CONTENTS

PLACING BLAME

XV

List of Figures
Table of Statutes
Table of Cases

XVll

XIX

A General Theory of the Criminal Law

Introduction

MICHAEL MOORE

I. A Theory of Criminal Law Theories

Part I. The Theory of Criminal Law's Function

2. Closet Retributivism
3. The Moral Worth of Retribution
4. Justifying Retributivism

Part

83

104
153

n. The Theory of the General Part:

The Theory of Responsibility

A. The Nature of Moral Responsibility

5. The Independent Moral Significance of Wrongdoing
B. The Elements of Re.\ponsibility:

Wrongdoing and CulpahWty

6.
7.
8.

9.
10.

I. The Nature of Wrongdoing

a. The Nature of Human Action
More on Act and Crime
b. The Nature of Causation
Causation, Rights-Violations, and Wrongdoing
Foreseeing Harm Opaquely
2. The Nature of Culpability
a. The Nature of Mental States
Prima Facie Moral Culpability
Mind. Brain. and the Unconscious

I R9

191

249

249

249

251

331

333

363

401

40 I

403

420

CLOSET RETRIBUTIVISM

Retributivism as a theory of punishment has enjoyed some resur

gence among the legal theorists of the past thirty years-in marked
contrast to the theory's earlier decline throughout most of this cen
tury, when it was often regarded as a remnant of our barbarous
past. I This chapter is designed to persuade those who have not
worked through any theoretical arguments in favour of retribu
tivism how they themselves might be retributivists----closet retribu
tivists, perhaps, but retributivists nonetheless. By 'retributivist' I
refer to one who believes that the justification for punishing a crim
inal is simply that the criminal deserves to be punished. What I
mean by 'closet' retributivism is a retributive theory held by those
who have not thought through a theory of punishment, but who
show themselves to be retributivists in the judgements they make
and the reasons for which they make them. By and large I do not
in this chapter attempt to show that such persons should be ret
ributivists; only to show that they are.
In order to fulfil the limited ambition of this chapter, as well as
the more ambitious undertaking of the next chapter, it is necessary
to develop a classification scheme for theories of punishment.
Having done this, I shall return to the kind of thought experiments
that make retributivism the most plausible theory of what many in
fact believe about punishment.
I See particularly Herbert Morris's splendid essays, 'Persons and Punishment',
The Monist, Vol. 52 (1968), 475-501, reprinted in his A" Guilt alld Innocence (Los
Angeles: University of California Press. 1976); and 'A Paternalistic Theory of
Punishment', American Philosophical Quarter(v, Vol. I R (1981), 263-71. See also the
very interesting essays by Herbert Fingarette, 'Punishment and Suffering',
Proceedin~s and Addresses of the American Philosophical Association. Vol. 50 ( 1977).
499-525; by Jeffrie Murphy. Retrihution. Justice and Therapy (Dordrecht: Reidel,
1979); and by Jean Hampton. 'The Retributive Idea', in Murphy and Hampton,
For~iveness (/nd Mercy (Cambridge: Camhridge University Press, 1988). See also
George Sher, 'Deserved Punishment', in his Desert (Princeton: Princeton University
Press, 1987); Wojciech Sadurski. Giving Desert Its Due (Dordrecht: Reidel, 1985);
Davis. 'Harm and Retribution', Philosophy and Puhlic Affairs. Vol. 15 (19861.
D6--66. See generally Galligan. 'The Return to Retribution in Penal Theory'. in
C. Tapper cd .. Crime, Proof and Punishment (London: Butterworth. 1981).

84

THE THEORY OF CRIMINAL LAW'S FUNCTION

I. THE PRIMA FACIE JUSTIFICATIONS OF PUNISHMENT

There is by now a familiar list of prima facie reasons given to jus

tify the institution of punishment. Such a list usually includes: inca
pacitation, special deterrence, general deterrence, denunciation,
rehabilitation, and retribution. A word about each of these reasons
is in order. Incapacitation is the simplest of theories, because, as the
name suggests, the good that punishment achieves is that it inca
pacitates an offender by doing something to him that prevents him
from committing further crimes. Special deterrence has a similar
aim in preventing crime, but a different means: punishing an
offender deters him from committing further crimes upon his
release. Likewise, general deterrence aims at the prevention of
crime by punishing an offender, except that those who are deterred
are others in the general population rather than the offender him
self. All three of these traditional theories of punishment share a
common goal thought to justify punishment, namely, the reduction
of crime.
The ideas captured in the theory labelled 'denunciation', some
times called the expressive theory of punishment,2 are somewhat
more complic'ated. One strand of this theory urges that punishment
must express society's condemnation because doing so educates cit
izens in the wrongfulness of the conduct that the criminal law
attempts to discourage. So stated, the denunciation theory is no
more than a somewhat broader form of the general deterrence the
ory: both aim at the prevention of crime, one by scaring people out
of it and the other by more subtle educational techniques. 3 Another
strand of the denunciation theory asserts that the denouncing of
crime via symbolic blaming coupled with harsh treatment, serves an
end distinct from the prevention of future criminal conduct.
Theorists of this stripe urge that crime must be denounced by pun
2 See, e.g., Joel Feinberg's 'The Expressive Function of Punishment', The Monist,
Vol. 49 (1965), 397--423, reprinted in his Doing and Deserving (Princeton: Princeton
University Press, 1970). H. L. A. Hart attributes this theory of punishment to Sir
James Fitzjames Stephen and to Lord Denning in his Punishment and Responsibility
(Oxford: Oxford University Press, 1968), 170-3.
3 Indeed, some include such educational techniques as a fonn of general deter
rence. See Johannes Andenaes, 'Does Punishment Deter Crime?', in G. Ezorsky, ed.,
Philosophical Perspective on Punishment (Albany, NY: State University of New
York Press, 1972); J. Andenaes, Punishment and Deterrence (Ann Arbor, MI:
University of Michigan Press, 1974).

CLOSET RETRIBUTIVISM

85

ishment, because doing so maintains a sense of social cohesion. If

punishments are inflicted, then citizens do not have the sense that
the social contract has been broken with impunity by others. The
good achieved on this branch of the theory is not the prevention of
future crimes; rather, it is thought that a sense of community is
itself a good thing that punishment helps to achieve.
Rehabilitation involves two quite different ideals of rehabilita
tion that are usually confused. These two rehabilitative ideals can
best be separated by thinking about two different ways of render
ing offenders non-dangerous. First, imagine that what is done is to
place offenders in extraordinarily awful places of detention, with
harsh treatment by inmates and guards. Here, non-dangerousness
is achieved because such offenders either become 'penitent', or they
are no longer willing to commit crimes because they are unwilling
to risk again such awful treatment. For comparison, imagine that
the same level of non-dangerousness can be achieved if prisoners
are placed in much nicer facilities, with kinder personnel (all of
them soft-spoken, in white coats, and manifesting sincere concern),
a place in which extensive therapy programmes are undertaken.
Imagine further that the second such programme, although much
more expensive than the first, not only makes the offender non
dangerous but also makes him a flourishing, happy, and self
actualizing member of our society.
The first sort of rehabilitative ideal is one that is achieved when
we make criminals safe to return to the streets. This sort of reha
bilitative theory justifies punishment, not by appeal to how much
better off the criminal will be at the end of the process, but rather
by how much better off all of us will be if 'treatment' is completed
because the streets will be that much safer. Such a theory seeks to
rehabilitate the criminal only as a cost-effective means of shorten
ing the expensive incarceration that would otherwise be necessary
to protect everyone against crime. The second sort of rehabilitative
ideal, by way of contrast, is a paternalistic theory. It seeks to reha
bilitate the offender, not just so that he can be returned safe to the
streets, but so that he can lead a flourishing and successful life.
Such a theory justifies punishment, not in the name of all of us, but
rather in the offender's own name; since it does so in his name, but
contrary to his own expressed wishes (few offenders want to be
punished), this kind of rehabilitative theory is paternalistic in char
acter.

86

87

THE THEORY OF CRIMINAL LAW'S FUNCTION

CLOSET RETRTBUTIVTSM

This paternalistic type of rehabilitative theory has no proper part

to play in any theory of punishment, even in the minimal sense of
constituting a prima facie justification of punishment. 4 There are
three reasons why this is so. First, such a paternalistic reform the
ory allocates scarce societal resources away from other, more
deserving groups that want them (such as retarded and autistic chil
dren or the poor) to a group that hardly can be said to deserve such
favoured status, and, moreover, does not want such 'benefits'. As a
simple matter of distributive justice it is difficult to argue that crim
inals should be favoured in the allocation of scarce social resources
in these ways. To make such distributive justice arguments go, one
has to claim that criminals suffer distinctive disadvantages entitling
them to extra societal resources. One idea of this sort is that crim
inal behaviour is thought to constitute a disease, so that as diseased
persons criminals deserve treatment. Yet the idea that 'crime is dis
ease' rests on unjustifiably bloated notions of health and sickness,
notions that conflate illness with badness. 5 An alternative idea hav
ing the same distributional implications is the thought that crimi
nals do merit extra social resources being devoted to their welfare
because they come predominantly from groups that have histori
cally received less than their due from our society. Yet this idea,
even if statistically true for general classes of people, ignores more
precise individuation of who it is in our society that deserves extra
social resources: those who are as disadvantaged economically and
socially as the average criminal but who commit no crimes. Why
allow criminal behaviour to serve as a proxy for social disadvan
tage, if one can use the latter criterion directly to guide remedial
legislation?6
A second reason that should give one pause about rehabilitation
as a prima facie reason for punishment stems from its paternalistic

nature. As John Stuart Mill explored in detail,7 paternalistic justi

fications of any piece of legislation should be regarded with suspi
cion. Criminals are not in the standard classes in society for which
paternalistic state intervention is appropriate, such as the severely
disordered, the young or others whose capacity for rational choice
is diminished; such a paternalistic theory is suspect on this ground
alone.
Third, such recasting of punishment in terms of 'treatment' for
the good of the criminal makes possible a kind of moral blindness
that is dangerous in itself. As C. S. Lewis pointed out years ago!!
adopting such a 'humanitarian' conceptualization of punishment
makes it easy to inflict treatments and sentences that need bear no
relation to the desert of the offender. We may do more to another
'for his own good' than we ever allow ourselves to do when we see
that it is really for our good for which we act. 9
Retributivism-the final theory used to justify punishment-is
the view that punishment is justified by the desert of the offender.
The good that is achieved by punishing, on this view, has nothing
to do with future states of affairs, such as the prevention of crime
or the maintenance of social cohesion. Rather, the good that pun
ishment achieves is that someone who deserves it gets it.
Punishment of the guilty is thus for the retributivist an intrinsic

4 A conclusion also reached, on somewhat different grounds, by Herbert Hart.

See Hart, supra n. 2, at 2fr-7.
5 Arguing against this attempted 'psychiatrization of ethics' is one of the princi
pal points of my book, Lalli and Psychiatry: Rethinking the Relationship (Cambridge:
Cambridge University Press, 1984), chs. 4-6.
6 A third idea, often confused with the two presented in the text, is that crimi
nals, unlike the rest of us, are passive victims of their own backgrounds who 'can't
help' but do what they do. This idea I call 'selective determinism' in ch. 12, and there
reject it as an incomprehensible metaphysics (a kind of patchy, on-again, ofT-again
free will metaphysics).

7 See below. ch. 16. where I distinguish paternalistic justifications (concerned with
the good of the criminal) from legal moralist justifications (concerned with having
the dictates of morality complied with, even though this may incidentally be good
for the criminal). If we draw this distinction, I am uncertain whether Herbert
Morris's revised retributive theory is truly paternalistic, as he himself labels it. See
Morris, 'A Paternalist Theory of Punishment', supra n. I.
8 C. S. Lewis, The Humanitarian Theory of Punishment'. Res Judicatae, Vol. 6
(1953),224-30.
9 Lewis's kind of argument against the rehabilitative ideal has limited efficacy
because it may only yield the conclusion that 'involuntary rehabilitative treatment'
'punishment', except for its purpose-is subject to certain non-treatment-justified
side-constraints. Even a rehabilitationist might concede, for example, that. no mat
ter how effective, behaviour-modification programmes using anectine therapy
(which paralyses the involuntary muscle movements in breathing while oxygen is
pumped into the patient, giving him the sensation of suffocating) are impermissible
because violative of basic human rights. And such side-constraints need not commit
the rehabilitationist to a mixed rehabilitationist/retributivist theory, because ret
ributivist punishments are also arguably subject to the same side-constraints. On the
latter possibility, see Jeffrie Murphy, 'Hatred: A Qualified Defense', in Murphy and
Hampton, n. I supra, at 1Ofr-7. Still, to keep involuntary rehabilitative treatment
fully within intuitively acceptable bounds, the rehabilitationist probably has to
become a mixed theorist because he has to use moral desert as a check on reforma
tive treatment.

~,

88

THE THEORY Of CRIMINAL LAW'S fUNCTION

CLOSET RETRIBUTIVISM

good, not the merely instrumental good that it may be to the utili
tarian or rehabilitative theorist.
Retributivism differs from a variety of views that are often
paraded as retributivist, but that in fact are not. Such views are typ
ically put forward by people who cannot understand how anyone
could think that moral desert by itself could justify punishment.
Such persons scramble about for other goods that punishment
achieves and label these, quite misleadingly, 'retributivism'. The
leading confusions seem to me to be seven in number.

for such additional reasons. That future crime might also be pre
vented by punishment is a happy surplus for a retributivist, but no
part of the just~fica~ion for pu.nishing.
.
3. Retributivlsm IS not the vIew that pUnishment of otTenders sat
isfies the desires for vengeance of their victims. In this view the
hann that is punishment is justified by the good it does psycholog
ically to the victims of crime, whose suffering is thought to have a
special claim on the structuring of the criminal justice system. 12
This is not retributivism. A retributivist can justify punishment as
deserved even if the criminal's victims are inditTerent (or even
opposed) to punishing the one who hurt them. Indeed, a retribu
tivist should urge punishment on all offenders who deserve it, even
if no victims wanted it.
4. Relatedly, retributivism is not the view that the preferences of
all citizens (not just crime victims) should be satisfied. A preference
utilitarian might well believe, as did Sir James Fitzjames Stephen,l3
that punishment should be exacted 'for the sake of gratifying the
feeling of hatred---call it revenge, resentment, or what you will
which the contemplation of such [criminal] conduct excites in
healthily constituted minds .. .', or that 'the feeling of hatred and
the desire of vengeance ... are important elements of human nature
which ought ... to be satisfied in a regular public and legal man
ner'. Yet a retributivist need not believe such things, but only that
morally culpable persons should be punished, irrespective of what
other citizens feel, desire, or prefer.
5. Relatedly, retributivism is not the view that punishment is jus
tified because without it vengeful citizens would take the law into
their own hands. Usually it is those who are hostile to retributivism,
such as Justice Marshall,'4 who link it to this indefensible idea.

I. First, retributivism is sometimes identified with a particular

measure of punishment such as lex talionis, an eye for an eye,1O or ,
with a kind of punishment such as the death penalty. Yet retribu
tivism answers a question prior to the questions to which these
could be answers. True enough, retributivists at some point have to
answer the 'how much' and 'what type' questions for punishments
of specific offences, and they are committed to the principle that
punishment should be graded in proportion to desert; but they are
not committed to any particular penalty scheme nor to any partic
ular penalty as being deserved. Separate argument is needed to
answer these 'how much' and 'what type' questions, after one has
described why one is punishing at all. It is quite possible to be a
retributivist and to be against both the death penalty and lex talio
nis, the idea that crimes should be punished by like acts being done
to the criminal.
2. Contrary to Anthony Quinton and others, II retributivism is
not 'the view that only the guilty are to be punished'. A retributivist
will subscribe to such a view, but that is not what is distinctive
about retributivism. The distinctive aspect of retributivism is that
the moral desert of an otTender is a sufficient reason to punish him
or her; the principle Quinton advocates makes such moral desert
only a necessary condition of punishment. Other reasons-typi
cally, crime prevention reasons-must be added to moral desert, in
this view, for punishment to be justified. Retributivism has no room
10 J. Wilson and R. Herrnstein, Crime and Human Nature (New York: Simon and
Schuster, 1985), 496.
" Quinton, 'On Punishment', Analysis, Vol. 14 (1954), 1933-42. Herbert Hart
calls this 'weak retributivism' (Hart, supra n. 2), and JefT Murphy and J. L. Mackie
call it 'negative retributivism' (Murphy, 'Retributivism and the State's Interest in
Punishment', in J. R. Pennock and J. W. Chapman, eds., Criminal Justice: Nomos
XXVII (New York: New York University Press, 1985), 159; Mackie, 'Morality and
thp Rplrihulive Emotions', Criminal Justice EtMcs (Winter/Spring, 1982),3-10.

89

12 Ted Honderich, Punishment: The Supposed Justifications (London: Hutchinson,

1969), 30.
U James Fitzjames Stephen, Liberty. Equality, Fra/ernity (Cambridge: Cambridge
University Press, 1967), 152. JelTrie Murphy seems intent on running retributivism
and this kind of preference-utilitarianism together in his recent essay, 'Getting Even:
The Role of the Victim', Sucial Philosuphy and Policy, Vol. 7 (1990), 209-25, at 212,
224-5. Murphy has come to blur this distinction because he has come to doubt that
he can motivate retributivism as an 'abstract concern for justice' (as opposed 10 a
concern for quelling the fires of vengeful emotions). Murphy's proper conclusion
should be that rctributivism is false, not that it is something that it is not.
14 See Thurgood Marshall's opinion in Gregg v. Georgia, 428 US 153 (1976).
Peggy Radin calls Marshall's kind of view 'revenge-utilitarianism'. Radin, 'Cruel
Punishment and Respect for Persons: Super Due Process for Death', Southcm
Cafifornian Lml' Rel'iClI". Vol. 53 (1980), 1143-85 at 1169-73. The argument is that

91

THE THEORY OF CRIMINAL LAW'S FUNCTION

CLOSET RETRIBUTIVISM

Punishment for a retributivist is not justified by the need to prevent

private violence, which is an essentially utilitarian justification.
Even in the most well-mannered state, those criminals who deserve
punishment should get it, according to retributivism.
6. Nor is retributivism to be confused with denunciatory theories
of punishment. 15 In this latter view punishment is justified because
punishment is the vehicle through which society can express its con
demnation of the criminal's behaviour. This is a utilitarian theory,
not a retributive one, for punishment is in this view to be justified
by the good consequences it achieves-either the psychological
satisfactions denunciation achieves, or the prevention of private
violence, or the prevention of future crimes through the educational
benefits of such denunciation. A retributivist justifies punishment
by none of these supposed good consequences of punishing.
7. Finally, retributivism should not be confused with a theory of
formal justice (the treating of like cases alike). Retributivism is not,
as McCloskey has urged,16 'a particular application of a general
principle of justice, namely, that equals should be treated equally
and unequals unequally'. True, a retributivist who also subscribes
to the principle of formal justice is committed to punishing equally
those persons who are equally deserving. However, the principle of
formal justice says nothing about punishing anybody for anything;
such a principle only dictates that, if we punish anyone, we must
do so equally. Why we should punish anyone is the question ret
ributivism purports to answer, a question not answered by the dis
tinct principle of formal justice. Moreover, notice that even if we
have punished some persons in the past, so that the principle of for
mal justice demands that relevantly similar persons be punished in
the future, that does not tell us the relevant similarities demanding
similar treatment. Like cases should indeed be treated alike, but it
takes some substantive theory of punishment to spell out when

cases are truly alike. A utilitarian who also subscribes to the prin
ciple of formal justice should think that the dangerousness and
deterrability of an offender (and of the class of potential offenders
that she instantiates) are the relevant respects in which one case is
relevantly like another.

90

the state must punish because otherwise private citizens will take the law into their
own hands, and that such private vengeance leads to chaos and disorder.
Punishment on such a view is justified by its ability to prevent these bad things.
Retributivism has nothing to do with this essentially forward-looking justification.
Moreover, this 'prevention of private vengeance' theory is to my mind not even a
prima facie justifying reason for punishment. The obvious thing to do if citizens are
going to violate the law by taking it into their own hands, is to deter those citizens
by punishing them.
I S See n. 2 supra.
16 H. J. McCloskey, 'A Non-Utilitarian Approach to Punishment', Inquiry. Vol.
8 (1965), 249-63.

Retributivism is a very straightforward theory of pllnishment: We

are justified in punishing because and only because offenders
deserve it. Moral responsibility ('desert') in such a view is not only
necessary for justified punishment, it is also sufficient. Such suffi
ciency of justification gives society more than merely a right to pun
ish culpable offenders. It does this, making it not unfair to punish
them, but retributivism justifies more than this. 17 For a retribu
tivist, the moral responsibility of an offender also gives society the
duty to punish. Retributivism, in other words, is truly a theory of
justice such that, if it is true, we have an obligation to set up insti
tutions so that retribution is achieved.

II. THE TWO PURE THEORIES OF PUNISHMENT

It is common to reduce the survivors on this list of prima facie jus

tifications of punishment to two general theories, the utilitarian the
ory and the retributive theory. To see how this is done, one need
only consider the good state of affairs that is to be achieved by
incapacitation, special deterrence, general deterrence, and rehabili
tation (to the extent that it is of the first sort of rehabilitative the
ory, and not the second). For all four of these rationales for
punishment share the prevention of crime as the good thing to be
achieved that justifies punishment. In each case, the ultimate justi
fication for inflicting the harm of punishment is that it is out
weighed by the good to be achieved, namely, the prevention of
future crimes by that offender or by others. This justification of an
institution by the social welfare that it will enhance, makes all such
theories instances of a utilitarian theory of punishment.
Thus, the denunciation theory of punishment is a second kind of
utilitarian theory of punishment, insofar as the good that it seeks
17 Compare K. G. Armstrong, 'The Retributivist Hits Back', in H. B. Acton, ed.,
The Philosophy of Punishment (London: Macmillan, St Martin's Press, 1969}, at
155-7, who conceives of retributivism as only giving the state the right to punish but
not the duty to do so. This is perhaps an eighth mischaracterization of retributivism.

92

THE THEORY OF CRIMINAL LAW'S FUNCTION

CLOSET RETRIBUTIVISM

to achieve is not simply the prevention of crime. To the extent one

grants intrinsic value to social cohesion, and does not regard that
as a value only because it contributes to the maintenance of public
order, the denunciation theory can be distinguished from the other
utilitarian theories just considered by the differing social good that
it seeks to achieve. Nonetheless, it is still a utilitarian theory, since
it outweighs the harm that is punishment by some form of net social
gain that punishment achieves.
Both the prevention of crime and the maintenance of social cohe
sion are types of collective goods. The general utilitarian theory of
punishment is one that combines these and any other form of col
lective goods that punishment might achieve, and calls them all a
'social gain'. Whenever the social gain outweighs the harm punish
ment causes to an offender or his family, such a theory would say
that there is a 'net social gain'. Such a vocabulary allows us a suc
cinct definition of any form of utilitarian theory: punishment is jus
tified if and only if there is some net social gain achieved by it.
A retributivist theory is necessarily non-utilitarian in character,
for it eschews justifying punishment by its tendency to achieve any
form of net social gain. Rather, retributivism asserts that punish
ment is properly inflicted because, and only because, the person
deserves it. That some people deserve punishment on such a theory,
is both a necessary and a sufficient condition justifying criminal
sanctions. A succinct definition of the retributivist theory of pun
ishment, paralleling that given of the utilitarian theory, is that pun
ishment is justified if and only if the person or persons receiving it
deserve it.

one of these merits any serious attention. There is first of all the
popular form of ~ixed th~ory which asser~s tha~ punishment is jus
tified if, and only If, there IS both a net SOCIal gam achieved by pun
ishment, and punishment is given to offenders who deserve it.
Giving just deserts and achieving a net social gain, in such a case,
are each individually necessary but only jointly sufficient conditions
justifying punishment. The second logically possible mixed theory,
would be one asserting that punishment is justified if, and only if,
there is a net social gain achieved, or punishment is given to offend
ers who deserve it. Such a theory has no name, because there is no
one. to my knowledge, who has ever adopted it. The reason such a
theory is unnamed and unclaimed is because it shares the defects of
each of the pure theories, utilitarianism and retributivism. 19 I shall
accordingly put this 'mixed theory' aside from further considera
tion.
The first kind of mixed theory itself has two branches. By far the
most usual and popular form of the theory asserts that we do not
punish people hecause they deserve it. 20 Desert enters in, this the
ory further asserts, only as a limit on punishment: we punish
offenders hecause some net social gain is achieved, such as the pre
vention of crime, but only if such offenders deserve it. It is, in other
words, the achieving of a net social gain that justifies punishment,
whereas the desert of offenders serves as a limiting condition on
punishment but no part of its justification. The alternative branch
of the mixed theory is just the converse of the branch just exam
ined: one would urge that we punish hecause offenders deserve it,
but only if there is some net social gain achieved by doing SO.21

III. THE MIXED THEORY OF PUNISHMENT

one will allow a big gain in utility to justify a punishment that is somewhat unjust.
See C. L. Ten, Crime, Guilt, and Punishment (Oxford: Clarendon Press, 1987), 79-8 I,
for an example.

Once one grants that there are two sorts of prima facie justifications
of punishment--effecting a net social gain (utilitarian), and giving
just deserts (retributivist)-one can also see that in addition to the
two pure theories of punishment there can also be mixed theories.
There are two logically possible mixed theories,'8 although only
IR If one abandons criterial theories of punishment (seeking necessary and suffi
cient conditions) for looser, criteriological theories (seeking only 'factors' that in
some unspecified combination justify punishment), then there are more than two
sorts of mixed theories. One might, for example, have a kind of sliding scale whereby

93

19 The main problem with the pure utilitarian theory of punishment is that it
potentially sacrifices the innocent in order to achieve a collective good. The main
problem with the pure retributivist theory of punishment is that it potentially
requires punishment of the guilty even if no further good is achieved by doing so.
The unnamed and unclaimed mixed theory would have hoth of these problems. C. L.
Ten also notes this about what he also regards as a 'very strange' theory. Id. at 77.
20 If one construes Hart's attempt to distinguish dilTerent levels of justification
('general aim', versus 'distribution', of punishment) to be no more than my distinc
tion between justification and limitation, his is the best-known exemplar of this view.
Hart, supra n. 2.
21 See Andrew von Hirsch. Doing Justice (New York: Hill and Wang. 1976), ch.
6, who distinguishes these two forms of conjunctive mixed theories.

THE THEORY OF CRIMINAL LAW'S FUNCTION

CLOSET RETRIBUTIVISM

In such a case, the roles of net social gain and desert are
simply reversed: giving offenders their just deserts serves as the
justification of punishment, and the achieving of a net social gain
as the limiting condition.
A cynic might view these two branches of the mixed theory as
nothing more than an uncomfortable shuffle by mixed theorists:
when accused of barbarism in punishing persons for retributivist
reasons, they assert the first branch of the theory (they do not pun
ish because some persons deserve it, but because of a collective
good that is achieved). When accused of immorality in imposing
harsh treatment on someone as a means of making everyone else
better off, such theorists shift to the other foot, and claim they do
not punish someone in order to achieve a net social gain, but only
to give offenders their just deserts. The reason the cynic has a point
here is that there is a sense in which the two branches of the the
ory are the same, namely, the sense that they justify exactly the
same kinds of treatment for all cases. The only difference in theo
ries is in the motivations of those who hold them. And while that
may make a difference in our moral judgements of those who hold
the different branches of the mixed theory of punishment, it does
not make a difference in terms of the actual social institutions and
judgements such theories will justify. For purposes of arguing
against the mixed theory in this chapter it will suffice to lump both
of these branches of the mixed theory together and call them 'the
mixed theory of punishment'.

tarian thesis that punishment is justified if and only if some net social
gain is achieved. I shall here mention only two such counter
examples: scapegoating and preventive detention. With regard to the
first. it might be recalled that D. B. Cooper successfully skyjacked an
aircraft some years ago, and that this successful, unsolved crime
apparently encouraged the mass of skyjackings that has cost
American society a great deal in terms of dollars, lives, and conve
nience. Cooper wore large sunglasses in his escapade, and there was
accordingly only a very limited description available of him. Imagine
that shortly after his skyjacking we had the benefit of the knowledge
that we now have by hindsight, and decided that it would better to
punish someone who looked like Cooper (and who had no good
alibi), in order to convince others that skyjacking did not pay. For
a consistent utilitarian, there is a net social gain that would be
achieved by punishing such an innocent person, and there is no a pri
ori reason that the net social gain in such a case might not outweigh
the harm that is achieved by punishing an innocent person.
The preventive detention kind of counter-example is very similar:
imagine that a psychiatrist discovers that his patient has extremely
dangerous propensities. His patient is also the accused in a crimi
nal trial. It turns out, however, that the accused is not guilty of the
crime for which he is charged, and in fact has committed no crime
whatsoever. Should a judge who, we may suppose, is the only one
who knows both that the man is dangerous and innocent, find the
accused guilty? Doing so wil1 prevent the predicted criminal behav
iour of the defendant because the latter will be incarcerated. On a
utilitarian theory, it is difficult to see why such a judgement would
not be perfectly appropriate, as long as the prediction is reliable
enough, and as long as the crimes predicted are sufficiently serious
that the good of their prevention outweighs the harm of punishing
that person, even though he has committed as yet no crime.
The general form of the argument arising from these kinds of
thought experiments is that of a reductio ad absurdum argument.
The argument has three premises:

94

IV. THE ARGUMENT AGAINST THE PURE UTILITARIAN

THEORY

In exploring one's thoughts leading to retributivism, it is perhaps

easiest to start with some standard kinds of thought experiments
directed against a pure utilitarian theory of punishment. A thought
experiment is essentially a device allowing one to sort out one's true
reasons for believing that certain propositions are true. To be suc
cessful, such a thought experiment need not involve any actual case
or state of affairs, nor need the cases envisioned even be very likely;
they only need be conceivable in order to test our own thoughts.
It is standard fare in the philosophy of punishment to assert, by
way of several thought experiments, counter-examples to the utili-

95

I. Punishment should be inflicted if, and only if, doing so

achieves a net social gain.
2. A net social gain would be achieved in this case by the inflic
tion of punishment.
3. Punishment should not be inflicted in this case.

96

THE THEORY OF CRIMINAL LAW'S FU NCTlON

Each of these premises corresponds to steps in both of the thought

experiments above. The first premise is simply a restatement of the
utilitarian theory of punishment. The second premise presupposes
that there are some cases where a net social gain can be achieved
by punishing an innocent person, and asserts that this is such a
case. The third premise asserts our intuition that such persons
ought not to be punished.
All three premises together yield a contradiction:
4. Punishment should not be inflicted and punishment should be
inflicted.
The first two premises have as their joint conclusion that the per
son should be punished; this conclusion, when conjoined with the
third premise, produces the contradictory conclusion.
To avoid this contradictory conclusion, there are only three pos
sibilities, corresponding to each of the three premises leading to it.
One could give up the third premise and simply admit that in such
cases the persons should be punished, despite their innocence. This
move is a rather implausible one, inasmuch as it commits one to
admitting that one will punish an entirely innocent person. The
Australian philosopher, Jack Smart, is one of the few utilitarians
willing to make this move. 22 This has earned Smart the following
entry in Dan Dennett's humorous glossary of terms constructed out
of philosopher's names:
Outsmart, v. To embrace the conclusion of one's opponent's reductio ad
absurdum argument. (As in) 'They thought they had me, but I outsmarted
them. I agreed that it was sometimes just to hang an innocent man.'23

The second possibility is to deny that there will be cases where there
will be a net social gain from punishing an innocent person. This
move is usually associated with the name of indirect utilitarianism
and involves the idea that one cannot make a general practice of
punishing the innocent, because then the harm of so doing (in terms
of demoralization costs in society, character degradation costs, and
the like) will outweigh any possible good to be achieved, even the
22 J. J. C. Smart, 'Utilitarianism: For', in J. J. C. Smart and Bernard Williams,
Utilitarianism: For and Against (Cambridge: Cambridge University Press, 1973),
67-73.
23 Daniel Dennett and Karl Lambert, The Philosophical Lexicon (7th edn., pri
vately printed, 1978),8.

CLOSET RETRIBUTIVISM

97

preventi?n.of skyJack.ing. 24 The ~roble~ with this response, popu

lar as it IS, IS that It falls to deal faIrly wIth the nature of the thought
experiment. That is, suppose that there are some risks of detection
of punishment of innocent persons and thus, some risks of demor
alization costs; such risk will only allow utilitarians to say that the
number of cases in which punishment of the innocent will maximize
utility is somewhat diminished. It does not foreclose as somehow
impossible that there are such cases. Such cases are conceivable,
and if in them one is still not willing to punish, one thereby shows
oneself not to be a utilitarian about punishment.
This brings us to the third possibility: one can simply give up the
first premise, that is, one can repudiate the utilitarian theory of
punishment. Such thought experiments, I think, when clearly con
ceived and executed, show all persons who are not prepared to 'out
smart' their own firmest intuitions that they are not pure
utilitarians about punishment.
V. ARGUMENTS AGAINST THE MIXED THEORY OF
PUNISHMENT

The argument against the pure utilitarian theory of punishment

does not by itself drive one into retributivism. For one can allevi
ate the injustice of the pure utilitarian theory of punishment by
adopting the mixed theory earlier mentioned. Since under the
mixed theory the desert of the otTender is a necessary condition of
punishment, it will follow from the mixed theory that in each of the
kinds of counter-examples considered (where punishment is not
deserved), punishment should not be given. There will be no con
tradictions generated, because the premises are consistent:
24 The response classically associated with John Rawls's well-known article, 'Two
Concepts of Rules', The Philosophical Review, Vol. 64 (1955), J-32. The variations
in formulation of indirect (or two-step) utilitarianisms since 1955 has been enor
mous. However, to the extent that any of such indirect utilitarianisms purport to
give us the truth conditions of moral propositions such as, 'It is sometimes just to
hang an innocent man'-and not merely the best decision-procedures justifying why
we should say, or induce others to believe, such propositions (see Moore, 'Three
Concepts of Rules', Harvard Journal of Law and Public Policy, Vol. 14 (1991),
771-95)-all such versions are subject to my objection in the text. For the most wor
ries such as detection, etc., can do is justify us in setting up rule-utilitarian decision
procedures (institutions); it does not relieve the utilitarian decision-maker from
confronting the blunt fact that his theory justifies the uncomfortable belief that some
innocents should be punished.

98

THE THEORY OF CRIMINAL LAW'S FUNCTION

I. Punishment should be inflicted if and only if doing so achieves

both a net social gain and gives an offender his just deserts.
2. A net social gain would be achieved in this case by the inflic
tion of punishment.
3. It is not the case that punishment would give an offender his
just deserts in this case.
4. Punishment should not be inflicted.
From the first three of these premises, the conclusion is deducible
that there should be no punishment. This is also what the fourth
premise asserts, so that there is no contradiction when one substi
tutes the mixed theory for the utilitarian theory of punishment in
the scapegoating and preventive detention thought experiments.
There is, nonetheless, another sort of thought experiment which
tests whether one truly believes the mixed theory, or is in fact a pure
retributivist. Such thought experiments are the kind that fill the edi
torial pages of our papers, when such opinions express outrage at
the lightness of sentence in a particular case, or the lightness of sen
tencing generally in the courts of some communities. Consider
some examples Mike Royko has used to get the blood to the eyes
of readers of his newspaper column:
The small crowd that gathered outside the prison to protest the execution
of Steven Judy softly sang 'We Shall Overcome'....
But it didn't seem quite the same hearing it sung out of concern for
someone who, on finding a woman with a flat tire, raped and murdered her
and drowned her three small children, then said that he hadn't been 'los
ing any sleep' over his crimes....
I remember the grocer's wife. She was a plump, happy woman who
enjoyed the long workday she shared with her husband in their ma-and-pa
store. One evening, two young men came in and showed guns, and the gro
cer gave them everything in the cash register.
For no reason, almost as an afterthought, one of the men shot the gro
cer in the face. The woman stood only a few feet from her husband when
he was turned into a dead, bloody mess.
She was about 50 when it happened. In a few years her mind was almost
gone, and she looked 80. They might as well have killed her too.
Then there was the woman I got to know after her daughter was killed
by a wolfpack gang during a motoring trip. The mother called me occa
sionally, but nothing that I said could ease her torment. It ended when she
took her own life.
A couple of years ago I spent a long evening with the husband, sister
and parents of a fine young woman who had been forced into the trunk of

CLOSET RETRtBUTlVISM

99

a car in a hospital parking lot. The degenerate who kidnapped her kept her
in the trunk, like an ant in a jar, until he got tired of the game. Then he
killed her.2~

Most people react to such atrocities with an intuitive judgement

that punishment (at least of some kind and to some degree) is war
ranted. Many will quickly add, however, that what accounts for
their intuitive judgement is the need for deterrence, or the need to
incapacitate such a dangerous person, or the need to reform the
person. My own view is that these addenda are just 'bad reasons
for what we believe on instinct anyway', to paraphrase Bradley's
general view of justification in ethics.
To see whether this is so, construct a thought experiment of the
kind Kant 26 originated. Imagine that these same crimes were done
in circumstances where there is no utilitarian reason to punish. The
murderer has truly found Christ, for example, so that he or she is
already reformed and thus is not dangerous; and the crime can go
undetected so that general deterrence does not demand punishment
(alternatively, we can pretend to punish and pay the person the
money the punishment would have cost us to keep his or her mouth
shut, which will also serve the ends of general deterrence). In such
a situation, should the criminal still be punished?
Or consider the case that Sanford Kadish and Stephen
Schulhofer include in their leading criminal law casebook, State v.
25 Mike Royko, 'Nothing Gained by Killing a Killer? Oh Yes, There Is'. Los
Angeles Times, 13 Mar. 1981, Sec. II. 7. (Reprinted by pennission of the Tribune
Media Services.) Jeff Murphy also has his favourite Mike Royko quote: 'It's great
to be back in Chicago where people still know how to hate.' Quoted in JelTrie
Murphy and Jean Hampton, Forgiveness and Mercy (Cambridge: Cambridge
University Press, 1988), at 88.
26 See Kant, The Metaphysical Elements of Justice (Indianapolis: Dobbs-Merrill,
J. Ladd trans., 1965), 102: 'Even if a civil society were to dissolve itself by common
agreement of all its members (for example. if the people inhabiting an island decided
to separate and disperse themselves around the world), the last murderer remaining
in prison must first be executed, so that everyone will duly receive what his actions
are worth and so that the blood-guilt thereof will not be fixed on the people because
they failed to insist on carrying out the punishment; for if they fail to do so, they
may be regarded as accomplices in this public violation of legal justice.' Jean
Hampton begins another such thought experiment from Dostoevsky's question of
what to do with the Russian nobleman who unleashes his dogs to tear apart an
eight-year-old child in the presence of the child's mother. Murphy and Hampton,
supra n. 25, at III. Dostoevsky in The Brothers Karamazov has Alyosha answer the
question in tenns that Royko and Kant would approve: ' "Shoot him!" Alyosha said
softly.'

THE THEORY OF CRIMINAL LAW'S FUNCTION

CLOSET RETRIBUTIVISM

Chaney,27 a case that also rouses our retributive juices. In Chaney,

the defendant was tried and convicted of two counts of forcible
rape, and one count of robbery. The defendant and a companion
had picked up the prosecutrix at a downtown location in
Anchorage. After driving the victim around in their car, the defen
dant and his companion beat her and forcibly raped her four times,
also forcing her to perform an act of fellatio with the defendant's
companion. During this same period of time, the victim's money
was removed from her purse, and she only then was allowed to
leave the vehicle after dire threats of reprisals if she attempted to
report the incident to the police.
Despite this horrendous series of events, the trial judge imposed
the minimum sentence on the defendant for each of the three
counts, and went out of his way to remark that he (the trial judge)
was 'sorry that the [military] regulations would not permit keeping
[the defendant] in the service if he wanted to stay because it seems
to me that is a better set up for everybody concerned than putting
him in the penitentiary'. The trial judge also mentioned that as far
as he was concerned, there would be no problem for the defendant
to be paroled on the very first day of his sentence, if the parole
board should so decide. The sentence was appealed by the state
under a special Alaska procedure, and the attorney-general urged
the Alaska Supreme Court to disapprove the sentence.
The thought experiment such a case begins to pose for us is as
follows: Imagine in such a case that the defendant after the rape
but before sentencing has got into an accident so his sexual and
aggressive desires are dampened to such an extent that he presents
no further danger of violence against women; if money was also
one of his problems, suppose further that he has inherited a great
deal of money, so that he no longer needs to rob. Suppose, because
of both of these facts, we are reasonably certain that he does not
present a danger of either forcible assault, rape, or robbery or
related crimes in the future. Since Chaney is (by hypothesis) not
dangerous, he does not need to be incapacitated, specially deterred,
or reformed. Suppose further that we could successfully pretend to
punish Chaney, instead of actually punishing him, and that no one

is at all likely to find out. Our pretending to punish him will thus
serve the needs of general deterrence and maintain social cohesion,
and the cost to the state will be less than if it actually did punish
him. Is there anything on the mixed theory of punishment which
would urge that Chaney nonetheless should really be punished? [
think not, so that if one's conclusion is that Chaney and people like
him nonetheless should be punished, one will have to give up the
mixed theory of punishment.
The argument structure is again that of a reductio, and is as fol
lows:

100

27 State v. Chaney, 477 P 2d 441 (Alaska Sup. Ct. 1970). Kadish and Schulhofer
presumably included Chaney for the same pedagogical purpose as those for which I
use it here. S. Kadish and S. Schulhofer. Cases and Materials on Criminal Low
(Boston: Little, Brown, 5th edn. 1989). 124-6.

tOt

I. Punishment should be inflicted if and only if doing so both

achieves a net social gain and gives an offender his just
deserts.
2. A net social gain would not be achieved in this case by the
infliction of punishment.
3. Punishment should be inflicted.
Again these three premises generate a contradiction:
4. Punishment should not be inflicted and punishment should be
inflicted.

;i
-~

~j

'fl

From the first two premises, it follows that there should be no pun
ishment; this contradicts the third premise that there nonetheless
should be punishment.
One has, again, the choice of giving up one of the three premises
of the argument. To give up the third premise is very unappealing
to most people for it requires that people like Chaney should not
be punished at all. Again, the tempting move is to assert that there
will be no cases in which both utility would not be served by pun
ishment and yet the offender deserves to be punished. One way to
assert this is to deny that one will be sure enough that the danger
is removed, or the ends of general deterrence served with only a
pretended punishment of such deserving people. But as with the
indirect utilitarian's defence of the pure utilitarian theory, this is
simply to misunderstand the nature of the thought experiment. One
only need think it conceivable that such dangers could be removed,
or such ends of deterrence served, in order to test the truth of one's
theory of punishment. And nothing in utilitarianism makes it even
plausible to think that utility is always maximized by the punish
ment of the guilty.

102

THE THEORY OF CRIMINAL LAW'S FUNCTION

If one is unwilling to give up the third premise (that punishment

should be inflicted), even admitting that this is a case where no net
social gain is achieved (thus accepting the second premise), then
one has to give up the first premise. One has to give up, that is, the
mixed theory of punishment.
C. L. Ten has recently urged that the mixed punishment theorist
can avoid this result. Ten acknowledges the kind of thought exper
iment that gives trouble to the mixed theorist here,2B but seeks to
modify the mixed theory by adding in considerations of formal jus
tice. Punishment in an individual case will be justified for Ten if and .~
only if: either (there is a net social gain from this punishment and .~!
the offender deserves it) or (there is a net social gain from punish
ment of other offenders and equal justice demands that this
offender be punished if those other offenders are punished).29 It is
this modification of the mixed theory that allows the mixed theo
rist to argue for 'punishment even though no overall utilitarian pur
pose is thereby served'.30 Yet this surely will not do, for Ten has
covertly smuggled a strong retributivist premise into his idea of
equal justice. Equality demands that like cases be treated alike, but
the crucial question is always, 'What are the relevant respects in
which one case must be like another case in order to be deserving
of like treatment?' Ten assumes that the only respect relevant here
is moral desert, yet surely a mixed theorist is committed to saying
that both moral desert and net social gain are relevant to punisha
bility so that a person whose punishment does not serve both is not
relevantly like a person whose punishment does serve both ends. 31

28 C. L. Ten, supra n. 18, at 47, considers a kind of Eichmann/Kant combination

example: '[Consider] the example of a Nazi war criminal who escapes to an unin
habited island where 30 years later he is found leading "an idyllic existence". While
he is still unrepentant for what he has done, he has no desire to cause further harm.
(Let US also assume that his punishment would have no general deterrent effect on
the behavior of others.)'
29 See id. at 50-I, 79-80.
JQ Id. at 79.
31 It is in addition unclear why only utility is needed in order to justify the ear
lier punishments Ten hypothesizes; a mixed theorist should think that earlier pun
ishments serve both utility and retribution. Finally, Ten's modified mixed theory
does not justify punishment of deserving persons unless there have been earlier pun
~-- ~-- : - ~ Mo.. i"tv "ould thus not be punished if his own pun

CLOSET RETRIBUTIVISM

103

VI. AN ARGUMENT FOR RETRIBUTIVISM?

If one follows the predicted paths through these thought experi

ments, the end result is that one finds oneself, perhaps surprisingly,
to be a retributivist. We might call this an argument from the back
door for retributivism, because the argument does not assert in any
positive way the correctness of retributivism. It only asserts that the
two theories of punishment truly competitive with retributivism,
namely, the pure utilitarian theory, and the mixed theory, are each
unacceptable to us. That leaves retributivism as the only remaining
theory of punishment that we can accept, assuming that we are not
willing to give up the institution of punishment entirely. In some
such way, one brings closet retributivists out of the closet to face
whether, in the light of day, they are comfortable with their ret
ributivist position. Whether they should be comfortable with their
retributivism is the topic of the next chapter.