Coding Strategies Underlying Visual Processing
Loading...
Authors
Emery, Kara
Issue Date
2021
Type
Dissertation
Language
Keywords
computational , neuroscience , perception , plasticity , vision
Alternative Title
Abstract
Acquiring and representing knowledge about our environment involves a variety of core neural computations. The coding strategies underlying visual perception highlight many of these processes, and thus reveal general design principles in perception and cognition. I will review three studies where I have used different computational frameworks and analyses to address open questions in visual coding. The first project uses factor analyses of individual differences in perception to demonstrate fundamentally different representational structures for the stimulus features of color and motion. In the second project, I have explored visual adaptation in the context of population coding to address controversies regarding which coding schemes are implicated by different patterns of adaptation aftereffects. In the third, I have explored these adaptation effects in the context of Bayesian inference. This approach accounts for the full gamut of known aftereffects within the context of physiologically plausible models and provides principled quantitative predictions for why and how much the system should adapt. Together, these projects draw on the power of formal computational approaches both for analyzing neural representations and for revealing the computations and coding principles on which they are based.