Figure 7: Dopamine release in the nucleus accumbens core tracks perceived saliency.

(A) The Kutlu-Calipari-Schmajuk (KCS) model. The model has 4 core components. 1) Associative component: Based on a Rescorla-Wager type prediction error term. 2) Attentional component: Mismatch between predicted/unpredicted stimuli increases novelty, and in turn, attention to all stimuli in the environment. 3) Perceived Saliency: Novelty, attention, and the physical intensity of a stimulus determine perceived saliency. 4) Behavioral response component: Perceived saliency is combined with associative strength to produce a prediction of an outcome. For operant responses, the value of an outcome is calculated as the difference between the unconditioned stimulus value before and after the operant response and predicts future responding in a probabilistic fashion. (B-F) Experiments from Figure 4 were replotted to map onto model simulations. Contingency switch from positive reinforcement to punishment represents a worse than expected outcome – i.e. a negative prediction error. (B) Model simulations of reinforcement behavior (grey line) overlaid with experimental data (blue line). Switching from positive reinforcement to punishment (denoted by dotted line) decreased simulated and actual nose pokes (r=0.79, p<0.0001; n=5). (C) Perceived saliency of (grey), and dopamine responses to (blue), the cue increased. (D) The dopamine response to the cue after the first punisher was increased (paired t-test, t₄=2.76, p=0.025). KCS model simulations show that perceived saliency (E, R=0.67, p<0.0001) matches dopamine response patterns during the contingency switch but prediction error does not (F, r=0.092, p=0.2264). (G-M) Optogenetics studies were designed to test whether behavioral responses change in response to increasing dopamine as predicted by changes in perceived saliency. (G) AAV.TH.Cre and AAV.DIO.ChR2 or eYFP were injected into the VTA to achieve specific expression of opsins in dopamine neurons. A fiberoptic was placed above the NAc core to stimulate dopamine release from terminals selectively in the NAc core. (H) A fear conditioning experiment was run where dopamine release was evoked during the cue on 25% of cue-shock pairings. (I) Increasing NAc core dopamine decreased freezing in the ChR2 group compared to eYFP controls (2 way-ANOVA, trial type x group interaction, F_(1,20)=17.84, p=0.0004; Sidak’ multiple comparison ChR2-Tone+Stim vs. eYFP-Tone+Stim, p=< 0.0001; ChR2-Tone+Stim vs. ChR2-Tone only p=< 0.0001; n=5-6) and tone only trials in the same animals (Sidak’ multiple comparison ChR2-Tone+Stim vs. ChR2-Tone only, p=0.0002). (J) Simulations from the KCS model show that this behavioral response is predicted by increased perceived saliency, but not other prediction-based parameters. (K) NAc core dopamine release was evoked at the time of the omitted shock during extinction. (L) Dopamine stimulation prevented fear extinction in the ChR2 group compared to eYFP controls (RM ANOVA, Group main effect, F_1,9=5.90, p=0.038; Last 4 trial block, unpaired t-test, t₉=3.32, p=0.0089; n=5-6). (M) KCS model simulations show enhancing perceived saliency of the omitted shocks prevents extinction of the conditioned response. Data represented as mean ± S.E.M. * p < 0.05, ** p < 0.01, **** p < 0.0001.