Betty J. House

Miniature Experiments in The Discrimination Learning of Retardates

Publisher Summary The area of discrimination learning can be a surprisingly complex one, with even the simplest classes of discrimination problems solved by relatively unsophisticated subjects. Two-choice visual discriminations by children of low developmental level reveal complexities not suggested by traditional stimulus–response (S–R) theories. Such theories describe the solution of discrimination problems by the formation of single-link associations—the subject learns to approach the positive cue and avoid the negative. But experimental and theoretical analyses of the discrimination learning of trainable retardates have led researchers to the conclusion that more than one associative link is formed. They infer a chain of at least two responses: a response of attending to relevant stimulus dimensions and an instrumental response of approach–avoidance to the cues of the relevant dimension. Some measure of the complexity of this idealized conception of the discriminative process can be had by counting the number of different, theoretically possible outcomes of a simple two-choice experiment.

Children'S Discrimination Learning Based on Identity or Difference

Publisher Summary Discrimination learning with identity and difference as relevant cues is a clear example of relational learning. Correctness of choice depends not upon absolute stimulus values but upon the relation between a stimulus and others of the display. To demonstrate successful solution of the problem, a subject must continue to respond correctly when specific elements of the display are replaced or their reward values reversed. An extensive literature on how children learn such problems is now available. Thus, this chapter evaluates this evidence with respect to the developmental theories.

Role of Positive and Negative Cues in the Discrimination Learning of Mental Defectives

This smdy inquires into the role of the positive and negative cues in the visual discrimination learning of a group of mentally deficient children. If the formation of a simultaneous, two-choice visual discrimination is o chronological age: mean 12, range 9-2 to 17; IQ (Stanford-Binet, 1916 revision): mean 44, range 20 to 64. There were 8 females and 10 males. An unbiased selection was made of the lowest mental age group attending school who were able to meet a performance criterion of successful pattern discrimination within 100 trials. Enumeration of Ss in the various diagnostic categories yielded 5 familial, 1 mongolian, 2 epileptic, and 9 cerebral-palsied or other brain-damaged.

Repetition of cues in children's oddity learning and transfer

Abstract The hypothesis that repetition of specific cues interferes with oddity learning was directly tested by comparing two groups of preschool (3 to 5 years of age) children, one with repeated cues and the other with new cues on every trial. Repetition of cues retarded oddity learning confirming previous speculation based on error analyses. Performance on postsolution transfer trials with repeated cues was not affected by conditions of acquisition. Significant age effects were found in oddity acquisition, while there was only weak evidence for agerelated performance differences on transfer trials. Transfer performance on trials with repetition of cues was not related to whether or not the children were informed of such repetition. The results were related to theories of oddity acquisition and transfer.

Attention to components or compounds as a factor in discrimination transfer performance.

Abstract Prior to being tested with standard reversal or extradimensional (ED) shift procedures, two groups of mentally retarded subjects (MA: 4 to years) were pretrained using two different methods. The Component group received two problems which could be solved easily only by attending to the form components. The Compound group was trained on two problems requiring attention to compound color-form cues. It was assumed that the two solution modes would carry over to the third problem, which was for both groups a standard form discrimination with color variable and irrelevant within settings. The fourth problem was either an ED shift or a reversal. The Component group learned reversal faster than ED shift whereas the opposite was true for the Compound group. Subproblem analyses of ED shift performance showed dependent settings for the Component group and independent settings for the Compound group. Results are consistent with the view that compound and component solutions of discriminative problems are achieved by attention to the abstract dimensions of compounds or components and that attention to either of these dimensions can be trained at higher or lower developmental levels even though the salience of the compound dimension may be initially greater for lower level subjects.

Discrimination of symmetrical and asymmetrical dot patterns by retardates

Abstract Retarded children performed on a series of 6-trial dot pattern discrimination problems. Each pattern consisted of 16 or 32 dots in a 16 × 16 cell matrix. In constructing the patterns the same basic subpatterns were either mirrored or repeated to produce symmetrical or asymmetrical arrays with equal redundancy. Symmetry produced a strong facilitating effect on discriminability not predicted by conventional statistical measures of information. Alternative redundancy measures are considered. The effect of IQ (range: 29–82) was surprisingly large, but there were no differences due to MA (range: 4–8 yrs.). Interactions of individual differences with experimental factors are discussed.

Some current issues in children's selective attention.

A large part of the history of learning theory of the last 25 years consists of devising ways to deal with stimulus selectivity. Earlier theories that assumed contiguity of stimulus and response to be a sufficient condition for formation of an association were found wanting. This paper has described some of the ways theorists have attacked the problem. These theoretical solutions were of two general types--two-stage theories with stimulus selection preceding association and theories that altered associative processing. Both types of theories successfully handled selective phenomena, and there is no clear basis at present for preferring one or the other. A question generally neglected by early attention theorists concerns the nature of a stimulus. How is the stimulus complex available to a subject divided into cues or elements that can be selectively responded to? Related questions are whether the experimenters perception of stimulus elements correspond to the subjects view, and whether elements are perceived the same by all subjects. The answer to the last two questions appears to be in the negative. Recent research relative to these questions with children, adults, and nonhuman organisms is reviewed.

Contiguity and imbeddedness factors in the discriminative learning of retardates

Abstract Previous studies show that imbedding a reward within a stimulus and spatial contiguity of stimulus and reward are factors which facilitate 2-choice discriminations. In the present study, the separate and joint effects of the two factors on discriminative learning are studied in a factorial design. Reward imbeddedness was manipulated by presenting the reward within a hollow box and was compared with a nonimbeddedness situation in which the reward was placed in a foodwell under the box. Contiguity was arranged by presenting copies of the discriminative stimulus in the goal situation (either the box or the foodwell) and was compared with noncontiguity in which the reward alone was presented in the goal situation Contiguity facilitates learning but imbeddedness does not reliably improve learning. The two factors interact significantly and in a direction which is consistent with previous findings.

A decremental effect of redundancy in discrimination learning

Abstract The finding was confirmed that pattern discrimination problems are more difficult with two different patterns on each stimulus card (double stimulus) rather than the usual one. performance on double stimulus problems having a single relevant dimension (color or form) was impaired if the two patterns on each card differed along the relevant dimension. Within-card differences along an irrelevant dimension did not reliably depress performance. Support was found for the general principle that additional relevant cues facilitate learning if they are attributes of a single stimulus but retard learning if they are spatially separated cues from the same relevant dimension. It is suggested that double stimuli divide attention or overload memory unless perceptually integrated.

RECALLS VERSUS TRIALS AS FACTORS IN SERIAL VERBAL LEARNING OF RETARDATES

A Markov model for verbal learning, developed by Miller and McGill, was applied to data from a serial learning experiment. Data were obtained from 38 retarded Ss (Mean MA: 105 mo.; Mean IQ: 66) who learned lists of six common monosyllabic words by the anticipation method. The model, which assumes that probability of recall is completely determined by number of previous recalls, corresponded closely to the data. The results were also consistent with Rocks theory that associations are formed in an all-or-nothing fashion.