Understanding the input-output function of principal cortical neurons and its role in network dynamics is a critical step toward deciphering how information is represented and encoded in the cortex. Pyramidal neurons act as sophisticated computational units, integrating the activity of thousands of synaptic inputs into a coherent output pattern. These computations largely occur in their elaborate dendritic trees, which receive these synaptic inputs and transform them into neural codes. However, the specific computations performed by dendrites in vivo during behaviorally relevant tasks remain poorly understood.
In this talk, I will present our findings on the dendritic mechanisms employed by layer 5 pyramidal tract (PT) neurons to encode motor information in vivo during dexterous motor tasks. By employing two-photon calcium imaging in head-fixed mice performing various motor tasks, alongside an experimental and analysis pipeline we developed, we achieved an unprecedented resolution in correlating the structural features of dendritic trees with their functional outputs.
Our results reveal that the dendrites of PT neurons exhibit highly localized and independent activity, storing distinct motor memories in specific dendritic compartments and generating parallel signals to control different aspects of movement. Additionally, I will present new data illustrating how motor information is reorganized within the dendrites of PT neurons during motor learning, and I will highlight the types of inputs that contribute to this process.