Cholinergic-dopaminergic synergy and antagonism shape prediction-error broadcasting

Abstract

Cholinergic and dopaminergic pathways control reinforcement learning, yet how their basal forebrain and midbrain populations interact remains unresolved. We simultaneously monitored both systems using dual-cell-type optogenetic tagging and neurotransmitter measurements in mice engaged in psychometric learning. Their activity contributed synergistically to reward prediction-error signaling. However, punishment-inhibited dopaminergic neurons showed antagonistic interactions with cholinergic neurons as well. Optogenetic stimulation and channelrhodopsin-assisted circuit mapping revealed a cholinergic-to-dopaminergic disynaptic inhibitory pathway through basal forebrain GABAergic neurons consistent with this antagonism. Chemogenetic suppression of cholinergic activity reduced dopaminergic responses to reward-predicting stimuli, attenuated punishment-induced dopamine suppression, and impaired learning. These findings reveal how a common underlying circuit architecture can flexibly shape prediction-error signaling through synergistic and antagonistic interactions, with implications for adaptive learning and neuropsychiatric disease.