Janssen 2009

AGENT-BASED MODELLING IN
ARCHAEOLOGY
KERSTIN KOWARIK
NATURAL HISTORY MUSEUM VIENNA

I. Introduction

Multi-Agent Simulations

Starting Point

I. Introduction

There are two approaches to the study of social

behaviour: Collect, observational, survey or other
forms of data and analyze them, possibly by
estimating a model; or begin from a theoretical
understanding of certain social behavior, build a
model of it, and then simulate its dynamics to gain a
better understanding of the complexity of a seemingly
simple social system. (Liao 2008: IX in Gilbert
2008)
Janssen 2009

I. Introduction

The idea of using computer-based modelling in the

social sciences dates back to the 1960s
Widespread use since the 1990s related to
complexity theory
Considered a valuable tool for the social sciences
by some
In the natural sciences simulations = basic
methodological tool
Regional preferences: Great-Britain, Northwest
Amercia

II. General Remarks

Model
A

model is a simplification smaller, less detailed, less

complex, or all of these together of some other
structure or system. (Gilbert / Troitzsch 2009: 2)

II. General Remarks

Modelling
One

creates some kind of simplified representation of

social reality that serves to express as clearly as
possible the way in which one believes that reality
operates. (Nigel Gilbert 2008: 2)

II. General Remarks

Modeling

II. General Remarks

Modelling
Mathematical
Every

and computer modelling require precision

aspect of the model needs to be laid out precisely

Model buidling highlights logical gaps and data gaps in the
verbal model
Relationships between system parts need to be identified
and described in detail

II. General Remarks

Simulation
Here:

simulations = computer based simulations

A particular type of modelling
Models that are expressed as computer programs

II. General Remarks

Formalization: computational model building as a

method of theory development
Understanding
Exploration:

explore our own assumptions systematically

Experimentation: what-if-scenarios

Experimentation

Simulation!

III. Agent-based Modelling (ABM)

What is Agent-based Modelling?

computational

method that enables a researcher to

create, analyze and experiment with models composed of
agents that interact within an environment. (Gilbert
2008: 2).

III. Agent-based Modelling (ABM)

Used in natural and social sciences

Artificial Societies
Development of ABM techniques is closely related
to:
Research

on complex, non-linear systems

Artificial Intelligence

III. Agent-based Modelling

Complex Systems
Systems

that are composed of many different parts

System behaviour cannot be explained by reduction to
its component parts.
Interaction of the system parts leads to new and
unpredictable behaviour on system level -> emergence
System behaviour cannot be explained through simple
cause and effect relations -> non-linear

III. Agent-based Modelling

Complexity Theory
Both

Systems Theory!

theories deal with systems.

But Systems Theory describes system behaviour by
breaking the system down to subsystems and
modelling those.
Systems Theory assumes that the relationship of
system parts is known and understood.
Complexity Theory assumes that relationships between
systems parts are not well known, that they are
unstable and non-linear and thus unpredictable.

III. Agent-based Modelling (ABM)

Agents
Agents

are either separate computer programs or, more

commonly, distinct parts of a program that are used to
represent social actors individual people, organizations
such as firms, or bodies such as nation nation-states.
(Gilbert 2008: 5)
Agents can interact intelligently with their environment,
be it social or other.

III. Agent-based Modelling (ABM)

Environment
Virtual

world in which agents interact

Most commonly a geographic space with features such
as houses, different soil types
Spatially

specific
In general not comparable to GIS-environments!
But

other spaces can be designed: knowledge space,

symbol space, networks.

III. Agent-based Modelling (ABM)

Abilities of Agents
Perception
Performance
Motion
Communication
Action

Memory
Policy

III. Agent-based Modelling (ABM)

What can we do with Agents?

Agents

can be given simple behavioural rules.

These rules can be chosen to guide interaction of the
agents with each other or of agents with the
environment or both.
They can be given knowledge and they are able to
learn.
On this basis they will be able to make independent
decisions.

III. Agent-based Modelling (ABM)

agents can be distinguished from other entities

that can express behavior by the fact that agents
have some cognitive decision-making abilities
(Altaweel 2006: 30)

III. Agent-based Modelling (ABM)

What can we do with Agents?

Define

a different set of rules for any single agent

Individuals, groups, institutions,
Different agent types can be put in one model
Individuals,

Two

households, tribes

model types

One

agent interacting with its environment.

Groups of agent

III. Agent-based Modelling (ABM)

Artifical Societies: Sugarscape

Epstein / Axtell 1996

III. Agent-based Modelling (ABM)

modelling from a special perspective

III. Agent-based Modelling (ABM)

Bottom Up

IV. ABM in Archaeology

Applications
Socio-ecological

dynamics
Spatial processes
Adaptation
Culture Change (long-time-perspective)
Social interaction
emergence of social complexity
evolution of sociality
decision making

IV. ABM in Archaeology

Specific Research Questions

Identifying drivers of historical change
Population aggregation and abandonment
Civilization collapse
Response to changing environmental and social conditions
Impact of ecological stress
Impact of population aggregation
System resilience
Mechanisms of adaptation
Emergence of territories
Settlement dynamics
Site selection

IV. ABM in Archaeology

Specific Research Questions

Resource management
Land use
Selfishness vs altruism
Food sharing
Alarm calls/protecting others
Social stratification
Emergence of centralized decision making
Emergence of specialization
Culture contact
Maintenance of ethnic boundaries
Formation of political entities

IV. ABM in Archaeology

Artificial Anasazi

Dean et al. 1999; Janssen 2009

IV. ABM in Archaeology

Artificial Anasazi

Janssen 2009

IV. ABM in Archaeology

Every year the following sequence of calculations is performed:
1) Calculate the harvest for each household
2) IF an agent derives not sufficient food from harvest and storage OR the
age is beyond maximum age of household THEN the agent is removed from
the system
3) Calculate the estimated harvest for next year based on corn in stock and
actual harvest from current year
4) Agents who expect not to derive the required amount of food next year
will move to a new farm location and a plot to settle nearby.
5) If a household is older than the minimum fertility age, there is a probability
pf, that a new household is generated. The new household will derive an
endowment of a fraction fcs of the corn stock.
6) Update water sources based on input data
7) Each agent, household, ages with one year.
(Janssen 2009: 5)
Janssen 2009

IV. ABM in Archaeology

Janssen 2009

IV. ABM in Archaeology

Janssen 2009

IV. ABM in Archaeology

Comparable ABMs
ENKIMDU:

Modeling Bronze Age Communities in Upper

Mesopotamia (Wilkinson 2007): model simulating the
impact of ecological stress on farming, herding and
trade behavior, very fine grained model

Wilkinson 2007, Fig. 4-5

IV. ABM in Archaeology

Comparable ABMs
Lake

Titicaca Basin (Griffin / Stanish 2007): The model

was designed to explain the role of agriculture,
competition, and trade leading to the political
consolidation observed in the Titicaca Basin.

Griffin / Stanish 2007, Fig. B-1

IV. ABM in Archaeology

Geographic Space in ABM

Elaborate
#what

Very

Data Collection

kind of data

simple surfaces

Combining ABM and GIS one focus of actual

research

Janssen 2009

IV. ABM in Archaeology

PatronWorld
Simulation

Graham 2009

Roman Social Life and Civil Violence

IV. ABM in Archaeology

Evolution of Organized Society (EOS)

explore

theories which account for the growth of social

complexity among the human population in the Upper
Paleolithic Period
emergence

of status, roles and leadership

formation of relationships between individuals
especially authority relationships

IV. ABM in Archaeology

Evolution of Organized Society (EOS)

Artificial

society: agents that were capable of selfinterested rational decision making, planning and
cooperation in the context of resource gathering
Environment: landscape with a population of mobile
agents and a scattering of resources which provide
energy for the agents

IV. ABM in Archaeology

Evolution of Organized Society (EOS)

Agents:

communicate with each other

do rudimentary planning

In the simulation agents could either:

acquire resources alone

construct plans involving other agents to secure resources collectively

Plans involved:

determine resources to be selected

select agents for work
calculate expected result and plan distribution
negotiate plan with other agents
choose from several plans -> select plan with greatest pay-off

IV. ABM in Archaeology

Evolution of Organized Society (EOS)

Results:

Some agents will repeatedly adopt plans proposed to them by

particular other agents.
They will come to see themselves as followers.
Very thinly distributed resources -> working alone greater
payoff.
Concentration of resources -> working together and building
extensive hierarchies paid off better
Concentration of agents and insufficient resources -> hierarchies
collapsed and disorganization ensued
If hierarchies once established were persistent chances of survival
were worse than when follower-leader relationships were only
temporary.

V. Conclusion

Possibilities
Explore our ideas about the past
Experiment with those ideas
Hypothesis testing
Maybe filling in some gaps

But
Time, effort, data
Epistemological problems
Modelling philosophy:

Real world models vs Stylized models

Exploration vs Prediction

Acknowledgements
Support and Organization:
Mario Bertieri, Gabriele Greylinger, Karina Grmer, Angelika Heinrich, Anton
Kern, Herbert Kritscher, Thomas Koch-Waldner, Franziska Nittinger, Gerlinde
Rattner, Wolfgang Taigner, Ingrid Viehberger.

Financial Support