The interplay of prior information and motion cues in resolving visual ambiguity in agent perception

Abstract

Agent perception, a complex cognitive task that involves interpreting an agent’s actions to infer their internal states and adjust our behavior accordingly, is a fundamental aspect of human cognition. Despite its importance, a significant gap persists in our understanding of how low-level factors, such as motion and form information, interact with top-down factors, such as prior information about the agents. To address this gap, we conducted two electroencephalogram (EEG) experiments investigating the interplay of prior knowledge and motion information on the temporal dynamics of agent perception in the human brain. We used human, android and robot agents engaged in various actions. The chosen agents were designed to form a set wherein form information alone could not easily resolve the ambiguity in agent identities. In the first experiment, participants were informed about agent identities be-fore the experiment (Prior Experiment), while in the second experiment, participants remained uninformed (Naive Experiment). We controlled the availability of motion information by presenting stimuli in either video and image formats. We recorded scalp EEG and utilized event-related potential (ERP) analysis and model-based representational similarity analysis (RSA) to uncover the temporal course of agent representation in the human brain. Our results revealed that the processing of agents in EEG depends on the availability of motion information and prior information. Specifically, in the Naive Experiment, agent information was available longer during the still condition than in the moving condition. In contrast, agent information was present for similar durations in still and moving conditions of the Prior Experiment. These findings suggest that prior knowledge and motion information interactively modulate the duration of the processing of agent information. Our results underscore the critical role of prior knowledge and motion cues in shaping the processing of agent information, highlighting the complex interplay between top-down modulation and bottom-up cues in driving the perception of agents and their actions. This study contributes to our understanding of the temporal dynamics of agent perception and the role of top-down and bottom-up processes in this complex cognitive task.