Proceedings of the Annual Meeting of the Cognitive Science Society

Adult learners can acquire non-adjacent dependencies (NADs) in visual sequences, yet learning is far stronger for human ac- tions than for dynamic objects (Lu & Mintz, 2023). We asked what makes human action stimuli special by separating two can- didate contributors: appearance-based structure (a visible body with form and texture) versus biological motion cues (joint trajectories). In Experiment 1, participants viewed full-body avatar actions for 6 minutes and were tested on matched point- light displays (PLDs); they rated NAD-consistent triplets as more familiar than positional foils, showing that knowledge learned from rich action displays transfers to sparse motion- only tests. Experiments 2 and 3 then asked whether motion cues alone can support learning when used during encoding. With PLDs throughout, learning was not detectable after 6 min- utes but emerged with extended exposure, suggesting that the human-action advantage stems from stronger support during encoding, not simply from perceiving the stimuli as actions.