We trained rats over 8 sessions to bar press for food rewards and recorded the rate of responding across training. A) Rats from all groups acquired an instrumental response, as shown by an increase in response rates across training sessions. B) After one of the two instrumental outcomes was devalued by specific satiety, rats performed a choice test with both levers available for responding in extinction. Rats with unilateral BLA lesions (uni-BLA), ipsilateral NAc and BLA lesions (ipsi), and lesions that disconnect the NAc shell from the BLA bilaterally (disc-shell), made significantly fewer responses on the lever that, during training, resulted in delivery of the now devalued outcome. In contrast, disconnection of the NAc core from the BLA (disc-core) impaired rats’ ability to use outcome value to guide actions, as shown by the lack of a devaluation effect. * = non-devalued greater than devalued at P < 0.05, paired t-test with Bonferroni correction. Error bars = ± 1 SEM (panel A) or 1 standard error of the difference (SED) (panel B), a measure of within-subjects variability.