BACKGROUND The prediction error hypothesis of dopamine action states that dopamine signals are necessary for the brain to update the predictive significance of cues. Yet, little is known whether D1 or D2 receptor-mediated signals in the nucleus accumbens core (AcbC) are required to learn a reversal of the predictive significance of cues. OBJECTIVE Here we examined the effects of a selective D1 or D2 receptor blockade in the AcbC on learning a reversal of previously acquired cue-reward magnitude contingencies. METHODS Rats were trained on a reaction time (RT) task demanding conditioned lever release with discriminative visual cues signalling in advance the upcoming reward magnitude (one or five food pellets). After acquisition, RTs were guided by cue-associated reward magnitudes, i.e. RTs of responses were significantly shorter for expected high vs low reward. Thereafter, cue-reward magnitude contingencies were reversed. Reversal learning was tested for 12 daily sessions with intra-AcbC micro-infusions being given on sessions 1-6. Subjects received pre-trial infusions of the selective D1 or D2 receptor antagonists, SCH23390 (0.5, 2 microg per side) or raclopride (1, 4 microg per side), or vehicle (0.5 microl). RESULTS Intra-AcbC infusion of SCH23390 (0.5, 2 microg) or raclopride (1, 4 microg) did not inhibit discrimination reversal learning, but the higher dose of each drug increased RTs of instrumental responses. CONCLUSIONS In a visual discrimination task as used here, D1 and D2 receptor-mediated signals in the AcbC seem to be unnecessary in updating the reward-predictive significance of cues, rather, they serve to activate instrumental behaviour.