Sensory computation through the lens of behavior
Animals are capable of fast, accurate movements in response to sensory inputs. House flies can respond to each others' turns in 30ms during a chase, major league baseball players can hit balls with ~100ms to estimate where and when the ball will cross the plate. Performances like this indicate that sensory circuits are efficient, estimating target motion quickly, and that precision is not lost in translation from vision to action. The mapping between sensation and behavior cannot be static if it is to remain efficient. In the natural world, where the statistics of sensory stimuli fluctuate over a large range, circuits need to adapt to maintain neural and behavioral sensitivity. My talk will explore how cortical neurons maintain information rates under changing conditions and how adaptive coding keeps our eyes on the ball.