Figure 1.

Experimental design. A, The plus maze had two start arms (east or west). On each trial, the unused start arm was blocked. The animal could enter one of two goal arms (north or south), only one of which was rewarded. B, The plus maze assessed four different tasks. Two spatial tasks were solved by entering the goal arm in one spatial location (north or south). Two response tasks were solved by executing a specific body turn (right or left) to enter a goal arm. A strategy switch occurred when the task was changed from spatial to response or vice versa. A spatial reversal occurred when the task was changed from one spatial task to the other. Each task included two paths [e.g., the north task was the combination of the west-to-north (WN) and east-to-north (EN) paths]. C, Along each path, the trial was divided into regions and marked by computer with behavioral flags: the start arm (SA), the choice point (CP), the goal arm (GA), and the reward cup (RE). D, Rats in the reversal group (n = 5) were trained initially to go north. They were then tested on five serial spatial reversals (north→south or south→north). Rats in the switch group (n = 4) were trained initially to go left and then tested on five serial strategy switches (left→north or north→left). A rat that completed five strategy switches was then tested on an additional north→south spatial reversal. E, Each reversal or strategy switch was divided into four trial stages. At the start of the testing day, the animal was tested for 20 trials in the previously rewarded task with pseudorandomly ordered start arms (never more than 3 in a row of 1 start arm) (SP, stable performance). Then the reward contingency was changed to the new task. The animal was tested on the new task with alternating pairs of start arms (east–east, west–west, etc.) until it performed eight in a row correct on the new task (learning). The animal was then tested for 20 trials with pseudorandomly ordered start arms on the new task (post-S/R). The following day, the animal was tested on the new task for another 20 trials with pseudorandomly ordered start arms (retention).