Chicago, IL 60637
United States
PhD Thesis Defense
Wednesday, October 15th, 2:00 pm
Kyle Bojanek, CNS
Palmer Lab
KCBD 1103
“Which bits matter? Information bottlenecks in vision”
Abstract: Biological sensors are inundated with high-dimensional, statistically complex inputs. As these sensors can only encode a finite amount of information, organisms must prioritize measuring behaviorally-relevant aspects of their input drive. To understand what constitutes a “good” measurement, it is necessary to understand what bits of information are most important for an organism’s survival. We study this problem in the retina, which has well-characterized inputs (visual stimuli) and outputs (retinal ganglion cell spikes). Rather than asserting a particular normative notion of good measurements, we examine what information is encoded in the population response of the retina and infer what these bits are best-representing. We construct a one-parameter family of optimal measurements and infer which one best describes retinal spiking activity. Across a range of driving statistics, there is strong agreement between the population response and one of the optimal measurement strategies from our one-parameter family. Interestingly, the particular value of the one-parameter family that best describes the population response changes with input statistics.