Sloan Swartz Summer 2005  — Abstracts

Tatyana Sharpee (UCSF)

Adaptive Decorrelation in the Primary Visual Cortex  

Sensory neuroscience seeks to understand how the brain encodes from

natural environments.  Strong multipoint correlations present in natural

visual, auditory or olfactory signals usually make it difficult to

correctly interpret neural coding of these inputs, and simplified stimuli

are used instead.  Does the brain’s coding strategy depend on the

stimulus ensemble? We apply a new information-theoretic method that

allows unbiased calculation of neural filters (receptive fields) from

responses to natural scenes or other signals with strong multipoint

correlations. We compare responses in the cat primary visual cortex to

natural and noise inputs matched for luminance and contrast. We find that

neural filters adaptively change with the input ensemble so as to

increase the information carried by the neural response about the

filtered stimulus. Adaptation affects the spatial frequency composition

of the filter, enhancing sensitivity to under-represented frequencies in

agreement with optimal encoding arguments. Adaptation occurs over 40

seconds to many minutes, longer than most previously reported forms of

adaptation.