Agent-based modeling, and the art of interdisciplinary science
The segregation model (Schelling 1978) explains how separate groups quickly form when individuals (simulated as green and red dots in NetLogo) prefer to have a certain amount of similar-looking people around them. Not too different from what happens in the scientific world.
During the first week of November the course with the catchy title ‘Agent-based modeling as a tool to answer complex problems: an interdisciplinary PhD course’ was held at the University of Oslo. I must admit I signed up for the course because of the first few words. Vaguely familiar with individual-based modeling I hoped I could pick up something useful for my PhD project. The interdisciplinary part was probably added to sell the course to the University administration, I thought. However, summing up my thoughts a week later, the interesting aspect of the course was as much in the interdisciplinarity as in the modeling itself.
But first, what is agent-based modeling (ABM)? In my notebook I’ve written, ‘a computational methodology that allows the analyst to create, analyze, and experiment with, artificial worlds populated by agents that interact in non-trivial ways’ (Cederman 2002). Or, if I were to explain it to my mother: ABMs are model of single agents (people, for example, or animals) who can change, interact and affect their environment. Instead of modeling the dynamics of a whole population, AMBs are useful to understand processes occurring between single agents that might have different characteristics and experience different environments. And if you wondered, yes, ABMs and IBMs (individual based models) are the same. For some reasons social scientists model ‘agents’, natural scientists ‘individuals’. An example of how researchers seem to emphasize differences rather than similarities.
During the week we got familiar with the basic principles of agent-based modeling and how to build simple models in the program NetLogo. This core of the course was given as thorough lectures and computer lab-sessions by Nils Weidmann, a German political (and computer) scientist. In addition to Nils’s sessions we were visited by different guest lecturers who gave us insight into how ABMs can be useful tools in research questions spanning from fisheries management to climate change policy.
Giovanni and I were the only biologists among the course participants, and we quickly became known as ‘the cod people’. N.C. Stenseth’s lecture on interdisciplinary marine research including an omnipresent power-point cod probably triggered this. But since Weidmann’s background is in political science the examples we dealt with during the course were mainly from this field, including models on social interactions, the diffusion of ideas within social networks, and ethnic conflict. Even so, it was always emphasized that any modeling technique can be applied to other topics, such as cod. Grateful for the consideration, I have to admit that I was tempted to say: Hey, even marine scientists don’t ALWAYS think about cod. In fact, I find ethnic conflict in Baghdad quite fascinating, even if it’s far away from the marine system I usually work with.
The final part of the course, which just briefly started during the hectic lunch breaks (filled with discussions, guest lectures and delicious sandwiches), was the group project work. In groups, we should come up with a personal research question and then, after the course itself, build an ABM to analyse the question in hand. We were strongly encouraged to form interdisciplinary groups, but the discussion quickly heated up: What does interdisciplinary really mean? One natural scientist in each group? Do we have to model cod? How can a political scientists contribute to that? I think the results of the discussions only partly met the aims of the course: although some new contacts were made most of us ended up with project topics within quite familiar grounds.
Interdisciplinary work is a challenge, it’s certainly easier to stay within our home fields. But if we’re not able to communicate across and outside academia, then our research will only have importance for the few that share our special interests. A trivial example, one of my flatmates is a linguist, and she was genuinely surprised that someone can do a whole PhD about cod. Only one fish species, how could someone work three years on that? And maybe a marine scientist would ask the same upon meeting a researcher working on a specific South-American writer, for example. Interdisciplinary courses provide good practice in communication between different scientific fields, and NorMER itself is also a great training field for that. But in everyday life, just talking to new people and escaping from the bubble of your research group a little bit can give interesting outcomes. At least my flatmate is slowly turning into a cod lover too.
No reason to stay away from ABMs: NetLogo even lets you use a fish shape as agents.