Computational bases of domain-specific action anticipation superiority in experts: Identifying and mapping kinematic invariants

Experts consistently demonstrate superior action anticipation within their domains, yet the specific computational mechanisms underlying this ability remain unclear. This study investigated the role of kinematic invariants processing in expert performance by recruiting table tennis players, volleyball players, and novices for two table tennis serve action anticipation tasks using normal and point-light displays. Employing the kinematic coding framework, we established encoding and readout models to predict actual action outcomes and participants' choices, respectively. Results showed that table tennis players consistently outperformed others in both tasks. Analysis of the intersection between encoding and readout models revealed that this superiority stems from experts' proficiency in identifying kinematic invariants and accurately mapping them to action outcomes. Notably, experts' invariant mapping ability showed consistency across display formats and correlated with their professional training duration. Our findings illuminate the computational bases of experts' superior action anticipation, highlighting the significance of internal models developed through sustained domain-specific experience.