Flexible updating of reward and punishment contingencies by VTA GABA neurons.

In dynamic environments where stimuli predicting reward or punishment unexpectedly change, it is critical to flexibly update behavior while preserving recollection of previous associations. Dopamine and gamma-aminobutyric acid (GABA) neurons in the ventral tegmental area (VTA) are implicated in reward and punishment learning, yet little is known about how each population adapts when the predicted outcome valence changes. To address this, VTA GABA and dopamine population calcium activity fluctuations were measured with fiber photometry while male and female rats learned to associate three discrete auditory cues with three distinct outcomes: reward, punishment, or no outcome within the same session. Contingency learning was determined by quantifying conditioned responding and probability of approach behaviors. After learning, the reward and punishment cue-outcome contingencies were reversed and subsequently re-reversed. The dopamine population displayed the expected adaptation to learning and contingency reversals by increasing the response to appetitive stimuli and decreasing the response to aversive stimuli. By contrast, the GABA population increased activity to all sensory events regardless of valence. Reversing learned contingencies selectively influenced GABA responses to the reward-predictive cue, prolonging increased activity both within and across sessions. Trial-by-trial analysis further confirmed that sustained GABA activity tracks contingency reversal. The valence-specific dissociations in the directionality and temporal progression of VTA GABA and dopamine neuronal activity suggest that these populations serve distinct roles during reward or punishment associative learning and reversal. These findings also describe a novel role for VTA GABA in behavioral flexibility.