Skip main navigation

New offer! Get 30% off one whole year of Unlimited learning. Subscribe for just £249.99 £174.99. New subscribers only. T&Cs apply

Find out more

Integrating theories into agent-based modelling

In this article we discuss the challenges of integrating different theories in ABM’s, and reflect on ABM’s as a new form of generative social science.
people playing with lego

Many societal issues display emergent phenomena (see the introductory course on People, Networks and Neighbours: Understanding Social Dynamics for an article on the social dynamics of this). Hypes, opinion dynamics and polarisations, the diffusion of new technology, these are all phenomena growing out of the manifold interactions between people.

To better understand the dynamics driving such societal issues, we can study the interactions between individuals and the group(s) they belong to. This addresses the so-called micro-macro dynamics in society. Because these dynamics can display tipping points, and may display unpredictable turbulences, understanding these are contributing to understanding and guiding societal change.

Examples of such changes are growing a society towards a more plant-based diet, the opinion formation and behavioural response to a disease (e.g., COVID, obesity), or migration and acculturation in relation to climate change and armed conflicts. Improving our understanding of these so-called micro-macro dynamics may help to think of policies preventing social dynamics from spinning out of control.

How to model people

A key challenge in the use of agent-based models is modelling relevant behavioural processes in a valid manner. We know that we are not only outcome maximisers, but also use several smart (bounded rational) strategies to make good-enough choices with our limited cognitive capacity. As humans we are equipped with different “fast and frugal” strategies to make choices in a world that is demanding constant choices from us. Gerd Gigerenzer describes this as an adaptive toolbox we have (formal models of heuristics) for solving problems in situations of uncertainty where an optimal solution is unknown. In other words, how people decide will depend on the decision content and context.

When we want to simulate the behaviour of people in an agent-based model, a key question is what mechanisms and associated theories can help us construct “artificial people” that perform the behaviours we are interested in? The answer to this question depends for a large part on the scope of the behaviour in question. Our dietary choices are, for example, very determined by the habits we have. This means that new information is less likely to be picked up, and people will be quite persistent in their behaviour, even if better alternatives are available. When we want to model this type of choices we need to assume a process in which the status quo plays an important role.

On the contrary, when buying a car people usually invest a lot of cognitive energy in scrutinising and comparing their options. Many models are available in a wide price range, and many attributes require some careful processing. For example, when a person is in doubt between a fuel or electric car, the available charging infrastructure definitely is a relevant aspect in decision making. However, for many people a car is also an expression of one’s taste and personality, and so the social context will be relevant to consider in a model. And mind that for many people a car is impractical or too expensive, and alternative mixes of modalities like public transportation, biking and car-sharing can be part of the decision context. So, when we think of modelling the dynamics of car purchasing in a time of energy transition, we may start with the Theory of Planned Behaviour, as this combines the multi-attribute perspective (attitudes), the influence of norms (subjective norm) and behavioural control (e.g., budget, charging infrastructure) as key drivers.

Agent-based modelling offers the computational tool for systematically studying emergent social dynamics. Hence, in recent years this methodology is increasingly being used to study a wide range of societal dynamics. In the Journal of Artificial Societies and Social Simulation a plethora of articles address a wide variety of societal phenomena being studied with agent-based models.

When a new topic or question arises where understanding the possible social dynamics comes in handy, it is smart to build upon earlier models that have been developed, used and published. The basic psychological and behavioural mechanisms are usually the same, and can be applied to new topics. As a natural consequence, models become available that combine different factors and processes, and that are generic enough to be used for different research topics.

This article is from the free online

Decision Making in a Complex World: Using Computer Simulations to Understand Human Behaviour

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now