To decipher the coding in superficial layers of primary sensory cortex we first developed a behavioural task that optimally engages computations in the circuits of interest and allows meaningful analysis of behaviourally-relevant variables. Then, the activity of hundreds of neurons in superficial primary somatosensory cortex (S1) was recorded simultaneously in mice that learned to perform the two-choice texture discrimination task. When the first contact is made between the texture and the whiskers, a group of neurons that we labelled “touch neurons” increase their firing rate transiently irrespective of the identity of the stimulus. Subsequently, we found that a subgroup of neurons in layer 2/3 (L2/3) S1 are reliably encoding the identity of the stimulus in their firing rate rather than low-level sensory features related to whisker kinematics. Furthermore, around 150 ms after the peak activity of the stimulus-encoding neurons, we found that a different subgroup of neurons present a behaviourally-relevant decision signal within the same circuit in L2/3 S1. Population-wide trajectories are therefore spanning the entirety from the first touch event to the computation of a behaviourally-relevant decision signal within the perceptual decision-making task. We show that the decision signal encoded is not directly driven by lick-related feedback and that it predicts behaviour and learning. These results substantially shift our current understanding of the function of primary somatosensory cortex from an area solely encoding stimulus information to an area directly involved in the decision-making process.
The results have been communicated at three scientific conferences, and the first manuscript will be submitted shortly.