Pre-existing visual responses in a projection-defined dopamine population explain individual learning trajectories

Area of Science:

• Neuroscience
• Computational Neuroscience

Background:

• Learning new tasks in complex environments presents challenges due to high dimensionality.
• Individual variability in learning speed is a consistent observation across various experimental paradigms.

Purpose of the Study:

• Investigate the neural mechanisms underlying initial task acquisition.
• Understand why some individuals learn new tasks faster than others.

Main Methods:

• Longitudinal recordings from dopaminergic (DA) axon terminals in mice during a visual decision-making task.
• Analysis of DA responses across the striatum, focusing on reward prediction error coding and pre-existing visual responses.

Main Results:

• DA responses tracked individual learning trajectories and showed widespread reward prediction error coding.
• Contralateral-side-specific visual responses were observed in DA terminals, particularly in the dorsomedial striatum (DMS), even before reward delivery.
• These pre-existing DA responses predicted learning extent and, when activated, improved contralateral performance.

Conclusions:

• Initial conditions of projection-specific and feature-specific DA signals explain individual learning differences.
• Functional heterogeneity across DA projections may bias learning towards specific task features, addressing the challenge of high-dimensional learning problems.