SCL Seminar: Jelena Grujić, 21 April 2017

 

You are cordially invited to the SCL seminar of the Center for the Study of Complex Systems, which will be held on Friday, 21 April 2017 at 14:00 in the library reading room “Dr. Dragan Popović" of the Institute of Physics Belgrade. The talk entitled:
 

Are people instinctively good or bad - neuroscience meets experimental game theory


will be given by Dr. Jelena Grujić (AI Lab, Vrije Universiteit Brussel, Belgium).

 

Abstract of the talk:

How do we decide to be involved in pro-social or anti-social behavior? This subject raised a lot of interest recently, with some papers claiming that pro-social behavior is intuitive and anti-social behavior requires deliberation, while some papers expressed a suspicion about the validity of this conclusion. As a main measure for what is intuitive and what deliberate, the researchers often use reaction times in an experiment for one or the other decision. In game theory experiments, no model is provided that would fit the distribution of the reaction time (RT) and usually only the mean value and the standard deviation are used. However, the RT distribution is clearly not Gaussian and therefore a proper model whose parameters have more clear neuroscientific interpretation is necessary in order to answer this question. Here we consider the Drift Diffusion Model, take it beyond the scope in which it was originally developed, and show that it can also explain the data of an iterated Prisoner’s Dilemma experiment with human subjects. In this way we show that the model is much more universal and could possibly be generalized to any binary-choice experiment. Knowing the exact shape of the distribution gives us a new tool to precisely describe the learning process in game theory experiments, isolating the difficulty of the task from personal bias and drop in attention. Using this new tool, we observe that although initially people’s intuitive decision is to cooperate, they are not naive and their biases change quickly in a non-cooperative environment.