We will also stream this seminar via Zoom. For the Zoom link, please contact Jessica Marcus: jmarcus@upenn.edu.

Bevil Conway
Senior Investigator, Sensation, Cognition, and Action Section
NEI

Concepts: origins & neural mechanisms

The ability to form and deploy concepts is a hallmark of intelligence. Some concepts such as string theory are sophisticated and must depend on learning, while other concepts like color categories can be more elementary. But are elementary concepts available at birth, or do they also require sensory experience and learning? I will take up three new lines of research that aim to make tractable the origin of concepts and their neural mechanisms, and I will welcome feedback as the work is unpublished. First, I will discuss behavioral work in macaque monkeys to determine whether they have consensus color categories. Because macaques have similar cone photoreceptors and central visual circuits to humans, if they possess consensus color categories, then language must not be a requirement. Conversely, if macaques do not have consensus color categories, then the consensus color categories observed in humans are unlikely to depend innate mechanisms of visual encoding. Second, I’ll describe fMRI experiments in macaques in which we aimed to determine the brain areas that underlie color-shape associations. Such associations are the basis for multivariate visual concepts (such as “yellow crescent” = “banana”). After monkeys spent four years interacting with a set of 2-D colored objects, we scanned their brains while they held in mind object cues (color or shape). Within-cue and cross-cue decoding was significant across much of the brain, including in primary visual cortex. Relative to within-cue decoding, cross-cue decoding increased progressively from posterior to anterior inferotemporal cortex and rhinal cortex, suggesting that the culmination of the ventral visual pathway is a locus for generating, storing, and accessing color-shape concepts. Finally, I will describe work in primary visual cortex that aims to recover spatial receptive fields at the center of gaze during active vision in which monkeys discriminate objects using color or shape cues. This work strives to uncover the impact of overt attention on visual encoding that we think is important for learning concepts.