Rules and mechanisms for efficient two-stage learning in neural circuits

TL;DR

This paper presents a new model of two-stage learning in neural circuits, specifically in songbirds, using stochastic gradient descent to optimize the matching between tutor signals and plasticity mechanisms for efficient learning.

Contribution

It introduces a novel framework linking tutor activity and plasticity rules, with predictive insights into the temporal structure of corrective signals in birdsong learning.

Findings

Derived conditions for matching tutor signals to plasticity mechanisms

Predicted the temporal structure of LMAN's corrective bias in birdsong

Framework applicable to other two-stage learning neural circuits

Abstract

Trial-and-error learning requires evaluating variable actions and reinforcing successful variants. In songbirds, vocal exploration is induced by LMAN, the output of a basal ganglia-circuit that also contributes a corrective bias to the vocal output. This bias is gradually consolidated in RA, a motor cortex analogue downstream of LMAN. We develop a new model of such two-stage learning. Using stochastic gradient descent, we derive how the activity in 'tutor' circuits (e.g., LMAN) should match plasticity mechanisms in 'student' circuits (e.g., RA) to achieve efficient learning. We further describe a reinforcement learning framework through which the tutor can build its teaching signal. We show that mismatches between the tutor signal and the plasticity mechanism can impair learning. Applied to birdsong, our results predict the temporal structure of the corrective bias from LMAN given a plasticity rule in RA. Our framework can be applied predictively to other paired brain areas showing two-stage learning.