Rats were trained in a shuttle-box to emit shuttle response (SBs) to a 5-s buzzer using three different routines: (1) classical conditioning (DP test), in which the buzzer was followed immediately by 1.5 mA foot-shocks on all trials regardless of responses; (2) an avoidance paradigm without stimulus pairing (DC test), in which the buzzer-shock interval was varied at random between 5 and 35 s on each trial, but in which each SB cancelled the next scheduled shock (avoidance contingency); (3) a typical two-way avoidance situation (DPC test), in which the stimuli were paired on every trial, as in DP, but in which each SB cancelled the next shock, as in DC. In all tests the intertrial interval ranged between 10 and 40 s, and sessions consisted of 50 trials.