Bayesian Integration in Sense of Agency: Understanding Self-Attribution and Individual Differences

Traditional models of the sense of agency provide theoretical frameworks to understand the processes underlying the sense of agency and the potential neural basis in the brain. However, there is a lack of understanding of where large individual differences in this subjective feeling when integrating multiple cues emerge from. This study developed a Bayesian integration model to explore individual differences in the sense of agency. We hypothesized that the variance of the likelihood distribution and the prior belief are associated with individual differences in sensitivity and the criterion of the sense of agency, respectively. More importantly, the variance of likelihood distribution likely originates from the precision of sensorimotor signals. Behavioral results showed that the weightings of integrating sensory cues in agency judgments varies substantially across individuals. Furthermore, the estimated parameters of our model successfully captured these differences, reflecting individual sensitivities and criteria for agency, compared to other alternative models including the additive logistic model, the fixed-weight integration model, and the signal detection theory model. The proposed Bayesian integration model provides critical insights into the mechanisms of the sense of agency and its variability, highlighting the model’s potential for understanding both typical and disordered agency experiences.