Learning deficit as a function of anxiety and task conditions

Thesis (Ph.D.)--Boston UniversitySince the chief characteristic of the psychoneuroses is anxiety, a better understanding of the way in which it operates to influence behavior or performance should lead to a better understanding of these disorders. Relatively few psychological investigations have been focused specifically on anxiety, but an approach to the experimental investigation of the effect of anxiety on behavior has been begun by some of the learning theorists. Mowrer's development of the concept of anxiety as drive seems to lead the way for a systematic investigation which has previously been lacking. Anxiety is conceived of as a secondary drive similar in some characteristics to the primary drives such as hunger and thirst in that it may motivate behavior and its reduction may act as reinforcement. In general, the effect of anxiety on performance seems to be related to the nature of the task or performance. In conditioning, considered to be a relatively simple learning situation, increased anxiety facilitates performance or leads to more rapid conditioning. In the relatively complex situation of serial-rote-learning or trial-and-error learning the effect of anxiety is dependent upon the complexity of the task. Montague has demonstrated that on a nonsense syllable list of low association value and high similarity, the anxious subject is inferior in performance to the non-anxious subject, but on a list of high association value and low similarity, the anxious subject's performance is superior to that of the non-anxious subject. Mowrer has suggested that one of the reasons for the beneficial effect of distribution of practice in verbal learning is the fact that the drive or motivation in such learning is some form of anxiety. In the investigations which showed poorer performance for anxious subjects, the learning was by massed trials. It seems possible that the massed practice might at least in part be responsible for the decrenent found in the performance of anxious subjects. The present study was intended to investigate the effects of distributed practice on the anxiety-produced decrement in serial-rote-learning or to test the prediction that the effect of anxiety on performance is a function of the difficulty of the task and the dietribution of practice. In the literature, it has been demonstrated that anxiety or fear in animals acts in a manner similar to drive in that its reduction is reinforcing and it intensifies any response tendencies present during its evocation. In human learning, also, anxiety has operated in a manner similar to drive in that it motivates behavior and its reduction acts as reinforcement. While in animal studies anxiety has been defined as a response to electric shock, in studies using human subjects, the definition has been in terms of psychiatric category or in terms of response to a questionnaire. Previous investigations where anxiety was defined either in terms of psychiatric classification or response to the Taylor Seale of Manifest Anxiety have demonstrated the phenomena noted above concerning conditioning and verbal learning. There is some evidence in the literature on learning for an explanation of the effect of anxiety in complex learning which falls in line with Mowrer's suggestion concerning the beneficial effect of distribution and its relation to anxiety. That is that inhibition, the variable affected by distribution, is related to the level of drive. Kimble investigated reminiscence in motor learning and showed evidence for the hypothesis that inhibitory potential, acting against drive, builds up more in highly motivated subjects than in those who are not highly motivated. Reynolds and others have also shown evidence that the amount of inhibition built up in learning is related to the level of drive as well as evidence for the concept of inhibition as a central process rather than a purely localized sort of muscular fatigue. There is also equally good evidence for a work decrement hypothesis to account for the beneficial effects of distribution in human verbal learning. If inhibition, then, builds up in relation to the level of drive such that high drive subjects accumulate more inhibition during learning than do low drive subjects, the high drive subjects should benefit more from rest periods during learning such that the greater inhibition would be allowed to dissipate. Also, if anxiety acts as drive, then anxious subjects should benefit more from distribution than non-anxious subjects provided the task is a difficult one requiring relatively many trials to learn. The experiment to test the prediction above involves the manipulation of three variables, level of anxiety, task difficulty, and degree of distribution, the dependent variable being performance. The statistical design required was a triple-classification analysis of variance in which significance of the triple interaction would be supporting of the prediction. In the experiment the anxious and non-anxious subjects were selected on the basis of high and low scores on the Taylor Scale of Manifest Anxiety. The anxious subjects were twenty students with scores of twenty-five or above and the non-anxious subjects were twenty students with scores of eleven or below. The tasks were verbal mazes of two levels of difficulty and the difficulty was determined by the number of deviations from a simple right-left sequence. The choice points for the difficult maze were L L R L R R L L R L R R R L R L L R R L. The choice points for the easy maze were R R L R L R L R R L R L L R L R R L R. There were two levels of distribution. In the massed condition the interval between the last item and the initial item was four seconds. In the distributed condition, that interval was fifty-nine seconds. The mazes were typed in capitals on white paper tape with the lefts to the left of the midline and the rights to the right of the midline. An asterisk at the beginning of each list was the initial item referred to above and was used to signal the beginning of the list. The mazes were presented on a memory drum allowing for a two second exposure of each item with a two second interval between the exposures. This allowed four seconds per item. The instructions were of the usual type for anticipation learning. Scoring was kept for each individual in terms of the number of trials to a criterion of two successive correct performances and the number of errors to criterion. Two levels of anxiety, two levels of difficulty, and two levels of distribution with forty subjects resulted in eight subgroups of five subjects each. Three separate analyses were carried out. In addition to analysis of the trials to criterion score and the error score, an analysis of the intial trial error score was carried out. Also comparisons of the subgroup means were made by means of the t test. In the analysis of initial trial error scores, no significant F ratios were found. There were no differences between the groups' performances at the beginning of learning and all differences could be attributed to processes going on during learning. A significant F ratio in terms of trials to criterion for the triple interaction of anxiety, difficulty, and distribution was found, supporting the prediction that the effect of anxiety on performance is a function of both task difficulty and degree of distribution. The F ratio for the triple interaction in terms of trials to criterion was 4.19, significant at the .05 level of probability. The difficulty variance and the variance of the anxiety-difficulty interaction were also significant and in line with expectation. Since the errors to criterion score was a Poisson type of distribution, a transformation score was used to carry out this analysis. The triple-interaction F was 34.7, significant beyond the .01 level of probability. In this analysis none of the other variance ratios was significant. This analysis also supported the experimental prediction. Comparison of the subgroup means both in terms of the trials to criterion score and the error transformation score showed that the anxious group did benefit more from distribution on the difficult task than did the non-anxious. Neither group benefited from distribution on the easy task. As in previous investigations, the trend here with massed practice was better performance for the anxious on the easy task and better performance for the non-anxious on the difficult task. The results, then, gave additional information on the operation of anxiety on performance and supported the use of the conceptual framework of anxiety as drive. A main conclusion from these results is that anxiety is not necessarily a disorganizing or destructive factor in performance. In fact, anxiety results in more potential variability of perfomance. The non-anxious perform within a more restricted range. While the results have direct implications for learning theory to the degree that anxiety is similar to drive and as such are an extension of learning theory, the results are not so directly applicable to psychotherapy. Psychotherapy involves a choice of what is to be learned or unlearned as well as techniques of learning. However, to the degree that psychotherapy is learning, the results are applicable. If anxiety is the main motivation in human learning as Mowrer has suggested, then the experimental results have far wider applicability than to psychotherapy. However, the results have immediate application in understanding an important clinical problem, the operation of anxiety on performance