State- and Circuit-Dependent Opponent Processing of Fear.

The presence of valence coding neurons in the basolateral amygdala (BLA) that form distinct projections to other brain regions implies functional opposition between aversion and reward during learning. However, evidence for opponent interactions in fear learning is sparse and may only be apparent under certain conditions. Here we test this possibility by studying the roles of the BLA→central amygdala (CeA) and BLA→nucleus accumbens (Acb) pathways in fear learning in male rats. First, we assessed the organization of these pathways in the rat brain. BLA→CeA and BLA→Acb pathways were largely segregated in the BLA but shared overlapping molecular profiles. Then we assessed activity of the BLA→CeA and BLA→Acb pathways during two different forms of fear learning-fear learning in a neutral context and fear learning in a reward context. BLA→CeA neurons were robustly recruited by footshock regardless of where fear learning occurred, whereas recruitment of BLA→Acb neurons was state-dependent because footshock only recruited this pathway in a reward context. Finally, we assessed the causal roles of activity in these pathways in fear learning. Photoinhibition of the BLA→CeA pathway during the footshock US impaired fear learning, regardless of where fear learning occurred. In contrast, photoinhibition of the BLA→Acb pathway augmented fear learning, but only in the reward context. Taken together, our findings show circuit- and state-dependent opponent processing of fear. Footshock activity in the BLA→Acb pathway limits how much fear is learned.