CONNECTOME ANALYSIS METHODS FOR A NEURAL CIRCUIT CENTRAL TO IMITATION LEARNING

Humans learn by observing others. Imitation learning is central to human cognition, yet its circuit mechanisms remain poorly understood.

The zebra finch songbird is a powerful model for studying imitation learning, as juveniles learn to sing by copying a tutor. However, it is unclear how the songbird evaluates its current song performance to better imitate their tutor.

Avalanche (Av), a small nucleus in the auditory cortex, is a compelling candidate for the evaluation of song performance. However, the lack of a complete wiring diagram (connectome) of Av remains a barrier to understanding how this region computes song evaluation.

Here, we plan to employ a high-throughput electron microscopy (EM) connectomics to reconstruct the connectome of Av at two stages of learning: the sensory phase, when juveniles memorise their tutor song, and the sensorimotor phase, when they practise their song. We present a systematic test of the cutting, milling, and imaging parameters of the EM-based connectomics pipeline, which demonstrates the feasibility of the proposed connectomes.

The sensory-phase connectome will allow us to investigate how synaptic architecture is shaped during learning. The sensorimotor-phase connectome will reveal how a mature Av circuit could generate the song evaluation signal.

Together, these datasets will test whether Av circuitry is sufficient to implement song evaluation. This work will provide the first circuit-level account of a core computation underlying imitation learning, with implications for corollary discharge and predictive coding frameworks.