Accumbal D1 and D2 medium spiny neurons control distinct learning parameters in complex behavior

Abstract

Value-based decision-making is at the core of nearly all motivated behaviors and requires the ability to associate outcomes with specific actions and make adaptive decisions about future behavioral action. At the core of value-based decision-making and reinforcement is the nucleus accumbens (NAc), which is integrally involved in learning, selecting, and executing goal-oriented behaviors. The NAc is a heterogeneous population primarily composed of D1 and D2 medium spiny projection (MSN) neurons that are thought to have opposing roles in behavior, in which D1 MSNs promote reward and D2 MSNs promote aversion. By expressing channelrhodopsin selectively in D1- and D2- populations (using D1-Cre and A2A-Cre mice) in the NAc core, we show that mice will nose poke for optical self-stimulation of both cell types, suggesting D2-MSN activity is not inherently aversive. While optogenetic approaches give some information about how cellular activation can modulate behavior, they eliminate the temporally specific neural activity patterns that encode information in behaving animals. To understand how real-time activity in these populations is linked to behavioral execution, we expressed the genetically encoded calcium indicator (GCaMP6f) within D1 and D2 MSNs coupled with in vivo fiber photometry to record from these cell populations in awake and behaving animals during pavlovian and operant conditioning tasks. Utilizing complex reinforcement schedules that allow dissociation of stimulus value, outcome, cue learning, and action, we show that D1 MSNs respond to the presence and intensity of unconditioned stimuli – regardless of value. Conversely, D2 MSNs respond to a mismatch between what is expected and what is received and thus encode errors in prediction in a value-independent fashion. To establish a causal link between the information encoded within these neuronal populations and behavioral output, we next expressed an inhibitory opsin (halorhodopsin) selectively in either D1- or D2-MSNs to inhibit these cells at specific time points during a fear conditioning task. We show that inhibition at select time points, either by inhibiting D2-MSN activity during the predictive cue or D1-MSN activity during the shock outcome, disrupts learning of the association. We thus provide foundational evidence for the discrete aspects of information that are encoded within these cellular populations.