A ventral tegmental area GABAergic projection to the ventral pallidum regulates value-based decision making in mice

Activity of the mesolimbic system is essential for adaptive performance of reward-related behaviors. Within this system, dopaminergic (DAergic) neurons play a critical role in driving motivation to obtain rewards and encoding predictions and error signals during reinforcement learning. However, activity of DAergic neurons shifts from reward presentation to predictive cues following cue-reward learning, leaving open questions about the mechanism of subjective reward value representation. Our previous studies suggest that activity of a GABAergic circuit originating from the ventral tegmental area (VTA) and projecting to the ventral pallidum (VP) scales with unconditioned reward value, independent of effort or associative cue-reward learning. Here, we demonstrate that activity in this pathway consistently reflects unconditioned reward value across extended cue-reward training, unlike DA activity, which undergoes dynamic changes toward the cue and away from a predicted reward. VTA-to-VP GABA activity tracks internal-state-dependent reward value, showing minimal response to water drinking in sated mice and strong activity after overnight dehydration. In a two-option probabilistic operant reward task (PRT), optogenetic activation of this pathway upon reward consumption biased decision-making toward the stimulation-paired option, even when its reward was of lesser value. These findings identify a previously uncharacterized circuit that encodes reward value and contributes to value-based decision-making. Significance StatementAdaptive behavior depends on accurate representation of reward value. While dopamine (DA) signals shift from rewards to predictive cues during learning, the neural encoding of unconditioned reward value has remained elusive. We identify a GABAergic projection from the VTA to the VP that stably encodes unconditioned reward value across extended training, yet tracks changes in internal state such as thirst. Unlike DA activity, this pathway consistently reflects reward consumption and, when stimulated, biases choice toward otherwise less-preferred options. These findings uncover a stable but state-sensitive mechanism of reward value encoding, providing a new framework for understanding value-based decision-making and its disruption in neuropsychiatric disorders.