Event:

The question of how neural populations encode complex, high dimensional stimuli has presented us with an interesting playground where optimality theories meet statistical inference from large-scale neural recordings. In this talk, I will briefly outline our recent efforts to make this connection statistically rigorous, and present a few vignettes related to efficient coding. This influential optimality theory, originally put forward to explain neural responses in the sensory periphery, can be productively extended into new regimes relevant for central neural processing: adaptive coding in the primary visual cortex, texture coding in higher-order visual cortices, and place encoding in the hippocampus. Our results suggest that resource and noise constraints might strongly influence neural representations across the entire processing hierarchy, making them amenable to theoretical prediction.