Abstract
Human observers are remarkably sensitive to the biological motion produced by other individuals, an ability that is vital to scene and person recognition as well as navigation. Behavioral and neurophysiological evidence has shown that biological motion occurs at a relatively higher level of visual processing, integrating the inputs from both object form and motion pathways. Recently, researchers have revealed distinct spatial biases in the perception of an object’s location and shape that are retinotopically selective and specific to an individual observer (e.g., Wang, Murai & Whitney, Proc Roy Soc, 2020; Zito et al., Front Behav Neurosci, 2016). Although these kinds of idiosyncratic distortions occur for low-level features, whether this happens at high levels of visual representation—after information is combined across dorsal and ventral streams, for example—remains unclear. Here, we tested whether there are idiosyncratic individual differences in the perception of biological motion at different visual field locations. We presented videos of point-light walkers with discrete heading directions throughout the visual field, and observers discriminated the walker heading direction (left/right). In the analysis, the point of subjective equivalence (PSE) was calculated based on the psychometric curve fitted onto each observer’s responses at each visual field location. The PSEs from each observer were correlated with their own PSEs from a second re-test session and also with other observers’ PSEs. The results revealed stable observer-specific biases across the visual field in perceived heading direction. The between-subject agreement was substantially lower than the within-subject consistency, demonstrating that each individual has their own unique spatial biases in biological motion perception. Our findings suggest that observers have distinct spatial heterogeneities in the representation of integrated form and motion information, and raise the possibility that low-level inhomogeneities in spatial vision may be inherited by subsequent visual processing, potentially biasing high level object and scene recognition.