Perceptual Science Series
Prediction of perceptual states under a 3D perspective visual illusion using patterns of motor variability
From Monday, March 24, 2014 - 11:00am
To Sunday, March 23, 2014 - 12:00am
Rutgers University, Graduate Program in Neuroscience, Center for Cognitive Science, Laboratory of Vision Research and De
We heavily rely on information from our visual system to help coordinate our daily interactions in the environment. Yet, our understanding of how top-down visual processes modulate the execution of fundamental goal-directed and transitional motor actions remains unclear. Here we utilize a robust 3D depth inversion illusion (3D-DII) to explore how top-down processes influence reach dynamics when participants grab toward a target embedded in a 3D scene, specifically asking whether their motor action is governed by the 3D-DII’s real or perceived geometry. We also explore how speed instructions may affect reach behavior.
Two 3D stimuli were used: (1) a proper-perspective in which perspective-painted cues were congruent with bottom-up signals of binocular disparity and motion parallax, and (2) a reverse-perspective, in which painted cues competed with bottom-up signals, eliciting bistable percepts of veridical depth and illusory reverse-depth in which concave parts are perceived as convex and vice versa.
We find that trajectories are significantly affected by the illusion when no speed instructions are given, particularly with respect to the lengths and curvatures that emerged from different hand orientations to avoid the illusory obstacle or to directly approach the target when viewed veridically. However, the addition of speed instructions modulated reach dynamics, suggesting significant changes in motor strategy. Variability analyses of the normalized peak velocity in the arm’s retraction also reveal informative distributions that blindly separate reaches performed under illusory and veridical states. This provides compelling evidence for the effect of top-down processes on somatosensation, allowing for the prediction of perceptual states based on self-emerging motor signatures.