Figure 6.

ORE. (a) Large reward (r = 1.5). Increasing the number of pre-reversal trials leads to a monotonic decrease in the number of post-reversal trials required to reach criterion. (b) Small reward (r = 0.75). Number of post-reversal trials to criterion varies non-monotonically with number of pre-reversal trials. Initially, the number of post-reversal trials required increases, but eventually the amount of pre-reversal training (≈140 trials) is sufficient to produce the ORE. Fifty simulations were run for each condition. The criterion for having learned the reversed task was at least 10 rewarded responses in the previous 12 responses. (Online version in colour.)