Abstract
Our visual system readily groups dynamic fragmented input into meaningful objects. Yet, how the brain represents such perceptual grouping remains unclear. Here we therefore used fMRI techniques to explore the precise topographic profile of brain activity underlying the perception of two bistable stimuli, each composed of an array of dynamic elements: the translating diamond (Lorenceau & Shiffrar, 1992) and the spinners (Anstis & Kim, 2011). When perceived globally, these stimuli appear as shape objects translating along a single motion path; when perceived locally, they appear as separate elements translating along individual motion paths. In two experiments, we measured brain activity while healthy observers viewed these stimuli and reported their percepts. Additionally, we estimated the population receptive field (pRF) for each voxel in visual cortex (Dumoulin & Wandell, 2008) and used these to back-project the brain activity during stimulus perception into visual space. We observed a large-scale reduction of V1 activity to the global state of either of these stimuli compared to the respective local one. This was accompanied by a general increase of activity in higher object-sensitive cortex—a pattern closely replicating previous work involving similar stimuli. Strikingly, whereas V1 deactivation was particularly prominent along the horizontal meridian for the translating diamond, the peak of V1 deactivation was confined to the center of the stimulus display for the spinners. These signatures roughly corresponded to the motion paths of the inferred shapes during the global percept, indicating they could serve as a label signaling a moving grouped entity. Our findings suggest a possibly crucial computational role of V1 during perceptual grouping of dynamic fragmented input and demonstrate that pRF-based back-projection techniques can potentially reveal hitherto undetected neural signatures of visual perception.
Meeting abstract presented at VSS 2017