Extending homeostasis to thought dynamics for a comprehensive explanation of mind-wandering.

Our thoughts are inherently dynamic, often wandering far from the current situation. Mind-wandering (MW), which is these thought transitions, is crucial for understanding the nature of human thought. Although previous research has identified various factors influencing MW, a comprehensive framework integrating these findings remains absent. Here, we propose that homeostasis has the potential to explain MW and validate the idea through simulations by replicating previous findings. We employed a homeostatic reinforcement learning model where independent drives for the task and others were assigned, and drive reduction became a reward and trained under sustained attention to the response task. To demonstrate that HRL agents can replicate key findings on MW, we had them perform a task widely used in MW research. We then analyzed their response tendencies and response times for validation. We confirmed that HRL agents behave consistently with the empirical results reported in human experiments, which suggest that MW could be under homeostatic control. Finally, we discuss the behavioral and neurobiological commonality between human thought and animal behavior and the possibility that the same principle, homeostasis, controls these phenomena.