Strategic Decision Making in Biological and Artificial Brains

The aim of this paper is twofold. First, it seeks to uncover the algorithms that humans and other animals employ for learning in decision-making strategies within non-zero-sum games, specifically focusing on fully observable iterated prisoners dilemma scenarios. Second, it aims to develop a new model to explain strategic decision-making which reflects previous neurobiological findings showing that different brain circuits are responsible for self-referential processing and understanding others. The model stems from the actor-critic framework and incorporates multiple critics to allow for distinct processing of both self and others state. We validate the biological plausibility and transferability of our algorithm through comparisons with experimental data from human on the iterated prisoners dilemma game.