Spatial hearing and perceptual inference in dynamic natural environments
* Presenting author
Abstract:
To successfully locate targets and sources of danger in natural environments, the brain must perform a complex inference, often relying on noisy and unreliable inputs such as natural sounds. To solve these inferential tasks efficiently, sensory systems must be adapted to the statistics of natural signals. In the first part of the talk, I will demonstrate how population codes for sound location in the mammalian auditory cortex can be interpreted as an implementation of such inference. I will then describe more general computational principles which may underlie sensory coding for perceptual inferences in dynamic environments which are constantly in flux.