Figure 3

Choice with probabilistic reinforcement. (a) Preoperative patterns of choice. There were no differences between the groups preoperatively. (b) The first three postoperative sessions. Transiently, AcbC-lesioned rats exhibited relative indifference between the two alternatives; their preference did not differ significantly from 50% at any large-reinforcer probability. As a result, AcbC-lesioned rats preferred the large, unlikely reinforcer more than shams did when its probability was 0.0625 and 0.125 (## p < .01, lesion × probability interaction; * p < .05, ** p < .01, comparison to shams at individual probabilities). However, both groups were influenced by the large-reinforcer probability (p ≤ .004). (c) The last three postoperative sessions on the same basic task. By this point, AcbC-lesioned rats preferred the large reinforcer less when its probability was 0.5 or 1 (## p < .01, interaction; * p < .05, simple effects). Again, both groups were influenced by the large-reinforcer probability (p < .001). (d) When the 4-pellet reinforcer and the 1-pellet reinforcer were both certain, all groups preferred the 4-pellet reinforcer, and when the 4-pellet reinforcer was always very unlikely (delivered with a probability of 0.0625) and the 1-pellet reinforcer was certain, all groups preferred the 1-pellet reinforcer, with no differences between AcbC-lesioned and sham-operated rats. This indicates that both groups discriminated the reinforcers themselves and discriminated their probability of delivery. (e) Choice following further training in which the large-reinforcer probability increased, rather than decreased, across each session. The pattern of choice is similar to c, in that AcbC-lesioned rats were risk-averse compared to shams, i.e. less likely to choose the large, unlikely reinforcer (### p < .001, interaction; * p < .05 and *** p < .001, simple effects). The similarity to c, despite the reversed task order, also indicates that subjects' choice reflected the probabilities in force rather than the order within a session.