Jiyoung Hwang¹, Jimin Ju^1,2, Sung-Ho Kim¹; ¹EWHA WOMANS UNIVERSITY, ²UNIVERSITY OF IOWA

People effortlessly recognize objects of various materials and predict their behavior from visual information, drawing on how different materials respond to external forces—fundamentally, pushes or pulls. Despite this, research on causal perception has largely underexplored dynamic interactions involving non-rigid objects, often treating spatiotemporal causality as independent of object properties. Using ambiguous separation events, Ju and Kim (2026) demonstrated that simple kinematic information can simultaneously shift material perception and reverse causal-role assignment: when a line segment elongates with a disc at its end and then separates, pre-separation acceleration evokes the impression of a rigid stick pushing the disc, whereas deceleration evokes that of a disc pulling an elastic band. To test whether these causal-role assignments arise from perceptual mechanisms, we employed a visual adaptation paradigm. Eighty participants viewed one of four adaptor types created by crossing acceleration profile (acceleration/deceleration) with the presence/absence of a 200-ms pause inserted during the line’s retraction at separation. They then judged whether ambiguous test events appeared to depict a push or a pull. Adaptors without a pause revealed bidirectional aftereffects: the acceleration adaptor (pushing-percept-dominant) increased subsequent pull responses, whereas the deceleration adaptor (pulling-percept-dominant) increased push responses. However, a 200-ms pause modulated these effects differentially depending on the acceleration profile. For the acceleration adaptor, the pause preserved the aftereffect. In contrast, for the deceleration adaptor, the pause increased subsequent pull responses, possibly because it disrupted the expected tension-release cue of elastic recoil, thereby inducing the perception of the line segment as rigid rather than elastic. These findings indicate that adaptation occurred at a level integrating motion segments into coherent force dynamics, rather than at the level of low-level motion features. More broadly, they suggest that push–pull interactions are perceptually encoded along a continuum that supports intuitive causal reasoning in everyday physical events.