Thalamocortical bursts encode reward contingencies and drive associative learning

Abstract

Abstract Learning requires adaptive changes in neuronal circuits, but how neurons encode learning content in their activity patterns to construct memories remains poorly understood. Using longitudinal multi-site recordings in freely moving male mice performing a sensory discrimination task, we discover the emergence of burst-coding neurons (BCNs) across cortical, thalamic, and extrathalamic regions. BCNs encoded task rules through the presence or absence of bursts, with their proportion increasing as learning progressed. Decoding analyses reveal that BCNs act as the principal carriers of rule information within the thalamocortical system. BCN burst rates scaled with stimulus valence, collapsed when contingencies were degraded, and inverted after repeated rule reversals, demonstrating that bursts dynamically track associative context during learning. Indeed, pharmacological and focal genetic suppression of thalamocortical bursting disrupted learning and task performance, establishing neuronal bursts as context-sensitive drivers of associative learning. These findings identify a burst-based neural code for stimulus–outcome associations in the thalamocortical system and provide causal evidence linking cellular firing dynamics to reward contingency learning.