Agent based models in social sciences
Lately I’ve been digging into the subject of agent based models of social sciences. Agent based models are defined by wikipedia as following
An agent-based model (ABM) is a computational model for simulating the actions and interactions of autonomous individuals in a network, with a view to assessing their effects on the system as a whole. It combines elements of game theory, complex systems, emergence, computational sociology, multi agent systems, and evolutionary programming. Monte Carlo Methods are used to introduce randomness.
What an agent based model actually does, is provide a "social life simulation". Every theory strips many factors out of reality and focuses on others in order to create a logical and concise way of interpreting a set of events. Every theory has to do that on based on a certain level of reduction. Psychological theories, for example, deal with the person and discard most other factors, while economic theories discard psychological factors.
No, we’re not talking about this kind of agent 
What agent based models do, is to try and emulate higher order events, based on lower order entities, called agents. An agent can actually be anything, like animals, people or robots. Since we are talking about social sciences here, we will focus on people.
Scientific American recently had this article on the recent economic crisis: After the Crash: How Software Models Doomed the Markets. In this article we read the following
The software models in question estimate the level of financial risk of a portfolio for a set period at a certain confidence level. As Benoit Mandelbrot, the fractal pioneer who is a longtime critic of mainstream financial theory, wrote in Scientific American in 1999, established modeling techniques presume falsely that radically large market shifts are unlikely and that all price changes are statistically independent; today’s fluctuations have nothing to do with tomorrow’s—and one bank’s portfolio is unrelated to the next’s. Here is where reality and rocket science diverge. Try Googling “financial meltdown,” “contagion” and “2008,” a search that reveals just how wrongheaded these assumptions were.
This modern-day tragedy could be framed not only as a major motion picture but also as a train wreck or plane crash. In aviation, controlled flight into terrain describes the actions of a pilot who, through inattention or incompetence, directs a well-functioning airplane into the side of a mountain. Wall Street’s version stems from the SEC’s decision to allow overreliance on risk software in the middle of a historic housing bubble. The heady environment permitted traders to enter overoptimistic assumptions and faulty data into their models, jiggering the software to avoid setting off alarm bells.
For its part, the quant community needs to undertake a search for better models—perhaps seeking help from behavioral economics, which studies irrationality of investors’ decision making, and from virtual market tools that use “intelligent agents” to mimic more faithfully the ups and downs of the activities of buyers and sellers. These number wizards and their superiors need to study lessons that were never learned during previous market smashups involving intricate financial engineering: risk management models should serve only as aids not substitutes for the critical human factor. Like an airplane, financial models can never be allowed to fly solo.
We need better models…
Well, the readers of Encefalus probably remember older articles that dealt exactly with the problem of merging psychology and economics(Some thoughts on a new micro-economic model and paradigm, through the integration of psychology into economics, Behavioral economics revisited in the face of the recent economic crisis). However, the basic notion of agent based models is founded upon something that most modern science ignores: the study of complexity and emergence.
Complexity is a term that is very loosely defined. A June 1995 Scientific American article (From Complexity to Perplexity) reports that there are more than 30 definitions of complexity. The elusiveness of this definition can mean many things. Many people might say that this proves that a science of complex systems is actually something that doesn’t offer many things. Others might argue that complexity is a term that relies too much on the nature of the observer.
Anyway, the definition of complexity isn’t something we will deal with right now. The phenomenon that makes complexity so interesting is emergence. The basic notion of emergence is that out of simple rules, complex systems arise. And these complex systems, while at their most basic levek might seem nearly chaotic, they actually give rise to certain patterns that are emerging through the apparent chaos.
Complexity…
So for example, let’s think once again about economics. What has allowed economists to create theories that work to some extent is the existence of emergence. Of course, the economic models we had until now, are far from perfect as we have said in previous articles(Behavioral economics revisited in the face of the recent economic crisis) and as it is obvious from the recent economic crisis. However, the fact still remains, that from simple interactions among people within a set of given rules, some patterns arise at a higher level that justify the creation of a science of economy.
However, while the emergent patterns can be studied by themselves, they don’t arise by themselves. They arise from the interactions that happen below them. At this lower level, lie the agents. Behavioral economics for example, can provide clues about the behavior of buyers and sellers in few persons setting. But how can we apply the new found knowledge into the domain of the market? Or, how can we predict future credit crunches? The only way to do that, is to create a representation of a real person, embed it with the features that we consider relevant for our study and then let it roam into our virtual world with the rules that with consider relevant, along with other agents, in order to see the emergent pattern of interaction arise.
Many people might consider that the emergent models can deprive science out of its humanity. However, the goal of science is to study, and if our goal is to study people, then we must make the necessary abstractions in order to create our models.
How can we predict future crunches? 
For example, someone might say that we don’t allow free will into the equation. Casting aside the philosophical question concerning free will, that we have written about in the past (Free will revisited in the face of quantum physics, Free Will – plain and simple), if you think about it, no-one is completely free, since we all operate within certain boundaries. And these boundaries are the tools that we use to make our models.
Another example that could be used for an agent based model, came to my mind while reading this article: The spread of disorder – can graffiti promote littering and theft? (or read this Can the can on the Economist that refers on the same topic). These article comment on a recent research concerning the Broken Windows Theory. This theory postulates (and the research that the articles above commented on, proved it to some extent) that if in a neighbourhood there are a few broken windows, then the neighbourhood will degrade further. People will break more windows and further vandalize the neighbourhood. Or if there are litter on the street, the people will tend to throw more litter.The explanation for this is simple social norms. People have a strong tendency to conform (and they usually do so at any chance ).
Anyway, this information is valuable for agent-based models of social life, since such an information, for example, could be used as a rule for the interaction among agents.
Santa Fe institute in California
I hope I made clear a few points about this novel approach to social sciences. Computational modelling is something that has been used in many other sciences. The closest relative to the kind of modelling we are talking about now, happens in biology where biologists also have to deal with ecosystems made of agents.
Of course there are many more things that we haven’t discussed, such as complexity and emergence. Concerning all the subjects we talked about today, the main supporters of this theory can be found in the Sante Fe Institute in California (Santa Fe Institute entry in wikipedia). This institute has been founded with the purpose to study complexity. If you want to learn more on this subject this is the place to look.
Furthermore, we have Princeton’s Press series in complexity and Stephen Wolfram’s A New Kind of Science.
Stephen Wolfram
I promise to return on this subject. Until next, keep thinking! 
