Abstract
The appearance of illusory contours, as in the Kanizsa square, is thought to result from the perceptual inference of unseen occluding objects. While the neural correlates of the formation of illusory contours have been previously described, little is known about how new sensory evidence affects the associated perceptual inference. Here, we investigated event-related potentials (ERPs) evoked by illusory contours (a perceptual hypothesis) and subsequent motion which broke figure completion (evidence disconfirming the hypothesis). Eleven participants performed an unrelated probe detection task as we recorded electrical scalp activity using EEG; simultaneously, task-irrelevant arrays of four inducers ("pacmen") were presented which either formed a Kanizsa square or were perceptually incomplete. After one second of static presentation, inducers rotated dynamically so as to either support the percept of an occluding surface (hypothesis-supporting) or break the Kanizsa illusion (hypothesis-violating). Consistent with previously observed correlates of perceptual completion (e.g. Murray, Foxe, Javitt, & Foxe, 2004), the initial static presentation of completed Kanizsa squares evoked more negativity than incomplete inducer arrays in lateral occipital electrodes, in the N1 component and between 250 and 350 ms following static presentation. In the dynamic phase, enhanced positivity was noted in frontoparietal electrodes between 200 and 300 ms after motion onset for hypothesis-violating (i.e. completion-breaking) inducer motion when compared to completion-supporting motion. Critically, this effect was attenuated for perceptually incomplete control arrays. The frontoparietal scalp distribution of the violation-related modulation implies the involvement of high-level regions of the cortical visual hierarchy in the interpretation of a visual scene, with feedback therefrom driving the associated perception of illusory contours.
Meeting abstract presented at VSS 2016