Talia Konkle (Harvard): The representational topography of the visual system

Event Details:

Monday, January 10, 2022
This Event Has Passed
Time
1:00pm to 2:00pm PST
Location
Contacts
mbct-center@stanford.edu
Event Sponsor
The Stanford Center for Mind, Brain, Computation and Technology
Add to calendar:
Image

Talia Konkle, PhD

Assistant Professor
Department of Psychology & Center for Brain Science
Harvard University

Host

MBCT

Abstract

The visual system transforms incoming patterns of light into useful representations of the physical world—what are these formats and what pressures govern their formation? My approach starts from the premise that cortical topography—that is, the spatial organization of responses across the cortical surface—is a critical source of information about the format of the underlying representation and the functional roles of brain regions.  I will present empirical work that reveals new factors in the large-scale organization of visual brain responses, and complementary computational work which provides new insight into why we have multiple early visual areas, and how the mosaic of category-selective regions are integrated with the surrounding cortex. Broadly, these results depart from the traditional emphasis on specialization and instead drive to an integrated domain-general account of visual representation, advancing new computational modeling approaches to explore the local and long-range architectural constraints that underlie the organization of the visual system.

Bio

Talia Konkle has been an assistant professor at Harvard since 2015. She received her Ph.D. from MIT in Cognitive Science and did her undergraduate work at UC Berkeley in Cognitive Science and Applied Math.

Curriculum Vitae

Related Papers

[1] Konkle, T. & Alvarez, G. A. (2021) bioRxiv. Beyond category-supervision: Computational support for domain-general pressures guiding human visual system representation. https://doi.org/10.1101/2020.06.15.153247

[2] Long, B., & Konkle, T. (2018). Proceedings of the National Academy of Sciences. Mid-level visual features underlie the high-level categorical organization of the ventral stream. http://www.pnas.org/cgi/doi/10.1073/pnas.1719616115