Computational bases of domain-specific action anticipation superiority in experts: Kinematic invariants mapping

Abstract

While experts consistently demonstrate superior action anticipation within their domains, the computational mechanisms underlying this ability remain unclear. This study investigated how the processing of kinematic invariants contributes to expert performance by examining table tennis players, volleyball players, and novices across two table tennis serve anticipation tasks using normal and point-light displays. Employing the kinematic coding framework, we established encoding and readout models to predict both actual action outcomes and participants' responses. Results showed that table tennis players consistently outperformed other groups across both tasks. Analysis of the intersection between encoding and readout models revealed a distinct mechanism: while both athlete groups showed enhanced ability to identify informative kinematic features compared to novices, only table tennis players demonstrated superiority in correctly utilizing these features to make precise predictions. This advantage in invariants mapping showed a positive correlation with domain-specific training experience and remained consistent across display formats, suggesting the development of a robust internal model through sustained domain-specific experience. Our findings illuminate the computational bases of domain-specific action anticipation, highlighting the significance of kinematic invariants mapping superiority in experts.