INTERMITTENT CONSEQUENCES AND PROBLEM-SOLVING - THE EXPERIMENTAL CONTROL OF SUPERSTITIOUS BELIEFS

Three groups of college students were asked to determine how points we re earned in a task that allowed the assessment of response variabilit y. All students received points for sequences of eight presses distrib uted across two keys (four presses on each key). One group received a point for each correct sequence, one group received points on a fixed- ratio 2 schedule, and one group received points on a random-ratio 2 sc hedule. There were no significant differences in nonverbal response va riability across the three groups, and the fixed-ratio 2 and random-ra tio 2 groups obtained equivalent point totals. However, participants i n the random-ratio group were significantly more likely to write verba l descriptions of the task that made reference to performance-conseque nce relations that were not in effect. The results demonstrate that su perstitious rule generation is more probable when consequences are ran dom and not merely intermittent.