Reversal Shift Research Articles

Problem Shifts and Hypothesis Behavior in Concept Identification

When college-age subjects were given thorough instructions stressing the nature of the solution of the problems and pointing out stimulus dimensions, a reversal shift was more difficult than an extradimensional shift, or there was no difference between them. When subjects were given brief instructions that they were to learn which stimuli fell into each of two categories, the reversal shift was easier. A hypothesis-sampling model accounts for the first set of data; Kendler and Kendler's mediational theory of concept learning accounts for the second.

言語水準と媒介過程

Kendler and D'Amato applied a mediational theory to predict that reversal shift would yield positive transfer for college students, and found, as predicted, that theyd behaved in a manner opposite to the albino-rats on which they had studied before, i. e. reversal shift was faster than nonreversal.Then, H. H. Kendler and T. S. Kendler used the same technique to study concept learning in lindergarten children. Its purpose was again to see whether the behavior of the children would be consistent with the single-unit or the mediational type of S-R theory.However, the results were that the group, as a whole, showed neither positive nor negative transfer in the test situation. Slow learners performed according to the single-unit theory: like the albinorats, they showed negative transfer for reversal shift. Only fast learners performed in accordance with the mediational theory; like the college students, they showed positive transfer for reversal shift.Therefore, Kendler's mediational theory which implies that the mediation is verbal process, was not fully confirmed yet. This was why we planned a series of experiments on kindergarten children and tried to make sure directly how the verbal or symbolic clue mediated between the external stimuus and the overt response.Method: Subjects-The Ss were kindergarten children (ages 3 to 6yrs. old). They were divided into two groups on the basis of their speech levels. Screening test trials were given to 243 children, and those who were able to express their choice verbally were put in the H-group. Those who were unable were put in the L-group. Then, each group was divided into two again, one for reversal shift (R-group) and another for nonreversal shift (NR-group).Then, half of each group was instructed to tell E. their choice before they actually pointed it with a finger, i. e. they were expected to use words, e. g.“little,” “big,” or “green.” Those groups were called “overt verbalization groups”(OV-groups). So our experimental groups numbered 8 in all and8 children in each group.We predicted the results as follows:1) When the Ss were not given any instruction to tell E. their choice, R-shift would be faster than NR-shift in the H-group, and NR-shift would be faster than R-shift in the L-group.2) When the Ss were on a high level in their speech, R-shift would be faster than NR-shift, and OV would have little or no effect on R-shift, but it would interfere with NR-shift.3) When the Ss were on a low level in their speech, OV would reinforce completing R-shift, but would have little effect on NR-shift, or impair it a bit. When the correct verbal clue was given, R-shift would be faster than NR-shift. When they were not given, NR-shift would be completed more quickly than R-shift.Differing from our predictions, the results of our experiments showed that R-shift was always faster than NR-shift not only in the H-group but in the L-group. The OV seemed to yield positive transfer for R-shift, but it was not statistically reliable.Discussion: As all the Ss. we used had already experienced an amount of test trials for screening, we found it possible to think that a. categorization set had been formed by it and it played the role of covert responses that were assumed to mediate between external stimuli and overt responses.

Factors affecting reversal and nonreversal shifts in rats and children.

Factors affecting reversal and nonreversal shifts in rats and children.

Children's shift performance in the absence of dimensionality and a learned representational response

The transfer performance of 6- and 8-year-old children on reversal (R) and half-reversal (HR) shifts was compared under treatments where the same or a new set of responses were used on the shift list. Performance on the reversal shift was superior to that on the half-reversal shift only for treatments where the responses were identical (I) on the two lists. Additional comparisons with performance under a control treatment (involving new stimuli and responses on List 1 and List 2) indicated strong positive transfer on the R-I treatment and negative transfer on the R-C treatment (where the responses were changed on List 1 and List 2). The pattern of results was functionally identical for both groups of children. The performance on the R-I treatment was attributable to the Ss use of a nonmediational “response-switching” strategy.

Effects of Experimentally Induced Irrelevant Stimuli on Reversal Shifts

Effects of Experimentally Induced Irrelevant Stimuli on Reversal Shifts

The effects of overlearning on children's nonreversal and reversal learning using unrelated stimuli

The purpose of the present experiments was to show that the response to explicit dimensions is not crucial to the difference in relative difficulty between nonreversal and reversal shifts in children. To eliminate the possibility of the response to explicit dimensions, the shift problems were defined using unrelated stimuli, and a successive discrimination procedure was employed. Grammar school children ages 7, 8, and 9 served as Ss. In Experiment I no overlearning was given, and nonreversal shifts were easier than reversal shifts. In Experiment II overlearning was given, and reversal shifts were easier than nonreversal shifts. The studies demonstrated that the response to explicit dimensions was not crucial to the change from easier nonreversal to easier reversal shifts during overlearning. Interpretations were offered in terms of learning mechanisms which rely on an increase in shift discriminability during overlearning and/or the creation of stimulus clusters.

Correlation of stimulus and response dimensions in paired-associate learning

Trigrams varying in association value were paired with numeral responses in paired-associate lists with 1:1 S-R assignments. When the pairings systematically correlated the stimulus and response dimensions, performance was facilitated relative to list pairings that were random with regard to positions on the dimensions. Second lists were presented with new stimuli, using either the same rule, a different rule, or a random arrangement, but ceiling effects obscured analysis of differential transfer. The procedure can be related to reversal shift phenomena, and has implications for serial learning.

Selection strategies and reversal-nonreversal shifts in verbal discrimination transfer

Verbal discrimination transfer with structured lists that permit the operation of an intrapair selection strategy was compared for reversal and nonreversal shift conditions. Reversal transfer was significantly superior to nonreversal transfer. However, reversal transfer did not differ significantly from a control condition in which the selection strategy was operative on the transfer list only.

Optional shift behavior of children as a function of age, type of pretraining, and stimulus salience

Optional shift behavior was studied within a factorial design involving four levels of age (nursery school children, first-, third-, and fifth-grade children), three types of pretraining (perceptual training, perceptual plus verbal labeling training, and a control condition), and two degrees of stimulus salience (objects vs. patterned stimulation). The perceptual and perceptual plus verbal pretraining procedures significantly and equally facilitated selection of optional reversal shift, and nursery children made fewer optional reversals than older Ss. Degree of stimulus salience as manipulated here had no effect on shift behavior. The findings are discussed in relation to perceptual, attentional, and verbal mediation theories of learning.

AN ONTOGENY OF OPTIONAL SHIFT BEHAVIOR

Between kindergarten and college age the probability that S will make an optional reversal shift (to the previously negative stimulus of the relevant dimension) increases with developmental level. Conversely, the probability that he will make an extradimensional shift (to the previously irrelevant dimension) or a nonselective shift declines with developmental level. This ontogeny was replicated on 3 sets of stimulus compounds composed of highly discriminable binary values on the color, form, or size dimension. An associated ontogeny of learning facility was also found which appeared due to a transition from an incremental to a hypothesis testing mode. The evidence was interpreted as consonant with the developmental mediation theory which assumes that the ontogenies reflect a transition from a singlestage, incremental learning process to a multistage process that includes symbolic representational responses.

Effects of overtraining on reversal and extradimensional shifts in schizophrenics.

Effects of overtraining on reversal and extradimensional shifts in schizophrenics.

Stimulus control and memory loss in reversal shift behavior of college students.

Stimulus control and memory loss in reversal shift behavior of college students.

Transfer from perceptual pretraining as a function of number of task dimensions

Children (average age 6 years 7 months) learned an initial discrimination (OL) and a reversal shift (RS) with 1 dimension relevant and either 0, 1, or 2 dimensions irrelevant following pretraining which required Ss to make same-different judgments to stimulus objects varying along the dimensions appearing in the transfer tasks. Pretraining did not affect performance in OL relative to a control condition, and performance degraded as a linear function of number of irrelevant dimensions. In RS, perceptually pretrained and control Ss did not differ under the condition of 0 dimensions irrelevant, but with increasing amounts of irrelevant stimulation the performance of the control Ss deteriorated in the same manner as in OL while the performance of perceptually pretrained Ss was unaltered.

The effect of verbalization of relevant and Irrelevant dimensions on discrimination-shift learning

This study explored both the effects of mixed verbalization on shift learning and the possible prepotent stimulus dimensions for 7-year-old children. Significant main effects were found for reversal and nonreversal shifts (p lt.05) and discrimination variables (p lt.05). Although no significant interaction was found, the data indicated that during transfer learning, verbalization of both dimensions facilitated the acquisition of a size discrimination to a greater extent than a brightness discrimination.

Reversal shift behavior and verbalization in two age groups of hearing and deaf children

Four groups each containing 24 children (hearing or deaf, 7 or 12 years of age) were given the Kendler “optional choice” reversal-nonreversal shift task to determine the relationship between language ability and performance in this task. Following completion of the task, Ss were asked to verbalize the concept. The results indicate no difference between hearing children and their deaf counterparts in either speed of learning the first or second concept, or in the proportion of “reversers”, but there was a significant difference between the two ages for all these measures. For the hearing children, verbal reports were not significantly related to task performance. Further analysis demonstrated a significant relationship between verbal report and speed of shifting to the second concept as well as between type of shift and speed of shifting.

Reversal and nonreversal shifts in neurologically handicapped children.

Reversal and nonreversal shifts in neurologically handicapped children.

Effect of variations in a reinforcement complex upon reversal shift performance of kindergarten children

Seventy-two kindergarten children were trained on a two-choice discrimination problem with each correct response being accompanied by a marble and five specifiable stimuli which served as secondary reinforcers. Following initial training, half of the Ss learned a reversal shift with the specifiable stimuli remaining with the response to the formerly positive cue while the other half learned the reversal with the stimuli shifted to the response to the new positive cue. Within each of these two groups, three stimulus-number (5, 3, or 1 of the original stimuli) conditions were used. Reversal speed was a negative function of the commonality of the reinforcement complex associated with the training and test tasks.

Economy of representation and reversal shift behavior

Two experiments sought to investigate the influence the number of words in a representational response has upon the speed of reversal learning. In Exp. I an attempt was made to control the length of the representational response by varying the number of words (2, 4, 8, or 20) in each of the two sorting categories. The expectation was that the representational response for S s sorting related (R) words, instances from the same natural-language conceptual category, would consist of one word (e.g., clothing ) regardless of the number of items in the category. In contrast the size of the representational response for S s sorting categories of unrelated (U) words was expected to be proportional to the number of words in each category. Although the results indicated that a reversal shift was executed more rapidly with R-as compared to U-words, no significant interaction effect was obtained between the relatedness of words and the number of items in each sorting category. In Exp. II the sorting categories consisted of the same number (20) of words but different numbers (1, 2, 5, and 10) of concepts. The expectation was that the size of the representational response would be proportional to the number of concepts within each sorting category. The results were consistent with the hypothesis that the more economical a representational response is, the more effective that response will be in mediating reversal learning.

The effect of dimensional preference on reversal shift behavior of young children

Twenty-eight kindergarten children served as Ss in a discrimination reversal task. Each S was given a preference test to determine the probability of attending to a particular dimension. It was hypothesized that dimensional preference would affect original learning and the tendency to reverse. The results supported the first hypothesis but not the second. It was further found that the effect of preference was primarily due to the performance of reversers. The data were interpreted in favor of a mediational explanation of developmental differences in discrimination learning.

Developmental analysis of reversal and half-reversal shifts.

Developmental analysis of reversal and half-reversal shifts.