Researchers trained head-fixed mice to categorize amplitude-modulated noise according to its temporal rate while repeatedly imaging neurons in the auditory cortex. They compared deep-layer L5 extratelencephalic neurons with neurons in layers 2/3 and another deep-layer group, L5 intratelencephalic neurons.

With learning, the L5 extratelencephalic neurons changed their sound tuning and developed clear category responses. The same neurons showed this code during the task but not when the mice passively listened to the sounds, indicating that the representation depended on the behavioral context.