2025 Volume 65 Issue 3 Pages 207-211
Computational modeling refers to a framework for representing the neural computational processes underlying behavior using mathematical models, such as reinforcement learning models, and for estimating their parameters and structure from behavioral data. The advantages of computational modeling lie in its ability to utilize information overlooked by traditional analysis methods, thereby enabling the estimation of latent computational processes and capturing detailed individual behavioral characteristics. Furthermore, simulations based on the estimated models can predict behavior in environments beyond experimental settings. In recent years, computational modeling has been increasingly used to characterize behaviors associated with mental and psychosomatic disorders. This article provides an overview of computational modeling, including its advantages and limitations, based on an example of modeling behavioral data from a two-armed bandit task using reinforcement learning models. The discussion also covers the potential contributions of computational modeling and pathways for its application in psychiatry and psychosomatic medicine.
