The mechanistic role of psychological expectations in human-humanoid robot cooperation

When collaborating, humans' decisions are influenced by the perceived cooperative willingness of their partners, and choices are continuously adjusted based on such willingness. In this study, we manipulated the cooperative willingness of different kinds of partners (human vs. humanoid robots) and employed a Rescorla-Wagner (RW) reinforcement learning model to investigate how psychological expectations affect human cooperation with nonhuman agents (n = 25 in the final analyses). The results showed similar overt behavioral performance, but distinct underlying learning dynamics and neural mechanisms depending on whether participants interacted with humans or humanoid robots. Specifically, the neural mechanisms underlying these decisions differed between partner types: Interactions with human agents elicited greater activation in reward-related regions of the putamen, whereas interactions with humanoid robot agents involved stronger activation in the precentral gyrus, a region implicated in embodied cognition. Moreover, participants exhibited distinct patterns of functional connectivity depending on the type of partner. Despite behavioral similarities, these findings highlight that psychological expectations shape decision making for human and humanoid robot partners through dissociable neural processes, reflecting type-specific cognitive representations. The findings advance our understanding of the psychological mechanisms underlying human-humanoid robot interaction and provide neuroscientific insight into the design of effective human-humanoid robot cooperation systems.