LEARNING RESHAPES CORTICAL DYNAMICS DURING SENSORY-GUIDED DECISION-MAKING

Perceptual decision-making arises from coordinated activity across brain regions that transform sensory information into goal-directed actions, with higher-order cortical areas playing a central role. The posterior parietal cortex (PPC) and the anterior lateral motor cortex (ALM) are key areas involved in linking sensory evidence to upcoming choice and action. Both encode behaviorally relevant sensory signals, with PPC more strongly representing sensory signals predictive of upcoming choices, whereas the ALM predominantly maintains persistent decision-related activity associated with motor planning.
These neural representations are strongly context-dependent: while sensory stimuli evoke little or no response in naïve animals, learning behaviorally relevant stimulus-outcome associations increases stimulus representations in trained animals. Learning-related activity is not only organized across cortical regions but also across cortical layers, with superficial layers biased toward motor preparatory activity and deep layers toward sensory responses. However, how single-cell representations across cortical layers reorganize during learning remains poorly understood.
Using longitudinal multiplane two-photon imaging in ALM and PPC, we tracked individual neurons across cortical layers in mice performing an auditory-guided decision-making task from naïve to expert stages. Quantifying sensory- and choice-selective activity revealed heterogeneous changes in single-cell response dynamics during learning. These included shifts toward earlier choice-related activity, neurons with stable response latency but increasing amplitude, the recruitment of stimulus-responsive neurons, and depth-dependent differences across cortical layers. Together, these results provide a longitudinal, single-cell-resolved view across cortical depths and regions of how learning reshapes the circuit mechanisms that transform sensory stimuli into decisions.