Extradimensional Shift Research Articles

Reversal and extradimensional shifts in abstract and concrete schizophrenics.

Reversal and extradimensional shifts in abstract and concrete schizophrenics.

Overtraining and stimulus change in extra-dimensional shifts

The hypothesis was tested that overtraining would differentially affect extradimensional shifts in which cues do or do not change over the shift transition. No evidence to support this hypothesis was obtained.

Effects of cue associations in concept shifts

An orthogonal design was employed to separate the effects of cue associations and cue dimensions in intradimensional (ID) and extradimensional (ED) shifts. Associations between relevant stimulus cues (words) were varied both between and within dimensions (categories). Two control groups were used to measure nonspecific transfer effects. The results obtained with 150 college students as S s were: (a) no difference between ID and ED shifts in trials or errors to criterion in the shift; (b) positive transfer with high association between the relevant cues in original learning and shift learning, and zero transfer with low association; and (c) the analyses of S s' hypotheses supported the results for trials and errors. The relevance of these data to mediational explanations of shift behavior is discussed.

Intradimensional and extradimensional shift performance of children in a differential conditioning task

The intradimensional and extradimensional shift performance of preschool children was compared with that of second graders in a differential conditioning (single cue presentation) situation. The younger children showed better ID than ED shift performance and their ID-ED difference was greater than that of the older children despite the fact that the differential conditioning paradigm prevents simultaneous comparison of dimensional cue values.

Intradimensional and extradimensional shifts in the monkey within and across sensory modalities.

Intradimensional and extradimensional shifts in the monkey within and across sensory modalities.

Tactual-kinesthetic discrimination shifts in retarded children

Mediational theories of discrimination learning have been supported by shift-performance data in visual-discrimination situations. The present study compared extradimensional (ED), intradimensional (ID) and reversal (R) shifts by retardates in a tactual-kinesthetic (active touch) task. The genearlity of a mediational approach was substantiated by reliably greater ease of the ID shift compared to the ED shift. ID-R differences were negligible. Acquisition-shift differences revealed negative transfer for ED shifts, positive transfer for ID shifts, and an apparent though nonsignificant positive transfer for R shifts. Form represented a more potent dimension than texture for acquisition of the problem by retarded Ss.

Effects of training level on reversal and extradimensional shifts

Reversal (RS) and extradimensional (EDS) shift groups were subdivided into undertrained (UT), criterion trained (CT), and overtrained (OT) conditions. Results indicated: (1) The RS was learned faster than the EDS in both the CT and OT conditions, but Ss in the UT conditions did not differ. (2) Ss in the UT condition took longer to learn the RS than Ss in either the CT or OT conditions, but no significant difference was found between the CT vs OT conditions. (3) The EDS group did not differ significantly with training level. The ORE was demonstrated, but the hypothesis that this ORE might be explained in terms of the increased awareness of incidental cues which has been found in OT was not supported.

Irrelevant stimulus dimensions and dimensional transfer in the discrimination learning of children

This experiment studied the learning of four extradimensional shifts by kindergarten children. These shifts differed with respect to the manner in which the irrelevant stimulus dimension of Problem 2 varied (within vs. between trials) and the status of this dimension in Problem 1 (relevant vs. constant). The main finding was that extradimensional shifts were learned faster when the irrelevant dimension of Problem 2 varied between trials and was constant within trials. This finding suggests that attention is not elicited so readily by dimensions which are constant within trials as by dimensions which vary within trials. The finding also allows for the consistent interpretation of many of the diserepant results of studies comparing reversal and extradimensional shifts. Performance was not affected by the status of the irrelevant Problem-2 dimension in Problem 1.

Intradimensional and extradimensional shifts in compound concept learning

Following pretraining in a two category concept task with one of two dimensions relevant, Ss were shifted to a second concept task requiring four response categories with two of three dimensions relevant. Training on the second (compound) concept involved either an intradimensional, an extradimensional, or a control shift relative to pretraining. Compound concept learning requiring an IS was shown to he superior to learning requiring an ES, and to yield positive transfer. Compound concept learning requiring an ES did not produce negative transfer.

Variables affecting the performance of preschool children in intradimensional, reversal, and extradimensional shifts.

Variables affecting the performance of preschool children in intradimensional, reversal, and extradimensional shifts.

Performance of preschool children on three discrimination shifts

After learning an initial two-dimensional discrimination, preschool children were given reversal (RV), intradimensional (ID), and extradimensional (ED) shifts. RV and ID shifts were learned faster than ED shifts, a finding contradictory to the mediational theory of Kendler & Kendler (1962).

The role of cue value, cue novelty, and overtraining in the discrimination shift performance of retardates and normal children of comparable discrimination ability

Twenty-four normal kindergartners and 24 adult retardates, matched on discrimination performance, were each given four 2-stage discrimination shift treatments. These constituted half the treatments of a 2 × 2 × 2 factorial design in which the 3 factors were (1) Stage-1 positive cue retained in Stage 2 vs. Stage-1 negative cue retained (P vs. N), (2) extradimensional shift vs. intradimensional reversal (ED vs. ID), and (3) criterion of 6 consecutive correct on Stage 1 vs. criterion plus 40 overtraining trials (CR vs. OT). The PED and NED conditions compared transfer effects of posotive and negative cues when the novel (new) cue was rewarded on 50% of the trials instead of 100% and 0%, as was the case with the PID and NID problems, respectively. That is, the PED and NED conditions permitted a test of cue value effects that were unconfounded with cue novelty effects. Color and form stimuli were projected only 3 × 4 inch panels, which S was instructed to press. The results (errors in Stage 2) showed little differential transfer to be associated with the retention of the positive or negative cue in either the ID or the ED shifts. However, there was a large Shift effect, the ID shifts being easier than the ED shifts ( p < .001) and also a significant Shift × Overtraining interaction, ED being impaired by overtraining but ID facilitated ( p < .025). In combination, the Shift main effect and the Shift × Overtraining interaction were taken as support for a 2-process interpretation of discrimination learning.

The effects of training and verbalization on reversal and extra-dimensional learning

The effects of undertraining, criterion learning, and overtraining on the subsequent ability of nursery school children to achieve reversal and extradimensional shifts were studied. Sixty-eight Ss were trained on a size sorting discrimination problem and then tested on a reversal shift or an extra-dimensional shift. Three conditions of training were employed, all of which allowed the subjects to solve the training problem but with different amounts of experience. The conditions were: (1) undertraining by informing the subjects of the solution, (2) undertraining as in condition 1, but then giving the Ss 20 additional overtraining trials, and (3) experience with the problem until S achieved an independent solution. The results demonstrated an inverse relationship between the number of trials on the training problem and the number of trials required for reversal learning. The use of language did not facilitate the reversal performance. No relation was found between degree of training and ease of extra-dimensional shifting.

Effects of overtraining and age on intradimensional and extradimensional shifts in children

Ninety-six kindergarten and 96 second-grade children were trained on a two-choice simultaneous discrimination with either color relevant and form irrelevant and variable or the converse. After reaching criterion, half of the Ss received 50 overtraining trials prior to presentation of Problem 2, while the remaining Ss received Problem 2 without overtraining on Problem 1. Problem 2, composed of all new stimuli, had either the same relevant dimension as Problem 1, intradimensional (ID) shift, or had as the relevant dimension the previously irrelevant dimension, extradimensional (ED) shift. Age had a significant effect on Problem 1 but not on Problem 2. The transfer data supported a mediational or two-stage interpretation of discriminative learning in that ID shifts were learned significantly faster than ED shifts and overtraining reliably improved performance. There was no Age × Shift interaction, indicating that younger children utilized some mediational process, dimensional in nature, as effectively as did older children.

Intradimensional and extradimensional shifts as a function of amount of training and similarity between training and shift stimuli.

Intradimensional and extradimensional shifts as a function of amount of training and similarity between training and shift stimuli.

Dimensional shifts and reversals in reterdate discrimination learning

The facilitative effect of overlearning on discrimination reversal was demonstrated with moderately retarded children. The relevant cues of the reversal were different from, but on the same dimension as, those of overlearning. The superior performance of an intradimensional over an extradimensional shift was also demonstrated, using completely new cues for both shift conditions to control for previous confounding of stimulus novelty in dimensional shift experiments. These results are not accounted for by single-link theories of discrimination learning but do support chaining theories in which the inferred, initial response of the chain is assumed to generalize within but not across stimulus dimensions.

EFFECT OF OVERTRAINING ON REVERSAL AND EXTRADIMENSIONAL SHIFTS.

EFFECT OF OVERTRAINING ON REVERSAL AND EXTRADIMENSIONAL SHIFTS.

Overtraining and discrimination shift behavior in children

After receiving 30 overtraining trials in an initial two-dimensional discrimination, 6 yr. old children accomplished a subsequent reversal shift faster than an extradimensional shift. Control Ss trained only to criterion did not differ in speed of learning the two types of shifts. These results are discussed in relation to similar experiments with infrahuman Ss.

Comment: Comparisons of reversal and nonreversal shift

It was argued that studies which employ comparisons between reversal (RV) and extradimensional(ED) shifts in order to test whether organisms use some media-tional process during discriminative learning are invalid for methodological reasons. However, adequate tests for mediation are possible from comparisons of intra-dimensional (ID) and ED shift problems. Data obtained from these comparisons support a mediational interpretation of discriminative learning.

Verbalization and optional reversal shifts among kindergarten children

Summary Previous research has found that the proportion of children who make optional reversal shifts increases with age. One explanation offered for this result is that as children get older they are more likely to respond to discrimination-learning situations by making covert mediating responses. These responses are of such a nature that they facilitate transfer from S+ to S−. In human Ss it is further assumed that these covert mediators are themselves verbal or are activated by verbal representational responses. The present research was designed to determine whether instructing kindergarten children overtly to label S+ and S− would have the effect assigned to the hypothetical covert response. Two experiments were performed in which children were presented with an initial discrimination involving a relevant and irrelevant dimension. After criterion was attained, a second discrimination was presented in which both dimensions were present and relevant and the reinforcement pattern was changed. This discrimination was called the optional shift because S had the option of reaching criterion on the basis of a reversal shift, an extradimensional shift, or a nonselective (inconsistent) shift. A test series which followed the attainment of criterion on the optional shift provided the means of inferring the basis of the optional shift. In each of the experiments one group of children was instructed to precede their choices during the initial discrimination with a sentence that labelled S+ and S− (V Group). Another group was run on the same discrimination but were given no instructions about verbalization (C Group). The major finding was that the V Group had significantly more optional reversers in both experiments. It was concluded that among kindergarten children overt verbal representation has a function similar to that assigned to the hypothetical covert mediating response. A three-stage analysis of the development of the mediational chain was proposed.