Beyond single-cell sensory representations in cortex.

Abstract

How is information encoded in the cerebral cortex, and how is it used? We review coding in sensory cortex, primarily in barrel cortex. Traditional single-cell analyses of barrel cortex have revealed features like somatotopy, tuning to angle of whisker deflection, and sensitivity to deflection velocity. A growing number of studies have, however, demonstrated mixed selectivity of individual neurons for sensory, movement, and task-related variables, re-emphasizing the need to consider population activity while investigating neural computation. Population representations may underpin an animal's ability to generalize behavior to novel stimuli and discriminate between stimuli in complex tasks. Additionally, neural 'subspaces' within high-dimensional population activity may facilitate distinct readouts by different downstream brain areas that have simple internal circuitry limiting their own computations. Population-level analyses are useful complements for understanding what primary sensory cortex encodes, how those codes are used, and how they are learned.