A pitfall for the expectancy theory of human eyelid conditioning.

Two simple eyeblink conditioning experiments with random intermittent reinforcement schedules were performed. In Experiment 1, subjects had to rate their expectancy for an unconditioned stimulus (US) on a seven-level scale prior to each trial. As anticipated, expectancy for US increased with a successive conditioned stimulus (CS) alone, and decreased with successive CS-US pairings. However, Experiments 1 and 2 showed that the frequency of eyeblink conditioned responses (CRs) evolved in a direction opposite to that of expectancy changes: CRs increased, whereas expectancy for US decreased, and vice versa. The possible effect of sensitization on eyeblink response was ruled out by the lack of a run effect in an unpaired control group in Experiment 2. These results tend to disconfirm the expectancy theory of conditioning. Although they were explicitly predicted by the conventional "strength" theory of conditioning, an alternative interpretation is proposed within a cognitive framework.