High-level visual specialization in the brain: linking single neurons to fMRI networks


March 28, 2017 - 12:00pm - 1:00pm
Northwest 243
About the Speaker
David Leopold
Speaker Affiliation: 

 Humans rely strongly on vision to read and interpret the world, including
complex social information conveyed through faces and bodies. The same is true
for nonhuman primates, where visual cognition mediates many aspects of group
interaction. The primate inferior temporal cortex features several well-studied
territories in which neurons respond categorically to faces. These "face
patches" are readily localized using fMRI and serve as important reference
points for investigating neural processing in the ventral visual pathway.
Recent advances in our laboratory have made it possible to record
longitudinally across weeks from individual neurons using chronic microwires.
In my talk, I will describe two experiments that rely upon this method. 
The first experiment measured responses to realistic face stimuli that were
morphed across different identities.  Neural responses were measured
before, during and after perceptual learning. The results provide new evidence
that the brain encodes face identity within a “caricaturization” framework, and
that this encoding is unchanging over time, even as perceptual abilities
sharpen. The second experiment probed the responses of face-selective neurons
during the free viewing of naturalistic videos. We compared single unit time
courses in a local population of neurons in face patch AF (<1 mm3) with fMRI
responses across the brain, yielding a unique correlational map for each
neuron.  With this approach, neighboring neurons showed a striking
diversity in their whole-brain functional maps.  Based on this mapping, we
classified neurons into workable subpopulations. The maps corresponding to
these subpopulations ranged from including only face patches to incorporation
large swaths of retinotopic visual cortex. Together, these findings shed new
light on a local population face-selective neurons, both in terms of their
coding principles, as well as their functional affiliation with other brain