The present study investigated the overtraining reversal effect in a probability learning situation. Groups of 15 Ss received either 100 or 200 training trials with either .90:.10 or .75:.25 schedules, and then received 100 trials with reversed probability schedules. There were no differences between the 100 and 200 training trial groups during the reversal trials, and thus, no evidence for an overtraining reversal effect. Previous reports of an overtraining reversal effect were attributed to differences in prereversallevels ofperformance.Juola & Hergenhahn (1967) have recently reported finding an overtraining reversal effect in a probability learning situation. In their study, Ss were given either 25 or 50 training trials with either .90: .10, .80:.20, or .70:.30 schedules, and then 100 trials with reversed probability schedules. Their resuIts showed that Ss receiving 25 training trials deviated more from the expected probability levels during the reversal trials than Ss receiving 50 training trials. Since other data (e.g., Estes & Straughan, 1954) suggest, particularly for the .80: .20 and .70:.30 schedules, that the Ss would not have reached the expected probability levels with so few training trials, it is possible that littIe or no overtraining occurred in the Juola and Hergenhahn study and that the differences between their 25 and 50 training trials groups can be attributed to other factors.The present experiment investigated the overtraining reversal effect in a probability learning situation where sufficient trials were given to produce stable performance levels prior to reversal. SUBJECTS The Ss were 60 students from introductory psychology elasses at Washburn University. DESIGN Fifteen Ss were randomly assigned to each of the four groups that resulted from the factorial combination of two levels of training trials (100 and 200) with two probability schedules (.90:.10 and .75 :.25). At the completion of training, all Ss received 100 trials with reversed probability schedules (.10:.90 and .25:.75). APPARATUS The apparatus consisted of three horizontallights and two toggle switches mounted on a screen separating S and E. The center light was used as a signal for the S to guess whether the 1.00
IS2e left or right light would be tumed on. The Ss used the toggte switches mounted below the left and right lights to indicate their guesses. The onsets and offsets of the Iights were controlled by silent Hunter timers. PROCEDURE Subjects were instructed that the purpose of the experiment was to "outguess" the E as often as possible by guessing whether E would turn on the left or right light. Ss were then given either 100 or 200 training trials with either .90:.l 0 or .75:.25 schedules. A trial consisted of the presentation ofthe center light for 2 sec, during which the Ss were required to indicate their guess, followed by the presentation of either the left or right light for 2 sec. The left and right lights were presented according to sequences which reproduced the probability schedules in blocks of 20 trials. ...