I’m an experimental behavioural ecologist. I deal, mainly, with information that we know because someone did an experiment. I trust that, within the parameters of the experiment, we have established some factual information that can then be carefully considered and discussed.
Behavioural modelling takes some parameters that we know of from experimental studies, makes some assumptions (sometimes many) and models an insight into what would happen if these assumptions were correct. The amount of information that the model offers depends on the amount of information that goes into it, but the more information you put in, the more assumptions you make (see diagram). This, as well as a recent interest in modelling animal behaviour, has left me a little puzzled. As a scientist I love to see a question answered; to have strong evidence that A happens because B, and maybe even C, say so. But, I often feel that modelling doesn’t offer such solace.
This diagram tries to explain the problem I have. I initially want my model to answer a specific question so I put in many parameters to isolate the trait I hope to see….BUT!…if I do this I am forced into making many assumptions, which I don’t want to do! So I remove some parameters and make fewer assumptions. Now, while I have confidence in the model, it really doesn’t tell me anything interesting!
However, there is a way…“The dotted line of modelling majesty” leads the way from a model that has made few assumptions but has the distinction of explaining a pattern and so is useful! This is the key to a good model. To do this, however, we shouldn’t make the model too specific. The question we want to answer should be broader than it would be when carrying out experimental behavioural studies. So let’s have a look at a study that has done just that!
In the early 1980’s Robert Axelrod proposed a computer tournament that would bring to light the evolution of cooperation. Based on the prisoner’s dilemna*, contestants would submit models to Axelrod that competed for success in his environment. The idea was simple; for each interaction that a model experiences it can COOPERATE or DEFECT. The agent would score depending on the outcome of the interaction (see table).
The key attribute of a model was how its behaviour is affected by the result of the previous interaction – if the last model you faced defected, what do you do? Do you retaliate and defect or offer peace and cooperate? For example one program, called “massive retaliatory strike”, will cooperate from the start…BUT as soon as the opponent defects, it will retaliate, and never cooperate again! This program did very badly! After each program had played all other programs 200 times there was a clear winner! This is surely an incredibly complicated piece of code that would take a while to explain! Actually, it was two lines of code.
The program, named “tit for tat”, was very simple. It started by being nice, cooperating. Its next move depended on what the other program did in the first round. If the opponent cooperated, then it would cooperate. However, if the opponent defected, it would defect. The key to this model was that it was always able to respond to the opponent’s actions. It wasn’t stuck defecting or cooperating; if the opponent always defected then so would tit for tat. This could be considered good defence, with tit for tat stopping the opponent from winning. If the opponent always cooperated so would tit for tat and both would be successful. Axelrod’s study offers an incredible insight into cooperation and can be used to look at a diversity of interactions within biology and culture. He posed a question that was simple but relevant to a lot of disciplines.
At the moment, still fairly new to modelling behaviour, I do not have an answer that satisfies the scientist in me. That is, my conclusion does not conclude my search for a more refined statement on how I should model. But, Axelrod’s paper has taught me that simplicity is key in a model…also, maybe get someone else to do it for you.
I would like to add this quick edit having spoken to a modeller friend of mine. I am not belittling modelling at all! I think the modelling of cultural evolution and behaviour is insightful and very interesting. I hope that I can gain a better understanding of the discipline and eventually write my own models. The confusion I have and discuss in the post relates to my inexperience in the area.
*Prisoner’s dilemma: Two members of a criminal gang are arrested and imprisoned. Each prisoner is in solitary confinement with no means of speaking to or exchanging messages with the other. The police admit they don’t have enough evidence to convict the pair on the principal charge. They plan to sentence both to a year in prison on a lesser charge. Simultaneously, the police offer each prisoner a bargain. Each prisoner is given the opportunity either to betray the other, by testifying that the other committed the crime, or to cooperate with the other by remaining silent. Here’s how it goes:
- If A and B both betray the other, each of them serves 2 years in prison
- If A betrays B but B remains silent, A will be set free and B will serve 3 years in prison (and vice versa)
- If A and B both remain silent, both of them will only serve 1 year in prison (on the lesser charge)