The value of anticipated events as a determinant of probability learning and extinction

A Bibliography of Experimental Extinction

An alphabetic listing of articles on extinction published from 1935 to 1970 is presented.

Effects of Event Schedules, Event Values, and Instructions on Ability of College Males to Discriminate Acquisition from Extinction

A probability learning task was used with 60 college males, in a 2 × 2 × 2 factorial design combining event schedules (50% and 100%), event values (high and low), and instructions (state and trial). Trial instructions required Ss to indicate whether they expected an event to occur on the next trial only while state instructions required Ss to indicate in effect whether they were in an experimental state of acquisition or extinction. The two sets of instructions produced distinctly different behavior in acquisition and extinction. Event value interacted significantly with event schedule in acquisition and with instructions in extinction. The partial reinforcement effect appeared under both instructional conditions. The event value by event schedule interaction was in the predicted direction but was nonsignificant. Utility of state instructions for studying acquisition and extinction phenomena would be enhanced by requiring Ss to respond under both state and trial instructions as here.

Effects of Negative Expectancy on Probability Learning

Ss predicted whether a red light or a white light would flash after a ready signal. Under neutral conditions Ss approximated the actual percentage of red lights in 100 trials. “Red” predictions were suppressed significantly when onset of the red light was accompanied by electric shock. Continuous and intermittent shock-light pairing were equally effective in lowering prediction rate.

Effects of Reinforcement Value on Expectancy Statements in Skill and Chance Situations

Previous investigations have frequently noted that rewarding events tend to enhance Ss' stated expectations of achieving the goal. However, these results usually have been obtained in gambling situations. In this study, both skill and chance situations were studied. Four groups were run: skill-nonreward, skill-reward, chance-nonreward, and chance-reward. After 10 trials in which the reinforcement sequence was the same for all groups, a reward was introduced for successful performance for one skill group and one chance group. The prediction was that, following success in a skill situation, introduction of a reward will result in a decrement in expectancies, while in a nonreward skill situation, there will be a corresponding increment. In reward and nonreward chance situations, however, there will be no difference in expectancy behavior. Results indicated a significant difference in expectancy behavior from Trial 10 to Trial 11 between skill-reward and skill-nonreward groups in the expected direction. The corresponding chance groups, however, showed similar results. Thus, the predicted interaction did not occur. The tendency of chance Ss to show results similar to those of skill Ss was hypothesized to be a function of a failure on the part of chance Ss to categorize the experimental task as chance controlled.