The role of the orbitofrontal cortex in human adaptive learning under strategic environments

This paper proposes an augmented learning model from a neuroscience perspective. This model contains brain activity data of the orbitofrontal cortex as a predictive variable of human strategic behavior. A Bayesian 3-layer perceptron, which shows the complex relationship between decision factors, was adopted to describe the learning behavior. However, the model's complexity creates the possibility of over tting. To avoid this problem, we adopt the Bayesian estimation and Akaike's Bayesian information criteria, which provide the statistical basis of the model selection, to select the model. Our experience shows that this model can better predict human strategic behavior than do existing behavioral learning models.