Informal Neuroscience Seminar - Zhaoping Li | Thursday, May 23, 2024 at 4 pm

Thursday, May 23, 2024

4 pm

Chen 100

Speaker:

Zhaoping Li

Prof. of Cognitive Science, University of Tuebingen

Head of Department of Sensory and Sensorimotor Systems,

Max Planck Institute for Biological Cybernetics

Faculty Host: Markus Meister

A well-known, but arguably less well-understood, observation is an attentional bottleneck that forces vision to select less than 1% of visual input information for deeper processing or recognition. It motivates two developing theories of primate vision that form the foundation of a new framework for understanding vision.  The first theory, the V1 Saliency Hypothesis (V1SH), suggests that primate V1 creates a saliency map to guide attention or gaze exogenously (via the superior colliculus), implying that the bottleneck already starts at V1.  I will briefly outline the theory and some of the extensive neural and behavioral evidence underpinning it. The second, Central-peripheral Dichotomy (CPD) theory, builds on V1SH and hypothesizes that central vision is specialized for seeing (recognizing), and peripheral vision for looking (shifting gaze/attention). Given the bottleneck starting from V1, the CPD theory additionally hypothesizes that mainly central, rather than peripheral, vision enjoys rich top-down feedback (from higher to lower visual cortical areas) that aids seeing in light of the bottleneck.

I will discuss CPD and motivate two theoretical predictions:  First, some visual illusions should be visible in peripheral, but not central, vision, because peripheral vision can be misled by the impoverished feed-forward information that makes it through the bottleneck from V1. Central vision avoids this fate by exploiting top-down feedback to query for more information from upstream areas. Second, central vision should be susceptible to the same illusions when the top-down feedback is compromised, for instance by backwards masking. Both predictions are confirmed in a reversed depth illusion created using random-dot stereograms. These behavioral confirmations suggest paradigms for investigating neural implementation of the CPD framework in behaving animals. I will also outline extensions of the CPD theory to other animal species and multiple/other senses.