生物运动
运动(物理)
感知
运动知觉
心理学
认知心理学
人工智能
神经科学
沟通
计算机科学
作者
Su Zhou,Yaqi Li,Shengyuan Wang,Yutong Zhang,Yongqi Li,Huichao Ji,Xiaowei Ding
摘要
One of the most primitive biological motions is walking. Human vision constantly processes walking movements to anticipate social interactions and avert potential collisions. Counterintuitively, when processing multiple walking biological motions, the visual system optimizes the perception through reference repulsion within a single motion (a bias away from the category boundary direction) and repulsive adaptation in a prolonged time (a bias away from the direction of preceding stimuli). However, how we uniquely perceive walking biological motion across short-term movements remains unclear. Here, by asking participants to adjust the direction until it matches the one they just saw, we uncovered the serial dependence (a bias toward the direction of preceding stimuli) in walking biological motion perception (Experiment 1). We found a similar effect for nonbiological motion (a rotating sphere) but with a greater amplitude (Experiment 2). Furthermore, serial dependence in biological motion coexisted with reference repulsion, while nonbiological motion coexisted with reference attraction. An additional experiment demonstrated an asymmetric mutual influence between biological and nonbiological motion: the attractive serial dependence could transfer between them and was greater from biological to nonbiological motion (Experiment 3). This asymmetry was significantly greater than that observed between inverted biological motion and nonbiological motion, suggesting that the effect is largely driven by the unique social significance of biological motion (Experiment 4). The results suggest that vision implements serial dependence when processing biological motion to maintain a relatively steady representation across time but in a less biased way than nonbiological motion to avoid too much deviation. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
科研通智能强力驱动
Strongly Powered by AbleSci AI