Hidden State Inference or Continuous Belief Updating during a Dynamic Visuomotor Skill

Adaptive sensorimotor behavior requires individuals to flexibly update their beliefs in response to changes in environmental context. Theories of predictive processing propose that such flexibility arises from hierarchical inference, where higher-level beliefs about hidden states shape lower-level perceptual and motor predictions. Here, we test whether behavior in a naturalistic interception task is better explained by continuous belief updating or by more abrupt shifts driven by state inference, consistent with hierarchical Bayesian learning. Twenty-three participants (9 females and 23 males) completed 160 trials of a VR-based interceptive task across two sessions. Participants attempted to return balls with varying bounciness, probabilistically cued by ball color and a contextual cue (court wall color) that reversed periodically. Gaze location prior to bounce was used as an index of perceptual inference and modeled using Bayesian and reinforcement learning frameworks. Continuous belief updating Bayesian models (the hierarchical Gaussian filter) outperformed associative learning models in predicting participant behavior. In addition, continuous belief updating, rather than discrete state, better explained participants' gaze behavior, despite the contextual state-inference model better anticipating the true structure of the task. Overall, while participants updated beliefs in a Bayesian manner, their behavior fell short of normative optimality, possibly due to limits on causal inference and model construction. This may reflect the specific demands of dynamic sensorimotor control, where the brain prioritizes flexible, real-time updating over explicit structural inference to support fluid, adaptive action.