wide range of current
views.IfocusonBicchierisfortworeasons.First,hersisoneofthemostrecentandpromising
proposals; second, and importantly, she argues that her model fares better than the
alternatives when it comes to prediction. In my discussion I pay special attention to whether,
andtowhatextent,thislastclaimisjustified.
Every known society has a multitude of social norms that regulate the behaviours of
individuals in a variety of situations. In a given social context, the same kind of social norm
mightproducedifferentbehaviouralpatternsacrossindividuals.Ifmostpeopleabidebysocial
norms, then norms can account for behavioural patterns observed in a population. And
different norms can account for behavioural variations across societies. Consequently, social
norms are important for predicting the behaviour of individuals. These observations call for
elucidation:

1)Whatisitmeantexactlybyasocialnorm?
2)Whatarethemechanismsthatregulatethepowersocialnormshavetoinfluencehuman
behaviour?
3)Whataretheconditionsunderwhichindividualsarelikelytofollowasocialnorm?

InGS,Bicchieriattemptstoaccountforthesethreeproblemsbyprovidingthefoundationsfor
anewmodelofhumanbehaviourbasedonaprecisecharacterizationofasocialnorm.

DefinitionAccordingtoBicchieri,aruleisasocialnorminapopulationifandonlyifasufficient
numberofpeopleinthatpopulation:

(i)knowthattheruleexistsandappliesinsituationsofacertaintypeS,and
(ii)prefertoconformtoit,insituationsoftypeS,ontheconditionthat
(a)itisbelievedthatasufficientnumberofothersconformtoitinsituationsoftypeS,
and,either
(b)itisbelievedthatasufficientnumberofothersexpectonetoconforminsituationof
typeS,or
(b`)itisbelievedthatasufficientnumberofothersexpectonetoconforminsituationof
typeS,preferconformity,andmaysanctiononeifonedoesnotconform(GS,p.11).

Suffice here to clarify three points about Bicchieris definition. First, a social norm is a set of
mutual expectations, and a communality of beliefs is a precondition for its existence. A social
norm like tipping for service in a restaurant has no reality other than our expectations that
others leave a tip in a restaurant, and that others expect us to tip in the same type of
circumstance. Second, a social norm cannot be simply identified with a recurrent collective
behavioural pattern. Taking a shower in the morning is not a social norm. I take a shower in
the morning whether or not I expect others to do the same. Moreover, a rule can be a social
norm in a population, even if compliance to it is not observed. Imagine the social norm in a
population that whoever first makes a proposal that something has to be done is directly
responsible for making sure that the proposal is carried out. During a seminar students may

3
WellknownenrichedmodelsareRabin(1993),Fehr&Schmidt(1999),Bolton&Ockenfels(2000).

Humana.MenteIssue10July2009

46

avoid suggesting a certain topic of discussion fearing that that social norm will be followed,
and,hence,theywouldhavetopreparethetalk.Inthissituationnobodyisviolatingthenorm.
Everybody is eluding it. Third, the conformity to a social norm is conditional: One has a
preference to conform to a norm N in a situation of type S, under the conditions that one
expects others to conform to N in S (empirical expectations), and one believes that others
think one ought to conform to N in S (normative expectations). This condition is discussed in
detaillater.Notice,forthemoment,thattheconditionalityofthepreferencetoconformityto
a norm lends itself to empirical testing. Were the expectations that underlie a social norm to
be different, we would predict behaviour to change in determinate ways. Clearly, unless we
have an account of when, how, and to what degree the expectations that constitute a social
norm affect behaviour, there is little hope in drawing precise, informative predictions. We
need understand the mechanism by which social norms influence our behaviour, and the
conditionsunderwhichindividualsarelikelytofollowasocialnorm.

Mechanism Bicchieri undertakes the first task by relying on findings from experiments in
cognitive psychology (GS, Ch.2). The mechanism works as follows. Subjects interpret and
categorizeagivencontextasafunctionofthesituationalcuesthatsparktheirattention.The
process of categorization relies on spreading activation. That is, the activation of the
representation of a certain concept spreads to representations of concepts related to it. E.g.,
when we are presented with the stimulus word tiger, we retrieve not just the representation
of a tiger, but also related representations like feline, predator, etc. Then, depending on how
subjects categorize the context, a script of a certain type activates. Scripts are cognitive
structuresweacquirethroughpersonalexperienceandhabitthatrepresentstoredknowledge
about people, objects, events, and roles relevant to the situation at hand. Scripts prompt
beliefs and expectations about social roles and sequences of actions appropriate in that
situation. A restaurant script, e.g., represents roles (waiters and diners) and sequences of
appropriate actions (diners enter the restaurant, wait to be seated at a table; waiters take
theirorder;dinerseat,askforthebill,pay,leaveatip,andleavetherestaurant).Socialnorms
areembeddedintoscripts(GS,p.94):Socialnormsareamongthesetofbeliefs,expectations,
andpreferencespromptedbyascript.Toseethismechanisminaction,consideranUGframed
intermsofthegainsfromatransactionbetweenabuyerandaseller.Player1issaidtobethe
seller;sheisendowedwith10.Player2issaidtobethebuyer.Atableregisterstheprofitof
the seller and of the buyer for each price (0, 1, 2,, 10) charged by the seller when the
buyerdecidestopurchase.Theprofitofthesellerisequaltothepriceshestates;theprofitof
thebuyeris10minusthatprice.Theprofitofeachiszeroifthebuyerrefusestopurchaseat
the price stated by the seller. In such context, the cues provided are likely to guide the
interpretation of the situation in terms of a market situation. Presumably, the categorization
marketexchangeactivatesascriptthatdefinesdeterminatemutualexpectationsunderlying
asocialnorm(ifitexists)amongtheplayers.
When compared with the results of an UG with standard instructions, it is found that the
buyerseller manipulation elicits lowered offers, whereas the rejections rates remains
unchanged (Hoffman et al 1994). Indeed, in western culture the right of sellers to quote a
higher price is not usually questioned, nor that of the buyer to decide to purchase or not to
purchase. Two points are worth noticing. First, consistently with Bicchieris account, the
example just provided unambiguously shows that context matters. Two UGs with the same
logicalstructures,butembeddedindifferentcontexts,withdifferentsituationalcues,arelikely
toelicitdifferentbehavioursbecausetheyarelikelytopromptdifferentsocialnorms.Second,
as acknowledged by Bicchieri, [t]he predictive power of a theory of norms therefore depends
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

47

on knowing which situational cues trigger which norm (GS, p.76). But it also depends on a
numberofotherconditions.Intheremainingofthissectionitisarguedthatthefulfillmentof
theseconditionsrepresentimportantchallengestothepredictivesuccessofhermodel.

ConditionsforpredictionTherearetwoconditionsunderwhichanindividualislikelytofollow
a social norm N in a context of type S. First, N is correctly identified. Second, individuals in S
havetherightkindofexpectations,andthereforeNexistsinS.
Identifying a norm There are at least three reasons why identifying N in a certain context
mightbeproblematic.Thefirstisthat,ifweexclusivelyfocusonthebehaviourofpeoplee.g.
in the UG, there might be alternative social norms, N, N`,,N, that entail the same
behavioural pattern. This is the problem of (deductive) underdetermination, and it would
threattheinformativenessofthepredictiongivenbythemodel.Ifwewantourmodeltogive
informativepredictions,weneedanassessmentofthesocialnormlikelytobeinplaceinthe
UG independent of the behaviour observed in the game itself. Independent measurement is
one of the remedies for limiting the bite of the underdetermination problem. With social
norms, however, independent measurement is made difficult because they might be too
vague, and consequently it might not be clear how to assess them independently of
behavioural data. This is the second problem. The third is that individuals of a population
mightbeunawaretobeinacontextwhereNapplies.Thishastodowiththesituationalcues,
crucialinprimingN.
Underdetermination Consider this slightly different form of the UG. Before the responder
hearstheoffer,shemustsetanacceptableofferrange:sheisaskedwhethershewouldaccept
a1000split,andthenwhethershewouldaccepta9010,a8020,a7030,etc,untilapointis
reached where she would accept anything higher. If player 1s offer is below her acceptable
range, her response would count as a rejection. Player 1s proposal is finally revealed. The
players belong to the same population, and there are no contextual cues that affect their
expectations. An experiment with Gypsies in Vallecas, Madrid, shows that in this situation
although 97% of proposers offered an equal split, as responders the Gypsies were willing to
accept completely unfair offers: The acceptance of thezero offer was the modal value (Pablo
etal2006).Inordertohavepredictedsuchbehaviouralpatternonthesideoftherespondent,
wewouldhaveneededtoknowaboutthesocialnormsexistinginthatpopulationrelevantto
anUGsituation.Onepossiblewaytomakesenseofthisbehaviouralpatternisbymeansofa
normofhospitality.Hospitalityseemstobeasocialnormthatprimarilyaffectsthebehaviour
of those who can offer: The proposer offers half and the other accepts whatever is offered.
Oncethisnormhasbeenrecognizedasobtaininginthatsituation,Bicchierismodelwouldnt
have problems in predicting the behaviour of the Gypsies. However, the same behavioural
patternisalsocompatiblewithadifferentsocialnormlikehelptheneedy.Accordingtothis
norm one would expect that if the proposer offers zero, then she is needy, and this would
motivate the respondent to accept a zero offer. With some imagination we can devise other
socialnormscompatiblewiththesamebehaviouralpattern.Becauseofunderdetermination,it
would be logically possible to find an infinite number of models that will accommodate the
same behavioural pattern without genuinely predict it. One way to tackle the problem is to
measure the expectations underlying the social norm we assume to be in place
independentlyof the observed behaviour. As convincingly argued by Larry Laudan in different
places (e.g. Laudan 1990), deductive underdetermination does not entail that the choice of a
model is underdetermined. One of the criteria used to decide which model is better, and
consequently to be preferred over rivals, is independent measuring of its assumptions.
Independent evidence about a social norm in a population would facilitate us to uncover the

Humana.MenteIssue10July2009

48

real expectations and motives behind the behaviour in an UG. And knowing motives and
expectationswouldenableustopreferamodelthatyieldsinformativepredictions.
VaguenessandIndependentMeasuresHowwemeasureasocialnormindependentlyofthe
observed behaviour is problematic since social norms are complex, vague, and might be
impossible to define them in a precise way. A norm of hospitality can be regarded as an
example of a vague social norm. I use vague meaning that hospitality cannot be precisely
definedbecauseithasmanydiverseaspects.Inatypicalsituationwherehospitalityexistswe
might expect that the host invites the guests in her house; the host entertains them with
kindness;sheoffersfoodanddrinks;theguestacceptswhateverisofferedwithgoodwill,etc.
Hospitalitymightexistinacontextevenifanyoneoftheseexpectationsismissingaslongas
people in that context share some of the other expectations. Yet, there are at least three
methods to measure a social norm: first, questioning people; second, inferring norms from
behaviour(otherthanthatdisplayedintheUG);third,indepthethnographicresearch.Eachof
thesemethodshasbiasesthatmayunderminethereliabilityofthemeasure.
WhichsplitproposedinanUGwouldbefair?,orWhatisthenormtofollowinsituation
oftypeS?arethekindofquestionsthatmayenableustodeterminethecontentofasocial
norm of fairness in a population. Questioning people, however, may yield an unreliable
measure because they may lie, may give the answer they suppose the research wants, may
understandthequestionindifferentways.
Apartfromthereasonsgivenaboveforwhyasocialnormcannotbesimplyidentifiedwith
a recurrent collective behavioural pattern, there are at least two further challenges in
measuringasocialnormfrombehaviour.First,toaccountforbehaviourbycitingasocialnorm
may lead to circularity if the norm is first measured from the behaviour in question. Second,
there is the problem of simultaneous attribution of belief and desire, namely the problem of
discriminating the respective roles played by an agents beliefs and desires in the production
of the actions we observe her to perform (Bradley unpublished). On the basis, e.g., of the
observation that a number of people leave a tip at a restaurant, we might attribute to them
thebeliefthatthereexistsasocialnormoftippingatarestaurant,andthedesiretofollowit.
Alternatively, we might attribute to them a belief that the service has been really good, and
the desire to reward good service regardless of there being a social norm of tipping at a
restaurant. The two alternatives entail different behaviours under different circumstances:
After an unsatisfactory service, according to the first beliefdesire attribution one may still
leaveatipfollowingthesocialnorm;instead,accordingtothesecondbeliefdesireattribution
onewillnot.Thechoiceastypicallyobserveddoesnotaloneallowustodecidebetweenthese
two alternatives, and the many other possible ones. To be sure, this is not a knockout
challenge since each hypothesis is testable in principle; that is, there are conditions under
which we can determine whether one of them is probably false. Nonetheless, the problem
remains to specify an underlying theory of beliefs and desires which enables us to
systematicallyinferonesmentalstatesfromherobservedbehaviour.
The third method, that of ethnographic research, has been adopted by a group of
economists and anthropologists who set out to study the foundations of human sociality
through classic economic experiments like the UG in fifteen smallscale societies in South
America,Asia,andAfrica(Henrichetaleds.2004).Foreachpopulationexperimentalresearch
was flanked by independent information concerning social context, political and economic
structure, religious beliefs, etc. It turned out that the betweengroup behavioural differences
in the UG were related to indicators of patterns of social and economic interaction (e.g. level
of market exchange, importance of anonymity in commercial transactions) that framed the
daily life of the population. Yet, there is no guarantee that a significant correlation will be
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

49

always found between behavioural patterns in the UG, and the socioeconomic variables that
researchers decide to study. This may be due either because there is in fact no correlation
between UG decisions and variables of socioeconomic patterns, or because the researchers
dontcollecttheappropriateinformation,orbecausetheinformationtheycollectfromcertain
individualsisnotrepresentativeofthewholepopulationbutofonlysomesubgroupwhichhas
tobeidentified.
Context and Cues The third condition that would enable us to predict that an individual is
likelytofollowanormisthattheindividualhastobeawaretobeinacontextwherethenorm
applies.Situationalcuesgovernthemappingbetweencontext,recognizedasbeingofacertain
type courtesy of the cues, and activation of the social norm appropriate to that context. But,
what drives peoples attention to situational cues? The key notion to tackle this question is
salience. A salient item stands out relative to neighbouring items, thereby sparking peoples
attention. A red dot surrounded by green ones is salient. Saliency is perhaps best understood
inthefieldofvisualperceptionwhereitisdefinedinfunctionofsuchcuesascolour,intensity,
orientation, and motion. In the case of norms, salient cues Bicchieri suggests (GS, p.112)
may involve a direct statement or reminder of the norm, observing others behaviour,
similarityofthepresentsituationtoothersinwhichthenormwasused,aswellashowoften
or how recently one has used the norm. Salient cues prime the expectations underlying the
norm.Conditionalontheseexpectations,certainbehaviourislikelytoensue.
Bicchieritestedhermodelbymanipulatingsalience.InanUG,peoplesattentionhasbeen
cued by information about the normative expectations of others that had played the game
before. The prediction obtained that more players would follow a norm of fairness, which in
thatcontextwoulddictateanequalsplit(Bicchieri&Xiao2008).
The potential problem for prediction here stems from a tension within the model. On the
onehand,Bicchierismodelmakesreferencetotypesituations.Ontheotherhand,theappeal
tosituationalcuesmakestheinterpretation\activationofsocialnormsdependentonspecific,
tokensituations. The interpretation of a social norm is local: Fairness, e.g., has different
meanings in different circumstances, depending on the people, objects and environment that
define the situation. A typesituation is a general type of situation like football match,
bargaining, theatreplay. A tokensituation is a specific situation of a certain type
performed in a particular context. Hamlet played at La Scala now, with a certain setting,
certain actors, and a certain audience, is a tokensituation of the type Hamlettheatreplay.
Wecanjudgethattheindividualiisinasituationofacertaintype;e.g.wecanjudgethatiisat
atheatreplayofHamlet.Butimakesdecisions,hasexpectations,followsasocialnorm,always
intokensituations.E.g.iexpectstosmilewhentheHamletsheisattendingisbeingperformed
by a company of funny comics wearing fancy dresses; instead, she expects something deep
when the play is being performed by a company of continental philosophers. The two
situationsareofthesametype,butdifferinsituationalcues.Thedifferenceislikelytoprompt
differentexpectations,preferencesandactions.ConsideragaintheUG.Ithasbeenshownthat
in a common UGtokensituation, a bargaining typesituation, attractive people are treated
differently by others in that they are offered more, and they are expected to give more
(Solnick & Schweister 1999). Now, in a tokensituation of a certain type there may be
innumerable situational cues that can spark ones attention. The attractiveness of the people
inabargainingtypesituationcanbeone.Infunctionofourpersonalhistory,wewillinterpret
thecontextinonewayoranother,wewillfindonepersonattractiveornot.Theinterpretation
ofthecontextwillaffectourexpectationsandpreferences,andhencemayprimethisorthat
normwemayfollowinthattokensituation.EvenifwecanjudgethatiisinatypesituationS,
we still need careful inquiry into the tokensituation. The potential problem for prediction is
herewithinformativeness.Forwewouldlikeamodelthatcouldbeappliedtotypesituations

Humana.MenteIssue10July2009

50

besides the tokenones from which it was deduced. We would need a general underlying
theory that specifies some systematic functional relationship between situational cues, which
are adjustable variables of the model, and experimental results. We would like something
analogue to a saliency map, which represents visual saliency of a corresponding scene, made
availablebyneurocognitiveresearchonvisualperception(e.g.Koch&Ullman1985).AsfarasI
know there are no such maps in the field of decisionmaking yet. The challenge is in
measuring,orquantifying,theinternalstateofanindividual,suchasherpersonalhistory,the
goals,beliefsandmotivationsshehasatatime.

Having the right kind of expectations One has a preference to conform to a norm N in a
situation of type S, provided N exists and has been correctly identified, under the conditions
that she expects others to conform to N in S (empirical expectations), and she believes that
others think she ought to conform to N in S (normative expectations). There are at least two
problemshere.First,theconditionsarenotsufficienttodetermineonesfollowingarule.This
maythreatthesecurityofthepredictionsthemodelgives.Second,thereisnohintastohow
(quantitatively) determine the likelihood of observed behaviours. This is a drawback for
predictiveinformativeness.
Theinsufficiencyofhavingtherightkindofexpectationsisapparent:Weoftenhaveother
personal motives for not following a norm. I may live in a society with a strong norm of
revenge. I expect that most people take vengeance on those who have wronged them, and I
believethatothersthinkIoughttotakerevengeonthosewhohavewrongedme.However,I
have also a stronger preference for not harassing others. Bicchieri acknowledges that the
presenceofanormofrevenge,anditssalienceinaparticularsituation,motivatemetoactin
a congruent manner, but my behaviour is ultimately explainable [and predictable] only by
referencetomypreferencesandexpectations(GS,p.22).Supposeweasktheresponderinan
UGwhatshebelievesisfairinthatUGsituation,andwhethersheexpectsotherstofollowit.
ThismightgiveusinformationthatthispersonbelievesanormNappliesinthatsituation,and
thiswouldleadustopredictheractioninaccordancewithN.However,suchactionmightstill
not occur, due to the presence of other motives and incentives which, unbeknownst to us,
bear on her decision more than her expectations related to N. To remedy this problem,
Bicchieri suggests (Bicchieri & Chavez 2008) that a finegrained account of individuals
sensitivity to specific norms would give us more reliable grounds to predict their behaviour
sinceitwouldenableustocomparethestrengthofconcurrentmotivationsinagivensituation
e.g. the preference to follow a norm versus other preferences that may overcome that
preference. How does the suggestion work exactly? Consider responders and proposers in an
UG. Each player can be of different types. In order to measure the expected normsensitivity,
call it K, of a type of proposers (P) we may ask responders about the expected type of offer
versustheofferthatisfair.Thenwewouldhaveindirectinformationabouttheexpectedvalue
ofKP:Ifthetwovaluesdiffer,theexpectedKPislow.Iftheyarethesame,theexpectedKPwill
be high or low depending on the (high or low) values of the expected type of offer and the
offer that is fair. If we ask proposers about the distribution of normative expectations on the
part of responders (R), and we then observe their offers, we can get information about the
expectedKR.(Bicchieri,personalcommunication).Iseethreeproblemswiththissuggestion.

First, Bicchieris reasoning entails that we can only have indirect evidence about ones K. The
information we obtain following her suggestion is a measure of the type of individuals
expected by the others. The grounds to ascribe a certain K to a responder in an UG would be
the proposers beliefs about the responders normative expectations, seeing whether the
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

51

proposersbehaveaccordingtotheirbeliefs.Theproblemiswhetherthiskindofevidencegives
usgroundsstrongenoughtoreliablyascribingKstorespondents.Mysuggestionisthatdirect
evidence about ones K would do a better job. For a proposer may believe e.g. that a
responder has strong normative expectations of fairness, and accordingly she makes a fair
proposal; yet, the respondents K doesnt match with the proposers expectations since she
alsoacceptsunfairoffers.Thecontrastbetweenhavingdirectevidencethattheindividualiis
sensitive to a norm and having indirect evidence thereof is between is having certain
expectations and preferences and her saying so, and anothers believing that i has certain
expectations.Adirectwayofknowingwhatarespondentbelievesandprefersisthusherown
sincere confession. But this would pose again the threat of accommodation since the model
wouldsimplydescribethesituationathand.
Second,thereisaproblemwithinformativenesssinceBicchierisapproachseemstohavea
difficultyinprovidingaquantitativeanswertothequestion:Howwellthemodelpredicts?We
would like to know about the predictive strength of a model by means of probabilities that it
would assign to the predicted outcomes. If our predictions depend on a parameter K, we
wouldliketoknowasystematicwaytodeterminetowhatdegreeoneissensitivetoanormin
a context. In default of such positive proposal the predictions of our model lacks in
informativeness.

The third problem is conceptually prior. It is that of determining quantifiable individuallevel


variables statistically correlated with players decisions. Bicchieris model takes into account
typesofindividuals,definedbytheirbeliefsabouttherelevantnorminatypeofcontext,their
expectations about their opponents, and their normsensitivity (Bicchieri, personal
communication). The model seems promising with predictions across groups of individuals
where we can take groupmembership as defining a type of individual. However, it is unclear
howitwouldpredictthebehaviourofspecificindividualswithinagroup.Itis,infact,possible
thatindividualsofthesametypebehavedifferently.ForthewayBicchiericonstruesatypeof
individual allows that they can have different motives and preferences. Although variation
between types of individuals can be accounted with economic, social, or cultural differences,
thesamedoesnotapplywithinthesametypeofindividuals.
Suchachargemayseemunfair:thesocialsciences(andtoagreatextentalsopsychology)
are less concerned with single individuals than with statistical tendencies across individuals.
Alsoinphysicsitiscustomarytotakeintoconsiderationthebehaviourofmanyparticlesrather
thanasingleone.
Mypointhowevercanbereadinaweakerway.Evenifthebehaviourofasingleindividual
withinagroupmaynotbethatinteresting,collectingappropriateindividuallevelinformation
systematically correlated to ones choice may be useful to make predictions more secure and
informativeacrossgroupsofindividualsafterstatisticalgeneralization.Asusualinscience,we
mayexamineasampleofacertainclassofindividualsandthengeneralizeourfindingstothe
class as a whole. This information may be neurobiological. The second part of this work
articulatessuchproposal.

III.NEUROBIOLOGICALLYINFORMEDMODELS

Twentyone years after Gth et al.s seminal work on the UG, a group of psychologists and
neuroscientists led by Alan Sanfey analyzed subjects with functional magnetic resonance
imaging (fMRI) as they played the UG. Sanfey et al.(2003) compared the brains of subjects
respondingto5050,6040offers(intheexperimentthetotalpietosplitwas$10),and9010,
8020 offers. Three brain areas were found to be differentially activated: the Dorsolateral

Humana.MenteIssue10July2009

52

prefrontalcortex,theanteriorcingulate,andtheanteriorinsula.Forthepurposeofthisessay,
thecrucialfindingisthecorrelationbetweentheactivationoftheanteriorinsulawithchoice
behaviour. Specifically: First, the activation of the anterior insula was significantly correlated
withtherateofrejection;Second,themagnitudeofactivationwasafunctionoftheamount
ofmoneyofferedtoparticipants(p.1756).Third,theactivationwasuniquelysensitivetothe
context (Ibid.): there was greater activation for a 8020 offer from human partners than the
sameofferfromcomputers.Fromthefirstfindingfollowsthatwhetherplayersrejectanoffer
ornotmaybepredictedwithacertainaccuracybytheleveloftheirinsulaactivity.
Sanfeyetalsexperimentisquitetypicalintheworldofsocalledneuroeconomics:From
a subject who confronts a choice problem D, a pair (d, x) is drawn. d is the alternative in D
chosenbythesubject;xisavectorofnumbersrepresentingtheactivitiesmeasuredinvarious
areasofthesubjectsbrainduringtheperiodbetweenthemomentsheispresentedwiththe
choicesetofDandthemomentshechoosesd.Afterstatisticalanalysis,acorrelationbetween
choicebehaviour d, and the activity of a specific brain area may become apparent. Building
upon this kind of correlation, it may be possible to integrate neurobiological variables into
behaviouralmodels.Thepredictivepowerofsuchmodelswouldbeboundtothesignificance
and robustness of the neurobiological finding. The remaining of this essay is a critical
assessment of the comparative merits, and limits, of the neurobiological approach to
modelling I wish to suggest. Security and informativeness of the predictions that this kind of
modelswouldgiveorientatemydiscussionaswithBicchieris.
SecuritySanfeyetalsexperimentinvolved19Americansubjectsthatcompleted10rounds
playing the game with a human partner, and 10 with a computer partner. Rubinstein (2006)
andHarrison(2008)crisplypointoutthattheinferencesdrawnfromexperimentsofthiskind
are highly problematic because of their small, culturally homogeneous samples, the
methodology of pooling subjects, and the heavy statistical machinery required to make raw
neurological data amenable to analysis. How can one feel comfortable about building upon
such weak grounds? I would like to make two points about Rubinsteins and Harrisons
critiques.
First, their charges about samples and statistical analyses are empirical in nature. As such,
there seems to be no principled reason why the methodological standards that enable us to
warrant the inferences from a neuroeconomic experiment couldnt be raised. Whereas
statisticalflawscanbesolvedwithcarefuldataanalysismadeexplicitandopentoscrutiny,the
problem with samples size can be faced in the long run, when many replications will tell us
about the robustness of regularities like that lying below Sanfey et als set of data. I use
replication, as opposed to exact repetition (see Radder 1996; Guala 2005, Ch.2). A
replicationinvolvessome(slightorradical)modificationoftheoriginalexperimentaldesign.A
nicereplicationofSanfeyetalsexperimentmightinvolvesubjectsfromdifferentcultures.At
the moment we can only speculate e.g. that in Israel, where lower offers are accepted more
often(Rothetal1991),wemayexpectlessinsulaactivityforthesamesizedofferthaninthe
US, and less rejection. If the regularity firstly observed survives across replications, then we
havereasontoconcludethatitisrobust:Itdoesntdependondetailsofthesituationoronthe
particular statistical assumptions used to derive the results. Therefore, by emphasizing the
evidential value of replications, and by noticing that we have no reason why we couldnt
expectreplicationsofneuroeconomicresults,andhigherstatisticalstandards,thebiteofthese
chargescanbelimited.
Second, a better way of evaluating Rubinsteins and Harrisons critiques is to ask whether
the evidential grounds for neurobiological predictive claims may be more reliable than the
grounds for psychological ones. Analyzing Bicchieris model we discussed some of the
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

53

evidentialproblemswithmeasurementandascriptionofexpectationstopeople.Iarguedthat:
First, because of the nature of social norms, secure predictions can be gained at the risk of
accommodation; Second, the conditionality of preference to follow a social norm seems to
openanepistemicgapbetweentheactualmotives,inatokensituation,thatultimatelyrender
a prediction correct\incorrect, and the information about typesituations and type of
individual,necessarytogroundthepredictionsofBicchierismodel.
The grounds for predictions for a model which integrates neurobiological parameters
consistintheempiricalfindingthattwoquantitativefactors(e.g.insulaactivationandrateof
rejectionsintheUG)arecorrelatedinsuchawaythatthebehaviouroftheoneforeshadows
thebehaviouroftheotherwithstatisticalsignificance.Thesecurityofthepredictionsisbond
to the stability of the functional specialization of the target neurobiological structure across
brains and situations. On the one hand, however, the activation of the target structure may
not be sufficient to warrant the prediction since it may be overcome, or blocked, by the
activities of other structures. However, the problem is not insoluble: We can identify to what
degree the insula activity is sufficient to warrant the prediction of rejection in the UG by
inquiringitsrelation\interactionwithotherbrainstructures.Ontheotherhand,theactivation
of the structure may not be necessary. As before, the problem can be solved with further,
accurate experimentation. In order to assess to what extent the insula activity is necessary in
predicting the rejection rate, it would be interesting, e.g, to study the behaviour in the UG of
neurologicalsubjectswithlesionsintheinsula;but,asfarasIknow,suchexperimenthasnot
beensetupyet.
Amoreseriousthreattothereliabilityoftheevidentialgroundsforneurobiologicallybased
predictions is the variability of brains across individuals. The problem, here, is not with token
vs. typesituations, which would be bypassed by focussing on brain activations since brain
activations are already sensitive to the tokensituation where a person acts. The problem,
instead,hasprimarilytodowiththeplasticityofthebrain:Plasticity,orneuroplasticity,isthe
capacityofthebraintoreorganizeneuralpathwaysbasedonnewexperiences.Bothstructure
andfunctionofdevelopingbrainsareshaped,tosomeextent,bothbytheenvironmentandby
culturalexperience.Differencesinneuralresponsestoagivenstimulusarealsolikelytoexist
All brains, then, are different from one another to some degree. To the extent that different
culturalnormsandpractices,anddifferentenvironmentsexistacrosssocialgroupsdifferences
in neural responses to a given stimulus are also likely to exist (Chiao & Ambady 2007). This
likelyvariationposesatleastonesubstantiveproblemtoamodelneurobiologicallyinformed.
Wemightexpecttofinddifferencesinthetypeofneuralactivitycorrelatedtothesametype
of behaviour across subjects. Were this actually so, then the basis for our neurobiologically
informedmodelofhumanbehaviourwouldbeintrinsicallyunstable.However,wehavereason
tobelievethatthebrainvarianceacrossgroupsisnotsodramatic.

First, it is worth noticing that despite notable progress in describing cultural variation at the
behavioural(andgenetic)level,relativelylittleisknownabouthowthestructureandfunction
ofthehumanbrainvaryacrosssubjectsandcultures.Hence,sweepingclaimsarenotjustified
bycurrentevidence.Second,althoughthereisnoarguingthatthereissomekindofplasticity
in our brains, there are clear limits on plasticity (Gazzaniga, Ivry, & Mangun 2002, ch.15).
Different types of neuroplasticity occur during certain critical periods, notably: 1. During
normal brain development when the brain begins to process sensory information, when it is
still immature; 2. After brain injury, especially when the lesion involves the somatosensory
system, to compensate for lost function; 3. Through adulthood underlying learning and
memory mechanisms. After these critical periods, the central nervous system is characterized
by a relative rigidity. Moreover, when it occurs, neuroplasticity seems to be particularly

Humana.MenteIssue10July2009

54

significant at microlevels of organization, e.g. at the synaptic level (Elmanet al. 1996; Quartz
& Sejnowski 1997). Here, plasticity involves subtle changes in the strength of synaptic
connections between neurons which underlie the mechanisms of learning and memory. For
thesereasons,itappearsunlikelythatwefinddramaticdifferencesinthetypeofbrainactivity
correlatedtothesametypeofbehaviouracrossadultindividualswhohavenotsufferedbrain
lesions the target of brain imaging experiments indeed. Brains are shaped, to some extent,
both by the environment and by cultural experiences. All brains, then, are different from one
another to some degree. To the extent that different cultural norms and practices, and
different environments exist across social groups, Given this relative stability in the brain
activity underlying a given type of behaviour, more reliable grounds for a neurobiologically
informedpredictivemodelaretobeexpectedwhenwewillhaveageneral,detailedtheoryof
the brain, and as we come to understand better how the instruments and techniques
employed by neuroscientists mediate their observations (see Bechtel forthcoming, for the
epistemologyofevidenceandinstrumentsinneuroscience).Atthemoment,aswithanyother
immaturescience,thereislotsofworktodointheneurosciences:Weknowagooddealabout
brain neuroanatomy, and cytoarchitecture. We know less about the physiology and
functioning of the brain than we know about other organs. Little is known about the
mechanismsunderlyingtheexerciseofourcognitiveabilities.Currently,thelevelofdetailand
qualityofourknowledgeaboutthebrainisessentiallyconstrainedbytheimprovementofthe
technologicalapparatususedduringexperimentation.Howandtowhatextentwewillbeable
tointegrateinacomprehensivetheoryofthebrainthehugeamountofdatacollectedduring
experimentsisstilltobedetermined.
Informativeness Consider the following experiment: People are asked to say aloud, sincerely,
their reasoning during an UG. The experimenter records everything, and at the end of the
game comes up with a model based on the notes she has written down. Her model is highly
successfulsinceitfitsexceptionallywellthedata:Itgetsthebehaviourofthosesubjectsright
withadegreeofaccuracyof99%1%beingtheerrorrateduetoherdisattention.Shouldwe
regard that a good predictive model? Clearly not. For it is a mere record, or description, of
what was going on during the UG. The evidence that that model purports to predict was
already used in the construction of the model itself. The evidence is written into the model:
This is a clear case where the model accommodates the evidence without predicting it. Now,
what of the objection that the situation of a brain imaging experiment during an UG is
analogoustothatcase?Intheimagingexperiment,sothechargewouldgo,wewouldbejust
witnessing the brain activity correlated to the thought processes leading to decision;
moreover, if mapped closely enough there may be a 100% correlation between patterns of
brain activity and observed behaviour. Hence, the evidence would be written into the model,
and,asbefore,themodelwouldmerelyaccommodatetheevidence.

There are at least three main reasons why the case of a neurobiologicallyinformed model is
unlike the one above. All three reasons draw on the special weight of independent
assessment. First, the insula, the brain structure that correlates with rejection in the UG, is
knowntobeinvolvedinavarietyofbehavioursbesidetheoneobservedintheUG.Significant
activationsoftheinsulahavebeenobservedinresponsetodisgustinggustatoryandolfactory
stimuli (Small et al. 2003; Zald & Pardo 2000), to the sight of disgusted facial expressions of
others(Phillipsetal.1997).Followingstimulationoftheanteriorinsula,ithasalsobeenfound
thatsubjectsreportfeelingnauseous(KrolakSalmonetal.2003).Ifwecomparethesefindings
withthoseofSanfeyetal.sfMRIexperimentduringtheUG,repulseisthefeaturecommon
to all the correlated observed behaviour. This commonality would lend predictive
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

55

informativenesstoourneurobiologicallyinformedmodelsinceitcouldbeappliedtosituations
besidetheUGfromwhichitwasdeducedtopredictrejection.Tothispointitmaybeobjected
thattheremarkbyasubjectIamgoingtorejecttheofferisalsoapplicabletoawidevariety
of situations besides the UG, e.g. selling a car or renting an apartment. So this doesnt
distinguish brain imaging studies from the imagined write down everything they say about
their reasoning study. Therefore, the point made above seems irrelevant. However, this
objectionseemstometooverlookafundamentaldifferencebetweenthetwocases.Suppose
theidealsituationwhereasubjectissincereinhersayingIamgoingtorejecttheoffer,there
is no physical impediment to the realization of her intention,andshe has no other intentions
thatmayoverridethatone.Contrastthissituationwithanotheronewhereasubjectsinsulais
significantly activated and there is no other activity in her brain that may override that one.
Now,consideringthefirstcase,itseemstomethatitisnotanempiricaldiscoverythatwhen
thesubjectstatesthatsheisgoingtorejecttheoffer,shehastheintentiontodoso,andshe
will do so regardless of the context of her utterance. Because her behaviour is conceptually
bound up with the meaning of the predicate to be going to, the write down everything she
says study would simply describe the situation at hand, and would not provide us with any
significant empirical discovery. Consider instead the second case. There we witness a neural
event correlated with a persons repulsebehaviour. Unlike the former case, the correlation is
notconceptual,butinductive:Wefirstobservethecorrelationinacertaincontext;then,given
numerousobservationsofthesamecorrelationinavarietyofcontextsbesidethefirstonewe
infer that a certain brain region is correlated to a certain type of behaviour. Unlike the first
study, then, we have made an empirical discovery, a discovery revisable in light of further
empiricalevidence.
The second main reason why the two cases differ is the following. Courtesy of the vast
array of technologies used in neuroscience we might use the data from one experiment to
predict independently what is likely to happen in another. We might activate or disrupt the
insula by using e.g. rTMS (repetitive magnetic stimulation, which temporarily disrupts brain
activity in a target region of interest), predict differences in rates of rejection that would
follow,andseewhetherwegetthemright.That,unfortunately,hasnotbeendone.
Finally, as emphasized by Worrall (2006), the interesting cases in which some evidence
follows from a model, all involve a general underlying theory which itself stands in need of
confirmation from evidence. Although, as indicated earlier, there is currently no
comprehensive theory of the brain, there is no reason why after accumulating evidence such
theory, which would specify the functional relationship of the insula with other brain
structuresandotherobservedbehaviours,cannotbearticulated.
Provided the model would not simply accommodate the data, and that it would give
quantitative predictions by relying on correlations whose significance can be determined
statistically,towhatextentwoulditmarkasubstantiveimprovementinpredictingbehaviour?
I would like to answer by borrowing an analogy from Rescher (1997, p.127). Take
meteorological forecasting in Britain. The easiest route to predictive success seems to stick
with the hypothesis that tomorrows weather will be the same as today. In recent years, this
prediction has had a rate of success of 75%. The British meteorological Office, courtesy of its
hightech resources, has been able to push this rate up to 85%. It doesnt look like a terrific
improvement. Nonetheless, it is comparatively significant and further improvable in the level
of detail. The same might be said for how a neurobiologicallyinformed model would fare in
the UG: It would give comparatively more precise predictions sensitive to potentially
significantdetailsignoredbyalternativemodels.

Humana.MenteIssue10July2009

56

IV.CONCLUSION

ArielRubinstein(2008)asksustoassumethefollowingscenario:Weareabletomapallbrains
ontoacanonicalbrain.Thefunctionsofthedifferentareasofthebrainarecrystalcleartous.
The machines used in experiments are cheap enough that thousands of subjects can be
experimented on. And finally the data are unambiguous and doublechecked. What would be
thepotentialroleofbrainstudiesineconomictheory?Thesuggestionthisessayhasdefended
isthatthebrainwouldmattertoprediction.
A neurobiologicallyinformed model would not be in opposition to other models such as
Bicchieris. It would provide predictions more secure and informative by incorporating
parameters neglected by current models, which may significantly correlate to human
behaviour.WhencomparedtomodelslikeBicchieris,becauseofthenatureofitsparameters,
the potential advantage of a neurobiologicallyinformed model is that its grounds would be
more easily warranted by independent lines of research without the risk of accommodation,
anditspredictionswouldbequantitativelyaccurate.
One way to summarize my argument is that by observing brain area x activation we may
bypass such problems as the specification problem in models like Bicchieris. The problem
of specification is the problem of identifying the norms that prime a certain behaviour. For
example,wemayhaveasituationsuchas:

NORMA>BRAINACTIVATIONx>UGBEHAVIOURY
NORMB>BRAINACTIVATIONx>UGBEHAVIOURY

Ifdifferentnormsactivatethesamebrainareas,andthesearecorrelatedwiththesametypes
of behaviour, then by observing the brain we would have an instrument for prediction more
simpleandeffective.Theimportanceofmakinggoodpredictionsisthusthereasontoexplore
alternative models informed by neurobiological evidence. The hope is that these models will
accountforanomaliesandmakeinterestingnewpredictions.
However,thescenariojustenvisagedisideal.Neuroscienceisstillanimmaturescience,and
theworktodobeforetryingtointegrateneurobiologicalparametersintoamodelofeconomic
decisionmaking is a lot. This essay has tried to argue that this work would be important and
worthwhile.
4

BIBLIOGRAPHY

Bechtel W. (forthcoming) The epistemology of evidence in cognitive neuroscience. In R.


Skipper Jr., C. Allen, R. A. Ankeny, C. F. Craver, L. Darden, G. Mikkelson, and R.
Richardson (eds.), Philosophy and the Life Sciences: A Reader. Cambridge, MA: MIT
Press.
Bicchieri C. (2006) The Grammar of Society: the Nature and Dynamics of Social Norms,
CambridgeUniversityPress.

4
The bulk of this paper was written at the London School of Economics (UK) during my M.Sc. Alex
Voorhoeve,Francesco Guala provided generous comments on previous drafts. Cristina Bicchieri helped
meverykindlybyansweringmanyquestionsIhadabouthertheory.AtanearlierstageIdiscussedsome
oftheideascontainedherewithMatteoMotterlini.Asincerethankyoutoallofthemfortheirhelp.The
usualdisclaimersabouttheremainingerrorsapply.
MatteoColomboWhatcanNeuroscienceoffertoEconomics?

57

Bicchieri C. & Chavez, A. (2008) Behaving as Expected: Public information and fairness
norms, unpublished. URL: <
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1082264>,July2008.
Bicchieri C. & Xiao, E. (2008) Do the Right Thing: But Only if Others Do So, Journal of
behaviouraldecisionmaking,21:118.
Bolton G., & Ockenfels, A. (2000) ERC: A Theory of Equity, Reciprocity and Cooperation,
AmericanEconomicReview90:16693.
BradleyR.(unpublished)AttributingMentalStates.
CamererC.(2003)BehavioralGameTheory:ExperimentsonStrategicInteraction,Princeton
UniversityPress.
Camerer C. (2007) Neuroeconomics: Using Neuroscience to Make Economics Predictions.
EconomicJournal,117:C2642.
Chiao J.Y. & Ambady N. (2007) Cultural neuroscience: Parsing universality and diversity
across levels of analysis. In Kitayama, S. and Cohen, D. (Eds.) Handbook of Cultural
Psychology,GuilfordPress,NY.
Elman J. L., Bates E. A., Johnson, M. H., KarmiloffSmith, A., Parisi, D. & Plunkett, K. (1996)
RethinkingInnateness.MITPress.
FehrE.&Schmidt,K.(1999)ATheoryofFairness,CompetitionandCooperation,Quarterly
JournalofEconomics114:81768.
Forster M. (2008) Prediction, in S. Psillos & M. Curd (eds.), Routledge Companion to
PhilosophyofScience,London:Routledge.
Gazzaniga M. Ivry R., Mangun G., (2002), Cognitive Neuroscience, W.W. Norton & Co. Inc.,
NewYork.
Guala F. (2005) The Methodology of Experimental Economics, New York: Cambridge
UniversityPress.
Guala F. (2008) Paradigmatic Experiments: The Ultimatum Game. Philosophy of Science,
forthcoming.
Gth W., Schmittberger, R., & Schwarze, B. (1982) An experimental analysis of ultimatum
bargaining.JournalofEconomicBehaviorandOrganization,3(4),367388.
HarrisonG.(2008)NeuroeconomicsACriticalReconsideration.EconomicsandPhilosophy,
24:303344.
Henrich J., Boyd R., Bowles, S., Camerer C., Fehr E., & Gintis H. (eds. 2004) Foundations of
Human Sociality: Economic Experiments and Ethnographic Evidence from Fifteen
SmallScaleSocieties.Oxford:OxfordUniversityPress.
Hoffman E., McCabe K., Shachat, K., & Smith, V. (1994) Preferences, property rights and
anonymityinbargaininggames.GamesandEconomicBehavior,7:346380.
Koch C. & Ullman S. (1985) Shifts in selective visual attention: towards the underlying
neuralcircuitry,HumanNeurobiology4:219227.
KrolakSalmonP.etal.(2003)Anattentionmodulatedresponsetodisgustinhumanventral
anteriorinsula.Ann.Neurol.53,446453.

Humana.MenteIssue10July2009

58

Laudan L. (1990) DeMystifying Underdetermination, in W. Savage, ed., Scientific Theories.


Pp.26797.Minneapolis:UniversityofMinnesotaPress.
Pablo B.G., Ramn, C.R., & Almudena, D. (2006) Si l lo necesita: Gypsy fairness in
Vallecas.ExperimentalEconomics,V9(3),253264.
Phillips M.L. et al. (1997) A specific neural substrate for perceiving facial expressions of
disgust.Nature389,495498.
Quartz S.R. & Sejnowski, T.J. (1997) The neural basis of cognitive development: A
constructivistmanifesto.BehaviouralandBrainSciences20:537596
Rabin M. (1993) Incorporating Fairness into Game Theory and Economics, American
EconomicReview83:1281302.
RadderH.(1996)InandabouttheWorld:PhilosophicalStudiesofScienceandTechnology,
Albany:StateUniversityofNewYorkPress.
RescherN.(1997)PredictingtheFuture.Albany:StateUniversityofNewYorkPress.
Roth A., Prasnikar V., OkunoFujiwara, M. & Zamir, S. (1991) Bargaining and Market
Behavior in Jerusalem, Lubljana, Pittsburgh and Tokyo: An Experimental Study,
AmericanEconomicReview81:10681095.
Rubinstein A. (2006).Comments on Behavioral Economics, in Advances in Economic
Theory(2005WorldCongressoftheEconometricSociety),EditedbyR.Blundell,W.K.
NeweyandT.Persson,CambridgeUniversityPress,2006,volII,246254
Rubinstein A. (2008) Comments on Neuroeconomics, Economics and Philosophy, 24: 485
494.
SallyD.&HillE.(2006)Thedevelopmentofinterpersonalstrategy:Autism,theoryofmind,
cooperationandfairness,JournalofEconomicPsychology27(1):7397.
Sanfey A. G., Rilling, J. K., Aronson, J. A., Nystrom, L. E., & Cohen, J. D. (2003) The neural
basisofeconomicdecisionmakingintheultimatumgame.Science,300(5626),1755
1758.
Small D.M. et al. (2003) Dissociation of neural representation of intensity and affective
valuationinhumangestation.Neuron39,701711.
Solnick S.J. & Schweitzer, M.E. (1999) The influence of physical attractiveness and gender
on ultimatum game decisions. Organizational Behavior and Human Decision
Processes79(3):199215
Thaler R. H. (1988) Anomalies: The Ultimatum Game, Journal of Economic Perspectives, 2,
pp.195206.
Worrall J. (2006) History and TheoryConfirmation in J. Worrall and C. Cheyne (eds)
Rationality and Reality: Conversations with Alan Musgrave. Pp. 3161 Kluwer
AcademicPublishers,2006.
Zald D.H. & Pardo, J.V. (2000) Functional neuroimaging of the olfactory system in humans.
Int.J.Psychophysiol.36,165181.

59

ANeuroeconomicPerspectiveonCharitableGiving

DavidYokum
*

dyokum@email.arizona.edu

FilippoRossi
^

filippor@email.arizona.edu

ABSTRACT

Psychologists and economists, particularly those assuming that people are rational egoists, have
struggled to understand the causes of voluntary donation for decades. Why would a person decide to
sacrificepartofhisorhermaterialpayoffinordertoincreasethewellbeingofothers?Inthefirstpart
of this paper, we outline a core set of possible motivations, and then consider how those motivations
can be used to construct behavioral models that can also be tested in terms of what we know about
brain function. We emphasize the role of otherregarding preferences and argue that there are moral
judgments, independent of any consideration of payoffs, that partially determine when and to whom
suchpreferencesexist.Inthesecondpartofthepaper,wearguethataneuroeconomicperspectivecan
help understand charitable giving, and then discuss recent neuroimaging studies that demonstrate this
potential.

1.THECHALLENGEOFVOLUNTARYDONATION

Homoeconomicus,thatrationalcreatureconcernedonlywithitspersonal,materialpayoffs,is
not an ideal candidate from whom to elicit a charitable donation. Unless there are offsetting
material benefits, it will refuse to contribute anything to the public good. In this paper, we
outline evidence demonstrating that people are more charitable than predicted by rational
egoism alone, and then explore alternative models that use otherregarding motivations to
explainthedifference.Attentionisgiventounpackingthefeaturesdeterminingwhenandto
whom such motivations exist and, in particular, how neuroscientific evidence can inform the
debate.
Two caveats before we get started. First, for the moment, we are restricting egoism to
meanaselfcenteredfocusonachievingmaterialgainsforoneself,withoutanyregardforthe
interestsofotherpeople.Amoreinclusivedefinitionofegoismwouldallowanyselfcentered
focus,materialorotherwise,tocount.Forexample,itmightbethatthedecisionmakerdoes
not expect a material payback, but instead wants to feel the pleasure of giving money to the
needy. This desire is still selfish in that it seeks to obtain personal benefit the pleasurable
sensation rather than advance the welfare of the beneficiary. We will return to this
possibilityindetaillater.
Second, egoism and altruism are defined differently by different researchers. For instance,
biologists often discuss selfishness and altruism in terms of evolutionary fitness without any
reference to intentionality. A different question, the one which we will focus on, is whether
the proximate, psychological system motivating the decision to act altruistically is selfish,
altruistic, or both. It is a question of what beliefs and desires are driving behavior. To put it

*
NeuralDecisionScienceLabDepartmentofPsychologyTheUniversityofArizona
^
NeuralDecisionScienceLabDepartmentofPsychologyTheUniversityofArizona

Humana.MenteIssue10July2009

60

loosely,dohumanshaveacognitivelyrepresented,ultimatedesireoftheformIwantothers
to be welloff a desire that exists independently of their concern for their own wellbeing?
Or is the only ultimate desire instead I want to be welloff, perhaps coupled with auxiliary
beliefs that acting charitably will somehow advance that goal? To say that humans act
charitably because they are psychologically motivated by altruism is to say that they possess
an ultimate desire to advance others wellbeing, irrespective of their own wellbeing. They
have otherregarding preferences. Nonaltruistic explanations of charity, in contrast, either
deny altruistic impulses altogether or explain them away as instrumental desires: they exist
onlytoadvanceultimatedesiresthatareselfish.
Returningtotheoriginaldiscussion,whateveritsdescriptivelimitations,theassumptionof
rational egoism has stimulated the creation of formal models, such as in game theory,
1

1
that
have precise, testable predictions about how persons will behave toward one another. This
provides a useful benchmark from which to compare real behavior and, as we will see, helps
orientresearchaimedatunderstandingunderlyingneuralmechanisms.Publicgoodgamesare
immediatelyrelevantforourpurposes.Withinatypicalpublicgooddesign,eachindividualis
givenamonetaryendowmentandmustdecidewhetherornottocontributesomeportionof
that endowment to a common pool from which all participants will benefit; the transferred
amount is often increased by a stated factor (Leydard, 1995; Gchter et al. 2009). For
example, each subject receives ten euro, and any money donated to the common pool is
multipliedbytwobeforebeingevenlyredistributedbacktotheparticipants.Iftherearefour
subjects,andeveryonecontributeshisorherentireendowment,thenthefinalcommonpool
is4[10x2]=80euro,andeachindividualreceivesback20euroatwofoldincreasefromthe
initial endowment. This is obviously a better outcome for all parties than if no one
contributes, for in that case each individual remains with only ten euro. Nevertheless, Homo
economicus will contribute nothing. The reason is that the highest payoff is achieved if one
defectswhileeveryoneelsecontributes.Inthatcasethepayoffis10+(3[10x2])/4=25euro
afiveeuroincreasefromwhenpersonallycontributingaswell.Theotheregotisticalplayers,
of course, will also realize the benefits of unilateral deviation. The end result is that no one
contributes to the common pool (Dawes & Thaler, 1988). Such an undesirable outcome is a
Paretodeficient equilibrium, since everyone would be better off if no one defected, but no
oneactuallydoessobecauseoftheevenhigherpotentialbenefitsofunilateraldeviation.
Paretodeficiency is a classic obstacle in many oneshot games, and it is also the basis of
the freerider problem. In the context of charitable giving, it means that each person is
predicted to refuse donating time or money, relying instead on the contributions of other
people thereby getting a freeride to a public good made possible by others donations.
However,realbehaviorinlaboratoryandnaturalisticsettingsnegatesthisprediction(Leydard,
1995; Camerer, 2003; Andreoni, 1995). Rather than transferring zero euro in the oneshot
publicgoodgame,forexample,subjectstypicallydonate4060%oftheirendowment(Marwell
&Ames,1981).Comparableratesarealsocontributedinthefirstroundofafinitelyrepeated
versionofthegame(Isaac&Walker,1988;Kim&Walker1984).
2

Charitable behavior is not restricted to economic games either. A recent survey in the
United States revealed that 89% of households donated an average of $1,620, or 3.1% of
household income, to charitable organizations, and almost half of all American adults
volunteered their time to participate in such groups (Independent Sector, 2001). Such

1
SeevonNeumannandMorgenstern(1944)andOsborneandRubinstein(1998).
2
Foramoregeneraldiscussiononthepublicgoodgame,seeLedyard(1995)andCamerer(2003).

DavidYokum,FilippoRossiANeuroeconomicPerspective

61

estimates fluctuate across surveys and countries (Andreoni, 2006; 2008), but in all cases
charitablegivingiswellabovethepredictedlevelofzerocontribution.
Initially it was thought that the egoism assumption could be preserved at the expense of
rationality. For one reason or another, people make mistakes and donate to the public good
against their own best interest. Such a hypothesis is partially consistent with data from
repeated public good games. Although subjects in an iterated public good game make first
round donations that are comparable to those in a oneshot interaction, the donation rate
begins to decay with each subsequent round until, in the final round, most subjects are
actually behaving as egotistical players (Dawes & Thaler, 1988; Fehr & Schmidt, 1999; Kim &
Walker, 1984).It was therefore hypothesized that subjects were initially confused by a rather
oddexperimentalsetting,andthatbehaviorduringlaterrounds,oncetherulesarelearned,is
moreinformativeofwhatwouldactuallyhappeninnaturalisticsettings.
Andreoni (1995) directly addressed the possibility of confusion. He used three different
publicgoodgames:(a)astandardversion;(b)aversionwhereplayerswerepaidaccordingto
therelativerankoftheirearnings(itwas,inotherwords,transformedintoazerosumgame);
and (c) a modified standard version with rank information (which was inconsequential with
respect to payment, and meant only to control for the effects of information about rank on
behavior). Note that in condition (b) there is no incentive to cooperate, due to its zerosum
nature. Donation in this case can therefore be attributed to confusion, for it benefits no one.
Theresultsreplicatedthecommonfindingforthestandardcondition(a):cooperationdecayed
with each additional round, although donation rates never dropped below 25%. Subjects in
condition (b), however, quickly and drastically altered their behavior, donating less than 10%
by the fourth round. Analyses across conditions revealed that confusion was indeed one
explanatory factor for early round behavior, but that it was unable to fully explain the
persistent willingness to donate. Moreover, the observed late round defection was best
explained, not by learning, but rather as a reaction to freeriding: players initially try to
cooperate,butaftertheirattemptsareexploitedbyothers,theyturntodefectionasaformof
retaliation.
3

The socalled restarting effect (Andreoni, 1988; Cookson, 2000) provides further
evidence against the confusion hypothesis. If subjects are learning that donation is against
their own interests with each subsequent round of a public good game, then it should not
matter if the game is interrupted and then restarted. If players are learning that the best
strategy,forarationalegoist,isnottodonatethenwewouldpredictthatcooperationshould
constantly decline across rounds of the repeated game. If, on the other hand, subjects are
sensitive to the decisions of their partners (rather than merely ignorant of the game rules),
then a new start may reopen possibilities for cooperation. The experimental data confirms
that donation rates, contrary to the confusion hypothesis, do in fact bump back up to high
rateswhenthegameisstoppedandrestarted.Theupshotisthatconfusionperseisunableto
explaintheobservedratesofcontribution.


3
SeealsoGoereeetal.(2002)foraninvestigationofthisissuewithintheframeworkofstochasticgame
theory.Theyusedaseriesof(quasi)oneshotpublicgoodgamesthatvaried:(a)thenumberofplayers
inagroup;(b)themagnitudeofhowmuchaplayerreceivedinreturnforcontributingoneunitofhisor
herendowmenttothepublicgood;and(c)howmucheachotherplayerinthegroupreceivedfromthe
unit contributed by the player. Their data revealed that both altruism and behavioral noise are
explanatoryfactorsofbehaviorinapublicgoodsetting.SeealsoPalfreyetal.(1997).

Humana.MenteIssue10July2009

62

Substantial evidence therefore indicates that Homo economicus is an endangered species,


and that for most people the assumption of raw egoism focused on material payoffs simply
does not hold. So why, then, do people voluntarily donate their time and money to a public
good even when there are not obvious offsetting material benefits? The terrain of candidate
explanationsiscomplex,spanningdisciplinessuchaseconomics,psychology,andevolutionary
biology,andthereareavarietyoftermsusedbydifferentauthorsthatoverlapinsomeways
but not others. In the next section, we attempt to outline several major conceptual
possibilities, but make no claim to being exhaustive. Our aim is to capture a set of essential
motivations, even if necessarily oversimplified, in order to begin constructing behavioral
modelsthatcanalsobetestedintermsofwhatweknowaboutbrainfunction.

2.COMPONENTSOFCHARITABLEDONATION

It is worth noting that the demise of Homo economicus in no way obliterates the role of
selfishness.Mostpeopleactcharitablyundertherightcircumstances,butnoteveryonedoes.
Andforthosewithotherregardingpreferences,thereisnoreasontothinkthatsomedegree
ofselfishnessnolongerexists.Themotivationsoutlinedhereshouldthereforebeunderstood
as expanding rather than replacing selfish models. People act charitably for selfish reasons at
times,butnonselfishreasonsarealsonecessarytoexplainthelevelofgenerositythatexists,
which is well above that predicted by selfish models, as we saw in the previous section.
Bearingthisinmind,wenowturntofourcomponentsofcharitabledecisionsthataredifferent
from seeking personal material payoffs and warrant special scrutiny: warmglow, preferences
relatedtofairness,reciprocity,anddeservedness.

2.1WARMGLOW

James Andreoni (1989; 1990) provides a useful distinction, often cited in the charitable giving
literature, between pure and impure altruism. The dichotomy captures the possibility that a
person might donate either because an increase in a public good is desirable per se or,
alternatively, because he or she experiences a sort of selfish, personal satisfaction from the
veryactofgiving.Toconsiderthisformally,letw
i
representpersoniswealth,andassumethat
thereisoneprivateandonepublicgood.Wealthcanbespenteitheronpersonalconsumption
oftheprivategood,denotedx
i
,orgivenasacharitabledonationtothepublicgood,denoted
g
i
. Let G represent the overall public good and be equal to the sum of all donations, that is,
person is donation (g
i
) plus everyone elses donation. People can then be assumed to have a
utility function U
i
= U
i
(x
i
,G,g
i
), which they seek to maximize. What this means is that people
derive satisfaction from their private consumption (x
i
), the level of public good (G), and their
personalcontributiontothatpublicgood(g
i
),andthustheirdecisionofhowmuchtodonate,
ifatall,willreflectabalancebetweenthesethreeconsiderations.
Apersonissaidtobepurelyaltruisticifheorshecaresaboutthestatusofthepublicgood
but not how it is achieved, except insofar as it affects their level of private consumption. The
utilityfunctioninthiscasereducestoU
i
=U(x
i
,G).Apersonisapureegoist,ontheotherhand,
ifheorshecaresonlyaboutprivateconsumptionandpersonalcontribution.Inthiscase,U
i
=
U(x
i
,g
i
). The g
i
term, as part of the utility function, captures a phenomenon referred to as
warmglow. It reflects a type of satisfaction that is independent of that derived from
achievingagivenlevelofG.Itisinsteadasatisfactionderivedimmediatelyfromthegivingact
itselfagoodfeelingfrompersonallyhelpingout,sotospeak.Someonewhoisneitherpurely
DavidYokum,FilippoRossiANeuroeconomicPerspective

63

altruisticnorpurelyegotistical,thatis,apersonwhoissensitivetobothGandg
i
,isconsidered
tobeanimpurealtruist.
Whatwewanttostresshereisthewarmglowcomponentofthemodel.Thevariablefor
privateconsumption,x
i
,representsthecontinuedpresenceofselfishmotivationsincharitable
decisions,aswestressedattheonsetofsectiontwo.Andthesections2.2and2.3belowcould
beinterpretedasfurtherelaboratingwhatisrepresentedbyG,inotherwords,whatitmeans
to have truly otherregarding preferences. Warmglow is unique, however, in that it does not
cleanly fit into either of the traditional categories of selfishness and otherregarding
preference.Thereisnomaterialpayoffatstake,anditseemsoddtocallsomeonewhoenjoys
helping others selfish. Nonetheless warmglow is, in an important sense, entirely selfish. The
act of giving is merely a means to the end of personal satisfaction, and at bottom the well
being of others is not a concern. To see that this is the case, consider the following decision:
should you personally donate one euro to a charitable group or allow the charitable group to
receive instead onethousand euro from someone else. The person motivated only by warm
glow will select the former, despite the substantial monetary loss to the charitable
organization.

2.2SOCIALPREFERENCES

Theoriesoffairnessassumethatsubjectshavepreferencesregardingthewayinwhichcertain
resourcesareallocated.Anallocationthatsatisfiesthesepreferencesiscalledfair.Weassume
thatanunfairallocationcanmotivateadecisionmakertoactinacompensatorymanner,that
is, to seek the obtainment of a fair allocation. In this sense, fairness enters as a motivational
force in a persons preferences (Camerer, 2003, p. 114). A theory of charitable giving should
thereforeaccommodatetherelationshipbetweenvoluntarydonationandthetheoryofjustice
ofthedonators.
In the economical literature, four main types of social preferences are usually discussed
(see Charness & Rabin, 2001, for a review): self interested, competitive, socialwelfare, and
differencesaversion motives. Selfishness was considered above. A competitive motive is not
particularlyinterestingfromthepointofviewofvoluntarydonation:ifasubjectspreferences
aresuchthatshepreferstobebetteroffthanheropponents,itisunlikelyforhertodonate.
Letusthenfocusonthelasttwopreferencetypes.

Consider the linear model below, adapted from Charness and Rabin (2001), which can be
usedasasimplifiedrepresentationofdifferentkindsofdistributionalpreferences:

U
i
(
i
,
j
)=
j
(r+s)+
i
(1rs) (1)

In a twoplayer interaction, equation (1) entails that the utilities derived from material
payoffs for players i and j, denoted
i
and
j
, respectively, depends on: (a) the magnitude of
the two payoffs and (b) the parameters and , which provide a way of modeling different
kindsofdistributionalpreferences.Thestrategicinteractionisdescribedthroughrands.Ifis
materialpayoffisbiggerthanjs(i.e.,
i
>
j
),thenr=1;otherwiseitiszero.Ifinstead
i
<
j
,
thens=1;otherwiseitiszero.

Andreoni and Miller (2002) and Charness and Rabin (2001) proposed a quasimaximin
model of socialwelfare preferences, where subjects are concerned about their material
payoff, the material payoff of the individual with the lowest payoff in the society, and finally
thepublicgood(intermsofthemodelin(1),thistranslatesintothefollowingconstraints:1

Humana.MenteIssue10July2009

64

> 0, 1/2). The upshot of this model is that subjects prefer more for themselves, are
morelikelytohelpwhentheyarebetteroff,andpreferParetoimprovements.
4
Charnessand
Rabin (2001) found that the quasimaximin model was capable of organizing the data from a
battery of thirtytwo dictator and response games, and thus that it may capture important
featuresofpeoplesdistributionalpreferences.
Another important theory has been proposed by Fehr and Schmidt (1999), one that is
particularly pertinent to the analysis of public good games.
5
They present a two parameter
model in which subjects are concerned about the differences between their payoff and the
payoffs of the other players; it is essentially a type of selfcentered inequity aversion. The
intuition behind Fehr and Schmidts proposal is that when subjects have more of a given
resource, they are more willing to sacrifice that resource in order to compensate for payoff
differences. On the other hand, when subjects are behind, they are willing to hurt their
opponents in order to shrink the distance between the material payoffs. This can be written
formallyas:

U
i
()=
i

i
max{
j

i
,0}
i
max{
i

j
,0} (2)

Theparameter
i
capturesthedegreetowhichsubjectscareabouttheinequalityofthepayoff
when they are ahead, while
i
represents their concern when they are behind. A reasonable
constraintisthat
i
shouldbegreaterthanorequalto
i
.
6
FehrandSchmidtalsoassumethat0

i
< 1, which distinguishes their theory from a competitive preferences model, where
subjects prefer to be better off than other players. This model is also capable of explaining a
widevarietyofexperimentaldata.

The above models are highly simplified, and there is no agreement on which is best for
representingsocialpreferences.CharnessandRabinthereforeprovidesomesoundadvice:

Toomuchwillbelostifexperimentalistsjumptooquicklytocalibratinghighlysimplifiedmodel[]
At this stage, models ought to be developed that help to interpret psychologically sound and
empiricallyprevalentpatternsofbehaviorcommoninabroadarrayofgames.(2001,p.821)

Whatismostimportantforourpurposesistheexplicitrepresentationofsomesortoffairness
preference, namely, a concern about how resources are distributed that is sensitive to the
payoffs of others. Whether this motivation reflects a desire to enhance the wellbeing of the
worstoff,tosimplyavoidinequity,both,orsomeothersortofotherregardingpreferenceis,
atthisstage,toearlytodetermine.However,asexperimentaliststacklethisissue,wethinkit
necessary to keep in mind that people are unlikely to have social preferences that are
uniformly applied across people and situations. Such preferences are not blind, so to speak,
andassuchmodelsofcharitablegivingshouldbesensitivetoqualificationsaboutwhenandto
whomcharitablemotivationsapply.Weturnnowtothispossibility.

4
SeealsoYaariandBarHillel(1984).
5
SeealsoFehrandFischbacher(2002)andBoltonandOckenfelds(2000).
6
It would be surprising, after all, if people cared about inequality when they were ahead more than
when they were behind, for in the latter case selfish motivations would also be predicted as a
motivatingfactortoachieveequality.Also,asFehrandSchmidtnote,thisassumptionisinlinewiththe
loss aversion literature (e.g., Tversky and Kahneman, 1991), which reflects the common saying that
lossesloomlargerthangains.

DavidYokum,FilippoRossiANeuroeconomicPerspective

65

2.3RECIPROCITYANDDESERVEDNESS

The theories outlined above do not specify anything about the qualities of the recipients of a
charitableact,butinsteadfocusexclusivelyontheirmaterialpayoffs.Nonetheless,issuessuch
as whether the potential recipients deserve your help or satisfy some moral criteria are
importantcandidatesforinclusionintheconsiderationofwhetherornottoactcharitably(see
Rabin, 1993; Dufwenberg et al., 2004; Charness & Rabin, 2001). As Dufwenberg et al. (2004)
note:

The assumption that individuals only care about final distributions implies that they must be
indifferentconcerninghowdistributionscomeabout.Thisisproblematicifinfactindividualsregard
information about their coplayers specific choices or intentions as important to their decision
making.(2004,p.26070)

What determines whether someone deserves help? In the economic literature, deservedness
isstrictlyrelatedtofairnessintheformofreciprocity(Rabin,1983;Dufwenbergetal.,2004).
Charness and Rabin (2001), for example, focus their attention on withdrawal reciprocity. The
idea is that subjects will withdraw their willingness to act charitably toward people who are
unwilling to sacrifice for the sake of fairness. They observed that this kind of reciprocation is
particularlyimportantinthecaseofsimpledictatorandresponsegames.
In the game theoretic literature, Rabin (1993, p. 298) developed the notion of fairness
equilibrium on the basis of three intuitions: (a) people are willing to sacrifice their own
material wellbeing to reciprocate kindness; (b) they are also willing to sacrifice wellbeing to
punish unkindness; and (c) the first two conditions significantly affect human behavior.
Dufwenberg and Kirchsteiger (2004) developed this notion for extensive form games,
7
and
their formulation provides an example of the kind of formalism that may be implemented in
ordertoqualifywhenandtowhomotherregardingpreferencesapply.

Call thesetofplayerisstrategies,withNthesetofplayers.DenotewithAthesetof
strategyprofiles(theCartesianproductofthesetsA
i
).DufwenbergandKirchsteigerdefinetwo
componentsoftheutilityfunctionforplayeri:hismaterialpayoff
i
andthereciprocitypayoff.
This latter component depends on the players beliefs at different stages of the game
(Battigalli&Dufwenberg,2009).Inparticular,thereare:(a)thesetofpossiblebeliefsofplayer
i on js strategies, B
ij
= A
j
and (b) player is beliefs about player js beliefs about player ks
strategy, C
ijk
= B
jk
= A
k
. Players update these beliefs throughout the game on the basis of the
previousactionsoftheiropponents.Forexample,b
ij
(h)representsanupdatingofB
ij
basedon
h,apossiblehistoryinthegame.
From here we can define a kindness index, which can be used to discriminate the players
withwhomtobekind.Following(inpart)Rabinsoriginalformulation(1993),Dufwenbergand
Kirchsteigerdefineareferencespayoff(
e
j
)astheaveragebetweenthehighestandthelowest
payoffsthatplayericangivetoplayerjbyplayingoneofhisefficientstrategies(Dufwenberg
etal.,2004,p.276).Thedistanceandsignofthematerialpayoffofjfromthereferencepayoff
definesthekindnessofitowardj:


7
Thistheorybelongstothemoregeneralframeworkofpsychologicalgametheory;seeGenakopoloset
al.(1989)andBattigalliandDufwenberg(2009).

Humana.MenteIssue10July2009

66

k
ij
(a
i
(h),(b
ij
(h))
ji
)=
j
(a
i
(h),(b
ij
(h))
ji
) ((b
ij
(h))
ji
) (3)

A further term (analogous to (3)) can be constructed that measures whether player i thinks
thatjwaskindwithi(
iji
).Throughthematerialpayoffofiandthekindnessfunctions,wecan
finallydefinethefollowingutilityfunction:

(4)

The first term stands for is material payoff. The second term is of particular interest here.
Notice that k
ij
is negative when i behaved unkindly with j, while
iji
is negative when j was
unkindwithi.Whenonlyoneofthetermsisnegative,thevalueof decreases;thiscaptures
the fundamental intuition underlying reciprocity: a mismatch between kind and unkind
behaviortriggersaviolationofreciprocity.Finally,Y
ij
isanexogenousterm,whichmodelshow
much player i cares about reciprocity (when it equals zero, the utility function reduces to the
materialpayoffcomponent).
8

The relation between this approach to reciprocity and voluntary giving is indirect, but
particularly relevant. In particular, Dufwenberg and Kirchsteigers theory is one of the most
developed ways of formally imposing constraints on otherregarding preferences. This
psychological game theoretic approach can, in principle, accommodate the imposition of
furtherconstraintsbasedonthepropertiesthatarecipientshouldhavefromthepointofview
of the donator in order to be eligible for a charitable act. This would constitute a novel and
important addition. What we have in mind, specifically, is that moral considerations are
relevant,considerationsthatarerelatedto(butsubtlydifferentfrom)desert,ortheideathat
apersondeservessomethingbaseduponhisorheractions.
As an example of data motivating this hypothesis, consider the debate surrounding the
system of welfare in America. This is an ideal test case, for the welfare system constitutes an
egalitarian redistribution of income among total strangers, and to the extent that it is
supported through votes it is a voluntary system it is, in other words, an act of charity.
Nonetheless, popular support for welfare is mixed, with many individuals adamantly opposed
to its continuation. Why might people oppose welfare or, to put it differently, refuse to be
charitable in this case? Surprisingly, income level, education, and a variety of demographic
variables are incapable of adequately discriminating between supporters and detractors;
votersarenotreallyconcernedaboutthecostofwelfareorfraud;andevenagreeingwiththe
idea that income should be evenly redistributed is not a very good predictor. The
overwhelminglysignificantpredictorisinsteadtheanswertothefollowingquestion:doesbad
luck cause poverty? Those who believe that poor people are somehow responsible for their
povertyarefarmorelikelytoopposewelfare(seeFongetal.,2005,forafulldiscussion).This
formofresponsibility,orratherirresponsibility,cantakeseveralforms.Fongetal.(2005),for
example,reportsurveyresultsindicatingthat,bymorethanafivetoonemargin,respondents

8
The model as presented here is only a partial representation. For the full structure, see Dufwenberg
andKirchsteigersoriginalpaper(2004).

DavidYokum,FilippoRossiANeuroeconomicPerspective

67

believe that welfare recipients could obtain employment if they tried, 70% of respondents
believeitismorefinanciallyrewardingtostayonwelfarethangetajob,57%believewelfare
encourageslaziness,and60%believewelfareencourageoutofwedlockchildbirth.
The welfare survey data reveal a refusal to financially support people who, it is believed,
areengaginginsociallyunacceptablebehaviors,suchasrefusingtocontributetothecommon
goodviaemploymentorhavingchildrenoutsideofmarriage.Tosomeextenttheoppositionto
welfare can be interpreted in terms of reciprocity. What is meant by reciprocity requires
specification. Most authors use reciprocity to mean reciprocal altruism in particular, or a sort
oftitfortatinwhichthebenefactorexpectstherecipienttosomehowrepaythefavor,evenif
unknowingly. This is not the most parsimonious explanation, however. After all, welfare
supports anonymous strangers that any given individual is unlikely to ever meet, and even
many very wealthy persons who presumably stand little to gain support welfare. More
pertinent is what is referred to as strong reciprocity (Gintis, 2000; Gintis et al., 2005). The
theory is that people tend to behave prosocially and punish antisocial behavior, at a cost to
themselves, even when the probability of future interactions is extremely low, or even zero.
This is fundamentally different from the weak reciprocity associated with reciprocal altruism,
foritremainsamotivatingfactorevenwhenfuturecompensationisunlikelyorimpossible.
Strongreciprocityisdifferentfrompurealtruism,however,foritisstillconditional.Itisjust
thattheconditionisnotofequivalentworthasthebeneficentact.Whatthismeansisthatthe
possible recipient must satisfy certain criteria, for instance (to continue with the welfare
example) be willing to work. But it is not the same as saying the potential recipient has a
legitimate claim to compensation in the standard sense of reciprocity. In other words, the
recipient has not actually provided an offsetting benefit to the benefactor, nor does the
benefactor ever expect the recipient to return the favor. Rather, the benefactor has a true
otherregardingpreference,oneinwhichheorsheisgenuinelyconcernedwiththepayoffof
another individual that is independent of his or her own payoff but that concern,
importantly,onlyexistsforthosewhosatisfytherelevantmoralcriteria.
Selfishness, warmglow, preferences related to fairness, reciprocity, and deservedness are
therefore all crucial components in the decision to act charitably. Other factors might be
important as well, but these motivations, incorporated into economic models such as those
outlinedabove,provideagenerativestartingpointfortheneuropsychologicalinvestigationof
charitable giving. Indeed, it is difficult to even begin research into the neurobiology of
charitablegivingwithouttheaidofpreciselyformulatedmodels.Havingoutlinedsuchmodels,
wenowturntotheverynewandverysmallneuroscientificliteratureoncharitablegiving(see
Mayretal.,2009,forareview).

3.NEURALEVIDENCE

Neuroeconomics is an experimental approach that couples the mathematical precision and


simplifying assumptions of economic models with the methods and data from cognitive
neuroscience(Glimcheretal.,2009).Neuraldataisusefulforthestudyofcharitablegiving,as
wellasothereconomicbehaviors,forseveralreasons.Tobeginwith,neuralevidenceplacesa
constraintontherangeofviablepsychologicaltheories,sinceproposedcognitivemechanisms
must align with possible neurophysiological function. In essence this works by enhancing the
epistemic criteria of internal consistency: not only must the proposed cognitive mechanism
align with behavioral data, but now it must also mesh with theories about different brain
regionsandthefunctionstheyperform.Second,brainactivityduringcharitablegivingislikely
tooverlapwithbrainactivityobservedinothercontextsandforwhichthereisalreadyagood
deal of theoretical understanding. This means that our understanding of brain function in

Humana.MenteIssue10July2009

68

general can be used to help interpret the neural signals observed during charitable behavior
and, potentially, stimulate hypotheses for testing at cognitive and behavioral levels. Finally,
neural evidence might reveal motivations for charitable giving that are not accessible
behaviorally, either because subjects are unwilling to report or unconscious of the actual
impetusfortheirbehavior.
This final point is especially relevant for uncovering the psychological motivations driving
charitablebehavior.Asmentionedpreviously,egoismcanbebroadlyconstruedtoentailnon
material selfish benefits, such as warmglow. This is a legitimate interpretation, but the
consequenceisthategoismbecomesseeminglyimpossibletorefuteonthebasisofbehavioral
data alone. One problem is that selfreported reasons claiming genuine altruism can almost
always be reinterpreted as an instrumental desire toward a selfish, ultimate desire, one that
subjects may be unable to access. And it is difficult to bypass this obstacle with behavioral
evidence other than selfreport because a decision maker, whether motivated by warmglow
or a true otherregarding preference, will likely act the same. Helping other people for their
ownsakeanddoingitonlytogeneratewarmglowfeelingsbothrequireactuallyhelping.This
is not to say that behavioral data are irrelevant, only that it is difficult to disentangle
motivationaltheoriesthatmakesimilarbehavioralpredictions.
But what if we could more directly observe the motivational systems at work? Sober and
Wilson (1998, p. 205208) discuss the example of marine bacterium that are obligate
anaerobes (they cannot survive in the presence of oxygen) as an analogy for the problem of
teasing apart psychological egoism from psychological altruism.They note that the bacterium
could avoid oxygen by utilizing a device that directly detects oxygen or, alternatively, it could
useamagnetosomesensitivetothegravitationalfieldoftheearth(oxygenismoreabundant
near the water surface, so gravity could be used to as a guide in swimming away from the
surface). But bacteria with either an oxygen detector or a magnetosome would behave the
same they would swim downward. However, assuming we knew enough about bacterial
anatomy, we could resolve this dilemma by dissecting the bacterium and observing what
devicesareinside.Doesithaveanoxygendetectorordoesithaveamagnetosome?Soberand
Wilsonnotethatthisispossible,inprinciple,forthequestionofhowhumanaltruismoperates
inthebrainbut,writingadecadeago,rightlyconcludedthatevenifneurobiologywillanswer
this question one day, it offers little guidance now (p. 207). This, fortunately, is beginning to
change.
The neural investigation of charitable giving specifically is only just beginning. Jorge Moll
andcolleagues(2006)conductedoneofthefirstfunctionalmagneticresonanceimaging(fMRI)
studies. Nineteen subjects were endowed with U.S. $128. Each person was then presented
withaseriesofrealcharitableorganizations,includingabriefmissionstatementforeach,and
giventheopportunitytoeitherdonateoropposeadonationtoeachorganizationfromhisor
herendowment.Decisionswerestrictlyanonymous,andsubjectswereawarethatrealmoney
was at stake, both for themselves and the stated charitable organization. The design entailed
several possible payoff conditions for each decision: pure monetary reward (YOU: $+2, ORG:
$0),noncostlydonation/opposition(YOU:$0,ORG:$5),costlydonation/opposition(accepting
YOU:$2,ORG:$5;refusingYOU:$2,ORG$5).Notethatthepuremonetaryrewardcondition
hasnoconsequencesforthecharitableorganization,andthereforethedecisionshouldreflect
solely egotistical preferences. The noncostly condition, on the other hand, has no personal
consequences,andthereforethedecisionshouldreflectsolelyotherregardingpreferencesfor
the charity. The costly conditions entail a conflict between egotistical and otherregarding
motivations.
DavidYokum,FilippoRossiANeuroeconomicPerspective

69

There were two main results. First, the ventral tegmental area (VTA) and striatum,
componentsofthebrainsrewardsystem,wereactivatedbybothpuremonetaryrewardsand
decisions to donate. This finding suggests that charitable giving and personal gains share the
anatomical systems underlying reward reinforcement and expectancy. Second, a direct
contrast between the pure monetary reward and donation conditions revealed that neural
activity was greater in the ventral striatum for donations and, most intriguingly, that the
subgenual area was uniquely activated during donations. To the extent that the donation
conditionelicitstheexperienceofrewardaboveandbeyondthemonetarygaincondition,the
interpretation of warmglow motivation is reasonable. In other words, the enhanced ventral
striatumactivitymightrepresenttheexperienceofutilityfromtheactofgivingperse.
But the presence of warmglow does not rule out the existence of other motivations as
well. The subgenual area has been repeatedly implicated in the expression of social
attachment(e.g.,Bartels&Zeki,2004;Aronetal.,2005)andthereleaseofneuromodulators,
such as oxytocin and vasopressin, which are likewise thought to have important social
functions (e.g., Zak et al., 2005). It is therefore tempting to speculate that subgenual activity
reflectsneuralevidenceofsomeformofconditionalaltruismorstrongreciprocity.Atthevery
least, the subgenual area is a neuroanatomical region outside of the putative reward system,
andthuscharitabledecisionsentailatypeofprocessingthatisperhapsmeaningfullydifferent
fromtheutilityprocessingofpersonalpayoffs. Suchaconclusionreallyisspeculativethough.
Aside from awaiting replication, the interpretation of subgenual activity is somewhat
complicated by the fact that the charitable organizations used in the study were deliberately
chosenonthebasisoftheirsupportoroppositiontosociallycontroversialissues,forinstance
euthanasia,guncontrol,andabortion.Theobservedactivitymightthereforereflectareaction
tothesocialissuethatisindependentofthedecisiontodonate.Thisseemsunlikely,however,
andaswearguedabovesocialvaluesandmoralassessmentsareimportantconsiderationsin
thedecisionofwhetherornottodonatetoaparticularpublicgood.Toputitdifferently,the
subgenual activity might very well reflect a reaction to stimulating socially values, but that
reactionexistsasarelevantfactorinthedecisionofwhetherornottobehavecharitably.
Harbaugh et al. (2007) conducted a second neuroimaging study, this time without the
additional variable of moral assessment and with a design that more clearly tested the
possibilityofotherregardingpreferences.NineteensubjectswereendowedwithUS$100and
given a series of payoff situations which they could either accept or reject. The charity was a
foodbank.Therewerealso,however,threeuniqueconditionsthatdidnotrequireadecision:
ataxcondition,inwhichsubjectmoneywasinvoluntarilytransferredtothefoodbank;apure
mandatory payment to the subject, at no cost to the charity; and a pure mandatory payment
to the charity, at no cost to the subject. Activation of the reward system was again found for
both personal gains and gains to the charity. There was also greater striatum activity during
voluntary rather than mandatory donations, again supporting the existence of warmglow.
However, in a contrast between the mandatory treatments, activation in the ventral striatum
was found for both personal monetary gains and the mandatory tax. This latter activation
cannotpossiblybeinterpretedasawarmgloweffect.Thesubject,afterall,didnotvolunteer
todonatethemoney.
The most impressive finding, however, was the predictive model that Harbaugh et al.
tested. Recall the impure altruism utility function, U
i
= U
i
(x
i
,G,g
i
). They used neural activity
during the pure mandatory payments to each subject as an indicator of his or her marginal
utility of money (x
i
), and activity during pure mandatory payments to the charity as an
indicator of the marginal utility derived from increases in the public good (G); the transfers
weremandatory,sog
i
(personaldonation)wasirrelevant.Thosewithalargerneuralresponse
tothemandatorypersonalpayoffwerelabeledegoists,andthosewithalargerresponseto

Humana.MenteIssue10July2009

70

the mandatory payoff to the charity altruists. As predicted, the labeled altruists were more
charitable,givingnearlytwiceasmuch.Theauthorsreasonablyinterpretedthisasevidenceof
a purely altruistic motive. The larger the neural response to increases in G, no matter the
source(thatis,regardlessofg
i
),themorelikelyoneistogivevoluntarily.

4.CONCLUSIONS

We have argued in this paper that selfish motivations are only one component of decisions
relatedtocharitablebehavior.Peoplealsovaluewarmglowfeelingsandhavegenuineother
regarding preferences, such as concerns about how resources are distributed and strong
reciprocity. Such otherregarding preferences are not blind, however, and are instead
constrained by a variety of criteria related to the deservedness of the potential recipient.
These constraints are not related to material compensation but are rather, in an important
sense,moralconsiderations.
The Moll et al. (2006) and Harbaugh et al. (2007) studies are merely a taste of what is to
come from the neuroeconomic approach. These two studies demonstrated that it is possible
to identify neuroanatomical regions uniquely activated during conditions of selfishness, pure
altruism, and warmglow. The data at present do not conclusively resolve the psychological
egoismversusaltruismdebatebyanymeans,butthisworkispromising.Dissectingthebrain,
as we might to see if a bacterium has an oxygen detector or magnetosome, is not as far
fetched as it once seemed. For example, the evidence that reward areas of the brain are
activatedbymerelywitnessingacharitableactiscompellingevidencethathumansdo,infact,
donate for otherregarding reasons, contrary to the egoism hypothesis. To appreciate the
uniqueforceofthesedata,considersomeonewhoclaimstoenjoythefactthatanothergroup
is benefitted by charity as an end in itself. An advocate of egoism could dismiss this claim as
disingenuous; perhaps the person merely says such things to earn the approval of the
experimenter, or is somehow deluding him or herself. But to also dismiss the neuroimaging
data, the critic would have to argue that the subject is somehow capable of selfgenerating
striatal activity despite not actually experiencing a reward, or perhaps postulate a source of
striatal activation other than reward. There might be such explanations, but this is a harder
casetomake,andnowtheburdenofproofwouldliewiththeproponentofegoism.

BIBLIOGRAPHY

AndreoniJ.(2008)Charitablegiving.InS.N.Durlauf,&L.E.Blume(Eds.),TheNewPalgrave
DictionaryofEconomics(2ndEditioned.)PalgraveMacmillan.
Andreoni J. (1995) Cooperation in PublicGoods Experiments: Kindness or Confusion? The
AmericanEconomicsReview,83(4),891904.
Andreoni J. (1989) Giving with impure altruism: applications to charity and Ricardian
equivalence.JournalofPoliticalEconomy,97,14471458.
Andreoni J. (1990) Impure altruism and donations to public good: a theory of warmglow
giving.EconomicJournal,100,464477.
Andreoni J. (2006) Philanthropy. In S.C. Kolm, & J. M. Ythier (Eds.), Handbook of Giving,
Reciprocity,andAltruism(pp.12011269)Amsterdam:NorthHolland.
Andreoni J. (1988) Why Free Ride Strategies and Learning in PublicGoods Experiments.
JournalofPublicEconomics,37,291304.
DavidYokum,FilippoRossiANeuroeconomicPerspective

71

Andreoni J., & Miller, J. (2002) Giving According to GARP: An Experimental Test of
ConsistencyofPreferencesforAltruism.Econometrica,LXX,73753.
Aron A., Fisher, H., Mashek, D. J., Strong, G., Li, H., & Brown, L. L. (2005) Reward,
motivation, and emotion systems associated with earlystage intense romantic love.
JournalofNeurophysiology,94,327337.
Bartels A., & Zeki S. (2004) The neural correlates of maternal and romantic love.
NeuroImage,21,11551166.
Battigalli P. & Dufwenberg M. (2009) Dynamic Psychological Games. Journal of Economic
Theory,144,135.
Bolton G. & Ockenfels A. (2000) ERC: A Theory of Equity, Reciprocity and Competition.
AmericanEconomicReview,90,16693.
Camerer C. Behavioral Game Theory. Experiments in Strategic Interaction. Princeton, New
Jersey:PrincetonUniversityPress.
Charness G. & Rabin M. (2001) Understending Social Preferences with Simple Tests. The
QuarterlyJournalofEconomics,117(3),81769.
Cookson R. (2000) Framing Effects in Public Goods Experiments. Experimental Economics ,
3,5579.
Dawes R. & Thaler, R. (1988) Anomalies: Cooperation. The Journal of Economics
Perspectives,2(3),18797.
Dufwenberg M. & Kirchsteiger G. (2004) A theory of sequential reciprocity. Games and
EconomicBehavior,47,268298.
Fehr E. & Fischbacher U. (2002) Why Social Preferences Matter: The Impact of NonSelfish
MotivesinCompetition.TheEconomicJournal,112(478),C1C33.
Fehr E. & Gchter S. (2000) Cooperation and Punishment in Public Good Experiments. The
AmericanEconomicReview,90(4),98094.
Fehr E., & Schmidt K. M. (1999) Theories of fairness, competition, and cooperation.
QuarterlyJournalofEconomics,114,817868.
Fong C. M., Bowles S. & Gintis H. (2005) Reciprocity and the Welfare State. In H. Gintis, S.
Bowles, R. Boyd, & E. Fehr (Eds.), Moral Sentiments and Material Interests (pp. 277
302)Cambridge,MA:MITPress.
Gchter S. & Herrmann B. (2009) Reciprocity, Culture and Human Cooperation: Previous
Insights and New CrossCultural Experiment. Philosophical Transactions of the Royal
SocietyB(364),791806.
Genakopolos J., Pearce D., & Stacchetti E. (1989) Psychological Games and Sequential
Rationality.GamesandEconomicBehavior,1,6079.
Gintis H. (2000) Strong Reciprocity and Human Sociality. Journal of Theoretical Biology ,
206,16979.
Gintis H., Bowles S., Boyd R., & Fehr E. (Eds.) (2005) Moral Sentiments and Material
Interests.Cambridge,MA:MITPress.
Glimcher P. W., Camerer C. F., Fehr E. & Poldrack R. A. (Eds.) (2009) Neuroeconomics:
DecisionMakingandtheBrain.SanDiego:ElsevierInc.

Humana.MenteIssue10July2009

72

GoereeJ.,HoltC.&LauryS.(2002)PrivateCostsandPublicBenefits:UnravelingtheEffects
ofaltruismandNoisyBehavior.JournalofPublicEconomics,83,255276.
Harbaugh W. T., Mayr U., & Burghart D. R. (2007) Neural responses to taxation and
voluntarygivingrevealmotivesforcharitabledonations.Science,316,16221625.
Herrmann B. & Gtcher S. (2008) Antisocial Punishment Across Societies. Science , 319,
13627.
IndependentSector.(2001)GivingandvolunteeringintheUnitedStates2001.Washington,
DC:IndependentSector.
IsaacR.M.&WalkerJ.M.(1988)Groupsizeeffectsinpublicgoodsprovision:thevoluntary
contributionsmechanism.QuarterlyJournalofEconomics103,179199.
KimO.&Walker,M.(1984)TheFreeRiderProblem:ExperimentalEvidence.PiblicChoice,
43,324.
Leydard J. (1995) Public goods: a survey of experimental evidence. In J. H. Kagel, & A. E.
Roth (Eds.), The handbook of experimental economics (pp. 111194) Princeton, NJ:
PrincetonUniversityPress.
Marwell G. & Ames, R. (1981) Economists Free Ride, Does Anyone Else? Journal of Public
Economics,15,295310.
MayrU.,Harbaugh,W.T.,&Tankersley,D.(2009)Neuroeconomicsofcharitablegivingand
philanthropy. In P. W. Glimcher, C. F. Camerer, E. Fehr, & R. A. Poldrack (Eds.),
Neuroeconomics: Decision Making and the Brain (pp. 303320) San Diego, CA:
ElsevierInc.
MollJ.,Krueger,F.Zahn,R.Pardini,M.,deOliveiraSouza,R.,&Grafman,J.(2006)Human
frontomesolimbic networks guide decisions about charitable donation. Proceedings
oftheNationalAcademyofSciences,103(42),1562315628.
OsborneM.,&Rubinstein,A.ACourseinGameTheory.Cambridge,MA:MITPress.
Ostrom E., Walker, J., & Gardner, R. (1992) Convenants With and Without a Sword: Self
GovernanceisPossible.AmericanPoliticalScienceReview,86(2),40417.
PalfreyT.,&Prisbrey,J.(1997)AnomalousBehaviorinPublicGoodsExperiment:HowMuch
andWhy?AmericanEconomicReview,87,82946.
Rabin M. (1993) Incorporating Fairness into Game Theory and Economics. American
EconomicReview,83,12811302.
Sober E., & Wilson, D. S. (1998) Unto Others: The Evolution and Psychology of Unselfish
Behavior.Cambridge,MA:HarvardUniversityPress.
Tversky A., & Kahneman, D. (1991) Loss Aversion in Riskless Choices: A Refernces
DependentModel.TheQuarterlyJournalofEconomics,CVI,103962.
Von Neumann, J., & Morgenstern, O. (1944) Theories of Games and Economic Behavior.
Princeton:PrincetonUniversityPress.
YaariM.,&BarHillel,M.(1984)OnDividingJustly.SocialChoiceandWelfare,1,124.
Zak P. J., Kurzban R. & Matzner W. T. (2005) Oxytocin is associated with human
trustworthiness.HormonesandBehavior,48,522527.
73

Wby Cognitivists About Practical Reason can Unly be
Semi-Cognitivists



Clocomo Mollo
*

gmollo[umlch.edu


A8S1kAC1

uebaLes abouL Lhe naLure and Lhe role of raLlonallLy are parL of Lhe core of phllosophy of economlcs.
Poldlng raLlonal bellefs and belng lnsLrumenLally raLlonal are some of Lhe prlmary deslderaLa LhaL
economlc Lheory demands when modelllng Lhe behavlor of an agenL. hllosophy lLself, and especlally
phllosophy of acLlon, has also explored analogous lssues and can cerLalnly provlde subsLanLlal lnslghLs
abouL Lhe naLure of pracLlcal raLlonallLy. An agenL ls demanded Lo hold raLlonal bellef abouL Lhe world:
LheoreLlcal raLlonallLy requlres Lhe agenL Lo hold bellefs LhaL are conslsLenL and coherenL. Cn a pracLlcal
level, Lhe agenL ls capable of lnLendlng: Lhls acLlvlLy brlngs abouL Lhe sLaLe of affalrs LhaL he deslres.
8aLlonallLy urges Lhe agenL Lo hold lnLenLlons conslsLenLly wlLh Lhe seL of bellefs LhaL he ls accepLlng.
racLlcal raLlonallLy ln parLlcular demands Lhe agenL Lo respecL Lhe conslsLency and coherence
requlremenLs for lnLenLlons.
ln Lhe followlng paper l am golng Lo explore and crlLlclze cognlLlvlsm abouL pracLlcal reason, one of Lhe
mosL dlspuLed poslLlon ln acLlon Lheory. CognlLlvlsLs clalm LhaL we can make sense of Lhe raLlonal
requlremenLs for conslsLency and coherence of lnLenLlons by appeallng excluslvely Lo Lhe requlremenLs
of conslsLency and coherence of bellefs. racLlcal raLlonallLy ls ln some sense reduclble Lo LheoreLlcal
raLlonallLy. l wlll explore Lhls clalm and assess wheLher Lhls ls an accepLable posslblllLy. And of course, l
wlll polnL ouL LhaL Lhere ls space for some crlLlclsm.


kCLCGUL

CognlLlvlsm abouL pracLlcal reason ls Lhe poslLlon accordlng Lo whlch we can explaln Lhe
requlremenLs of conslsLency and coherence of lnLenLlons ln Lerms of Lhe requlremenLs of
conslsLency and coherence of bellefs. My maln lnLeresL ls Lo analyze Lhe valldlLy of Lhe
cognlLlvlsL's clalm LhaL demands of LheoreLlcal raLlonallLy (18) can fully explaln all demands of
pracLlcal raLlonallLy (8) and ln parLlcular Lo LesL wheLher Lhe cognlLlvlsL accounL of bellef-
lnvolvemenL of lnLenLlon can be successful ln explalnlng boLh conslsLency and coherence of
lnLenLlons.

l wlll sLarL by conslderlng whaL are assumed Lo be Lhe requlremenLs of pracLlcal and LheoreLlcal
raLlonallLy. l wlll use examples Lo lllusLraLe how exacLly Lhese requlremenLs work and l wlll
make clear whaL Lhey are supposed Lo accompllsh. l wlll Lhen conslder Lhe relaLlonshlp
beLween requlremenLs of 18 and requlremenLs of 8. 1hls wlll be followed by an explanaLlon
of how cognlLlvlsLs abouL lnLenLlons have aLLempLed Lo clalm LhaL raLlonal requlremenLs of
pracLlcal reason are grounded ln (or derlve from) raLlonal requlremenLs for LheoreLlcal reason.
l wlll conslder Lhe general case where cognlLlvlsLs aLLempL Lo show LhaL (1) locooslsteocy of
loteotloo coo be exploloeJ lo tetms of locooslsteocy of bellefs, and (2) locobeteoce of loteotloos

+
l would llke Lo Lhank rofessor Sarah 8uss (unlverslLy of Mlchlgan, Ann Arbor) for Lhe long and exclLlng
dlscusslons abouL Lhls Loplc and Lhe referees from Pumana.MenLe for Lhelr helpful crlLlclsms.
*
ueparLmenL of hllosophy, unlverslLy of Mlchlgan

Bumana.Nente - Issue 1u- }uly 2uu9

74

coo be exploloeJ by locobeteoce of bellefs. l wlll conslder cases (1) and (2) and assess Lhelr
plauslblllLy.
As for clalm (1), l wlll toke lnLo accounL an ob[ecLlon lnsplred by 8runero's kldnapplng
scenarlo
1
. l wlll make some lnLerpreLlve clarlflcaLlon of Lhe case - l wlll dlsLlngulsh beLween
lnLendlng noL Lo A and noL lnLendlng Lo A - and demonsLraLe LhaL Lhe cognlLlvlsL ls able Lo
provlde a sLrong case ln favour of hls poslLlon no maLLer whlch lnLerpreLaLlon we adopL.
Secondly, l wlll conslder clalm (2) and explaln how 8raLman
2
has ob[ecLed Lo Lhls poslLlon.
Pe shows how Lhere ls a posslblllLy Lo hold false bellefs abouL lnLenLlons: Lhls wlll creaLe a
slLuaLlon ln whlch we have coherence of bellefs ln 18 and lncoherence of lnLenLlons ln 8. 1hls
occurrence challenges Lhe cognlLlvlsL 's ldea LhaL coherence of lnLenLlons ls grounded ln
coherence of bellefs. l wlll ob[ecL Lo 8raLman by assesslng wheLher holdlng a false bellef abouL
lnLenLlon ls a genulne posslblllLy.
1he resL of Lhe paper wlll presenL an alLernaLlve aLLack agalnsL cognlLlvlsm. lf we assume
LhaL (a) Lhe sLandard formulaLlon of coherence of lnLenLlon ls correcL and (b) Lhe SLrong 8ellef
1hesls holds, l show LhaL we have good ground Lo dlspuLe Lhe clalm LhaL lncoherence of
lnLenLlons can be reduced Lo lncoherence of bellefs.
My concluslon ls LhaL lL ls noL posslble Lo be a compleLe cognlLlvlsL: conslsLency of
lnLenLlons ls sysLemaLlcally explalned from a cognlLlvlsL's perspecLlve and Lhe accounL ls able
Lo address a number of ob[ecLlons. lf however we conslder coherence of lnLenLlons, we can
see how no clalm abouL bellef-lnvolvemenL LhaL a cognlLlvlsL supporLs can posslbly reduce
lncoherence of lnLenLlons Lo lncoherence of bellefs.

1nLCkL1ICAL kA1ICNALI1 AND kAC1ICAL kA1ICNALI1: LACn nAS I1S CWN DLMANDS

lf an agenL wanLs Lo be raLlonal abouL hls own bellefs, he needs Lo respecL Lhe Lwo maln
requlremenLs of LheoreLlcal raLlonallLy: conslsLency of bellefs (CC8) and coherence of bellefs
(CCP8). lf a seL of bellef honours boLh demands of LheoreLlcal raLlonallLy Lhen Lhe seL ls
consldered raLlonal.
8ellef ConslsLency requlres LhaL all bellefs are noL muLually excluslve lf consldered LogeLher.
Cne posslble formulaLlon ls:

cO8. 1beotetlcol totlooollty JemooJs tbot lf ooe belleves p, ooJ belleves p-p
, tbeo ooe most oot belleve p.

We can lllusLraLe Lhls requlremenL by uslng an example.

1beotetlcol totlooollty tepoltes tbot ll yoo bolJ bellefs.
8ellef 1. jlts tololoq]
8ellef 2. lf jlts tololoq] tbeo NO1 j5tteets ote Jty] (l.e. 5tteets ote wet)

1nN

(lf yoo woot to tespect tbe cO8 tepoltemeot ) yoo Mu51 NO1 bolJ tbe bellef.
8ellef J. j5tteets ote Jty]

1
8runero, !. AgalnsL CognlLlvlsm abouL racLlcal 8aLlonallLy. lbllosopblcol 5toJles, AugusL 2008.
2
8raLman, M. lnLenLlons, 8ellefs, racLlcal, 1heoreLlcal. ln !. 1lmmerman, !. Sorupskl, S. 8oberLson, eds.
5pbetes of keosoo, Cu.
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

73


l Lake Lhls requlremenL as a wlde scope requlremenL: lf l hold bellefs 1, 2, 3 aL Lhe same Llme -
and Lherefore my bellefs are noL conslsLenL- Lhen ln order Lo be conslsLenL l musL dlscard
elLher bellef 1, bellef 2 or bellef 3.
3

Coherence of bellefs (CCP8 or 8ellef Closure) ls Lhe requlremenL LhaL you need Lo belleve
Lhe loglcal consequences of your bellefs. 8raLman eL al.
4
refer Lo lL as explanaLory coherence:
Lhe ldea ls LhaL each bellef x needs Lo have an explanaLlon (l.e. a serles of oLher bellefs ?,Z,W)
LhaL are able Lo accounL for bellef x. Cne posslble formulaLlon ls:

cOn8. 1beotetlcol totlooollty tepoltes tbot lf ooe belleves p, ooJ belleves tbot
p-p, tbeo ooe most belleve -p.

Agaln, we can use Lhe same example:

1beotetlcol totlooollty tepoltes tbot ll yoo bolJ bellefs.
8ellef 1. jlts tololoq]
8ellef 2. lf jlts tololoq] tbeo NO1j5tteets ote Jty] (l.e. 5tteets ote wet)

1nN

(lf yoo woot to tespect tbe cOn8 tepoltemeot ) yoo Mu51 bolJ tbe bellef.
8ellef J. NO1j5tteets ote Jty]

Cn Lhe oLher hand, pracLlcal raLlonallLy deflnes Lhe requlremenLs for a raLlonal sysLem of
lnLenLlon and of lnLenLlons and bellefs. lf an agenLs wanLs Lo be raLlonal abouL hls own bellefs
and lnLenLlons, he needs Lo respecL Lhe Lwo maln requlremenLs of pracLlcal raLlonallLy:
conslsLency of lnLenLlons (CCl) and coherence of lnLenLlons (CCPl). lf a seL of lnLenLlons and
bellefs honours Lhe demands of pracLlcal raLlonallLy Lhen Lhe seL ls consldered raLlonal.
ConslsLency of lnLenLlons explalns how lnLenLlons and bellefs are compaLlble lf Laken
LogeLher:

cOl. ltoctlcol totlooollty tepoltes tbot lf ooe loteoJs to x, ooJ belleves tbot lf
ooe wlll x tbeo ooe wlll oot y, tbeo ooe most oot loteoJ to y.

We can use an example Lo clarlfy Lhe noLlon:

ltoctlcol totlooollty tepoltes tbot ll yoo bolJ.
1. loteotloo . jl loteoJ to ttovel to ltoqoe tbls oftetoooo]
2.8ellef . jlf l ttovel to ltoqoe tbls oftetoooo, l wlll oot ttovel to New otk
tbls oftetoooo]


3
Cenerally speaklng, belng a wlde scope requlremenL slmply means LhaL Lhere are Lhree ways Lo
remove yourself from an lrraLlonal sLaLe: as menLloned you can dlscard elLher 1, 2 or 3. 1he scope of Lhe
requlremenLs ls noL aLLached Lo a speclflc aLLlLude - Lhls would be a narrow scope requlremenL- buL
ranges over a dls[uncLlon, ln Lhls case, of bellefs. lor a more deLalled LreaLmenL of Lhe sub[ecL see
8roome 1999, 2007 and kolodny 2003.
4
8raLman, M. lnLenLlons, 8ellefs, racLlcal, 1heoreLlcal. ln !. 1lmmerman, !. Sorupskl, S. 8oberLson, eds.
5pbetes of keosoo, Cu.

Bumana.Nente - Issue 1u- }uly 2uu9

76

1nN

(lf yoo woot to tespect tbe cOl tepoltemeot) yoo Mu51 NO1 bolJ tbe loteotloo.
J.loteotloo. jl loteoJ to ttovel to New otk tbls oftetoooo]

l Lake Lhls requlremenL also as a wlde scope requlremenL: lf l hold 1, 2, 3 aL Lhe same Llme
(and Lherefore my lnLenLlons and bellefs are noL conslsLenL) Lhen ln order Lo be conslsLenL l
musL dlscard elLher 1, 2 or 3.
Coherence of lnLenLlons (CCPl or Means-Lnd coherence) explalns whaL lnLenLlons we
should hold glven our bellefs and our oLher lnLenLlons:

cOnl. ltoctlcol totlooollty tepoltes tbot lf ooe loteoJs to x, ooJ belleves tbot lf
ooe wlll x, tbeo ooe most loteoJ to y, tbeo ooe most loteoJ to y.

Agaln we can use an example Lo clarlfy Lhe poslLlon:

ltoctlcol totlooollty tepoltes tbot ll yoo bolJ.
1.loteotloo. jl loteoJ to ttovel to ltoqoe tbls oftetoooo]
2. 8ellef. jlf l ttovel to ltoqoe tbls oftetoooo, tbeo l most loteoJ to boy o
tlcket to ltoqoe]

1nN

(lf yoo woo to tespect tbe cOnl tepoltemeot) yoo Mu51 bolJ tbe loteotloo.
J.loteotloo. jl loteoJ to boy o tlcket to ltoqoe]

Agaln l Lake CCPl Lo be a wlde scope requlremenL: lf l hold 1 and 2 and fall Lo hold 3, Lhen l am
ln a slLuaLlon of lncoherence of lnLenLlons. ln order Lo hold coherenL lnLenLlons l can elLher
abandon 1, abandon 2 or form lnLenLlon 3.

CCGNI1IVISM A8CU1 kAC1ICAL kLASCN

ls Lhere a relaLlonshlp beLween Lhe requlremenLs of LheoreLlcal raLlonallLy and Lhe
requlremenLs of pracLlcal raLlonallLy? CognlLlvlsLs generally Lhlnk LhaL we can appeal Lo
LheoreLlcal raLlonallLy Lo explaln Lhe requlremenLs of pracLlcal raLlonallLy.
1helr approach lncludes Lwo dlfferenL momenLs: ln Lhe flrsL place Lhey formulaLe a poslLlon
LhaL ls able Lo relaLe bellefs and lnLenLlons. Secondly, Lhey explaln why glven Lhls poslLlon lL ls
posslble Lo clalm LhaL Lhe raLlonal requlremenLs of pracLlcal raLlonallLy are grounded ln raLlonal
requlremenLs for LheoreLlcal raLlonallLy.

1here are Lhree dlfferenL ways of conslderlng Lhe relaLlon beLween bellefs and lnLenLlons. Lach
vlew has a dlfferenL ldea of whaL ls deflned as bellef-lnvolvemenL l.e. Lhe expllclL relaLlon
beLween lnLenLlon and Lhe bellef LhaL ls lnvolved when Lhe lnLenLlon ls formed.
Accordlng Lo Lhe ldenLlLy 1hesls (or ure CognlLlvlsm) Lhe lnLenLlon l lnLend Lo A" ls a bellef
LhaL l WlLL do A". lnLenLlons are some sorL of bellefs: Lhey colnclde wlLh a speclal klnd of
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

77

bellef. 1hls ls Laken Lo be Lhe sLrongesL poslLlon of Lhe Lhree. 1hls vlew can be found ln Lhe
early works of Parman (1976) and ln Lhe work of velleman
3
.
A dlfferenL poslLlon ls Lhe SLrong 8ellef 1hesls: Lhls clalms LhaL Lhe lnLenLlon l lnLend Lo A"
lovolves Lhe bellef LhaL l WlLL do A". 1he ldea ls LhaL lf l am ln Lhe sLaLe of lnLendlng Lo A, Lhen
l am also ln Lhe sLaLe of bellevlng LhaL l wlll A. unllke Lhe ldenLlLy 1hesls, an lnLenLlon ls a
separaLe enLlLy from a bellef. 1hls poslLlon flgures ln Lhe laLer work of Parman (1987).
LasLly, Lhe Weak 8ellef Lhesls clalms LhaL Lhe lnLenLlon l lnLend Lo A" lovolves Lhe bellef
LhaL lL ls posslble LhaL l WlLL do A". Poldlng an lnLenLlon requlres holdlng Lhe bellef LhaL lL wlll
be posslble LhaL l wlll perform Lhe acLlon LhaL l am lnLendlng. 1hls ls clearly Lhe weakesL
poslLlon.
ln general we can say LhaL Lhe relaLlon beLween lnLenLlon and bellef polnLed ouL by one
Lheory or Lhe oLher wlll Lell you how Lo derlve a cerLaln bellef from Lhe correspondlng
lnLenLlon. lor lnsLance lf you hold Lhe lnLenLlon l lnLend Lo A" Lhen, accordlng Lo Lhe SLrong
8ellef 1hesls, you do hold Lhe bellef LhaL l wlll A".
As polnLed ouL earller, a deflnlLlon of Lhe bellef-lnvolvemenL ls Lhe flrsL sLep Loward
explalnlng Lhe requlremenLs of pracLlcal raLlonallLy by appeallng Lo Lhe requlremenLs of
LheoreLlcal raLlonallLy. lL ls now Llme Lo see exacLly how Lhls movemenL ls accompllshed.


1. CCNSIS1LNC CI IN1LN1ICNS IS LkLAINLD 8 CCNSIS1LNC CI 8LLILIS

Conslder flrsL of all Lhe general case of lnconslsLency of lnLenLlons by uslng Lhe example l
presenLed earller ln Lhe LexL. lor Lhe sake of clarlLy l wlll dlsLlngulsh bellefs and lnLenLlons ln
Lwo separaLe columns.



8LLlLlS ln1Ln1lCnS

(b) 8ellef:
[lf l Lravel Lo rague Lhls afLernoon,
l wlll noL Lravel Lo new ?ork Lhls
afLernoon]

(a) lnLenLlon:
[l lnLend Lo go Lo rague Lhls
afLernoon]

(c) lnLenLlon:
[l lnLend Lo Lravel Lo new ?ork Lhls afLernoon]


1hls ls clearly a case of lnconslsLency ln pracLlcal reason: lf l lnLend Lo go Lo rague Anu l hold
Lhe bellef LhaL lf l Lravel Lo rague Lhls afLernoon Lhen l won'L Lravel Lo new ?ork, Lhen l cannoL
raLlonally hold Lhe lnLenLlon Lo Lravel Lo n? Lhls afLernoon.

LeL's now assume Lhe SLrong 8ellef 1hesls
6
and see how cognlLlvlsLs relaLe Lhe requlremenL of
conslsLency of lnLenLlon ln pracLlcal reason wlLh Lhe requlremenL of conslsLency of bellefs ln

3
velleman, u. ltoctlcol keflectloo, unlverslLy of Chlcago ress.
6
1he enLlre paper wlll focus on Lhe SLrong 8ellef 1hesls verslon of cognlLlvlsm. 1hls ls Lhe poslLlon ln
beLween Lhe radlcal and Lhe weak poslLlons and, as Lhe reader wlll see laLer ln Lhe LexL, lL seems Lhe
mosL promlslng when lL comes Lo successfully address cerLaln ob[ecLlons agalnsL cognlLlvlsm. l wlll
brlefly show how Lhe ldenLlLy 1hesls can be problemaLlc buL l wlll noL Lake lnLo conslderaLlon Lhe Weak
8ellef 1hesls.

Bumana.Nente - Issue 1u- }uly 2uu9

78

LheoreLlcal reason. 1he bellef-lnvolvemenL prlnclple ln Lhe case of Lhe SLrong 8ellef Lhesls ls:
tbe loteotloo l loteoJ to A lovolves o bellef tbot l wlll Jo A. lf we conslder Lhe example, Lhen
Lhe lnLenLlon [l lnLend Lo go Lo rague Lhls afLernoon] (a) locloJes ln some sense Lhe bellef [l
wlll go Lo rague Lhls afLernoon] (a*) by Lhe SLrong 8ellef 1hesls. 8y Lhe same Loken [l lnLend
Lo Lravel Lo new ?ork Lhls afLernoon] (c) locloJes Lhe bellef [l wlll Lravel Lo new ?ork Lhls
afLernoon] (c*). ln a schemaLlc form:


8LLlLlS ln1Ln1lCnS

(o*) 8ellef.
jl wlll ttovel to ltoqoe]

by SLrong 8ellef
1hesls
(a) lnLenLlon:
[l lnLend Lo Lravel Lo rague Lhls
afLernoon]
(b) 8ellef:
[lf l Lravel Lo rague Lhls
afLernoon, l wlll noL Lravel Lo
new ?ork Lhls afLernoon]


(c*) 8ellef.
[l wlll ttovel to New otk tbls
Aftetoooo ]

by SLrong 8ellef
1hesls
(c) lnLenLlon:
[l lnLend Lo Lravel Lo new ?ork Lhls
afLernoon]


lL ls aL Lhls polnL clear how Lhe lnconslsLency of lnLenLlon LhaL occurred beLween (a), (b) and (c)
can be explalned by Lhe lnconslsLency of bellefs (a*), (b) and (c*).

1nL kIDNAING CASL

LeL me now Lake lnLo conslderaLlon one speclflc ob[ecLlon Lo cognlLlvlsm. 8runero
7
polnLs ouL
LhaL Lhe SLrong 8ellef 1hesls enLalls LhaL bellevlng LhaL one wlll x ls oecessoty for lnLendlng Lo
x. Pe Lhen clalms LhaL Lhls ls problemaLlc: lL ls ln facL posslble Lo show LhaL bellevlng one wlll x
ls noL sufflclenL for lnLendlng Lo x.
8
1hls means LhaL 8runero lnLerpreLs Lhe SLrong 8ellef 1hesls
as saylng LhaL bellevlng LhaL one wlll x ls sofflcleot for lnLendlng Lo x. 1hls laLLer poslLlon ls
vulnerable Lo a number of dlfferenL counLerexamples. 8runero descrlbes cases ln whlch lL ls
posslble Lo belleve LhaL one wlll x wlLhouL hlm lnLendlng Lo x. Suppose a gangsLer ls now
kldnapplng me. As he Lles me up l hear from hls drlver LhaL we are golng Lo Lhe gangsLer's
hldeouL. ln Lhls case l do noL lnLend Lo go Lo Lhe gangsLer's hldeouL buL l belleve LhaL l wlll go
Lo Lhe gangsLer's hldeouL
9
. We hold a bellef LhaL l wlll x wlLhouL lnLendlng Lo x. 1hls ls conLrary
Lo Lhe clalm LhaL bellevlng LhaL one wlll x ls sufflclenL for lnLendlng Lo x. 8runero ls rlghL ln
clalmlng LhaL bellevlng one wlll x ls oot sofflcleot for lnLendlng Lo x. 8uL ls lL really Lrue ln Lhe

7
8runero, !. AgalnsL CognlLlvlsm abouL racLlcal 8aLlonallLy. lbllosopblcol 5toJles, AugusL 2008.
8
lbld. AccotJloq to tbe 5ttooq 8ellef 1besls, bellevloq tbot ooe wlll x ls oecessoty fot loteoJloq to x. 8ot
ptoblems fot tbe obove ottempt to ose 8c jbellef cooslsteocy] to explolo lc jloteotloo cooslsteocy] otlse
ftom tbe foct tbot bellevloq ooe wlll x ls oot sofflcleot fot loteoJloq to x. lll otqoe bete tbot (o) bellevloq
ooe wlll x ls oot sofflcleot fot loteoJloq to x j.].
9
lbld. 5otely l Jo oot loteoJ to qo to tbe qooqstets blJeoot, eveo tbooqb l belleve l wlll.
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

79

flrsL place LhaL - as 8runero seems Lo lnLerpreL lL - Lhe SLrong 8ellef 1hesls enLalls LhaL holdlng
a bellef ls sofflcleot for lnLendlng?

l belleve LhaL Lhe SLrong 8ellef 1hesls cannoL be undersLood as 8runero lnLerpreLs lL: 1here ls
no elemenL ln lLs deflnlLlon LhaL would suggesL how holdlng a bellef would be sofflcleot Lo
produce an lnLenLlon. All lL clalms ls LhaL tbe loteotloo l loteoJ to A lovolves tbe bellef tbot l
wlll Jo A. lrom Lhls formulaLlon lL ls clear LhaL Lhere ls ln facL a necesslLy relaLlon, namely LhaL
lnLendlng Lo A" necessarlly requlres a bellef LhaL l wlll A". 1here ls however no sufflclency
relaLlon LhaL Lransplres from Lhe orlglnal deflnlLlon: noLhlng Lells us LhaL formlng a bellef ls
sufflclenL Lo produce a correspondlng lnLenLlon. ln hls ob[ecLlon, 8runero seems Lo lnLerpreL
Lhe deflnlLlon of Lhe SLrong 8ellef 1hesls as tbe loteotloo l loteoJ to A ONl lovolves o bellef
tbot l wlll Jo A. lf we conslder hls lnLerpreLaLlon, Lhen we can deduce a sufflclency relaLlon
beLween bellef and lnLenLlon: lf an lnLenLlon Lo x only lnvolves a bellef Lo x, Lhen a bellef Lo x
can be sald Lo be sufflclenL Lo produce Lhe lnLenLlon Lo x. 1hls lasL formulaLlon ls equlvalenL Lo
Lhe ldenLlLy 1hesls buL lL ls lnaccuraLe as an lnLerpreLaLlon of Lhe SLrong 8ellef 1hesls.
ln concluslon, 8runero's kldnapplng case - as lL ls presenLed- ls cerLalnly a good
counLerexample Lo Lhe ldenLlLy 1hesls buL lL ls lnadequaLe Lo challenge Lhe SLrong 8ellef
1hesls.

l sLlll belleve LhaL Lhe kldnapplng case ls an lnLeresLlng example and l wanL Lo propose an
alLernaLlve way ln whlch lL can sLlll challenge cognlLlvlsm. As we have seen, cognlLlvlsLs clalm
LhaL lL ls posslble Lo explaln lnconslsLency of lnLenLlons by appeallng Lo lnconslsLency of bellefs.
ln Lhe kldnapplng case we have a slLuaLlon ln whlch l do nC1 lnLend Lo go Lo Lhe gangsLer's
hldeouL buL l do belleve LhaL l wlll go Lo Lhe gangsLer's hldeouL.
lf we assume Lhe SLrong 8ellef 1hesls and aLLempL Lo see whlch bellef ls lnvolved ln (a) Lhen
we wlll have a slLuaLlon as Lhe followlng:


8LLlLlS ln1Ln1lCnS


(o*) 8ellef.
jl wlll NO1 qo to tbe
qooqstets blJeoot

(b) 8ellef:
[l wlll go Lo Lhe gangsLer's
hldeouL]



by SLrong 8ellef 1hesls

(a) lnLenLlon:
[l do noL lnLend Lo go Lo Lhe
gangsLer's hldeouL]

1hls ls problemaLlc ln LhaL we can observe conslsLency ln pracLlcal reason LhaL ls nC1 reflecLed
ln conslsLency ln LheoreLlcal reason slnce (a*) and (b) are lnconslsLenL. We have no reason Lo
say LhaL lnLenLlons are lnconslsLenL, buL lf we apply Lhe SLrong 8ellef 1hesls Lhen we cannoL
doubL LhaL we are holdlng lnconslsLenL bellefs. 1hls ls clearly a problem for cognlLlvlsm.

"IN1LNDING NC1 1C A"

lL ls posslble Lo respond Lo Lhls ob[ecLlon on dlfferenL grounds. LeL's now focus on how we can
lnLerpreL lnLenLlon (a) ln Lhe example. rlma facle, by l do noL lnLend Lo go Lo Lhe gangsLer's

Bumana.Nente - Issue 1u- }uly 2uu9

80

hldeouL" l mlghL mean l lnLend noL Lo go Lo Lhe gangsLer's hldeouL" or l lnLend Lo absLaln
from golng Lo Lhe gangsLer's hldeouL". l have formed my lnLenLlon and my lnLenLlon ls Lo noL
go Lo Lhe gangsLer's hldeouL. Slnce we do have an lnLenLlon, a cognlLlvlsL would leglLlmaLely
apply Lhe SLrong 8ellef 1hesls Lo undersLand whaL bellef ls lnvolved by LhaL lnLenLlon. ln a
schemaLlc form:

As polnLed before, we see how Lhls lnLerpreLaLlon presenLs a problem for Lhe cognlLlvlsL's
accounL: we are presenLed wlLh lnconslsLenL bellefs (a*) and (b) and wlLh lnLenLlon (a). We
have Lhus a case ln whlch lnconslsLency of bellefs ls noL reflecLed ln lnconslsLenL lnLenLlons.

AL Lhls polnL l belleve Lhe cognlLlvlsL can sLlll provlde a response Lo Lhls counLerexample.
1hls would requlre Lo analyze Lhe exacL kloJ of bellef LhaL ls lnvolved by lnLenLlon (a).
velleman
10
suggesLs an lnLeresLlng dlsLlncLlon beLween how we undersLand lnLenLlons and
how we undersLand predlcLlons. undersLandlng someone's lnLenLlon Lo do A requlres
undersLandlng Lhe evldences LhaL shows LhaL he wanLs Lo do A. Cn Lhe oLher hand,
undersLandlng a predlcLlon LhaL he wlll do A lnvolves undersLandlng Lhe evldences LhaL he wlll
do A loJepeoJeotly of Lhe wlll of Lhe agenL Lo do A.

LeL's now see how Lhls dlsLlncLlon can help cognlLlvlsLs Lo respond Lo Lhe kldnapplng case.
1he quesLlon ls wheLher bellef (a*) ls Lhe correcL formulaLlon of Lhe klnd of bellef lnvolved ln
Lhe lnLenLlon [l ln1Lnu nC1 Lo go Lo Lhe gangsLer's hldeouL]. lollowlng velleman's dlsLlncLlon
we would need Lo be clear on Lhe wlll of Lhe agenL when aLLempLlng Lo undersLand hls
lnLenLlons. uoes (a*) really helps undersLandlng Lhe evldences LhaL shows wheLher Lhe agenLs
wanLs (or doesn'L wanL) Lo go Lo Lhe gangsLer's hldeouL or noL? l belleve LhaL we can produce
an alLernaLlve formulaLlon of Lhe bellef lnvolved ln (a):








10
velleman, u. letbops tbe Jlffeteoce ls tbot oo oqeots loteoJloq to Jo sometbloq ls ootmolly to be
exploloeJ by bls wootloq to Jo lt, wbeteos bls pteJlctloq tbot be wlll Jo sometbloq ls to be exploloeJ by
bls bovloq evlJeoces tbot be wlll. ltoctlcol keflectloo, unlverslLy of Chlcago ress. ages 94-93.

8LLlLlS ln1Ln1lCnS


(o*) 8ellef.
jl wlll NO1 qo to tbe
qooqstets blJeoot

(b) 8ellef:
[l wlll go Lo Lhe gangsLer's
hldeouL]


by SLrong 8ellef 1hesls

(a) lnLenLlon:
[l ln1Lnu nC1 Lo go Lo Lhe
gangsLer's hldeouL]
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

81




We have here applled Lhe SLrong 8ellef 1hesls and formulaLed Lhe correcL klnd of bellef LhaL ls
lnvolved ln lnLenLlon (a). 8ellef (a**) expresses Lhe wanLs of Lhe agenL holdlng lnLenLlon (a): lf
Lhe agenL could choose, Lhen he wlll noL go Lo Lhe gangsLer's hldeouL l.e. he wlll noL go Lo Lhe
gangsLer's hldeouL by hls own wlll. lrom a cognlLlvlsL perspecLlve, Lhls ls Lhe exacL bellef LhaL ls
lnvolved by lnLenLlon (a): lL helps us undersLandlng Lhe evldences LhaL shows wheLher or noL
Lhe agenL ls wllllng Lo go Lo Lhe gangsLer's hldeouL. 8ellef (b) on Lhe conLrary ls noL derlved
from Lhe agenL's declslons or wanLs buL lL ls a predlcLlon glven a cerLaln sLaLe of Lhe world
(namely Lhe facL LhaL Lhe agenL ls Lled up slLLlng ln Lhe gangsLer's car). lL lnvolves
undersLandlng Lhe evldences LhaL he wlll go Lo Lhe gangsLer's hldeouL loJepeoJeotly of hls wlll
Lo go Lo Lhe gangsLer's hldeouL.

lL ls now flnally clear how (a**) and (b) are conslsLenL. 8y dlsLlngulshlng beLween klnds of
bellefs Lhe cognlLlvlsL has one way Lo successfully respond Lo Lhe kldnapplng case and
lnvalldaLe lLs role as a counLerexample.

"NC1 IN1LNDING 1C A"

1here ls however a second posslble lnLerpreLaLlon of l do noL lnLend Lo go Lo Lhe gangsLer's
hldeouL" (a) ln Lhe kldnapplng example. 1hls mlghL ln facL mean LhaL Lhe agenL folleJ Lo form
Lhe lnLenLlon Lo A. ln oLher words Lhe agenL ls lncapable Lo produce Lhe lnLenLlon l lnLend Lo
A". Pe does nC1 embark ln Lhe process of lnLenLlon formaLlon aL all and Lhere ls no lnLenLlon
whaLsoever LhaL appears ln hls pracLlcal reasonlng.

ln Lhe llghL of Lhe laLLer formulaLlon of (a), we can conslder lLs lmpllcaLlons for cognlLlvlsm. 8y
l do noL lnLend Lo go Lo Lhe gangsLer's hldeouL" we mean ls LhaL l do noL form Lhe lnLenLlon Lo
go Lo Lhe gangsLer's hldeouL. 1here ls no lnLenLlon whaLsoever ln Lhe plcLure:








8LLlLlS

ln1Ln1lCnS

(o**) 8ellef.
jl wlll NO1 qo to tbe
qooqstets blJeoot 8 M
OwN wlll]

(b) 8ellef.
jl wlll qo to tbe qooqstets
blJeoot]

by SLrong 8ellef 1hesls

(a) lnLenLlon:
[l do nC1 lnLend
Lo go Lo Lhe gangsLer's hldeouL]

8LLlLlS ln1Ln1lCnS

(b) 8ellef: [l wlll go Lo Lhe
gangsLer's hldeouL]




(a) lnLenLlon: [.]

Bumana.Nente - Issue 1u- }uly 2uu9

82


lL ls evldenL LhaL Lhe kldnapplng case does noL pose any problems for cognlLlvlsm: we lack an
lnLenLlon LhaL could be reduced Lo a bellef Lhrough a bellef-lnvolvemenL prlnclple. under Lhls
lnLerpreLaLlon Lhe lssue abouL conslsLency does noL arlse ln Lhe flrsL place. 1he cognlLlvlsL ls
Lherefore lmmune Lo Lhe counLerexample.
We have consldered Lhe cognlLlvlsL's clalm LhaL conslsLency of lnLenLlons can be explalned
by conslsLency of bellefs. We have LesLed Lhe clalm by provldlng a counLerexample Lo lL. We
have also seen how Lhe cognlLlvlsL has dlfferenL ways Lo respond Lo Lhe challenges and flrmly
defend hls maln clalm. lL seem Lherefore plauslble Lo conclude LhaL Lhere are solld cases LhaL
Lhe cognlLlvlsL can employ Lo show LhaL we can explaln Lhe conslsLency of lnLenLlons ln Lerms
of conslsLency of bellef. l am noL here clalmlng LhaL Lhls ls Lhe rlghL way of explalnlng
conslsLency of lnLenLlons ln pracLlcal reason. WhaL l am Lrylng Lo convey ls LhaL Lhe cognlLlvlsL
seems Lo have Lhe Lools Lo defend hls explanaLlon of conslsLency of lnLenLlons ln pracLlcal
reason.

2. CCnLkLNCL CI IN1LN1ICNS IS LkLAINLD 8 CCnLkLNCL CI 8LLILIS

CognlLlvlsLs clalm LhaL lncoherence ln pracLlcal reason can be reduced Lo lncoherence ln
LheoreLlcal reason: lf we percelve an lncoherence of lnLenLlons Lhen Lhere musL be a parallel
lncoherence ln Lhe sysLem of bellefs LhaL Lhe agenL ls holdlng. 1hls laLLer wlll be able Lo explaln
Lhe former.
LeL's conslder Lhe general case of lncoherence of lnLenLlons as 8raLman
11
presenLs lL. We
have seen LhaL pracLlcal raLlonallLy demands LhaL lf l lnLend Lo L" and l hold Lhe bellef LhaL lf
one wlll L, Lhen one wlll lnLend Lo M" (M ls Lhe means l musL Lake Lo accompllsh end L), Lhen l
am requlred Lo form Lhe lnLenLlon l lnLend Lo M". lf you do noL lnLend Lo M Lhen your
lnLenLlons are lncoherenL. 1ake Lhe lnlLlal example of lncoherence of lnLenLlons and rephrase lL
Lo flL 8raLman's formulaLlon. lf l hold Lhe lnLenLlon [l lnLend Lo go Lo rague Lhls afLernoon] (a),
hold Lhe bellef LhaL [buylng a LlckeL ls necessary Lo go Lo rague] (b) and [lnLendlng Lo buy a
LlckeL ls necessary Lo buy a LlckeL](c) and sLlll hold [l do noL lnLend Lo buy a LlckeL] (d), Lhen my
lnLenLlon-bellef sysLem ls lncoherenL. LeL me represenL Lhe slLuaLlon by hlghllghLlng lLs Lwo
maln elemenLs, bellefs and lnLenLlons. l wlll use agaln Lhe lefL column for bellefs and Lhe rlghL
column for lnLenLlons:

8LLlLlS ln1Ln1lCnS

(b) 8ellef:
[8uylng a LlckeL ls necessary Lo Lravel Lo
rague]

(c) 8ellef:
[lnLendlng Lo buy a LlckeL ls necessary Lo buy
a LlckeL]
12


(a) lnLenLlon: [l lnLend Lo Lravel Lo rague Lhls
afLernoon]


(d) lnLenLlon: [l do noL lnLend Lo buy a LlckeL]

11
8raLman,M. lnLenLlon, 8ellef, racLlcal, 1heoreLlcal. ln !. 1lmmerman, !. Sorupskl, and S. 8oberLson,
eds. 5pbetes of keosoo, Cu.
12
l am lncludlng bellef (c) Lo be falLhful Lo Lhe characLerlzaLlon LhaL 8raLman provldes. lL can be polnLed
ouL LhaL Lhls bellef ls by no means requlred Lo make Lhe sysLem of bellefs/lnLenLlons coherenL as lL does
noL appear ln Lhe orlglnal formulaLlon of CCPl presenLed earller ln Lhe LexL.
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

83



Clven Lhe formulaLlon of coherence of lnLenLlons lllusLraLed above, lL ls now relaLlvely
sLralghLforward Lo see how Lhls ls a case of lncoherenL lnLenLlons. l do noL lnLend Lo buy a
LlckeL when Lhe demands of pracLlcal raLlonallLy requlre me Lo do so.

LeL's now assume Lhe SLrong 8ellef 1hesls and see how cognlLlvlsLs relaLe Lhe requlremenL
of coherence of lnLenLlon ln pracLlcal reason wlLh Lhe requlremenL of coherence of bellefs ln
LheoreLlcal reason. 1he poslLlon clalms LhaL: Lhe lnLenLlon l lnLend Lo A" lnvolves Lhe bellef
LhaL l wlll do A". lf we conslder Lhe example menLloned above, Lhen we can see how by Lhe
SLrong 8ellef 1hesls Lhe lnLenLlon [l lnLend Lo Lravel Lo rague Lhls afLernoon] (a) lnvolves Lhe
bellef [l wlll Lravel Lo rague Lhls afLernoon] (a*). 8y Lhe same Loken [l do noL lnLend Lo buy
Lhe LlckeL] (d) lnvolves Lhe bellef [l wlll noL buy Lhe LlckeL] (d*). ln a schemaLlc form:


lrom a cognlLlvlsL's sLandpolnL, we can Lherefore observe how Lhe lncoherence of lnLenLlons
LhaL do noL respecL Lhe demands of pracLlcal raLlonallLy - seL (a), (b), (c), (d) - can be seen as
lncoherence of bellefs LhaL do noL respecL Lhe demands of LheoreLlcal raLlonallLy - seL (a*), (b),
(c), (d*).
8raLman challenges Lhls accounL by presenLlng a case ln whlch we observe coherence of
bellefs and, aL Lhe same Llme, lncoherence of lnLenLlons.

8kA1MAN AGAINS1 CCGNI1IVISM

ln loteotloo, 8ellef, ltoctlcol, 1beotetlcol "
13
8raLman argues LhaL lL ls noL posslble Lo reduce a
case of lncoherence ln pracLlcal raLlonallLy Lo a case of lncoherence ln LheoreLlcal raLlonallLy.
CognlLlvlsm seems Lo allow Lhe posslblllLy of holdlng a false bellef abouL one's own lnLenLlon:

13
8raLman, M. lnLenLlons, 8ellefs, racLlcal, 1heoreLlcal. ln !. 1lmmerman, !. Sorupskl, S. 8oberLson, eds.
5pbetes of keosoo, Cu.
8LLlLlS

ln1Ln1lCnS

(o*) 8ellef.
jl wlll ttovel to ltoqoe]


(b) 8ellef:
[8uylng a LlckeL ls necessary
Lo Lravel Lo rague]

(c) 8ellef:
[lnLendlng Lo buy a LlckeL ls
necessary Lo buy a LlckeL]

(J*) 8ellef.
jl wlll NO1 boy o tlcket]

by SLrong 8ellef 1hesls











by SLrong 8ellef 1hesls

(a) lnLenLlon:
[l lnLend Lo Lravel Lo rague
Lhls afLernoon]









(d) lnLenLlon:
[l do noL lnLend Lo buy a LlckeL]


Bumana.Nente - Issue 1u- }uly 2uu9

84

Lhls wlll generaLe a slLuaLlon ln whlch even Lhough someone's bellefs are coherenL, her
lnLenLlons wlll be lncoherenL. 1hls ls how 8raLman argues Lo supporL Lhls clalm:

Suppose l lnLend L and know LhaL L requlres boLh M and LhaL l lnLend Lo M. lf l sLlll do noL
lnLend M my lnLenLlons suffer from means-end lncoherence. 8uL suppose LhaL l ln facL do noL
lnLend M, l neverLheless falsely belleve LhaL l lnLend M. So my bellefs are LhaL L, LhaL L requlres
boLh M and LhaL l lnLend M, and LhaL M. 1here ls no lncoherence (Lhough Lhere ls falslLy) ln Lhls
sLrucLure of bellefs. So means-end coherence ls noL bellef coherence.
14


lf cognlLlvlsm allows Lhe posslblllLy of holdlng a false bellef regardlng one's lnLenLlons (l.e. l
falsely belleve l lnLend M when ln facL l don'L) Lhen we have a slLuaLlon where we have
cobeteoce lo oot system of bellefs ooJ ot tbe some tlme we bolJ locobeteot loteotloos ('l lnLend
Lo L' and 'l do noL lnLend M'). Slnce coherence of lnLenLlons derlves (accordlng Lo cognlLlvlsm)
from coherence of bellefs Lhen coherence ln one's bellefs sysLem should be teflecteJ ln one's
sysLem of lnLenLlons. Accordlng Lo 8raLman, Lhe posslblllLy of holdlng a false bellef precludes
Lhls parallellsm. ln a schemaLlc form:


As we can observe, by lnLroduclng false bellef (e) we are faced wlLh a perfecLly coherenL
sysLem of bellefs and aL Lhe same Llme wlLh a sysLem of lncoherenL lnLenLlons. 1hls would
counL as a counLerexample Lo Lhe explanaLlon LhaL cognlLlvlsLs provlde Lo reduce lncoherence
of lnLenLlons Lo lncoherence of bellefs.



14
lbld.
8LLlLlS

ln1Ln1lCnS

(a*) 8ellef:
[l wlll Lravel Lo rague (L)]

(b) 8ellef:
[8uylng a LlckeL (M) ls
necessary Lo Lravel Lo rague]

(c) 8ellef:
[lnLendlng Lo buy a LlckeL ls
necessary Lo buy a LlckeL]




(e) jlAl5] 8ellef.
jl loteoJ to boy tbe tlcket]

(f) 8ellef.
jl wlll boy tbe tlcket]

by SLrong 8ellef 1hesls
















(a) lnLenLlon: [l lnLend Lo Lravel
Lo rague Lhls afLernoon]









d) lnLenLlon:
[l do noL lnLend Lo buy a LlckeL]

uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

83


AGAINS1 8kA1MAN

llrsL of all lL ls necessary Lo esLabllsh whaL ls Lhe lnLerpreLaLlon of 8raLman's clalm ln facL l do
noL lnLend M" or, ln my example, l do noL lnLend Lo buy Lhe LlckeL" (d). 1wo are Lhe
posslblllLles lf we follow Lhe dlsLlncLlon LhaL l presenLed earller beLween holdlng Lhe lnLenLlon
l lnLend noL Lo A" and falllng Lo form Lhe lnLenLlon l lnLend Lo A".
Cn one hand 8raLman mlghL mean 'l lnLend nC1-M' or 'l lnLend Lo nC1 buy a LlckeL'. lf Lhls
ls Lhe case, Lhen hls poslLlon cannoL posslbly hold. Slnce Lhe agenL ls holdlng an lnLenLlon Lo
noL perform an acLlon, Lhe cognlLlvlsL mlghL clalm LhaL (glven LhaL lnLenLlon lnvolves bellef) Lhe
agenL needs Lo have a bellef of Lhe klnd l wlll noL buy Lhe LlckeL". 1hls wlll produce
lncoherence also ln LheoreLlcal reasonlng and lL would dlsmlss 8raLman's counLerexample. 1he
lnLerpreLaLlon however mlsses Lhe very ldea of whaL we have descrlbed as lncoherence.
Cn a second lnLerpreLaLlon Lhe senLence l do noL lnLend Lo buy Lhe LlckeL" (d) mlghL be
paraphrazed as l dld noL form Lhe lnLenLlon Lo buy Lhe LlckeL". 1hls ls cerLalnly more plauslble
and ls able Lo susLaln 8raLman's llne of reasonlng. 1he relevance glven Lo Lhe facL LhaL
lncoherence of lnLenLlons enLalls my falllng Lo form Lhe rlghL lnLenLlon wlll be laLer helpful Lo
supporL Lhe crlLlque LhaL l wlll move agalnsL cognlLlvlsm. As for now leL's conslder how we can
crlLlclse 8raLman from a dlfferenL perspecLlve even lf we allow Lhls second lnLerpreLaLlon.
lollowlng Lhe second lnLerpreLaLlon, Lhe flnal plcLure would be:


?our lnLenLlons are lncoherenL ln LhaL you lAlL Lo form Lhe lnLenLlon LhaL you musL form
accordlng Lo Lhe demand of pracLlcal raLlonallLy. AL Lhe same Llme your bellefs are coherenL ln
LhaL you uC lC8M Lhe bellef LhaL you musL form accordlng Lo Lhe demand of LheoreLlcal
8LLlLlS

ln1Ln1lCnS

(a*) 8ellef:
[l wlll Lravel Lo rague (L)]

(b) 8ellef:
[8uylng a LlckeL (M) ls
necessary Lo Lravel Lo rague]

(c) 8ellef:
[lnLendlng Lo buy a LlckeL ls
necessary Lo buy a LlckeL]




(e) jlAl5] 8ellef.
jl loteoJ to boy tbe tlcket]

(f) 8ellef.
jl wlll boy tbe tlcket]

by SLrong 8ellef 1hesls
















(a) lnLenLlon: [l lnLend Lo Lravel
Lo rague Lhls afLernoon]









(d) lnLenLlon: [.]
[should be: l lnLend Lo buy a
LlckeL}


Bumana.Nente - Issue 1u- }uly 2uu9

86

raLlonallLy. lf we accepL Lhls lnLerpreLaLlon, Lhen we can appreclaLe how 8raLman's case ls a
counLerexample Lo cognlLlvlsm.

1here are Lwo maln llnes of reasonlng LhaL can be used Lo aLLack 8raLman's counLerexample.
Cn one hand, a cognlLlvlsL mlghL show LhaL a closer analysls of Lhe example presenLed wlll
reveal coherence ln pracLlcal reason LogeLher wlLh coherence ln LheoreLlcal reason. Larller ln
Lhls paper we consldered dlfferenL verslons of cognlLlvlsm: ldenLlLy 1hesls, SLrong 8ellef 1hesls
and Weak 8ellef 1hesls. lf 8raLman's plcLure ls a counLerexample Lo cognlLlvlsm Lhen lL needs
Lo challenge all Lhree verslons of cognlLlvlsm. lL ls clear LhaL Lhe counLerexample challenges Lhe
SLrong 8ellef 1hesls and Lhe Weak 8ellef 1hesls. 8uL whaL abouL Lhe ldenLlLy 1hesls? As we
have seen, accordlng Lo Lhe ldenLlLy 1hesls (or ure CognlLlvlsm) Lhe lnLenLlon l lnLend Lo A"
a bellef LhaL l WlLL do A". lnLenLlons are some sorL of bellefs: Lhey colnclde wlLh a speclal
klnd of bellef. 1he bellef l wlll buy Lhe LlckeL" (f) Lhe same as Lhe lnLenLlon l lnLend Lo buy
Lhe LlckeL". We are Lhus holdlng Lhe lnLenLlon l lnLend Lo buy Lhe LlckeL" ln pracLlcal reason as
Lhe coherence requlremenL would command us Lo do. 1hls ls preclsely Lhe case because my
lnLenLlon l lnLend Lo buy Lhe LlckeL" ls Lhe same as my bellef l wlll buy Lhe LlckeL". lf Lhls ls
Lrue, Lhen we are presenLed wlLh coherence of bellefs and coherence of
lnLenLlons. 1hls ls conLrary Lo whaL 8raLman ls clalmlng. 1here ls Lherefore aL leasL one verslon
of cognlLlvlsm LhaL ls noL effecLlvely challenged by 8raLman's counLerexample.

A second way Lo aLLack 8raLman ls Lo show LhaL we are acLually presenLed wlLh lncoherence of
bellefs ln LheoreLlcal reason as well as lncoherence of lnLenLlons ln pracLlcal reason. 8raLman
clalms LhaL whaL makes bellefs coherenL - even when lnLenLlons are lncoherenL - ls Lhe
posslblllLy of holdlng a false bellef abouL one's lnLenLlon. ln Lhe followlng parL l wlll aLLack Lhls
clalm and show LhaL Lhere ls no genulne posslblllLy for Lhe agenL Lo hold false bellefs abouL
fuLure lnLenLlons. lf my argumenL ls correcL Lhen 8raLman's counLerexample Lo cognlLlvlsm wlll
lose lLs appeal.
LeL's begln by breaklng Lhe ldea of 'falsely bellevlng LhaL l lnLend Lo buy Lhe LlckeL' lnLo lLs
maln componenLs. l am now golng Lo make some baslc assumpLlons LhaL wlll help supporLlng
Lhe resL of Lhe argumenL. llrsL of all lL ls cruclal Lo dlsLlngulshlng bellevlng and lnLendlng and
deLermlne whlch ls Lhe funcLlonal role of each elemenL ln our menLal economy. 8ellevlng ls a
cognlLlve sLaLe: l am ln a menLal sLaLe ln whlch l have reasons Lo belleve someLhlng abouL Lhe
world. l have evldences LhaL l Lake as supporLlve for my bellef and l accepL Lhls bellef as Lrue.
1he role of a cognlLlve sLaLe ls Lo accepL or re[ecL lnformaLlon abouL Lhe world on Lhe basls of
evldences. lnLendlng, on Lhe oLher hand, ls a moLlvaLlonal sLaLe: lL affecLs my lmmedlaLe
behavlor and, generally speaklng, lL lnlLlaLes Lhe acLlon LhaL ls Lhe conLenL of my lnLenLlon. l
Lake lnLenLlon Lo be Lhe flrsL sLep Loward a consclous acLlon. lLs role ls Lo move my behavlor
Loward a speclflc dlrecLlon. 1hls ls an exLremely broad dlsLlncLlon buL lL ls helpful Lo
undersLand exacLly Lhe klnd of bellef LhaL we hold when we hold a bellef abouL our own
lnLenLlon. lrom Lhls characLerlzaLlon lL follows LhaL l cannoL be ln a moLlvaLlonal sLaLe LhaL
applles Lo Lhe fuLure excluslvely: slnce lnLenLlon ls lmmedlaLely moLlvaLlonal, Lhen my
lnLenLlons can only concern my presenL behavlor.
13
l can also be ln a cognlLlve sLaLe LhaL Lells

13
l can lnLend Lo perform an acLlon ln Lhe fuLure buL Lhls lnLenLlon ls noL dlsconnecLed from my presenL
acLlons. l can lnLend Lo boll an egg ln Lhe fuLure buL Lhls lnLenLlon has an lmmedlaLe moLlvaLlonal lmpacL
on Lhe means LhaL l lnLend Lo use now Lo reach my goal. l Lake Lhe lnLenLlon Lo perform a fuLure acLlon
Lo be necessarlly connecLed wlLh Lhe presenL moLlvaLlonal sLaLe of Lhe agenL. 1hls ls compaLlble wlLh
8raLman's ldeal of fuLure-dlrecLed lnLenLlons buL lL ls noL Lhe focus of Lhe dlscusslon. See 8raLman 1987.
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

87

me someLhlng abouL my fuLure moLlvaLlonal sLaLe. 1hls ls preclsely Lhe case ln whlch l belleve
LhaL l lnLend Lo buy Lhe LlckeL. l am now ln a cognlLlve sLaLe: l belleve someLhlng abouL Lhe
world, ln Lhls speclal case l belleve someLhlng abouL my fuLure self. l am ln a cognlLlve sLaLe
whose conLenL ls a fuLure moLlvaLlonal sLaLe. l am holdlng Lhe bellef LhaL l lnLend Lo buy Lhe
LlckeL: Lhls ls a parLlcular klnd of bellef, namely a bellef oboot my own fuLure lnLenLlons.

LeL's now examlne how a cognlLlve sLaLe - bellef - abouL a moLlvaLlonal sLaLe can be made
Lrue or false. ln oLher words whaL l am asklng ls how exacLly can l Lruly or falsely belleve LhaL l
lnLend Lo buy Lhe LlckeL. 8raLman argues LhaL Lhe maln problem wlLh cognlLlvlsm ls LhaL lL
allows cases ln whlch we can hold false bellefs abouL lnLenLlons l.e. cases where an agenL
falsely belleve he lnLends Lo x when he ln facL he does noL lnLend Lo x. unforLunaLely he does
noL fully explaln how cognlLlvlsm can allow such a posslblllLy and ln parLlcular how exacLly
bellef abouL lnLenLlons can be made Lrue or false. We can however lnfer Lhe underlylng plcLure
LhaL 8raLman seems Lo lmply when he clalms LhaL we can hold false bellefs abouL lnLenLlons.
Conslder Lhe example ln whlch l hold Lhe bellef LhaL l lnLend Lo drlnk Lhe coffee LhaL ls ln fronL
of me ln one mlnuLe. As Lhe Llme comes l fall Lo lnLend Lo drlnk Lhe coffee. l mlghL clalm LhaL l
am falsely bellevlng LhaL l lnLend Lo drlnk Lhe coffee ln one mlnuLe: l am lacklng LhaL very
moLlvaLlonal sLaLe when Lhe Llme comes. 1he presence of Lhe lnLenLlon Lo drlnk Lhe coffee
when Lhe proper Llme comes ls whaL makes Lhe bellef abouL my lnLenLlon Lrue or false. lf, ln
facL, l formed Lhe lnLenLlon Lo drlnk Lhe coffee, Lhen l would have LruLhfully held Lhe bellef LhaL
l lnLended Lo drlnk Lhe coffee ln one mlnuLe. 1he bellef abouL my lnLenLlon ls made Lrue or
false by Lhe presence of LhaL very lnLenLlon. 1he bellef l belleve l lnLend Lo drlnk Lhe coffee ln
one mlnuLe" ls made Lrue or false by Lhe presence or Lhe absence of Lhe lnLenLlon l lnLend Lo
drlnk Lhe coffee" when Lhe proper Llme comes. Pence Lhe posslblllLy of holdlng a false bellef
abouL one's lnLenLlon l.e. LhaL l lo foct Jo oot loteoJ M when aL Lhe same Llme l
oevettbeless folsely belleve tbot l loteoJ M.
16


l belleve LhaL Lhls ls a faulLy characLerlzaLlon and LhaL Lhere ls no genulne posslblllLy Lo hold a
false bellef abouL one's lnLenLlon lf we conslder Lhls plcLure. 1o show Lhls l wlll conslder Lwo
maln cases, Lhe synchronlc case and Lhe dlachronlc case. LeL's sLarL wlLh Lhe synchronlc case:
Lhe bellef abouL lnLenLlon Lo do x and Lhe lnLenLlon Lo do x lLself occur aL Lhe same Llme. 1hlnk
of an example ln whlch l lnlLlaLe an acLlon by my lnLendlng and aL Lhe same Llme l am fully
aware of lL. We can say LhaL Lhe agenL ls ln Lhe cognlLlve sLaLe l belleve aL L
1
LhaL l lnLend Lo
buy Lhe LlckeL aL L
1
". LeL's say LhaL L
1
ls exacLly now: l am ln Lhe moLlvaLlonal sLaLe l lnLend Lo
buy Lhe LlckeL aL L
1
". AL Lhe same Llme, by observlng myself as belng moved by LhaL lnLenLlon, l
have good evldence for Lhe presence of my lnLenLlon and Lherefore good grounds Lo supporL
Lhe bellef - l.e. be ln Lhe cognlLlve sLaLe - l belleve aL L
1
LhaL l lnLend Lo buy Lhe LlckeL aL L
1
". lf
aL L
1
l held Lhe bellef l belleve aL L
1
LhaL l lnLend Lo buy Lhe LlckeL aL L
1
" wltboot aL Lhe same
Llme Lhe presence of Lhe lnLenLlon l lnLend Lo buy Lhe LlckeL aL L
1
", Lhen l would have a
slLuaLlon ln whlch l falsely belleved abouL my lnLenLlon. ls Lhls lasL case a genulne posslblllLy? l
am lncllned Lo say no: lf l do noL hold Lhe lnLenLlon Lo do x aL L
1,
Lhen l do noL have any reason
Lo form a bellef LhaL l hold Lhe lnLenLlon Lo do x aL L
1.
1hls would mlsrepresenL whaL ls golng on
ln my pracLlcal reason. ln general, Lo raLlonally hold a bellef abouL an lnLenLlon, one requlres
Lhe presence of Lhe lnLenLlon LhaL Lhe bellef ls speclfylng. ln order Lo see wheLher LhaL
lnLenLlon ls presenL Lhe agenL needs evldences. l belleve LhaL many elemenLs can counL as
evldences LhaL you are/are noL holdlng an lnLenLlon: Lhese lnclude belng sponLaneously aware

16
8raLman, M. lnLenLlons, 8ellefs, racLlcal, 1heoreLlcal. ln !. 1lmmerman, !. Sorupskl, S. 8oberLson, eds.
5pbetes of keosoo, Cu.

Bumana.Nente - Issue 1u- }uly 2uu9

88

of your own lnLenLlon, consclously observlng you own sysLem of lnLenLlons by careful
lnLrospecLlon, or seelng yourself lnlLlaLlng an acLlon rlghL now. lf by lnLrospecLlon l cannoL
flnd any lnLenLlon Lo llfL Lhe cup ln fronL of me now - or l do noL see myself lnlLlaLlng now Lhe
acLlon of llfLlng Lhe cup ln fronL of me - lL would be lrraLlonal Lo hold Lhe bellef l belleve now
LhaL l lnLend Lo llfL Lhe cup ln fronL of me now". ln Lhe synchronlc case we are ln a slLuaLlon ln
whlch Lhe presence of Lhe lnLenLlon l lnLend Lo do x now" supporLs Lhe bellef l belleve now
LhaL l lnLend Lo do x now". 1he absence of such lnLenLlon does noL allow Lhe bellef abouL LhaL
lnLenLlon Lo emerge ln Lhe flrsL place. ln Lhls case raLlonallLy demands LhaL bellefs abouL
lnLenLlon are elLher supporLed by acLual lnLenLlon or are noL formed aL all. 1herefore ln Lhe
synchronlc case Lhere ls no posslblllLy Lo hold a false bellef abouL lnLenLlon ln Lhe sense LhaL
8raLman emphaslzes.

1hls ls Lhe slmplesL case and 8raLman mlghL reply LhaL lL does noL reflecL Lhe scenarlo LhaL he
was presenLlng. WhaL he ls concerned wlLh, as we have seen, are cases of bellefs abouL fuLure
lnLenLlons. ln Lhose cases, he wanLs Lo clalm LhaL we can hold false bellefs abouL fuLure
lnLenLlons. LeL's call Lhls caLegory dlachronlc cases l.e. slLuaLlon ln whlch Lhe bellefs abouL
lnLenLlon and Lhe lnLenLlon lLself occur ln separaLe Llmes. 1ake for example Lhe bellef: l
belleve now LhaL l lnLend Lo buy Lhe LlckeL aL L
2
" - where L
2
ls ln Lhe fuLure. l am ln a cognlLlve
sLaLe now LhaL expresses someLhlng abouL my moLlvaLlonal sLaLe aL L
2
. ln order for 8raLman's
counLerexample Lo work we need Lhe posslblllLy Lo falsely belleve now LhaL l lnLend Lo buy Lhe
LlckeL aL L
2
. ln 8raLman's plcLure, Lhe absence or presence of Lhe lnLenLlon aL L
2
ls whaL makes
Lhe bellef Lrue or false. 8uL ls Lhls a real posslblllLy?
1here are only Lwo momenLs aL whlch Lhe bellef l belleve now LhaL l lnLend Lo buy Lhe
LlckeL aL L
2
" could be made Lrue or false: Lhese are when Lhe 'now' ls ln Lhe pasL - call lL L
1
-
and when Lhen 'now' colncldes wlLh L
2
. 1he flrsL quesLlon ls: Can l hold a false bellef aL L
1
? l
would say LhaL glven Lhe very naLure of Lhe bellef aL L
1
, LhaL bellef cannoL be Lrue or false aL L
1
.
1he LruLh maklng condlLlon of Lhe bellef aL L
1
has noL yeL happened: Lhe occurrence or noL
occurrence of Lhe lnLenLlon aL L
2
wlll deLermlne wheLher Lhe bellef

ls Lrue or false. AL L
1
we are
clearly noL ln a poslLlon Lo asslgn a LruLh value Lo LhaL bellef. lf Lhls argumenL ls correcL Lhen aL
L
1
we cannoL posslbly clalm LhaL we are holdlng a false (or Lrue) bellef abouL our fuLure
lnLenLlon. 1he bellef ls nelLher Lrue nor false ln LhaL Lhe LruLh maklng condlLlon has noL
happened yeL. lf we conslder Lhe behavlor of Lhe agenL aL L
1
we see LhaL he acLs aL L
1
as lf he
wlll acLually have Lhe lnLenLlon aL L
2
. 1hls would noL make Lhe bellef lLself Lrue or false: all we
can say ls LhaL Lhere ls no way Lo dlsLlngulsh Lhe behavlor of an agenL aL L
1
LhaL wlll acLually
form Lhe lnLenLlon aL L
2
from Lhe behavlor of a second agenL aL L
1
LhaL wlll fall Lo form Lhe
lnLenLlon aL L
2
. 8oLh agenLs are holdlng aL L
1
a bellef LhaL ls nelLher Lrue nor false.
LeL's say LhaL L
2
comes - l.e. 'now' ls L
2
- and LhaL l do nC1 form Lhe lnLenLlon Lo buy Lhe
LlckeL. Can l hold a false bellef aL L
2
? Can l falsely belleve now -l.e. aL L
2
- LhaL l lnLend Lo buy
Lhe LlckeL aL L
2
? lf 8raLman ls rlghL, slnce l dld noL form an lnLenLlon Lhen my bellef ls false. 8uL
ls Lhls a real posslblllLy? As Lhe reader can noLlce, Lhls ls Lhe same as a synchronlc case l.e. a
case ln whlch bellef abouL lnLenLlon and lnLenLlon lLself happen aL Lhe same Llme. 1he absence
of Lhe lnLenLlon Lo buy Lhe LlckeL wlll glve no evldences Lo raLlonally hold a bellef abouL LhaL
lnLenLlon. ln oLher words lf l do noL lnLend Lo buy Lhe LlckeL aL L
2
Lhen Lhere ls no posslblllLy
LhaL l can raLlonally hold Lhe bellef l belleve aL L
2
LhaL l lnLend Lo buy Lhe LlckeL aL L
2
".
lmaglne a case ln whlch you are ln love wlLh a perfecL sLranger LhaL you happen Lo see
everyday aL Lhe coffee shop where you go each mornlng. l mlghL consclously form Lhe
lnLenLlon Lo Lalk Lo Lhe sLranger Lhe nexL day. l am aware of my menLal process and l mlghL
come Lo Lhlnk LhaL l belleve now LhaL l lnLend Lo Lalk Lo her Lomorrow. 1he followlng day l
uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

89

enLer Lhe coffee shop, see Lhe person and my emoLlonal reacLlon prevenLs me Lo accompllsh
whaL l dld lnLend Lo do Lhe prevlous day: my volce dlsappears and l cannoL Lhlnk of anyLhlng Lo
say. uoes LhaL mean LhaL yesLerday - or Loday - l falsely belleved LhaL l lnLended Lo Lalk Lo her
when l acLually dld noL lnLend Lo do so? Would Loday's falllng Lo accompllsh whaL l lnLended Lo
do make yesLerday's bellef false? l do noL Lhlnk so: yesLerday my bellef could noL be elLher Lrue
or false: Lhe LruLh maklng condlLlon - wheLher l wlll or won'L Lalk Lo her Lomorrow - dld noL
happen yeL. 1oday as l walk ln Lhe cafeLerla and mlserably fall Lo Lalk Lo LhaL person, l do noL
have any evldence Lo accepL Lhe bellef l belleve now LhaL l lnLend Lo Lalk Lo her now"
anymore.
1hls ls much ln llne wlLh SeLya's clalm LhaL Lhe mlnd ls noL LransparenL Lo lLself "
17
: we
mlghL be [usL wrong abouL assesslng our fuLure psychologlcal sLaLes. WhaL l am clalmlng ls LhaL
we elLher are noL ln Lhe poslLlon Lo say wheLher Lhe assessmenL ls correcL - we are holdlng
now a bellef abouL a fuLure lnLenLlon- or whenever we aLLaln such poslLlon Lhen lL would be
lrraLlonal Lo hold LhaL bellef wlLhouL proper evldences.

ln concluslon Lhere ls no space Lo hold a false bellef abouL lnLenLlon: lL ls ln facL Lhe case LhaL
elLher your bellef abouL lnLenLlon cannoL assume any LruLh value or LhaL you do noL have any
ground Lo form a cerLaln bellef abouL lnLenLlon. ConLrary Lo whaL 8raLman clalms, l do noL
belleve LhaL holdlng a false bellef abouL lnLenLlon ls a genulne posslblllLy. lf Lhe argumenL ls
correcL, Lhen Lhe counLerexample LhaL 8raLman presenL agalnsL Lhe cognlLlvlsL loses lLs appeal.
lf Lhls ls Lhe case Lhen Lhe cognlLlvlsL ls safe: lL ls ln facL Lhe case LhaL we have lncoherence ln
pracLlcal reason - we fall Lo form Lhe lnLenLlon Lo buy Lhe LlckeL - and lncoherence ln
LheoreLlcal reason - we fall Lo form Lhe bellef LhaL we wlll buy Lhe LlckeL.

Suppose now for Lhe sake of Lhe argumenL LhaL Lhe prevlous reasonlng ls compleLely mlsLaken:
leL's granL LhaL holdlng a false bellefs abouL lnLenLlon ls a genulne posslblllLy. lf Lhls ls Lhe case,
l belleve LhaL 8raLman wlll sLlll have Lo solve one lmporLanL problem.
llrsL of all, Lhe cognlLlvlsL mlghL clalm LhaL Lhe mere lnLroducLlon of bellef (e) does noL
make Lhe sysLem of bellefs coherenL or lncoherenL. 8ellef (e) ls so Lo speak neuLral: Lhe bellef
l belleve LhaL l lnLend Lo do x" ls noL Lhe klnd of bellef LhaL we need Lo form Lo produce a
coherenL seL of bellefs ln LheoreLlcal reason. 1he bare lnLroducLlon of (e) ln Lhe seL of bellefs
does noL make bellefs coherenL. 8raLman seems Lo be aware of Lhls: whaL makes Lhe sysLem of
bellefs coherenL ls Lhe presence of bellef (f) l wlll buy Lhe LlckeL". 8uL how ls bellef (f)
produced?
Cne can argue LhaL bellefs l wlll Lravel Lo rague Lhls afLernoon" (a*) and 8uylng Lhe
LlckeL ls necessary Lo Lravel Lo rague" (b) are sufflclenL Lo produce Lhe bellef l wlll buy Lhe
LlckeL" (f) and make Lhe seL of bellefs coherenL. 8raLman however clalms LhaL lL ls (e) - Lhe
posslblllLy Lo form false bellefs abouL one's lnLenLlon - Lhe key elemenL LhaL makes Lhe sysLem
of bellefs coherenL. 1he challenge for 8raLman ls Lo explaln whaL ls Lhe exacL role of (e) Lo
make Lhe seL of bellefs coherenL: from Lhe prevlous conslderaLlons lL seems LhaL we can easlly
hold a coherenL seL of bellefs even wlLhouL Lhe presence of (e). 1here musL be Lherefore some
LheoreLlcal movemenL LhaL wlll make Lhe posslblllLy of holdlng a false bellef abouL lnLenLlon a
cruclal elemenL for maklng bellefs coherenL. unless Lhls lssue ls addressed, l do noL Lhlnk LhaL
Lhe counLerexample LhaL 8raLman presenLs ls successful ln challenglng cognlLlvlsm.




17
SeLlya, k. CognlLlvlsm AbouL lnsLrumenLal 8eason. 117, 2007.

Bumana.Nente - Issue 1u- }uly 2uu9

90

AN AL1LkNA1IVL A11ACk AGAINS1 CCGNI1IVISM

1he ldea presenLed ln Lhe flrsL crlLlque Lo 8raLman ls whaL led me Lo belleve LhaL Lhere mlghL
be an alLernaLlve way Lo ob[ecL Lo cognlLlvlsm. lL seems Lo me LhaL Lhe problemaLlc move ls
done when connecLlng LheoreLlcal reason wlLh pracLlcal reason. l do noL see parLlcular
problem ln reduclng (a) Lo (a*) buL Lhere ls cerLalnly an amblgulLy ln lnLerpreLlng (d) l do noL
lnLend Lo buy Lhe LlckeL". 1he lnLerpreLaLlons we can posslbly provlde Lo (d) mlghL prevenL Lhe
posslblllLy of lLs LranslaLlon ln Lhe sphere of bellefs.
1he general case for lncoherence presenLed earller ln Lhe LexL, shows how lnLenLlon (d) [l
do noL lnLend Lo buy Lhe LlckeL] - and lnLerpreLed as l lnLend nC1 Lo buy Lhe LlckeL" - ls
reduced by Lhe SLrong 8ellef 1hesls Lo Lhe bellef (d*) [l wlll noL buy Lhe LlckeL]. 1hls ls Lhe case
LhaL Lhe cognlLlvlsL endorses. ln Lhe earller example, lncoherence of lnLenLlons ln pracLlcal
reason ls reduced Lo Lhe lncoherence of bellef (a*), (b), (c) and (d*). We have a slLuaLlon ln
whlch lncoherence of lnLenLlons ln pracLlcal reason ls explalned by lncoherence of bellef ln
LheoreLlcal reason.
l wanL Lo clalm LhaL Lhls ls noL Lhe correcL lnLerpreLaLlon. lncoherence of lnLenLlons occurs
because l foll Lo form Lhe lnLenLlon l lnLend Lo buy Lhe LlckeL". 8elng lncoherenL ls cerLalnly
noL Lhe same as lnLendlng -or Lo fotm an lnLenLlon Lo do- Lhe opposlLe of whaL l was
demanded Lo lnLend by Lhe coherence requlremenLs of pracLlcal reason. lncoherence, on Lhe
conLrary, perLalns Lhe facL LhaL we foll to fotm an lnLenLlon for Lhe means LhaL our bellefs
would requlre us Lo lnLend ln order Lo accompllsh Lhe end. A slLuaLlon of lncoherence of
lnLenLlons ls a slLuaLlon ln whlch l hold (a), (b), (c) and fall Lo produce Lhe correcL lnLenLlon l
lnLend Lo buy Lhe LlckeL".
18
So lf we wanL Lo see whlch lnLenLlons l am holdlng when l am
lncoherenL we wlll flnd LhaL Lhere ls oo loteotloo wbotsoevet regardlng Lhe acLlon buylng Lhe
LlckeLs". ln a schemaLlc form:


18
AL Lhls polnL some crlLlcs would challenge Lhe ldea LhaL lncoherence of lnLenLlons ooly amounLs Lo fall
Lo form Lhe rlghL lnLenLlon whlch ls demanded by Lhe coherence requlremenL of raLlonallLy. 1hey can
argue LhaL one can ln facL be lncoherenL also by formlng an lnLenLlon wlLh Lhe wrong conLenL . 1hls ls
noL an lssue: even lf we form an lnLenLlon wlLh Lhe wrong conLenL, aL Lhe same Llme we are sLlll falllng Lo
form Lhe rlghL lnLenLlon. 1hls laLLer ls Lhe key Lo lncoherence. When we conslder lncoherence, lL seems
Lo me LhaL Lhere ls no dlfference beLween Lhe slLuaLlon ln whlch (a) l fall Lo form Lhe requlred lnLenLlon
and (b) l fall Lo form Lhe requlred lnLenLlon ooJ l form an lnLenLlon wlLh Lhe wrong conLenL. ln boLh
clrcumsLance we are lncoherenL because we fall Lo form Lhe correcL lnLenLlon.
8LLlLlS ln1Ln1lCnS

(o*) 8ellef.
l wlll qo to ltoqoe

(b) 8ellef:
[8uylng a LlckeL ls necessary
Lo go Lo rague]


(c) 8ellef:
[lnLendlng Lo buy a LlckeL ls
necessary Lo buy a LlckeL]

by SLrong 8ellef 1hesls











(a) lnLenLlon:
[l lnLend Lo go Lo rague Lhls
afLernoon]








uiacomo Nollo - Why Cognitivists about piactical Reason can only be Semi-Cognitivists

91


ln (d) we observe an absence of lnLenLlon. 1he agenL, even Lhough holdlng bellefs/lnLenLlons
(a), (b) and (c), falled Lo form Lhe correcL lnLenLlon ln conformlLy wlLh Lhe requlremenL of
coherence of lnLenLlon.
Slnce Lhere ls no lnLenLlon upon whlch we can apply Lhe SLrong 8ellef 1hesls Lhen Lhere
can'L be a parallel bellef LhaL wlll be able Lo produce lncoherence of bellefs. CognlLlvlsLs ln facL
lack a prlnclple LhaL connecLs non-formed-lnLenLlons wlLh bellefs: Lhere ls no relaLlon LhaL ls
made evldenL regardlng Lhe bellef lnvolvemenL of lnLenLlons LhaL are falled Lo be produced. lf
Lhls ls correcL Lhen lL ls noL Lhe case, glven Lhe classlcal formulaLlon of bellef lnvolvemenL ln
any cognlLlvlsL accounL, LhaL lncoherence of lnLenLlons ln pracLlcal reason can be posslbly
explalned wlLh lncoherence of bellefs ln LheoreLlcal reason.

1he cognlLlvlsL mlghL respond LhaL ln Lhls case you sLlll have lncoherence of bellefs: Lhe
demand of LheoreLlcal raLlonallLy requlres you Lo derlve Lhe bellef LhaL you wlll buy Lhe LlckeL
(d*) lf you hold bellefs (a), (b) and (c). 1here ls no bellef (d*) and Lherefore we have
lncoherence ln LheoreLlcal reasonlng. AL Lhe same Llme no lnLenLlon (d) has been formed even
Lhough lLs producLlon was requlred by Lhe coherence requlremenL of pracLlcal reason. ?ou can
see how Lhe absence of my lnLenLlon Lo buy Lhe LlckeL occurs LogeLher wlLh Lhe absence of my
bellef LhaL l wlll buy Lhe LlckeL. 1he cognlLlvlsL mlghL go furLher and clalm LhaL whenever we fall
Lo form Lhe bellef requlred by Lhe demand of LheoreLlcal raLlonallLy, Lhen we also fall Lo form
Lhe lnLenLlon requlred by Lhe demand of pracLlcal raLlonallLy. Pe mlghL lnslsL on Lhe facL LhaL
Lhere ls a slLuaLlon of LheoreLlcal lncoherence every Llme we encounLer pracLlcal lncoherence.
We can Lherefore observe lncoherence of bellefs and lncoherence of lnLenLlons.
1he problem wlLh Lhls response ls LhaL lL does noL supporL Lhe orlglnal cognlLlvlsL clalm LhaL we
can undersLand lncoherence ln pracLlcal reason by reduclng lL Lo lncoherence ln LheoreLlcal
reason. Lven lf we concede LhaL Lhe cognlLlvlsL ls able Lo demonsLraLe Lhe consLanL coexlsLence
of LheoreLlcal lncoherence wlLh pracLlcal lncoherence, Lhen lL does noL follow LhaL one ls
expllcable by Lhe oLher. Pe dld noL show how whaL wenL wrong ln cases of pracLlcal
lncoherence ls reduclble Lo whaL wenL wrong ln cases of LheoreLlcal lncoherence. 1he
cognlLlvlsL sLlll needs Lo explaln how exacLly Lhe fallure ln pracLlcal reason ls connecLed wlLh
Lhe fallure ln LheoreLlcal reason or how exacLly lncoherence ln pracLlcal reason [usL ls
lncoherence ln LheoreLlcal reason.
As l undersLand Lhe cognlLlvlsL plcLure, Lhe only avallable lnsLrumenL Lo connecL lnLenLlon
and bellef ls a bellef-lnvolvemenL prlnclple: dependlng on Lhe verslon of cognlLlvlsm - ldenLlLy
1hesls, SLrong 8ellef 1hesls or Weak 8ellef 1hesls - Lhls prlnclple wlll provlde a brldge beLween
pracLlcal reason and LheoreLlcal reason. lnLenLlons wlll be Lherefore relaLed Lo Lhelr
correspondlng bellef. When we conslder cases of pracLlcal lnconslsLency, Lhe cognlLlvlsL has
Lhe proper lnsLrumenLs Lo explaln how exacLly lnconslsLency ln pracLlcal reason can be
explalned by lnconslsLence ln LheoreLlcal reason. lf however we conslder cases of pracLlcal
lncoherence, Lhe cognlLlvlsL does noL have Lhose LheoreLlcal Lools LhaL can provlde a
saLlsfacLory accounL ln favour of Lhe clalm LhaL we can explaln lncoherence ln pracLlcal reason
by lncoherence ln LheoreLlcal reason. As l polnLed ouL earller ln Lhe LexL, a bellef-lnvolvemenL
prlnclple cannoL posslbly be applled when an lnLenLlon ls absenL.










(d) lnLenLlon: [.]
(l do noL form Lhe lnLenLlon Lo
buy Lhe LlckeL)

Bumana.Nente - Issue 1u- }uly 2uu9

92

1here ls ln facL a full LheoreLlcal gap: we would requlre Lhe cognlLlvlsLs Lo provlde a 'non-
bellef lnvolvemenL prlnclple" LhaL would explaln how Lhe absence of an lnLenLlon somehow
lovolves Lhe absence of a correspondlng bellef. ln lLs currenL formulaLlon, Lhe cognlLlvlsL
accounL cannoL explaln how Lhe fallure Lo form an lnLenLlon ls connecLed wlLh Lhe fallure Lo
form a bellef. lL seems Lherefore LhaL cognlLlvlsm cannoL posslbly provlde an accounL for
lncoherence of lnLenLlon ln Lerms of lncoherence of bellefs.

CCNCLUSICN

Can you now posslbly be fully cognlLlvlsL regardlng all Lhe requlremenLs of pracLlcal raLlonallLy?
Clven Lhe ldea of bellef-lnvolvemenL, can you explaln Lhe raLlonal requlremenLs of conslsLency
and coherence ln pracLlcal reason by appeallng Lo Lhe raLlonal requlremenLs of LheoreLlcal
reason?
My answer ls no. 1he cognlLlvlsL's plcLure coo provlde a sound and sLrucLured explanaLlon
on how Lo derlve lnconslsLency of lnLenLlons from lnconslsLency of bellefs. Powever, as lL ls
formulaLed, lL coooot explaln how Lo reduce lncoherence of lnLenLlons Lo lncoherence of
bellefs. Clven Lhe presenL debaLe, a reasonable poslLlon Lo hold ls Lo be a seml-cognlLlvlsL: Lhls
vlew endorses cognlLlvlsm regardlng Lhe conslsLency requlremenL buL cannoL accepL
cognlLlvlsm when lL comes Lo explaln Lhe coherence requlremenL. We have seen how Lhe Lwo
raLlonal requlremenLs presuppose dlfferenL sLrucLural elemenLs, a facL LhaL ls parLlcularly
evldenL when we conslder fallure of respecLlng Lhose demands. lf we assume Lhe bellef-
lnvolvemenL prlnclple alone, Lhen we can see how lL can be applled on Lhe conslsLency
requlremenL and lL cannoL posslbly be applled Lo Lhe coherence requlremenL.


8I8LICGkAn

8raLman M. (forLhcomlng) lnLenLlons, 8ellefs, racLlcal, 1heoreLlcal. ln !. 1lmmerman, !.
Sorupskl, S. 8oberLson, eds. 5pbetes of keosoo, Cu.
8raLman M. (1987) loteotloo, lloos ooJ ltoctlcol keosoo. Parvard unlverslLy ress.
8roome !. (1999) normaLlve 8equlremenLs. kotlo, 12, 398-419.
8roome !. (2007) Wlde or narrow Scope? MloJ, 116, 339-370.
8runero !. (2008) AgalnsL CognlLlvlsm abouL racLlcal 8aLlonallLy. lbllosopblcol 5toJles,
AugusL.
kolodny n. (2003) Why be raLlonal? MloJ, 114, 309-363.
SeLlya k. (2007) CognlLlvlsm AbouL lnsLrumenLal 8eason. tblcs 117.
velleman u. (2007) ltoctlcol keflectloo, unlverslLy of Chlcago ress.



93

ComparingPreferences

MauroRossi
*

mauro.rossi@umontreal.ca

ABSTRACT

The orthodox view in economics is that interpersonal comparisons (ICs) of preferences present
insurmountableepistemicdifficultiesand,thereby,havenoscientificlegitimacy.Arecentlineofthought
argues against this position by investigating how ordinary people make ICs of preferences. The
underlying idea is that the problem of ICs can be solved if we can find scientific evidence showingthat
ICs can be reliably made in everyday life. In this paper, I provide an assessment of this strategy. I
considerfourargumentsattemptingtoshowthattheconditionsforhavingreliableICscanbesatisfied.
Iarguethatalltheseargumentsfailandrejectthisstrategyasunsuccessful.

The susceptibility of one mind may, for what we


know, be a thousand times greater than that of
another. But, provided that the susceptibility was
different in a like ratio in all directions, we should
neverbeabletodiscoverthedifference.Everymind
is thus inscrutable to every other mind, and no
common denominator of feeling seems to be
possible.
1

1. INTRODUCTION

It is commonplace that, in everyday life, we compare preferences belonging to different


peoplewithrespecttotheirintensity.Wetypicallymakesuchcomparisonswithrelativeease.
Moreover,weoftendonotfindinterpersonalcomparisonsofpreferencesmoredifficultthan
intrapersonal comparisons, that is, comparisons involving our own preferences.
2
Things
change as soon as we consider the matter from a theoretical point of view. Several authors
claimthatinterpersonalcomparisons(ICshenceforth)ofpreferencesareeitherimpossible,
3
or
meaningless
4
or, at least, that they are not factual claims, but rather normative statements.
5

Themainreasonfortheoreticalscepticismisthat,whiletheempiricalevidenceissufficientfor
establishingcomparisonsabouttheintensityofasingleindividualspreferences,itappearsto

*
PostdoctoralFellow,CentredeRechercheenthiquedel'UniversitdeMontral(CRUM).
1
S.Jevons,TheoryofPoliticalEconomy,4
th
edition,Macmillan,London1911(1871,1
st
ed.),p.14.
2
SeeD.Davidson,Judginginterpersonalinterests,inFoundationsofsocialchoicetheory,ed.byJ.Elster
A. Hylland, Cambridge University Press, Cambridge 1986, pp. 195211, reprinted as Interpersonal
ComparisonsofValues,inD.DavidsonProblemsofRationality,OxfordUniversityPress,Oxford2004,p.
59.
3
See S. Jevons, op. cit, p. 14. For an early reaction against the impossibility claim, see I.D.M. Little, A
CritiqueofWelfareEconomics,2
nd
ed.,ClarendonPress,Oxford1957(1950,1
st
ed.),ChapterIV.
4
SeeK.Arrow,SocialChoiceandIndividualValues,2
nd
ed.,Wiley,NewYork1963(1951,1
st
ed.),p.9.
5
SeeL.Robbins,AnEssayontheNatureandSignificanceofEconomicScience,Macmillan,London1932,
p.139.

Humana.MenteIssue10July2009

94

be insufficient for establishing comparisons about the intensity of different individuals


preferences.Inturn,thishasledvariousareasofscientificresearchmostnotably,economics
todismissICsofpreferencesasscientificallyillegitimate.
Thisrejectionhasnotbeenwithoutrepercussions.Itistruethateconomictheorydoesnot
needICsofpreferencestoexplainthebehaviourofthemaineconomicagents,i.e.consumers
and firms. On the other hand, if ICs of preference strengths are not allowed, welfare
economicsisunabletosettledistributiveconflictsbyconsideringtherelativeimportancethat
different individuals attach to competing states of affairs. In social choice theory, it is
impossibletoaggregateindividualpreferencesinordertoobtainasocialrankingofalternative
states of affairs, which satisfies, amongst the others, a condition of nondictatorship. Even
worse, if ICs turn out to be impossible, the meaningfulness of various ethical doctrines e.g.
preference utilitarianism and at least some versions of objective list theories of wellbeing, is
entirely compromised. Finally, the rejection of ICs has a considerable impact on more applied
spheres, such as health care and policy making, which require the use of normatively
significant measures indicating how different peoples preferences compare in terms of
strength.
In the course of the years, the constraints posed by the rejection of ICs have started to
appear intolerable. As a consequence, several strategies have been explored to give ICs new
scientific legitimacy. In this paper, I want to examine one of them in particular. Broadly
speaking, the main idea underlying this strategy is that, by investigating how ordinary people
makeICsofpreferencesineverydaylife,itispossibletofindasolutiontothemaintheoretical
problems concerning ICs. Consider the common view of our comparative practice. It is
generallyheldthatICsofpreferencesarebasedontheascriptionofpreferenceswithspecific
contentandintensitytootherpeopleandtoourselves.Ifthispictureiscorrect,thesuggestion
isthatthesolutiontotheproblemofICsshouldbebasedontheanalysisoftheconditionsthat
ought to be satisfied in order for us to make reliable ascription of preference strengths to
otherpeopleandtoourselves.
Thisapproachinvitesustoexaminetwodifferentkindsofquestion:thequestionofmental
ascription, that is, the question of how ordinary people assigns mental terms to other people
andtothemselves;andthequestionofthemeaningofmentalstates,thatis,thequestionof
whatordinarypeoplemeanwhentheyemploymentalterms.Broadlyspeaking,therearetwo
main theories of mental ascription: Theory Theory (TT) and Simulation Theory (ST). According
totheformer,ordinarypeopleascribementalstatesbymeansofaTheoryofMindthatthey,
moreorlesstacitly,possess.Accordingtothelatter,ordinarypeopleascribementalstatesto
others by trying to simulate, or replicate, their mental activity. On the other hand, there are
two main theories of the meaning of mental states in contemporary philosophy of mind:
(commonsense)functionalismandexperientialism.Accordingtofunctionalism,themeaningof
amentalstateisgivenbythesetofcausallawsinwhichsuchamentalstatefiguresandwhich
relate it to inputs, other mental states and behavioural outputs. Instead, according to
experientialism, the meaning of a mental state is given by the more or less conscious
experiencesthatthesubjecthasofit.
Given that the problem of comparing different peoples preferences concerns a specific
type of mental states (i.e. preferences) and one of their properties (i.e. strength), one would
expecttheexistenceofalargeliteratureinphilosophyofmindconnectingtheproblemofICs
to these fields of research. Instead, and quite surprisingly, philosophers of mind have almost
completely ignored the issue of ICs. Alvin Goldman constitutes the only significant exception.
Indeed, in his Simulation and Interpersonal Utility, Goldman attempts to bring the problem
MauroRossiComparingPreferences

95

ofICsinlinewithcurrentdebatesinphilosophyofmindandepistemology.
6
Forthepurposeat
stake,Goldmansapproachhastwolimitations:itfocusesmainly onICsofhappinessanditis
toospecific,asitconsidersonlySTasatheoryofmindreadingandexperientialismasatheory
ofthemeaningofmentalstates.
In this paper, I want to extend Goldmans approach by focusing explicitly on ICs of
preferencestrengthandbyconsideringthemaintheoriesofmindreadingandofthemeaning
of mental states. I shall pursue a twofold goal. First, I shall individuate the conditions that
oughttobesatisfiedinordertohavereliableICsofpreferencestrengthwithrespecttoallthe
main accounts of how ordinary people make ICs of preference strength. Second, I shall
examine some arguments attempting to show that these conditions can, at least in principle,
besatisfied.Bysodoing,Ibelieveitispossibletohaveafullassessmentofthestrategyunder
consideration. Ultimately, I shall claim that this strategy is unsuccessful. I shall offer an
argumentbyelimination.NomatterwhichaccountweadoptofhowordinarypeoplemakeICs
of preference strength, all the arguments proposed in the literature fail to show that we can
make reliable ICs of preference strength. Obviously, my analysis does not entail that no such
argumentexists.Nevertheless,thecurrentstateofresearchaboutICslegitimatesamoderate
form of scepticism: given that no solution explored so far proves to be successful, it might as
wellbethecasethatICsofpreferencespresentanunsolvableproblem.
Ishallproceedasfollows.Insection2,IshallillustratetheproblemofICsofpreferencesin
more detail. Following Goldman, I shall specifically focus on its epistemological dimension,
according to which the problem is whether or not we can have scientific knowledge of, or, at
least,scientificallyjustifiedbeliefsabout,howdifferentpeoplespreferencescompareinterms
ofstrength.Insection3,Ishallpresent,respectively,themaintheoriesofmindreadingandthe
maintheoriesofthemeaningofmentalstatespresentincontemporaryphilosophyofmind.In
section 4, I shall consider the issue of scientific justification. Although scientific justification
may require the satisfaction of several requirements (e.g. publicity, replicability,
measurability), for simplicity here I shall focus only on one of them, namely, the requirement
that the relevant mechanisms on which ordinary peoples comparative practice is based be
reliable for the purpose of making ICs. In section 6, I shall discuss five different arguments
attempting to show that this condition can be satisfied. I shall argue that all these arguments
fail.Ishallsummarisemyfindingsintheconclusion.

2. THEPROBLEM

The problem of ICs of preferences presents several dimensions. However, two of them are
especially important for the previously mentioned areas of scientific research. The first is the
metaphysical dimension. The relevant question is whether or not there are facts about ICs of
preference strength. The second is the epistemological dimension. The relevant question is
whether or not we can have epistemic access to these (alleged) facts about ICs of preference
strength. These dimensions are often confused in the literature. However, they differ in
importantrespectsandtheyshouldbedistinguished.Themetaphysicalquestionisclearlyprior
to the epistemological question. If there is no fact of the matter about ICs of preference
strength, then no question of epistemic access arises. Although the former is a contentious
issue, in what follows, I shall simply presuppose that we can positively address the
metaphysical question and focus instead on the epistemological question about ICs. Broadly
speaking, this is the question of whether or not we can have knowledge of or, at least,
justified, ICs of preference strength. More narrowly, it is the question of whether or not we

6
SeeA.Goldman,SimulationandInterpersonalUtility,Ethics,4,1995,pp.709726.

Humana.MenteIssue10July2009

96

can have scientific knowledge of, or scientifically justified beliefs about, how different
peoplespreferencescompareintermsofstrength.Thisissueisparticularlyimportant.Indeed,
ifwecannothavescientificknowledgeof,orscientificallyjustifiedbeliefs,abouthowdifferent
peoples preferences compare in terms of strength, welfare economics, social choice theory,
normativeandappliedethicsremainintroubleevenifthereisafactofthematteraboutICsof
preferences. It is thus important to see why ICs of preference strength raise specific
epistemologicalworries.
AstheproblemofICsofpreferencesoftenarisesinthecontextofeconomics,itisusefulto
startourillustrationfromthisareaofanalysis.Ineconomics,preferencesaredefinedasbinary
relationsR,thatis,relationsbetweentwoitems.Theitemsincludedinthepreferencedomain
vary according to different decision theories.
7
For the purpose at stake, it is not necessary to
committoanyspecificontology.Inwhatfollows,Ishallrefertotheobjectofpreferenceasthe
option that an individual faces. We can then say that an individual prefers an option x to
another option y and express this in the following way: x R y. If preferences satisfy certain
axioms,theycanberepresentedbyanumericalfunction.Morespecifically,iftheysatisfythe
ordering axioms, i.e. completeness and transitivity, they can be represented by an ordinal
utilityfunction,uniqueuptoamonotoneincreasingtransformation.Ifpreferencessatisfythe
expected utility axioms, i.e. the ordering axioms, the independence axiom and the
Archimedean axiom, they can be represented by an interval utility function, unique up to a
positive affine transformation. An interval function is a cardinal function. While an ordinal
function is supposed to preserve simply the order of the individuals preferences, a cardinal
functionissupposedtopreserveadditionalinformation.Morespecifically,acardinalfunction
is supposed to capture the degree to which an individual prefers one option rather than
another, or, in other words, the intensity of the individuals preferences for alternative
options. Typically, in order to fix a cardinal scale of measurement, it is (necessary and)
sufficient to fix two points on the scale, namely, the zero and the unit. One feature of an
intervalscaleofmeasurementisthatboththezeroandtheunitarearbitrary.Thus,assigning
the utility value 0 to one option does not mean that the individual prefers that option with
zero intensity. It is simply an arbitrary choice, which fixes one of the relevant points on the
cardinalscaleofmeasurement.
Let us now consider an example. Suppose there are two individuals, i and j, and four
optionsx,y,w,zA.Individualirankstheoptionsinthefollowingway:xR
i
yR
i
wR
i
z.Onthe
other hand, individual j ranks the options in the following way: w R
j
z R
j
x R
j
y. Suppose we
represent their preferences on an interval scale. In particular, we assign the value 1 to the
mostpreferredoptionandthevalue0totheworstoptionintheirpreferencerankings.Wecan
then assign a value that represents the intensity of their preferences for the other options,
wherethesevaluesarerelativetothebestandtheworstoptionsineachindividualsranking.
Supposeitisthecasethatu
i
(y)=u
j
(x)=0.6.Supposeitisalsothecasethatu
i
(x)u
i
(y)=u
j
(w)
u
j
(x) = 0.4. In the first case, can we conclude that individual i prefers option y with the same
strength with which individual j prefers option x?
8
In the second case, can we conclude that

7
More specifically, preferences range over either acts, or propositions, or prospects. For acts, see L.
Savage, The Foundations of Statistics, Wiley, New York 1954. For propositions, see R. Jeffrey, The Logic
ofDecision,2
nd
ed.,UniversityofChicagoPress,Chicago1983(1965,1
st
ed.).Forprospects,orlotteries,
seeJ.vonNeumannandO.Morgenstern,TheoryofGamesandEconomicBehavior,PrincetonUniversity
Press,Princeton1944.
8
Thisisthecaseofaninterpersonalcomparisonofutilitylevels.ICsofutilitylevelsarejudgmentsofthe
form:u
i
(x)u
j
(y).
MauroRossiComparingPreferences

97

the difference in strength of individual is preference for option x over y is the same as the
differenceinstrengthofindividualjspreferenceforoptionwoverx?
9

The answer to both questions is negative: the empirical evidence is not sufficient to
conclude that identical utility values represent identical preference strengths. The reason is
the following. As we have seen before, in order to fix an interval scale of measurement it is
(necessary and) sufficient to fix an arbitrary zero and an arbitrary unit. In the case of
preferences,wefixthescaleofmeasurementbyassigningthevalue0totheworstoptionand
thevalue1tothebestoption.However,thisisonlysufficienttofixthescaleofmeasurement
foreachindividual.Itisnotsufficienttofixacommonscaleforbothindividuals.Inorderfor
thepreferencescaletobeacommonone,itmustbethecasethatbothindividualsprefertheir
best option and their worst option with the same strength. The problem is that the evidence
doesnottellusanythingatallabouthowdifferentindividualspreferencesfortheirbestand
worst options compare in terms of strength. As it is typically put, the evidence is consistent
with the case where individual i prefers the best (the worst) option with intensity ten times
greaterthanj.
10

ThepreviousexampleshowsthatICsofpreferencesareunderdeterminedbytheempirical
evidence.Atfirstsight,thisposesathreattothepossibilityofhavingscientificallyjustifiedICs.
At least, this seems to be the case if we adopt an evidentialist theory of epistemic
justification.
11
The argument is straightforward. If all the possible empirical evidence is
insufficient to determine ICs of preference strength, then, if the empirical evidence is what
makesICsofpreferencestrengthjustified,itfollowsthatICscannotbejustified.Ontheother
hand,ifthisistheonlysourceoftheproblem,theepistemologicalproblemofICsmightaswell
haveapositivesolution.
12
Onereasonisthatevidentialismisnottheonlytheoryofepistemic
justification. According to reliabilism, for instance, what makes a belief justified is not the
empiricalevidence,butthereliabilityoftheprocessesbymeansofwhichthebeliefinquestion
isformed.
13
Thus,evenifallthepossibleempiricalevidenceisinsufficienttodetermineICsof
preference strength, these can nonetheless be justified, provided that they are acquired
through reliable processes, that is, through processes that tend to produce true beliefs. The
resultisthat,ifwehaveindependentgroundstopreferreliabilismtoevidentialismasatheory

9
This is the case of an interpersonal comparison of utility differences. ICs of utility differences are
judgmentsoftheform:u
i
(x)u
i
(y)/u
j
(w)u
j
(z)=,forsome.
10
It is worth noticing that the problem is independent from which empirical evidence one considers.
Economists typically take choice behaviour, under conditions of both certainty and uncertainty, to be
the only admissible evidence for the ascription of individual preferences. However, as List has recently
shown(SeeC.List,AreInterpersonalComparisonsofUtilityIndeterminate?,Erkenntnis,58,2003,pp.
229260), it is not possible to compare different individuals preference strengths in a meaningful way
evenifweextendthesetofadmissibleevidencebyincludingtheindividualslatencyofchoice,i.e.the
time delay between the presentation of the option and the actual choicemaking, the probability of
choice, i.e. the probability of choosing one option rather than another the probability of choice (See I.
Waldner, The Empirical Meaningfulness of Interpersonal Utility Comparisons, The Journal of
Philosophy, 4, 1972, pp. 87103), their verbal expressions (See J. Harsanyi, Cardinal Welfare,
Individualistic Ethics, and Interpersonal Comparisons of Utility, The Journal of Political Economy, 63,
1955, pp. 309321), their expressive reactions (See R. Weintraub, Do Utility Comparisons Pose a
Problem?, Philosophical Studies, 92, 1998, pp. 307319), their facial expressions, body temperature
andotherproxies(SeeC.List,op.cit).
11
For a paradigmatic statement of this position, see R. Feldman and E. Conee, Evidentialism,
PhilosophicalStudies,48,1985,pp.1534.
12
SeeA.Goldman,SimulationandInterpersonalUtility,op.cit.
13
Foraparadigmaticstatementofthisposition,seeA.Goldman,WhatisJustifiedBelief?,inJustification
andKnowledge,ed.byG.Pappas,KluwerAcademicPublisher,Reidel1979,pp.123.

Humana.MenteIssue10July2009

98

ofepistemicjustification,wemightbeabletosolvetheepistemologicalproblemofICs,evenif
ICsareunderdeterminedbyallthepossibleempiricalevidence.
Let us go back to the goal of this paper. I said in the introduction that I am interested in
assessing a specific strategy, which attempts to find a solution to the main theoretical
problemsconcerningICsofpreferencestrengthbyinvestigatinghowordinarypeoplemakeICs
of preferences in everyday life. It is now clear that the problem about ICs of preference
strength with which I am concerned is the problem of whether or not we can have scientific
knowledge of, or scientifically justified beliefs about, how different peoples preferences
compareintermsofstrength.Inthispaper,Ishalladoptreliabilismasmytheoryofepistemic
justification.Accordingtoreliabilism,inorderforabelieftobejustified,itisnotnecessarythat
weknowthatitisreliablyacquired.Itisonlysufficientthat,asamatteroffact,suchabeliefis
reliably acquired. However, for other purposes, this may not be enough. For instance, if we
wanttogiveICsofpreferencestrengthnewscientificlegitimacy,itmayindeedbeimportantto
know whether or not ICs of preferences can be reliably made. The relevant question is thus
whether or not there is scientific evidence that ICs of preference strength can be reliably
made. As the strategy under consideration specifically focuses on how ordinary people make
ICs of preferences in everyday life, this question can be further specified as the question of
whether or not there is scientific evidence that ordinary people can reliably make ICs of
preferencestrengthineverydaylife.
14
Thiswillbetheobjectofmypaper.

3. THEORIESOFMINDREADING

Thissectionisdividedintwoparts.Inthefirst,Ishallbrieflyillustratehowthemaintheoriesof
mindreadingpresentedintheliteratureexplaintheascriptionofmentalstatestoatarget.This
involves considering what ordinary people mean when they use mental terms. In the second
part,Ishallattempttoclarifyhowthesetheoriesmightexplaintheascriptiontoatarget,and
theinterpersonalcomparison,ofpreferencestrengths.
Letusstartwiththefirstpart.Differentdisciplinesformulatetheirexplanatoryaccountsof
peoples mindreading capacity at different levels of description, i.e. the personal, the sub
personalandthephysicallevel.Mostofthephilosophicalliteratureisconcernedwiththesub
personallevelofdescription.Thecommonstrategyistoconceivethemindasasystem,where
mental states and processes are characterized functionally. The goal is then to identify the
underlyinginformationprocessingmechanismsthatneedtobepostulatedinordertoexplain
our mindreading capacity. In the course of the years, two main approaches have emerged:
Theory Theory (TT) and Simulation Theory (ST). TT characteristically accounts for the
mindreading capacity by positing cognitive processes that exploit an internally represented
"knowledgestructure"typicallyabodyofrulesorprinciplesorpropositionswhichservesto
guide the execution of the capacity to be explained.
15
In short, TT explains mental ascription

14
SeeA.Goldman,SimulationandInterpersonalUtility,op.cit.
15
See S. Stich and S. Nichols, Folk Psychology: Simulation or Tacit Theory?, Mind and Language, 7,
1992,pp.3536.SeealsoS.StichandS.Nichols,SecondThoughtsonSimulation,inFolkPsychology:The
TheoryofMindDebate,ed.byM.DaviesT.Stone,Blackwell,Oxford1995,pp.87108,S.StichandS.
Nichols, Cognitive Penetrability, Rationality and Restricted Simulation, Mind and Language, 12, 1997,
pp.297326,S.Nicholsetal.,VarietiesofOffLineSimulation,inTheoriesofTheoriesofMind,ed.byP.
Carruthers P. Smith, Cambridge University Press, Cambridge 1996, pp. 3974, and S. Nichols and S.
Stich,Mindreading:AnIntegratedAccountofPretence,SelfAwareness,andUnderstandingOtherMinds,
OxfordUniversityPress,Oxford2003.
MauroRossiComparingPreferences

99

byarguingthatthefolkspossessaTheoryofMind(ToM),towhichtheyhaveamoreorless
consciousaccess.
16

Asfarasthemeaningofmentalstatesisconcerned,TTisgenerallyassociatedwithanalytic
functionalism.Accordingtofunctionalism,themeaningofamentalstateisgivenbythesetof
causallawsinwhichthatmentalstatefigures.Suchcausallawsspecifyhoweachmentalstate
isrelatedtoenvironmentalinputs,othermentalstatesandbehaviouraloutputs.Itmaybethe
casethattheagentemployingmentalconceptsisincapableofspecifyingalltheseconstitutive
causal relations. Indeed, this may require a sophisticated analysis. If we think of the defining
causal relations as part of the ToM that the agent possesses, then we can say that such a
theory operates tacitly, or, equivalently, that the theory is tacit. As far as mental ascription is
concerned,then,TTassumesthatthefolksascribementalstatestootherpeoplebyobserving
externalevents(i.e.inputsandoutputs)andinferringtherelevantmentalstatesbyreference
tothecausalrelationspostulatedbytheToMthattheypossess.
ST offers an alternative account of mental ascription. The basic idea is that mental
ascriptioninvolvesindividuatinganotherindividualstargetedmentalstatesbyimaginingbeing
subject to the same mental states to which she is subject. The ST approach to mindreading
comes in different forms. In this paper, however, I shall primarily focus on the account
advocated by Alvin Goldman.
17
According to Goldman, first, the simulator asks herself what
mentalstatesshewouldhaveifsheweresubjecttotheinitialmentalstatesofthesimulated
agent. By so doing, she feeds her own informationprocessing mechanisms with pretend
inputs, which supposedly correspond to the other persons initial mental states. These

16
TherearetwovariantsoftheTTapproachtomindreading,namely,thescientifictheorytheory(STT)
andthemodularitytheory(MT).Accordingtotheformer,theToMthatthefolksuseformindreadingis
both learnt and stored in the mind in the same way as scientific theories are. In the course of their
development, children proceed as little scientists, formulating hypotheses on the basis of the
information available and revising them in the light of new data. According to the latter, the ToM is
neither learnt nor stored in the same way as scientific theories are, but it is rather included in one or
moreinnatemodules.Forthecurrentpurpose,however,wecanignorethedistinctionbetweenthetwo
approaches. See H. Wellman, The Childs Theory of Mind, MIT Press, Cambridge, Mass. 1990, J. Perner,
Understanding the Representational Mind, MIT Press, Cambridge, Mass. 1991, J. Gopnik and H.
Wellman,Whythechildstheoryofmindreallyisatheoryofmind,MindandLanguage,7,1992,pp.
145171, J. Gopnik and H. Wellman, The theory theory, in Mapping the Mind: Domain Specificity in
Cognition and Culture, ed. by L. Hirschfiled S. Gelman, Cambridge University Press, New York 1994, J.
Gopnik and A. N. Meltzoff, Words, Thoughts and Theories, MIT Press, Cambridge, Mass. 1997, for a
defenceoftheSTTapproach.SeeA.Leslie,Pretenceandrepresentation:Theoriginsoftheoryofmind,
PsychologicalReview,94,1987,pp.412426,A.Leslie,Someimplicationsofpretenseformechanisms
underlyingthechildstheoryofmind,inDevelopingTheoriesofMinds,ed.byJ.AstingtonP.HarrisD.
Olson, Cambridge University Press, Cambridge 1988, pp. 1946, A. Leslie, Pretending and Believing:
Issues in the theory of ToMM, Cogniton, 50, 1994, pp. 211238, A. Leslie, How to acquire a
representational theory of mind, in Metarepresentation: A Multidisciplinary Perspective, ed. by D.
Sperber,OxfordUniversityPress,NewYork2000,pp.197223,A.LeslieandT.German,Knowledgeand
ability in theory of mind: Oneeyed overview of a debate, Mental Simulation, in ed. by M. Davies T.
Stone,Blackwell,Oxford1995,pp.123150,andS.BaronCohen,Mindblindness:AnEssayonAutismand
TheoryofMind,MITPress,Cambridge,Mass.1995,foradefenceoftheMTapproach.
17
See,inparticular,A.Goldman,InterpretationPsychologized,MindandLanguage,4,1989,pp.161
185, A. Goldman, In defense of the simulation theory, Mind and Language, 7, 1992, pp. 104119, A.
Goldman, The mentalizing folk, in Metarepresentations, ed. by D. Sperber, Oxford University Press,
Oxford 2000, A. Goldman, Simulation theory and mental concepts, in Simulation and Knowledge of
Action, ed. by J. Dokic J. Proust, John Benjamins, Amsterdam 2002, pp. 120, A. Goldman, Simulating
Minds,OxfordUniversityPress,Oxford2006.

Humana.MenteIssue10July2009

100

mechanisms run offline and produce pretend outputs. The simulator then introspectsthese
mentalstatesandrecognisesthemasbelongingtoacertaintype.Finally,sheascribesthem
byanalogytothesimulatedagent.
18

One crucial feature of Goldmans account is its emphasis on introspection. Indeed, third
personalmentalascriptionpresupposesthatthesimulatoriscapableofintrospectingherown
mentalstatesinordertoascribementalstatestoanotherindividual.Ifthisisthecase,theST
explanation of mindreading requires an account of ordinary peoples introspective capacity.
One suggestion is that the simulator recognises her own mental states with respect to the
functionthattheyoccupyinhermindsystem.Goldmangivestworeasonstorejectthisoption.
The first is that functionalism characterises mental states as dispositions, which may involve
relationshipswitheventstowhichthepersonhasnotyetepistemicaccess(e.g.futureevents)
orwillneverhaveepistemicaccess(e.g.inthecaseofsubjunctiverelationships).Thesecondis
that, in order for an individual to recognise her own occurrent mental states, functionalism
requiresthatshebeabletoidentifytheindefinitelylargenumberofotherattitudestowhich
heroccurrentmentalstatesarerelatedasamatterofdefinition.Inturn,thisseemstoburden
selfascriptionwithexcessivecomputationalrequirements.
19

In the light of these problems, Goldman suggests that a more plausible account explains
selfascriptionintermsofthecapacityofinternallydetectingmentalpropertiesthatareboth
categorical and non relational (or at least nonmassively relational). There are two candidate
types of properties satisfying these criteria: phenomenological and nonphenomenological
properties. In his Simulation and Interpersonal Utility, Goldman opts for the former
candidate. Accordingly, the simulator recognises her mental states on the basis of their
phenomenology,thatis,onthebasisofwhatitisliketohavetheminspecificcircumstances.
Effectively,thismeansembracinganexperientialistviewofthemeaningofmentalstates,that
is the traditional view that mental language gets its meaning, primarily and in the first
instance, from episodes of conscious experience of which the agent is more or less directly
aware.
20
ItfollowsthatthecorrespondingSTaccountofmentalascriptionisassociatedwitha
viewofmentalstatesasphenomenologicallyrealstates,towhichtheagenthasintrospective
privileged,althoughnotinfallibleaccess.
21

Wecannowexaminemorecloselyhowthetwomainmindreadingapproachesexplainthe
folks capacity to ascribe preferences to other individuals and to compare them in terms of
strength. Let us consider TT first. As seen above, TT is typically associated with a functionalist
understanding of mental states. Preferences can be defined in functionalist terms as mental
states that are causally related to certain inputs, and that, in combination with other mental
states,producecertainbehaviouraloutputs.Accordingtosomeauthors,decisiontheoryisthe

18
SeeA.Goldman,InterpretationPsychologized,op.cit.
19
See A. Goldman, The Psychology of Folk Psychology, Behavioral and Brain Sciences, 16, 1993, pp.
1528,andA.Goldman,Simulationtheoryandmentalconcepts,op.cit.
20
SeeA.Goldman,SimulationandInterpersonalUtility,op.cit,p.712.
21
See A. Goldman, The Psychology of Folk Psychology, op. cit. It is worth noticing that Goldman has
recently changed his mind about the meaning of mental states. At present, he defends the view that
mental concepts pick out categorical properties of mental states that are nonphenomenal properties.
See A. Goldman, Simulating Minds, op. cit. In what follows, I shall still focus on the version of ST
associated with experientialism for two reasons. The first is that this is the most developed version of
ST. The second is that Goldmans most recent account of the meaning of mental states is not yet
elaborated in sufficient details to provide a basis for an accurate discussion of the problem of ICs of
preferencestrength.
MauroRossiComparingPreferences

101

researchprogrammethatattemptstospecifysomeoftheserelevantrelations.
22
Inparticular,
decision theory conceives preferences as mental states that lead to choices, in combination
withbeliefsanddesires.
23
Thus,ifwedefinepreferencesinfunctionalistterms,thepropertyof
preferential strength can be conceived as the causally efficacious property, which leads an
individual to behave in a certain way, when subject to specific circumstances and in the
presenceofotherbeliefsanddesires,inaccordancewiththecausallawsdefiningthenotionof
preference.
Suppose now that an observer, e.g. a judge, wants to compare another individuals
preferences with her own in terms of strength. The first step concerns thirdperson mental
ascription. The judge observes the relevant external events (i.e. instances of the inputtypes
andoutputtypesthatareincludedinthedefinitionofpreference)andinfersboththecontent
and the strength of the other individuals relevant preferences, by reference to the causal
relations postulated by the ToM that she possesses. The second step concerns firstperson
mental ascription. Orthodox TT suggests that firstperson mental ascription entirely parallels
thirdpersonmentalascription.Thismeansthatselfascriptionisbasedoninferencesmediated
by the ToM that the judge possesses. Less orthodox TT approaches relax this position by
conjecturing that firstperson mental ascription involves the use of recognitional devices or
mechanisms which either make the use of the ToM invisible, but not completely irrelevant,
or confine it to certain specific purposes and ends up with the selfascription of both a
specificcontentandaspecificstrengthtoonesownpreferences.Thelaststageconcernsthe
interpersonalcomparisonofpreferences.Thejudgehasformedabeliefabouttheintensityof
the other individuals preferences and a belief about the intensity of her own preferences.
Straightforwardly,shecannowcombinethosebeliefstomakeaninterpersonalcomparisonof
preferencestrengths.
Let us now consider ST. As seen above, Goldmans ST version is associated with an
experientialist understanding of mental states.
24
Preferences can be defined in experientialist
terms as mental states that give raise to certain experiences in a subject. It may be the case
that there is no unique phenomenal experience that different individuals have in common
whentheyareinapreferencestate.However,itisenoughthatthereisafamilyofexperiences
that are sufficiently similar to constitute a preferencetype. According to an experientialist
understanding, then, preference strength is a felt property, a qualitative experience of the
individualthathaspreferences.Thesubjecthasintrospectiveaccessandcandiscriminatethe
strengths of his preferences. As such, preference strength is a real psychic magnitude, whose

22
See D. Lewis, Philosophical Papers. Vol. 1, Oxford University Press, Oxford 1986, P. Pettit, Decision
Theory and Folk Psychology, in Foundations of Decision Theory, ed. by M. Bacharach S. Hurley, Basil
Blackwell, Oxford 1991, pp. 147175, and P. Pettit, Preference, Deliberation and Satisfaction, Royal
InstituteofPhilosophySupplement,81,2006,pp.131154.
23
Roughly speaking, there are three possible ways to conceive the relationship between desires and
preferences.First,onecanbeeliminativistaboutpreferencesandclaimthatthenotionofpreferencesis
syncategoramatic.Itissimplyawaytoconvenientlydescribeanindividualsdesiresandtheirrelations.
However,therearenorealmentalstatescorrespondingtopreferences.Second,onecanbereductivist
andclaimthatpreferencesarerealmentalstatesbutmentalstatesthatreducetodesiresinonesense
oranother,e.g.theyconstituteaspecific,e.g.relational,classofdesires.Finally,onecanmaintainthat
preferences are derivative on desires, in the sense that they are related to, and determined by, them;
but they do not reduce to desires, except in the loose sense that they are both proattitudes of some
sort. I think that the functionalist position fits more comfortably with the latter position, which I shall
therebyadoptinwhatfollows.
24
SeeparticularlytheaccountofferedbyA.Goldman,SimulationandInterpersonalUtility,op.cit.

Humana.MenteIssue10July2009

102

meaning arises from points or intervals on the experiential scale


25
that the term denotes,
andwhichthesubjectexperiencesandcanintrospectivelydiscriminate.
Once again, suppose that one individual, e.g. the simulator, wants to compare another
individualspreferenceswithherownintermsofstrength.Thefirststepconcernsthirdperson
mental ascription. The simulator asks herself which content and intensity her preferences
would have if she had the simulated agents initial mental states. This involves recreating in
imaginationthesamequalitativeexperiencesoftheindividualwhosepreferenceshewantsto
compare. Then, the simulator discriminates the intensity of these experiences through
introspectionandclassifiesthemasexperiencesofpreferenceswithaspecificintensity.Lastly,
she ascribes such preference strengths to the other agent by analogy. The second step
concerns firstperson mental ascription. Selfascription proceeds by direct introspection. The
simulator detects her own preferences, discriminates their intensities and ascribes them to
herself. Finally, the last stage concerns the interpersonal comparison of preferences. Once
again, the simulator has formed a belief about the intensity of the other individuals
preferences and a belief about the intensity of her own preferences. She can then combine
thosebeliefstomakeaninterpersonalcomparisonofpreferencestrengths.
It is worth noticing that both TT and ST set only minimal conditions for the possibility of
forming beliefs about how different peoples preferences compare in terms of strength. In
both cases, the explanation of how ordinary people make ICs of preference strength is
consistent with the possibility that these beliefs are systematically mistaken. Two questions
arise. First, what conditions should be satisfied in order for ordinary people to make reliable
ICs of preference strength, within a TT and a ST account of their mindreading capacity?
Second,cantheseconditions,atleastinprinciple,besatisfied?

4. CONDITIONSFORRELIABLEICSOFPREFERENCESTRENGTH

InthissectionIwanttoconsiderthefirstofthepreviousquestions.LetusstartwithTTfirst.If
ordinary people make ICs of preference strength as the TT approach suggests, there are two
requirements that must be satisfied for such ICs to be reliably made. First, the judges
inferencesaboutpreferencestrengthsmustbebasedonthecorrectinputs.Second,theymust
bebasedonthecorrecttheoryaboutthecomparedindividualsmind.
The first requirement is straightforward. If the evidence that the judge usesis not correct,
thensheislikelytoreachwrongconclusionsabouttheobservedagentspreferencestrengths.
Thesecondrequirementholdsthatthejudgesinferencesmustbebasedonthecorrecttheory
aboutthecomparedagentsmind.ThismeansthatthecausallawsthatformtheToMthatthe
judge possesses must correctly represent, or at least very closely approximate, the way in
which the relevant informationprocessing mechanisms of the targeted agents work. For
simplicity,letusrefertothisrequirementastheconditionofToMtomindsimilarity.
Let us now move to ST. Summarising Goldmans own position and a large literature on ST,
wecandistinguishthreerequirementsthatmustbesatisfiedforICstobereliablymade.First,
simulation must be based on the correct inputs. Second, the simulator and the simulated
agentsmustbesimilarattheleveloftherelevantinformationprocessingmechanisms.Third,
thesimulatorsrelevantinformationprocessingmechanismsmustoperateinthesamewayin
imaginationasinreality.
The rationale underlying the first requirement is identical to the TT case. If the simulator
feedshisinformationprocessingmechanismswithincorrectinputs,thensheislikelytoreach

25
SeeA.Goldman,SimulationandInterpersonalUtility,op.cit,p.713.
MauroRossiComparingPreferences

103

wrong conclusions about the simulated agents preference strengths. One problem arises. In
Goldmans account, the inputs are pretend mental states corresponding to the agents actual
mental states. Thus, in order for simulation to be reliable, it must be the case that the
simulatorcancorrectlyindividuatetheintensityoftheagentsactualmentalstates.However,
thistaskpresentsthesamedifficultiesassociatedwiththecomparisonofdifferentindividuals
preference strengths. Thus, the assumption that the simulator can feed the correct inputs
intohisofflinesystemsimplybegsthequestion.
In the light of this problem, it seems better to take the simulated agents environmental
circumstances,ratherthanpretendmentalstates,asinputsofsimulation.Thesimulatordoes
not begin by asking herself what preferences she would have if she were to have another
individuals initial mental states. Instead, she begins by asking herself what preferences she
would have if she were in the other individuals initial circumstances. ST must be thus
complemented with causal knowledge about the relations between environment and mental
statessuchasbeliefsanddesires.Moreover,itmustbecomplementedwithknowledgeabout
the history of the simulated agent, which should be used to identify which environmental
circumstances constitute relevant inputs in specific situations, amongst the infinite ones that
the mere observation of the simulated agents situation allows one to consider. Clearly, this
movesSTtowardsamorehybridformulation.
Considernowthesecondrequirement,accordingtowhichthesimulatorandthesimulated
agent must be similar at the level of the relevant informationprocessing mechanisms. Let us
refer to it as the assumption of interpersonal psychological similarity. The rationale of this
requirement is intuitive. Even if the simulator feeds her informationprocessing mechanisms
with the correct inputs, she will reach the wrong conclusions about the agents mental states
unless they are psychologically similar in the respects that matter for forming preference
strengths.
Finally, consider the third requirement. Even if the assumption of interpersonal
psychologicalsimilarityissatisfied,sothatthesimulatorformspreferencesthataresimilarto
thoseofthetargetindividualonthebasisofactualcausalcircumstances,itmightstillbethe
casethatsheformspreferencesthatareradicallydifferentfromthoseoftheotherindividual
on the basis of imagined causal circumstances. After all, simulation works offline, whereas
interpersonal psychological similarity is a thesis about online informationprocessing
mechanisms. In order for simulation to be reliable for ICs, the previous requirement must be
complementedwiththerequirementthattheofflineworkingoftheindividualsmindsystem
approximatesitsonlineworking.

5. FIVEARGUMENTS

In this section I want to examine whether or not the conditions for having reliable ICs of
preference strength can be met. There is now considerable evidence that the third
requirement for the reliability of simulation can be satisfied.
26
Thus, here, I shall mainly focus
on the first and the second requirements for the reliability of both TT and ST. Following and
extending Goldmans analysis, I shall consider five arguments, which I shall call, respectively,
the argument from mindreading predictive success, the argument from neuroscience, the
argument from evolution, the argument from scientific practice and the argument from
nativism.

26
See,forinstance,G.CurrieandI.Ravenscroft,MentalSimulationandMotorImagery,Philosophyof
Science,64,1997,pp.161180.

Humana.MenteIssue10July2009

104

Theargumentfrommindreadingpredictivesuccessclaimsthatthefactthatmindreadingis
reliableforpredictivepurposesprovidesprimafacieevidencethatmindreadingisreliablealso
for the purpose of making IC judgments. The reason is that IC judgments are based on the
samementalascriptionsthatleadustoreliablebehaviouralpredictions.Theobjectionagainst
this argument is that, even if we grant that mindreading is reliable for predicting an agents
behaviour, we cannot conclude that mindreading is also reliable for making ICs of preference
strength.LetusseewhybyconsideringTTandSTinmoredetail.
Consider TT first. The judge ascribes preference strengths to herself and to the target,
which best predict their respective behaviour on the basis of the available evidence. The
problem is that it is possible for the judge to make both correct behavioural predictions and
incorrect ICs of preference strength. For instance, the following is one possible reason. As
preferences are characterised in functionalist terms, their intensity is relative to the intensity
of other mental states, e.g. other preferences. However, such mental states may never
becomemanifestinovertbehaviour.Itfollowsthattwoindividualspreferencestrengthsmay
appear identical, from the point of view of all the possible empirical evidence, and yet be
different, insofar as they are relative to the intensity of mental states that never generate
behaviouraloutputs.Clearly,thejudgecanmakesuccessfulbehaviouralpredictionsevenifnot
all of the agents mental states become manifest in overt behaviour. On the other hand, her
mindreading activity may be unreliable for making ICs, because there is no way to show that
theinputsthatarerelevantforICsarereallycorrect.
ConsiderSTnow.Onceagain,reliablebehaviouralpredictionsareconsistentwithdifferent
and incompatible IC judgments. This means that although it may be true that empirically
observed success at empathybased predictions of behaviour does go some distance toward
supportingpsychologicalisomorphism,
27
itisnottruethatpredictivesuccessgoesfarenough
in showing that such a psychological isomorphism is high enough to lead to correct ICs of
preference strength. One problem is that the simulator and the simulated agent may differ
with respect to one of the relevant informationprocessing mechanisms, despite the fact that
thisdifferenceneverbecomesmanifestatthepredictivelevel.Forinstance,itmaybepossible
that the simulated agent responds to the environmental inputs by forming desires with
intensity ten times greater than the simulators. If the evidence about the two individuals is
perfectly identical, the simulator can make successful predictions of the other agents
behaviourandyetincorrectICsofpreferencestrength.
Tosummarise,successatpredictinganagentsbehaviourrequiresbothalessfinegrained
individuation of the relevant inputs and a looser degree of similarity than those required for
havingreliableICsofpreferencestrength.Atbest,predictivesuccessshowsthatmindreading
is reliable for predictive purposes. However, it does not offer a reason to think that
mindreadingisreliablealsoformakingICsofpreferencestrength.
The argument from neuroscience claims that, if it is possible to establish welldefined
correlationsbetweentheintensityofthejudgesneuralactivationduringpreferenceascription
and the intensity of the agents neural activation during preference formation, then it is
possible to claim that the mechanisms underlying mindreading are isomorphic to the
mechanisms underlying preference formation. This argument specifically fits the ST
mindreadingapproach.Indeed,assimulationemploysofflinethesamemechanismsthatare
activated during online preference formation, the hypothesised neural correlation should
figure amongst the predictions of the theory. If a correlation of that sort can be robustly

27
SeeA.Goldman,SimulationandInterpersonalUtility,op.cit,p.724.
MauroRossiComparingPreferences

105

established,itmaythusbepossibletovindicatetheassumptionofinterpersonalpsychological
similarity.
This argument faces some problems. To begin with, although there is evidence that some
mental states, e.g. disgust, are located in specific brain regions and, thereby, that different
individualsundergoingthosestatessharecommonneuralproperties,thesameisnottruefor
other mental states, like preferences. Perhaps, this is simply a problem of limited empirical
evidence.Itmightaswellbethecasethatonedayscientificresearchwilldiscovertheneural
correlatesofpreferences.Ifso,theargumentfromneurosciencemayenjoyagoodfate.Even
if we grant this possibility, however, I believe that the prospects of success are dim. The
existence of a common neural region dedicated to preference formation does not imply, per
se, that an identical neural activation corresponds to the formation of preferences with
identicalstrengthacrossindividuals.
Consider the experientialist understanding of preferences associated with ST. If we grant
the possibility that the qualitative character of experiences is not fully accounted by their
neurophysiological character, it is clear that identical neuronal activation across individuals
may correspond to preference experiences that are significantly different at the level of
strength.Thisisthesameasadmittingthatinterpersonalisomorphyatthephysicalleveldoes
notnecessarilyimplyinterpersonalisomorphyatthesubjectivelevel.
Thingsdonotchangeifweadoptafunctionalistunderstandingofpreferences.Accordingto
it,preferencestrengthsareindividuatednotonlywithrespecttoexternalinputsandoutputs,
butalsowithrespecttoothermentalstates.Crucially,thesestatescanbebothoccurrentand
nonoccurrent.Theproblemisthatneuralactivationregistersonlyoccurrentmentalstates.In
order to conclude that different individuals preference strengths are the same when their
neuronsfirewiththesameintensity,weneedtoassumethattheyareidenticalwithrespectto
all those nonoccurrent mental states which might impact on their occurrent preference
strengths. However, we have no epistemic reason to accept this ceteris paribus assumption.
Once again, the result is that interpersonal isomorphy at the physical level does not imply
interpersonalisomorphyatthefunctionallevel.
If the previous points are correct, the hypothesised correlation between the judges
mindreading mechanisms and the agents preference formation mechanisms does not prove
thatICscanbereliablymade.Thereasonisthefollowing.Althoughtheactivationoftheneural
region dedicated to mindreading may covary with the activation of the neural region
dedicated to preference formation, it may still be the case that the judge does not get the
other individuals preference strengths right. In other words, interpersonal neural correlation
maybeconsistentwithsystematicerrorsaboutpreferencestrengthattribution.Therefore,the
argumentfromneurosciencedoesnotsupportthehypothesisthatICsofpreferencestrength
canbereliablymade.
The argument from evolution claims that evolutionary pressure might have favoured the
development of a close isomorphism between the observers ToM and the targets
informationprocessing mechanisms, in the case of TT, or between the simulators and the
simulatedagentsinformationprocessingmechanisms,inthecaseofST.Thereasonisthatthis
wouldhavemaximisedtheexpectedfitnessofthemembersofarelevantgroupbyendowing
them with competitively advantageous features for the typical environment encountered by
the group. Roughly speaking, individual fitness is assessed with respect to the (probabilistic)
propensityofspreadingonesowngenesintothenextgeneration.Clearly,thispropensitymay
beenhancedbytheindividualscapacitytocorrectlypredict,explainorinterpretothergroup
members behaviour. If this is the case, evolutionary pressure might have favoured the
development of a strong convergence in the working of different individuals mindreading
mechanisms and, thereby, a high level of intersubjective agreement about the outcomes of

Humana.MenteIssue10July2009

106

mindreading. The problem is that, as I shall illustrate below, intersubjective agreement


presupposes a less demanding degree of similarity than the one required for having reliable
IUC judgments. Given that a higher degree of similarity would be unnecessarily costly, it
follows that the argument from evolution does not support the claim that ICs of preference
strengthcanbereliablymade.
Let us consider TT first. Suppose there are two individuals trying to mindread each other.
Suppose that the ToM that each of them uses closely represents the working of the other
individuals mind. Finally, suppose that the empirical evidence about their preferences is
identical. On the basis of the ToM in their possession, they will both conclude that they have
the same preference strengths. However, it may as well be the case that the first individuals
occurrent preference strengths are relative to a different set of nonoccurrent mental states,
which never become manifest. The consequence is that intersubjective agreement is
consistent with the case where ICs of preferences are unreliably made. Intersubjective
agreementmerelyrequiresthattheindividualsmakesimilarassumptionsaboutasuitableset
of nonoccurrent mental states, i.e. those that have a chance of becoming manifest in overt
behaviour.
Let us now consider ST. Suppose that there are two individuals simulating each others
mentallife.Supposethatbothindividualsarecompletelyidenticalattheleveloftherelevant
informationprocessing mechanisms, except for the fact that the first individual forms desires
with intensity ten times greater than the second individuals, when responding to the same
environmentalstimuli.Iftherelevantevidenceisthesame,bothindividualswillconcludethat
they have identical preference strengths. Indeed, both individuals ascribe preferences to the
other on the basis of their own cognitive machinery, under the assumption that the other
individual is similar to them in the relevant respects. However, by so doing, they make
incorrect ICs of preference strength, because, by stipulation, the intensity of the first
individuals desires is ten times greater than the intensity of the second individuals desires.
Intersubjective agreement requires a looser degree of interpersonal psychological similarity
thantheonerequiredformakingreliableICs.
The argument from scientific practice starts from the observations that, on the one hand,
mostscientifictheoriesareunderdeterminedbytheempiricalevidence,likeICsofpreference
strength,andyetthat,ontheotherhand,thereareoftenreasonstopreferonetheoryrather
than another on grounds of simplicity, parsimony and nonarbitrariness. These are pragmatic
virtues that characterise scientific practice and that help break the underdetermination of
scientific theories by the empirical evidence. Applied to ICs of preference strength, the
argumentworksinthefollowingway.Supposethattwodifferentjudges,inthecaseofTT,or
simulators, in the case of ST, reach intersubjective agreement about other individuals
preference strengths. It is true that preference strengths may be relative to other mental
states that never become manifest or that there may be hidden interpersonal differences
concerning the relevant informationprocessing mechanisms, but if all the empirical evidence
is otherwise the same, the best explanation of the mindreaders intersubjective agreement is
that the other individuals preference strengths are really the same. In both cases, the
recommended conclusion looks like the simplest, the most parsimonious and the least
arbitraryhypothesis.
The objection against this line of thought is that it contrasts with the very explanatory
practiceinthemindreadingliterature.InthecaseofTT,thebestexplanationofintersubjective
agreement in mindreading is that different judges take their theories about other peoples
mindtobecorrect,ontheonehand,andtheceterisparibusassumptionaboutnonoccurrent
mental states to be satisfied, on the other hand. This does not mean that the ToM in their
MauroRossiComparingPreferences

107

possession or the inputs that they consider are really correct. Likewise, in the case of ST, the
best explanation of intersubjective agreement is that different simulators take other
individualstobejustlikethemattheleveloftherelevantinformationprocessingmechanisms.
Onceagain,thisdoesnotmeanthatsimulatorsandsimulatedagentsarereallyalike.Inother
words, the mindreading literature explains intersubjective agreement by holding that the
judges or the simulators merely take the reliability requirements to be satisfied, without
holding that they really are. If this the case, the argument from scientific practice does not
showthatICsofpreferencestrengthcanbereliablymade.
The argument from nativism is specifically advocated by Goldman and pursues an analogy
with Chomskys nativist approach in linguistics.
28
Chomskys analysis starts from the
observation that children belonging to the same community end up acquiring the same
grammar. This fact is particularly striking because grammar acquisition is radically
underdetermined by the empirical evidence. According to Chomsky, it is not plausible to
assume that children use purely pragmatic criteria, such as simplicity and parsimony, in order
to learn a common grammar amongst the infinitely many possible ones that are consistent
with the available empirical evidence. Instead, Chomsky suggests that children possess an
innate and universal body of knowledge, which guides them in the process of language
learning. Such a body of knowledge is not only important during the development process.
Indeed, it is the very body of knowledge on which the grammaticality judgments of adult
competentspeakersarebased.
Goldman invites us to conceive the problem of ICs in analogy with linguistics. The starting
point is the observation that different observers reach frequent intersubjective agreement
aboutICsofpreferencestrength.Thisfactisparticularlystrikingbecause,aswehaveseen,ICs
are radically underdetermined by the empirical evidence. As the analogy with linguistics
suggests,itisnotplausibletoassumethatdifferentobserversformthesamebeliefs,amongst
the infinite ones licensed by the empirical evidence, on the basis of purely pragmatic
considerations.
29
Rather, it is more plausible to hold that they form the same beliefs on the
basisofthepossessionaninnateandhighlyrepresentativeToM,inthecaseofTT,andinnate
and highly similar informationprocessing mechanisms, in the case of ST. According to
Goldman, if the nativist hypothesis gives linguistics epistemic respectability, so does it with
ICsofpreferencestrength.
30

The crucial concept is that of innateness. The question of what innateness is has
generatedaparticularlyintensephilosophicaldebateinthepastfewyears.
31
Althoughthereis
an evident lack of agreement in the literature, the most recent positions suggest taking
nativism as equivalent to psychological primitivism.
32
Roughly speaking, innate cognitive
capacities are psychological primitives. In turn, psychological primitives are entities or
processes that, on the one hand, are mentioned in the best psychological explanations of
human behaviour; and, on the other hand, whose acquisition cannot be explained by any
psychological theories, but only by a theory at a lower level. The important issue is to see

28
SeeN.Chomsky,Rulesandrepresentations,BasilBlackwell,Oxford1980.
29
Cfr. J. Harsanyi, Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons of Utility, op.
cit, and J. Harsanyi, Rational Behaviour and Bargaining Equilibrium in Games and Social Situations,
CambridgeUniversityPress,Cambridge1977.
30
SeeA.Goldman,[1995],pp.725726.
31
See F. Cowie, Whats Within? Nativism Reconsidered, Oxford University Press, Oxford 1999, P.
Griffiths,Whatisinnateness?,Monist,85,2002,pp.7085,R.Samuels,Nativismincognitivescience,
Mind & Language, 17, 2002, pp. 233 265, M. A. Khalidi, Innate Cognitive Capacities, Mind and
Language,22,2007,pp.92115.
32
SeespeciallyF.Cowie,op.cit,andR.Samuels,op.cit.

Humana.MenteIssue10July2009

108

whetherornotthereisscientificevidencesupportingthenativisthypothesisinthecaseofICs
ofpreferencestrength.
The first objection is that, even if such evidence exists, it does not show that ICs of
preference strength can be reliably made. After all, nativism is a hypothesis about the
acquisition of the mindreading capacity rather than a hypothesis about its reliability. In order
toturntheargumentfromnativismintoanargumentoftherightkind,weshouldreformulate
it as claiming that the reliability of the relevant informationprocessing mechanisms
constitutesaprimitiveinatheoryofhowordinarypeoplemakeICsofpreferencestrength,in
the sense that the best explanation of their comparative capacity and of their frequent
intersubjective agreement requires such property. In the light of the previous discussion,
however,wecaneasilyrejecteventhisalternativeversionofGoldmansargument.Aswehave
seen above, the best explanation of peoples comparative practice and intersubjective
agreement merely requires the assumption that ordinary people take the reliability
requirementstobesatisfied,nottheyreallyare.Therefore,eventheargumentfromnativism
fails.

6.CONCLUSION

SolvingtheproblemofICsofpreferencestrengthisofvitalimportanceforwelfareeconomics,
social choice theory and ethics. It is therefore not surprising that several attempts have been
made in the course of the years. In this paper, I examined one strategy in particular, which
tries to solve the problem of ICs by looking more closely at how ordinary people make ICs of
preference strength in everyday life. The question that I specifically considered is whether or
not there is scientific evidence that ordinary people make ICs of preference strength in a
reliable way. I discussed five arguments, which offer various reasons to think that the answer
tothisquestionisaffirmative.Inthispaper,Iarguedthattheyallfail.Thisdoesnotmeanthat
no positive argument exists, which can vindicate the strategy under scrutiny or, more
generally, solve the problem of ICs. However, given that all the solutions explored so far turn
out to be unsuccessful, this paper provides prima facie reason to think that scepticism about
ICsmaybeheretostay.

BIBLIOGRAPHY

ArrowK.(1995)SocialChoiceandIndividualValues,2
nd
ed.,Wiley,NewYork1963(1951,1
st

ed.).
BaronCohen, S. Mindblindness: An Essay on Autism and Theory of Mind, MIT Press,
Cambridge,Mass.,foradefenceoftheMTapproach.
ChomskyN.(2002)Rulesandrepresentations,BasilBlackwell,Oxford1980.
Cowie, F. Whats Within? Nativism Reconsidered, Oxford University Press, Oxford 1999, P.
Griffiths,Whatisinnateness?,Monist,85,pp.7085.
Currie G. and Ravenscroft I. (1997) Mental Simulation and Motor Imagery, Philosophy of
Science,64,pp.161180.
Davidson D. (1986) Judging interpersonal interests, in Foundations of social choice theory,
ed.byJ.ElsterA.Hylland,CambridgeUniversityPress,Cambridge,pp.195211.
MauroRossiComparingPreferences

109

DavidsonD.(2004)ProblemsofRationality,OxfordUniversityPress,Oxford.
FeldmanR.andConeeE.(1985)Evidentialism,PhilosophicalStudies,48,1985,pp.1534.
Goldman A. (1979) What is Justified Belief?, in Justification and Knowledge, ed. by G.
Pappas,KluwerAcademicPublisher,Reidel,pp.123.
GoldmanA.(1989)InterpretationPsychologized,MindandLanguage,4,pp.161185.
GoldmanA.(1992)Indefenseofthesimulationtheory,MindandLanguage,7,pp.104119.
Goldman A. (1993) The Psychology of Folk Psychology, Behavioral and Brain Sciences, 16,
pp.1528.
GoldmanA.(1995)SimulationandInterpersonalUtility,Ethics,4,1995,pp.709726.
GoldmanA.(2000)Thementalizingfolk,inMetarepresentations,ed.byD.Sperber,Oxford
UniversityPress,Oxford.
GoldmanA.(2002)Simulationtheoryandmentalconcepts,inSimulationandKnowledgeof
Action,ed.byJ.DokicJ.Proust,JohnBenjamins,Amsterdam,pp.120.
GoldmanA.(2006)SimulatingMinds,OxfordUniversityPress,Oxford.
Gopnik,J. and Wellman H. (1992) Why the childs theory of mind reallyis a theory of mind,
MindandLanguage,7,pp.145171.
Gopnik J. and Wellman H.(1994) The theory theory, in Mapping the Mind: Domain
Specificity in Cognition and Culture, ed. by L. Hirschfiled S. Gelman, Cambridge
UniversityPress,NewYork.
Gopnik J. and Meltzoff A. N. (1997) Words, Thoughts and Theories, MIT Press, Cambridge,
Mass.
Harsanyi J. (1955) Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons
ofUtility,TheJournalofPoliticalEconomy,63,pp.309321.
Harsanyi J. (1977) Rational Behaviour and Bargaining Equilibrium in Games and Social
Situations,CambridgeUniversityPress,Cambridge.
Jeffrey R. (1983) The Logic of Decision, 2
nd
ed., University of Chicago Press, Chicago (1965,
1
st
ed.).
JevonsS.(1911)TheoryofPoliticalEconomy,4
th
edition,Macmillan,London(1871,1
st
ed.).
Khalidi,M.A.(2007)InnateCognitiveCapacities,MindandLanguage,22,2007,pp.92115.
LeslieA.(1987)Pretenceandrepresentation:Theoriginsoftheoryofmind,Psychological
Review,94,1987,pp.412426.
LeslieA.(1988)Someimplicationsofpretenseformechanismsunderlyingthechildstheory
of mind, in Developing Theories of Minds, ed. by J. Astington P. Harris D. Olson,
CambridgeUniversityPress,Cambridge,pp.1946.
LeslieA.(1994)PretendingandBelieving:IssuesinthetheoryofToMM,Cogniton,50,,pp.
211238.
LeslieA.(2000)Howtoacquirearepresentationaltheoryofmind,inMetarepresentation:
AMultidisciplinaryPerspective,ed.byD.Sperber,OxfordUniversityPress,NewYork,
pp.197223.

Humana.MenteIssue10July2009

110

Leslie A. and German T. (1995) Knowledge and ability in theory of mind: Oneeyed
overview of a debate, Mental Simulation, in ed. by M. Davies T. Stone, Blackwell,
Oxford,pp.123150.
LewisD.(1986)PhilosophicalPapers.Vol.1,OxfordUniversityPress,Oxford.
List L. (2003) Are Interpersonal Comparisons of Utility Indeterminate?, Erkenntnis, 58, pp.
229260.
LittleI.D.M.(1957)ACritiqueofWelfareEconomics,2
nd
ed.,ClarendonPress,Oxford(1950,
1
st
ed.).
NicholsS.etal.(1996)VarietiesofOffLineSimulation,inTheoriesofTheoriesofMind,ed.
byP.CarruthersP.Smith,CambridgeUniversityPress,Cambridge,pp.3974.
Nichols, S. and Stich, S. (2003) Mindreading: An Integrated Account of Pretence, Self
Awareness,andUnderstandingOtherMinds,OxfordUniversityPress,Oxford.
PernerJ.(1991)UnderstandingtheRepresentationalMind,MITPress,Cambridge,Mass.
PettitP.(1991)DecisionTheoryandFolkPsychology,inFoundationsofDecisionTheory,ed.
byM.BacharachS.Hurley,BasilBlackwell,Oxford,pp.147175.
Pettit P. (2006) Preference, Deliberation and Satisfaction, Royal Institute of Philosophy
Supplement,81,2006,pp.131154.
RobbinsL.(1932)AnEssayontheNatureandSignificanceofEconomicScience,Macmillan,
London.
SamuelsR.(2002)Nativismincognitivescience,Mind&Language,17,pp.233265.
SavageL.(1954)TheFoundationsofStatistics,Wiley,NewYork.
Stich S. and Nichols S. (1992) Folk Psychology: Simulation or Tacit Theory?, Mind and
Language,7,pp.3536.
Stich S. and Nichols S. (1995) Second Thoughts on Simulation, in Folk Psychology: The
TheoryofMindDebate,ed.byM.DaviesT.Stone,Blackwell,Oxford,pp.87108.
StichS.andNicholsS.(1997)CognitivePenetrability,RationalityandRestrictedSimulation,
MindandLanguage,12,1997,pp.297326.
von Neumann J. and Morgenstern O. (1944) Theory of Games and Economic Behavior,
PrincetonUniversityPress,Princeton.
Waldner I. (1972) The Empirical Meaningfulness of Interpersonal Utility Comparisons, The
JournalofPhilosophy,4,pp.87103.
WeintraubR.(1998)DoUtilityComparisonsPoseaProblem?,PhilosophicalStudies,92,pp.
307319.
WellmanH.(1990)TheChildsTheoryofMind,MITPress,Cambridge,Mass.



111

Does Freedom of Cboice cause Satisfaction?

Motqbetlto 8otteto
*^

MargherlLa.8oLLero[hhs.se


A8S1kAC1

1hls paper lnvesLlgaLes Lhe relaLlon beLween freedom of cholce and lndlvldual well belng, as measured
by saLlsfacLlon. l look aL analyse Lhe hypoLhesls of a poslLlve correlaLlon beLween Lhe Lwo varlables,
conLrolllng for oLher relevanL covarlaLes and unobserved heLerogenelLy. 1he resulLs show LhaL freedom
ls slgnlflcanLly correlaLed wlLh saLlsfacLlon, a flndlng robusL Lo several alLernaLlve model speclflcaLlons.


IN1kCDUC1ICN

Well belng has recenLly encounLered renewed lnLeresL as a pollcy ob[ecLlve, moLlvaLlng Lhe
flourlshlng of a new branch of economlcs, known as economlcs of happlness. arL of Lhls
llLeraLure ls focused on uncoverlng Lhe deLermlnanLs of well belng, ofLen undersLood as
saLlsfacLlon over one's llfe or feellng of happlness. 8y now, however, vlrLually no sLudy has
lnvesLlgaLed Lhe relaLlvely lnLulLlve quesLlon of wheLher freedom maLLers for lndlvldual well-
belng. lf one undersLands freedom as freedom of cholce
1
(loC), Lhe exlsLlng Lheory endorses a
poslLlve answer (aLLalnx, xu, 1990, Sen, 1999 and Sugden, 1998). rovldlng emplrlcal evldence
ln supporL of Lhls clalm encounLers a Lwofold problem. llrsL, lL ls noL yeL clear how Lo correcLly
measure freedom, slnce Lhere ls no unlque deflnlLlon of lL. Second, freedom ls mosL llkely Lo be
endogenous wlLh oLher measures of well belng, and Lhe relaLlon ls wldely blurred by Lhe
presence of unobserved heLerogenelLy. uoes more freedom cause hlgher saLlsfacLlon or ls
belng saLlsfled wlLh one's llfe Lhe cause of hlgher self reporLed freedom? 1o whaL exLenL are
unobservable eLhlcs or value sysLems affecLlng Lhe relaLlonshlp under sLudy?
ln Lhls paper l wlll mosLly concenLraLe on Lhe problem of unobservables. As for Lhe
LheoreLlcal parL, l wlll be measurlng loC as AuLonomy lreedom (Sugden, 1998, 8aveLLa and
eraglne, 2006, 8aveLLa and Cuala, 2003). AuLonomy lreedom ls a meLrlc of freedom wlLh a
formal characLerlzaLlon (8aveLLa and eraglne, 2006), malnLalnlng LhaL Lhe essence of freedom
ls havlng opporLunlLles Lo choose from ln a consclous way. More lmporLanLly ln Lhls conLexL,
daLa on AuLonomy lreedom are collecLed ln a publlcly avallable daLaseL
2
(for a [usLlflcaLlon,
see 8aveLLa eL al., 2008). Some prellmlnary research (8aveLLa eL al., 2008) reporLs LhaL loC ls
correlaLed wlLh several lmporLanL soclo-demographlc varlables, as lncome, employmenL,
marlLal sLaLus and rellglon. 1hls evldence ls suggesLlve LhaL loC may be relevanL for soclal
sLudles, furLher moLlvaLlng Lhe lnLeresL ln how lL relaLes Lo lndlvldual well belng.
Well belng ls a broad concepL wlLh a wlde array of deflnlLlons and measures. ln whaL
follows, l wlll measure lL as saLlsfacLlon. lrom a phllosophlcal perspecLlve, ln facL, belng free Lo
choose lnduces an lndlvldual Lo be "Lhe masLer of one's own llfe" (Mlll, 1839), by allowlng hlm
or her Lo Lake Lhe cholces more approprlaLe Lo hls or her llfesLyle. Lmplrlcally, Lhls means LhaL

*
ueparLmenL of Lconomlcs SLockholm School of Lconomlcs
^
1he auLhor Lhanks Lhe Wallander/Pedellus loundaLlon for flnanclal supporL.
1
Where freedom of cholce means Lhe exLenL Lo whlch an lndlvldual has opporLunlLles allowlng hlm Lo
Lake meanlngful cholces.
2
1he World values Survey and Lhe Luropean values Survey.

Bumana.Nente - Issue 1u- }uly 2uu9

112

loC should be expecLed Lo lmpacL measures of lndlvldual saLlsfacLlon wlLh one's own llfe,
where saLlsfacLlon comes from Lhe appralsal of one's consclous cholces (8aveLLa and Cuala,
1998).
Culded by Lhe above conslderaLlons, l wlll lnvesLlgaLe Lhe hypoLhesls LhaL loC relaLes
poslLlvely wlLh Lhe saLlsfacLlon LhaL an lndlvldual feels wlLh hls or her llfe. Whlle l do noL
address Lhe problem of endogenelLy, l Lry Lo deal wlLh Lhe unobserved heLerogenelLy lssue by
conLrolllng for flxed effecLs. 1he resulLs show LhaL Lhere ls a sLrong, poslLlve correlaLlon
beLween Lhe loC and saLlsfacLlon, holdlng across all speclflcaLlons, wlLh and wlLhouL flxed
effecL conLrols.
1he remalnder of Lhe paper ls organlzed as follows: Lhe nexL secLlon qulckly revlews Lhe
prevlous research. AfLer LhaL, Lwo secLlons dlscuss Lhe daLaseL and Lhe emplrlcal sLraLegy.
SecLlon 3 provldes Lhe resulLs and Lhe followlng one commenLs Lhe emplrlcal evldence. A
concludlng secLlon follows.

kLVICUS kLSLAkCn

1he exlsLlng llLeraLure ls relevanL for Lhe presenL work ln Lwo respecLs. llrsL, Lhey wlll gulde Lhe
cholce of Lhe relevanL varlables Lo be lncluded ln Lhe regresslon of loC on saLlsfacLlon. Second,
Lhey wlll provlde a benchmark agalnsL whlch Lo LesL Lhe conslsLency of Lhe esLlmaLes.
AlmosL all sLudles flnd LhaL lncome lncreases wlLh saLlsfacLlon (Schnys, 1998, lrey and
SLuLzer 2002), and a slmllar correlaLlon holds for belng employed, especlally self-employed
(Cswald, 1997), havlng a hlgher educaLlon and llvlng ln a blg clLy. SaLlsfacLlon ls also sLrongly
poslLlvely relaLed wlLh belng marrled and belng male, whlle lL ls sLlll debaLable wheLher lL
lncreases or noL wlLh havlng chlldren (lrey and SLuLzer, 2006). Several works show LhaL
saLlsfacLlon ls a convex funcLlon of age, wlLh a peak down around Lhe mld-forLles. Moreover,
belng rellglous ls a deLermlnanL of saLlsfacLlon (Clark and Lelkes, 2003). lor a more
comprehenslve revlew of Lhe sLudles, see 8[ornskov eL al. (2008). 1o my knowledge, no sLudy
lncludes a measure of loC among Lhe covarlaLes.
LsLabllshlng a clear axlomaLlc measure for loC has been on Lhe agenda of Soclal Cholce
Lheory for many years. aLLanalx and xu (1990) suggesLed LhaL loC should be measured by Lhe
number of opLlons avallable Lo an lndlvldual. AlLernaLlvely, Sen proposed Lo look aL Lhe
number of opLlons avallable Lo and preferred by Lhe declslon maker (Sen, 1991).
ln Lhls work, l wlll measure loC as AuLonomy lreedom, a concepLuallzaLlon due Lo Sugden
(1998) and 8aveLLa eL al. (1997, 2003). 1hey malnLaln LhaL a meLrlc of freedom should capLure
Lhe exLenL Lo whlch a cholce seL allows Lhe declslon maker Lo come Lo choose an opLlon x
raLher Lhan Lo whaL exLenL lL glves access Lo (ex posL preferred) opLlons, wlLhouL any
requlremenL on Lhe declslon process (Mlll, 1898, 8aveLLa and Cuala, 1997, 2003). 1here are
Lwo reasons why l favour Lhls noLlon. lL ls an lnLeresLlng pollcy varlable because lL capLures Lhe
value of havlng opporLunlLles, whlch recenLly has been cenLral for many pollLlcal goals
3
and lL ls
posslble Lo work wlLh lL emplrlcally by uslng Lhe followlng quesLlon from Lhe Luropean value
survey
4
.

3
See accepLance speech dellvered by presldenL-elecL 8arak Cbama,
hLLp://www.markeLwaLch.com/news/sLory/LexL-barack-obamas-accepLance-
speech/sLory.aspx?guld=78323A921u-6L3l-4103-8A81-23A1ACL29L8L7u
4
ln Lhe orlglnal work, Lhe auLhors refer Lo Lhe World value Survey, buL Lhe quesLlon ls unchanged ln Lhe
more handy Luropean value Survey.
Naigheiita Botteio - Boes Fieeuom of Choice cause Satisfaction.

113

Some people feel Lhey have compleLely free cholce and conLrol over Lhelr llves, whlle oLher
people feel LhaL whaL Lhey do has no real effecL on whaL happens Lo Lhem. lease use Lhls
scale where 1 means "none aL all" and 10 means "a greaL deal" Lo lndlcaLe how much freedom
of cholce and conLrol you feel you have over Lhe way your llfe Lurns ouL (for a [usLlflcaLlon,
8aveLLa eL al., 2008). Moreover, prellmlnary work shows LhaL Lhls measure has some sLaLlsLlcal
slgnlflcance ln relaLlon wlLh oLher soclo-demographlc varlables (8aveLLa eL al., 2008), buL Lhe
lssue of lLs relaLlon wlLh saLlsfacLlon ls noL sLudled Lhere
3
.

1nL DA1A

1he daLa LhaL l am uslng are Laken from Lhe worldwlde lnLegraLed Luropean values Survey
6
, a
survey LhaL gaLhers lnformaLlon on how people percelve Lhemselves and Lhe socleLy around
Lhem, slmllar ln splrlL Lo LurobaromeLer
7
and Lhe lSS survey
8
. 1he sample has a cross secLlon
dlmenslon conslsLlng of more Lhan 230,000 lndlvlduals from roughly 60 counLrles worldwlde,
and a Llme serles of four waves, 1981-1984, 1989-1992, 1994-1997, and 1999-2003.
1o measure well belng, l wlll be uslng daLa on saLlsfacLlon, as gauged by Lhe answer Lo Lhe
quesLlon All tbloqs cooslJeteJ, bow sotlsfleJ ote yoo wltb yoot llfe os o wbole tbese Joys?
8espondenLs provlded an answer ranglng from 1 Lo 10, wlLh 1 meanlng LoLal dlssaLlsfacLlon
and 10 compleLe saLlsfacLlon. AparL from loC, l wlll be uslng Len oLher covarlaLes, whose
descrlpLlve sLaLlsLlcs are dlsplayed ln llgure 1.

Covar|ate Cbs. Mean Std. Dev. Var|ance Sma||est Largest
SA1 263097 6.61 2.48 6.18 1 10
loC 248224 6.64 2.44 3.97 1 10
lnC 228938 4.67 2.47 6.13 1 11
LML 239689 3.18 2.13 4.64 1 8
1CWn 181216 4.88 2.31 6.32 1 9
Luu 187668 4.44 2.27 3.19 1 8
CPlLu 244024 1.93 1.70 2.90 0 20
M.S1A1uS 263038 2.62 2.17 4.73 1 7
8l81P 264839 1933 17.46 304.83 1881 1988
SLx 267660 1.3 .49 .24 1 2
8LL_0 238328 1.3 .36 .31 1 4
8LL_1 204721 36.2 12.81 164.16 1 86
CL 193331 3.36 2.26 3.13 1 10
I|g. 1. uescrlpLlve SLaLlsLlcs


MosL of Lhem are self explanaLory. Marrled ls a dummy Laklng value 1 lf slngle and 2 lf Lhe
respondenL ls marrled. Sex ls coded wlLh 1 belng 'Male' and 2 'lemale'. 8LL_0 sLands for Lhe
quesLlon loJepeoJeotly of wbetbet yoo qo to cbotcb ot oot, woolJ yoo soy yoo ote o tellqloos
petsoo?, wlLh posslble answers: 1 'A rellglous person', 2 'noL a rellglous person', 3 'A convlnced

3
1he auLhors regress loC on a seL of covarlaLes (sex, age, marlLal sLaLus, educaLlon level, lncome and
employmenL sLaLus, pollLlcal orlenLaLlon, LrusL level, lndlvldual's supporL ln markeL-orlenLed lnsLlLuLlons,
rellglous bellefs), buL wlLhouL accounLlng for flxed effecLs.
6
hLLp://www.europeanvalues.nl/.
7
hLLp://ec.europa.eu/publlc_oplnlon/.
8
hLLp://www.lssp.org/.

Bumana.Nente - Issue 1u- }uly 2uu9

114

aLhelsL' and 4 'CLher answer'. llnally, CL corresponds Lo Lhe quesLlon: lo polltlcol mottets,
people tolk of "tbe left" ooJ "tbe tlqbt." now woolJ yoo ploce yoot vlews oo tbls scole,
qeoetolly speokloq?, whose answers are coded wlLh 1 meanlng 'LefL' and 10 '8lghL'.

1nL LMIkICAL S1kA1LG

Cne way Lo esLlmaLe emplrlcally Lhe relaLlon beLween loC and SaLlsfacLlon ls Lo Lry Lo conLrol
as much as posslble for Lhe large unobservable heLerogenelLy lylng ln Lhe background. More
speclflcally, lL ls llkely LhaL some of Lhe unobservable varlables are connecLed wlLh Lhe effecL of
belng ln a cerLaln counLry (counLry effecL) and belng lnLervlewed ln a cerLaln wave (Llme
effecL). Conslder for lnsLance Lhe parLlclpanLs ln Lhe wave 1989-1992: Lhe effecL produced on
Lhelr reporLed answers by Lhe fall of Lhe 8erlln wall ls per se unobservable, buL lL ls posslble Lo
capLure lL by lncludlng Llme effecLs. Slmllarly, counLry effecLs accounL for Lhose eLhlcs or values
LhaL are speclflc Lo a cerLaln area and affecL Lhe relaLlon beLween loC and SaLlsfacLlon ln ways
LhaL are lnLrlnslcally unobservable.
Clven Lhe panel naLure of Lhe daLaseL LhaL l am uslng, whlch Lracks a cross secLlon of
counLrles across Llme, Lhe naLural Lechnlque Lo employ ls Lo esLlmaLe a model wlLh flxed
effecLs. l wlll lnvesLlgaLe Lhe parLlal effecL of Lhe observable lndependenL varlables descrlbed ln
llgure (1) on SaLlsfacLlon assumlng Lhe followlng model

L[SA1 x, c]
SA1
lL
= x
lL
+c
l
+u
lL
(1)

where x
lL
ls a vecLor conLalnlng Lhe observable covarlaLes
9
, c
l
a vecLor of unobservable random
varlables and u
lL
a ldlosyncraLlc error Lerm, where l refers Lo Lhe counLry and L Lo Lhe wave.
Accordlng Lo Lhe Lheory, Lo conslsLenLly esLlmaLe models llke (1), Lhe explanaLory varlables
x
lL
need Lo be uncorrelaLed wlLh Lhe error Lerm, alLhough only once Lhe c
l
's are accounLed for
10
.
Moreover, slnce Lhe panel daLa l am uslng ls unbalanced, ln Lhe sense LhaL some counLrles
drop off Lhe cross secLlon ln some perlods, conslsLency requlres also Lhe x
lL
Lo be uncorrelaLed
wlLh Lhe drop-off paLLern.
lormally, Lhe sLrlcL exogenelLy assumpLlon reads

L(u
lL
x
l
, s
l,
c
l
)=0 L(s
lL
, x
ls
u
lL
)=0 s,L=1,2...,1 (2)

where s
lL
ls selecLlon lndlcaLor vecLor s
l
(s
l1
, s
l2
, ...,s
l1
), such LhaL s
lL
=1 lf (SA1
lL
, x
lL
) obLalned,
zero oLherwlse.
lnLulLlvely, Lhls assumpLlon ls llkely Lo be saLlsfled. 1he lndependence of Lhe covarlaLes wlLh
Lhe error Lerm should be sufflclenLly warranLed by Lhe condlLlonlng on Lhe flxed effecLs, and
Lhe correlaLlon wlLh Lhe aLLrlLlon paLLern should be zero, glven LhaL drop-offs ln Lhe daLaseL are
lnLrlnslcally random, belng due Lo Lechnlcal consLralnLs ln Lhe survey deslgn.
l wlll esLlmaLe model (1) by runnlng four dlfferenL regresslons, whose resulLs are dlsplayed
ln Lhe nexL secLlon. llrsL, l wlll esLlmaLe a basellne regresslon, whlch capLures Lhe relaLlon
beLween SaLlsfacLlon and loC wlLhouL accounLlng for unobservable heLerogenelLy:


9
loC, lncome, Slze of 1own, LducaLlon, LmploymenL, MarlLal sLaLus, number of Chlldren, 8ellglous
personor noL, ollLcal self poslLlonlng, Sex, ?ear of 8lrLh.
10
sLrlcL exogenelLy assumpLlon (see Wooldrldge, p. 379).
Naigheiita Botteio - Boes Fieeuom of Choice cause Satisfaction.

113

SA1
lL
=
0lL
+loC
l
+x
lL

x
+du
lL


(3)

1hen l wlll add a counLry flxed effecL dC
l


SA1
lL
=
0lL
+loC
l
+x
lL

x
+dC
l

C
+du
lL
(4)

where dC
l
ls a vecLor of counLry dummles, so LhaL dC
l
=1 lf l=C and zero oLherwlse. 1he Lhlrd
model speclflcaLlon accounLs for Llme effecLs d1
L
ln a slmllar way

SA1
lL
=
0lL
+loC
l
+x
lL

x
+d1
L

L
+du
lL
(3)

where d1
L
ls a vecLor of Llme dummles such LhaL ds
L
=1 lf L=1 and L=1,2,3,4 and zero oLherwlse.
llnally, l have a compleLe speclflcaLlon lncludlng counLry flxed effecLs, Llme flxed effecLs and
counLry speclflc Llme effecLs, where Lhe laLLer capLures all Lhe unobservables relaLed wlLh
belng ln a cerLaln counLry ln a cerLaln wave

SA1
lL
=
0lL
+loC
l
+x
lL

x
+dC
l

C
+d1
L

L
+d(C
l
1
L
)
lL
+du
lL
(6)

where obvlously d(C
l
1
L
) ls a dummy such LhaL d(C
l
1
L
)=1 lf l=C and L=1 and zero oLherwlse.
1hls procedure recovers Lhe parLlal effecL of loC on Lhe dependenL varlable, and Lhe
esLlmaLes wlll be conslsLenL as long as Lhe exogenelLy assumpLlon ls saLlsfled.

kLSUL1S

1he resulLs from regresslon (3)-(6) are dlsplayed ln Lhe followlng 1ables. As Lhe LheoreLlcal
research on loC suggesLs, Lhe esLlmaLes lmply a large, poslLlve and sLaLlsLlcally slgnlflcanL
assoclaLlon beLween loC and SaLlsfacLlon. lor each speclflcaLlon, Lhe esLlmaLed coefflclenL
varles a llLLle ln slze accordlng Lo Lhe esLlmaLlon Lechnlque used (CLS, roblL, roblL wlLh
survey opLlon). Powever, for each Lechnlque, esLlmaLes do noL vary across Lhe dlfferenL
speclflcaLlons, remalnlng slgnlflcanL, of Lhe same slgn and of abouL Lhe same slze.
LsLlmaLlon resulLs for (3) are shown ln 1able (1). As menLloned above, Lhe magnlLude of
Lhe relaLlve effecL of loC drops from 0.34 Lo 0.2 when (3) ls esLlmaLed wlLh a mulLlnomlal
problL esLlmaLlon Lechnlque raLher Lhan wlLh pooled CLS. 1hls was expecLed, slnce CLS LreaLs
Lhe regressors as lf Lhey were conLlnuous varlables, whlch lnflaLes Lhe resulLlng esLlmaLes.


CCLLu CLS

8C8l1

8C8l1 Sv?

loC 0.340** 0.200** 0.198**
(73.93) (80.88) (13.39)
lnC 0.122** 0.039** 0.060**
(31.39) (31.36) (3.94)
LML -0.007 -0.006** -0.007
(1.66) (2.90) (1.38)
1CWn 0.002 0.011** 0.012
(0.43) (6.03) (1.19)
Luu 0.034** 0.003* 0.006
(7.68) (2.38) (0.72)
CPlLu -0.003 0.021** 0.020
(0.69) (6.30) (1.71)

Bumana.Nente - Issue 1u- }uly 2uu9

116

M.S1A1uS -0.047** -0.007** -0.008
(10.00) (3.10) (1.31)
8l81P -0.000 -0.001** -0.001
(0.43) (4.37) (0.79)
SLx 0.108** 0.036** 0.037**
(6.37) (6.32) (2.93)
8LL_0 -0.211** -0.049** -0.043
(10.44) (3.16) (1.34)
8LL_1 -0.004** 0.012** 0.011**
(3.42) (28.28) (3.43)
CL 0.066** 0.041** 0.040**
(13.36) (19.39) (6.20)
CCnS 4.290**
(3.41)
Cbs 39898 39898 39898

# counLry 62
8obusL z
* slgn.3, ** slgn. 1

8asellne regresslon


1able 2 dlsplays resulLs from model (4). Addlng counLry flxed effecLs does noL alLer Lhe
esLlmaLes LhaL l found ln Lhe basellne model. 1he relaLlon beLween loC and SaLlsfacLlon
remalns presenL, wlLh Lhe same poslLlve slgn and baslcally Lhe same magnlLude, wlLh Lhe
esLlmaLed coefflclenL varylng beLween 0.17 and 0.34. 1he coefflclenLs on Lhe counLry dummles
are noL all zero and some are slgnlflcanL aL 3, buL Lhe sLandard errors are qulLe large,
suggesLlng evldence of mulLlcolllnearlLy and warnlng of overemphaslzlng Lhe resulLs.


CLS CCLLu

8C8l1

8C8l1 Sv?

loC 0.338** 0.173** 0.174**
(18.22) (14.86) (13.06)
lnC 0.121** 0.038** 0.036**
(9.34) (9.72) (9.19)
LML -0.007 -0.000 0.000
(0.83) (0.03) (0.06)
1CWn 0.002 -0.001 0.000
(0.21) (0.24) (0.02)
Luu 0.034** 0.014** 0.013*
(3.79) (2.81) (2.61)
CPlLu -0.003 -0.002 -0.003
(0.33) (0.33) (0.37)
M.S1A1uS -0.048** -0.023** -0.026**
(6.02) (3.89) (6.03)
8l81P -0.000 -0.001 -0.001
(0.13) (1.21) (0.83)
Naigheiita Botteio - Boes Fieeuom of Choice cause Satisfaction.

117

SLx 0.107** 0.037** 0.038**
(3.89) (4.33) (4.64)
8LL_0 -0.213** -0.116** -0.113**
(4.32) (3.28) (3.16)
8LL_1 -0.003* -0.003* -0.002*
(2.19) (2.21) (2.23)
CL 0.066** 0.033** 0.033**
(3.83) (6.27) (3.82)
ConsLanL 4.082
(1.68)
Cbs 39898 39898 39898

# counLry 62
8-squared 0.168
L sLaL **slgn.1, *slgn. 3

CounLry llxed LffecLs


1he same conslderaLlons follow by looklng aL Lhe resulLs obLalned accounLlng for Llme effecLs
raLher Lhan counLry effecLs, as can be seen from 1able 3. Agaln, Lhe assoclaLlon beLween loC
and SaLlsfacLlon remalns poslLlve and slgnlflcanL, beLween 0.19 and 0.4. uummles for wave 1
and 3 are auLomaLlcally dropped for colllnearlLy reasons, and Lhe resulLs show LhaL only Lhe
second wave has had a slgnlflcanL poslLlve lmpacL on SaLlsfacLlon. 1hls may be explalned by Lhe
facL LhaL Lhe second wave ls Lhe one carrled ouL rlghL afLer Lhe fall of Lhe 8erlln wall. Powever,
sLandard errors are large, so Lhe lnLerpreLaLlon ls noL clear.


CCLLu CLS

8C8l1

8C8l1_Sv?

loC 0.419** 0.200** 0.198**
(92.08) (13.31) (13.71)
lnC 0.142** 0.061** 0.062**
(36.19) (6.37) (6.38)
LML -0.018** -0.003 -0.006
(4.06) (1.13) (1.30)
1CWn 0.022** 0.009 0.010
(3.84) (0.92) (1.03)
Luu 0.012** 0.004 0.003
(2.72) (0.48) (0.61)
CPlLu 0.034** 0.018 0.017
(3.11) (1.48) (1.43)
M.S1A1uS -0.009 -0.003 -0.003
(1.76) (0.84) (0.93)
8l81P -0.001* -0.001 -0.001
(2.07) (0.63) (0.62)
SLx 0.119** 0.038** 0.060**
(6.34) (2.79) (3.04)
8LL_0 -0.064** -0.043 -0.038
(3.10) (1.46) (1.27)

Bumana.Nente - Issue 1u- }uly 2uu9

118

8LL_1 0.028** 0.013** 0.013**
(31.38) (4.63) (4.33)
CL 0.086** 0.042** 0.041**
(19.39) (6.31) (6.00)
ConsLanL 3.449**
(2.74)
Cbs 39898 39898 39898
number of
wave 3

8-squared 0.237
8ob L sLaL * slgn 3, ** slgn 1
1lme llxed LffecLs


llnally, model speclflcaLlon (6) does noL glve rlse Lo slgnlflcanL alLeraLlons of Lhe esLlmaLlon
resulLs found ln Lhe basellne model. Pere Lhe marglnal effecL of loC ranges beLween 0.17 and
0.33.


CCLLu CLS

8C8l1

8 Sv? (W)

8 Sv? (C)

loC 0.336** 0.174** 0.173** 0.173**
(18.07) (14.76) (14.96) (14.96)
lnC 0.127** 0.060** 0.039** 0.039**
(9.66) (9.98) (9.42) (9.42)
LML -0.004 0.001 0.001 0.001
(0.60) (0.24) (0.32) (0.32)
1CWn -0.006 -0.004 -0.003 -0.003
(0.82) (1.32) (0.93) (0.93)
Luu 0.031** 0.012** 0.012* 0.012*
(3.39) (2.63) (2.33) (2.33)
CPlLu -0.003 -0.002 -0.003 -0.003
(0.34) (0.32) (0.60) (0.60)
M.S1A1uS -0.048** -0.023** -0.026** -0.026**
(6.00) (3.84) (6.04) (6.04)
8l81P 0.000 -0.001 -0.000 -0.000
(0.13) (1.03) (0.73) (0.73)
SLx 0.100** 0.034** 0.036** 0.036**
(3.33) (3.97) (4.33) (4.33)
8LL_0 -0.217** -0.117** -0.114** -0.114**
(3.12) (6.04) (3.88) (3.88)
8LL_1 -0.004 -0.002* -0.002* -0.002*
(1.92) (2.00) (2.04) (2.04)
CL 0.063** 0.034** 0.033** 0.033**
(3.83) (6.22) (3.73) (3.73)
Cons 3.304
(1.32)
Cbs 39898 39898 39898 39898
Naigheiita Botteio - Boes Fieeuom of Choice cause Satisfaction.

119


# counLry 62
8-squared 0.174
8obusL L * slgn aL 3, ** slgn aL 1
1ab|e 4. CompleLe


WlLh regard Lo Lhe oLher covarlaLes, Lhe esLlmaLed parLlal effecLs are ln llne wlLh Lhe resL of Lhe
llLeraLure. lncome ls always slgnlflcanL and poslLlvely relaLed Lo saLlsfacLlon. LmploymenL
sLaLus dlsplays a poslLlve correlaLlon as well
11
. 1he coefflclenLs on educaLlon, marlLal sLaLus,
rellglon and sex are slgnlflcanL and poslLlve across all speclflcaLlons, as one would have
expecLed. l do noL flnd any effecL for Lhe number of chlldren buL Lhls covarlaLe ls generally
consldered conLroverslal. l cannoL flnd a slgnlflcanL correlaLlon for age, buL Lhls may noL be
surprlslng glven LhaL l lmpose a llnear speclflcaLlon. Surprlslngly, pollLlcal orlenLaLlon ls hlghly
slgnlflcanL across all speclflcaLlons, wlLh a poslLlve correlaLlon beLween belng rlghL-wlnged and
belng saLlsfled wlLh one's llfe.

DISCUSSICN

1hls secLlon dlscusses some crlLlcal lssues LhaL may have arlsen wlLh Lhe emplrlcal
meLhodology, Lhe daLa and Lhe lnLerpreLaLlon of Lhe resulLs.
1he meLhodology LhaL l am uslng bolls down Lo lnvesLlgaLlng Lhe effecL of loC on
SaLlsfacLlon conLrolllng for how much heLerogenelLy as posslble. 1hls lmplnges on Lhe lmpllclL
assumpLlon LhaL saLlsfacLlon levels are lnLerpersonally comparable
12
. Whlle Lhls may be ln
prlnclple quesLlonable, Lhe assumpLlon ls commonly accepLed ln Lhe economlcs of happlness
llLeraLure (8[ornskov eL alll, 2008), and even has some psycho-neurologlcal foundaLlons.
WlLh regard Lo Lhe daLa, lL ls lmporLanL Lo warranL LhaL Lhere are no sysLemaLlcally mlsslng
observaLlons, arlslng from Lhe respondenLs' lnablllLy or unwllllngness Lo answer some
quesLlonnalre lLems. lf mlsslng observaLlons follow a paLLern relaLed Lo some unobservable
varlable, Lhen Lhe regresslon esLlmaLes wlll be blased and lnconslsLenL. Powever, by looklng aL
Lhe daLaseL Lhe problem does noL seem a pervaslve one: less Lhan 3,000 respondenLs dld noL
reporL an answer Lo Lhe quesLlon on saLlsfacLlon, over more Lhan 230,000 lnLervlewed.
1he cholce of flxed effecLs as a modelllng Lechnlque ls raLher naLural Lo deal wlLh parL of
Lhe heLerogenelLy problem. Powever, lL can be Lhe case LhaL Lhe flxed effecL dummles wlll Lake
up parL of Lhe effecL on saLlsfacLlon LhaL lnsLead ls Lo be aLLrlbuLed Lo oLher covarlaLes, mosL
lmporLanLly Lo loC. unforLunaLely, Lhere ls no way Lo solve Lhls lssue, whlch loses relevance
conslderlng LhaL Lhe esLlmaLed coefflclenL for loC ls always slzeable and slgnlflcanL.
lor flxed effecLs models, Lhe conslsLency of Lhe resulLs depends cruclally on Lhe exogenelLy
of Lhe covarlaLes wlLh respecL Lo Lhe error Lerm, alLhough allowlng poslLlve correlaLlon
beLween Lhe explanaLory varlables and unobservables. As menLloned ln SecLlon 4, assumpLlon
(2) ls qulLe llkely Lo hold lf we look conLemporaneously aL Lhe covarlaLes and Lhe error Lerm,
belng Lhls random ln a glven Llme perlod. Zero correlaLlon beLween lagged errors and
covarlaLes may be Lrlckler Lo defend: some respondenLs may sysLemaLlcally reporL lower loC ln
L ln face of a negaLlve shock ln L-1. Powever, sLrlcL exogenelLy requlres zero correlaLlon once
Lhe c
l
s have been accounLed for: conLrolllng for Lhe unobservables ls llkely Lo wash ouL Lhls
klnd of feedbacks and make assumpLlon (2) llkely Lo be meL.

11
LmploymenL ls coded such LhaL hlgher values correspond Lo lncreaslng levels of unemploymenL.
12
noLe LhaL Lhls does noL lmply any assumpLlon abouL cardlnallLy of Lhe caLegorles.

Bumana.Nente - Issue 1u- }uly 2uu9

120

Conslderlng Lhe acLual esLlmaLes, Lhere are few general commenLs Lo be made. llrsL, lL may
be dlfflculL Lo exLend Lhe resulLs Lo samples wlLh dlfferenL characLerlsLlcs. 1he flxed effecLs
conLrol for unobservables buL Lhese remaln a sorL of black box. Also, Lhe effecLs of Lhe long
Lerm levels of Lhe observed varlables wlll no longer be capLured ln Lhe resulLs, whlch wlll raLher
provlde an esLlmaLe of Lhe effecLs of varlaLlons ln Lhe observed varlables.
Looklng aL Lhe ouLpuL, Lhe sLandard errors more Lhan double ln Lhe speclflcaLlons lncludlng
Lhe counLry dummles, (4) and (6). 1hls evldence ls suggesLlve LhaL Lhere ls mulLlcolllnearlLy
among Lhe counLry dummles. 1o conLrol for Lhls posslblllLy, l compuLe Lhe varlance lnflaLlon
lacLor (vll) of Lhe varlables enLerlng speclflcaLlon (2) and (4)
13
. 1he vll shows for each varlable
how much lLs varlance ls lnflaLed by mulLlcolllnearlLy: as long as Lhe vll ls less Lhan 10 Lhere ls
no need Lo worry abouL Lhe growlng sLandard errors. AlLhough lL ls posslble Lo compuLe Lhe
vll only for Lhe CLS esLlmaLlons, Lhe resulLs clearly show LhaL when addlng Lhe counLry
dummles Lhe average vll ralses sharply from 1.13 Lo more Lhan 7. SLandard errors, however,
Lake overall reasonable values, Lhus Lhe paLLern dlscovered above, alLhough calllng for cauLlon,
does noL weaken Lhe maln resulLs.
llnally, lL ls lmporLanL Lo polnL ouL LhaL whlle Lhls analysls ls suggesLlve of a causal relaLlon
runnlng from loC Lo SaLlsfacLlon, lL ls by no means evldence for lL. 1he emplrlcal lnLulLlon ln
favour of causallLy comes from Lhe robusLness of Lhe esLlmaLes Lhrough Lhe dlfferenL
speclflcaLlon (3)-(6). 1he prellmlnary evldence on a poslLlve and slgnlflcanL effecL of loC on
SaLlsfacLlon comes from Lhe basellne regresslon, whlch does noL conLrol for any unobservable
effecL. Addlng Lhe flxed effecLs does noL affecL Lhe esLlmaLes of Lhe coefflclenL, whlch keep
havlng Lhe same slgn and slgnlflcance. 1o furLher LesL Lhe robusLness of Lhls resulL, for each
model speclflcaLlon l run a pooled CLS esLlmaLlon, a mulLlnomlal problL and a mulLlnomlal
problL LhaL accounLs for Lhe survey-naLure of daLa. Whlle Lhe resulLs agaln remaln unchanged,
Lhe causallLy lnLerpreLaLlon agaln remalns only a posslblllLy. 8everse causallLy from SaLlsfacLlon
Lo loC ls an ob[ecLlon dlfflculL Lo counLer aL Lhls sLage, and l cannoL rule ouL Lhe hypoLhesls of
an exLernal, unaccounLed varlable LhaL drlves boLh dependenL and lndependenL varlables. ?eL,
lf one accepLs Lhese caveaLs, Lhe resulLs supporL Lhe noLlon LhaL loC maLLers for well belng,
suggesLlng LhaL furLher research ln Lhls area may be worLh pursulng.

CCNCLUSICNS

1hls paper assesses Lhe lmporLance of Lhe relaLlonshlp beLween freedom of cholce and
lndlvldual saLlsfacLlon, as a measure of well belng. l lnvesLlgaLe Lhe hypoLhesls LhaL lreedom of
Cholce has a poslLlve lmpacL on saLlsfacLlon, as lL ls lmplled by LheoreLlcal sLudles ln Lhe Soclal
Cholce llLeraLure. 1he resulLs show LhaL freedom lndeed correlaLes poslLlvely wlLh saLlsfacLlon,
alLhough glven Lhe few caveaLs dlscussed above. WhaL Lhe resulLs do noL show ls Lhe dlrecLlon
of causallLy ln Lhe relaLlonshlp under sLudy, whlch remalns an open lssue for fuLure research.


8I8LICGkAn

8aveLLa, S. and v. eraglne (2006) Measurlng auLonomy freedom. 5oclol cbolce ooJ
welfote, 26:31-43.

13
vll ls Lhe reclprocal of Lhe Lolerance of a varlable, whlch ls Lhe proporLlon of Lhe varlance for Lhe
varlable ln quesLlon LhaL ls noL due Lo oLher explanaLory varlables: a Lolerance close Lo 0 lndlcaLes rlsk of
mulLlcolllnearlLy.
Naigheiita Botteio - Boes Fieeuom of Choice cause Satisfaction.

121

8aveLLa, S. and l. Cuala (2003) AuLonomy freedom and dellberaLlon. Iootool of 1beotetlcol
lolltlcs, 13:423443.
8aveLLa, S., M. 8oLLero, u. Malmone Ansaldo aLLl and . navarra (2008) 8ettet tboo ltee.
wby Aotooomy lovots lteeJom, 5occess ooJ 5elf-steem, Mlmeo.
8[rnskov, C., A. ureher and !. A.v. llscher (2006) Cross-CounLry ueLermlnanLs of Llfe
SaLlsfacLlon: Lxplorlng ulfferenL ueLermlnanLs across Croups ln SocleLy, wotkloq
popets 06-145, kCl Swlss Lconomlc lnsLlLuLe, L1P Zurlch.
Clark, A. and C. Lelkes (2003) uellver us from evll: rellglon as lnsurance. lk wotkloq lopet
06/0J, Luropean CenLer for Soclal Welfare ollcy and 8esearch.
lrey, 8. and A. SLuLzer (2002) nopploess ooJ cooomlcs. now tbe cooomy ooJ lostltotloos
A ect well-beloq, rlnceLon unlverslLy ress, rlnceLon.
lrey, 8runo and Alols SLuLzer (2006) MlspredlcLlng uLlllLy and Lhe ollLlcal rocess, ln
Ldward !. McCaffery and !oel Slemrod (eds) 8ebovlotol lobllc lloooce. new ?ork:
8ussell Sage loundaLlon, pp. 113-140.
Mlll, !.S. (1839) Oo llbetty, !ohn W. arker and Son, Cxford unlverslLy ress, 1991.
Cswald, A.!. (1997) Papplness and Lconomlc erformance, 1be cooomlc Iootool 107, 1813-
1831.
aLLanalk, . and ?. xu (1990) Cn 8anklng CpporLunlLy SeLs ln 1erms of lreedom of Cholce.
kecbetcbes cooomlpoes Je loovolo, 44:173-98.
Schyns, . (1998) CrossnaLlonal dlfferences ln happlness: economlc and culLural facLors
explored. 5oclol loJlcotot keseotcb 43, 3-26.
Sen, A. k. (1988) lreedom of cholce: ConcepL and conLenL. otopeoo cooomlcs kevlew
32:269-294.
Sen, A. k. (1991) Welfare, reference and lreedom. Iootool of cooomettlcs 30:13-29.
Sen, A.k. (1999) uevelopmeot os lteeJom, Cxford: Cxford unlverslLy ress.
Sugden, 8. (1998) 1he meLrlc of opporLunlLy. coo lbllos 14:307-337.
Sugden, 8. (2004) 1he CpporLunlLy CrlLerlon: Consumer SoverelgnLy WlLhouL Lhe
AssumpLlon of CoherenL references, Ak 94, 4: 1014-1033.
Wooldrldge, !.M. (2001) cooomettlc ooolysls of ctoss sectloo ooJ pooel Joto, Ml1 ress,
Cambrldge MassachusseL lnLroducLlon.







Bumana.Nente - Issue 1u- }uly 2uu9

123

Understanding Marx's Normative Critique of
Capitalism: Reification revisited
+

volkoo tJom
*

volkancldam[hoLmall.com

MarxlsL LradlLlons of LhoughL barely pay any aLLenLlon Lo Lhe normaLlve dlmenslon of Marx's crlLlque
of caplLallsm ln copltol. ln Lhls respecL Lukcs' semlnal essay on relflcaLlon ls an excepLlon. SlLuaLlng
Marx wlLhln Cerman ldeallsm, Lukcs deflnes Lhe fundamenLal normaLlve concern of Marx's wrlLlngs
as an lnvesLlgaLlon of Lhe condlLlons of radlcal freedom. Clven lLs cenLral role ln Marx's crlLlque, lL ls
necessary Lo flnd an answer Lo Lhe quesLlon of how Lo concepLuallze freedom. Lukcs flnds LhaL
answer ln a relnLerpreLaLlon of Lhe lnLroducLory chapLers of copltol, whlch seemlngly explores only Lhe
economlc loglc of caplLallsm. AlLhough Lhls ls a frulLful LheoreLlcal move, l argue LhaL Lukcs'
concepLuallzaLlon of Marx's undersLandlng of freedom, lmpllclL ln hls concepL of relflcaLlon, ls
problemaLlc. SLlll, Lukcs' lnLerpreLlve approach provldes concepLual resources Lo revlse Lhe concepL
of relflcaLlon wlLh a vlew Lo develop a beLLer undersLandlng of Marx's normaLlve crlLlque.


lL has always been conLroverslal Lo look for Marx's normaLlve crlLlque of caplLallsm ln hls
Magnum Cpus copltol. lor such a readlng challenges Lhe sclenLlflc-lmage of Marx, who has
analyzed Lhe laws of developmenL and dynamlcs of caplLallsm, and foreseen lLs necessary
demlse. 1hls lmage of a poslLlvlsL sclenLlsL
1
was flrmly esLabllshed by Lngels and hls boLh
soclal-democraLlc and LenlnlsL followers, who bullL Lhelr case on Marx's seemlngly
'ob[ecLlvlsL' language ln copltol. 1hls parLlcular readlng of Marx became Lhe domlnanL
dlscourse ln MarxlsL debaLes. 1he Lerm, 'orLhodox Marxlsm' has generally been used Lo
denoLe Lhls dlscourse, even Lhough lL ls dlfflculL Lo glve an exacL deflnlLlon of Lhe Lerm. 1hls ls
due Lo Lhe facL LhaL wlLhln Lhe LradlLlon of Marxlsm 'orLhodox Marxlsm' has ofLen been
uLlllzed as Lhe mosL convenlenL ldeologlcal weapon Lo dlscredlL Lhe opposlng fracLlon. 1hus
desplLe Lhelr radlcally dlfferenL lnLerpreLaLlons boLh Lukcs ln Cescblcbte ooJ
klosseobewotselo (London: 8ed SLar ress, 2000) [1923] and AlLhusser and 8allbar, aL leasL
lmpllclLly, ln Lhelr keoJloq copltol (London: verso, 1979) [1968] declare Lhelr own readlng of

+
l would llke Lo Lhank lgdem idam for her very helpful commenLs, suggesLlons, and correcLlons. My
work on Lhls paper owes a greaL deal Lo conversaLlons wlLh her. Also, l am graLeful Lo Marco Sollnas
for hls encouragemenL. llnally l would llke Lo Lhank uenlz Lrcan for smlllng aL Lhe world wlLh hls blg
brlghL eyes.
*
hd CandldaLe ln hllosophy, !ohann Wolfgang CoeLhe-unlverslLaL
1
WlLh 'poslLlvlsL lmage' l don'L refer Lo any parLlcular school of LhoughL wlLhln Marxlsm. Any MarxlsL
approach LhaL adopLs Lhe oplnlon LhaL soclal sclences are besL Lo be modeled afLer 'naLural sclences'
and search for an unanlmous meLhod ls however essenLlally poslLlvlsL. 1he classlcal poslLlvlsL approach
can be found ln lrledrlch Lngels, ulalecLlcs of naLure" ln Motx/oqels collecteJ wotks, (new ?ork:
lnLernaLlonal ubllshers, 2003), volume 23, p. 313-390 and Soclallsm uLoplan and SclenLlflc" ln
Motx/oqels 5electeJ wotks, (Moscow: rogress ubllshers, 1970), volume 3, p. 93-131. MosL
promlnenL Among Lhose who read (accuse) Marx as a poslLlvlsL are C.A Cohen, kotl Motxs 1beoty of
nlstoty. A uefeoce. (new !ersey: rlnceLon unlv. ress, 1978), and AlbrechL Wellmer, ctltlcol 1beoty of
5oclety, (new ?ork: 1he ConLlnuum ubl. Corp, 1971). lor a crlLlcal overvlew of poslLlvlsm-debaLe and
a successful refuLaLlon of Lhls clalm see, !ames larr, Marx and poslLlvlsm" ln 8all / larr (ed), Aftet
Motx, p. 217-233 (Cambrldge, Cambrldge unlverslLy ress, 1984).

Bumana.Nente - Issue 1u- }uly 2uu9

124

Marx as orLhodox. ln Lhls paper, l call boLh soclallsL (and evenLually soclal-democraLlc)
movemenL of Lhe Second-lnLernaLlonal and Lenlnlsm as orLhodox poslLlons. 8oLh soclallsLs of
Lhe Second lnLernaLlonal and LenlnlsLs have clalmed auLhorlLy on Lhe ttoe lnLerpreLaLlon of
Marx's work and even Lhough Lhelr vlew on how Lo overcome caplLallsm dlffers conslderably,
Lhey share some baslc lnLerpreLaLlve assumpLlons LhaL draw on Lngels' wrlLlngs raLher Lhan
LhaL of Marx.
1he problem wlLh orLhodox Marxlsm ls Lwofold. Cn Lhe one hand, because of lLs
mlsgulded undersLandlng of sclence LhaL condemns any normaLlve [udgmenL as ldeologlcal,
orLhodox Marxlsm Lends Lo cover up Lhe normaLlve foundaLlons upon whlch lLs own pro[ecL
of emanclpaLlon ls based - as a resulL, lLs lnLerpreLaLlve approach Lo copltol lacks any
normaLlve lnLeresL. Cn Lhe oLher hand, slnce Lhe LoLal absence of normaLlve argumenLs ls
lmposslble wlLhln a ctltlcol - leL alone a tevolotloooty - Lheory of socleLy, orLhodox Marxlsm
flnds lL necessary Lo lmporL a crlLlcal normaLlve perspecLlve Lo lLs analysls. 1hls perspecLlve ls
elLher borrowed from Lhe earller wrlLlngs of 'humanlsL' Marx or, aL lLs worsL, ls drawn ouL
from a common-sense moral lnLulLlon, whlch has been crlLlclzed by Marx ln all hls wrlLlngs.
1hus, for lnsLance, Lhe ldeal-Lyplcal orLhodox-MarxlsL leglLlmaLes hls/her pollLlcal agenda
wlLh Lhe followlng rheLorlcal quesLlon: ls lL noL evldenL LhaL Lhe caplLallsL form of explolLaLlon
ls 'un[usL' and 'lnhuman'? Marx's own answer Lo Lhls quesLlon ln copltol however would noL
exacLly serve Lo orLhodox MarxlsL's lnLeresLs. lor, accordlng Lo Marx, caplLallsL form of
explolLaLlon ls boLh jost
2
and, as a hlsLorlcal phenomenon, all Loo human.
1he sLrucLurallsL readlng of AlLhusser can be besL undersLood as a crlLlc of Lhls double
sLraln ln orLhodox Marxlsm, sclenLlsm, on Lhe one hand, humanlsm or common sense
morallLy on Lhe oLher. And yeL, AlLhusser's sLraLegy of seLLlng a radlcal break beLween Lhe
'humanlsL' early wrlLlngs and Lhe sclenLlflc laLer wrlLlngs of Marx, perhaps lnadverLenLly,
sLrengLhened Lhe dlscourse of orLhodox Marxlsm. 1hls ls Lhe case desplLe Lhe facL LhaL
AlLhusser's undersLandlng of sclence slgnlflcanLly dlffers from LhaL of orLhodox Marxlsm.
3

ln many ways, lL ls posslble Lo suggesL LhaL an earller aLLempL Lo challenge Lhe ascendancy
of an amoral, sclenLlflc" readlng of copltol had proved Lo be more successful Lhan
AlLhusser's approach because, unllke AlLhusser's, Lhls approach dealL wlLh Lhe problem of
orLhodox Marxlsm dlrecLly. Accordlng Lo Lhls vlew, Lhe normaLlve foundaLlons of Marx's crlLlc
of caplLallsm had Lo be uncovered and Lhls could only be done, lf we read copltol wlLh

2
See Lhe ChapLer on Worklng uay, where he wrlLes on Lhe anLlnomy of 'rlghLs' ln volume l (karl
Marx, uos kopltol ln Marx/Lngels Werke, (8erlln: uleLz verlag) volume 23, p. 209) and hls deflnlLlon of
[usLlce ln volume lll (karl Marx, uos kopltol ln Marx/Lngels Werke (hereafLer MLW) (8erlln: uleLz
verlag) volume 23, p. 331). ln order Lo explaln an oxymoron such as a ,[usL explolLaLlon' LhaL occuples
a cenLral place wlLhln Marx' crlLlc of caplLallsm, lL ls necessary Lo unearLh Lhe normaLlve foundaLlons of
Lhe crlLlque. lor a plauslble accounL LhaL argues agalnsL Lhls necesslLy see, Allen Wood, kotl Motx,
(Cxon: 8ouLledge, 1981). Accordlng Lo Wood lL ls posslble Lo dlfferenLlaLe beLween 'moral' and
'amoral' goods. Pe suggesLs LhaL Lhe Lerm 'explolLaLlon' refers Lo a sLaLe LhaL ls deLrlmenLal Lo a
naLural, amoral good, Lo well-belng. 1hus he concludes, [usL explolLaLlon ls noL an oxymoron aL all. lL
seems however LhaL Wood here presenLs a very ArlsLoLellan argumenL, ln order Lo deflne ,amoral'
goods and Lhus polnLs Lo a normaLlve foundaLlon wlLhouL namlng lL.
3
See AlLhusser, keoJloq copltol. Accordlng Lo AlLhusser, Marx ls Lhe lnlLlaLor of a oew sclence and
meLhodology and noL a poslLlvlsL sclenLlsL who analyses Lhe eLernal laws of socleLy. A brlef look aL Lhe
neo-AlLhusserlan LexLs would show, however, Lhe slmllar normaLlve concluslons drawn by Lhls llne of
LhoughL and orLhodox Marxlsm. See !an Poff (among oLhers ed.), uos kopltol oeo leseo, (MunsLer:
WesLfallsches uampfbooL, 2006).
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

123

dlfferenL phllosophlcal lenses Lhan Lhose of Lngels.
4
ln order Lo subsLanLlaLe Lhls approach
Lukcs developed hls concepL of relflcaLlon
3
LhaL prlmarlly resLs on Lhe lnLerpreLaLlon of
Marx's crlLlque of Lhe caplLallsL mode of soclal medlaLlon
6
presenLed ln Lhe openlng chapLers
of copltol. Lukcs convlnclngly argued LhaL any aLLempL Lo undersLand Marx's crlLlque of
caplLallsm ln copltol, whlch dlsregards hls phllosophlcal sLance wlLhln Lhe LradlLlon of
Cerman ldeallsm, ls lnadequaLe. 8ead wlLhln Lhe conLexL of MarxlsL debaLes, Lukcs' analysls
of relflcaLlon sLands ouL noL only by lLs phllosophlcal rlgor and cogency, buL also by lLs unlque
aLLempL Lo search for a normaLlve sLandard wlLhln copltol wlLh a vlew Lo hlghllghL Marx's
normaLlve crlLlque of caplLallsm.
7

ln Lhls paper, l argue LhaL alLhough Lukcs was prlnclpally rlghL ln hls lnLerpreLlve
approach Lo copltol, he was unsuccessful ln capLurlng Lhe essence of Marx's normaLlve
crlLlque of caplLallsm. lL ls due Lo Lhls fallure LhaL hls accounL of revoluLlonary pollLlcs, whlch
ls based on Lhe analysls of relflcaLlon, succumbs Lo a pollLlcs of macro-sub[ecL - a form of
pollLlcs LhaL has been Lhe ob[ecL of leglLlmaLe concern and has ralsed doubLs abouL Lhe
crlLlcal poLenLlal of Lhe concepL of relflcaLlon.
8
AlLhough Lukcs' parLlcular accounL of Marx's
relaLlon Lo Cerman ldeallsm ls responslble for Lhls ouLcome, l suggesL LhaL a cerLaln Pegellan
readlng of copltol, whlch Lukcs occaslonally seems Lo pursue, can noL only help us Lo save
Lhe revoluLlonary core of Lukcs' analysls of relflcaLlon buL also conLrlbuLe Lo a beLLer
undersLandlng of Marx's normaLlve crlLlque of caplLallsm. 1hus, ln whaL follows, l propose a
rereadlng of Marx's openlng argumenLs ln copltol Lo reLhlnk relflcaLlon and Lo brlng Lo llghL
lLs crlLlcal poLenLlal.

4
non-MarxlsL wrlLers dlscovered very early LhaL Marx' analyses of pollLlcal economy has a normaLlve
crlLlcal core Lo lL. Ceorg Slmmel's analysls ln lbllosopble Jes CelJes (Ceorg Slmmel, lbllosopby Jes
CelJes (lrankfurL am Maln: Suhrkamp, 1989) [1900]) can be lnLerpreLed as a polemlc agalnsL Marx
(See below). Slmllarly, lL would noL be far-feLched Lo clalm LhaL Max Weber's renowned analysls of
roLesLanL LLhlcs Lakes lL lnsplraLlon dlrecLly from copltol. 1he flrsL analysls of copltol, almlng Lo
uncover Lhe normaLlve foundaLlons of Lhe Work belongs Lo lranz eLry, uet 5ozlole Cebolt Jet
Motxcbeo wetttbeotle (!ena: C. llscher, 1916).
3
Marx colns Lhe Lerm 'relflcaLlon', vetJloqllcbooq, ln volume lll of copltol ln reference Lo hls analysls
of Lhe ,feLlsh charakLer of commodlLy'. ln Lhls conLexL Lhe Lerm llLerally denoLes Lhe converslon of
soclal relaLlons Lo a relaLlon beLween Lhlngs. More accuraLely soclal relaLlons acqulre a relfled, tbloqly,
quallLy, whlle aL Lhe same Llme Lhe relaLlons beLween Lhlngs, commodlLles, Lake Lhe form of a soclal
relaLlon. As Marx calls lL: lL ls an enchanLed, perverLed (verkehrLe), Lopsy-Lurvy world, ln whlch
Monsleur le CaplLal and Madame la 1erre do Lhelr ghosL-walklng as soclal characLers and aL Lhe same
Llme dlrecLly as mere Lhlngs. (MLW 23, p. 838). As l wlll demonsLraLe below Lukcs Lakes Lhls Lerm
and develops hls Lheory of relflcaLlon ln hls semlnal essay 8elflcaLlon and Lhe Conscloussnes of Lhe
roleLarlaL (Ceorg Lukcs, ule verdlngllchung und das 8ewussLseln des roleLarlaLs ln Cescblcbte
ooJ klosseobewosstselo).
6
8y 'soclal medlaLlon' l refer Lo a form of soclal relaLlon LhaL plays a consLlLuLlve role for sub[ecL
poslLlons. A consLlLuLed sub[ecL poslLlon denoLes a parLlcular aLLlLude Lowards oLher sub[ecLs and Lhe
ouLslde world. 1hls aLLlLude ln reLurn depends upon a parLlcular concepLuallzaLlon of freedom.
7
Cnly recenLly lnnovaLlve aLLempLs have been made Lo unearLh Lhe foundaLlons of Marx' normaLlve
crlLlc of caplLallsm ln copltol LhaL go beyond Lhe analysls of Lukcs. See, Ceorg Lohmann, loJlffeteoz
ooJ Cesellscboft (lrankfurL am Maln: Suhrkamp, 1991) and Molshe osLone, 1lme loboot ooJ 5oclol
uomlootloo (Cambrldge, Camobrldge unlverslLy ress, 1993).
8
Pabermas ls among Lhe mosL emlnenL Lhlnkers LhaL polnLed Lo Lhls problem. See Pabermas, uet
pbllosopblscbe ulskots Jet MoJetoe, (lrankfurL am Maln: Suhrkamp, 1988), p. 93-103 and Pabermas,
1beotle Jes kommoolkotlveo nooJelos 8J.2, (lrankfurL am Maln: Suhrkamp, 1981), ChapLers 6 and 8.

Bumana.Nente - Issue 1u- }uly 2uu9

126

1hls readlng draws parallels beLween Lhe unfoldlng of Marx's crlLlque and Pegel's
dlscusslon of dlalecLlcally unfoldlng knowledge models ln hls lbeoomeooloqy of 5pltlt.
9
My
argumenL ls LhaL when read Lhrough Lhe lenses of lbeoomeooloqy, Lhe openlng chapLers of
copltol do provlde a crlLerlon for Lhe normaLlve crlLlque of caplLallsL socleLy. 1hls crlLerlon, l
clalm, ls Lo be found ln Lhe dlalecLlc of recognlLlon: for Pegel, as well as for Marx, Lhe
dlalecLlc of recognlLlon serves as a model for Lhe lnLersub[ecLlve medlaLlon LhaL ls necessary
for Lhe subslsLence of a Lrue, l.e. free sub[ecLlvlLy. 1hus, llke Lukcs', my readlng of copltol
aLLrlbuLes nelLher an amoral, poslLlvlsL undersLandlng of sclenLlflc sLance Lo Marx's crlLlque of
caplLallsm, nor searches for exLernal normaLlve sLandards ln Lhe earller wrlLlngs of Marx LhaL
wlll supposedly make sense of Lhe ob[ecLlvlsL analysls of hls laLe work. l also share Lukcs'
vlew LhaL Marx's normaLlve crlLlque of caplLallsm ln copltol LargeLs prlmarlly Lhe unfreedom
speclflc Lo caplLallsL socleLy and LhaL ln Marx's accounL Lhe concepLlon of freedom plays a
fundamenLal role ln deflnlng oLher (moral and eLhlcal) values such as [usLlce, equallLy and a
concepLlon of good llfe.
10
l parL ways wlLh Lukcs, however, on Lhe lssue of how Lo
concepLuallze Lhe undersLandlng of freedom LhaL ls Lo serve as Lhe normaLlve sLandard
/crlLerlon of Marx's crlLlque of CaplLallsm. Accordlng Lo my accounL, Lo fully grasp Marx's
undersLandlng of freedom, lL ls essenLlal Lo reLhlnk Lhe concepL of relflcaLlon Lhrough Lhe
Pegellan dlalecLlcs of recognlLlon.
1o elaboraLe Lhese polnLs, ln Lhe flrsL parL of Lhls paper l Lurn Lo Lwo lnLerpreLaLlve
accounLs of Lukcs' concepL of relflcaLlon LhaL underllne Lukcs' lndebLedness Lo Pegellan
phllosophy. PerberL Marcuse and Axel PonneLh, boLh members of lrankfurL School, flnd lL
necessary Lo have recourse Lo Lhe concepL of relflcaLlon ln search for a normaLlve sLandard
LhaL enables a subsLanLlal crlLlque of modern caplLallsL socleLy on phllosophlcal grounds. My
own lnLerpreLaLlon of Luccs draws on Lhe lnslghLs provlded by Lhese Lhlnkers. ln Lhe second

9
1hls Pegellan readlng ls noL Lo be confused wlLh a form of Pegellanlsm parLlcularly prevalenL ln
Cermany LhaL reads copltol ln lLs relaLlon Lo Pegel's 5cleoce of loqlc, whlch Lakes lLs orlglns from whaL
can be called as 'value-lorm-Analysls School' of Pans-Ceorg 8achhaus, ulolektlk Jet wettfotm.
uotetsocbooqeo zot Motxcbeo Okooomlektltlk [1974] (lrelburg: a lra, 1997), PelmuL 8elchelL, 2ot
loqlscbeo 5ttoktot Jes kopltolbeqtlffs bel kotl Motx (lrankfurL am Maln: Lurop. verl.-AnsL., 1970) and
Mlchall Pelnrlch, ule wlsseoscboft vom wett. ule Motxcbe ktltlk Jet polltlscbeo Okooomle zwlscbeo
wlsseoscboftllcbet kevolotloo ooJ klosslscbet 1toJltloo. (MunsLer: WesLfallsches uampfbooL, 1991).
uesplLe very valuable conLrlbuLlons Lo Lhe quesLlon of laLe Marx' meLhodologlcal conslderaLlons ln
Lhelr relaLlon Lo Pegel's 5cleoce of loqlc, Lhe focus of Lhese works remaln Lo be a beLLer undersLandlng
of Lhe crlLlque of pollLlcal economy by Marx. ln oLher words Lhe underlylng assumpLlon ls
absoluLely ln llne wlLh orLhodox Marxlsm's (especlally wlLh Lenlnlsm's) clalm LhaL laLe Marx use of
dlalecLlcal meLhod of presenLaLlon ln copltol ls yeL anoLher proof of hls crlLlque of caplLallsm on purely
sclenLlflc / amoral grounds. lor recenL examples of Lhls approach see Leo Lelerko, uas 'wlrkllche
1rlebwerk des kaplLals' und selne 8ezlehung zu Pegels Loglk" ln Poff (ed.), uos kopltol oeo leseo and
vaz[ulln, vlkLor A., "LnLwlcklung sysLemaLlsch denken. Lln verglelch der dlalekLlschen Loglk bel Pegel
und Marx" ln: ueotscbe 2eltscbtlft fot lbllosopble, 33:2, S. 203-18, (2003).
10
A recenL debaLe beLween nancy lraser and Axel PonneLh can be Laken as Lhe proof of how
fundamenLally lmporLanL Lhe lssue aL sLake here ls for conLemporary emanclpaLory pollLlcs. See nancy
lraser / Axel PonneLh, umvettellooq oJet Aoetkeooooq? loe polltlscb-pbllosopblscbe koottovetse.
(lrankfurL am Maln: Suhrkamp, 2003) Whlle lraser opLs for a llberal/soclal democraLlc double-Lrack
pollLlcs of redlsLrlbuLlon ooJ recognlLlon ln order Lo malnLaln a more epool Lerms of pollLlcal
parLlclpaLlon, PonneLh ln hls more Marxlan approach calls for a reLhlnklng of Lhe concepLlon of
fteeJom ln modern caplLallsL socleLles, whlch would only Lhan make emanclpaLory pollLlcs posslble.
1hen as for Marx also for PonneLh Lhe ldeal dlalecLlc of recognlLlon presenLs a mode of soclal
medlaLlon LhaL can serve as an orlenLaLlon for conLemporary pollLlcal acLors.
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

127

parL of Lhe paper, l glve a brlef analysls of Lukcs' concepL of relflcaLlon presenLed ln hls
semlnal essay 8elflcaLlon and Lhe Consclousness of roleLarlaL". Pere l demonsLraLe LhaL
Lukcs' LexL carrles wlLhln lL Lhe concepLual resources for an alLernaLlve undersLandlng of
relflcaLlon. ln my concludlng remarks, drawlng on Lhese resources, l provlde a brlef accounL
of whaL a Pegellan readlng of Lhe lnLroducLory chapLers of copltol mlghL look llke. Such an
accounL ls cruclal because only a Pegellan readlng of Lhls klnd can boLh save Lhe
revoluLlonary core of Lukcs' analysls of relflcaLlon and conLrlbuLe Lo a beLLer undersLandlng
of Marx's normaLlve crlLlque of caplLallsm.

WrlLlng rlghL afLer Lhe 8olshevlk 8evoluLlon, Lukcs, ln hls famous essay 8elflcaLlon and Lhe
Consclousness of roleLarlaL", challenges Lhe phllosophlcal foundaLlons of Lhe economlc
deLermlnlsL Marxlsm of Lhe Second lnLernaLlonal. Whlle maklng phllosophlcal argumenLs Lo
lead revoluLlonary Marxlsm away from Lngels' sclenLlflc dlalecLlcal maLerlallsm, Lukcs also
Lrles oot Lo counLer Lhe laLLer slmply by puLLlng forward a naive anLhropologlcal-humanlsL
lnLerpreLaLlon of Marx's wrlLlngs. ln oLher words, he does noL wanL Lo offer a Lheory of
allenaLlon LhaL ls lnevlLably based on an essenLlallsL undersLandlng of human naLure. Lukcs
develops hls concepL of relflcaLlon malnly Lo address Lhese concerns.
As l wlll demonsLraLe below, Lukcs ends up falllng Lo accompllsh Lhls goal due Lo Lhe
phllosophlcal cholces he makes. Lukcs' fallure, however, ls producLlve because, even Lhough
hls answers Lo Lhem are lnadequaLe, he glves a cenLral sLaLus Lo Lwo lnLerdependenL and
cruclal quesLlons ln hls analysls: llrsL, whaL ls Lhe relaLlon of Marx's phllosophy Lo Cerman
ldeallsm ln general? And second, whaL ls Marx's poslLlon wlLh respecL Lo Lhe cenLral eLhlcal
pursulL of Lhls LradlLlon, namely Lhe lnvesLlgaLlon of Lhe condlLlons of exlsLence of radlcal
lndlvldual freedom? ln many ways, lL ls posslble Lo suggesL LhaL WesLern Marxlsm ln general,
and CrlLlcal 1heory ln parLlcular has emerged and developed ln response Lo Lhese quesLlons
formulaLed by Lukcs. 8ecause Lukcs' answer Lo Lhese quesLlons ls Lo be found ln hls
concepLlon of relflcaLlon, a new assessmenL of Lhls concepL has aLLalned a cenLral role ln Lhe
self-undersLandlng of MarxlsL LradlLlons of LhoughL.
11

lL ls cruclal Lo noLe LhaL Lhose who dlrecLly engage wlLh Lhe concepL of relflcaLlon, ln one
way or anoLher Lurn Lo Pegel's phllosophy. 1hus, for lnsLance, PerberL Marcuse ln hls
reconsLrucLlon of Lhe lbeoomeooloqy of 5pltlt uses Lhe concepL Lo characLerlze Lhe flrsL Lhree
secLlons of Pegel's work.
12
Accordlng Lo Marcuse's lnLerpreLaLlon, all LruLh clalms made by
Lhe succeedlng models of knowledge ln Lhese lnLroducLory secLlons fall on Lhe same grounds:
All Lhese models clalm Lo secure Lhelr presumable knowledge lndependenLly of any hlsLorlcal
deflnlLe soclal pracLlce.
13
1herefore, Marcuse argues, Pegel aLLempLs Lo show LhaL man can
know Lhe LruLh only lf he breaks Lhrough hls 'relfled' world. We borrow Lhe Lerm 'relflcaLlon'
from Lhe MarxlsL Lheory, where lL denoLes Lhe facL LhaL all relaLlons beLween men ln Lhe
world of caplLallsm appear as relaLlons beLween Lhlngs [.]. 1he commodlLy, for lnsLance,
embodles ln all lLs quallLles Lhe soclol relaLlons of labor [.] (lbld, p. 112). Pereby Marcuse
seLs Lhe concepL of relflcaLlon rlghLly ln relaLlon Lo Pegel's lbeoomeooloqy as well as Lo

11
See Lhe classlcal work of MarLln !ay, Motxlsm ooJ 1otollty. 1be AJveototes of o coocept ftom lokcs
to nobetmos (Cambrldge: ollLy ress, 1984).
12
PerberL Marcuse, keosoo ooJ kevolotloo, (1941) [Cxford: 8ouLledge & kegan aul, 1986].
13
See also Lhe more recenL sLudy of 1erry lnkard, who bullds on Lhls lnslghL of Marcuse. 1erry
lnkard, neqels lbeoomeooloqy. 1be 5oclollty of keosoo, (Cambrldge: Cambrldge unlverslLy ress
1994).

Bumana.Nente - Issue 1u- }uly 2uu9

128

Marx's analyses of commodlLy feLlshlsm. A few pages laLer, however, commenLlng on Pegel's
masLer and slave ulalecLlc, Marcuse equaLes Lhe poslLlon Laken by Marx ln hls 1844
ManuscrlpLs wlLh Lhe laLer crlLlque formulaLed ln Lhe flrsL chapLer of CaplLallsm, wlLh
commodlLy feLlshlsm or ln Lukcs' words wlLh relflcaLlon: ln 1844 Marx sharpened Lhe baslc
concepLs of hls own Lheory Lhrough a crlLlcal analysls of Pegel's lbeoomeooloqy of MloJ. Pe
descrlbed Lhe 'allenaLlon' of labor ln Lhe Lerms of Pegel's dlscusslon of masLer and servanL
[.] Marx makes reference Lo Pegel's deflnlLlve lnslghL, whlch dlsclosed Lo hlm LhaL lordshlp
and bondage resulL of necesslLy from cerLaln relaLlons of labor, whlch are ln Lurn,
relaLlonshlps ln a 'relfled' world (lbld. 113). 1hls conflaLlon of Lwo dlsLlncL concepLs, LhaL ls,
of 'allenaLlon' and 'relflcaLlon' by Marcuse polnLs Lo an lnner Lenslon ln Lukcs' own analyses.
When he wroLe hls essay, Lukcs dld noL yeL know abouL Lhe exlsLence of 1844
ManuscrlpLs of Marx. And yeL, as l argue below, ln hls accounL of revoluLlonary overcomlng
of caplLallsm, bulldlng up on 1besls oo leoetbocb and !ohann CoLLlleb llchLe's phllosophy,
Lukcs reached Lo slmllar concluslons wlLh young Marx's Lheory of allenaLlon. Accordlng Lo
Lukcs, Lhe 'proleLarlan sLandpolnL' ls, llke Lhe sLandpolnL of Lhe 'slave' ln Pegel's accounL
(presumably, for Pegel does noL argue exacLly ln Lhls llne), lncllned Lo freedom. lor, Lhrough
hls work - ln Lukcs case also pollLlcal work - slave bullds /educaLes hlmself Lo be Lhe ttoe
sub[ecL of hlsLory.
14

WlLh Lhls llne of argumenL Lukcs prlvlleges, as noLed by !urgen Pabermas,
13
a cerLaln
producLlvlsL mode of acLlon over oLhers, and ln dolng so, he comes very close Lo a poslLlon
occupled by 'humanlsL' crlLlclsms of caplLallsm, whlch aLLach a cenLral lmporLance Lo Lhe role
played by labor / producLlve acLlon ln human flourlshlng. PumanlsL crlLlcs of caplLallsm argue
LhaL modern (caplLallsL) dlvlslon of labor and work processes ls responslble for human belng's
olleootloo from (Lhelr own) naLure. 8ecause Lukcs ls aware of Lhe ahlsLorlcal and essenLlallsL
naLure of such a crlLlcal accounL, he Lrles Lo dlsavow Lhls poslLlon Lhrough lengLhy dlscusslons
ln hls essay. uesplLe all hls efforLs, however, as we wlll see, he cannoL avold Lo fall back lnLo
Lhls perspecLlve.
16
neverLheless, Lhls Lenslon ln hls essay does noL prevenL Lukcs from Lrylng
Lo esLabllsh hls own Lheory of telflcotloo ln opposlLlon Lo a Lheory of olleootloo.
8y polnLlng Lo Lhe dlfferenL accenL puL on Lhe normaLlve groundlng of Lhe Marxlan
crlLlque of caplLallsm ln a Lheory of olleootloo ln conLrasL Lo a Lheory of telflcotloo, a
dlfference, whlch, as we have seen, was dellberaLely evaded by Marcuse, l do noL wanL Lo
suggesL LhaL Lhere ls 'a radlcal break' beLween 'young Marx' and 'maLure Marx.' As wlLh
every Lhlnker, however, lL ls posslble Lo observe gradual shlfLs beLween dlfferenL works of a
glven auLhor.
17
Pere lL sufflces Lo noLe LhaL Lhe cenLral role aLLached Lo 'allenaLed labor' as
Lhe normaLlve ground for Lhe crlLlque of caplLallsm ln 1844 Mooosctlpts, leaves lLs place Lo a
crlLlque of a hlsLorlcally deLermlned form of a soclol meJlotloo ln copltol, lnformed by Lhe
Pegellan Lheory of lnLersub[ecLlvlLy. AlLhough l am crlLlcal of Marcuse's erasure of Lhe
dlfference beLween a Lheory of allenaLlon and a Lheory relflcaLlon ln hls reconsLrucLlon of
Marx's work ln relaLlon Lo lbeoomeooloqy of tbe 5pltlt, l share Lhe fundamenLal premlse of
hls lnLerpreLaLlon: 8elflcaLlon ls besL undersLood as a concepLual lnnovaLlon LhaL shows and

14
lor a deLalled accounL of Lukcs' usage of Pegellan MasLer and Slave ulalecLlc ln hls essay see
Andrew AraLo, Lukcs' 1heory of 8elflcaLlon" ln 1elos 11, Sprlng 1972, p. 38f.
13
Pabermas, lblJem.
16
lor Lhe dlscusslon of Pumanlsm See Lukcs, lblJ., p. 148f and 208f. Lspeclally Lhe explanaLlon of Lhe
prlvlleged sLaLus of Lhe 'sLandpolnL of roleLarlaL' (ascrlbed class-consclousness of proleLarlaL) ls by
conLrasL made from Lhe perspecLlve of a Lheory of allenaLlon. See 173f.
17
8y far Lhe besL accounL of Lhe poslLlons LhaL Marx Lakes ln hls earller wrlLlngs ls glven by uanlel
8rudney, Motx ottempt to leove lbllosopby, (Cambrldge: Parvard unlverslLy ress, 1988).
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

129

hlghllghLs Lhe lndebLedness of Marx's analysls of commodlLy feLlshlsm Lo Lhe openlng
secLlons of lbeoomeooloqy, speclflcally, l suggesL, Lo Lhe chapLer on 'lorce and
undersLandlng'.
8ecenLly, Axel PonneLh also revlslLed Lukcs' concepL ln hls work. unllke LradlLlonal
lnLerpreLaLlons, ln hls book enLlLled 8elflcaLlon",
18
PonneLh refuses Lo analyze Lhe concepL
ln Lerms of lLs relaLlon Lo Cerman ldeallsm, ln order noL Lo hazard Lhe consequences of Lhe
meLaphyslcal assumpLlons LhaL Lhen have Lo be made. 1hus, he successfully avolds falllng
back Lo a producLlvlsL model of sub[ecLlvlLy LhaL frames Lukcs' problemaLlc dlscusslon of Lhe
'ascrlbed class-consclousness' of proleLarlaL, whlch has burdened Lhe MarxlsL Lheory and
praxls ever slnce. As a resulL, lnsLead of con[urlng Lukcs' eschaLologlcal lmagery of Lhe
macro- sub[ecL and ob[ecL of hlsLory LhaL wlll overcome Lhe sLaLe of relflcaLlon once agaln,
PonneLh focuses on soclal phenomena LhaL helped Lukcs Lo characLerlze Lhe ob[ecL of hls
crlLlclsm from a soclologlcal and acLlon-LheoreLlcal perspecLlve. PonneLh argues LhaL even
sLrlpped from lLs ldeallsL presumpLlons, on a purely soclologlcal level, Lukcs' concepL goes
beyond Lhe currenL uLlllzaLlons of slmllar concepLs, such as nussbaum's 'ob[ecLlflcaLlon',
whlch elLher descrlbe cerLaln Lypes of human behavlor LhaL vlolaLe eLhlcal prlnclples or polnL
Lo a form of lnsLrumenLal raLlonallLy LhaL colonlzes Lhe emoLlonal llfe of Lhe acLor.
19
1he
radlcal core of Lukcs' argumenL, PonneLh suggesLs, resLs on Lhe observaLlon LhaL ln modern
socleLles acLors are compelled Lo Lake a conLemplaLlve aLLlLude Lowards Lhe reallLy, whlch
Lhey consLlLuLe, and ln whlch Lhey parLlclpaLe and lnLeracL. ln Lhls readlng, relflcaLlon refers
Lo a dlsLorLed human praxls, whlch ls Lo be conLrasLed wlLh an lnLersub[ecLlve aLLlLude on Lhe
parL of Lhe sub[ecL, and as such, can be relnLerpreLed as loslng slghL of a prlmordlal form of
relaLlon Lo Lhe world, as forgeLfulness of recognlLlon".
AlLhough l share PonneLh's vlew LhaL Lhe concepL of recognlLlon consLlLuLes Lhe
normaLlve horlzon of a non-relfled soclal world, ln whaL follows, l refuse Lo adopL hls
soclologlcal and acLlon-LheoreLlcal approach LhaL seLs, ln PonneLh's own words, a more
'exlsLenLlal mode' of recognlLlon as Lhe sLandard of crlLlque of a relfled world.
20
lnsLead, l
argue LhaL Lukcs' own phllosophlcal assumpLlons are adequaLe and sufflclenL Lo undersLand
Lhe concepL of relflcaLlon wlLhln Lhe framework of a Lheory of recognlLlon. lor Lhe
phllosophlcal assumpLlons ln Lukcs' analysls of commodlLy feLlshlsm provlde us wlLh Lhe
concepLual resources Lo develop an alLernaLlve readlng, whlch unllke Lukcs' own accounL,
seLs Lhe Pegellan dlalecLlc of recognlLlon as Lhe sLandard of Lhe crlLlque of caplLallsm. When
lnLerpreLed ln Lhls manner, Lhe concepL of relflcaLlon conLlnues Lo provlde a powerful
crlLlque of a mode of soclal medlaLlon speclflc to caplLallsL socleLy, as lnLended by Lukcs,
raLher Lhan referrlng Lo a dlsLorLed (allenaLed) human praxls.


ln order Lo uLlllze Lhese lnslghLs provlded by Marcuse and PonneLh Lo develop an alLernaLlve
lnLerpreLaLlon of Lukcs' concepL of relflcaLlon, lL ls flrsL necessary Lo glve an accounL of hls
renowned essay, 8elflcaLlon and Lhe Consclousness of roleLarlaL". LeL me begln LhaL
accounL wlLh a brlef dlscusslon on Lhe argumenLaLlve sLrucLure of Lhe LexL. urawlng on
Marx's meLhod ouLllned ln hls lnLroducLlon Lo (Crundrlsse)",
21
Lukcs dlvldes hls essay ln

18
Axel PonneLh, vetJloqllcbooq, (lrankfurL am Maln: Suhrkamp, 2003).
19
MarLha nussbaum, Cb[ecLlflcaLlon, ln Alan Soble (ed.) 1be lbllosopby of 5ex (Maryland: 8owman &
LlLLlefleld, 2002).
20
See Axel PonneLh, lblJ. p. 46.
21
See karl Marx, MLW 42, p. 19-43.

Bumana.Nente - Issue 1u- }uly 2uu9

130

Lhree ma[or parLs. Accordlng Lo Marx, a sclenLlflc analysls beglns wlLh a naive pre-concepLual
percepLlon of an emplrlcal whole (hls example ls 'populaLlon'). lrom Lhls percepLlon of a
chaoLlc gaLherlng of facLs one has Lo move analyLlcally Lo slmple absLracL concepLs. A
sclenLlflc Lheory of socleLy brlngs Lhese absLracL concepLs LogeLher so as Lo reconsLrucL a
concepLually lnformed concreLe LoLallLy. A consclous use of processes of Lhlnklng, absLracLlon
and reconsLrucLlon are essenLlal for Lhe sclenLlflc meLhod. lor, only lf sclenLlflc efforL ln Lhese
processes can be ralsed Lo consclousness, Lhe Lheory can become lmmune Lo ahlsLorlcal
generallzaLlons.

lollowlng Lhls llne of Lhlnklng, ln Lhe flrsL parL of hls essay, followlng Marx and drawlng on
Max Weber and Ceorg Slmmel, Lukcs provldes a soclologlcal analysls of everyday llfe wlLhln
caplLallsL socleLy Lo glve a plcLure of Lhe LoLallLy LhaL ls Lo be overcome Lhrough revoluLlon.
1hls soclologlcal analysls seLs Lhe ground for a concepLual, LhaL ls, pbllosopblcol dlscusslon on
Lhe foundaLlons of Lhe dlverse phenomena broughL under Lhe Lerm telflcotloo. Pence, ln Lhe
second parL of hls essay Lukcs' dlscusslon ls flrmly slLuaLed wlLhln Lhe LradlLlon of Cerman
ldeallsm. Pere Lukcs puLs forward hls radlcal clalm and lnLroduces Lhe cenLral dlscusslon of
Lhe sub[ecL-LheoreLlcal condlLlons of Lhe exlsLence of lndlvldual freedom: Modern crlLlcal
phllosophy sprlngs from Lhe relfled sLrucLure of consclousness. (Lukcs, lbld., p. 122) llnally,
ln Lhe Lhlrd and flnal parL of hls essay, followlng Lhe sclenLlflc meLhod ouLllned ln CtooJtlsse,
Lukcs draws dlrecL pollLlcal consequences from hls phllosophlcal analysls and aLLempLs Lo
consLrucL an accounL of a ttoe LoLallLy. 1hus, he argues LhaL Lhe relfled world of caplLallsm
can only be dlssolved and opptoptloteJ, lf Lhe proleLarlaL seLs lLself as Lhe sub[ecL of soclo-
hlsLorlcal reallLy lnsLead of slmply belng Lhe producL of caplLallsL soclal order. 1he momenL of
revoluLlon ls Lhe momenL, ln whlch 'Lhe relfled sLrucLure of consclousness' glves way Lo Self-
consclousness.
!usL llke lLs argumenLaLlve sLrucLure, Lhe subsLanLlve conLenL of 8elflcaLlon and Lhe
Consclousness of roleLarlaL" ls also lndebLed Lo Marx's Lhlnklng. Pence, Lukcs beglns hls
essay wlLh a brlef analysls of Marx's commodlLy feLlshlsm. Pe correcLly polnLs ouL LhaL Lhe
Lerm commodlLy for Marx ls noL only, or even prlmarlly, an economlc caLegory. lnsLead,
Lukcs suggesL, 'commodlLy' refers Lo a soclal medlaLlon speclflc and cenLral Lo caplLallsm.
Pe, Lhen, seLs ouL Lo elucldaLe Marx's somewhaL obscure crlLlclsm of Lhls soclal medlaLlon by
focuslng on Lhe dlscusslon on 'commodlLy feLlshlsm,' whlch even Loday presenLs Lhe key Lo
dlverse lnLerpreLaLlons of copltol.
lL ls of LheoreLlcal slgnlflcance Lo underllne LhaL one of Lhe ma[or goals of Lukcs' accounL
of commodlLy feLlshlsm ls Lo challenge Lhe orLhodox MarxlsL vlew, whlch lnLerpreLs
commodlLy feLlshlsm slmply as an lnverLed sLaLe of soclal relaLlons. Accordlng Lo Lhe
orLhodox MarxlsL vlew, soclal relaLlons beLween human belngs oppeot to be relaLlons
beLween Lhlngs. 1hus, lL ls argued, Lhls lnverLed sLaLe of affalrs, whlch conceals Lhe Lrue
relaLlonshlp beLween human belngs, has Lo be correcLed Lo glve way Lo a genulne and
consclous, a JecommoJlfleJ, relaLlonshlp beLween human belngs LhaL defy a soclal relaLlon
medlaLed by commodlLles. 1hls lnLerpreLaLlon, however, ls a slmpllflcaLlon of Marx's
argumenL and consequenLly mlsleadlng. lor, Marx's crlLlque based on Lhe feLlsh characLer of
commodlLy ls Lwofold. Cn Lhe one hand, Marx crlLlclzes Lhe pollLlcal economlsLs, who presenL
a hlsLorlcal and soclal relaLlonshlp as a naLural order, for falllng vlcLlm Lo Lhe ldeology
emanaLlng from Lhe commodlLy form. Cn Lhe oLher hand, Marx dlrecLs hls maln crlLlque
Lowards a form of soclal medlaLlon LhaL seLs Lhe sub[ecLs of Lhls relaLlonshlp as aLomlsLlc
lndlvlduals. 1he exlsLence of Lwo worlds, LhaL ls Lhe world of human belngs and Lhe world of
commodlLles, whlch are lnverslons of each oLher, ls a precondlLlon LhaL holds Lhls
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

131

problemaLlc soclal medlaLlon LogeLher. Accordlng Lo Marx, a Lwo-world perspecLlve ls
sympLomaLlc of such an lndlvlduaLlon process LhaL undermlnes Lhe freedom of Lhe sub[ecL
fundamenLally. ln oLher words, Marx does noL dlrecLly crlLlclze Lhe foct LhaL soclal medlaLlon
Lakes place behlnd Lhe back of Lhe producers vlo commodlLles - for, Lhls ls yeL anoLher
symptom of Lhe fundamenLal unfreedom LhaL emanaLes from Lhe lndlvlduaLlon process -,
buL Lhe very form of Lhls soclal medlaLlon.
22


8ulldlng on Lhls more complex undersLandlng of commodlLy feLlshlsm Lukcs deflnes
telflcotloo as a speclflc problem of modern caplLallsm ln conLrasL Lo olleootloo, whlch clalms
valldlLy for a much broader hlsLorlcal perlod and ls usually undersLood ln Lerms of a fall from
some golden age. lnsLead, Lukcs uses Lhe concepL of relflcaLlon Lo glve an expresslon Lo Lhe
effecLs of commodlLy feLlshlsm ln all spheres of caplLallsL socleLy. 1hus, looklng for
concepLual means Lo explaln a broader range of phenomena ln modern everyday llfe, Lukcs
brlngs Max Weber and Ceorg Slmmel lnLo Lhe dlscusslon.
1he prlnclple of formal equallLy", on whlch Lukcs bullds hls Lheory of relflcaLlon, ls an
absLracLlon from Marx's commodlLy-form analysls. 1he concepL of formal equallLy" noL only
helps Lukcs Lo ldenLlfy an essenLlal feaLure of Lhe soclal medlaLlon speclflc Lo bourgeols
socleLy buL also enables hlm Lo lncorporaLe Lhe raLlonallzaLlon Lheory of Weber lnLo hls
crlLlque of caplLallsm. 1hrough Lhe lncreaslng raLlonallzaLlon LhaL ls pecullar Lo modern
world, Lukcs argues, lnsLrumenLal-raLlonallLy ls lnsLlLuLlonallzed ln every sphere of llfe.
arallel Lo Lhe commodlLy produclng absLracL labor, whlch generaLes a formal equallLy
beLween Lhe producLs so LhaL Lhey can be exchanged, lnsLrumenLal raLlonallLy, Loo, seLs
everyLhlng lnLo a relaLlonshlp of formal equallLy. ln order Lo funcLlon smooLhly, modern
socleLy relles on lnsLrumenLal raLlonallLy Lo reduce everyLhlng Lo a cerLaln form of ob[ecLlvlLy,
whlch can, Lhen, be colcoloteJ. 1herefore, Lukcs concludes, noL only ln Lhe assembly llne,
buL vlrLually ln every sphere of llfe, where, ln Pabermas' words, lnsLrumenLal raLlonallLy
colonlzes Lhe llfe-world, Lhe lndlvlduals are sub[ecLed Lo an ob[ecLlve scheme of raLlonallzed
acLlvlLy.
1he lnsLlLuLlonallzaLlon of lnsLrumenLal raLlonallLy compels lndlvlduals Lo a cootemplotlve
stooce, because, Lo use Lukcs' own words, 'Lhe acLlvlLy of Lhe agenL becomes less and less
acLlve' as Lhe agenL acLs slmply ln conformlLy wlLh Lhe glven laws of Lhe parLlcular sphere ln
whlch he flnds hlmself. 1he noLlon of conLemplaLlve sLance ls Lhe key Lo undersLand Lukcs'
concepL of relflcaLlon. 1o explaln Lhls polnL lL ls necessary Lo quoLe Lukcs aL some lengLh:

1he conLemplaLlve sLance adopLed Lowards a process mechanlcally conformlng Lo flxed laws
and enacLed lndependenLly of man's consclousness and lmpervlous Lo human lnLervenLlon, l.e.
a perfecLly closed sysLem, musL llkewlse Lransform Lhe baslc caLegorles of man's lmmedlaLe
aLLlLude Lo Lhe world: lL reduces space and Llme Lo a common denomlnaLor and degrades Llme
Lo Lhe dlmenslon of space [.] 1hus Llme sheds lLs quallLaLlve, varlable, flowlng naLure, lL freezes
lnLo an exacLly dellmlLed, quanLlflable conLlnuum fllled wlLh quanLlflable 'Lhlngs', lL becomes
space. (Lukcs, lbld., p. 101)

Already wlLh Lhls Lhesls Lhe concepLual LranslLlon Lo Lhe phllosophlcal dlscusslon of relflcaLlon
ls provlded for, whlch wlll slLuaLe Lhe problem wlLhln Cerman ldeallsm. lor here, wlLhouL
Lrylng Lo conceal lL, Lukcs lnLroduces Lhe ldea LhaL Lhe modern caplLallsL raLlonallzaLlon
dlsLorLs even 'Llme', LhaL ls, Lo puL lL ln kanLlan Lermlnology, 'Lhe sub[ecLlve condlLlon under

22
Pence, Marx does noL propose a pollLlcal sLraLegy agalnsL Lhe llls of 'commodlflcaLlon' as proposed
by crlLlcs lnsplred by Lhe orLhodox marxlsL lnLerpreLaLlon, buL raLher demands Lhe overcomlng of Lhls
form of lndlvlduaLlon process Lhrough Lhe consLlLuLlon of an lnLersub[ecLlve one.

Bumana.Nente - Issue 1u- }uly 2uu9

132

whlch all our lnLulLlons Lake place'.
23
ln oLher words, deprlvlng Lhe LranscendenLal Sub[ecL of
lLs lnward lnLulLlon, caplLallsL socleLy reverses Lhe achlevemenL of kanL's Copernlcan
8evoluLlon" ln phllosophy, whlch glves Lhe prlmary role ln undersLandlng, or more correcLly,
ln consLrucLlng Lhe reallLy, Lo Lhe knowlng Sub[ecL.
8efore golng lnLo Lukcs' dlscusslon of Lhe concepL of relflcaLlon ln Lhe phllosophlcal parL
of hls essay, lL ls necessary Lo Lake lnLo accounL a slgnlflcanL problem LhaL arlses aL Lhls
[uncLure of hls argumenL. 1he key poslLlon, whlch Lhe noLlon of conLemplaLlve sLance aLLalns
ln Lukcs' analysls, reveals how deeply he ls lnfluenced by Ceorg Slmmel's lnqulry lnLo Lhe
modern forms of sub[ecLlvlLy ln hls book lbllosopby of Mooey.
24
WrlLLen expllclLly as a
polemlc agalnsL Marx, Slmmel's sLudy ls also one of Lhe flrsL works Lo appreclaLe Lhe
normaLlve conLenL of Marx's analysls of Lhe soclal medlaLlon ln commodlLy-form.
23
Accordlng
Lo Slmmel's readlng, Lhe soclal medlaLlon lnherenL ln Lhe worklngs of money economy brlngs
lnLo belng an aLomlsLlc form of lndlvldual freedom LhaL manlfesLs lLself as an aLLlLude of
loJlffeteoce Lowards oLhers. 1hls aLLlLude of lndlfference however, Slmmel argues, ls flrsL Lo
be undersLood on purely descrlpLlve Lerms and for LhaL reason cannoL be [udged as
presumably Marx does, only ln negaLlve Lerms. lor Slmmel, lL ls posslble Lo have a more
balanced vlew of Lhe modern form of freedom, lf Lhe enabllng and empowerlng quallLles
generaLlng from Lhe aLLlLude of lndlfference are Laken lnLo conslderaLlon. lL can be argued
LhaL whlle approprlaLlng Slmmel's analysls Lukcs re[ecLs Lhls balanclng acL. lor alLhough he
doesn'L explaln why, ln Lukcs' accounL Lhe conLemplaLlve sLance ls necessarlly Lhe
undeslrable ouLcome of caplLallsL soclallzaLlon. And yeL, Lukcs' sLrlcLly negaLlve evaluaLlon
of Lhe conLemplaLlve sLance requlres a normaLlve sLandard. Slnce Lukcs ldenLlfles kanL's
phllosophy as a parL of Lhe problem of relflcaLlon, lL would be self-defeaLlng for hls argumenL
Lo bulld hls normaLlve crlLlque on Lhe grounds LhaL caplLallsL socleLy presenLs a dlsLorLlon of
Lhe capaclLles of LranscendenLal sub[ecL as deflned by kanL. 1hus, aL Llmes Lukcs Lrles Lo
solve Lhls problem of normaLlvlLy by presenLlng Lhe conLemplaLlve sLance as a paLhologlcal
sLaLe, as an ouLcome of dehumanlzaLlon broughL by caplLallsm: 1he worker's faLe ls Lyplcal
of socleLy as a whole ln LhaL Lhls self-ob[ecLlflcaLlon, Lhls LransformaLlon of a human funcLlon
lnLo a commodlLy reveals ln all lLs sLarkness Lhe dehumanlzed and dehumanlzlng funcLlon of
Lhe commodlLy relaLlon. (lbld., p. 104) Powever, Lhls ls a slmllarly self-defeaLlng argumenL.
As hls lengLhy dlscusslon ln Lhe Lhlrd parL of hls essay, whlch presenLs bls 'orLhodox MarxlsL'
crlLlque of relflcaLlon as an alLernaLlve Lo whaL he Lakes Lo be Lhe problemaLlc accounL of Lhe
humanlsL noLlon of allenaLlon makes lL clear, Lukcs ls aware of Lhls problem. 1here remalns
only one oLher opLlon Lo subsLanLlaLe Lukcs' negaLlve normaLlve [udgmenL of conLemplaLlve
sLance. Lukcs Lrles Lo pursue Lhls opLlon ln hls phllosophlcal dlscusslon of relflcaLlon, where
he enLers Lhe realm of Cerman ldeallsm.
WlLhln Lhe scope of Lhls paper lL ls noL posslble Lo glve a deLalled accounL of Lukacs'
engagemenL wlLh Cerman ldeallsm. Sufflce lL Lo say LhaL ln a bold aLLempL he brlngs LogeLher

23
kanL, ktltlk Jet teloeo vetoooft, (Pamburg: Melner verlag, 1998), p. 109f.
24
Ceorg Slmmel, lbllosopble Jes CelJes, [lrankfurL am Maln: Suhrkamp, 1989] (1900).
23
ln facL Slmmel does noL conceal hls concern ln wrlLlng hls sLudy, even Lhough Lhe maln opponenL
Marx's name ls menLloned only Lhree Llmes ln Lhe LexL. ln lnLroducLlon he wrlLes: MeLhodologlcally
Lhe baslc lnLenLlon can be expressed ln Lhe followlng manner. 1he aLLempL ls made Lo consLrucL a new
sLory beneaLh hlsLorlcal maLerlallsm such LhaL Lhe explanaLory value of lncorporaLlon of economlc llfe
lnLo Lhe causes of lnLellecLual culLure ls preserved, whlle Lhese economlc forms Lhemselves are
recognlzed as Lhe resulL of more profound valuaLlons and currenLs of psychologlcal or even
meLaphyslcal pre-condlLlons. (lbld, p. 12) ln oLher words Slmmel wanLs Lo hold prlmordlal facLs
agalnsL Marx's crlLlque of caplLallsm based on value-form analysls.
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

133

Lhe phllosophlcal agendas of several phllosophers, among Lhem prlmarlly kanL, Schlller,
llchLe, Pegel and Marx, around a relaLlvely slmple Lhesls. l wlll conclude Lhls parL of Lhe
paper by elucldaLlng Lhls Lhesls and polnLlng ouL lLs unexplored poLenLlal.
Cn mosL general Lerms Lukcs bullds hls argumenL on Lhe fundamenLal assumpLlon LhaL
our aLLlLude Lowards Lhe world depends upon how we percelve, comprehend and/or grasp
(beqtelfeo) lL. Slnce our aLLlLude towotJs Lhe world deLermlnes how we acL lo Lhe world, Lhe
quesLlon of percepLlon ls of normaLlve value. Cn Lhls assumpLlon, Lukcs beglns Lhe second
parL of hls essay wlLh Lwo lnLerdependenL clalms. llrsL, commenLlng on kanL's second
preface ln hls flrsL ctltlpoe, where he brlngs forward Lhe meLaphor of Copernlcan
8evoluLlon,
26
Lukcs argues as follows: Modern phllosophy seLs lLself Lhe followlng problem:
lL refuses Lo accepL Lhe world as someLhlng LhaL has arlsen (or e.g. has been creaLed by Cod)
lndependenLly of Lhe knowlng sub[ecL, and prefers Lo concelve of lL lnsLead as lts owo
ptoJoct (lbld, p. 123). 1he deflnlng feaLure of kanL's undersLandlng of sub[ecLlvlLy ls a
parLlcular undersLandlng of freedom, LhaL ls, auLonomy. Lukcs broadens Lhls concepLlon of
freedom Lo lnclude a noLlon of self-deLermlnaLlon, by hlghllghLlng Lhe conLrlbuLlon of Lhe
sub[ecL ln consLrucLlng Lhe reallLy. WlLh Lhe noLlon of knowledge as 'producL', here Lukcs
emphaslzes hls genulne MarxlsL and arguably 'maLerlallsL' concern. Second, a couple of pages
laLer Lukcs puLs forward a complemenLary clalm: WhaL ls novel abouL modern raLlonallsm
ls lLs lncreaslngly lnslsLenL clalm LhaL lL has dlscovered Lhe ptloclple whlch connecLs up all
phenomena whlch ln naLure and socleLy are found Lo confronL manklnd. Compared wlLh Lhls,
every prevlous Lype of raLlonallsm ls no more Lhan a pottlol system. (lbld, p. 123) 1o cuL a
long dlscusslon shorL, Lhls prlnclple ls Lo be found ln Lhe noLlon of Sub[ecL, or Lo puL lL more
Lechnlcally ln Lhe concepL of 5elf-cooscloossoess. 1he way we percelve and/or consLrucL Lhe
world as well as our aLLlLude Lowards lL depends upon our undersLandlng and consLrucLlon of
Lhls very consLlLuenL Sub[ecL. 1hus, Lukcs reformulaLes Lhe fundamenLal quesLlon of
Cerman ldeallsm: Pow are we Lo undersLand Lhe Sub[ecL, so LhaL lL does noL conLradlcL lLs
deflnlLlon, namely LhaL lL ls free? Cr Lo puL lL ln Marxlan Lerms: WhaL are Lhe condlLlons of
exlsLence of lndlvldual freedom?
ln answerlng Lhese quesLlons Lukcs' LexL provldes concepLual resources for an alLernaLlve
undersLandlng of relflcaLlon, whlch he does noL pursue hlmself. ln order Lo dlsLlngulsh Lhls
alLernaLlve beLLer, lL ls necessary Lo challenge Lhe apparenL llnearlLy of Lukcs' accounL of
Cerman ldeallsm. lor Lhough lL may seem LhaL here Lukcs ls slmply mapplng Lhe
phllosophlcal paLh leadlng Lo Marx, he acLually draws dlfferenL concepLual resources from
dlfferenL phllosophlcal sysLems wlLhln Cerman ldeallsm. 1hls explalns Lhe cenLral role glven
Lo llchLe as opposed Lhe marglnal role of Pegel ln hls sub[ecL-LheoreLlcal aLLempL Lo
subsLanLlaLe hls negaLlve normaLlve [udgmenL of conLemplaLlve sLance. Slnce lL ls Lhe
deflnlLlon of Lhe Sub[ecL and lLs normaLlve horlzon LhaL deLermlnes Lhe ulLlmaLe deflnlLlon of
relflcaLlon as a sLaLe LhaL ooqbt Lo be overcome, Lhe cenLral role aLLached Lo llchLe ls cruclal.
Lukcs' rellance on llchLe ls, lf parLly, Lhe resulL of hls zealous aLLempL Lo slLuaLe Marx ln
Cerman ldeallsL LradlLlon. lL can also be seen as a backlash of hls romanLlc anLl-caplLallsm.
27

ln any case, Lukcs ln Lhls parL of hls essay draws malnly on an earller LexL of Marx raLher
Lhan on Lhe secLlon on 'commodlLy feLlshlsm' ln copltol, from whlch, as we saw, he orlglnally
derlves hls concepL of relflcaLlon. ln Lhe flrsL Lhesls on leuerbach, young Marx colns Lhe
concepL of raxls. Lukcs brlngs Lhls concepL ln a frulLful exchange wlLh llchLe's 'Sub[ecL of
AcLlon' and concludes LhaL aL Lhe momenL Lhe sub[ecL-ob[ecL of hlsLory, LhaL ls, Lhe

26
See kanL, lblJ, p. 13-40.
27
Cn Lhls polnL see Lhe deLalled analysls of Mlchael Lwy, ltom komootlclsm to 8olsbevlsm, (London:
nL8, 1979).

Bumana.Nente - Issue 1u- }uly 2uu9

134

proleLarlaL becomes consclous of lLs consLlLuLlng force, Lhe condlLlons of exlsLence of
lndlvldual freedom wlll be esLabllshed.
1he eschaLologlcal Lone of such an accounL ls deeply problemaLlc and has been crlLlclzed
by many oLhers. 1he same accounL can also be crlLlclzed for lLs pollLlcal lmpllcaLlons, for lL
brlngs wlLh lL a serles of quesLlons LhaL remaln relevanL for a cerLaln LradlLlon of MarxlsL
pollLlcs. 1o name a few: WhaL ls Lhe dlfference beLween a !acobln We" and a llchLean-
roleLarlan Sub[ecL? Can Lhe self-consclousness of Lhe proleLarlaL be delegaLed? Who decldes
when Lhe proleLarlaL ls maLure for a revoluLlon? And flnally, how do we know, wheLher Lhe
proleLarlaL has reached Lo self-consclousness, lf our answer ls noL slmply Lo say LhaL Lhe proof
of puddlng ls ln eaLlng lL? All Lhese quesLlons refer Lo a normaLlve deflclL ln Lukcs' accounL,
desplLe hls aLLempL Lo provlde for a powerful normaLlve crlLlque of caplLallsm. lf Lhls ls Lhe
case, Lhen lL ls Lukcs' deflnlLlon of relflcaLlon LhaL needs Lo be crlLlclzed. lor lL ls Lhls
deflnlLlon, whlch ls responslble for Lhe normaLlve deflclL ln quesLlon.
1o puL lL slmply, wlLh hls deflnlLlon of relflcaLlon, Lukcs LargeLs a stote of a Sub[ecL, whlch
ls Lo be exchanged wlLh anoLher sLaLe. 1he acLlve 5object ls Lhe proper Sub[ecL, slnce lL ls
capable of keeplng ln conLrol of Lhe reallLy LhaL lL poslLs, whereas Lhe conLemplaLlve sLance
lacks Lhls capaclLy and tbetefote ends up wlLh a relfled reallLy beyond lLs reach. undersLood
ln Lhese Lerms, relflcaLlon Lurns ouL Lo be Lhe loss of a fanLasy of omnlpoLence. 1hls sLaLe of
omnlpoLence, however, ls arguably lLself a relfled sLaLe. lor Lhls fanLasy defles any medlaLlon
beLween sub[ecL and ob[ecL, poslLlng an lmmedlaLe ldenLlLy beLween Lhe Lwo. ln oLher
places, however, Lukcs seems Lo polnL Lo a dlfferenL undersLandlng, parLlcularly when he
glves a more Pegellan accounL of Lhe supercedlng (Aufhebung) of relflcaLlon. 1hus, for
lnsLance, desplLe Lhe facL LhaL Lukcs sLlll reads Pegel Lhrough Lhe lenses of llchLe and
consequenLly asslgns a cruclal role Lo Lhe masLer and slave dlalecLlcs, aL cerLaln polnLs ln hls
accounL, he seems Lo suggesL LhaL Lhe overcomlng of relflcaLlon ls posslble Lhrough a form of
soclal medlaLlon, whlch would consLlLuLe Lhe sub[ecLs as consLlLuenLs of Lhe reallLy LhaL Lhey
Lake parL ln. 1hls soclal medlaLlon lnvolves a reflecLlve self-consclousness LhaL requlres an
lnLersub[ecLlve relaLlonshlp Lhrough Lhe consLrucLlon of a plural ldenLlLy, a 'We' LhaL would
noL undermlne Lhe lndlvldual dlfferences of lLs members. Lukcs refers noL only Lo llchLe, buL
also repeaLedly Lo Lhe lbeoomeooloqy of 5pltlt ln supporL of Lhls argumenL and speclflcally Lo
Lhe LranslLlon from Lhe chapLer on 'lorce and undersLandlng', where Lhe ahlsLorlcal and
monologlcal concepLuallzaLlon of Lhe sub[ecL ls crlLlclzed, Lo Lhe chapLer on 'Self-
conscloussness'.
28
1hus Lukcs seLs Lhe problem as LhaL of flndlng Lhe approprlaLe Sub[ecL
LhaL would sLrlp off Lhe llmlLaLlons puL on Lhe monologlcal Sub[ecL. 8ecause of hls essenLlally
llchLean approach, however, he puLs hls emphasls on how Lhls approprlaLe Sub[ecL would
overcome Lhe separaLlon beLween sub[ecL and ob[ecL, lnsLead of clarlfylng Lhe
lnLersub[ecLlve medlaLlon requlred for Lhls overcomlng. 1hus brlnglng Pegel and llchLe
LogeLher ln one paragraph Lukcs wrlLes:

Pere ln our newly-won knowledge where, as Pegel puLs lL ln Lhe lbeoomeooloqy, Lhe Lrue
becomes a 8acchanLlc orgy ln whlch no one escapes belng drunk, reason seems Lo have llfLed
Lhe vell conceallng Lhe sacred mysLery [.] and dlscovers [.] LhaL lL ls lLself Lhe soluLlon Lo Lhe
rlddle. 8uL here, we flnd [.] Lhe declslve problem of Lhls llne of LhoughL: the prob/em of the
subject of the oction, the subject of the qenesis. lor Lhe unlLy of sub[ecL and ob[ecL, of LhoughL
and exlsLence, whlch Lhe 'acLlon' underLook Lo prove and Lo exhlblL, flnds boLh lLs fulflllmenL
and lLs subsLraLum ln Lhe unlLy of Lhe genesls of Lhe deLermlnanLs of LhoughL and of Lhe hlsLory
of Lhe evoluLlon of reallLy. 8uL Lo comprehend Lhls unlLy lL ls necessary boLh Lo dlscover Lhe slLe

28
Compare Lukcs, lbld., p.137f, 161f., 191f., 211f.
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

133

from whlch Lo resolve all Lhese problems and also Lo exhlblL Lhe 'we' whlch ls Lhe
sub[ecL of hlsLory, LhaL 'we' whose acLlon ls ln facL hlsLory. (lbld, p. 161)

Lukcs refers above flrsL Lo Pegel ln polnLlng ouL Lhe slgnlflcance of Lhe concepLuallzaLlon of
self-consclousness ln reallzlng whaL he deems Lo be Lhe radlcal freedom of sub[ecL. lnsLead of
followlng Pegel's deflnlLlon of self-consclousness LhaL presupposes an lnLersub[ecLlve
medlaLlon beLween sub[ecLs, however, he Lurns Lo llchLe and deflnes Lhe essenLlal feaLure of
a free sub[ecLlvlLy wlLh Lhe noLlon of 'Sub[ecL of AcLlon'. And yeL, he relnLroduces Lhe
Pegellan perspecLlve Lo accounL for Lhe hlsLorlcal genesls of Lhe 'Sub[ecL of AcLlon'. 1hls
hlsLorlcal sub[ecL, 'we', ls Lhe 'proleLarlaL as a class'. Agaln referrlng Lo Pegel, Lukcs wrlLes:

1he lndlvldual can never become Lhe measure of all Lhlngs. lor, when Lhe lndlvldual confronLs
ob[ecLlve reallLy he ls faced by a complex of ready-made and unalLerable ob[ecLs whlch allow
hlm only Lhe sub[ecLlve responses of recognlLlon or re[ecLlon. Cnly Lhe class can relaLe Lo Lhe
whole of reallLy ln a pracLlcal revoluLlonary way. (1he 'specles' cannoL do Lhls as lL ls no more
Lhan an lndlvldual LhaL has been myLhologlzed and sLyllzed ln a splrlL of conLemplaLlon) (lbld.,
p. 211).

As Pegel argues, and Lukcs agrees, a plural ldenLlLy, a 'we', cannoL be consLlLuLed ln
reference Lo a pre-glven naLure of Lhe 'specles', or ln reference Lo human naLure. 1he
revoluLlonary sub[ecL LhaL wlll overcome Lhe dlvlslon beLween sub[ecL and ob[ecL, has Lo
come lnLo belng Lhrough a process of soclal and hlsLorlcal medlaLlon. lnsLead of reflecLlng
upon Lhe process of Lhls medlaLlon, Lhe necesslLy of whlch ls Lhus grounded ln hls Lheory of
relflcaLlon, Lukcs chooses, however, Lo focus on Lhe copoclty of Lhls revoluLlonary sub[ecL Lo
approprlaLe Lhe reallLy. As a resulL boLh relflcaLlon and lLs overcomlng are presenLed as
stotes of a slngular and/or a plural sub[ecL raLher Lhan referrlng Lo Lhe process of lLs
consLlLuLlon ln and Lhrough soclal and hlsLorlcal relaLlons. 1hus lnsLead of followlng Pegel's
perspecLlve and Marx's analysls of commodlLy and crlLlque of commodlLy feLlshlsm ln
deflnlng Lhe mode of soclal medlaLlon LhaL would overcome Lhe aLomlsLlc sub[ecLlvlLy of
lndlvlduals llvlng ln caplLallsL socleLy, Lukcs puLs hls hopes ln a surrogaLe of a plural
'ldenLlLy', LhaL ls, a revoluLlonary parLy LhaL has Lo play Lhe role of a 'Sub[ecL of AcLlon'.
Lukcs does noL clarlfy whaL form of lndlvlduaLlon process has Lo replace Lhe relfled soclal
relaLlons. 1hus, aL Lhe end of hls essay, ln hls response Lo Lhe quesLlon of how Lo organlze Lhe
proleLarlaL Lo consLlLuLe lL as Lhe sub[ecL-ob[ecL of hlsLory, Lukcs ends up wlLh no cholce
oLher Lhan Lo quoLe Lngels afflrmaLlvely: 1he proof of Lhe puddlng ls ln Lhe eaLlng [.] 1hls
puddlng, however, ls Lhe maklng of Lhe proleLarlaL lnLo a class: Lhe process by whlch lLs class
consclousness becomes real ln pracLlce (lbld, p. 217). WlLh Lhls Lurn Laken, Lukcs' concepL
of relflcaLlon falls Lo capLure Lhe essence of Marx's normaLlve crlLlque of caplLallsm, Lo whlch
we now Lurn.

1he flrsL chapLer of copltol glves a very LlghLly knlL accounL of how we are Lo undersLand Lhe
commodlLy as Lhe mosL absLracL caLegory and commodlLy-form as Lhe mosL elemenLary form
of caplLallsL relaLlons of producLlon.
29
As Marx argues, Lhe key Lo undersLandlng Lhls absLracL

29
Marx polnLs Lo Lhe cenLrallLy of Lhe flrsL chapLer ln ltefoce to tbe fltst eJltloo of copltol as follows:
1he value-form, whose fully developed shape ls Lhe money-form, ls very elemenLary and slmple.
neverLheless, Lhe human mlnd has for more Lhan 2,000 years soughL ln valn Lo geL Lo Lhe boLLom of lL
all, whllsL on Lhe oLher hand, Lo Lhe successful analysls of much more composlLe and complex forms,

Bumana.Nente - Issue 1u- }uly 2uu9

136

caLegory lles ln lLs fotm, LhaL ls, ln Lhe form of soclal medlaLlon, whlch Lransforms labor lnLo
voloe (MLW 23, p. 94f.). Marx beglns hls analysls wlLh an lndlvldual commodlLy, Lhen moves
on Lo Lhe analysls of a relaLlon lnlLlally beLween Lwo commodlLles, whlch glves way Lo an
analysls of Lhe wotlJ of commoJltles. 1wo dlfflculLles arlse ln lnLerpreLlng Lhls chapLer, lf
Marx's meLhod of exposlLlon ls noL Laken lnLo conslderaLlon.
llrsL, ln Marx's accounL Lhere ls an apparenL lnconslsLency wlLh respecL Lo whaL
consLlLuLes value. lor Marx suggesLs LhaL value-creaLlng labor ls boLh an absLracL energy
quanLum and obsttoct lobot, undersLood as a producL of Lhe soclal medlaLlon necessary ln
order Lo produce commodlLles. Second, lL becomes exLremely dlfflculL Lo make sense of
Marx's very deLalled exposlLory concepLual accounL LhaL move from a slngle commodlLy Lo
Lhe world of commodlLles. WlLh respecL Lo Lhe second polnL, many commenLaLors flnd refuge
ln Marx's sLaLemenL aL Lhe very beglnnlng of hls analysls of Lhe commodlLy-form. Slnce ln Lhe
beglnnlng of Lhls secLlon Marx clalms Lo have already solved Lhe rlddle of commodlLy-form,
Lhe remalnlng flfLeen pages LhaL consLlLuLe Lhe value-form analysls ls frequenLly lgnored by
Lhe readers of Marx.
30
l suggesL LhaL a readlng whlch draws a parallel beLween Lhls chapLer
and Lhe chapLers of 'ercepLlon' and 'lorce and undersLandlng' of Lhe lbeoomeooloqy of tbe
5pltlt solves Lhese dlfflculLles.
1he explanaLory power of such a readlng can be summarlzed as follows. llrsL, Lhrough Lhls
readlng lL ls posslble Lo Lrack Marx's meLhodologlcal lnLeresL ln hls exposlLlon (uarsLellung)
and accounL Lhe seemlng lnconslsLencles ln hls dlscusslon. ln Lhe flrsL exposlLory sLep of Lhls
chapLer, a slngle commodlLy, as a tbloq, ls presenLed as Lhe ob[ecL of analysls, whereas ln Lhe
value-form analysls LhaL follows Lhls accounL a totloool-telotloo ls aL sLake. Marx searches, l
argue, for Lhe adequaLe model of knowledge Lo comprehend and fully develop Lhe ob[ecL of
hls analysls. ln dolng so, Marx, [usL llke whaL Pegel does, ln Lhe lbeoomeooloqy of 5pltlt,
ascrlbes Lhe lnconslsLencles and conLradlcLlons wlLhln hls exposlLlon Lo Lhe lnsufflclencles of
Lhe models of knowledge. undersLood ln Lhls way, Lhe lnsufflclency of Lhe flrsL model ls
superceded ln Lhe second model. 1hls ls why, for lnsLance, Lhe flrsL deflnlLlon of voloe-
cteotloq lobot as energy quanLum dlffers from Lhe second deflnlLlon of obsttoct lobot as
producL of soclal relaLlons. lor Lhls second deflnlLlon follows from Lhe flrsL deflnlLlon's
lnsufflclenL undersLandlng of Lhe absLracLness of labor.
Second, when read Lhrough Lhe lenses of Lhe lbeoomeooloqy of tbe 5pltlt, lL becomes
posslble Lo dlscern Lhe subsLanLlve conLenL of Lhe exposlLory developmenL of Lhe ob[ecL of
analysls ln copltol from a slngle commodlLy Lo Lhe world of commodlLles. Marx
concepLuallzes Lhe flrsL, emplrlcal ob[ecL of hls analysls as a tbloq wlLh speclol poolltles, LhaL
ls, boLh as a Lhlng Lo be consumed and as a Lhlng of value. ln dolng so, he deflnes
'commodlLy' ln Lerms of Lhe ob[ecL of Pegel's chapLer, 'ercepLlon'. 1he lnsufflclency ln
concepLuallzlng Lhls Lhlng wlLh lLs double naLure ls superceded by Lhe conslderaLlon of a
telotloo beLween commodlLles, whlch can only be Lhe ob[ecL of a totloool concepLuallzaLlon.

Lhere has been aL leasL an approxlmaLlon. Why? 8ecause Lhe body, as an organlc whole, ls easler Lo
sLudy Lhan are Lhe cells of LhaL body. ln Lhe analysls of economlc forms, moreover, nelLher
mlcroscopes nor chemlcal reagenLs are of use. 1he force of absLracLlon musL replace boLh. 8uL ln
bourgeols socleLy, Lhe commodlLy-form of Lhe producL of labour - or value-form of Lhe commodlLy -
ls Lhe economlc cell-form. 1o Lhe superflclal observer, Lhe analysls of Lhese forms seems Lo Lurn upon
mlnuLlae. lL does ln facL deal wlLh mlnuLlae, buL Lhey are of Lhe same order as Lhose dealL wlLh ln
mlcroscoplc anaLomy. (MLW 23, p. 12)
30
See among many such lnLerpreLaLlons LrnsL Mlchael Lange's oLherwlse lnslghLful commenLary.
LrnsL Mlchael Lange, WerLformanalyse, CeldkrlLlk und dle konsLrukLlon des leLlschlsmus bel Marx" ln
Neoe nefte fot lbllosopble, PefL 13, p. 1-46.
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

137

lollowlng Lhe Pegellan dlscusslon of 'lorce and undersLandlng,' Marx characLerlzes
commodlLles ln Lhls relaLlon as reallzaLlons of an lmmaLerlal fotce, so LhaL Lhey aLLaln a
phenomenal exlsLence, whlch ls a form". lL ls only afLer esLabllshlng Lhls LhaL Marx beglns Lo
analyze Lhe commoJlty-fotm. arallel Lo Pegel's model of knowledge ln lorce and
undersLandlng", Lhe relaLlon among commodlLles lmprlnLs Lhem wlLh Lhelr voloe-fotm, and
embodles a law, low of voloe, whlch regulaLes Lhe world of commodlLles.
31
1hls world
presupposes lLs lovetteJ lmage, Lhe world of human belngs, who exchange commodlLles ln
order Lo make use of Lhelr maLerlal, noL pbeoomeool, quallLles. 1he medlaLlon beLween Lhese
Lwo worlds consLlLuLes Lhe soclal relaLlon called copltol ln Lhe proceedlng chapLers.
MosL lmporLanLly drawlng Lhls parallel beLween Lhe openlng chapLers of copltol and Lhe
lbeoomeooloqy of 5pltlt makes lL posslble Lo accounL for Lhe normaLlve core of Marx's
crlLlque of caplLallsm. WlLhln Lhe scope of Lhls paper, lL ls lmposslble Lo fully presenL Lhe
deLalled lnLerpreLaLlve analysls, whlch ls necessary Lo subsLanLlaLe Lhls polnL. WlLh Lhls ln
mlnd, ln my concludlng remarks, l wlll llmlL my dlscusslon Lo provldlng a brlef accounL of Lhe
concluslons drawn from such an analysls.
ln boLh Pegel's accounL of fotce and Marx's analysls of commoJlty-fotm a dlsLlncL form of
sub[ecL-relaLlon ls LhemaLlzed. 1he normaLlve sLandard of Marx's crlLlque of caplLallsm ls
conLalned wlLhln Lhe crlLlque of Lhls form of sub[ecL-relaLlon. ln Lhe analysls of voloe-fotm
Marx descrlbes a soclal medlaLlon, a socleLy, whlch ls based on a radlcally lndlvlduallsLlc,
aLomlzed form of sub[ecLlvlLy. 1hls ls noL [usL a descrlpLlve accounL LhaL denoLes, say, Lhe
loJlffeteoce of Sub[ecLs agalnsL each oLher. Marx, followlng Pegel's accounL of fotce,
descrlbes a world of moooJs as proposed by Lelbnlz. 8oLh for Pegel and Marx Lhe form of
Sub[ecLlvlLy descrlbed by Lelbnlz enLalls an undersLandlng of freedom and lndlvlduallLy LhaL ls
conLradlcLory wlLhln lLself. A monadologlcal concepLlon of freedom ls deflclenL because lL
proves Lo be a fully deLermlned relaLlon of necesslLy beLween sub[ecLs as Lhe monadlc
lndlvlduallLy glves way Lo an absLracL unlversal form of equallLy, whlch obllLeraLes all forms of
dlfference. ln Marx's analysls of voloe-fotm, Lhe deLalled exposlLlon of Lhls deflclenL form of
Sub[ecLlvlLy - deflclenL, because Lhe self-reflecLlon of commodlLles presuppose an absoluLe
Sub[ecL (l.e. Money), whlch annlhllaLes Lhe sub[ecL sLaLus of commodlLles - conLalns, aL Lhe
same Llme, lLs lmmanenL, Jetetmlooot oeqotloo, LhaL ls, Lhe proper form of sub[ecLlvlLy. 1hls
proper form of sub[ecLlvlLy and Lhe concepLlon of freedom and lndlvlduallLy LhaL lle aL lLs
core can be deflned as a Lrue, non-deflclenL, relaLlonshlp of lnLersub[ecLlvlLy. ln
lbeoomeooloqy of 5pltlt, Pegel deflnes Lhls reclprocal relaLlonshlp of lnLersub[ecLlvlLy wlLh
Lhe help of dlalecLlcs of recognlLlon. ln modellng caplLallsL mode of soclal medlaLlon afLer a
dlalecLlcal relaLlon LhaL necessarlly leads Lo Lhe dlalecLlc of recognlLlon, Marx seLs Lhe
sLandard of hls normaLlve crlLlque of caplLallsm as recognlLlon.
1hls brlefly skeLched alLernaLlve readlng of Lhe openlng argumenLs of copltol, Lhus,
conLrlbuLes Lo a beLLer undersLandlng of Marx's normaLlve crlLlque of caplLallsm. lL also
enables us Lo reLhlnk Lukcs' concepL of relflcaLlon along Lhe llnes argued by Marcuse and
PonneLh. 8oLh Pegel ln hls chapLer 'lorce and undersLandlng' ln Lhe lbeoomeooloqy, and
Marx ln hls analysls of commodlLy-form, provlde a crlLlque of a telfleJ soclal world based on a
speclflc soclal medlaLlon. 1hey boLh emphaslze LhaL Lhe reallLy percelved by a Lelbnlzlan
monad-Sub[ecL, whlch ls Lhe ouLcome of such a deflclenL medlaLlon, ls a reallLy beyond Lhe
conLrol of Lhe sub[ecL by deflnlLlon. 8ecause, Pegel and Marx argue, lnsofar as Lhe monad-
Sub[ecL clalms Lo approprlaLe Lhe reallLy as lLs own, lL has Lo poslL lL lndependenLly of any
hlsLorlcally deflnlLe soclal pracLlce, LhaL ls, as a relfled Lhlng ln lLself. 1hus, Lhe reallLy aLLalns a

31
1he parallel concepL Pegel uses aL Lhls [uncLure ls ruhlges 8elch der CeseLze", 'Lranqull klngdom of
laws'. See Pegel lbooomeooloqle Jes Celstes (Pamburg: Melner verlag 1988), p. 111.

Bumana.Nente - Issue 1u- }uly 2uu9

138

deLached quallLy, and becomes a power over Lhe monad-Sub[ecL renderlng lLs relaLlon Lo
oLher sub[ecLs lnessenLlal. lor, Lhe monad-Sub[ecL acqulres lLs lndlvlduallLy / ldenLlLy Lhrough
Lhe medlaLlon of an omnlpoLenL reallLy, namely Lhrough Cod. ln caplLallsL socleLy, Marx
argues, Lhe consLlLuenLs of socleLy are lmprlsoned ln such a form of lndlvlduallLy. 1he soclal
relaLlon among aLomlzed lndlvlduals can be characLerlzed by an aLLlLude of loJlffeteoce, for
Lhelr lndlvlduallLy / ldenLlLy ls qlveo Lhrough Lhe medlaLlon of an omnlpoLenL sub[ecL, l.e.
money. ln Lhls sense, Lhls soclal medlaLlon crlLlclzed by Marx can be called re|f|cat|on proper.
1hls lmplles LhaL Lhe revoluLlonary praxls agalnsL relflcaLlon ls Lo re[ecL all soclal relaLlons LhaL
succumb Lo a deflclenL mode of lnLersub[ecLlvlLy.

lor Lhe revoluLlonary praxls, Lhen, Lhe dlalecLlcs of recognlLlon, whlch expresses Lhe ldeal
relaLlonshlp of lnLersub[ecLlvlLy, serves as Lhe norm for a free lndlvlduallLy. ln oLher words, all
soclal medlaLlons LhaL fall Lo adhere Lo Lhe normaLlve ldeal of recognlLlon are relfled.
undersLood as such, Lhls revlsed concepLlon of relflcaLlon reLalns Lhe revoluLlonary lmpeLus
of Lukcs' crlLlclsm of caplLallsm, whlle, aL Lhe same Llme, provldlng a crlLlcal sLandpolnL for
Lhe consLlLuLlon of a new socleLy, ln whlch ln Pegels words Lhe unlLy of Lhe dlfferenL
lndependenL self-consclousness whlch, ln Lhelr opposlLlon en[oy perfecL freedom and
lndependence: 'l' LhaL ls 'We' LhaL ls 'l' (lbld,. p.127) ls posslble.

8I8LICGkAn

AlLhusser L. and 8allbar L. (1979) keoJloq copltol, London: verso [1968].
AraLo A. (1972) Lukcs' 1heory of 8elflcaLlon ln 1elos 11, Sprlng 1972.
8ackhaus P. (1997) ulolektlk Jet wettfotm. uotetsocbooqeo zot Motxcbeo
Okooomlektltlk, lrelburg: a lra [1974].
8rudney u. (1988) Motxs ottempt to leove lbllosopby, Cambrldge: Parvard unlverslLy
ress.
Cohen C.A (1978) kotl Motxs 1beoty of nlstoty. A uefeose, new !ersey: rlnceLon unlv.
ress.
Lngels l. (2003) ulalecLlcs of naLure ln Motx/oqels collecteJ wotks, volume 23, new
?ork: lnLernaLlonal ubllshers, p. 313-390.
Lngels l. (1970) Soclallsm uLoplan and SclenLlflc ln Motx/oqels collecteJ wotks, volume
3, Moscow: rogress ubllshers, p. 93-131.
larr !. (1984) Marx and poslLlvlsm ln 8all / larr (ed) Aftet Motx, Cambrldge: Cambrldge
unlverslLy ress, 1984, p. 217-233.
lraser n. and PonneLh A. (2003) umvettellooq oJet Aoetkeooooq? loe polltlscb-
pbllosopblscbe koottovetse, lrankfurL am Maln: Suhrkamp.
Pabermas !. (1981) 1beotle Jes kommoolkotlveo nooJelos, 8d. 2, lrankfurL am Maln:
Suhrkamp.
Pabermas !. (1988) uet pbllosopblscbe ulskots Jet MoJetoe, lrankfurL am Maln:
Suhrkamp.
Pegel C.W.l (1988) lbooomeooloqle Jes Celstes, Pamburg: Melner verlag.
volkan iuam - 0nueistanuing Naix's Noimative Ciitique of Capitalism

139

Pelnrlch M. (1991) ule wlsseoscboft vom wett. ule Motxcbe ktltlk Jet polltlscbeo
Okooomle zwlscbeo wlsseoscboftllcbet kevolotloo ooJ klosslscbet 1toJltloo,
MunsLer: WesLfallsches uampfbooL.
Poff !. (among oLhers ed.) (2006) uos kopltol oeo leseo, MunsLer: WesLfallsches
uampfbooL .
PonneLh A. (2003) vetJloqllcbooq, lrankfurL am Maln: Suhrkamp.
!ay M. (1984) Motxlsm ooJ 1otollty. 1be AJveototes of o coocept ftom lokcs to
nobetmos Cambrldge: ollLy ress.
kanL l. (1998) ktltlk Jet teloeo vetoooft, Pamburg: Melner verlag.
Lange L. M. (1978) WerLformanalyse, CeldkrlLlk und dle konsLrukLlon des leLlschlsmus bel
Marx ln Neoe nefte fot lbllosopble, PefL 13, p. 1-46.
Lelbnlz C.W. (1998) MoooJoloqle, SLuLLgarL: 8eclam.
Lwy M. (1979) ltom komootlclsm to 8olsbevlsm, London: nL8.
Lohmann C. (1991) loJlffeteoz ooJ Cesellscboft, lrankfurL am Maln: Suhrkamp.
Lukcs C. (2000) ule verdlngllchung und das 8ewussLseln des roleLarlaLs ln Cescblcbte
ooJ klosseobewotselo, London: 8ed SLar ress [1923].
Marcuse P. (1986) keosoo ooJ kevolotloo, Cxford: 8ouLledge & kegan aul [1941].
Marx k. (2001) uos kopltol ln Motx/oqels wetke (MLW) 8and 23-23, 8erlln: uleLz verlag
[1890].
Marx k. (2003) Motx/oqels wetke (MLW) 8and 42, 8erlln: uleLz verlag.
nussbaum M. (2002) Cb[ecLlflcaLlon, ln Alan Soble (ed.) 1be lbllosopby of 5ex, Maryland:
8owman & LlLLlefleld.
eLry l. (1916) uet 5ozlole Cebolt Jet Motxscbeo wetttbeotle, !ena: C. llscher.
lnkard 1. (1994) neqels lbeoomeooloqy. 1be 5oclollty of keosoo, Cambrldge: Cambrldge
unlverslLy ress.
osLone M. (1993) 1lme loboot ooJ 5oclol uomlootloo, Cambrldge, Cambrldge unlverslLy
ress.
8elchelL P. (1970) 2ot loqlscbeo 5ttoktot Jes kopltolbeqtlffs bel kotl Motx, lrankfurL am
Maln: Lurop. verl.-AnsL.
Seserko L. (2006) uas 'wlrkllche 1rlebwerk des kaplLals' und selne 8ezlehung zu Pegels
Loglk ln Poff(ed.) uos kopltol oeo leseo, MunsLer: WesLfallsches uampfbooL.
Slmmel C. (1989) lbllosopble Jes CelJes, lrankfurL am Maln: Suhrkamp [1900].
vaz[ulln v. A. (2003) LnLwlcklung sysLemaLlsch denken . Lln verglelch der dlalekLlschen
Loglk bel Pegel und Marx ln: ueotscbe 2eltscbtlft fot lbllosopble, 33:2, S. 203-18.
Wellmer A. (1971) ctltlcol 1beoty of 5oclety, new ?ork: 1he ConLlnuum ubl. Corp.
Wood A. (1981) kotl Motx, Cxon: 8ouLledge.



Bumana.Nente - Issue 1u- }uly 2uu9





















141

Recensione
Filosofiadell'economia
DanielM.Hausman
AAVV,Filosofiadellescienze,acuradiNiclaVassallo,Einaudi,Torino,2003

AlbertoBinazzi
alberto.binazzi@humanamente.it

Nel2003lacasaeditriceEinaudipubblicavaFilosofiadellescienze,un'antologiacuratadaNicla
Vassallo, nota filosofa italiana che, col suo lavoro, tesseva un metatesto sui problemi
epistemologicisollevatidaogniscienza.Nellibro,checomprendearticolimonograficidedicati
alla filosofia della biologia, alla filosofia della chimica e alla filosofia della scienze cognitive,
compariva anche la traduzione in lingua italiana dell'articolo di Daniel HausmanPhilosophy of
Economics(1998).Iltestoinquestione,perquantodatato,ancoraoggiunadellepichiare
introduzioni alla materia. Hausman infatti risulta un eccellente didatta, non per niente
curatoreasuavoltadidiversihandbooksemanualiusaticonprofittodaglistudentiedacoloro
chesiavvicinanoperlaprimavoltaaquestapurgiovanedisciplina.
L'autore spiega con molta efficacia come le questioni dinteresse filosofico offerte dal
pensiero economico siano molteplici: il rapporto tra condotta economica, welfare e diritti, la
naturadellarazionalit,lebasiepistemologicheemetodologichedellescienzesociali,illegame
tra le varie scuole economiche in relazione al pi ampio problema dello statuto delle scienze
naturali. Numerose sono state, soprattutto nel secolo appena trascorso, le proposte di
soluzione al problema sollevato da John Stuart Mill sulla natura dei principi fondanti della
scienzaeconomica,suscettibili,secondoilfilosofobritannico,diesserestabilitilegittimamente
tramitelintrospezione.Moltediquesteproposte,secondoHausman,tracuiletesidiPopper,
Lakatos, Friedman, non sono state in grado di affrontare correttamente il tema della
inadeguatezzaempiricadelleteorieeconomiche.Negliultimidecenni,poi,sonostateavanzate
nuovelettureepistemologicheemetodologiche,comequellediMcCloskey,Baugh,Caldwelle
Rosenberg, anche molto critiche nei confronti della pretesa da parte delleconomia di
costituirsiatuttiglieffetticomeunascienzaempirica.
Leconomia contemporanea appare oggi come una disciplina assai differenziata al suo
interno. In parte complementari alla visione ortodossa o neoclassica, si sono sviluppate
numerose scuole che testimoniano, secondo Hausman, unevidente non omogeneit tra le
varie posizioni: si osservano varie correnti, tra cui la scuola austriaca, quella istituzionalista,
quella marxiana, le dottrine socioeconomiche, quelle comportamentiste, postKeynesiane,
neoRicardiane, la scuola di Chicago, la neoistituzionalista. Se tale frammentazione ha un
interesse specifico per il filosofo, le principali implicazioni epistemologiche che lo riguardano
concernono, in particolare, il dibattito sulla supposta neutralit delleconomia nei confronti
della politica, il problema della plausibilit del vocabolario mentalistico mutuato dalla folk
psychology per le spiegazioni di tipo causale riguardanti le scelte individuali, il problema del
naturalismo (leconomia desidera diventare scienza naturale, ma, allo stesso tempo, i testi
economici sono pieni di teoremi dimostrati deduttivamente) leccessiva idealizzazione e
astrazione che non tiene conto delle interferenze di natura sociale e culturale, e, infine, il
rapportochelanozionedicausalitintrattieneconlastatisticaelametodologia.
IlrifiutodellapropostadiMilldilegittimarelautonomiadelleconomiaattraversoilricorso
allintrospezione e al metodo deduttivo, ha stimolato numerose ricerche che hanno
evidenziatocomeifenomenieconomicisianoinfluenzatidamolteplicifattori(sociali,culturali,

Humana.MenteIssue10July2009

142

comportamentali, politici) estrinseci alle teorie economiche stesse e non ancora


adeguatamente considerati. In questa luce, le teorie economiche non sono suscettibili di
essere messe alla prova dellesperienza come qualsiasi altra teoria scientifica, a causa di
fenomeni di interferenza che comportano numerose assunzioni di sostegno necessarie per
derivare implicazioni opportunamente falsificabili. I tentativi, infine, da parte di Rosenberg e
McCloskey, di rinunciare al potere predittivo delle teorie economiche insieme al rifiuto delle
questionimetodologiche,picheoffrireconcretealternativealletesidiMill,hannoaffermato,
negandola, linsolubilit del problema posto dal filosofo britannico. Celato dietro nuove
declinazioni teoretiche, resta, dunque, ancora irrisolto il Mills problem, e, cio, la difficolt
diconciliarelepremessedelleconomiaconimetodieglistrumentidellescienzeempiriche.


143

Book Review
Tbe New Spirit of Capitalism
Luc 8olLansk, Lve Chlapello
verso, London & new ?ork, 2007, 636 pp.

Ceotqe noll


1hls hlsLorlcal, soclologlcal and concepLual sLudy, flrsL publlshed ln
lrance ln 1999, Lakes as lnsplraLlon Lhe soclal slLuaLlon of Lhe
lndusLrlallsed WesL ln Lhe mld-1990s. CaplLallsm was flourlshlng:
reLurns on lnvesLed caplLal were conslderably hlgher Lhan ln Lhe
early 80s. Cn Lhe oLher hand, unemploymenL, lnequallLles and
homelessness were also on Lhe rlse. ?eL, whereas comparable
clrcumsLances ln Lhe lnLer-war years and ln Lhe 60s had provoked a
concerLed and redoubLable crlLlclsm of Lhe prevalllng economlc
order, ln Lhe 90s crlLlclsm had largely glven way Lo an economlc or
uarwlnlsLlc faLallsm - a rueful (or enLhuslasLlc) accepLance LhaL 'Lhe
laws of Lhe markeL' were lndomlLable. 1hls dlagnosls of a crlsls of
crlLlclsm ls, desplLe varlous knee-[erk reacLlons Lo Lhe presenL
flnanclal dlfflculLles, as perLlnenL Lo Lhe slLuaLlon ln WesLern
Lurope and Lhe u.S. Loday as lL was Len years ago Lo Lhe lrench
case on whlch 8olLanskl and Chlapello concenLraLe Lhelr aLLenLlon.
1he presenL crlsls of crlLlclsm, Lhe auLhors argue, musL be undersLood ln connecLlon wlLh a
prevlous crlsls of caplLallsm, or, more preclsely, a crlsls of a prevlous spltlt of caplLallsm, whlch
erupLed mosL vlslbly ln Lhe proLesL movemenLs of 1968. CaplLallsm, undersLood as 'an
lmpetotlve to oollmlteJ occomolotloo of copltol by fotmolly peocefol meoos' (p. 4), alLhough lL
lncludes a normaLlve, or [usLlfylng, componenL ln lLs deflnlLlon, cannoL by lLself moLlvaLe
people Lo parLlclpaLe acLlvely - wheLher as lnvesLors or as wage-workers - ln Lhe process of
accumulaLlon. Cn Lhe conLrary, 8olLanskl and Chlapello argue LhaL Lhe rlse ln Lhe sulclde raLe
from 22.9/100,000 ln 1977 Lo 31.6/100,000 ln 1994 (p. 423) can be aLLrlbuLed Lo Lhe 'anomle'
whlch resulLs when people are dlrecLly exposed Lo Lhe caplLallsL lmperaLlve wlLhouL an
lnLermedlary 'splrlL'. ueslre for essenLlal goods, or even luxurles, ls also noL enough, Lhe
auLhors clalm, Lo moLlvaLe Lhe klnd of engagemenL wlLh one's work whlch ls requlred ln
modern producLlon and servlces. So caplLallsm needs a moLlvaLlng splrlL, and musL lmporL Lhls
splrlL from Lhe area of soclal reallLy ouLslde Lhe sphere of accumulaLlon.
1he splrlL of caplLallsm aL a glven polnL ln Llme can be called an ldeology, buL Lhls should noL
be Laken Lo lmply LhaL lL ls an lnevlLable eplphenomenal accompanlmenL of Lhe economlc
'basls', LhaL lL ls necessarlly a false represenLaLlon of reallLy, or LhaL one soclal group
consclously uses lL ln order Lo manlpulaLe anoLher. 8aLher, Lhe splrlL of caplLallsm musL draw
on normaLlve orders, or- ln Lhe Lermlnology of 8olLanskl and 1hevenoL's earller book, Oo
Iostlflcotloo - 'clLles', whlch are noL only endorsed as correcL sLandards of [usLlce across
dlfferenL groups ln Lhe socleLy ln quesLlon, buL furLhermore form Lhe crlLerla for success ln
lnsLlLuLlonallsed 'LesLs', whlch as a consequence counL as 'leglLlmaLe LesLs' raLher Lhan bare
'LesLs of sLrengLh' (p. 31). ln Lurn, because Lhese LesLs Lake a concreLe, lnsLlLuLlonal form, Lhey
have a - parLlally consLralnlng - lmpacL on Lhe process of accumulaLlon.
1hls book's lnLrlgulng Lhesls ls LhaL caplLallsm has evolved over Lhe lasL cenLury noL [usL ln
response Lo lnLernal pressures (efflclency, compeLlLlon) and Lechnologlcal advances, buL also

Bumana.Nente - Issue 1u - }uly 2uu9

144

by lncorporaLlng large sLrands of Lhe crlLlclsm Lo whlch lL has been sub[ecL lnLo lLs own
moLlvaLlng splrlL. 1lme and agaln, crlLlclsm of Lhe caplLallsL lmperaLlve lLself for lLs
unaccepLable consequences, almlng Lo overLhrow or abollsh Lhe caplLallsL economlc sysLem, ls
relnLerpreLed as a [usLlflcaLlon for a new form of... caplLallsm.
1o subsLanLlaLe Lhelr clalms Lhe auLhors Lurn ln Lhe flrsL place Lo Lhe managemenL llLeraLure
of Lhe 1960s and 1990s, whlch, ln prescrlblng ways ln whlch a flrm should be run, provldes
managers noL only wlLh Lhelr own moLlvaLlons buL also wlLh [usLlflcaLlons Lhey can use when
explalnlng feaLures of organlsaLlon Lo Lhe workforce. ln subsequenL hlsLorlcal chapLers lL ls
shown LhaL Lhe 'LesLs' lnLroduced by flrms and by governmenL leglslaLlon progresslvely
embodled Lhe 'splrlLs' advocaLed by Lhe llLeraLure dlrecLed aL managers ln Lhese Lwo perlods.
1he splrlL of caplLallsm ln Lhe 60s deflned lLself ln opposlLlon Lo Lhe caplLallsm of Lhe flrsL
half of Lhe LwenLleLh cenLury. lLs deflnlng feaLures can be undersLood as an lncorporaLlon of
Lhe subsLanLlal crlLlclsm dlrecLed aL caplLallsm ln Lhe lnLer-war years by Luropean soclallsL and
communlsL parLles, as well as Lhe Lrade unlons. 1hls crlLlclsm was largely whaL 8olLanskl and
Chlapello call 'soclal crlLlque' - crlLlclsm of poverLy and lnequallLles ln socleLy, on Lhe one hand,
and, on Lhe oLher hand, crlLlclsm of egoLlsLlc explolLaLlon of one soclal group by anoLher.
Accordlng Lo Lhls splrlL of caplLallsm, proflL should be dlsLrlbuLed falrly accordlng Lo
quallflcaLlon and poslLlon ln Lhe flrm, as opposed Lo nepoLlsLlcally, as lL ofLen was ln earller,
famlly-based caplLallsm, and a [ob ln a large flrm should provlde a worker wlLh llfelong
securlLy. 1he splrlL of caplLallsm ln Lhese years had much ln common wlLh Lhe eLhos of Lhe
welfare sLaLe.
8uL lL ls ln opposlLlon Lo Lhls (second) splrlL of caplLallsm LhaL Lhe new (Lhlrd) splrlL of
caplLallsm deflnes lLself, and whereas Lhe former draws above all on Lhe soclal crlLlque of Lhe
lnLer-war years, Lhe laLLer draws predomlnanLly on whaL Lhe auLhors call 'arLlsLlc crlLlque',
whlch acqulred a formldable publlc volce for Lhe flrsL Llme ln 1968. ArLlsLlc crlLlque ls crlLlclsm
of 'lnauLhenLlc' acLlvlLles, emoLlons and forms of llfe, on Lhe one hand, and of oppresslon
resLrlcLlng Lhe lndlvlduals' freedom, auLonomy and creaLlvlLy, on Lhe oLher. lncorporaLlng
crlLlclsm of lndusLrlal caplLallsm as hlerarchlcal and dehumanlslng, Lransformlng workers lnLo
mere machlnes for carrylng ouL lnsLrucLlons, Lhe new splrlL of caplLallsm advocaLes self-
managemenL and versaLlllLy ln Lhe workplace and a new ldeal of Lhe manager as an lnsplrlng
'coach' or 'leader', who geLs Lhe workers Lo do whaL ls needed, noL by lssulng orders, buL by
converLlng Lhem Lo Lhe cause, so LhaL Lhe flrm's lnLeresLs become lnLeresLs.
1hls laLesL splrlL of caplLallsm can, argue 8olLanskl and Chlapello, be besL undersLood as
embodylng Lhe values of a new normaLlve order, 'Lhe pro[ecLual clLy', noL seen before, and noL
yeL fully arLlculaLed - leL alone lnsLlLuLlonallsed. 1hls clLy, drawlng lndlrecLly on neo-
nleLzschean phllosophy (ueleuze), lrench and Amerlcan soclology and Lhe language of
lnformaLlon Lechnology, acknowledges LhaL Lhe world of work ls now a 'connexlonlsL' world, a
mesh of lnLerllnklng neLworks ln whlch Lhe successful acLor ls Lhe one who has enough
Lransferable skllls, conLacLs and moblllLy repeaLedly Lo come ouL on Lop ln Lhls clLy's chlef LesL:
endlng one 'pro[ecL' and beglnnlng anoLher.
Whlle Lhe flexlblllLy prlzed ln Lhe new connexlonlsL world may - ln Lhe splrlL of a dlsarmed
arLlsLlc crlLlque - enhance Lhe auLonomy of rovlng managers, and cerLalnly enhances Lhe
revenue of lnvesLors on Lhe lnLernaLlonal flnanclal markeLs, Lhlngs are noL so rosy for much of
Lhe workforce on new flexlble conLracLs wlLh Lhelr concomlLanL lnsecurlLy. ln 1993 67 of
lrench workers on parL-Llme conLracLs had accepLed Lhese 'for wanL of someLhlng beLLer' (p.
238).
8uL Lhe same arLlsLlc crlLlque LhaL conLrlbuLed Lo formlng Lhe pro[ecLual clLy also helped Lo
generaLe Lhe crlsls of crlLlclsm, whlch ls Lhe book's sLarLlng polnL. lor one Lhlng, crlLlclsm of
Book Review - The new Spiiit of Capitalism

143

auLhorlLarlan hlerarchy was from 1968 onwards dlrecLed as much aL Lhe Lrade unlons and Lhe
.C.l. as aL lndusLry. lor anoLher, measures requlred Lo furLher auLonomy are lnLrlnslcally
llkely Lo dlmlnlsh securlLy, and vlce versa, so LhaL Lhere ls a concepLual Lendency for Lhe arLlsLlc
crlLlque and Lhe soclal crlLlque Lo undermlne one anoLher.
Powever, ln Lhe mosL fasclnaLlng parL of Lhelr sLudy Lhe auLhors ouLllne a model for how
Lhe LradlLlonal soclal crlLlclsm of explolLaLlon can be adapLed Lo Lhe connexlonlsL world,
crlLlclslng lL from wlLhln Lhe normaLlve framework of Lhe pro[ecLual clLy lLself. lL can be shown
LhaL acLors wlLh more flexlblllLy and moblllLy (mulLlnaLlonals, flnanclal flrms) rely for Lhelr
proflLablllLy on Lhe facL LhaL oLher acLors (sLaLes, local workforces) have less. 1hls facL can Lhen
ground clalms for more recognlLlon and remuneraLlon on Lhe parL of less moblle and versaLlle
acLors. 1he auLhors perhaps do noL make clear how much Lhe LradlLlonal model of explolLaLlon
by a proprleLor class of a worklng class can sLlll be broughL Lo bear on a connexlonlsL world, as
Lhey Lend Lo blur Lhe dlsLlncLlon beLween whaL klnd of crlLlclsm ls [usLlfled and whaL klnd of
crlLlclsm ls, or can mosL easlly be, lnsLlLuLlonallsed (unlons, laws).
1hls orlglnal and lmaglnaLlve book, whlch goes a long way Lowards maklng conLemporary
llfe lnLelllglble, ends wlLh a subLle and percepLlve dlscusslon of new ways ln whlch feaLures of
work and consumpLlon can be experlenced as lnauLhenLlc.

































Bumana.Nente - Issue 1u - }uly 2uu9



147

Critica della ragione economica.


Tre saggi di Kabneman, McFadden, Smitb
A cura dl MaLLeo MoLLerllnl e Masslmo laLLelll almarlnl
ll SagglaLore, Mllano, 2003

Albetto 8loozzl
alberLo.blnazzl[humana-menLe.lL

l recenLl svlluppl dl rlcerca ln economla cognlLlva e sperlmenLale,


ponendo come loro prlnclpale oggeLLo dl sLudl ll comporLamenLo
effeLLlvamenLe osservaLo dell'aLLore economlco, hanno lncrlnaLo
la flducla ln una concezlone dl razlonallLa asLraLLa e formale a
favore dl una vlslone problemaLlca del processl declslonall umanl
ln condlzlonl dl razlonallLa llmlLaLa. Cll arLlcoll raccolLl ln quesLo
volume - la verslone ln lLallano dl un lavoro dl Mcladden apparso
su !ournal of 8lsk and uncerLalnLy lnLlLolaLo kozlooollt pet
ecooomlstl e le Lraduzlonl delle Nobel lectotes dl kahneman e
SmlLh - raccolgono l conLrlbuLl dl Lre sLudlosl lmpegnaLl ln quesLo
nuovo seLLore dl rlcerca che lnLegra l meLodl classlcl dell'economla
con quelll Lradlzlonall della pslcologla cognlLlva e delle
neurosclenze. Come ben slnLeLlzzano MoLLerllnl e almarlnl nella
posLfazlone del volume, gll economlsLl classlcl assumono
(lmpllclLamenLe ed espllclLamenLe) l'eslsLenza dl leggl dell'economla rlspeLLo alle quall le
mlcrodevlazlonl lndlvlduall dovuLe a lsLanze pslcologlche" sono conslderaLe rumore dl fondo
dlreLLo ad auLocompensarsl a llvello macroeconomlco.
1
L proprlo l'evldenza sperlmenLale con l
suol numerosl rlsulLaLl che ha messo ln crlsl l prlnclpl LaclLamenLe assunLl dagll economlsLl.
Scopo del curaLorl e evldenzlare l'lmporLanza della relazlone dl complemenLarlLa Lra pslcologla
ed economla - capace dl lnLegrare le meLodologle Lradlzlonall ln economla aLLraverso l'uso
dell'esperlmenLo conLrollaLo ln laboraLorlo - e accogllere, cosl, la valenza pragmaLlca e
anLldogmaLlca dl quesLe rlcerche lnLerdlsclpllnarl, pur rlmanendo fedell all'lmposLazlone
orlglnarla dl kahneman e 1versky che conceplsce le Leorle cognlLlve non necessarlamenLe
alLernaLlve a quelle classlche.

La Leorla comporLamenLale delle declslonl (behavloral declslon Lheory) - osserva Mcladden -
ha orlglne con ll lavoro dl von neumann e MorgensLern sulle scelLe ln condlzlonl dl lncerLezza e
con la Leorla del glochl. L'approcclo maLemaLlco ha avuLo ll preglo dl rendere appeLlblle
l'anallsl formale e asslomaLlca ln economla e pslcologla" lncoragglando la sperlmenLazlone ln
laboraLorlo per conLrollare la valldlLa descrlLLlva degll assloml.
2
Cll lmporLanLl conLrlbuLl delle
sclenze cognlLlve hanno recenLemenLe evldenzlaLo come l processl declslonall umanl slano
lnfluenzaLl da elemenLl dl naLura pslcologlca quall percezlonl, credenze, aLLegglamenLl, gludlzl.
1all rlsulLaLl hanno messo ln crlsl ll modello dell'uomo dl Chlcago, un agenLe che sceglle con

1
ctltlco Jello toqlooe ecooomlco. 1te soqql. koboemoo, McloJJeo, 5mltb, a cura dl MaLLeo almarlnl e
Masslmo laLLelll almarlnl, ll SagglaLore, Mllano 2003, pp.236.
2
lvl, pp.27.

Bumana.Nente - Issue 1u - }uly 2uu9

148

l'unlco flne dl masslmlzzare preferenze razlonall, prlmlLlve e coerenLl, unlformandosl a rlgorosl
prlnclpl sLaLlsLlcl bayeslanl (una specle ln vla dl esLlnzlone, come l'ha deflnlLa l'auLore).
1ra l lavorl degll sclenzlaLl cognlLlvl, le rlcerche dl kahneman e 1versky hanno affasclnaLo e
sgomenLaLo allo sLesso Lempo gll economlsLl. l loro rlsulLaLl evldenzlano delle lncongruenze Lra
ll comporLamenLo osservaLo e ll modello dell'uomo dl Chlcago: le prlnclpall dlscrepanze
concernono un'elevaLa varlablllLa nell'uLlllzzo dell'lnformazlone, l'uso dl eurlsLlche, l'lncapaclLa
dl masslmlzzare le preferenze, la senslblllLa al conLesLo, la vlolazlone degll assloml della scelLa
razlonale, deformazlonl perceLLlve, effeLLl dl dlssonanza cognlLlva. Le nuove evldenze
sperlmenLall, osserva Mcladden, convergono verso una concluslone molLo severa: la falslLa del
modello dell'uomo dl Chlcago a causa dl lrregolarlLa cognlLlve dovuLe a effeLLl dl conLesLo, dl
punLo dl rlferlmenLo, dl dlsponlblllLa, dl supersLlzlone, dl processo, dl prolezlone. CuesLl
rlsulLaLl dlmosLrano come gll esserl umanl non rlescano a reperlre e a elaborare le
lnformazlonl ln modo coerenLe e cl genera una varleLa dl anomalle, compreso ll
comporLamenLo che rende l consumaLorl vulnerablll a essere sfruLLaLl dal mercaLl".
3
Agll
economlsLl - conclude Mcladden - non resLa che far evolvere l'uomo dl Chlcago nella
dlrezlone dell'uomo dl kahneman-1versky" adoLLando, dl conseguenza, lpoLesl dl razlonallLa
plu blande dl quelle classlche e modlflcando l'anallsl economlca ln funzlone dl quesLe.
4


ll sagglo dl kahneman lnLlLolaLo Moppe Jl kozlooollt llmltoto. loJoqloe sol qloJlzl e le scelte
lotoltlve lllusLra l rlsulLaLl dl una lunga collaborazlone con Amos 1versky (1937-1996) su Leml dl
pslcologla delle credenze e anallsl delle scelLe ln condlzlone dl lncerLezza. 1ra gll argomenLl
affronLaLl, eurlsLlche e dlsLorslonl del gludlzlo, effeLLl dl ftomloq, dl sallenza e ancoragglo,
Leorla del prospeLLo, eurlsLlche proLoLlplche, llluslonl cognlLlve. L'ldea cenLrale sosLenuLa dallo
pslcologo lsraellano e che l penslerl dlvergono ln una dlmenslone dl accesslblllLa manlfesLa nel
dlfferenLe funzlonamenLo del processl menLall dl Llpo lnLulLlvo e dellberaLo. un ruolo
prlvlleglaLo vlene aLLrlbulLo alle lnLulzlonl: secondo kahneman, le lnLulzlonl rappresenLano una
modallLa dl elaborazlone dell'lnformazlone lnLermedla Lra l due poll cosLlLulLl dalle operazlonl
auLomaLlche della percezlone e quelle dellberaLe del raglonamenLo. 1ale generale dlsLlnzlone
lndlvldua due Llpl dl slsLeml (1, 2) dl funzlonamenLo globale della menLe umana.
A un esLremo Lrovlamo ll slsLema perceLLlvo e lnLulLlvo (slsLema 1) le cul operazlonl sono
auLomaLlche, raplde e poco cosLose da un punLo dl vlsLa compuLazlonale. All'opposLo, le
rlsposLe del slsLema 2 sono dellberaLe, lenLe, serlall, cosLose ln Lermlnl dl carlco cognlLlvo e, dl
norma, le persone le lnLraprendono solo se hanno una qualche preclsa raglone per farlo".
3
L
complLo dello sclenzlaLo cognlLlvo, secondo kahneman, mosLrare come molLepllcl condlzlonl dl
scelLa slano lnfluenzaLe noLevolmenLe dal slsLema 1. Cll effeLLl gesLalLlcl osservaLl nell'amblLo
della pslcologla della percezlone, lnfaLLl, sono ln azlone anche ln presenza dl slLuazlonl
cognlLlve che lncludono l'elaborazlone dl sLlmoll plu asLraLLl. Lcco, slnLeLlcamenLe, alcunl
rlsulLaLl sperlmenLall: gll effeLLl dl ftomloq nel processo declslonale e nella rlsoluzlone del
probleml sl manlfesLano quando dlfferenLl descrlzlonl dello sLesso complLo evldenzlano una
noLevole dlscrepanza Lra l rlsulLaLl, la Leorla del prospeLLo lndlca che le varlazlonl e le
dlfferenze sono molLo plu accesslblll del llvelll assoluLl, le eurlsLlche del gludlzlo offrono una
cornlce lnLerpreLaLlva per molLl errorl slsLemaLlcl nelle credenze e nelle preferenze. 1all effeLLl
documenLano un forLe allonLanamenLo dalla loglca esLenslonale-normaLlva della credenza e
della scelLa: la sosLlLuzlone dl aLLrlbuLl proLoLlplcl ad aLLrlbuLl esLenslonall sembra essere una

3
lvl, pp.64.
4
lvl, pp.66.
3
lvl, pp.84.
Recensione - Ciitica uella iagione economica

149

caraLLerlsLlca lncompaLlblle sla con l prlnclpl bayeslanl, sla con l'assloma della masslmlzzazlone
dell'uLlllLa.

vernon SmlLh, nel suo lnLervenLo lnLlLolaLo kozlooollt costtottlvlsto e tozlooollt ecoloqlco
osserva come la rlcerca nel campo della pslcologla economlca abbla porLaLo alla luce molLl
esempl ln cul venlvano vlolaLe le lpoLesl dl razlonallLa del modello sLandard o cosLruLLlvlsLa.
LredlLaLo da uescarLes, Pobbes, 8acone, quesLo assume che gll agenLl possledano una
compleLa lnformazlone sul guadagnl e dlmenLlca l'lmporLanza del ruolo del conLesLo e della
memorla auLoblograflca degll aLLorl economlcl. Secondo SmlLh, appare plu opporLuno
muoversl verso ll conceLLo dl razlonallLa ecologlca, una razlonallLa che emerge da processl
evoluzlonlsLlcl, culLurall e blologlcl. ll modello cosLruLLlvlsLa, lnfaLLl, per mezzo della Leorla
razlonale, rappresenLa una slLuazlone economlca LramlLe un albero dl gloco lnLeraLLlvo
asLraLLo. Al conLrarlo, ll conceLLo ecologlco dl razlonallLa sl lnLerroga sulla naLura della sLruLLura
formale delle slLuazlonl rappresenLaLe. Cuall praLlche soclall sl presLano a essere splegaLe da
quesLl modelll e non alLre? SmlLh accoglle, lnolLre, le poLenzlallLa offerLe dagll esperlmenLl ln
laboraLorlo, dove e posslblle cosLrulre razlonalmenLe del conLrofaLLuall, come nella sLorla
economlca" per verlflcare ed esamlnare la valldlLa dl quelll che sono conslderaLl Leoreml
dlmosLraLl per vla deduLLlva.
6
ll suo gludlzlo flnale sul modello sLandard dl razlonallLa e assal
crlLlco: l'adeslone alla Lesl cosLruLLlvlsLa, olLre a essere prlva dl valldlLa emplrlca, lmpllca, ln
manlera lmmoLlvaLa, che ognl agenLe sla un cosLruLLlvlsLa esaLLamenLe nello sLesso senso ln
cul lo slamo nol ln quanLo Leorlcl".
7
ua qul, allora, occorre rlparLlre lnLegrando la rlcerca
classlca ln Leorla economlca con l conLrlbuLl offerLl dall'economla cognlLlva e sperlmenLale.

























6
lvl, pp.130.
7
lvl, pp.139.

Bumana.Nente - Issue 1u - }uly 2uu9


131

Book Review
Tbe Culture of tbe New Capitalism
8lchard SenneLL
?ale unlverslLy ress, new Paven and London, 2006

Motco 5olloos
marcosollnas[Lln.lL


8uL Lhe flexlblllLy Lhey celebraLe does noL glve, lL cannoL glve, any
guldance for Lhe conducL of an ordlnary llfe. 1he new masLers have
re[ecLed careers ln Lhe old Lngllsh sense of Lhe word, as paLhways
along whlch people can Lravel, durable and susLalned paLhs of
acLlon are forelgn LerrlLorles." 1hls was Lhe blLLer concluslon of Lhe
nearly classlc 1be cottosloo of cbotoctet, prlnLed ln 1998 (1be
cottosloo, p. 147). AfLer Lhe more emlnenLly eLhlcal-moral
lnLerlude of kespect lo oo Aqe of loepoollty (2003), 8lchard SenneLL
Look up agaln where he had lefL off ln 1be coltote of tbe New
copltollsm (whlch was glven as Lhe CasLle LecLures aL ?ale
unlverslLy ln 2004), he agaln brlngs up for dlscusslon Lhe
consequences of Lhe new lnsLlLuLlons' flexlblllLy: 1he fragmenLlng
of blg lnsLlLuLlons has lefL many people's llves ln a fragmenLed
sLaLe: Lhe places Lhey work more resembllng Lraln sLaLlons Lhan
vlllages, as famlly llfe ls dlsorlenLed by Lhe demands of work" (1be coltote, p. 2). Pe also beglns
Lo oppose Lhls culLure wlLh Lhe ldeal of crafLsmanshlp" agalnsL whlch Lhe emerglng soclal
order mlllLaLes, as celebraLed ln ctoftsmoo, publlshed lasL year (1be coltote, pp. 4, 194-197,
103 ff.).
8uL leL us geL back Lo Lhe effecL of Lhe new lnsLlLuLlons. 1he flrsL alm of 1be cottosloo was
Lo show Lhe personal consequences of work ln Lhe new caplLallsm", l.e. LhaL Lhe condlLlons
of Lhe new economy feed lnsLead on experlence whlch drlfLs ln Llme, from place Lo place, from
[ob Lo [ob. [.] shorL-Lerm caplLallsm LhreaLens Lo corrode [.] characLer, parLlcularly Lhose
quallLles of characLer whlch blnd human belngs Lo one anoLher and furnlshes each wlLh a
sense of susLalnable self" (1be cottosloo, pp. 26-27). SenneLL Lakes up hls dlscourse agaln from
Lhls polnL: Lhe culLural ldeal requlred ln new lnsLlLuLlons damages many of Lhe people who
lnhablL Lhem" (1be coltote, p. 3). laylng a ma[or parL ln boLh books ls Lhe facL LhaL Lhe flexlble
lnsLlLuLlons, flrsL of all enLerprlses and Lhe markeL, no longer acL ln Lhe long buL Lhe shorL Lerm,
and Lhls LransformaLlon lnvolves boLh self-employed professlonals LhaL work ln Lhe sphere of
Lhe new economy as well as employees. lf, as a resulL, some analyses come Lo overlap, Lhe
sphere ls noneLheless exLended as now Lhe welfare lnsLlLuLlons and Lhe publlc realm are also
consldered. 1hls exLenslon corresponds Lo Lhe adopLlon of an enlarged perspecLlve: Lhe
analysls alms Lo skeLch ouL Lhe culLural ldeal and Lhe embodlmenL of Lhe ldeal man, demanded
and promoLed by Lhe lnsLlLuLlons of new caplLallsm. ln shorL, aL Lhe core ls Lhe new culLural
model" or Lhe culLure of Lhe new caplLallsm" and lLs effecL on socleLy (pp. 3-8). And Lhls
provldes Lhe sLrucLure for Lhe volume: analysls of Lhe LransformaLlons ln Lhe lnsLlLuLlons
broughL abouL by Lhe shorLened framework of Llme, developed ln Lhe flrsL long chapLer
(8ureaucracy"). 1he relaLlon beLween fears abouL belng made redundanL or lefL behlnd and
Lhe LalenL requlred by Lhe skllled socleLy", l.e. Lhe relaLlon beLween work, unemploymenL and
8llJooq ln Lhe second chapLer (1alenL and Lhe SpecLer of uselessness"). 1he relaLlon beLween

Bumana.Nente - Issue 1u - }uly 2uu9

132

consumer behavlour and pollLlcal aLLlLudes ln Lhe Lhlrd chapLer (Consumlng ollLlcs"). llnally,
a brlef concluslon LhaL alms Lo ouLllne some publlc remedles for Lhe llls of Lhe new lnsLlLuLlonal
model (Soclal CaplLallsm ln our 1lme").
8eyond a cerLaln dls[olnLedness, one of Lhe Lhreads LhaL seems Lo me Lo unlfy Lhe analysls
of Lhe above menLloned spheres ls Lhe crlLlcal sLaLe of new caplLallsm's culLural model: lLs
negaLlve effecL on dlfferenL levels. 1he analysls lndeed alms Lo show LhaL Lhe new model noL
only produces anxleLy and sLress ln Lhe worker of Lhe flexlble enLerprlses (pp. 32, 63-68, 127,
181), buL also presenLs some deflclLs and malfuncLlons ln Lhe producLlveness and organlzaLlon
of Lhose enLerprlses, whlch ofLen prove Lhelr undolng" (pp. 43, 64-71, 136), as well as a
scourge on Lhe welfare lnsLlLuLlons and Lhe pollLlcal models (pp. 163 ff., 173-178). And lL ls Lhls
Lransversal negaLlvlLy whlch could perhaps be glven furLher examlnaLlon. ln Lhe sense LhaL lf
SenneLL clearly wrlLes LhaL Lhere ls a neL class dlvlde beLween Lhose who proflL from Lhe new
economy and Lhose ln Lhe mlddle who do noL" (p. 131), and LhaL Lhe new sysLem brlngs a more
and more radlcal soclal and economlcal lnequallLy (34 ff.), lL neverLheless seems Lo me LhaL ln
Lhe analysls of Lhe funcLlonlng of Lhe new sysLem more emphasls could be puL on Lhe
dlfference and even sLruggle beLween Lhe dlfferenL lnLeresLs, needs and soclal expecLaLlons aL
sLake. llrsL of all beLween Lhe perspecLlve of Lhe workers below ellLe levels", lncreaslngly
sub[ugaLed Lo Lhe new labour markeL mechanlsm, l.e. Lo Lhe process of casuallzaLlon" of Lhe
labour force (pp. 48-31, 61-62, 76-77), and Lhe perspecLlve of Lhe Lop execuLlves, consulLanLs
and more generally Lhe ellLe. Whlle Lhe new mechanlsms could ln some way have an anxleLy-
lnduclng effecL on Lhese workers, aL Lhe same Llme Lhey guaranLee hlgh lncomes and relaLlvely
good fuLure expecLaLlons (pp. 34, 80-81).
And so dlfferenL lnLeresLs come Lo lnLeracL ln varlous ways wlLh Lhe lnLeresLs of Lhe owners
and Lhe old- sLyle sLockholders, wlLh Lhelr alms Lo lncrease dlvldends and reduce labour cosLs
by recrulLlng deunlonlzed and cheap young workers (pp. 48-49, 93-97, 134), whlle also
sacrlflclng quallLy Lo achleve hlgh proflLs" (p. 106). ln lLs Lurn a perspecLlve dlfferenL from LhaL
of Lhe new" sLockholders, who lnsLead are looklng Lo make a proflL ln a shorL Llme (pp. 38-40).
ln brlef, lf Lhe lnnermosL confllcLs of Lhese perspecLlves were polnLed ouL ln a clearer and more
arLlculaLe way, Lhe negaLlve effecL of Lhe new model could perhaps be beLLer deflned, more
speclflc, and less Lransversal. lrom here one could Lhen also furLher examlne Lhe cruclal
correlaLed quesLlon of Lhe asserLlon of Lhe new model. WlLh regard Lo Lhls, SenneLL wrlLes LhaL
Lhe new culLural model" ls LhaL of Lhe new economy and LhaL, alLhough lL ls sLlll only a small
parL of Lhe whole economy [.] lL does exerL a profound moral and normaLlve force as a
cuLLlng-edge sLandard for how Lhe larger economy should evolve", lL has also a culLural
lnfluence far beyond lLs numbers", from a complemenLary vlewpolnL, and ln slmple Lerms:
Lhe avaLars of a parLlcular klnd of caplLallsm have persuaded so many people LhaL Lhelr way ls
Lhe way of Lhe fuLure" (pp. 10-12). Powever, no answers are glven Lo Lhe quesLlons abouL Lhe
way ln whlch Lhls persuaslon" has Laken place, nor are Lhere any reasons why and how Lhe
exLenslon of Lhe new values ls [so] broad", desplLe Lhelr negaLlve effecL (p. 182).
1o conclude, ln order Lo Lry Lo beLLer hlghllghL Lhe confllcLual characLer of Lhe dlfferenL
perspecLlves aL sLake and aL Lhe same Llme Lo Lry Lo lllumlnaLe Lhe process leadlng Lo Lhe
asserLlon of Lhe new model, lL seems Lo me LhaL lL may be useful Lo lnLerpreL Lhe worklng of
Lhe new order as Lhe exerclslng of a new form of hegemony. An approach LhaL could
conLrlbuLe boLh Lo examlnlng Lhe dlfferenL power relaLlons beLween Lhe soclal parLners aL
sLake and Lo analyslng Lhe processes of obLalnlng consenL. And an approach LhaL could
lnLegraLe ln LhaL crlLlcal mlnd-seL of eLhnographers" Lhrough whlch Lhe ldeologlcal
proposals", wrlLes SenneLL, reveal Lhelr eluslve characLer (p. 11). ln oLher words, l belleve LhaL
whaL could perhaps arlse from Lhls concepLual-meLhodologlcal lnLegraLlon ls a renewed form
Book Review - The Cultuie of new Capitalism

133

of crlLlclsm LhaL can unmask Lhose power mechanlsms underlylng Lhe hegemony exerclsed by
Lhose new masLers" LhaL celebraLe flexlblllLy and have re[ecLed careers ln Lhe old Lngllsh
sense of Lhe word" and whlch, ln splLe of Lhe negaLlve effecL of Lhe new order on Lhe mass of
clLlzens-workers, neverLheless ends up achlevlng wldespread consenL ln publlc llfe.













































Bumana.Nente - Issue 1u - }uly 2uu9





133

Book Review
Equilibrium in Economics: Scope and Limits
valerla Moslnl
8ouLledge, London, 2007

Clovoool coslol
glovannl.caslnl[humana-menLe.lL


1he noLlon of equlllbrlum plays a fundamenLal role ln economlc
sclence. ?eL, scholars ofLen adopL or refer Lo lL uncrlLlcally. 1hls
slLuaLlon led Lo Lhe flourlshlng of an ample debaLe ln conLemporary
economlcs abouL Lhe value of models based on Lhe concepL of
equlllbrlum. 1he book under revlew offers an ln-depLh analysls of
Lhls noLlon by lnvesLlgaLlng some lmporLanL quesLlons: Pow dld Lhe
noLlon of equlllbrlum enLer lnLo economlcs? Whlch meanlng(s) has
lL been used wlLh? WhaL klnd of (pasL, presenL, and, posslbly,
fuLure) role does and can lL play ln Lhls fleld?
1he volume conslsLs of a collecLlon of papers by auLhors comlng
from dlfferenL academlc backgrounds (economlsLs, malnly, buL also
hlsLorlans, soclologlsLs and phllosophers), who Lhereby address Lhe
lssue from dlfferenL polnLs of vlew. 1he work ls dlvlded lnLo Lhree
parLs. 1he flrsL Lwo of Lhem, '1he lnLerplay of equlllbrlum noLlons
beLween Lhe naLural sclences and economlcs' and 'Lqulllbrlum ln pre-neoclasslcal economlcs',
have an hlsLorlcal orlenLaLlon. 1hey lnvesLlgaLe, respecLlvely, Lhe emergence of boLh
LheoreLlcal and meLhodologlcal lnLerchanges beLween economlcs and naLural sclences and Lhe
lssue of how Lhe noLlon of equlllbrlum have been used and characLerlzed by some of Lhe
founders of economlcs. 1he lasL and rlchesL parL, 'Lqulllbrlum ln presenL-day economlc Lheory
and pracLlce', examlnes Lhe use of Lhe noLlon of equlllbrlum ln presenL economlc Lheory and
pracLlce. ln whaL follows, l shall dlscuss each of Lhese parLs ln more deLall.
1he Lhree papers ln Lhe flrsL parL address Lhe lssue of Lhe lnfluence of oLher sclenLlflc flelds
ln Lhe developmenL of Lhe noLlon of equlllbrlum ln economlcs. 1he paper by CraLLan-Culnness
deals wlLh Lhe LheoreLlcal lnLerchanges beLween mechanlcs and economlcs. More speclflcally,
lL analyses how Lhe noLlons of equlllbrlum, deflned ln mechanlcs, have lnfluenced Lhe work of
some of Lhe faLhers of marglnallsm, llke !evons, Walras and areLo. 1he paper shows how
economlcs lnherlLed many dlfferenL noLlons of equlllbrlum from mechanlcs and how
dlsproporLlonaLe Lhe lnfluence of such noLlons has been, glven Lhe very dlfferenL sclenLlflc
naLure of dlsclpllnes such economlcs, on Lhe one hand, and mechanlcs/physlcs, on Lhe oLher
hand. 1he second paper, by 8ensaude-vlncenL and Moslnl, focuses on Lhe LheoreLlcal lnLerplay
beLween chemlsLry and economlcs and analyses Lhe Lwo maln noLlons of equlllbrlum
developed ln chemlsLry by Lavolsler (a sLaLlc noLlon) and Le ChaLeller (a dynamlc noLlon) ,
respecLlvely. llrsL, Lhe auLhors sLress LhaL Lavolsler held LhaL Lhe same noLlons of balance
mechanlsms and equlllbrlum were generally aL play ln naLure, and hence presenL boLh ln
chemlsLry and ln economlcs. Secondly, Lhey analyze how Lhe noLlon of equlllbrlum developed
by Le ChaLeller has been LranslaLed lnLo economy by Samuelson - an example of a haphazard
use of analogy beLween Lwo radlcally dlfferenL dlsclpllnes. 1he Lhlrd paper, by !arvls and
Moslnl, looks aL Lhe lnLerchanges beLween blology and economy. lL analyses, on Lhe one hand,
Lhe noLlon of equlllbrlum as homeosLasls and, on Lhe oLher hand, a LheoreLlcal economlc-

Bumana.Nente - Issue 1u - }uly 2uu9

136

derlved noLlon of equlllbrlum assoclaLed wlLh evoluLlonlsm and connecLed Lo cosL-beneflL
analyses. 1o summarlze, Lhls flrsL parL ls cenLered on Lhe LheoreLlcal 'exchanges' beLween
economlcs and oLher dlsclpllnes. We are faced wlLh a rlch lnLerplay, whlch shows how Lhe
mulLlfarlous has been Lhe noLlon of equlllbrlum lnLroduced, slnce Lhe beglnnlng, ln economlcs.
All Lhese Lhree papers ralse lnLeresLlng quesLlons abouL Lhe value and Lhe correcL use of
analogy beLween dlfferenL flelds.
1he second parL, also consLlLuLed by Lhree papers, deepens Lhe analysls of how equlllbrlum-
connecLed noLlons have been LreaLed by pre-neoclasslcal economlsLs. 1he flrsL paper, by ulxon
and Wllson, shows LhaL Lhe selflsh, Pobbeslan, neoclasslcal economlc agenL cannoL be equaLed
Lo Lhe SmlLhlan economlc agenL. lndeed, SmlLh's noLlon of 'slmpaLhy', LogeLher wlLh Chalmers'
noLlon of 'characLer', and wlLh Lhe more recenL work of Mead, suggesLs a noLlon of economlc
agency LhaL denles Lhe neoclasslcal reducLlon of human agency Lo 'parLlcle-llke behavlour'. 1he
second paper, by van den 8erg, dedlcaLed Lo lsnard's work, shows how Lhe Lendency Lowards
an uLlllLy-maxlmlzaLlon approach was already mlLlgaLed, ln Lhe LhoughL of Lhls elghLeen-
cenLury auLhor, by Lhe appeal Lo dlmenslons of human agency oLher Lhan pure egolsm,
Lhrough Lhe emphasls placed on Lhe role of morallLy and 'vlrLuous hablLs' ln Lhe
characLerlzaLlon of raLlonal behavlor. llnally, Lhe paper by vaLln revlses Lhe work of CournoL,
sLresslng Lhe fundamenLal lnfluence of Llebnlz's LhoughL over hls ldeas.
1he Lhlrd and lasL parL, composed by seven papers, ls devoLed Lo Lhe analysls of Lhe role of
Lhe noLlon of equlllbrlum ln conLemporary economlc Lheory. A blg porLlon of Lhese papers ls
dedlcaLed Lo hlghllghLlng Lhe amblgulLles assoclaLed wlLh Lhe noLlon of equlllbrlum ln
economlcs and Lo proposlng posslble paLhs of dlsamblguaLlon ln order Lo resLore clarlLy ln Lhe
dlscusslon abouL equlllbrlum. 1he papers by Lawson, by Samuels and by Chlck are clear
examples. Lawson sLarLs by analyzlng Lhe amblguous role of Lhe noLlon of equlllbrlum ln
economlc Lheorlzlng. Pe dlsLlngulshes Lwo posslble approaches Lowards equlllbrlum, accordlng
Lo whlch such concepL refers, respecLlvely, Lo a real properLy of Lhe economlc sysLems (Lhe
'onLlc' approach), or slmply Lo a useful Lool ln Lhe lnvesLlgaLlon of such sysLems (Lhe 'eplsLemlc'
approach). Lawson belleves LhaL much of Lhe confuslon ln Lhe debaLe abouL equlllbrlum sLems
from conflaLlng Lhls dlsLlncLlon and confoundlng 'clalms abouL models and Lhelr properLles
wlLh properLles of Lhe reallLy LhaL Lhe models puLaLlvely alm Lo represenL'.
ln Lhelr papers, Samuels and Chlck Lry Lo dlsenLangle Lhe varlous aspecLs of Lhe noLlon of
equlllbrlum LhaL are mlxed ln conLemporary economlcs. Samuels hlghllghLs varlous
dlchoLomles, whlch dlfferenL characLerlzaLlons of equlllbrlum move from, focuslng on Lhe
approach adopLed Lowards equlllbrlum (seen elLher as an elemenL of reallLy or as a useful
modellng Lool) or on lLs role ln argumenLaLlve sLrucLures. Chlck proposes a classlflcaLlon of Lhe
many dlfferenL noLlons of equlllbrlum aL play ln economlcal Lheorlzlng and argues LhaL Lhe
equlllbrlum-conLroversy should be reconsldered wlLh respecL Lo a beLLer deflned noLlon of
equlllbrlum.
8ackhouse sLarLs from a sLrong lnsLrumenLallsL polnL of vlew abouL equlllbrlum: economlcs
ls flrsL of all a maLLer of problem-solvlng, and Lhe debaLe abouL Lhe role of equlllbrlum has Lo
be LreaLed case by case. 1he quesLlon wheLher Lhe noLlon of equlllbrlum ls useful or noL
depends prlmarlly on lLs role wlLhln models LhaL successfully LargeL parLlcular problems.
1he paper by 1homas dlscusses how economeLrlcs deals wlLh Lhe concepL of equlllbrlum,
concludlng LhaL Lhls speclflc fleld can elaboraLe lnLeresLlng economlc models LhaL do noL
employ such a noLlon.
lreeman polnLs Lo Lhe mulLlfarlous use of Lhe concepL of equlllbrlum ln economlcs, whlch
conLrasLs wlLh Lhe use of such concepL ln Lhe naLural sclences. lreeman also suggesLs LhaL, ln
Lhe former fleld, Lhe concepL of equlllbrlum ls used ln a dogmaLlc, qulLe rellglous, way.
Book Review - Equilibiium of Economics

137

llnally, sLarLlng from a represenLaLlve lnsLanLlaLlon of Lhe concepL of equlllbrlum ln
neoclasslcal economlcs, l.e., Lhe noLlon proposed by Lucas, uenls examlnes Lhe cenLral role
LhaL equlllbrlum galned ln Lhe neoclasslcal approach and argues LhaL Lhls led Lo overesLlmaLlng
Lhe sLaLlc analysls of Lhe sysLem aL Lhe expense of lLs dynamlc analysls. uenls concludes by
sLresslng Lhe need for Lhe fuLure recovery of a dlalecLlcal approach Lo Lhe sLudy of Lhe
economlc sysLem.
1hls book presenLs varled and rlch conLrlbuLlons, whlch look aL Lhe noLlon of equlllbrlum
from very dlfferenL polnLs of vlews, movlng from an hlsLorlcal Lo an eplsLemologlcal approach.
Surely, Lhls collecLlon achleves aL leasL one resulL: lL convlnces Lhe reader LhaL Lhe debaLe
abouL equlllbrlum ln economlcs needs Lo be re-assessed from lLs foundaLlons. Lqulllbrlum ls a
concepL LhaL has been lnLroduced ln economlcs under Lhe urge of varlous and dlfferenL
exLernal leverages. Such manlfold characLerlzaLlon keeps vlLlaLlng Lhe currenL usage of Lhe
noLlon. ln order Lo address properly Lhe quesLlon abouL Lhe role of equlllbrlum ln economlcs,
we need Lherefore Lo address Lhe prellmlnary quesLlon abouL Lhe meanlng of Lhe noLlon of
equlllbrlum lLself. 1he deflnlLlon and classlflcaLlon of Lhe varlous noLlons aL play, of Lhelr
eplsLemologlcal and funcLlonal role ln LheoreLlcal economlcs, ls Lhe flrsL sLep for a good grasp
of Lhe lssue.
AL Lhe same Llme, Lhe rlchness and varleLy of Lhe conLrlbuLlons appear also as Lhe maln
drawback of Lhe book. uesplLe an evldenL efforL of coordlnaLlng and lnLegraLlng Lhe papers,
Lhe presence of so many dlfferenL perspecLlves makes lL dlfflculL for Lhe reader Lo galn a
conslsLenL overvlew on Lhe lssue and Lo selecLlng a research rouLe approprlaLe for faclng Lhe
many unlocked problems.


1ab|e of Contents

v. Moslnl - lnLroducLlon: Lhree ways of looklng aL economlc equlllbrlum.

lAk1 l. 1be lotetploy of epolllbtlom ootloos betweeo tbe oototol scleoces ooJ ecooomlcs.
l. CraLLan-Culnness - Lqulllbrlum ln mechanlcs and Lhen ln economlcs, 1860-1920: a good
source for analogles?
8. 8ensaude-vlncenL and v. Moslnl - 8eLween economlcs and chemlsLry: Lavolsler's and Le
ChaLeller's noLlons of equlllbrlum.
L. !arvls and v. Moslnl - 1he ublqulLy of Lhe noLlon of equlllbrlum ln blology, and lLs relaLlon
wlLh equlllbrlum ln economlcs.

lAk1 ll. polllbtlom lo pte-oeoclosslcol ecooomlcs.
W. ulxon and u. Wllson - 'SympaLhy', 'characLer' and economlc equlllbrlum.
8. van deb 8erg - Lconomlc equlllbrlum ln Lhe lrench LnllghLenmenL: Lhe case of A.n. lsnard.
l. vaLln - lnfluences on Lhe economlc Lheory of A.A. CournoL: mechanlcs, physlcs and blology.

lAk1 lll. polllbtlom lo pteseot-Joy ecooomlc tbeoty ooJ ptoctlce.
1. Lawson - 1enslons ln modern economlcs: Lhe case of equlllbrlum analysls.
8. L. 8ackhouse - Lqulllbrlum and problem solvlng ln economlcs.
W. !. Samuels - Lqulllbrlum analysls: a mlddlebrow vlew.
!. 1homas - Lqulllbrlum ln economlcs, sLablllLy and sLaLlonarlLy ln economeLrlcs.
v. Chlck - Lqulllbrlum ln economlcs: some concepLs and conLroversles.
A. lreeman - Peavens above: whaL equlllbrlum means for economlcs.
A. uenls - 1he hyposLaLlsaLlon of Lhe concepL of equlllbrlum ln neoclasslcal economlcs.

Bumana.Nente - Issue 1u - }uly 2uu9



139

Etica ed economia: il rapporto possibile


A cura dl 8. Clovanola e l. 1oLaro
LM Ldlzlonl, adova, 2008

cbloto tbosl
chlara.erbosl[humana-menLe.lL

tlco eJ ecooomlo. ll toppotto posslblle e un llbro che raccoglle


scrlLLl dl varl auLorl. ll fllo conduLLore del volume e l'lndaglne del
rapporLo eslsLenLe Lra eLlca ed economla al nosLrl glornl. L'lnLeresse
per quesLo rapporLo nasce dalla volonLa dl lnLerrogarsl su alcunl
aspeLLl degll aLLuall meccanlsml economlcl, naLl sopraLLuLLo dalle
dlnamlche connesse alla loglca della globallzzazlone e dalla volonLa
dl provare a comprenderll. ln parLlcolare LuLLl l conLrlbuLl sono
accomunaLl dal LenLaLlvo dl lndlvlduazlone dl modelll dell'economla
rlspeLLosl della dlgnlLa della persona, dlgnlLa che spesso l'economla
dlmenLlca. Come lnfaLLl sosLlene 1oLaro nel suo arLlcolo dl aperLura
del llbro, la colpa dell'economla e non solo quella dl essersl lsolaLa
da LuLLl gll amblLl dello spazlo dell'umano ma anche dl preLendere dl
conLenerll LuLLl, con ll cosLanLe obleLLlvo dl assorblre e lnglobare
ognl aspeLLo della persona. L ln buona parLe sl pu dlre che cl sla rlusclLa. Cosl lnfaLLl l'auLore
descrlve ll comporLamenLo dell'uomo occldenLale, corroso a Lal punLo dalle vorLlcose
dlnamlche economlche da essere lncapace dl modlflcare comporLamenLl e aLLegglamenLl che
sLanno dlvenLando sempre plu perlcolosl e dannosl per ll nosLro planeLa:

erche ll nosLro modo dl vlvere come donne e uomlnl dell'CccldenLe cl conduce a rlLenercl degnl
dl conslderazlone e dl auLosLlma solo se slamo ln grado dl essere sempre plu produLLorl e
consumaLorl dl benl maLerlall ad alLo lnvesLlmenLo energeLlco? erche abblamo affldaLo la nosLra
lmmaglne dl umanl al possesso crescenLe dl benl lndlvlduall che, a flanco dl opporLunlLa dl scelLa
che dlpendono escluslvamenLe da nol, comporLano lo spreco o l'uso abnorme dl rlsorse che
apparLengono anche ad alLrl o sono dl rllevanza comune?
1


ll llbro dunque sl apre con l'arLlcolo dl lrancesco 1oLaro, let ooo mlsoto etlco-oottopoloqlco
Jellecooomlo, dedlcaLo ln parLlcolare alla quesLlone energeLlca ed ecologlca, dlvenLaLa cosl
urgenLe e pressanLe da chlamare l'economla ad una vera e raplda lnverslone dl Lendenza e ad
una correzlone. 1oLaro sl dlchlara a favore dl un conLenlmenLo dell'economla da aLLuarsl per
ln modo dlfferenzlaLo a seconda che rlguardl paesl dell'lpersvlluppo e quelll del soLLosvlluppo.
Se lnfaLLl e vero che, ln mancanza dl llllmlLaLe fonLl energeLlche, uno svlluppo del paesl
emergenLl ll quale rlcalcasse ll modello dl svlluppo persegulLo dal paesl rlcchl condurrebbe al
collasso ll planeLa Lerra
2
e vero anche che e molLo dlfflclle domandare al paesl che vogllono
usclre dalla poverLa una condoLLa vlrLuosa ln quanLo essl sl senLono dopplamenLe defraudaLl

1
l. 1oLaro, 8. Clovanola (a cura dl), tlco eJ ecooomlo. ll toppotto posslblle, Ldlzlonl Messaggero
adova, adova, 2008, p. 43.
2
lvl, p. 44.

Bumana.Nente - Issue 1u - }uly 2uu9

160

dal paesl rlcchl sla per la soLLrazlone dl maLerle prlme sublLa nel passaLo, sla per l'evenLuale
rlnuncla forzaLa ad uno svlluppo che ormal bussa anche alla loro porLa
3
.
ll secondo arLlcolo dl SebasLlano MaffeLLone, kespoosoblllt Jl lmpteso e sosteolblllt, sl
lncenLra sul conceLLo dl sosLenlblllLa come ll parameLro plu slgnlflcaLlvo della responsablllLa
soclale e come unlca posslblllLa conLro l'ossesslva rlcerca delle lmprese del proflLLo a breve
Lermlne.
ll Lema della sosLenlblllLa e cenLrale anche nell'arLlcolo dl Clan Lulgl 8rena, tlco Jlmpteso
tto ecooomlo eJ ecoloqlo, ln cul sl sosLlene che la responsablllLa per l'amblenLe non e mal solo
dell'lmpresa ma dell'lnLera socleLa che e chlamaLa ad oLLenere la cooperazlone dl LuLLl l popoll.
CuesLa per non pu che eslsLere su basl dl glusLlzla, rlducendo le drasLlche dlsuguagllanze
economlche. Sono esse lnfaLLl che

offrono la posslblllLa dl conLlnuare ad lnqulnare alle lmprese del paesl svlluppaLl, sempllcemenLe
sposLando l loro lmplanLl ln paesl che non possono permeLLersl ll lusso dl occuparsl
dell'amblenLe. L la slLuazlone dl emergenza del paesl ancora ln vla dl svlluppo ll auLorlzza
anzlLuLLo ad lndusLrlallzzarsl, senza curarsl dell'lnqulnamenLo: le persone vengono prlma
dell'amblenLe e devono essere anzlLuLLo superaLe le slLuazlone dl poverLa.
4

ll quarLo sagglo dl Pelen Alford, lmpostozlooe Jel lovoto oello ptospettlvo Jello lobotem
exetceos, e dedlcaLa alla vlslone del lavoro nella prospeLLlva dell'enclcllca lobotem exetceos.
L'auLrlce meLLe ln evldenza l'lmporLanza della dlmenslone soggeLLlva del lavoro che lnclude gll
aspeLLl morall e splrlLuall della persona.
ll sagglo dl SLefano Zamagnl, uemoctozlo, llbett posltlvo, svlloppo, e dedlcaLo al modello dl
democrazla dellberaLlva, la sola ln grado dl favorlre un processo dl democrazla economlca
grazle all'aLLuazlone dl un mercaLo plurale e ad una fondamenLale campagna culLurale
lncenLraLa sulla flgura del consumaLore eLlcamenLe responsablle.
ll llbro sl chlude con ll sagglo dl 8enedeLLa Clovanola, kozlooollt etlco, tozlooollt
ecooomlco e tlccbezzo oottopoloqlco, che lndaga l fondamenLl eLlcl ed anLropologlcl
dell'economla.
ur dunque nella dlverslLa del Leml affronLaLl, ll llbro cerca dl mosLrare l'lnsosLenlblllLa del
modelll economlcl aLLuall che conLlnuano ad accenLuare dlsuguagllanze e confllLLl e che sLanno
dlvenLando sempre plu deleLerl per l'amblenLe e per ll planeLa, lncapace dl far fronLe al vorLlce
economlco che porLa lnLrlnsecamenLe con se lnqulnamenLo dell'aLmosfera, rarefazlone delle
rlsorse ldrlche, dlsLruzlone delle foresLe e del paLrlmonlo naLurale. ua una parLe, lnfaLLl,
aumenLa la popolazlone mondlale, aumenLano l consuml anche da parLe del paesl ln vla dl
svlluppo e sl dlffonde una Lecnologla la cul poLenza ha gla porLaLo l paesl plu avanzaLl a uno
sfruLLamenLo lnvaslvo e dlsLruLLlvo delle rlsorse naLurall
3
, dall'alLra lnvece la naLura non
rlesce a smalLlre e a rlclclare ln Lempo le rlsorse consumaLe. CuesLa slLuazlone che dlvenLera
nel Lempo sempre plu drammaLlca rlchlede l'lnLervenLo e la moblllLazlone dl LuLLl. L lnfaLLl gll
auLorl concordano LuLLl nel sosLenere che un'lnverslone dl Lendenza dell'economla sl pu
avere solo con un lnLervenLo programmaLo e condlvlso a ragglo mondlale e con plu
consapevolezza e cosclenza del comporLamenLl quoLldlanl dl LuLLl gll lndlvldul, ln parLlcolare dl
quelll del paesl rlcchl.


3
lvl, p. 43.
4
lvl, p. 88.
3
lvl, p. 80.

Recensione - Etica eu economia: il iappoito possibile

161


lnLroduzlone (lrancesco 1oLaro - 8enedeLLa Clovanola)
er una mlsura eLlco-anLropologlca dell'economla (lrancesco 1oLaro)
8esponsablllLa d'lmpresa e sosLenlblllLa (SebasLlano MaffeLLone)
LLlca d'lmpresa Lra economla ed ecologla (Clan Lulgl 8rena)
lmposLazlone del lavoro nella prospeLLlva della lobotem xetceos (Pelen Alford)
uemocrazla, llberLa poslLlva, svlluppo (SLefano Zamagnl)
8azlonallLa eLlca, razlonallLa economlca e rlcchezza anLropologlca (8enedeLLa Clovanola)



























Bumana.Nente - Issue 1u - }uly 2uu9



163

Economia e concezione dell'uomo


AA.vv. - a cura dl . Crasselll
lranco Angell, Mllano, 2007

5clllo 8elloccl
scllla.belluccl[humana-menLe.lL


ll volume curaLo da lerlulgl Crasselll raccoglle le relazlonl
presenLaLe da alcunl docenLl
1
duranLe una ClornaLa dl SLudlo
LenuLasl presso la lacolLa dl Lconomla dell'unlverslLa degll SLudl dl
erugla ll 22 novembre 2006. Suddlvlso LemaLlcamenLe ln Lre
parLl
2
, offre un amplo venLagllo dl prospeLLlve rlguardo al penslero
economlco conLemporaneo. CuesLe vengono affronLaLe, cosl come
e posslblle lnLulre dalla parLlzlone del volume, sla da un punLo dl
vlsLa Lecnlco (azlendale e glurldlco) sla da una poslzlone plu
speculaLlva e Leorlca, preLLamenLe economlca, da cul provengono
la magglor parLe del conLrlbuLl.
lmmedlaLamenLe evldenLl sono due caraLLerlsLlche che
apparLengono alla molLepllclLa del saggl: l'urgenza dl rlpensare l
fondamenLl della scuola economlca molostteom, ossla quella
neoclasslca, e la presa ln esame della muLaLa realLa conLemporanea che rende necessarlo Lale
rlpensamenLo. La correnLe molostteom sl fonda, lnfaLLl, su una concezlone dell'aLLore
economlco, che vlene usualmenLe chlamaLo bomo oecooomlcos
3
, le cul pecullarlLa conslsLono
ln un aLLegglamenLo egolsLa ed edonlsLa volLo alla masslmlzzazlone dell'uLlllLa o del proflLLo, a
seconda che sl LraLLl dl un consumaLore o dl un produLLore. negll ALLl qul pubbllcaLl sl hanno,
lnvece, molLe proposLe dl allonLanamenLo da quesLa conflgurazlone che sl e rlvelaLa fallace
come base dl elaborazlone dl Leorle e sLraLegle. Lssa, lnfaLLl, non Llene conLo del
comporLamenLl effeLLlvl che vengono condlzlonaLl dall'emoLlvlLa, dalla paura, dalla morale
lndlvlduale e da alLrl faLLorl 'devlanLl' che orlenLano le scelLe ln una quanLlLa dl dlrezlonl non
sempre, o megllo, quasl mal, razlonalmenLe glusLlflcablll dalla loglca del mero proflLLo
(Crasselll, Slgnorelll, vlsagglo)
4
.
Ma non solLanLo sul comporLamenLo del slngolo lndlvlduo sl concenLra la crlLlca alla
correnLe neoclasslca delle doLLrlne economlche. Lssa lnvesLe la sfera azlendale, anallzzando le
posslblll dlnamlche dl parLeclpazlone del lavoraLorl all'lmpresa (uvallc, MonLrone), la sfera
soclale, conLemplando la crescenLe rlchlesLa dl comporLamenLl 'eLlcl' nella flnanza (Moronl,
ClglloLLl) come nella gesLlone delle rlsorse pubbllche (Clannonl e Arcelll) e lnserendo nel

1
er l'elenco compleLo degll auLorl sl veda l'lndlce del volume rlporLaLo ln calce.
2
arLe prlma: nomo oecooomlcos, razlonallLa, reclproclLa. arLe seconda: 8esponsablllLa soclale, equlLa,
parLeclpazlone. arLe Lerza: 8esponsablllLa d'lmpresa, ldenLlLa lndlvlduale, ordlnamenLo glurldlco.
3
L probabllmenLe vllfredo areLo nel suo Manuale del 1906 ad avere uLlllzzaLo per prlmo l'espresslone
laLlna bomo oecooomlcos, la formulazlone del conceLLo corrlspondenLe e lnvece plu anLlca dal momenLo
che appare nel sagglo dl !ohn SLuarL Mlll Oo tbe uefloltloo of lolltlcol cooomy ooJ oo tbe MetboJ of
lovestlqotloo ltopet to lt del 1836". Marco 8occacclo ln cooomlo e coocezlooe Jell'oomo, lranco
Angell, 2007, p. 33.
4
1ra parenLesl sono lndlcaLl gll auLorl dl alcunl LesLl presenLl nel volume al cul conLenuLo sl fa
rlferlmenLo nell'lndlcazlone degll argomenLl LraLLaLl.

Bumana.Nente - Issue 1u - }uly 2uu9

164

dlbaLLlLo elemenLl asslologlcl e slmbollcl quall la graLulLa e ll dono (MonLesl). lnflne sl approda
anche al dlbaLLlLo glurldlco rlspeLLo al dlrlLLl e al doverl (8udelll, Clamplerl), al probleml legaLl
alla globallzzazlone del mercaLl per l paesl ln vla dl svlluppo (CerreLa) e alla gesLlone del benl
culLurall (Cosl).
ln concluslone pu essere opporLuno rlporLare un'osservazlone ln merlLo al ruolo della
formazlone cul sono depuLaLl l posslblll svlluppl dl una nuova doLLrlna economlca:

L'unlverslLa rappresenLa qulndl ll luogo prlvlleglaLo per creare culLure professlonall plu avanzaLe,
ancoraLe ed allmenLaLe da una consapevole dlmenslone eLlca ln grado dl gesLlre un slsLema dl
responsablllLa soclale.
3


refazlone - lerlulgl uaddl


resenLazlone: presenza e rlllevo delle concezlonl dell'uomo ln economla - .Crasselll

arLe prlma - nomo oecooomlcos, razlonallLa, reclproclLa.

1. Alle radlcl della sclenza economlca. Lconomla e concezlonl dell'uomo - C.Calzonl
2. nomo oecooomlcos e uomo eLlco: la sLrana coppla - M. 8occacclo
3. Sclenza economlca e caLegorla dl persona: noLa su un non rapporLo" - l. MusoLLl
3. 8revl noLe su alcune concezlonl dell'uomo ln economla - . Crasselll
4. 8azlonallLa e sclenza economlca - M. Slgnorelll
3. l paradossl della graLulLa ln economla - C. MonLesl
6. Lgolsmo vetsos alLrulsmo nella Leorla neoclasslca.
7. Cuall lmpllcazlonl dl pollLlca economlca? - M.vlsagglo
8. ClLre l'approcclo welfarlsLa: ll dlbaLLlLo degll economlsLl e l'evldenza per l'lLalla su
saluLe, equlLa e accesso al servlzl sanlLarl - M.Clannonl e L: Arcelll
9. L'uomo proLagonlsLa dell'agrlcolLura nell'epoca dell'economla
a. della conoscenza - 1. Sedlarl
10. La sempllclLa come punLo dl vlsLa unlflcanLe - M. agllaccl
11. Crlsl del penslero o llmlLe della modernlLa?
a. L'urgenza della quesLlone anLropologlca - M.Moschlnl
12. L'alLro: L'Lscluso - L. Lussana

arLe seconda - 8esponsablllLa soclale, equlLa, parLeclpazlone

13. ll flnallsmo dell'lmpresa Lra economla e soclallLa - A. MonLrone
14. LLlca e corporaLe governance: dalla leqol compllooce all' etblcol compllooce - C.
Chlrlelelson
13. ll modello parLeclpaLlvo - M. uvallc
16. La Leorla del glochl comporLamenLall: verso una rappresenLazlone reallsLlca del
comporLamenLl sLraLeglcl? - C. ueveLag
17. L'eLlca nel servlzl flnanzlarl - 8: Moronl
18. La flnanza eLlca, sLrumenLo dell'economla soclale - L. Llno

3
lerlulgl uaddl, ln cooomlo e coocezlooe Jell'oomo, lranco Angell, 2007, p.13.
Recensione - Economia e concezione uell'uomo

163

19. ll consumo dl prodoLLl eLlcl" e ll comporLamenLo del consumaLore responsablle - M.
ClglloLLl
20. La naLura della conoscenza - M. aoll
21. arLe Lerza - 8esponsablllLa d'lmpresa, ldenLlLa lndlvlduale, ordlnamenLo glurldlco
22. nomo oecooomlcos e lndlvlduo. ConLesLazlonl e mlnacce - 8.Chlarelll
23. Alla rlcerca della quadraLura del cerchlo": ll prlnclplo dl sussldlarleLa - S. 8udelll
24. Condlzlone degll lmmlgraLl ed aslmmeLrle
23. nella sfera del dlrlLLl e degll obbllghl - 8. Clamplerl
26. LqulLa conLraLLuale e mercaLo concorrenzlale - l. Scagllone
27. l codlcl eLlcl: proflll sLorlcl e funzlone - l. ascuccl
28. La nuova fronLlera del codlcl eLlcl. 8llevanza glurldlca aLLuale - M. CerreLa
29. ll nuovo accordo sul caplLale dl 8asllea 2, l'eLlca degll lmprendlLorl e l raLlng" nella
concesslone del credlLo - C. CaLLl
30. Sull' uomo glurldlco LrlbuLarlo" - M. verslgllonl
31. ualla conservazlone" pubbllca del benl sLorlco - arLlsLlcl alla regolazlone" e
valorlzzazlone" del benl ed evenLl culLurall - u. Cosl


































Bumana.Nente - Issue 1u - }uly 2uu9



167

Freakonomics. Il calcolo dell'incalcolabile


SLeven LevlLL, SLephen !. uubner
Sperllng & kupfer, Mllano, 2008

cbloto tbosl
chlara.erbosl[humana-menLe.lL


nell'esLaLe 2003 SLephen !. uubner, glornallsLa e scrlLLore, fu
lncarlcaLo dal new ?ork 1lmes dl lnLervlsLare SLephen u. LevlLL,
glovane e brlllanLe economlsLa dell'unlverslLa dl Chlcago e vlnclLore
del !ohn 8aLes Clark Medal, premlo assegnaLo al mlgllor glovane
economlsLa degll SLaLl unlLl soLLo l quaranL'annl. ua quesLo lnconLro
e dalla loro collaborazlone nasce lteokooomlcs, un llbro lnsollLo e
lnusuale che sl pone l'obleLLlvo dl raschlare la superflcle della vlLa
conLemporanea ln cerca delle sue dlnamlche plu profonde
1
.
L un llbro che parla dl economla, ma ll meLodo dl lndaglne
uLlllzzaLo dagll auLorl e assoluLamenLe esLraneo a quello accademlco
dl cul sl avvalgono gll economlsLl: sl servono lnfaLLl del meLodo
lnduLLlvo, ma basano le loro anallsl sclenLlflche su una vasLa raccolLa
dl daLl sLaLlsLlcl, apparenLemenLe svlncolaLl gll unl dagll alLrl e prlvl dl
relazlone Lra dl loro. uubner e LevlLL appllcano ll loro meLodo a realLa molLo lonLane dalle
quesLlonl economlche e spesso ne derlvano rlsposLe sorprendenLl, sconcerLanLl e molLo
orlglnall. Cll auLorl, lnfaLLl, cercano dl Lrovare moLlvazlonl che spleghlno perche la maggloranza
degll spacclaLorl dl droga vlva ancora ln casa con l genlLorl, perche ll llvello dl crlmlnallLa negll
SLaLl unlLl sla dlmlnulLo ln segulLo all'lnLroduzlone della legge sull'aborLo o come ll nome dl
baLLeslmo possa lnflulre sulla vlLa soclale dl una persona o esporla alla dlscrlmlnazlone razzlale.
Le argomenLazlonl porLaLe a sosLegno delle loro lpoLesl sl allonLanano qulndl dagll amblLl
Lradlzlonall dell'economla, per sfoclare nella soclologla, nella pslcologla e nella crlmlnologla.
lteokooomlcs affronLa qulndl Leml Lra loro dlverslsslml, parLendo per da alcunl punLl
ferml: gll lncenLlvl sono la pleLra angolare della vlLa moderna e saperll caplre e ll punLo
essenzlale per comprendere qualslasl problemaLlca, la saggezza convenzlonale e spesso faLLa
dl luoghl comunl che pur sembrando lnconLroverLlblll possono essere confuLaLl, effeLLl anche
speLLacolarl hanno spesso cause soLLlll e lonLane nel Lempo, l coslddeLLl esperLl, dal
crlmlnologo all'agenLe lmmoblllare, sfruLLano a proprlo uso e consumo ll vanLagglo lnformaLlvo
dl cul godono, sapere che cosa mlsurare rende ll mondo meno complesso dl quanLo non
appala e lmparando a sLudlare l daLl nel glusLo conLesLo sl possono splegare fenomenl
alLrlmenLl lncomprenslblll.

uleLro quesLe premesse c'e l'ldea che ognl cosa abbla un volLo nascosLo, che pu essere
esploraLo solo meLLendo ln crlsl ll senso comune e rlfluLando le splegazlonl plu ovvle. Come gla
deLLo, l casl che sl soffermano ad anallzzare non rlenLrano nel Leml Llplcl delle sclenze
economlche, bensl rlguardano fenomenl soclall e culLurall: dal meccanlsml lmmoblllarl al

1
S. u. LevlLL, e S. !. uubner, lteokooomlcs. ll colcolo Jelllocolcoloblle, Sperllng e kupfer, Mllano, 2003, p.
xxxlx.


Bumana.Nente - Issue 1u - }uly 2uu9

168

rapporLo genlLorl-flgll. Cll auLorl sl servono del plu soflsLlcaLl sLrumenLl anallLlcl che le sclenze
economlche poLessero offrlre
2
per caplre dlnamlche soclall che non sono mal sLaLe anallzzaLe,
per sLudlare nessl Lra cose che nessuno avrebbe mal poLuLo lmmaglnare, per calcolare
l'lncalcolablle - come dlce appunLo ll soLLoLlLolo del llbro.
Cla nell'lottoJozlooe gll auLorl mosLrano ll loro parLlcolare modo dl procedere, cercando dl
offrlrcl ll nesso Lra la rlduzlone della crlmlnallLa mlnorlle negll u.S.A degll annl '90 e
l'lnLroduzlone della legge sull'aborLo. negll annl '90, lnfaLLl, la crlmlnallLa mlnorlle avrebbe
dovuLo sublre una grande lmpennaLa, secondo ll calcolo del magglorl crlmlnologl e pollLologl.
ln realLa, anzlche reglsLrare la LanLo LemuLa escolotloo dl vlolenza, la crlmlnallLa comlncl a
dlmlnulre. L'enLlLa dl quell'lnverslone dl Lendenza era lmpresslonanLe. L'lncldenza degll
omlcldl commessl da adolescenLl, anzlche reglsLrare quell'lmpennaLa del 100 per cenLo o
anche solLanLo del 13, secondo lo scenarlo meno caLasLroflco dl !ames Alan lox, nel glro dl 3
annl era crollaLa del 30 per cenLo
3
. 1uLLl cercarono le moLlvazlonl dl quesLa lnverslone dl
Lendenza e le Lrovarono nelle cause plu svarlaLe: dalla grande espanslone economlca al
molLlpllcarsl dl leggl sul possesso dl arml, dalle nuove sLraLegle dl pubbllca slcurezza al
mlglloramenLo delle Lecnlche lnvesLlgaLlve, dall'aumenLo del numero dl pollzloLLl
all'lnLroduzlone dl pene plu severe.
Ma per gll auLorl la vera moLlvazlone sLa alLrove e rlguarda quello che era successo
venL'annl prlma ad una glovane donna dl uallas, norma McCorvey. rlva dl lsLruzlone, dedlLa
all'uso dl alcol e dl sLupefacenLl, dopo avere gla dovuLo dare ln adozlone due bamblnl, nel 1970
sl rlLrov dl nuovo lnclnLa e voleva aborLlre. Ma nel 1exas, come ln quasl n LuLLl gll sLaLl del
aese all'epoca, l'aborLo era lllegale. La donna dlvenne cosl lo sLrumenLo dl un processo
colleLLlvo flnallzzaLo a chledere la legallzzazlone dell'aborLo. ll 22 gennalo 1973 la CorLe
Suprema sl pronuncl a favore della donna, permeLLendo l'lnLerruzlone volonLarla dl
gravldanza su LuLLo ll LerrlLorlo degll SLaLl unlLl. CvvlamenLe per lel era Lroppo Lardl: aveva gla
messo al mondo ll blmbo e l'aveva daLo ln adozlone ma, senza volerlo, con la sua sLorla, flnl
con l'alLerare ln modo sosLanzlale ll corso degll evenLl.
Ma cosa c'enLra LuLLo quesLo con ll crollo della crlmlnallLa? Secondo gll auLorl, e evldenLe
che non LuLLl l bamblnl sono uguall: un blmbo naLo ln un conLesLo soclale dlsaglaLo ha molLe
plu probablllLa dl dlvenLare un dellnquenLe rlspeLLo a coloro che nascono ln condlzlonl plu
vanLagglose e che sono bamblnl voluLl e amaLl. roprlo a segulLo dell'lnLroduzlone della legge
sull'aborLo, molLlsslme donne ln condlzlonl dl dlsaglo soclale hanno poLuLo accedere
all'lnLerruzlone volonLarla dl gravldanza lmpedendo cosl la nasclLa dl quel bamblnl che
avrebbero avuLo magglor probablllLa dl flnlre nelle maglle del crlmlne. Cuel bamblnl non
nacquero mal e quesLo ebbe un poLenLe effeLLo: ad annl dl dlsLanza, proprlo quando quesLl
ragazzl avrebbero poLuLo complere ll loro debuLLo come dellnquenLl ln erba, ll Lasso dl
crlmlnallLa comlncl a dlmlnulre
4
.
Cll auLorl soLLollneano come una Lesl dl quesLo Llpo possa dlsLurbare molLo e come sla, da
un punLo dl vlsLa eLlco, sconvolgenLe acceLLare che ll crollo della dellnquenza e sLaLo un
beneflclo collaLerale della legallzzazlone dell'aborLo. non occorre chlamare ln causa moLlvl
rellglosl o morall per Lrovare agghlacclanLe che una Lragedla prlvaLa dlvenga un beneflclo per la
colleLLlvlLa
3
. Cll auLorl sono ben consapevoll che ll Lema dell'aborLo e esLremamenLe dlfflclle
da LraLLare, ma non possono fare alLro che concludere che eslsLe uno sLreLLo legame Lra

2
lvl, p. xLll.
3
lvl, p. xxlx.
4
lvl, p. xxxl.
3
lvl, p. 143.
Recensione - Fieakonomics

169

legallzzazlone dell'lnLerruzlone volonLarla dl gravldanza e rlduzlone della dellnquenza. u'alLra
parLe, come sosLengono nell'lnLroduzlone del loro llbro, la morale rappresenLa ll modo ln cul
sl vorrebbe veder glrare ll mondo, menLre l'economla rappresenLa ll modo ln cul glra
davvero
6
. Se qulndl la loro lpoLesl pu sembrare lnacceLLablle da un punLo dl vlsLa eLlco,
dlvenLa assoluLamenLe congruenLe sul plano della pura anallsl del daLl. CuesLo e solo uno del
molLepllcl esempl che sl possono fare per mosLrare ll modo dl procedere del due auLorl e cloe
ll LenLaLlvo dl Lrovare nessl Lra quesLlonl apparenLemenLe svlncolaLe le une dalle alLre,
andando olLre ll buon senso, prendendo le dlsLanze dalla saggezza convenzlonale e dlffldando
sempre delle splegazlonl plu ovvle e preconfezlonaLe. lnfaLLl dleLro l'apparenza del pensare
comune sl celano convlnzlonl superflclall ed errorl che solo un'lndaglne accuraLa e
approfondlLa pu smascherare.
Cll auLorl non seguono un fllo conduLLore ben preclso ma spazlano nel plu svarlaLl amblLl.
Leggendo ll llbro, sl scoprlra cosl che, a gludlcare dal numero dl mlnorl che muolono ognl anno,
e plu perlcoloso avere ln casa una plsclna che un'arma da fuoco. AnnualmenLe lnfaLLl muolono
300 bamblnl al dl soLLo del 10 annl per annegamenLo ln plsclna, menLre sono 100 quelll che
rlmangono ucclsl da un'arma da fuoco, ln un paese che dove ne clrcolano plu dl 200 mlllonl.
Cppure sl scoprlra che la maggloranza degll spacclaLorl dl crack vlve ancora con l genlLorl
perche sl LraLLa dl un'aLLlvlLa assal poco reddlLlzla e che la sLruLLura socleLarla su cul sl basa lo
spacclo e ldenLlca a quella delle grandl azlende sLaLunlLensl. LevlLL sosLlene lnfaLLl che lo
spacclo dl crack funzlona ln larga mlsura come qualslasl alLra aLLlvlLa dl lmpresa ln senso
caplLallsLa: per fare soldl blsogna essere vlclno al verLlce
7
. non cl sl deve qulndl sLuplre se uno
spacclaLore dl droga vlve ancora ln casa con la mamma.
Cuando un glovane doLLorando ln soclologla, Sln venkadesh, alllevo dl !.W. Wllson, rlesce
ad enLrare ln possesso del llbrl conLablll dl una gang, nel quall sono rlporLaLl quaLLro annl dl
aLLlvlLa dl spacclo dl droga, da quesLl llbrl emerge che l 120 Lop manager della gang
guadagnano clrca 300.000 dollarl l'anno, menLre per LuLLl gll alLrl membrl la reLrlbuzlone orarla
e dl 7 dollarl all'ora per gll ufflclall e dl 3 dollarl e 30 cenLeslml per l soldaLl sempllcl, cloe ben al
dl soLLo dl un salarlo mlnlmo. 8lmanere ln casa con l genlLorl e qulndl una scelLa obbllgaLa. ln
plu le condlzlonl dl lavoro degll spacclaLorl sono molLo perlcolose con una probablllLa dl venlre
ucclsl dl 1 su 4. erche dunque acceLLare un slmlle lavoro? La rlsposLa dl LevlLL e che la molla
e sempre ll deslderlo dl affermarsl ln un campo esLremamenLe compeLlLlvo, ma che al verLlce,
fa guadagnare un sacco dl soldl. er non parlare del compenso soLLo forma dl glorla e dl
poLere
8
. ln quarLlerl dove ll 36 del mlnorl vlve al dl soLLo della soglla dl poverLa, ll 78 vlve
ln famlglle monoparenLall, solo ll 3 e ln possesso dl un LlLolo dl sLudlo superlore, ll senLlero
che conduce ad un lavoro onesLo e quasl lnvlslblle. L'alLernaLlva e lo spacclo, le cul modallLa
sono analoghe a quelle dl ognl alLra professlone-mlLo: una gran massa dl concorrenLl per un
numero llmlLaLlsslmo dl preml
9
. Anche gll spacclaLorl sl sconLrano qulndl con le rlglde leggl del
mercaLo del lavoro: se l'offerLa dl manodopera e abbondanLe, dl sollLo la paga e bassa.
Come per LuLLe le professlonl-mlLo che hanno molLa rlchlesLa, per cercare dl arrlvare al
verLlcl, blsogna parLlre dal basso, con LuLLl l rlschl e sacrlflcl che quesLo comporLa. Anche ln
quesLo caso qulndl l'lnLenLo degll auLorl e dl mosLrare come dleLro convlnzlonl
apparenLemenLe fondaLe e dleLro la coslddeLLa saggezza convenzlonale sl nascondano realLa

6
lvl, p. xL.
7
lvl, p. 96.
8
lvl, p. 98.
9
lvl, p. 99.


Bumana.Nente - Issue 1u - }uly 2uu9

170

dlverse da lndagare secondo crlLerl anLlconvenzlonall - ma slcuramenLe plu verlLlerl - capacl dl
mosLrare davvero le dlnamlche soclall e dl comprendere fenomenl alLrlmenLl lnsplegablll.


refazlone all'edlzlone lLallana
8reve noLa espllcaLlva
refazlone alla nuova edlzlone ampllaLa e agglornaLa
lnLroduzlone. 1uLLo ha un volLo nascosLo
Che cosa accomuna maesLrl elemenLarl e loLLaLorl dl sumo
Ma come sl fa a paragonare ll ku klux klan a un'accollLa dl agenLl lmmoblllarl?
erche gll spacclaLorl vlvono ancora a casa della mamma?
uove sono andaLl a flnlre LuLLl l crlmlnall?
Che cosa conLraddlsLlngue un perfeLLo genlLore?
ll perfeLLo genlLore (parLe seconda): con un alLro nome, la mla SamanLha, apparlrebbe
alLreLLanLo carlna?
Lpllogo
ConLenuLl exLra
8lngrazlamenLl
noLe


171

Recensione
Capitalismo totale
Peich la finanza minaccia la uemociazia
!ean eyrelevade
lsbn edlzlonl, Mllano, 2008

klccotJo lotl
rlccardo.furl[humana-menLe.lL


ln Lempl dl crlsl economlca l slsLeml d'lnformazlone dl massa cl
parlano dl crollo del mercaLl', svaluLazlone delle azlonl' e bolle
speculaLlve', descrlvendo con alcunl Lermlnl Lecnlcl un mondo che
nella plu banale delle lpoLesl appare dlsLanLe ed lnlnfluenLe dal
sussegulrsl quoLldlano della vlLa dl nol clLLadlnl. Sl caplsce che la
slLuazlone economlca non e delle mlgllorl ma raramenLe vengono
fornlLe splegazlonl esausLlve sulle cause e gll effeLLl delle crlsl o su
quall sono l meccanlsml e prlnclpl che regolano le aLLlvlLa flnanzlarle.
La noLlzla che plu spesso salLa all'occhlo rlguarda la Lal lmpresa che
ha gonflaLo ll bllanclo o ll manager LruffaLore dl azlonlsLl, cosl che ll
danno vlene clrcoscrlLLo all'lnLerno dl un evenLo parLlcolare e
conLlngenLe dovuLo alla sprovveduLezza o malafede dl qualcuno del
proLagonlsLl della flnanza mondlale evlLando dl lnLerrogarsl sulla
naLura dl quesLl evenLl: se, lnvero, slano una lnevlLablle conseguenza
della sLruLLura del copltollsmo flooozlotlo.
ln copltolllsmo totole dl !ean eyrelevade posslamo Lrovare una lnLeressanLe ed accesa
anallsl rlguardo al doraLo mondo della flnanza, l meccanlsml del slsLema ed ll ruolo del suol
aLLorl prlnclpall. oco plu dl 100 paglne dense e placevolmenLe scorrevoll che, senza ollsml o
soflsLlcaLe argomenLazlonl, rendono espllclLl conceLLl Lroppo spesso celaLl da osLlcl Lecnlclsml.
ll llbro dl eyrelevade, gla presldenLe e ammlnlsLraLore delegaLo dl CredlL Lyonnals, nonche
conslgllere del governo Mauroy (1981-83), colnvolge ln una dupllce accusa sla ll caplLallsmo
flnanzlarlo che lnLelleLLuall ed esperLl dl flnanza: ll prlmo reo, con LuLLl l proLagonlsLl che lo
anlmano, dl essere causa delle sue sLesse crlsl e della accenLuaLa dlseguagllanza e mal
dlsLrlbuzlone delle rlcchezze mondlall, l secondl per Lacere, nella mlgllore delle lpoLesl - o
mlsLlflcare, nella pegglore -, alla comunlLa mondlale, quella che e la realLa economlca globale.
ConLenuLl pollLlcamenLe scorreLLl, ma obleLLlvl perche conforLaLl da lnconLesLablll daLl,
rendono ll LesLo uno sLrumenLo che, usaLo con la dovuLa aLLenzlone, permeLLe dl caplre quanLo
la prevlslone dl cresclLa dl un'azlone possa lnfluenzare dlreLLamenLe la nosLra vlLa, ll nosLro
posLo dl lavoro.
La Lesl dell'economlsLa francese e chlara e dlreLLa: slamo enLraLl ln un era dl caplLallsmo
LoLale, compleLamenLe slegaLo dalla socleLa ed ln cul ll flne unlco e l'lncremenLo dl valore.
ll legame Lra ll mercaLo azlonarlo e le lmprese, gla lnLrlnsecamenLe Lenue, non e mal sLaLo
cosl unlvoco nel confronLl del prlmo.
nel prlml caplLoll, ll nosLro auLore descrlve ll conLesLo sLorlco e culLurale della nasclLa del
mercaLo azlonarlo, la fondazlone LeoreLlca e la funzlone praLlca cul doveva assolvere.

Bumana.Nente - Issue 1u - }uly 2uu9

172

Lsso era un luogo vlrLuale ln cul sl moneLlzzava ll valore delle lmprese che poLeva essere
llberamenLe scamblaLo: ln Lal modo le lmprese avevano sempre a dlsposlzlone una cerLa
quanLlLa dl caplLale fresco da poLer relnvesLlre, ln rlcerca o occupazlone, a seconda del
prodoLLo e delle proprle sLraLegle. Sl LraLLava dl un slsLema sLruLLuraLo ln modo Lale che
poLesse, ldealmenLe, generare un clrcolo vlrLuoso dl denaro llquldo.
L doveroso preclsare che eyrelevade sl rlferlsce prlnclpalmenLe al modello llberlsLa degll
SLaLl unlLl, culla del mercaLo e orlglne del nuovo modello dl llberlsmo LoLale.
uunque ll mercaLo nasce ln funzlone dell'lmpresa. Cll azlonlsLl necesslLano dell'lnLervenLo
lnLermedlarlo degll lsLlLuLl dl credlLo, che garanLlscono le rendlLe e decldono gll lnvesLlmenLl da
fare per le lmprese. una volLa descrlLLl l pregl e l dlfeLLl dl quesLo modello, deLLo teoooo,
l'auLore ce ne racconLa ll decllno e ll passagglo ad una forma monopollzzaLa dal poLere
azlonarlo. uopo la crlsl del 1929 sl e lnnescaLo un processo che ha desLlLulLo dl valore l'lsLlLuLo
bancarlo, aLLraLLl dalle luslnghe dl magglorl rendlLe, l rlsparmlaLorl sl sono accasaLl a lsLlLuLl dl
gesLlone del caplLale speclallzzaLl che non lnLerpreLano plu ll ruolo dl lnLermedlarl ma che sono
alle loro dlreLLe dlpendenze. ll loro unlco scopo e l'lncremenLo delle rendlLe per l loro cllenLl.
aradossalmenLe ll mercaLo, oggl, non e plu fonLe dl sosLenLamenLo per le lmprese: quesLe
sono smembraLe, LrasferlLe o chluse su declslonl delle assemblee degll azlonlsLl, LalvolLa sulla
base dl qualche punLo ln plu o ln meno dl lncremenLo apparso su un dlsplay lumlnoso.
L'accusa dell'economo francese sl manlfesLa nel daLl rlporLaLl nel caplLolo cenLrale, ln cul
l'auLore orlenLa la lenLe d'lngrandlmenLo sugll lngranaggl del slsLema, chl sono gll azlonlsLl?
Come vlene sparLlLa la rlcchezza mondlale? un daLo su LuLLl: ll 93 delle rlcchezze del planeLa
vlene conLrollaLo dal mercaLo e sparLlLo ln clrca 300 mlllonl dl azlonlsLl. Conslderando che
slamo 6 mlllardl, fare le deblLe proporzlonl dlvenLa lmbarazzanLe. 1uLLavla, ancora, ll nosLro
auLore non e soddlsfaLLo, vuole vedercl chlaro sulle declamaLe regolamenLazlonl del mercaLo,
sulla funzlone degll lsLlLuLl dl anallsl d'lmpresa, l conslgll dl ammlnlsLrazlone e la sLruLLura della
cotpotote qovetooote.
Le crlLlche dl eyrelevade sono ammonlmenLl verso quesLo slsLema lnlquo che arrlcchlsce
pochl a scaplLo anche degll alLrl: cos'e un falllmenLo, se alla flne sl pu vendere LuLLo? La
provocazlone, ln quesLl Lempl dl globallzzazlone, spera dl ragglungere anche gll lnLelleLLuall,
Lroppo assenLl dal dlbaLLlLo che dovrebbe sLablllre la roLLa economlca mondlale, ed l governl,
sempre plu schlavl dell'economla flnanzlarla - se non connlvenLl nello svendere l dlrlLLl e la
llberLa del proprl clLLadlnl.


refazlone
SalvaLore 8raganLlnl

lnLroduzlone
1. La flne del modello renano
2. 1eorla e praLlca del poLere azlonarlale
3. Cll azlonlsLl: chl sono?
4. MercaLl e cresclLa
3. Che fare?
Concluslonl



173

Ldoardo lerrarlo e AA. vv.


LlLhos edlLrlce, 8oma, 2009

klccotJo lotl
rlccardo.furl[humana-menLe.lL


Olkooomlo e ll LlLolo del nuovo lavoro dl Ldoardo lerrarlo per la collana lllosofla" della LlLhos
LdlLrlce per la quale e auLore dl 1estlmoolote (2006) e vocl Jello feoomeooloqlo (2007). Cuale
sua Lerza opera della collana, Olkooomlo sl rlvela essere un progeLLo amblzloso ed orlglnale
che fln dalle prlme paglne mosLra ll caplllare e mlnuzloso lavoro crlLlco, ermeneuLlco e
semanLlco del nosLro auLore.
lerrarlo lnLroduce l'argomenLo deflnendo cosa sl lnLenda con ll Lermlne lnLrodoLLo ln
fllosofla da ArlsLoLele e composLo dl due parole: olkos e oomos. Sl LraLLa delle regole o norme
dl ammlnlsLrazlone della casa, della 'proprleLa', del 'fare' - un 'fare' cosl dlsLlnLo dall'aglre
(ptoxls) pollLlco.
AmmlnlsLrare, Lrarre proflLLo, governare, cl fa noLare lerrarlo, sono ancora oggl conceLLl
chlave che cosLlLulscono l'essenza dell'economla (Lermlne ln cul e LraslaLo l'orlglnale
arlsLoLellco) e l'economla, a sua volLa, non e che l'orlglnarlo 'fare' (che rlguarda la molLlLudlne
delle aLLlvlLa umane, polche l''economlco' sempre sl da nel 'fare') ln cul l'uomo produce se
sLesso, ovvero: l'olkonomla come onLologla dell'essere-nel-mondo.
ll volume sl cosLlLulsce dl due parLl: la prlma, fruLLo del lavoro del nosLro auLore, ha come
scopo quello dl dare un senso alla domanda se l'olkonomla possa sosLlLulre l'onLologla o sl
possa cosLlLulre come onLo-economla.
nel cercare una rlsposLa, aLLraverso un'anallsl fenomenologlca dl preglo, lerrarlo Lrova
alleaLl lllusLrl, del callbro dl Peldegger e uerrlda, che fornlranno gll sLrumenLl ed l prlml lndlzl
all'lnfaLlcablle rlcerca dell'auLore, una rlcerca che sl svlluppa aLLraverso l plu lmporLanLl
conceLLl e modelll fllosoflcl.
La sLraLegla aLLuaLa nel LesLo dl quesLa prlma parLe, MeLaflslca ed economla", prende le
mosse dalla deflnlzlone dl due caraLLerlsLlche dell'economlco: la sua lndeflnlblllLa e la sua
pervaslvlLa, esso e dunque lndeflnlblle, perche nel suo senso plu amplo sl dlssolve nella
mlrlade delle praLlche umane, e pervaslvo, perche la mlrlade dl praLlche umane sl decllna
molLo spesso e ln molLe parLl degll elemenLl dell'economlco.
Cvunque e ln qualche modo sl manlfesLl un moLlvo o un momenLo dl regolazlone,
ammlnlsLrazlone, gesLlone, dlsposlzlone, dlsclpllna - sono le parole sLesse a dlrcelo - un
procedlmenLo economlco e gla ln corso."
1

L'obleLLlvo e dlmosLrare come sla dlfflclle, se non lmposslblle, deflnlre cosa non e
economlco, cosa c'e 'olLre' l'olkonomla, come l''olLre' dell'onLologla e qulndl 'olLre' la
meLaflslca heldeggerlana.
Clkonomla dunque non come porLaLo dell'onLologla, ma cosLlLuLlva dl essa: precedenLe
anche alla prospeLLlva Leologlca, la quale poLrebbe essere lnLerpreLaLa come orlglnarla forma
d'economla 'sacrallzzaLa'. er lerrralo devono cosl essere lndagaLl l nessl Lra olkonomla ed
onLologla, olkonomla e meLaflslca. Avvalendosl degll sLrumenLl della rlcerca dl Peldegger sl

1
L. lerrarlo, Olkooomlo, LlLhos LdlLrlce, 8oma 2009, p. 14.

Bumana.Nente - Issue 1u - }uly 2uu9

174
arrlva alla concluslone, alLrlmenLl ragglunLa anche dal fllosofo Ledesco, che e nel prlclplo dl
raglone dl Lelbnlz che va maLurandosl ll conceLLo dl un onLo-olkonomla.
La seconda parLe, cosLlLulLa da una raccolLa dl saggl dl varl auLorl, sl occupa delle
decllnazlonl dell'economlco nella fllosofla del nosLro Lempo" e sl avvale dl conLrlbuLl che
fanno da approfondlmenLo e corollarlo alla prlma parLe. er brevlLa dl spazlo non ml e
posslblle clLarll LuLLl pur se LuLLl molLo lnLeressanLl e merlLevoll dl sLudlo, vorrel per meLLere
ln evldenza alcune LemaLlche, ln parLlcolare quelle LraLLaLe ln Lconomla e Leo-logla negaLlva",
dl Clorgla 8ordonl, LLlca/Lconomla" dl Lmlllano CamerLonl, Clorglo Agamben: economla e
nuda vlLa" dl Andrea ue SanLls.
ln concluslone e doveroso preclsare che ll LesLo che sl ha davanLl non e per neoflLl ma
presuppone una profonda conoscenza delle LemaLlche fllosoflche che rlguardano la meLaflslca
e l'onLologla, la passlone susclLaLa da cerLl lnLerrogaLlvl pu LuLLavla suppllre alcune mancanze
e, anche se rlmangono necessarle aLLenzlone e concenLrazlone, ne vale la pena. 8uona leLLura.


arLe prlma
MeLaflslca ed economla

Che cos'e la meLaflslca?
8aglone e fondamenLo
L'essenza della Lecnlca
La fondazlone della meLaflslca ln ArlsLoLele
L'essere e ll dlvlno
Lelbnlz. CnLoLeologla e monadologla
LslsLenza e raglone
La macchlna del mondo
lnformazlone e asslcurazlone sulla vlLa
kanL: rlfondazlone o decosLruzlone della meLaflslca?
L'onLoLeologla alla luce della fllosofla crlLlca
1ecnlca della naLura. L'analogla del fondamenLo
rlnclpl economlcl della meLaflslca
Pegel. La meLaflslca come sclenza della loglca
ll Logllersl del fondamenLo
Lconomla e onLo-loglca dlaleLLlca

arLe seconda
uecllnazlonl dell'economlco nella fllosofla del nosLro Lempo

er una pragmaLlca dell'economlco: da Marx a Pegel Jl Nooe coototote
Che cos'e l'economla pslchlca? 8lflesslonl sull'opera dl lreud Jl Clollo 1osslcl
ldenLlLa ed economla della presenza ln MarLln Peldegger Jl utoqo koccbl
ll prlnclplo dl raglone Jl AlessooJto lotlo
Lconomla e Lecnlca Jl uoolele ue 5ootls
1ra Lerra e clelo. Lconomla e poeLlca ln MarLln Peldegger Jl 1eteso 8ettlol
Lconomla e onLologla nel penslero dl Lmmanuel Levlnas Jl AlessooJto cotool
Asslcurazlone e flne della sLorla ln !an aLocka Jl klccotJo lopotosso
Recensione - 0ikonomia

173
Ceorges 8aLallle: economla generale ed economla rlsLreLLa Jl Ambto Cootoletl
CnLologla ed economla ln !aques uerrlda Jl MoJJoleoo locotelll
Lconomla e Leo-logla negaLlva Jl Clotqlo 8otJool
Sacrlflclo ed economla Jl 5omootbo Motozzello
LLlca/Lconomla Jl mlllooo comettool
Senso e valore. nleLsche-ueleuze: economla dl un lnconLro Jl loco Cotl
Lconomla e governa menLallLa ln Mlchel loucaulL Jl Mlcbele 5poo
La macchlna dell nslcurezza
L'urbanlsmo come sclenza della Lrasformazlone pollLlca Jl Atloooo loJesetto
Clorglo Agamben: economla e nuda vlLa Jl AoJteo ue 5ootls
Sul conceLLo dl economla ln eLer SloLerdl[k Jl Aotoolo loccl
lllosofla ed economla
una rlflesslone a parLlre da AmarLya Sen Jl mlllooo lppolltl
er una cura dell'hablLaL.
enslerl sull'economla globale e sullolkooomlo preglobale Jl 5tefooo Moscblettl



































Bumana.Nente - Issue 1u - }uly 2uu9



177

Commentaiy
Karl Marx's revisited

Mottlo 5ottlet
*



8ereadlng ChapLer 26 (1be so colleJ otlqlool occomolotloo) ln
secLlon vll (1be occomolotloo ptocess of tbe copltol) from karl
Marx's copltol lL became clear Lo me LhaL hls analysls of Lhe
hlsLorlc and economlc evenLs LhaL led Lo Lhe separaLlon beLween
Lhe dlsposal over Lhe means of producLlon and Lhe dlsposal over
Lhe labour power was characLerlzed by meLaphorlc noLlons and
raLher vague language alLogeLher. lL also became clear Lo me LhaL
Lhese pages neverLheless conLaln a deep lnslghL ln Lhe sLrucLure of
Lhe caplLallsL sysLem and a klnd of baslc common sense poslLlon.
1hls poslLlon ls hldden ln hls analysls from Lhe chlld perlod" of
manufacLurlng lndusLry on Lo Lhe year 1867, Lhe year ln whlch he
flnallzes Lhe copltol. 1hls poslLlon can be seen as a premonlLlon
and a foreslghL of Lhe flnanclal slLuaLlon of our days and lL also
opens an lnslghL lnLo Lhe sLrucLure of conLemporary caplLallsm.
Marx derlves hls argumenL ln hls analysls flrsL from book ll of Adam SmlLh's Ao lopolty loto
tbe ootote ooJ cooses of tbe weoltb of Notloos where Lhe llberal LheoreLlclan LreaLs Lhe
prevlous accumulaLlon". 8uL Lhere ls only an unclear alluslon Lo SmlLh's Lheorem of Lhe
naLure, accumulaLlon and employmenL of sLock. Marx, all bomo symbollcos, ls overcome
subsequenLly by an economlc-Lheologlcal comparlson. 1hls comparlson shows noL only LhaL
Marx puLs hls pollLlcal economy" wlLhln Lhe greaL flow of Lhe LradlLlon of hlsLorlcal ldeas, lL
may also suggesL a radlcal deparLure from Lhls LradlLlon. ?eL Lhe meLaphors remaln amblguous.
noLlonally Marx lnLroduces a Lerm by whlch he polnLs ouL Lhe meanlng of Lhe separaLlon
beLween means of producLlon and labour LhaL has Lheologlcal connoLaLlons.

1hls prlmlLlve accumulaLlon plays ln ollLlcal Lconomy abouL Lhe same parL as Lhe orlglnal sln ln
Lheology. Adam blL Lhe apple, and Lhereupon sln fell on Lhe human race. lLs orlgln ls supposed
Lo be explalned when lL ls Lold as an anecdoLe of Lhe pasL. ln Llmes long gone-by Lhere were Lwo
sorLs of people, one, Lhe dlllgenL, lnLelllgenL, and, above all, frugal ellLe, Lhe oLher, lazy rascals,
spendlng Lhelr subsLance, and more, ln rloLous llvlng. 1he legend of Lheologlcal orlglnal sln Lells
us cerLalnly how man came Lo be condemned Lo eaL hls bread ln Lhe sweaL of hls brow, buL Lhe
hlsLory of economlc orlglnal sln reveals Lo us LhaL Lhere are people Lo whom Lhls ls by no means
essenLlal. (8eglnnlng of ChapLer 26)

1he separaLlon LhaL creaLes caplLal ouL of money, he conLlnues, musL proceed ln Lhe followlng
way:

Cn Lhe one hand (tbete ote, M.5.), Lhe owners of money, means of producLlon, means of
subslsLence, who are eager Lo lncrease Lhe sum of values Lhey possess, by buylng oLher
people's labour-power, on Lhe oLher hand, free labourers, Lhe sellers of Lhelr own labour-
power, and Lherefore Lhe sellers of labour. (nexL page)

*
LmerlLus rofessor (Mannhelm), dlrecLor of Lhe Peldelberg School Pans-Ceorg Cadamer"


Bumana.Nente - Issue 1u - }uly 2uu9

178


Marx conLlnues and he sLaLes Lhe sysLemaLlc change Lowards Lhe caplLallsL sysLem ln a
somewhaL dogmaLlc way:

1he caplLallsL sysLem pre-supposes Lhe compleLe separaLlon of Lhe labourers from all properLy
ln Lhe means by whlch Lhey can reallze Lhelr labour. As soon as caplLallsL producLlon ls once on
lLs own legs, lL noL only malnLalns Lhls separaLlon, buL reproduces lL on a conLlnually exLendlng
scale. (same page)

ConLlnually exLendlng ls Lhe scale ln whlch Lhe caplLallsL producLlon progresses. 1here are no
seLbacks and Lhere ls no rupLure ln Lhe scale of Lhe developmenL of caplLallsL producLlon.
So far Lhe exposlLlon of Marx. 1here exlsL of course crafLsmen Loday who own Lhelr means
of producLlon. ln hls book 1he CrafLsman" 8lchard SenneLL, 2008, deals wlLh Lhls
phenomenon and shows clearly LhaL Lhe separaLlon beLween means of producLlon and labour
power ls noL aL all radlcally reallzed ln our days. Maybe we shall look upon Lhe free-lance home
compuLer labourer as worklng ln Lhe chlld perlod" wlLh hls own dlsposal over means of
producLlon and Lhe dlsposal over hls own labour power, worklng laLe aL nlghL or whenever he
pleases Lo do so. 1he exLendlng scale" Loward Lhe caplLallsL sysLem ls lllusLraLed by Marx very
opulenLly ln Lhe subsequenL secLlons of ChapLer 26: 1he colonlal sysLem and sLaLe debLs, Lhe
represslve load of Laxes and Lhe commerclal wars characLerlze Lhe rlse of Lhe caplLallsL sysLem
from chlld perlod" Lo large lndusLry". Pe sLaLes Lhls developmenL ln 1867. Slnce Lhen Lhe
colonlal explolLaLlon (Lhrough Lhe World 8ank), Lhe sLaLe debLs ln Lhelr volume, Lhe Lax load
have all rlsen from Lhe sprouL of Lhe manufacLurlng perlod" or even from 1867 Lo our laLe
caplLallsm of Loday ln a glganLlc unforeseen measure.

LaLer on Marx as a propheLlc auLhor Louches casually Lhe core of Lhe caplLallsL accumulaLlon
process as lL emerges ln Lhe orlglnal accumulaLlon (otsptooqllcbe Akkomolotloo):

lL (tbe coloolol system, M.5.) was Lhe sLrange Cod who perched hlmself on Lhe alLar cheek by
[owl wlLh Lhe old Cods of Lurope, and one flne day wlLh a shove and a klck overLhrew Lhem all.
lL proclalmed surplus-value maklng as Lhe sole end and alm of humanlLy. (page 782)

1he flnallLy of all human llfe was, accordlng Lo Marx, reduced Lo surplus-value maklng, and no
oLher culLural acLlvlLy was henceforLh pursued ln Lhe caplLallsL sysLem. 1hls hlsLorlcal breach
was seen by Marx ln Lhe mosL radlcal way and, of course, Lhls observaLlon ls noL coherenL wlLh
Lhe facLs of llfe of Lhe socleLy, noL aL Lhe Llme of Lhe dlvlslon beLween means of producLlon and
labour power, noL aL Lhe Llme ln whlch Lhe copltol was wrlLLen (1867), nor ln Lhe Llme
LhereafLer. We know many non-surplus orlenLed acLlvlLles ln our culLure: Lhe arLs, Lhe crafLs,
educaLlon, eLc.
lf Lhls ls an exaggeraLlon, Lhere ls sLlll anoLher observaLlon of prlmordlal characLer added Lo
Lhe surplus-value orlenLaLlon. Cur auLhor analyslng Lhe prlmlLlve accumulaLlon" has anoLher
argumenL aL hand Lo explaln Lhe hlsLory of economlc pollLlcs.

1he sysLem of publlc credlL, l.e. of naLlonal debLs, whose orlgln we dlscover ln Cenoa and
venlce as early as Lhe Mlddle Ages, Look possesslon of Lurope generally durlng Lhe
manufacLurlng perlod. 1he colonlal sysLem wlLh lLs marlLlme Lrade and commerclal wars served
as a forclng-house for lL. 1hus lL Look rooL ln Polland. naLlonal debLs, l.e., Lhe allenaLlon of Lhe
SLaLe - wheLher despoLlc, consLlLuLlonal or republlcan - marked wlLh lLs sLamp Lhe caplLallsL
era. 1he only parL of Lhe so-called naLlonal wealLh LhaL acLually enLers lnLo Lhe collecLlve
Commentaiy - Kail Naix's Capital ievisiteu

179

possesslons of modern peoples ls Lhelr naLlonal debL. ubllc debLs become Lhe Creed of Lhe
caplLal. And Lhe emerglng debLs of Lhe sLaLe sLands for Lhe sln agalnsL Lhe Poly SplrlL, whlch
cannoL be pardoned [.]. (page 783)

A noLlon wlLh Lheologlcal lmpllcaLlons ls used by Marx Lo lllusLraLe Lhe lrrefuLablllLy of Lhe
obllgaLlon Lo pay back a debL. ?eL Lhe meLaphor ls somewhaL problemaLlc. 1he sln agalnsL Lhe
Poly SplrlL" ls menLloned here. 1he blbllcal conLexL of Lhls noLlon ls lnLeresLlng. 1he harlsees
accuse !esus LhaL he wanLs Lo exorclze sln by uslng 8elzebub, Lhe super devll. !esus answers,
LhaL he represenLs Lhe LruLh and Lhe Poly SplrlL and Lhose who oppose Lhe Poly SplrlL cannoL
be pardoned. lL ls a klnd of an oaLh or curse LhaL ls uLLered by !esus agalnsL Lhe doubLlng
harlsees (MaLhew 12, 31-32). WhaL does Lhls noLlon mean when used for characLerlzlng Lhe
sLaLe debLs? llrsL of all Lhe lmpllcaLlon of so sLrong a blbllcal noLlon means LhaL Marx wanLs Lo
sLaLe LhaL he ls absoluLely sure ln ralslng Lhe argumenL of Lhe sLaLe debLs and Lhe lrrefuLablllLy
of Lhe obllgaLlon Lo pay Lhem back. ?eL slnce Lhe mlddle ages Lhe credlLors belleved LhaL one
could consolldaLe sLaLe debLs by sLaLe credlLs, by lendlng ouL money from Lhe Lreasury of Lhe
sLaLe as a sLop-gap loan. lor Marx LhaL ls a breech of Lhe baslc prlnclple of loglc whlch ls
underllned by hls recourse Lo Lhe meLaphor of Lhe sln agalnsL Lhe Poly SplrlL". ln hls argumenL
sLaLe debLs form only a sLrucLural elemenL ln Lhe prlmlLlve accumulaLlon and noL any more ln
Lhe caplLallsL sysLem. WhaL he could foresee ls LhaL Lhey do play a ma[or role also Loday ln our
fully grown lndusLrlal socleLles.

1he publlc debL becomes one of Lhe powerful levers of prlmlLlve accumulaLlon. Colonlal sysLem,
publlc debLs, heavy Laxes, proLecLlon, commerclal wars eLc., Lhese chlldren of Lhe Lrue
manufacLurlng perlod lncrease glganLlcally durlng Lhe lnfancy of modern lndusLry. (page 783)

noL only do Lhey lncrease glganLlcally durlng Lhe lnfancy of modern lndusLry as seen ln Lhe year
1867, buL Lhe lnfancy" of Lhe caplLallsL sysLem lasLed unLll our days and Lhe elemenLs Marx
saw ln Lhe "chlld perlod" are sLlll wlLh us and Lhey are growlng lnLo a super-glganLlc scale ln our
days. 1he basls of Lhe colonlal sysLem are for Marx Lhe credlL glven by Lhe sLaLe Lo Lhe colonlal
companles, Lhe debL-maklng by Lhe sLaLe ln order Lo provlde Lhls credlL, Lhe sureLy warranL or
Lhe ball-ouL", as Lhe sLaLe debL ls euphemlsLlcally called Loday.
1he sLrucLure of Lhe economlc developmenL knows many forms LhaL Marx would puL ln hls
sysLemaLlc approach Lo hlsLory lnLo Lhe chlld perlod", ln Lhe manufacLurlng phase or lnLo Lhe
orlglnal prlmlLlve accumulaLlon phase. 1he core of hls pollLlcal economy" ls revealed by Lhls
argumenL, namely LhaL one cannoL spend, whaL hasn'L been earned beforehand. lf Lhere ls a
breach agalnsL Lhls rule Lhe scale" of Lhe caplLallsL sysLem" emerges conLlnually", Lhereby
produclng allenaLlon" whlch ls Lhe foll of Moo" or Lhe foll of AJom".
1here are of course many breaches of Lhe obllgaLlon Lo pay back sLaLe debLs. 1here ls
lnflaLlon and Lhere are flnanclal reforms" wlLh Lhe lssue of a new currency LhaL make sLaLe
debLs shrlnk and Lhey have been adopLed ever so ofLen ln economlc hlsLory slnce 1867.
Marx appears ln hls analysls of Lhe chlld perlod of caplLallsm" as a surprlslngly conservaLlve
economlc Lhlnker, who holds fasL Lo Lhe prlnclple LhaL one cannoL spend, whaL one hasn'L
earned beforehand. 1be foll of moo ln Llmes of ptlmltlve occomolotloo and ever slnce ls a
perslsLlng drama whlch Lakes place ln Lhe breach of Lhls obllgaLlon.







Bumana.Nente - Issue 1u - }uly 2uu9
















181

Commentaiy
La grande trasformazione.
Le origini economicbe e politicbe della nostra epoca
karl olanyl
Llnaudl LdlLore, 1orlno, 1974

Clovoool loocool
glovannl.pancanl[humana-menLe.lL


Se ll penslero dl karl olanyl desLa ancor oggl un vlvo lnLeresse lo
dobblamo senza dubblo ln buona parLe a lo qtooJe
ttosfotmozlooe. elaboraLa sul flnlre degll annl '30, quesL'opera
della maLurlLa dl olanyl e, come noLo, cosLlLulLa da un suggesLlvo
e complesso lnLrecclo dl soclologla, sLorla, economla, pollLlca e,
sopraLuLLo, anLropologla.
ll complLo dl chl scrlve e qulndl reso parLlcolarmenLe
complesso dall'elaborazlone dell'opera sLessa, che peralLro non
dlsLlngue sempre ln manlera espllclLa l dlversl planl del dlscorso,
quesLo per nlenLe Loglle alla chlarezza del volume, come alla
lucldlLa dl sguardo dl olanyl, che ln quesL'opera e affllaLlsslmo.
La LraLLazlone de lo qtooJe ttosfotmozlooe rlguarda
l'lndlvlduazlone del meccanlsmo lsLlLuzlonale che ha permesso alla
clvllLa del xlx secolo dl crollare: quesLo meccanlsmo vlene
lndlvlduaLo nel llbero mercaLo.
olanyl espone acuLamenLe la nasclLa del caplLallsmo aLLraverso lo svlluppo e le modlflche
lsLlLuzlonall e leglslaLlve avvenuLe ln lnghllLerra nel secolo xvll mosLrando come, da un
conLrasLo d'lnLeressl lnlzlale che opponeva parLl dlsLlnLe della socleLa, sl sla orlglnaLa quella
splnLa al camblamenLo che ha deLermlnaLo un'lnarresLablle reazlone a caLena. LlemenLl
cosLlLuLlvl ne sono sLaLl la leglslazlone sulla poverLa (le cosldeLLe loot's low), l'Act of
5ettlemeot (dlsposlzlonl medlanLe le quall sl poneva un llmlLe al mercaLo del lavoro) e ll loro
necessarlo superamenLo, che porL alla creazlone dl un equlllbrlo lnsLablle, del LuLLo nuovo e
pecullare, che flnl per dlffondersl su scala globale.
ll llbero mercaLo, secondo olanyl, non avrebbe comunque poLuLo reslsLere
lndeflnlLamenLe all'lnLerno del resLo della socleLa, senza che quesLa prendesse delle mlsure per
dlfendersl, Lall mlsure avrebbero dovuLo essere una llmlLazlone al llbero mercaLo e la
dlsarLlcolazlone della vlLa lndusLrlale.
Anche cosl, per, soLLo un dlverso profllo, esso avrebbe messo ln perlcolo ll resLo della
socleLa: una slffaLLa conLraddlzlone e quella che porLera lo svlluppo del slsLema dl mercaLo ad
arresLarsl, Lrasclnandosl dleLro le socleLa che sl basavano su dl esso.
La forLe llnearlLa del dlscorso dl olanyl a quesLo punLo comlncla ad aprlrsl ad anallsl
sLorlco-anLropologlche plu parLlcolaregglaLe: se e vero, e lo e, che alcune Lesl dl olanyl
appalono superaLe nella mlsura ln cul sono sLaLl superaLl l LesLl degll sLorlcl cul fa rlferlmenLo
(esemplo calzanLe e ll caso delle opere dl lrenne, che resLano comunque a LuLL'oggl
suggesLlve ed lmporLanLl per chlunque sLudl sLorla medlevale), non e lnvece faclle superare la
lezlone che olanyl Lrae dalle sue vasLlsslme leLLure.
una dl quesLe lezlonl, esposLa ne lo qtooJe ttosfotmozlooe, conslsLe nella Lesl dl fondo
secondo cul vl e un LraLLo dl unlclLa e dl lnnaLurallLa nelle socleLa dl mercaLo: dopo un'anallsl

Bumana.Nente - Issue 1u - }uly 2uu9

182

delle plu remoLe orlglnl soclall dl quello che pu essere conslderaLo l'lnlzlo del processo sLorlco
che ln lnghllLerra deLermlnera lo svlluppo della prlma 8lvoluzlone lndusLrlale, olanyl sposLa la
sua vlsuale dlmosLrando abllmenLe ed ln manlera sLrlngenLe come l fenomenl economlcl
pogglno sempre su basl valorlall exLra-economlche.
CuesLa ldea, che non sl pu dlre essere ne facllmenLe conLesLablle ne banale - o almeno
non lo e oll'lotetoo dl una socleLa dl mercaLo -, cl permeLLe oggl un confronLo con quelle
caLegorle dl lungo perlodo che domlnano l rapporLl economlcl rendendoll, per cosl dlre,
lmmersl (embeJJeJ. leLLeralmenLe sprofondaLl") nell'lnLera vlLa soclale.
CuesLe ldee, come molLe alLre, danno al volume dl olanyl un valore che, come rlsulLera
evldenLe, e largamenLe lndlpendenLe dall'anallsl soclologlca della dlnamlca degll scambl dl benl
da parLe degll lsolanl delle 1robrland o dalla vlslone dl ArlsLoLele sul rapporLo Lra economla
domesLlca ed economla dl vendlLa, flnallzzaLa ad un guadagno.
ConvlncenLe come rlcosLruzlone conceLLuale e anche la successlva esposlzlone dl come sla
avvenuLo ll predomlnlo del rapporLl economlcl sulla socleLa, dl cul erano parLe speclale: quesLa
asslmllazlone da parLe del mercaLo, porLaLa alle sue esLreme conseguenze, rende
evldenLemenLe posslblle quella conLraddlzlone secolare che Lrova la sua plu plena espresslone
nel rapldlsslmo e vlolenLo muLamenLo che le lsLlLuzlonl llberall europee sublrono negll annl '30.
Ld e qul ll cuore delle anallsl dl olanyl: l'lnLera sLruLLura del volume gravlLa aLLorno al
fasclsmo come posslblle e necessarlo sbocco soclale dl quella conLraddlzlone secolare Lra una
sosLanza umana e naLurale" da una parLe e l'asslmllazlone dell'lnLera socleLa da parLe della
sfera economlca dall'alLra.
1uLLo cl sarebbe sLaLo posslblle proprlo a segulLo del secolo della ace del CenLo Annl,
l'CLLocenLo, duranLe ll quale ll mercaLo auLoregolaLo era dlvenuLo la fonLe e la maLrlce
dell'lnLero slsLema: su dl esso, lnfaLLl, pogglava lo sLaLo llberale. La base aurea, lnolLre, alLro
non era che ll LenLaLlvo dl espandere ll mercaLo su scala lnLernazlonale. A sua volLa, l'equlllbrlo
Lra le poLenze era dlvenLaLo una sovrasLruLLura che sl sosLeneva sulla base aurea - ed ln parLe
operanLe su dl essa.
ll complesso equlllbrlo naLo dalla lnLerazlone Lra quesLe quaLLro lsLlLuzlonl vlene a erodersl
ed lnflne a crollare con la dlssoluzlone del Lerzo polo della 1rlpllce Alleanza e, con ll concorso
dell'alLa flnanza, comlnclano quelle fluLLuazlonl dlsLruLLlve che porLeranno prlma ad una serle
dl crlsl economlche nel LrenLennlo 1879-1909, pol al generarsl delle precondlzlonl alla Crande
Cuerra. CuesLa, che e ancora una guerra Lra poLenze oLLocenLesche, apre, con la sua flne,
qualcosa dl nuovo, un muLamenLo che e anche la deflnlLlva affermazlone della poLenza e del
domlnlo soclale lnconLrasLaLo della sfera economlca sul resLo della socleLa, per mezzo
dell'abollzlone della parLeclpazlone democraLlca nella vlLa pubbllca: quesLo, per olanyl (che
comunque ne da varle e preclse deflnlzlonl) e ll fasclsmo.
roprlo su quesLo punLo l'anallsl dl olanyl sl fa lncredlbllmenLe avvlncenLe e credlblle: del
fasclsmo vengono colLe varle sfumaLure - e chl volesse approfondlre quesLo aspeLLo Lrovera un
anLecedenLe a lo qtooJe ttosfotmozlooe ln 1be esseoce of fosclsm, pubbllcaLo ben undlcl annl
prlma, nel 1933: ln quel LesLo, sl afferma conclsamenLe che dopo l'abollzlone della sfera
pollLlca democraLlca resLa solo la vlLa economlca, ll caplLallsmo organlzzaLo nel dlversl seLLorl
dell'lndusLrla dlvenLa l'lnLera socleLa: quesLa e la soluzlone fasclsLa" ( pp. 392).
ll fasclsmo sl conflgurerebbe qulndl come ll prodoLLo dl una socleLa dl mercaLo che sl rlfluLa
dl funzlonare e che sl rlfluLa dl farlo ln base alla scoperLa della sLrldenLe conLraddlzlone che
l'ldea sLessa dl llbero mercaLo auLoregolaLo conLeneva gla lo ooce.
CuesLa vlslone, che pu apparlre sempllclsLlca, e arrlcchlLa da un'lnLegrazlone che agll occhl
dl un leLLore lLallano ha una pregnanza parLlcolare: olLre a quanLo gla deLLo, lnfaLLl, ll fasclsmo
e anche una sorLa dl reazlone emoLlva, quasl uno sfogo lsLerlco e momenLaneo, radlcaLo ln
Commentaiy -La gianue tiasfoimazione

183
ognl comunlLa lndusLrlale. ua quesLo punLo dl vlsLa ll fasclsmo, conLrarlamenLe a come sl e
auLodeflnlLo, non e LanLo un movlmeoto, quanLo una mosso o una teoJeozo loteote dl una
socleLa e dl LuLLo un mondo che vede crollare l pllasLrl della proprla vlLa e che vede rlsucchlare,
lnsleme ad essa, anche la proprla ldenLlLa.
ln quesLo senso ll fasclsmo e, con buone probablllLa, la plu rlusclLa comblnazlone dl
nlchlllsmo e dogmaLlsmo mal apparsa all'lnLerno della socleLa occldenLale nel suo complesso.
SLraordlnarlamenLe peneLranLe, arLlcolaLo e suggesLlvo ll volume dl olanyl rlesce a
superare l'lnadeguaLezza sLorlca dl alcune sue parLl, con un affresco dello svlluppo della socleLa
occldenLale convlncenLe ed addlrlLLura profeLlco, come ln quelle parLl dove accenna alla
dlsLanza crescenLe Lra l'uomo e la naLura, ovvero a come la loglca del llbero scamblo vada ad
lnLaccare l'equlllbrlo dell'ecoslsLema Lrasformando ll suo amblenLe ln un deserLo".







































Bumana.Nente - Issue 1u - }uly 2uu9



183

Commentaiy
Problemi epistemologici dell'economia
Ludwlg von Mlses
Armando LdlLore, 8oma, 1988

Albetto 8loozzl
alberLo.blnazzl[humana-menLe.lL


l saggl raccolLl ln quesLo volume rappresenLano ll necessarlo e prellmlnare sLudlo per quel
correLLo esame del probleml che ho LenLaLo dl porLare a Lermlne ln nomoo Actloo. A 1teotlse
oo cooomlcs." CuesLe le parole lnLroduLLlve dl Ludwlg von Mlses ln un LesLo che e conslderaLo
un classlco dell'eplsLemologla dell'economla e dell'eplsLemologla ln generale. MolLepllcl le
quesLlonl anallzzaLe: la naLura dell'lnduzlone, ll rapporLo Lra Leorla ed esperlenza, la
problemaLlca della comprenslone del faLLl sLorlco-soclall, ll legame Lra soclologla e sLorla,
alcunl argomenLl fondamenLall dl economla pollLlca come la Leorla soggeLLlva del valore e la
converLlblllLa del benl caplLall. uleLro le dlspuLe dl pollLlca economlca, lnfaLLl, vl sono probleml
eplsLemologlcl e meLodologlcl non eludlblll: non e fuorl luogo far noLare che le quesLlonl
asLraLLe dl loglca e meLodologla hanno una sLreLLa lnfluenza sulla vlLa dl ognl lndlvlduo e sul
desLlno dl LuLLa la nosLra culLura".
1
Alllevo dlreLLo dl Carl Menger ed Lugen von 8hm-8awerk,
von Mlses consldera complLo dell'economla la formulazlone dl cl che egll chlama 'leggl
esaLLe', leggl, cloe, che non necesslLano dl esperlenza alcuna per essere confermaLe o refuLaLe.
La sua e una poslzlone arlsLoLellca, secondo cul sl pu arrlvare a leggl vallde sempllcemenLe
per vla deduLLlva. uell'lmposLazlone mengerlana, olLre al prlmaLo del meLodo deduLLlvo, von
Mlses recupera l'ldea dl sclenza esaLLa caraLLerlzzaLa dal rlfluLo dl ognl fllosofla della sLorla,
come dl poslzlonl poslLlvlsLlche, emplrlsLlche, colleLLlvlsLlche, lnsleme alla prlmarleLa del
procedlmenLo composlLlvo e dell'lndlvlduallsmo meLodologlco (concezlone che vede nelle
azlonl del slngoll lndlvldul e non ln enLlLa meLaflslche l'orlglne lnLenzlonale o non lnLenzlonale
del fenomenl soclall). ComplLo prlnclpale della sua anallsl e quello dl dlsLlnguere la Leorla a
prlorl dalla sLorla e dalla sclenza emplrlca, dlmosLrando l'assurdlLa del LenLaLlvl della scuola
sLorlca e lsLlLuzlonallsLa dl rlconclllare cl che e loglcamenLe lncompaLlblle".
2
8lvendlcando
l'auLonomla della sclenza economlca sla da approccl sLorlclsLlcl che sperlmenLall, von Mlses
afferma la leglLLlmlLa loglca della sclenza che ha come suo oggeLLo le leggl unlversalmenLe
vallde dell'azlone umana, leggl lndlpendenLl dal Lempo, dal luogo, dalla razza, dalla nazlonallLa,
dalla classe dl apparLenenza dell'aLLore economlco".
3

L'azlone per Mlses conslsLe nell'economlzzare l mezzl dlsponlblll per la reallzzazlone del flnl
prescelLl. Le sclenze naLurall, lnvece, nel loro processo dl lndlvlduazlone dl regolarlLa
dell'esperlenza chlamaLe leggl, non espongono un dlsegno o cause flnall. Allo sLesso modo, l
llmlLl degll approccl sLorlclsLlcl ln economla sl palesano nell'lncapaclLa dl fare prevlslonl sul
fuLuro, cosl come per lo sLorlco dell'arLe e lmposslblle prevedere, sulla base del passaLo, gll sLlll
dl moda fuLurl. Come preclsa Lorenzo lnfanLlno nella posLfazlone, le Leorle sclenLlflche sono, ln
opposlzlone a qualslasl approcclo sLorlclsLa, a-Leleologlche, non hanno preLese dl reallsmo e sl
soLLraggono alla LenLazlone dl conslderare gll evenLl umanl come dlreLLl verso un desLlno. Le
rlflesslonl eplsLemologlche dl von Mlses, osserva lnfanLlno, olLre a rlbadlre gll errorl commessl

1
L. von Mlses, ltobleml eplstemoloqlcl Jellecooomlo, Armando LdlLore, 8oma 1988, pp.30.
2
lvl, pp.27.
3
lvl, pp.26.

Bumana.Nente - Issue 1u - }uly 2uu9

186

dagll approccl sLorlclsLlcl e neoposlLlvlsLl ln economla, affermano la falllblllLa delle Leorle
sclenLlflche, ll rlfluLo della loglca lnduLLlva e ll prlmaLo del daLo Leorlco per l'lndaglne
sclenLlflca. lnolLre, e posslblle rlnLracclare nella doLLrlna mlseslana alcunl rlsulLaLl anLlclpaLorl
delle Lesl eplsLemologlche popperlane.
4

Che ruolo assegnare, oggl, alla Leorla dell'azlone se confronLaLa con l recenLl svlluppl dl rlcerca
ln economla cognlLlva e sperlmenLale? 1ra le aLLuall llnee dl rlcerca dl quesLo nuovo seLLore dl
sLudl, Lrovlamo ll problema del rapporLo Lra maLemaLlca e realLa, l'anallsl del modelll dl
razlonallLa, l'appllcazlone dl rappresenLazlonl menLall e mappe cognlLlve al processl declslonall,
l fenomenl dl caLegorlzzazlone e blases, ll ruolo delle slmulazlonl al calcolaLore, l'uLlllzzo dl
Lecnlche dl braln lmaglng. SeLLorl dl rlcerca, alcunl, molLo lonLanl dalla concezlone arlsLoLellca
dl von Mlses, che, sl, enucleava l'elemenLo prasseologlco dell'aglre economlco, ma,
conLemporaneamenLe, lo depurava da qualslasl componenLe dl naLura cognlLlva. ln un punLo
le luclde argomenLazlonl dl von Mlses manLengono un rapporLo dl grande aLLuallLa con l
recenLl svlluppl cognlLlvl ln economla: l rlcercaLorl nel campo dell'economla sperlmenLale e
cognlLlva conLemporanea sarebbero concordl nel rlLenere l loro rlsulLaLl plu come apparLenenLl
a una sclenza cognlLlva lnLeressaLa alle quesLlonl economlche, pluLLosLo che come una vera e
proprla speclallzzazlone della sclenza economlca. 8enche sla connoLaLo come un approcclo
forLemenLe crlLlco nel confronLl del modello neoclasslco, quello alla kahneman-1versky non
vlene conslderaLo dagll auLorl sLessl come lndlpendenLe dagll approccl normaLlvl classlcl e ne
rlduclblle a quesLl. Mlses e chlaro su quesLo punLo: L'economla comlncla laddove la pslcologla
flnlsce".
3
CuesL'ulLlma, lnfaLLl, occupandosl dl faLLl pslchlcl, non ha a che vedere con la Leorla
dell'azlone umana, prlva dl elemenLl rlconduclblll alla soggeLLlvlLa. L'economla, conceplLa come
ramo della prasseologla, e, sl, Leorla dell'azlone umana, ma ne e, per deflnlzlone, azlone
razlonale. L ll conceLLo dl razlonallLa cul von Mlses aderlsce e quello classlco: non eslsLono
azlonl o-razlonall ln economla, l'azlone razlonale colnclde con l'aLLo sLesso dl economlzzare.
1uLLo cl che posslamo conslderare azlone umana, osserva Mlses, perche va olLre ll
comporLamenLo meramenLe reaLLlvo degll organl del corpo umano, e razlonale. L ancora, ll
conflne che separa l'economlco dal non economlco, non sl pu cercare nell'amblLo dell'azlone
razlonale, ma colnclde con la llnea che separa l'azlone dalla non azlone".
6
A un prlmo sguardo,
dunque, le llnee dl rlcerca svlluppaLe dal cognlLlvlsLl appalono loglcamenLe lncompaLlblll con la
Leorla dell'azlone mlseslana, auLonoma, allo sLesso Lempo, sla da lmposLazlonl maLemaLlzzanLl
(noLa e la sua concezlone dell'equlllbrlo come espedlenLe Leorlco e non come LenLaLlvo dl
descrlvere la realLa) che pslcologlsLe (rlfluLo dl splegazlonl lnerenLl a daLl pslchlcl).
l rapporLl Lra economla e pslcologla hanno caraLLerlzzaLo da sempre la rlcerca economlca.
7

Anche SchumpeLer ha osservaLo come gll economlsLl sl slano sempre fabbrlcaLe a loro
placlmenLo le osservazlonl sul processl pslchlcl che rlLenevano opporLune nonosLanLe
rlconoscessero la presenza nell'anallsl economlca dl probleml che avrebbero poLuLo
beneflclare del meLodl elaboraLl dagll pslcologl.
8
A paLLo, per, dl deflnlre chlaramenLe ll
conflne Lra le due dlsclpllne, evlLando uLlllzzl sLrumenLall. er esemplo, la legge del compensl
decrescenLl della Lerra sl rlferlsce a un faLLo che sl poLrebbe chlamare flslco, ma cl non
slgnlflca che slamo obbllgaLl a conslderarla un'asserzlone relaLlva alla flslca. SlmllmenLe, se
affermlamo che ll deslderlo dl clbo dlmlnulsce ln manlera lnversamenLe proporzlonale alle

4
L.von Mlses, lo meotollt ootlcopltollstlco, Armando LdlLore, 8oma 1988, pp.9.
3
lvl, pp.32.
6
lvl, pp.134.
7
M.novarese, S.8lzzello, cooomlo 5petlmeotole, Mondadorl, Mllano 2004, pp.8.
8
!.SchumpeLer, 5totlo Jellooollsl ecooomlco, unlversale SclenLlflca 8orlnghlerl, 1orlno 1976, pp.33.
Commentaiy -Pioblemi epistemologici uell'economia

187

porzlonl successlvamenLe lngerlLe, con cl non sLlamo prendendo nulla ln presLlLo dalla
pslcologla, bensl formullamo un faLLo dl esperlenza comune. ln quesLo senso, per SchumpeLer,
non c'e molLa pslcologla nelle proposlzlonl economlche. Cl non lmpllca che l'economla debba
dlslnLeressarsl della rlcerca ln pslcologla, ma parlare dl leggl pslcologlche ln economla
cosLlLulrebbe un flagranLe abuso. L'economlsLa classlco fa a meno dl Leorle pslcologlche e
afferma che l Leoreml dell'economla hanno ll caraLLere dl leggl sclenLlflche".
9

La Leorla dell'azlone mlseslana polche prende ln conslderazlone solo l'essenzlale, e formale,
asslomaLlca e le azlonl sono conceplLe non nella loro forma concreLa, ma come cosLruzlonl
formall".
10
Lssa e a-prlorl, non emplrlca. Cosl come la loglca e la maLemaLlca, non derlva
dall'esperlenza, e anLecedenLe all'esperlenza. L, per cosl dlre, la loglca dell'azlone e
dell'aLLo".
11
L evldenLe qul l'lnfluenza kanLlana nella concezlone dl aLLore economlco che
conosce cl dl cul la naLura e la sLruLLura del nosLrl sensl e delle nosLre menLl cl rendono
conLo" e nella fede ln una forma loglca lmmuLablle e senza Lempo della raglone. ln che modo,
per, dobblamo receplre ll LenLaLlvo mlseslano dl enucleare l'elemenLo loLenzlonale, e
rlLenerlo, allo sLesso Lempo, 'a prlorl', ovvero, precedenLe a qualslasl esperlenza umana?
nemmeno gll assloml della maLemaLlca e della loglca sono, ad oggl, conslderaLl a-prlorl,
flgurlamocl quelll dell'economla. Sl pensl a Culne, per esemplo, fllosofo che ha affermaLo
come la dlsLlnzlone Lra verlLa anallLlche e slnLeLlche, se appllcaLa senza resLrlzlonl a LuLLl gll
enunclaLl Leorlcl, sla, ln realLa, una scomposlzlone lnvallda da un punLo dl vlsLa loglco. L fanno
bene, evldenLemenLe, a dublLarne gll economlsLl cognlLlvl e sperlmenLall che hanno avuLo ll
merlLo dl meLLere ln prlmo plano la complesslLa della nozlone dl razlonallLa, Lermlne, quesLo,
fondamenLale per LuLLo ll penslero economlco. L uLlle noLare come le varle llnee dl rlcerca,
sopraLLuLLo ln amblLo mlcroeconomlco, slano rlconduclblll a dlfferenLl lmpleghl della nozlone
dl razlonallLa. La concezlone classlca dl razlonallLa assume, lnfaLLl, la forma della soluzlone dl
un problema dl oLLlmlzzazlone ln relazlone alla rlcerca dell'lnLeresse personale. 1ale modello
ha sublLo noLevoll Lrasformazlonl lndlvlduablll ln due devlazlonl lnLerne al paradlgma
neoclasslco: ll conceLLo dl razlonallLa del rapporLo mezzl-flnl, che uLlllzza modelll probablllsLlcl
(sla soggeLLlvlsLlcl che oggeLLlvlsLlcl) e ll conceLLo dl razlonallLa llmlLaLa, per cul la razlonallLa -
procedurale e aglLa ln slLuazlonl dl lnformazlone non dlsponlblle senza cosLo - non pu essere
deflnlLa ln Lermlnl dl oLLlmallLa. lnflne, un Lerzo fllone, che lndaga la naLura sLraLeglca del
conceLLo dl razlonallLa opponendosl alla vlslone Lradlzlonale o parameLrlca, sl serve, ln
masslma parLe, della Leorla del glochl. Anche von Mlses crlLlca la concezlone classlca dell'bomo
oecooomlcos.
12
L'economla moderna non comlncla dall'azlone dell'uomo d'affarl, ma da quella
del consumaLorl, dall'azlone, cloe, dl LuLLl". ln quesLa poslzlone slamo forse auLorlzzaLl a
lndlvlduare una qualche afflnlLa con l recenLl svlluppl cognlLlvl? Se presLlamo ascolLo al recenLe
premlo nobel vernon L. SmlLh e alla sua severa crlLlca al modello dell'uomo dl Chlcago,
probabllmenLe sl, ma, domandlamocl, von Mlses ne avrebbe accolLo facllmenLe le relaLlve
lmpllcazlonl? Sebbene l conLrlbuLl dl von Mlses slano conslderaLl LuLLora ln gran parLe aLLuall,
daLo che molLl del suol Leml prlnclpall - l dlrlLLl dl proprleLa, le regole della responsablllLa,
l'efflcacla del mercaLl, l'lnuLlllLa dell'lnLervenLlsmo, la supremazla dell'lndlvlduo - sono dlvenuLl
degll elemenLl lmporLanLl nella Leorla e nella praLlca mlcroeconomlca", egll non poLeva
lmmaglnare come sl sarebbe svlluppaLo clnquanLa annl plu Lardl lo sLudlo delle declslonl

9
L.von Mlses, op.clL., pp.104.
10
lvl, pp.41.
11
lvl, pp.40.
12
lvl, pp.179.

Bumana.Nente - Issue 1u - }uly 2uu9

188

umane.
13
uopoLuLLo, osserva SmlLh, la concezlone dl von Mlses rlfleLLe l'approcclo
meLodologlco unlversale del suo Lempo che conslderava l'economla necessarlamenLe una
sclenza non sperlmenLale.
14
1ale leLLura reLrospeLLlva poLrebbe non rlspecchlare
adeguaLamenLe la poslzlone dl von Mlses, ln quanLo, lo abblamo gla rlcordaLo, egll conceplva
l'approcclo sperlmenLale ln economla, al parl dl quello sLorlclsLa, loglcamenLe lncompaLlblle
con la Leorla dell'azlone umana. Se conslderlamo leglLLlma quesL'ulLlma oblezlone, ln che
modo valuLare affermazlonl come l'economla sperlmenLale e dl forLe sosLegno alla Leorla dl
von Mlses del prezzl dl mercaLo, ma anche alla Leorla dell'equlllbrlo ln condlzlonl sLazlonarle o
ln camblamenLo dlnamlco"?
13
Conosclamo la poslzlone crlLlca dl Mlses nel confronLl del
conceLLo dl equlllbrlo ed egll non poLeva cerLo lmmaglnare che l colleghl sperlmenLall
avrebbero cercaLo dl verlflcarla ln laboraLorlo, ma e proprlo quesLo l'obleLLlvo dl molLa rlcerca
ln economla sperlmenLale: porLare alla luce LramlLe l'esperlmenLo conLrollaLo ln laboraLorlo,
l'elemenLo dl cosLruzlone ldeale" dl molLe Leorle economlche, Lra cul, appunLo, l'equlllbrlo.
von Mlses conslderava l'azlone umana come consapevolmenLe lnLenzlonale, e, allo sLesso
Lempo, conceplva l mercaLl capacl dl auLoregolarsl lndlpendenLemenLe dalle scelLe consapevoll
e dellberaLe degll lndlvldul. Con la sua Leorla dell'azlone, la doLLrlna dl von Mlses ha
lnfluenzaLo profondamenLe gran parLe del penslero economlco del llberlsmo rlformlsLa
anglosassone. 8asLl pensare a Llonel Charles 8obblns, economlsLa brlLannlco suo
conLemporaneo che nel celebre 5oqqlo sollo oototo e sol slqolflcoto Jello scleozo ecooomlco
deflnlra l'economla come lo sLudlo del comporLamenLo umano caraLLerlzzaLo da un generale
lmperaLlvo dl economlclLa polche domlnaLo dalla prlvazlone. La naLura dlnamlca e vollLlva
dell'agenLe economlco e un LraLLo cosLanLe del penslero mlseslano. ln lo Meotollt
ootlcopltollstlco, sagglo del 1936 ln cul anallzza le radlcl e le conseguenze del dlffuso
pregludlzlo anLlcaplLallsLlco, von Mlses afferma che gll uomlnl desLlnaLl a prevalere sugll alLrl
sono quelll che megllo rlescono a fare l conLl con l'lncerLezza lnslLa nel muLamenLo
lsLlLuzlonallzzaLo" e che non Lemono dl confronLarsl con la preoccupazlone derlvaLa dalla
lmprevedlblllLa del fuLuro.
16
ll LraLLo plu caraLLerlsLlco dell'uomo e che egll, con un'aLLlvlLa a
cl flnallzzaLa, non cessa mal dl cercare ll mlglloramenLo del proprlo benessere".
17
L per
reallzzare l proprl scopl, olLre alla necesslLa dl resplrare arla llbera", depuraLa, cloe, da
pregludlzl ldeologlcl che hanno l'unlco obleLLlvo dl lnLrodurre un flnallsmo coaLLlvo nella vlLa
degll uomlnl, l'aLLore economlco mlseslano deve essere ln grado dl effeLLuare scelLe razlonall. l
recenLl conLrlbuLl ln economla cognlLlva hanno evldenzlaLo come, ln realLa, l comporLamenLl
degll aLLorl economlcl non slano sempre soLLo ll conLrollo del penslero razlonale e, sopraLLuLLo,
lnLenzlonale. Come rlconosce SmlLh, Mlses soLLovaluLava enormemenLe ll funzlonamenLo del
processl menLall lnconscl".
18
La rlcerca neurosclenLlflca ha dlmosLraLo come anche l fenomenl
dl presa dl declslone ln slLuazlonl dl lncerLezza slano affronLaLl dal cervello al dl soLLo dl un
llvello dl accesslblllLa cosclenLe. A coloro che sl affldano alla supremazla della raglone nella
Leorla della scelLa, rlsulLa arduo acceLLare ll ruolo (evoluLlvo) eserclLaLo dalle emozlonl
nell'azlone umana. una buona parLe dl esserl umanl, evldenLemenLe, non sceglle ln base a
paragonl basaLl su preferenze, anche se ln deLermlnaLl conLesLl sarebbe opporLuno. Csserva
SmlLh, l loro cervelll conservano rlsorse dl penslero aLLenzlonale, conceLLuale e slmbollco

13
v.L.SmlLh, klflesslool so lozlooe omooo, clopoootoool Jopo, lsLlLuLo 8runo Leonl Cccaslonal aper,
brunoleonlmedla.servlngfreedom.neL/C/61_SmlLh.pdf, n.61, novembre 2008, pp.2.
14
lbldem.
13
lvl, pp. 3.
16
L.von Mlses, op.clt, pp.14.
17
lvl, pp.23
18
v.L. SmlLh, op.clt., pp.4.
Commentaiy -Pioblemi epistemologici uell'economia

189

perche sono scarse le procedure per delegare la magglor parLe delle declslonl a processl
auLonoml (lncluse le emozlonl) che non rlchledono aLLenzlone conscla".
19
Sl poLrebbe perflno
dlscuLere, alla luce del rlsulLaLl delle neurosclenze conLemporanee, se le nosLre declslonl slano
sempre guldaLe da conLenuLo menLale, o per dlrla con Mlses, se slano sempre preceduLe da
penslero. urLroppo, non abblamo oggl una Leorla unlLarla del penslero che cl consenLa dl
argomenLare adeguaLamenLe sull'aLLuallLa dl quesLe poslzlonl. Cl che sembra comlnclare a
farsl sLrada e una concezlone ecologlca dl razlonallLa che dlsLlngue Lra una razlonallLa Leorlca e
una evoluLlva che lmpara dagll errorl. ll dlbaLLlLo sulla razlonallLa degll aLLorl economlcl,
suggerlamo, dovrebbe essere esLeso anche a quella degll economlsLl sLessl: le recenLl rlcerche
ln economla cognlLlva, lnfaLLl, hanno avuLo ll merlLo dl porLare alla luce l noLl errorl dl
lnLrospezlone del Leorlcl che aLLrlbulscono -su base asslomaLlca- una compleLa lnformazlone
agll agenLl economlcl.
Consapevoll della necesslLa dl un magglore approfondlmenLo dl quello qul slnLeLlcamenLe
preso ln esame, posslamo conslderare ll LesLo dl Mlses come porLaLore dl ldee e conceLLl dl
grande aLLuallLa, sopraLLuLLo per la sua 'proleLLablllLa' verso seLLorl dl rlcerca anche molLo
lonLanl da quelll che Mlses sLesso avrebbe poLuLo (o voluLo) lmmaglnare.






























19
v.L.SmlLh, 8azlonallLa cosLruLLlvlsLa e razlonallLa ecologlca ln ctltlco Jello toqlooe ecooomlco. 1te
soqql. koboemoo, McloJJeo, 5mltb, a cura dl MaLLeo MoLLerllnl e Masslmo laLLelll almarlnl, ll
SagglaLore, Mllano 2003, pp.203.

Bumana.Nente - Issue 1u - }uly 2uu9























191

Commentaiy
Sulla natura e l'importanza della scienza economica
Llonel 8obblns
1orlno, u1L1, 1947

5clllo 8elloccl
scllla.belluccl[humana-menLe.lL


Llonel 8obblns, con quesL'opera, propone una Lesl per la
fondazlone della sclenza economlca che, olLre ad essere sLaLa
accolLa dalle Lradlzlonl successlve con un amplsslmo marglne dl
approvazlone, lnLende emanclpare la sLessa aLLraverso un
rlconosclmenLo alLo: dl dlsclpllna auLonoma e sclenLlflcamenLe
fondaLa.
Alla daLa dl pubbllcazlone dl quesLo che lul volle chlamare
'sagglo', ll 1932, ll penslero economlco, lnfaLLl, aveva ancora forLl
connoLazlonl dl caraLLere pollLlco e soclale da cul senLlva la
necesslLa dl svlncolarsl, non cerLo perche ne rlsulLasse un processo
meno rlcco o declsamenLe arldo, bensl per poLer acqulslre
sLrumenLl dl valuLazlone che comporLassero un magglor rlgore
loglco e sclenLlflco. 8obblns LenL quesL'operazlone affldandosl al
meLodo deduLLlvo, per lndlcare una vla d'lndaglne che poLesse porLare nella dlrezlone del
rlsulLaLo deslderaLo. er quanLo cl possa apparlre ln conLrasLo con le Leorle dl alcunl
economlsLl, le cul eredlLa dl penslero sono evldenLemenLe presenLl ln 8obblns ( Wleser,
Weber, keynes, Lra gll alLrl), egll effeLLlvamenLe organlzz ln un slsLema ldee che gla nel
decennl precedenLl erano venuLe a cosLlLulrsl come 'sclenza economlca', ma che, prlve dl una
reale e coerenLe slsLemaLlzzazlone, non poLevano dlrsl Lall.
A fondamenLo della sua Leorla, l'auLore pose una deflnlzlone dl economla che e rlmasLa
famosa flno ad oggl: L'Lconomlca e la sclenza che sLudla la condoLLa umana come una
relazlone Lra scopl e mezzl scarsl appllcablll ad usl alLernaLlvl.
1

8ecuperaLo ll conceLLo dl scarslLa dal penslero dl AugusLe Walras, l'auLore lndlvldua l
quaLLro caraLLerl fondamenLall" del comporLamenLo umano che segnerebbero ll conflne del
campo dl lnLeresse dell'economlsLa:

Cll scopl sono molLepllcl, ll Lempo e l mezzl per consegulrll sono llmlLaLl e sono capacl dl usl
alLernaLlvl, nello sLesso Lempo, gll scopl hanno dlversa lmporLanza...Cra, la molLepllclLa degll
scopl non ha lo s un necessarlo lnLeresse per l'economlsLa...ne la solo llmlLazlone del mezzl e
pet s sufflclenLe a dare orlglne a fenomenl economlcl... ne, ancora, l'appllcablllLa alLernaLlva dl
mezzl scarsl e condlzlone sufflclenLe...Ma quando ll Lempo e l mezzl per consegulre gll scopl sono
llmlLaLl e sono susceLLlblll dl appllcazlone alLernaLlva, e gll scopl possono essere dlsLlnLl ln ordlne
d'lmporLanza, allora la condoLLa assume necessarlamenLe la forma dl una scelLa...ha un aspeLLo
economlco.
2



1
Llonel 8obblns, Sollo oototo e l'lmpottoozo Jello scleozo ecooomlco,1orlno, u1L1, 1947, p. 20.
2
Llonel 8obblns ln Op.clt., pp. 16-17.

Bumana.Nente - Issue 1u - }uly 2uu9

192

La sempllclLa del posLulaLl che reggono la sLruLLura basllare del raglonamenLo economlco
sarebbe Lale da assurgere ad una sorLa dl auLoevldenza lndlscuLlblle, non necesslLanLe dl prove
emplrlche, essendo fruLLo delle quoLldlane esperlenze dl vlLa. A Lal proposlLo e posslblle
muovere una crlLlca al prlnclpl che vallderebbero Lale raglonamenLo: 8obblns, lnfaLLl,
lnserendosl ln una prospeLLlva soggeLLlvlsLa, sl afflda all'lnLrospezlone, Lrascurando dl chlarlre
con magglor efflcacla ll ruolo della verlflca. uesLa curloslLa che proprlo quesLo aspeLLo sla
sLaLo LrascuraLo, o megllo osservaLo da un punLo dl vlsLa poco 'sclenLlflco', nel complesso dl un
lavoro che Lende ad allonLanare l'economla dal laccl che la legano ad alLre dlsclpllne e dalle
lmpllcazlonl dl ognl penslero morale e valorlale.
L'auLore, pur concenLrandosl sulla relazlone Lra mezzl e flnl, lnfaLLl, dlsconosce per enLrambl
la posslblllLa dl cosLlLulrsl ad oggeLLo d'lnLeresse speclflco per l'economla, glacche nlenLe pu
permeLLere all'economlsLa dl esprlmere un gludlzlo sugll unl o sugll alLrl. non e ll prlnclplo
dell'economlclLa che deLermlna necessarlamenLe ll comporLamenLo umano ln vlsLa del
ragglunglmenLo dl uno scopo, anzl, molLl alLrl faLLorl, quall la pollLlca, la rellglone e slmlll, sl
rlvelano ben plu lnfluenLl nell'operare una scelLa pluLLosLo che un'alLra.
Cl che pu rlsulLare lncoerenLe e l'ammlsslone, da parLe dl 8obblns, dl alcunl elemenLl dl
valuLazlone che poco sembrano rlspeLLare un'obleLLlva sclenLlflclLa. Lgll sosLlene, lnfaLLl, che
..benche non sla vero che le proposlzlonl dell'economla anallLlca slano fondaLe su dl una
qualslasl parLlcolare pslcologla, LuLLavla esse lmpllcano lnconLroverLlbllmenLe elemenLl dl
naLura pslcologlca - o per megllo dlre, pslchlca.
3
Le splegazlonl che vengono addoLLe a
glusLlflcazlone dl quesLa lncluslone conceLLuale appalono mancanLl dl una loglca davvero
efflcace. Come nel caso succlLaLo della superflulLa dl prove emplrlche per deLermlnare la
valldlLa del posLulaLl Leorlcl, anche qul sl ha un rlferlmenLo forLe all'lnLerlorlLa e addlrlLLura ad
elemenLl pslchlcl che non vengono mal quallflcaLl ln modo chlaro.

uopo LuLLo, ll nosLro complLo e dl splegare cerLl aspeLLl della condoLLa lndlvlduale, ed e molLo
dlspuLablle che quesLo possa farsl ln Lermlnl che non lmpllchlno nessun elemenLo pslchlco. L
cerLlsslmo che, sla o non sla con cl soddlsfaLLo ll deslderlo della masslma ausLerlLa dl llnguagglo,
cerLl vocaboll come scelLa, lndlfferenza, preferenza, e slmlll, nol ln realLa ll loteoJlomo ln Lermlnl
dl esperlenza lnLerlore.

L ancora

non e ln realLa posslblle lnLendere l conceLLl dl scelLa o delle relazlonl fra mezzl e scopl, che sono l
conceLLl cenLrall della nosLra sclenza, ln Lermlnl dl osservazlonl dl daLl esLernl. ll conceLLo dl
condoLLa rlvolLa ad uno scopo, nel nosLro senso, non lmpllca necessarlamenLe un lndeLermlnlsmo
flnale, ma conLlene, nella caLena della splegazlone causale, anelll che sono pslchlcl, non flslcl...ll
faLLo che quesLl daLl abblano essl sLessl naLura dl gludlzl dl valore non rende necessarlo che
debbano essere valuLaLl come Lall. Lssl non sono gludlzl dl valore compluLl dall'osservaLore. Cl
che lmporLa alle sclenze soclall non e se l gludlzl lndlvlduall dl valore slano cottettl nel senso
deflnlLlvo della fllosofla del valore, ma se essl slano fotmotl e se slano anelll essenzlall nella
caLena della splegazlone causale.
4


Se, da una parLe, l'esposlzlone dl quesLo penslero e prlva dl una sLruLLura fllosoflcamenLe
acceLLablle, glacche nlenLe glusLlflca come sl possano assumere ln quallLa dl 'daLl' elemenLl
lnLerlorl, ne come sl proceda al flne dl oLLenere da un'lnlzlale gludlzlo dl valore un elemenLo
neuLro dl valuLazlone causale, e alLresl sorprendenLe noLare come ln un'opera cosl daLaLa

3
lbldem, p.103.
4
lbldem, pp. 108,109.
Commentaiy -Sulla natuia e l'impoitanza uella scienza economica

193

slano presl ln conslderazlone aspeLLl che oggl cosLlLulscono parLe dl molLa rlcerca ln campo
economlco.
8lfluLando dl assumere un meLodo dl lndaglne meramenLe maLemaLlco e conLrapponendosl
gla al modello classlco dl bomo oecooomlcos, 8obblns sl avvlclna ad una concezlone dell'uomo
e del suol processl menLall meno asLraLLa che ln passaLo. Sebbene l'ldea dl sclenza economlca
come dlsclpllna lndlpendenLe sla alla base dell'lnLero lavoro, oggl cl e posslblle conslderare
quesL'aspeLLo ln una prospeLLlva dlversa, anacronlsLlca, se sl vuole, ma nel senso dl una
prevlslone e dl una preparazlone. l progressl conLemporanel delle neurosclenze, lnfaLLl, hanno
flnlLo con l'lnLeressare dlreLLamenLe gll sLudl sul comporLamenLo economlco degll lndlvldul e l
loro rlsulLaLl vengono appllcaLl alla rlcerca economlca ln vlrLu dl un muLamenLo conceLLuale a
cul, leggendo le sue parole, sl pu credere l'auLore non sl sarebbe LrovaLo lmpreparaLo.
AlLro Lema che emerge dal LesLo e ll rapporLo Lra Lconomla e 1ecnologla, 8obblns rlmane
famoso, appunLo, anche per aver LraLLaLo quesLo aspeLLo con la sollLa energla e per aver
formulaLo, ln proposlLo, un penslero speclflco. Lgll assume che anche la Lecnlca debba essere
conslderaLa un 'daLo', polche da essa dlpendono alcunl meLrl dl valuLazlone. 1ale affermazlone
dlpende ancora da una vlslone ampla rlguardo a cl che pu lnLeressare le sclenze
economlche: LalvolLa la Lecnlca (o Lecnologla) produce osclllazlonl nel prezzl delle mercl,
LalvolLa da essa dlpendono occupazlone o dlsoccupazlone, ma sempre la La sLorla degll
sLrumenLl e la sLorla del genere umano3. ln noLa a quesLa proposlzlone legglamo

Le dlsLlnzlonl da me adoLLaLe ln quesLo punLo sono slmlll a quelle adoLLaLe dal doLL. S18lCL...[Lgll]
cerca dl presenLare la lnLerpreLazlone maLerlallsLlca come una Leorla prlmlLlva dl cl ch'egll
chlama uoteooJetooq(muLamenLo del daLl)..Lende a nascondere la deflclenza lnslLa ln quella
doLLrlna ln quanLo rlfluLa dl Lener conLo del muLamenLl nelle valuLazlonl flnall, salvo a conslderarll
come derlvazlonl da camblamenLl dal laLo dell'offerLa. lo, lnvece, desldero vlvamenLe dl mosLrare
la dlsLlnzlone fondamenLale che passa fra qualslasl splegazlone della sLorla sorgenLe dall'anallsl
economlca, quale nol la conosclamo, e la splegazlone LenLaLa dall'lnLerpreLazlone maLerlallsLlca.
6


non e dlfflclle vedere ln quesLe parole un'lnLenzlone che sl pone al dl la delle dlspuLe
accademlche, pur dovendosene occupare dlffusamenLe. 1uLLo ll LesLo e cosLellaLo dl
osservazlonl che sl oppongono o seguono le Leorle dl pensaLorl conLemporanel a 8obblns.
CuesLo cl rende mlsura dl quanLo allora fosse accesa la dlspuLa lnLorno alla 'naLura delle
sclenze economlche', dl quanLo non sl sla LraLLaLo dl un lmpulso solo lndlvlduale, ma soclale e
culLurale, rlusclre ln quesL'lmpresa.
L'alLra osservazlone che e posslblle proporre, rlguarda la ll penslero lnLero dell'auLore. vero
e che egll LenLa dl sfugglre le maglle dl quesLe alLre suggesLlonl, ma, conLemporaneamenLe, e
lmposslblle pensare alle sue affermazlonl come slegaLe dal fermenLo culLurale e pollLlco
dell'epoca, cul, peralLro, sono lapallsslanamenLe lnLrecclaLe. ln rlsposLa alla 'lnLerpreLazlone
maLerlallsLlca' che vorrebbe l muLamenLl negll scopl e nelle valuLazlonl dlpendere dal
muLamenLl nelle condlzlonl Lecnlche dl produzlone e offerLa, legglamo:

1uLLo cl pu esser vero o falso, chlacchlerlcclo pseudo-hegellano o lnLulzlone profonda ...L una
proposlzlone dl ordlne generale clrca la causazlone del movenLl umanl, che, dal punLo dl vlsLa
della Sclenza economlca, e pura meLaflslca... L'Lconomlca pu ben fornlre un lmporLanLe
sLrumenLo per la splegazlone della sLorla, ma non v'e nulla nell'anallsl economlca che cl auLorlzzl
ad affermare che LuLLa la sLorla debba essere splegaLa ln Lermlnl 'economlcl', se ll vocabolarlo
'economlco' vlene adoperaLo come equlvalenLe dl LecnlcamenLe maLerlale. L'lnLerpreLazlone

3
lbldem, p.32.
6
lbldem, p.32.

Bumana.Nente - Issue 1u - }uly 2uu9

194

maLerlallsLlca della sLorla e glunLa ad esser chlamaLa lnLerpreLazlone economlca della sLorla,
perche sl pensava che oggeLLo dell'Lconomlca fossero 'le cause del benessere maLerlale'. una
volLa che sl e compreso che cl non e, l'lnLerpreLazlone maLerlallsLlca deve reggersl da se o
cadere, la sclenza economlca non porge nessun sosLegno alle sue doLLrlne, ne posLula ln nessun
punLo le relazlonl ch'essa afferma. ual punLo dl vlsLa della Sclenza economlca, l muLamenLl nelle
valuLazlonl relaLlve sono daLl prlmarl.
7
















7
lbldem, pp. 33 e segg.

193

Commentaiy
Storia della economia
!ohn k. CalbralLh
8lzzoll LdlLore, 1orlno, 1990, 368 pp.
Il grande crollo
!ohn k. CalbralLh
8ollaLl 8orlnghlerl, 1orlno, 1991, 218 pp.

Clovoool loocool
glovannl.pancanl[humana-menLe.lL


nonosLanLe ll noLevole numero dl sLorle dell'economla pubbllcaLe
negll ulLlml venLl annl, ll volume dl CalbralLh rlmane, a suo modo,
una leLLura uLlle: a dlfferenza, lnfaLLl, dl molLe opere slmlll la 5totlo
Jello ecooomlo dl CalbralLh unlsce due aspeLLl dlversl che sono
sollLamenLe separaLl, ovvero l'approcclo sLorlco e quello Leorlco.
Come e chlarlLo nel prlmo caplLolo (uoo sqootJo ol poesoqqlo) ll
LenLaLlvo dl CalbralLh e espllclLamenLe quello dl conslderare
l'economla come un rlflesso del mondo ln cul sl sono svlluppaLe
quelle sLesse Leorle: ln quesLo modo Adam SmlLh e leLLo nel
conLesLo delle prlme fasl della rlvoluzlone lndusLrlale, che a sua
volLa e lnLerpreLaLa aLLraverso alcune delle ldee cenLrall dl SmlLh.
CuesLo approcclo clrcolare deflnlsce bene cosa CalbralLh
lnLenda per sLorla dell'economla: non una sLorla degll economlsLl e
del loro penslero (o perlomeno non solo), ma una sLorla lnLegraLa
che esamlnl ll passaLo ln LuLLe le sue varlablll economlche e sLorlche prlnclpall.
nello svlluppo del LesLo sono qulndl presl ln conslderazlone aspeLLl dlverslsslml: parLendo
dal prlml Lre caplLoll dedlcaLl rlspeLLlvamenLe al mondo classlco, al medloevo ed al nascenLe
mercanLlllsmo, sl arrlvano ad esamlnare parLl de lo tlccbezzo Jelle oozlool dl SmlLh. ua quesLo
punLo ln pol ll LesLo sl fa progresslvamenLe plu denso: ln effeLLl la cosLruzlone del volume dl
CalbralLh e armonlcamenLe progresslva ed ll LesLo dlvlene sempre plu rlcco e denso col passare
del caplLoll, una vasLa parLe e qulndl dedlcaLa all'economla del '900, che occupa olLre un Lerzo
del volume, menLre quasl alLreLLanLa e occupaLa dalla seconda meLa dell'800.
CalbralLh cl gulda aLLraverso la formazlone del penslero economlco llberale classlco, con le
sue lacune ed lposLaLlzzazlonl, la Legge 8ronzea del salarl dl 8lcardo e la legge dl Say, flno all'
energlco e peneLranLe" dlssenso dl karl Marx, (cul e dedlcaLo un caplLolo, l'undlceslmo, breve
ma molLo lsLruLLlvo), LuLLo quesLo, assleme all'lnevlLablle lnconLro- sconLro Lra Marx e keynes,
la lucldlLa causLlca dl veblen, ll rlgore dl von Mlses ed Payek, la sLruLLura e la naLura del
caplLallsmo e molLlsslml alLrl Leml sono racconLaLl da CalbralLh ln modo leggero ma mal vuoLo,
anche rlcorrendo al rlcordl personall dl CalbralLh sLesso, che nella sua lunghlsslma vlLa
professlonale ha conoscluLo dlreLLamenLe molLe delle personallLa descrlLLe negll ulLlml dlecl
caplLoll almeno del volume.
ComplesslvamenLe l'opera e molLo godlblle, la leLLura rlsulLa abbasLanza scorrevole ed
esenLe da deLLagll Lecnlcl, anche se la parLe rlguardanLe la moneLa e la sua evoluzlone (ll cap.
xll) e pluLLosLo arLlcolaLa: la rlcosLruzlone sLorlca, come deLLo, e anche (e forse sopraLuLLo)
conceLLuale e quesLo dona al volume una cerLa compaLLezza.

Bumana.Nente - Issue 1u - }uly 2uu9

196

ll volume sl chlude con due brevl caplLoll, ln cul sl Llrano le flla del superamenLo del
penslero keyneslano e della dlffuslone delle ldee dl MllLon lrledman.
La dlsLlnzlone, oggl sconLaLa, Lra mlcro e macroeconomla, la perslsLenza orLodossa del
penslero classlco nella mlcroeconomla e dl come quesLo penslero sla andaLo ln crlsl con la
complesslLa crescenLe dell'lndusLrla sLessa, dove l'equlvalenza classlca del cosLo marglnale del
salarl e del rlcavo marglnale dlvenLa una carlcaLura lnveroslmlle": LuLLl quesLl Leml Lrovano un
abbozzo nella parLe concluslva del volume.
Cuella dl CalbralLh e un'opera che pur essendo sLaLa scrlLLa olLre venL'annl fa, manLlene
ancora una freschezza noLevollsslma, speclalmenLe nel suol caplLoll flnall, dove sl fa plu lnLensa
la rlflesslone sulla naLura dell'economla, del mercaLo e della burocrazla azlendale.
un Lema parLlcolarmenLe lmporLanLe affronLaLo ln quesLl ulLlml caplLoll e quello relaLlvo al
crac del 1929: vlene rlcosLrulLa l' oplnlone e l'aLmosfera allora prevalenLe Lra gll economlsLl e
sono descrlLLe le conseguenze soclall prodoLLe dal crac.
CuesLo fondamenLale Lema e per megllo affronLaLo ne ll qtooJe ctollo. nella vasLlsslma
leLLeraLura sulle cause e lo svolglmenLo del crollo dl Wall SLreeL dell'oLLobre 1929 ll LesLo dl
CalbralLh sl e lmposLo fln dall'usclLa, alla flne del 1934, come un classlco.
CalbralLh sLesso, lnfaLLl, essendo naLo nel 1908 e sopraLLuLLo avendo avuLo un ruolo
pollLlco soLLo la presldenza 8oosevelL dlsponeva nel glornl del pegglor dlsasLro economlco
della sLorla moderna dl un osservaLorlo prlvlleglaLo: glovanlsslmo, a soll 32 annl, era sLaLo
nomlnaLo responsablle del conLenlmenLo del prezzl da parLe del governo.
ll LesLo, la cul composlzlone fu sLlmolaLa nlenLemeno che da ArLhur Schleslnger !r., sl
compone dl dlecl caplLoll preceduLl da una breve prefazlone e da una dlverLenLlsslma
lnLroduzlone, scrlLLa nel Llplco sLlle brlllanLe ed affllaLo dl CalbralLh che rlcosLrulsce la genesl
del llbro ed lnsleme alcune vlcende caplLaLegll nello sLesso perlodo, fra cul un lnconLro/sconLro
con !oseph Mc CarLhy.
La sLorla che prelude al crollo dl Wall SLreeL e per alcunl aspeLLl sorprendenLe slmlle a
quella che gll SLaLl unlLl - e conseguenLemenLe ll mondo lnLero - sl sLanno Lrovando a vlvere
oggl: come oggl, lnfaLLl, all'orlglne della recesslone del 1929 c'e lo scopplo della bolla
lmmoblllare ln llorlda (una bolla che per oggl colnvolge numerosl paesl europel), e come
allora sl rlpropongono gll sLessl meLodl, come quello adoLLaLo da
Charles onzl, un lmprendlLore dl 8osLon che Lruff con ll boom
lmmoblllare molLlsslme persone, un meLodo porLaLo oggl alle
esLreme conseguenze nel caso del crac Madoff, ln cul sono sLaLl
bruclaLl ben 30 mlllardl dl dollarl.
AlLreLLanLo e forse ancora plu lnLeressanLe e ll faLLo che l'ulLlma
parLe della presldenza Poover assomlgll noLevolmenLe alla flne
della presldenza dl Ceorge W.8ush, un faLLo quesLo recenLemenLe
messo ln luce Lra gll alLrl anche dal premlo nobel per l'economla
aul Samuelson.
ll boom azlonarlo amerlcano ebbe ll suo lmpulso declslvo
probabllmenLe ad effeLLo dl una rlvaluLazlone della sLerllna lnglese
nel 1927, declsa dal governo Churchlll: quesLa manovra deLermln
una llquldlLa lmprovvlsa a dlsposlzlone dl chlunque volesse
lnvesLlre ln 8orsa negll uSA.
SlmllmenLe a quello che era successo ln llorlda, dove sl era rlusclLl a svlncolare ll dlrlLLo
all'lnvesLlmenLo dagll onerl del possesso concreLo aLLraverso ll meccanlsmo del compromesso
- un slsLema che llberava da molLl rlschl dl mercaLo, almeno lnlzlalmenLe - cosl anche nella
8orsa dl new ?ork accadde qualcosa dl slmlle con l coslddeLLl LlLoll a rlporLo". ll mercaLo fu
Commentaiy -}ohn K. ualbiaith

197

cosl llbero dl glrare a vuoLo, senza rlschl apparenLl: molLlsslme banche accordarono credlLo agll
speculaLorl ed al loro agenLl dl camblo, fln quando ll slsLema non comlncl a mosLrare delle
crepe.
negll ulLlml mesl del 1928 ll mercaLo comlncl a dare segnall dl nervoslsmo e comlnclarono
a manlfesLarsl quelle amblgulLa e compromlsslonl che porLeranno al crac dopo poco plu dl un
anno.
ll segreLarlo del 1esoro Andrew Mellon persegul una pollLlca dl compleLo lolssez-folte,
menLre l'unlca manovra davvero radlcale che avrebbe poLuLo forse sgonflare la bolla senza
farla esplodere, ovverosla la rlchlesLa al Congresso dell'auLorlzzazlone ad lnLerrompere
dlreLLamenLe le conLraLLazlonl a rlporLo rlponendone le condlzlonl dl legallLa nelle manl del
Congresso sLesso al flne dl un regolamenLazlone non fu aLLuaLa da parLe dell'organo
compeLenLe, ll Conslgllo del CovernaLorl della 8lserva lederale, esso sLesso ln parLe composLo
da speculaLorl.
La 8orsa era conslderaLa sempre meno come la reglsLrazlone dl prospeLLlve azlendall dl
lungo Lermlne e sempre plu come ll luogo ln cul sl complvano macchlnazlonl arLlflclose, dove sl
creavano dall'oggl al domanl enorml rlcchezze sulla base dl vocl" e dl lnformazlonl rlservaLe.
CuesLo comporL un dlsLacco crescenLe Lra l'lnformazlone pubbllcamenLe dlsponlblle
sull'andamenLo del mercaLo (lnformazlone che sl scoprl ln segulLo essere lnfllLraLa dagll sLessl
poLerl economlcl che erano colnvolLl nelle operazlonl dl 8orsa) e la realLa.
Al culmlne del nervoslsmo un gruppo formaLo dal magglorl flnanzlerl della nazlone declse dl
comprare ln modo masslcclo LuLLl l LlLoll soLLovaluLaLl, salvando ll mercaLo per quello che pu
essere conslderaLo ll canLo del clgno della 8orsa dl new ?ork: solLanLo pochl glornl dopo dl
fronLe ad una nuova ondaLa dl panlco, lo sLesso gruppo dl flnanzlerl sl rlunl e declse esso sLesso
dl vendere. lu ll panlco.
ll prezzo del MarLedl nero dl Wall SLreeL, come noLo, fu durlsslmo da pagare sla per gll SLaLl
unlLl che per ll resLo del mondo: dlecl annl dl depresslone economlca, dlsoccupazlone dl massa
ed ll sorgere del nazlsmo ln Cermanla.
ll LesLo dl CalbralLh ha espllclLamenLe valore come LesLlmonlanza sLorlca e non sl occupa
delle conseguenze soclall del 1929, come lnvece accade ln 5totlo Jello ecooomlo, come LesLo dl
sLorla dell'economla LenLa per dl meLLercl ln guardla da posslblll nuovl casl dl Lracollo della
8orsa e dalle loro nefasLe conseguenze, a quesLo proposlLo e assal lsLruLLlvo leggere le ulLlme
paglne dell'ulLlmo caplLolo, coose e cooseqoeoze per Lrarne l deblLl lnsegnamenLl sulla naLura e
lo svlluppo del caplLallsmo.

















Bumana.Nente - Issue 1u - }uly 2uu9

199

Commentaiy
Micromotives and Macrobebavior
1homas Schelllng
W.W norLon and Company, 1978

loolo Moffezloll
*



My goal ls noL so much Lo provlde a summary of 1homas C.
Schelllng's Mlctomotlves ooJ Moctobebovlot, buL raLher Lo Lry Lo
clarlfy some of Lhe lssues LhaL Schelllng deals wlLh from a
phllosophlcal polnL of vlew. 1wo quesLlons lmmedlaLely arlse. llrsL,
WhaL sorL of perspecLlve do we have ln mlnd when we speak of
provldlng a phllosophlcal analysls of Mlctomotlves ooJ
Moctobebovlot? Second, ls Lhls perspecLlve adequaLe for offerlng an
accounL, as far as posslble compleLe, of a book wlLh such conLenL?
lndeed, Mlctomotlves ooJ Moctobebovlot seems noLhlng buL an
lnformal lnLroducLlon Lo formal meLhods ln soclal sclences. Such
formal meLhods are Lhe Lools used Lo analyse Lhe maln sub[ecL
maLLer of Lhe book: Lhe lnLerrelaLlonshlp beLween Lhe lndlvlduals
and Lhe soclal aggregaLe Lhey comprlse. ln parLlcular, Schelllng's
pro[ecL ls Lhe LheoreLlcal explanaLlon of how lndlvldual purposlve behavlour (mlcromoLlves)
can deLermlne group equlllbrlum or a dlsequlllbrlum process (macrobehavlor). ln oLher words,
Lhe book explores Lhe relaLlon beLween Lhe behavlour characLerlsLlcs of Lhe loJlvlJools who
comprlse some soclal aggregaLe, and Lhe characLerlsLlcs of Lhe oqqteqote.
1

8oLh Lhese concepLs, lndlvlduals and aggregaLe, deserve some words of clarlflcaLlon. Cn Lhe
one hand, Schelllng's lndlvlduals are goal-dlrecLed, namely, Lhey behave ln a way LhaL we
mlghL call purposlve: Lhey have preferences, pursulng goals, mlnlmlzlng efforL or
embarrassmenL or maxlmlzlng vlew or comforL.
2
8arely do people care abouL Lhe ouLcome of
Lhe aggregaLe, Lyplcally Lhelr own declslons and Lhelr own behavlour are moLlvaLed by Lhelr
own lnLeresL. ln Lhls perspecLlve, Lhe enLlre book can be read as a sorL of generallzed rlsoner's
ullemma: behlnd any acLlon Lhere ls a uLlllLy-based reason Lo acL. 1hls ls a purely game-
LheoreLlcal assumpLlon. Schelllng argues LhaL, even lf lL ls someLlmes mlsleadlng, lL ls
lndlspensable for Lhe analysls of Lhe relaLlonshlp beLween lndlvldual goals and aggregaLe
behavlour.
Cnce Lhe game's scenarlo ls flxed, some lnLeresLlng consequences follow. Cne ls LhaL
prlvaLely opLlmlzed declslons mlghL lead ln aggregaLe Lo ouLcomes LhaL are undeslrable for
everyone. As Schelllng says aL Lhe beglnnlng, Lhere are several reasons we mlghL lnLeresL
ourselves ln whaL lL ls LhaL Lhose people [are] dolng, or [Lhlnk] Lhey [are] dolng, or [are] Lrylng
Lo do [...]. Cne ls LhaL we do noL llke Lhe resulL.
3

Cn Lhe oLher hand, Lhe aggregaLe, whlch resulLs from goal-dlrecLed lndlvldual behavlour, ls
noL merely an exLrapolaLlon from Lhe lndlvldual. Cf course, Lhere are some easy cases ln whlch

*
hu SLudenL, unlverslLy of llorence
1
1homas C. Schelllng, Mlctomotlves ooJ Moctobebovlot, W. W. norLon and Company, 1978. p.13. lLallcs
ln orlglnal.
2
1homas C. Schelllng, clt., p. 17.
3
1homas C. Schelllng, clt., p. 12.

Bumana.Nente - Issue 1u - }uly 2uu9

200

Lhls happens, e.g. Lhe case ln whlch l choose an acLlon no maLLer whaL Lhe oLhers declde Lo do.
lor lnsLance, lf we know LhaL every drlver Lurns hls llghLs on aL sunseL, lL ls easy Lo guess whaL
Lhe feaLures of Lhe aggregaLe are: all Lhe car llghLs ln a local area wlll Lurn on aL abouL Lhe same
Llme. 8uL Lhe mosL lnLeresLlng slLuaLlons are Lhose ln whlch people's behavlour depends on Lhe
behavlour of oLher people, LhaL ls, when Lhelr behavlour ls noL (or noL only) consLralned by
some klnd of exLernal resource buL also by Lhe behavlour of oLher lndlvlduals. lor lnsLance, ln
Lhe example above, lL ls no longer Lrlvlal Lo derlve Lhe feaLures of Lhe aggregaLe lf we know
LhaL every drlver Lurns hls llghLs on only when Lhe oncomlng cars have already done so. 1he
aggregaLe ls now a sorL of sysLem of lnLeracLlons ln whlch lndlvlduals respond Lo an
envlronmenL, whlch lnclude oLher people's responses Lo Lhelr own envlronmenL.
AnoLher lmporLanL feaLure of Lhe mlcromoLlves underlylng lndlvldual acLlons ls Lhe facL LhaL
Lhey are noL markeL moLlvaLlons. Schelllng wrlLes LhaL qulLe a number of acLlvlLles operaLe
almosL enLlrely ouLslde Lhe purvlew of Lhe markeL, so Lhey cannoL even be expecLed Lo recelve
markeL-llke beneflLs or LreaLmenL. Some examples are Lhe languages we choose Lo speak, Lhe
person we declde Lo marry, whom we llve near, whlch games we llke Lo play, eLc.
Schelllng clalms LhaL Lhe behavlour he ls concerned wlLh ls noL economlcal sttlcto seoso. Pls
lnLeresL ls ln aggregaLe behavlour when markeL mechanlsms are noL operaLlng, namely, ln all
Lhose slLuaLlons ln whlch Lhose who are lnvolved are noL volunLary parLlclpanLs: Lrafflc [ams,
sex-dlsLrlbuLlon ln campus dlnlng hall, sendlng of ChrlsLmas cards, membershlp ln brldge or
Lennls clubs, raclal composlLlon of nelghbourhoods, and so on.
A very lllumlnaLlng example of Lhls klnd of behavlour ls Lhe one Schelllng beglns hls book
wlLh: Lhe commonly observed seaLlng paLLern ln a publlc lecLure hall, where Lhe audlence ls
packed lnLo Lhe back rows leavlng four, flve or slx rows ln fronL of Lhe lecLure hall compleLely
empLy. Any aLLempL Lo explaln such a slLuaLlon uslng Lhe Lechnlques of economlcs wlll fall, noL
because Lhese Lechnlques are lnadequaLe, buL because Lhe explanaLlon ls, ln a sense,
lncompleLe.
1he reason for Lhls fallure ls LhaL Lhe lndlvlduals are noL lnvolved ln LhaL slLuaLlon ln vlrLue
of Lhe facL LhaL Lhey wanL Lo be lnvolved, buL lL slmply so happens LhaL Lhey are. Schelllng's alm
ls Lhen Lo enlarge Lhe scope of soclal sclence beyond Lhe analysls of economlc behavlour and
Lo show us LhaL economlcs [ls] a large and lmporLanL speclal case, raLher Lhan a model for all
soclal phenomena.
4


LeL me skeLch ln shorL Lhe sequence of Lhe book's Loplcs.
CPA1L8 CnL ls a sorL of synopsls of Lhe enLlre monograph and lL ls deflnlLely Lhe mosL
lnLeresLlng parL for Lhe purpose of glvlng, as we sald above, a phllosophlcal accounL of
Schelllng's work. 1he ChapLer presenLs whaL ls Lo be consldered Lhe maln quesLlon of Lhe book:
lf Lhe lndlvlduals acL purposlvely buL Lhe behavlour of oLhers elLher consLralns whaL Lhe
lndlvlduals can do or ls parL of Lhe lndlvlduals' preferences, whaL klnd of aggregaLe behavlour
wlll resulL?
CPA1L8 1WC develops Lhe models of nlne slLuaLlons where Lhe behavlour of Lhe aggregaLe
ls a loglcal necesslLy, unaffecLed by Lhe speclflc goals or behavlour of Lhe lndlvlduals. lndeed,
Lhey do noL even merlL Lhe appellaLlve of models", Lhey are merely deflnlLlonal sLaLemenLs":
Lhelr belng Lrue does noL depend on Lhe way lndlvlduals acL or on whaL Lhelr goals are, Lhe
LruLh of such sLaLemenLs derlves dlrecLly from Lhe deflnlLlon of Lhe Lerms LhaL occur wlLhln
such sLaLemenLs.

4
1homas C. Schelllng, clt., p. 27.
Commentaiy -Niciomotives anu Naciobehavioi

201

lor lnsLance, Lelephonlng ls a Lyplcal acLlvlLy Lhrough whlch one can lnfluence Lhe
behavlour of oLher people. now, surely lL ls Lrue LhaL none of us ever makes [usL as many calls
as he recelves, buL lL ls also Lrue LhaL lf Lhe sysLem of calllng people on Lhe phone presupposes
a sender and a recelver, Lhen Lhe number of lncomlng calls ls equal Lo Lhe number of
ouLcomlng ones. 1herefore, ln Lhe aggregaLe, lL ls Lrue LhaL Lhe LoLal number of recelved calls
equals Lhe LoLal number of made calls, alLhough lL could happen (and ln a large number of
cases lL really does happen) LhaL we lndlvldually recelve more calls Lhan we make, or vlce
versa.
1hls example lllusLraLes an lmporLanL feaLure of a large class of sLaLemenLs, whlch Lhe
economlsLs are very lnLeresLed ln: proposlLlons LhaL are Lrue ln Lhe aggregaLe buL noL ln
deLall, and are Lrue lndependenLly of how people behave.
3
1he reason of Lhe emphasls on
such sLaLemenLs ls LhaL economlcs ls manly concerned wlLh exchanges of equlvalenL values.
AlLhough we don'L need any maLhemaLlcal Lool Lo recognlse LhaL Lhe example of
Lelephonlng ls Lrue, Schelllng polnLs ouL LhaL even ln Lhls case one flnds surprlses ln Lhe
relaLlonshlp beLween Lhe mlcromoLlves level and Lhe macrobehavlor level. Conslderlng how
banal Lhese proposlLlons sound, lL ls asLonlshlng how many hours of commlLLee meeLlngs have
been spenL on proposals Lo mlx men and women ln dormlLorles, or blacks and whlLes, or
freshmen and sophomores, ln ways LhaL vlolaLed Lhe slmple arlLhmeLlc prlnclple LhaL no
maLLer how you dlsLrlbuLe Lhem, Lhe numbers ln all dormlLorles have Lo add up Lo Lhe
numbers LhaL Lhere are.
6
When Lhe soclal slLuaLlons become even a llLLle more complex,
lnLulLlon ls noL enough Lo reallse LhaL Lhere are very slmple and lnescapable maLhemaLlcal
laws, llke Lhe one menLloned above, whlch underlle Lhese slLuaLlons.
CPA1L8 1P8LL lnLroduces some models, or famllles of models, ln whlch lndlvldual goals and
behavlours do affecL aggregaLe behavlour. Schelllng's analysls prlmarlly focuses on such
models.
Cne famlly ls LhaL of Lhe crlLlcal mass" models. CrlLlcal mass models descrlbe a large
number of phenomena LhaL become self-susLalnlng once Lhe level of acLlvlLles passes some
mlnlmum level. WhaL lL ls common Lo all [crlLlcal mass slLuaLlons] ls Lhe way people's
behavlor depends on bow mooy are behavlng a parLlcular way, or how much Lhey are behavlng
LhaL way.
7
Schelllng glves Lhe academlc example of a dylng semlnar", l.e. Lhe semlnar whose
popularlLy depends on lLs popularlLy, lL wlll llve or dle, dependlng upon wheLher Lhe sufflclenL
condlLlon ls saLlsfled, namely, upon wheLher Lhose Laklng parL ln lL conslder Lhe number of
oLher parLlclpanLs sufflclenL, relaLlve Lo some sLandards. ln oLher words, lf enough oLher
people are dolng someLhlng, we wlll make Lhe same cholce. no one wanLs Lo be Lhe only
person aL a club. So we go Lo a club only lf we expecL enough oLher people Lo be Lhere Lo make
lL en[oyable.
1he mosL lmporLanL subclasses of crlLlcal mass models are lemons" and Llpplng" models.
1he former lnvolve buyers and sellers who respond ln dlfferenL ways Lo Lhe acLual proporLlon
of bad used car, l.e. Lhe lemons", ln Lhe Lermlnology of Akerlof.
8
lf lemons are rare, Lhe
markeL wlll be sLable because of exchanges wlll muLually advanLageous for boLh buyers and
sellers. 8y conLrasL, when Lhe number of lemons lncreases and when lL exceeds Lhe crlLlcal
number, Lhe enLlre markeL may dlsappear. 1he laLLer models descrlbe Lhe lnsLablllLles ln Lhe
raclal composlLlon of nelghbourhoods. lL was observed LhaL Lhe enLrance of a few members

3
1homas C. Schelllng, clt., p. 49.
6
1homas C. Schelllng, clt., p. 39.
7
1homas C. Schelllng, clt., p. 94. lLallcs ln orlglnal.
8
Ceorge A. Akerlof, 1he markeL of 'lemons': CuallLy uncerLalnLy and Lhe markeL mechanlsm, Ooottetly
Iootool of cooomlcs, 84:3, 1970.

Bumana.Nente - Issue 1u - }uly 2uu9

202

of a mlnorlLy lnLo a nelghborhood ofLen causes some among Lhe formerly homogeneous
populaLlon Lo leave, or Lo slgn of leavlng. 1helr deparLure lefL openlngs, so more member of
Lhe mlnorlLy could enLer, Lhe lncrease ln new resldenLs lnduced more of Lhe old Lo leave [...]
Some of Lhe deparLure mlghL be moLlvaLed by Lhe mlnorlLy enLranLs who have already arrlved,
some by Lhe bellef LhaL Lhe process, would conLlnue [...].
9

1wo oLher classes of models LhaL are sLrlcLly relaLed Lo crlLlcal mass models refer Lo,
respecLlvely, Lhe commons"-Lype of processes and Lhe expecLaLlon-deLermlned processes.
More speclflcally, Lhe former class refers Lo all Lhose slLuaLlons ln whlch Lhe lndlvlduals acL
purposlvely and, by dolng so, produce an aggregaLe LhaL ls consldered undeslrable by
everyone. 1he rlsoner's ullemma ls Lhe mosL famous and clear example of one slLuaLlon of
Lhls klnd. !. Plrshlelfer
10
has polnLed ouL LhaL Lhe [uxLaposlLlon of commons processes wlLh Lhe
crlLlcal mass phenomena ls qulLe mlsleadlng. lndeed rlsoner's ullemma slLuaLlons show
cycllcal sLrucLures LhaL are Lyplcal of Lhe class of phenomena descrlbed by crlLlcal mass models.
erhaps Lhls ls Lhe reason why Schelllng ls lnLeresLed Lhese slLuaLlons ln Lhls ChapLer.
Cn Lhe oLher hand, expecLaLlon-deLermlned processes refer Lo a class of processes LhaL has
Lhe sLrucLure of self-fulfllllng prophecles: Lhe general ldea ls LhaL cerLaln expecLaLlons are of a
such characLer LhaL Lhey lnduce Lhe klnd of behavlor LhaL wlll cause Lhe expecLaLlon Lo be
fulfllled. [...] lor example, lf a parLlcular mlnorlLy ls consldered lncapable of holdlng responslble
poslLlons, Lhey wlll noL be hlred for responslble poslLlons, Lhey wlll have no opporLunlLy for
responslble poslLlons, and lacklng any such experlence, Lhey may lndeed be lncapable.
11

CPA1L8 lCu8 and CPA1L8 llvL examlne Lhe appllcaLlon of Lhe models presenLed above for
Lwo klnds of problems: segregaLlon and lnLegraLlon. ln parLlcular, ChapLer four deals wlLh
segregaLlon and lnLegraLlon wlLh respecL Lo sex and race, whlch are consldered dlscreLe
varlables. lnsLead, ChapLer flve ls dedlcaLed Lo an analysls of segregaLlon and lnLegraLlon when
Lhe varlables of Lhe sysLem are conLlnuous: age and lncome.
CPA1L8 Slx ls qulLe lndependenL from Lhe oLher chapLers. lL was publlshed before Lhen
12

and lL could be read lndependenLly of Lhe enLlre book. lL conslders a posslble slLuaLlon - one
LhaL mlghL soon become a reallLy, ln Lhe auLhor's oplnlon - where parenLs would have Lhe
opporLunlLy of selecLlng Lhelr chlldren's genes.
CPA1L8 SLvLn ls abouL rlsoner's ullemma wlLh coallLlons, Lhls ChapLer geLs raLher
Lechnlcal, wlLh Lhe help of a loL of maLhemaLlcs. As lerglorglo Cdlfreddl polnLed ouL ln hls
lnLroducLory noLe Lo Lhe flrsL lLallan edlLlon of Mlctomotlves ooJ Moctobebovlot
13
, Lhe
dlfference beLween game Lheory as flrsL ploneered by nash, SelLen and Parsanyl and Lhe very
same Lheory as developed by Schelllng and Aumann ls LhaL whlle Lhe flrsL economlsLs llmlLed
Lhemselves Lo Lhe analysls of confllcLs ln whlch Lwo or more players have no posslblllLy of
maklng coallLlons, Schelllng and Aumann sLudled confllcLs wlLh coallLlons. 1hls ls Lhe reason
why Schelllng and Aumann recelved Lhe nobel rlze ln Lconomlcs.
1he resulL of a baslc rlsoner's ullemma slLuaLlon ls LhaL prlvaLely opLlmlzed declslons
(mlcromoLlves) lead Lo an aggregaLe (macrobehavlor) LhaL ls undeslrable for everyone. ls Lhls
resulL lnescapable? 1haL ls, ls lL always Lrue LhaL, lf we play a rlsoner's ullemma-based game,

9
1homas C. Schelllng, clt., p. 101.
10
!. Plrshlelfer, 8evlew of Mlctomotlves ooJ Moctobebovlot by 1homas C. Schelllng, Iootool of cooomlc
lltetotote, Sol 18, no. 3 (SepL., 1980), pp. 1092-1094.
11
1homas C. Schelllng, clt., p. 113.
12
Chooslng our chlldren's genes, ln Mack Llpkln and eLer 1. 8owley (eds), Ceoetlc kespooslblllty,
lenum ress, 1974.
13
lerglorglo Cdlfreddl, Noto lottoJottlvo a 1homas C. Schelllng, Mlctomotlvozlool Jello vlto pootlJlooo,
8omplanl, 2008.
Commentaiy -Niciomotives anu Naciobehavioi

203

Lhe cooperaLlve cholce ls sLrlcLly domlnaLed by Lhe non-cooperaLlve one? Schelllng's answer ls
negaLlve. Pls concern now ls Lo exLend Lhe deflnlLlon of Lhe baslc verslon of Lhe game (l.e. Lhe
Lwo-player verslon), ln order Lo [caLch] Lhe splrlL of Lhe rlsoner's ullemma
14
, when lL ls
played by more Lhan Lwo persons. Schelllng makes Lwo hypoLheses: flrsL, an lndlvldual ls
always beLLer off, Lhe more oLher people choose Lhelr unpreferred alLernaLlve, second, Lhe
lndlvldual's own preference ls consLanL no maLLer how many among Lhe oLhers choose one
way or Lhe oLher. 1hese assumpLlons lead Lo Lhe deflnlLlon of unlform MulLl-ersons rlsoner's
ullemma (Mu), namely a slLuaLlon ln whlch:

1. 1here are people, each wlLh Lhe same blnary cholce and Lhe same pay-offs.
2. Lach person has a preferred cholce regardless of whaL Lhe oLhers do and
everybody prefers Lo make Lhe same cholce.
3. Whlchever cholce a person makes, Lhe more people among Lhe oLhers choose
Lhelr unpreferred alLernaLlve, Lhe beLLer off LhaL person ls.
4. 1here ls a number , greaLer Lhan 1, such LhaL lf Lhe number of Lhe lndlvlduals
chooslng Lhelr unpreferred alLernaLlve ls equal or hlgher Lhan , Lhen Lhese
lndlvlduals are beLLer off Lhan lf Lhey had all chosen Lhelr preferred alLernaLlve,
buL lf Lhe number of Lhe lndlvlduals chooslng Lhelr unpreferred alLernaLlve ls
less Lhan , Lhen Lhls ls noL Lrue anymore.

ln Lhls scenarlo, Lhe cenLral lssue ls Lhe parameLer . WhaL ls lL, exacLly? lL represenLs Lhe
mlnlmum slze of any coallLlons LhaL can galn by absLalnlng from Lhe preferred cholce. lL ls Lhe
smallesL dlsclpllned group LhaL, Lhough resenLful of Lhe free rlders, can be proflLable for Lhose
who [oln (Lhough more proflLable for Lhose who sLay ouL).
13

1he resL of Lhe ChapLer ls a sorL of dlagrammaLlc varlaLlon on Lhls baslc Lheme, namely Lhe
posslblllLy of coallLlons ln rlsoner's ullemma conLexLs.

LeL me now Lry Lo offer Lhe phllosophlcal analysls LhaL l have menLloned aL Lhe beglnnlng.
1he flrsL conslderaLlon ls a meLhodologlcal one. Schelllng's approach, as he hlmself sLaLes, ls
noL a markeL approach. 1he aLLenLlon Lo markeL-free conLexLs leads lmmedlaLely Lo a serles of
meLhodologlcal reflecLlons abouL Lhe consequences of such an analysls, l.e. Lhe dlvergence of
Lhe 'equlllbrlum' from Lhe 'opLlmum'.
An equlllbrlum ls a slLuaLlon ln whlch several Lhlngs LhaL have been lnLeracLlng are
evenLually ln balance, aL resL. 1yplcally, and ln splLe of rlsoner's ullemma slLuaLlons, from
equlllbrlum analysls economlsLs derlve Lhe concluslon LhaL self-lnLeresLed mlcromoLlves ofLen
lead Lo surprlslng and soclally useful coordlnaLlon ln Lhe aggregaLe, a sysLem ln equlllbrlum ls
Lhus a good one. 1hls ls Lhe old and fasclnaLlng ldea by Adam SmlLh of an lnvlslble hand.
Schelllng does noL Lry Lo assess wheLher a SmlLhlan approach Lo economlc phenomena ls
rlghL or wrong, whaL he does ls slmply Lo argue LhaL, when a SmlLhlan approach ls applled Lo
non-markeL conLexLs of behavlour, lL ls surely wrong. Cne can sLlll belleve LhaL someLhlng ln
Lhe ldea of an lnvlslble hand ls Lrue, buL one has also Lo admlL LhaL Lhls ls Lrue only ln slLuaLlons
dlfferenL from Lhose LhaL Schelllng conslders.
Schelllng lnslsLs on Lhe facL LhaL equlllbrla are slmply convenlenL LheoreLlcal resulLs, Lhere
ls noLhlng parLlcularly aLLracLlve abouL an equlllbrlum.
16
noL every slLuaLlon LhaL ls ln
equlllbrlum ls a [usL or an opLlmal slLuaLlon, Lhe body of a hanged man ls ln equlllbrlum when lL

14
1homas C. Schelllng, , p. 217.
13
1homas C. Schelllng, , p. 218.
16
1homas C. Schelllng, , p. 26.

Bumana.Nente - Issue 1u - }uly 2uu9

204

flnally sLops swlnglng, buL lL ls hard Lo acknowledge LhaL someone could say LhaL Lhe man ls all
rlghL. lrom Lhls meLhodologlcal polnL of vlew, Mlctomotlves ooJ Moctobebovlot can be
consldered as an aLLempL Lo provlde a large number of formal Lools for all Lhe slLuaLlons where
reachlng Lhe equlllbrlum ls noL Lhe general and deflnlLlve purpose LhaL we are Lrylng Lo
achleve. Powever, ln order Lo recognlse Lhe asymmeLry beLween equlllbrla and deslred
aggregaLe resulLs, we need Lo look beyond a purely markeL-orlenLed approach. 1hls ls Lhe goal
of Schelllng's book.
1he second conslderaLlon concerns Lhe rlsoner's ullemma. Schelllng shows how we can
exLend Lhls class of games ln such a way LhaL Lo [oln a coallLlon ls no more Lhe klnd of cholce
we do noL prefer Lo make. 1he fallure of cooperaLlve cholce ln Lhe baslc verslon of rlsoner's
ullemma ls due Lo Lhe facL LhaL lndlvlduals acL solely Lo promoLe Lhelr own lnLeresLs and Lo Lhe
facL LhaL Lhls behavlour ls raLlonal. lrom Lhls perspecLlve, lL ls lrraLlonal Lo [oln a coallLlon and
Lo keep Lhe agreemenLs. 1he raLlonal mlcromoLlve ls Lo free rlde. Lven lf everyone recognlsed
LhaL Lhe aggregaLe resulLlng from raLlonal mlcromoLlves ls noL Lhe deslrable aggregaLe, Lhls
acknowledgmenL would noL consLlLuLe a sufflclenL reason for chooslng Lhe unpreferred
alLernaLlve of cooperaLlng.
neverLheless, ln observed real-llfe slLuaLlons, coallLlons do work. Pow can we explaln LhaL?
1he scepLlc may suggesL LhaL Lhe lnLrlnslc llmlL of game Lheory ls Lhe lncapablllLy of glvlng an
adequaLe deflnlLlon of raLlonallLy. lf we adopL a purely game LheoreLlc concepL of raLlonallLy,
coallLlons are lrraLlonal. 1he scepLlc derlves Lhe concluslon LhaL whaL we need ls a sorL of
converslon" of raLlonallLy.
17
lL seems LhaL Lo make Lhe cooperaLlve cholce raLlonal we have Lo
change our concepL of raLlonallLy. ln oLher words, lf we wanL Lo capLure Lhe loglc underlylng
coallLlons we have Lo glve up game Lheory. CLherwlse, Lhe expresslon rlsoner's ullemma
wlLh coallLlon" would be a loglcal conLradlcLlon llke Culne's round square cupola on 8erkeley
College.
18

ls Lhe scepLlc rlghL? Schelllng mlghL reply LhaL lL ls cerLalnly Lrue LhaL Lhe rlsoner's
ullemma, ln lLs baslc verslon, does noL capLure Lhe splrlL of all Lhose soclal conLexLs ln whlch
we have a preference Lo [oln a coallLlon raLher Lhan sLaylng ouL. ln oLher words, Lhere ls
cerLalnly a sense ln whlch Lhe rlsoner's ullemma ls noL adequaLe Lo Lhe facLs. Powever, Lhls
does noL mean LhaL cooperaLlve cholces are lrraLlonal or LhaL Lhelr raLlonallLy exceeds Lhe
resources of game Lheory. !olnlng a coallLlon can be raLlonal noL ln Lhe llghL of a dlfferenL
concepL of raLlonallLy, buL ln Lhe llghL of Lhe very same ldea of raLlonallLy LhaL we flnd ln Lhe
rlsoner's ullemma. 1hus, Lhere ls no need Lo abandon Lhe game-LheoreLlc framework. We [usL
have Lo reflne lL.
Schelllng's exLenslon of Lhe rlsoner's ullemma ln lLs Mu verslon has Lo counL as a proof of
Lhe facL LhaL lL ls posslble Lo undersLand Lhe raLlonallLy of coallLlons from Lhe perspecLlve of
Lhe game LheoreLlc noLlon of raLlonallLy. ln Mu slLuaLlons people acL purposlvely llke ln Lhe
baslc verslon of Lhe game. WhaL ls dlfferenL ls Lhe soclal conLexL ln whlch Lhey acL, and noL Lhe
lndlvldual reasons leadlng Lhem Lo perform ln Lhe way Lhey do. ln Schelllng's Lermlnology,
whaL ls new ln Mu slLuaLlons ls Lhe aggregaLe macrobehavlor and noL Lhe lndlvldual
mlcromoLlves. lndlvldual mlcromoLlves are sLlll self-lnLeresLed and raLlonallLy ls sLlll goal-
orlenLed, buL Lhe aggregaLe macrobehavlor sLrlcLly depends on how many mlcromoLlves
comprlse Lhe aggregaLe. lrom Lhls perspecLlve we can undersLand beLLer Lhe reason why
Schelllng conslders Lhe case of commons" (Lyplcally caLegorlzed as rlsoner's ullemma
slLuaLlons) ln ChapLer Lhree whlle deallng wlLh crlLlcal mass models. 1he parameLer k

17
!ean PampLon, nobbes ooJ tbe 5oclol coottoct 1toJltloo, Cambrldge unlverslLy ress, 1986.
18
W. v. C. Culne, Cn WhaL 1here ls, ln ltom o loqlcol lolot of vlew, Parvard unlverslLy ress, 1933.
Commentaiy -Niciomotives anu Naciobehavioi

203

menLloned above and represenLlng Lhe mlnlmum number of people who have Lo choose Lhelr
unpreferred alLernaLlve ln order Lo make coallLlon Lhe preferred cholce ls consldered a polnL of
crlLlcal mass. lL seems LhaL Schelllng's ls concerned, among oLher Lhlngs, wlLh a sorL of new
formulaLlon of Lhe rlsoner's ullemma, LhaL ls, wlLh Lhe aLLempL Lo lnLegraLe Lhe rlsoner's
ullemma wlLhln Lhe large famlly of crlLlcal mass models.
Schelllng's revlslLed rlsoner's ullemma suggesLs a flnal conslderaLlon concernlng Lhe
general meLhodology of game Lheory afLer Lhe appearance of Mlctomotlves ooJ
Moctobebovlot ln Lhe laLe '70s and lLs role wlLhln Lhe Lheory lLself. 1here are several reasons
why Mlctomotlves ooJ Moctobebovlot was consldered revoluLlonary. 1he frlendly approach Lo
Lhe sub[ecL, whlch makes lL a book noL [usL for Lhe bookshelf or Lhe docLoral semlnar, buL for
Lhe bedslde Lable, surely counLs as one of Lhese reasons.
19
lL also seems Lo me LhaL behlnd
Lhe auLhor's lnLenLlon Lo wrlLe a fun book Lo read, a clear poslLlon abouL Lhe meLhodology of
soclal sclences ls acLlve. lndeed, Lhe real novelLy of Mlctomotlves ooJ Moctobebovlot was a
compleLely dlfferenL aLLlLude Lowards Lhe general meLhods LhaL, ln Schelllng's perspecLlve,
game Lheory had Lo employ. We can hlghllghL Schelllng's change of perspecLlve by asklng Lhe
quesLlon: does lndlvldual behavlour cause sysLems, or do sysLems cause lndlvldual
behavlour?
20
Cr, ln oLher words: whaL ls Lhe relaLlonshlp beLween facLs and model? Whlch one
of Lhese Lwo elemenLs - facLs or model - has Lhe prlorlLy? uoes game Lheory derlve facLs from
Lhe formal model, or ls Lhe model derlved from Lhe facLs?
Schelllng's poslLlon ls ln favour of Lhe prlorlLy of facLs over Lhe model. As 8lchard
Zeckhauser clalms, Schelllng sLudles a real-world problem and develops a concepLual model.
Pe Lhen Lakes LhaL concepLual model back Lo a dozen real-world problems Lo see how lL
applles, and Lhen rlcocheLs back Lo reflne Lhe model. Pe keeps Lhe process golng unLll he ls
happy wlLh hls model, and saLlsfled wlLh hls lnslghLs lnLo Lhe problems LhaL mosL lnLeresL
hlm.
21
WhaL we wanL Lo do now ls clarlfylng Lhe phllosophlcal moLlvaLlon leadlng Schelllng Lo
Lhls meLhodologlcal aLLlLude.
lf we look aL Lhls Loplc from a phllosophlcal polnL of vlew, we see LhaL we face here Lhe
classlcal problem posed by Lhe dlchoLomy beLween deducLlve and lnducLlve meLhods. 1he
phllosophlcal debaLe abouL Lhe dlsLlncLlon beLween deducLlon and lnducLlon ls vasL. Clven Lhe
llmlLed space avallable, lL wlll noL posslble Lo analyse Lhe maLLer ln greaL depLh. neverLheless
some words of clarlflcaLlon are needed.
1he concepLs of deducLlon and lnducLlon remlnd us lmmedlaLely of an old dlsLlncLlon. l am
referrlng Lo Lhe dlsLlncLlon LhaL ldenLlfles deducLlon wlLh Lhe lnference from sLaLemenLs
referrlng Lo general concepLs Lo sLaLemenLs referrlng Lo parLlcular cases, and, conversely,
lnducLlon wlLh Lhe lnference from parLlcular sLaLemenLs Lo more general sLaLemenLs. 1he
developmenL of probablllLy Lheory and sLaLlsLlcs allowed conLemporary phllosophy of sclence
Lo reflne Lhls dlsLlncLlon
22
. Accordlng Lo lL, when we Lalk abouL deducLlon and lnducLlon we Lalk
abouL Lwo dlfferenL Lypes of lnference, namely, Lwo ways Lo derlve a concluslon from cerLaln
premlses. lf Lhe concluslon follows from Lhe premlses wlLh Lhe force of loglcal necesslLy, we

19
1erry Connolly, 8evlew of Mlctomotlves ooJ Moctobebovlot by 1homas C. Schelllng, AdmlnlsLraLlve
Sclence CuarLerly, vol. 24, no. 3 (Sep., 1979), p. 304.
20
uel 1aebel and Charles Llder, uoes Lhe lndlvldual 8ehavlor Cause SysLems or do SysLems Cause
lndlvldual 8ehavlor? utboo Affolts Ooottetly, vol. 13 no. 2, uec. 1979, pp. 229-233.
21
8lchard Zeckhauser, 1homas Schelllng, 8lcocheL 1hlnker, ln 8oberL uodge, 1be 5ttoteqlst. 1be llfe ooJ
1lmes of 1bomos 5cbellloq, urlLan ress, new Pampshlre, 2006, pp. vll-xll.
22
See C.C. Pempel, . Cppenhelm, 5toJles lo tbe loqlc of xploootloo, ln hllosophy of Sclence, 13,
pp. 133-173, 1948, W.C. Salmon, loot uecoJes of 5cleotlflc xploootloo, MlnnesoLa 1989.

Bumana.Nente - Issue 1u - }uly 2uu9

206

say LhaL such an lnference ls deducLlve, oLherwlse, lf Lhe concluslon follows from Lhe premlses
wlLh a cerLaln degree of probablllLy, Lhe lnference ls lnducLlve.
lL ls my oplnlon LhaL Schelllng's revoluLlonary conLrlbuLe Lo game Lheory can be appreclaLed
only lf we conslder Lhe laLLer way of concelvlng Lhe dlsLlncLlon beLween deducLlon and
lnducLlon. Maklng Lhls dlsLlncLlon ln Lerms of loglcal deducLlon and deducLlon wlLh a cerLaln
degree of probablllLy, lnsLead of Lhe classlcal explanaLlon ln Lerm of lnference from parLlcular
Lo general (lnducLlon) and from general Lo parLlcular (deducLlon), glves us Lhe opporLunlLy Lo
reflecL on Lhe role of sLaLlsLlcs wlLhln Schelllng's work. SLaLlsLlcs ls lndeed Lhe sclence LhaL
derlves concluslons wlLh a cerLaln degree of probablllLy. lf we geL an lnslghL lnLo Schelllng's
meLhod of analysls, we see LhaL concluslons abouL Lhe macrobehavlor ln Lhe aggregaLe are
sLaLlsLlcally - and noL loglcally - derlved from lndlvldual mlcromoLlves. 1he presence of
sLaLlsLlcal lnferences ln Mlctomotlves ooJ Moctobebovlot ls ln facL pervaslve. 8efore
Mlctomotlves ooJ Moctobebovlot, game Lheory, llke any maLhemaLlcal Lheory, was a
deducLlve sclence. AfLer flxlng Lhe models, sLaLemenLs concernlng facLs were Lhereby derlved
ln a purely deducLlve way. 1he employmenL of sLaLlsLlcal Lools ln Mlctomotlves ooJ
Moctobebovlot radlcally changed Lhe scenarlo. 1he model has now Lo be ad[usLed Lo Lhe facLs,
and noL vlce versa. 1he formal apparaLus ls sub[ecL Lo a conLlnuous process of reflnemenL ln
order Lo adequaLely descrlbe soclal phenomena LhaL are essenLlally varlable. 1hus, Schelllng's
sLraLegy ls Lo sLarL wlLh lndlvldual mlcromoLlves and Lo derlve, wlLh a cerLaln degree of
probablllLy, Lhe resulLlng aggregaLe macrobehavlor.
AL Lhls polnL we sLarL Lo see Lhe consequences of Schelllng's lnducLlve approach. nelLher
Lhe feaLures of formal models are consldered lmmuLable nor soclal phenomena are derlved
from Lhem ln a merely loglcal way. lormal models are sub[ecL Lo revlslons. lf Lhe models are
sLaLlsLlcally deLermlned, lL ls clear LhaL Lhe answer Lo Lhe quesLlon abouL prlorlLy, lf any,
beLween model and facLs ls LhaL facLs have Lhe prlorlLy over Lhe model. 8uL Schelllng's
preference for Lhe lnducLlve meLhod would be a petltlo ptloclpl lf we dld noL focus on Lhe facL
LhaL Lhls preference ls grounded on and [usLlfled by Lhe use of sLaLlsLlcs. 1he ground for
adopLlng an lnducLlve meLhod ls Lhus Lhe employmenL of sLaLlsLlcs.

ln Lhls scenarlo we can beLLer undersLand Lhe role of Schelllng's examples. 1hese are noL only a
dldacLlc sLraLegy Lo make Lhe book fun Lo read, buL Lhey are parL of a dlsLlncLlve meLhodology.
PlsLorlcally, Lhe endorsemenL of an lnducLlve meLhod ls Lhe elemenL whlch dlsLlngulshes
Mlctomotlves ooJ Moctobebovlot from Lhe oLher game LheoreLlcal approaches Lo soclal
phenomena. 8ecenL conLrlbuLes llke Malcolm Cladwell's 1be 1lpploq lolot
23
or
lteokecooomlcs
24
by uubner and LewlLL cannoL be undersLood wlLhouL Lhe lnLermedlaLe role
of Schelllng's Mlctomotlves ooJ Moctobebovlot.





23
Malcolm Cladwell, 1be 1lpploq lolot. now llttle 1bloqs coo Moke o 8lq ulffeteoce, 8ack 8ay 8ooks,
2002.
24
SLeven LevlLL and SLephen !. uubner, lteokecooomlcs. A koqoe cooomlst xplotes 1be nlJJeo 5lJe of
vetytbloq, Wllllam Morrow/Parper Colllns, 2003.

207

Commentaiy
Etica ed economia
AmarLya Sen
LaLerza LdlLore, 8oma-8arl, 2002

8edazlone
redazlone[humana-menLe.lL


8asaLo su una rlelaborazlone delle koyet lectotes LenuLe dall'auLore
all'unlverslLa dl 8erkeley nel 1986, quesLo breve volume dl AmarLya
Sen vuole essere ll punLo dl conglunzlone Lra due campl
LradlzlonalmenLe afflnl eppure sempre plu dlsLanLl quall l'eLlca e
l'economla.
CuesLl due amblLl del sapere non s