Skip to 0 minutes and 7 seconds This interview is about applications of agent-based modelling with the teacher of this week’s activity, Wander Jager. Wander, welcome. Hi, Lex. What applications for agent-based modelling can be envisioned in the future? Well, I think in future, a lot is possible, obviously, but if we look at the shorter time scale, I think in the forthcoming years, we’ll all see the first applications that target actual systems in society. For example, talking about innovation diffusion of products and so on. And I think at the later stage, which is currently already in development, is the integration of agent-based models with other models. We talk, for example, about coastal management. We think about landslides. We think about economy.
Skip to 0 minutes and 56 seconds I think that we will see that agent-based modelling will become a tool to be integrated in other models as well. What is critical for using these models? Well, what really is critical are two issues. It’s the input of the models and the use of the output of the models. And if you consider the input, I always emphasise that it’s very important to use very sound theory in the models to make a model theoretically sound. And I think that psychology has a lot to contribute to developing these models and making it more applicable. I will give you one example of how things can go wrong.
Skip to 1 minute and 37 seconds Because a couple of years ago, I encountered an agent-based model which was aiming to study criminal behaviour. And the essential assumption in the model was that of your homo economicus. So imagine that we had a model here where every person in the world would wake up in the morning facing the decision either to engage in criminal acts today or to be a decent civilian. And the choice was made on the basis of perfect information on how many people engaged in criminal behaviour yesterday and the chances of being arrested. Well, this model was being used for policy, so you can imagine that using these kinds of input and assumptions in the models is really risky.
Skip to 2 minutes and 29 seconds So that’s the first thing about input. It’s very important to have decent theory in your model and to use, also, good data to parameterise your model. The second point was on the output of the model. And that’s also very important that you get output of your model that’s understandable. What I think is very important that a simulation tool is capable of creating narratives of future developments. So it is easy to communicate– possible for sites of future developments using the model. So that are the very important things I see for the near future. So in summary– in conclusion, what in your view can agent-based modelling contribute to the development of the social sciences, in particular.
Skip to 3 minutes and 18 seconds Well, for the social sciences, it’s a very important tool, I assume, because modelling forces you to be very precise in defining the processes. So you go beyond a more linguistic description of processes but you actually make it happen. You make it run in your model. And then also allows for connecting different ideas from psychology because perhaps you know psychology is a relatively young science. And we have a lot of theories on different aggregational levels– on different levels of description. And if you use agent-based models, you’re forced to collect them in some way. And sometimes that’s really very difficult.
Skip to 4 minutes and 0 seconds And in that respect, a developed agent-based model very often confronts you with questions for additional field redevelopment and for additional empirical data collection. So in this way, I think agent-based modelling also capitalises the integration of social science and the further theory development in the field. Thank you very much, Wander, for this interview.
Interview: Applications for agent based models (Wander Jager)
In this interview Lex Hoogduin discusses the possibilities for using agent based models with Wander Jager.
© University of Groningen