Journal of Abnormal Psychology 1978, Vol. 87, No. 5, 541-5S3

Variables Affecting Stimulus Fading and Discriminative Responding in Psychotic Children Arnold Rincover University of North Carolina at Greensboro Overselective attention in autistic children has proven detrimental to learning when prompt-fading techniques are used, as autistics often respond exclusively to the prompt (the extra guiding stimulus) and fail to learn about the training stimuli. Experiment 1 was therefore designed to investigate stimulus variables in prompt fading that might reduce the attentional requirements for discrimination learning. Two variables were assessed: (a) distinctive versus nondistinctive feature fading, which signified whether a prompt was a feature contained only in S+ or contained both in S+ and in S—, and (b) within- versus extrastimulus fading, which signified whether the prompt was superimposed on S+ during fading or presented spatially separate from S+. The results showed '(a) significant main effects for both variables, due to the success of the within-stimulus and the distinctive feature conditions, and (b) the combination of within-stimulus and distinctive feature fading was the most effective procedure. Since children were required to respond only to a single (pretrained) feature during the highly successful within-stimulus, distinctive feature fading, Experiment 2 was conducted to assess whether children were still responding only to that pretrained feature after fading. The results of probe trials revealed that discriminative responding was maintained when the pretrained feature was made irrelevant, showing that children attended to multiple features of S+, but that it was disrupted when the whole letter containing the pretrained feature was made irrelevant, showing that children still learned about a very restricted portion (i.e., the first letter) of the S+ word. The implications of these results for the treatment of autistic children, and for a theory of overselective attention in autism, are discussed. A major area of research in discrimination learning deals with changes in an organism's responding as a function of stimulus (as opposed to reinforcement) variables. One of the most successful techniques developed through This research was supported by Grant G007802084 from the Office of Education, U.S. Department of Health, Education, and Welfare. The author wishes to acknowledge Richard Leitenen and Norrnan Kravitz for their aid in running several treatment conditions and collecting reliability measures, Ivar Lovaas for his insightful discussions, and Jay Solnick and Crighton Newsom for their helpful comments on an earlier version of this manuscript; and very special thanks to Robert Koegel for Ms valuable assistance and comments throughout this project. Reprints may be obtained from Arnold Rincover, Department of Psychology, University of North Carolina, Greensboro, North Carolina 27412.

this research is called prompt fading. Since Pavlov (1927) first demonstrated the effectiveness of fading, these procedures have been used successfully, with a wide variety of subjects and learning tasks, to teach discriminations which were otherwise very difficult or not learned at all. Fading has been used success fully with animals (Lawrence, 1952, 1955; Terrace, 1963a, 1963b), autistic children (Ferster & DeMyer 1 962 ; Koegel & Rin_ in,f r>o -rr i cover > 1974' Metz> 196S! Rmcover & Koegel, 1977a; Risley & Wolf, 1967), retarded children (Dorry & Zeaman, 1973; Sidman & Stod. , _ , ,0 ^r> dard > ^66, 1967; Touchette, 1968, 1969, 1971), normal children (Cheney & Stein, 1974; Moore & Goldiamond, 1964; Storm & _. , . ,..„, _, , „ „. «n*(cf. Lipinski & Nelson, 1974). The median reliability for recording subjects' responses was 100% (range: 93.8%-100%).

Results Stimulus variables influencing jading. Table 1 shows the results for individual children on each of the four fading procedures. A 1 entered in the table signifies that the child acquired the target discrimination (i.e., five consecutive correct responses on the final fading step) and a 0 entered in the table signifies that the child did not acquire the target discrimination (i.e., erred at the same fading step on two consecutive attempts). In general, the children learned the target discriminations when the within-stimulus fading procedures were used: All eight target discriminations were acquired in the WSDF conditions, and four of eight discriminations were acquired with WSNDF fading. On the other hand, the children generally did not learn the target discriminations when the extrastimulus procedures were employed: only one of eight discriminations was acquired during ESDF fading, and no discriminations were acquired with ESNDF fading. Significant main effects were found, using the Cochran Q test,1 for each of the variables within- versus extrastimulus (p < .001) and distinctive versus nondistinctive feature (p < .01); the interaction effect did not approach significance. These results clearly show that the variable within- versus extrastimulus fading contributed to the number of discriminations acquired. Both of the within-stimulus conditions produced significantly more learning than did either of the extrastimulus conditions. These

RINCOVER

data also show that the variable distinctive versus nondistinctive feature significantly influenced the number of discriminations acquired. However, it is noteworthy that one of the nondistinctive feature conditions (WSNDF) was also quite successful. Finally, the combination of within-stimulus and distinctive feature fading was clearly the most effective training procedure, as it produced significantly more learning than any of the other combinations; further, it appeared to be an "errorless" fading procedure, as no more than a single error occurred for any child. It should also be noted that there were no significant differences found between naive (undergraduates) versus knowledgeable (experimenter and graduate assistant) trainers on any of the four training conditions, using Fisher's exact probability test (Siegel, 1956, pp. 96-104). Analysis of failures during extrastimulus fading. A major result reported above is that the within- versus extrastimulus variable had a great effect on the number of discriminations learned. Since the pretrained feature was removed on the final step of extrastimulus fading but was still present after within-stimulus fading, one might expect that errors during extrastimulus fading typically occurred on the final fading step. The data for the extrastimulus conditions reveal a fairly consistent pattern of responding. In general, the children continued to respond correctly as the level of exaggeration was reduced, but then erred on the final fading step. A total of IS target discriminations

1 An analysis of variance for a two-factor experiment with repeated (dichotomous) measures on both factors (Winer, 1969, pp. S29-SS8) produced similar results: Significant main effects were found for each of the variables within- versus extrastimulus (p < .01) and distinctive versus nondistinctive feature (p < .025); the interaction effect was not significant (p

Variables affecting stimulus fading and discriminative responding in psychotic children.

Journal of Abnormal Psychology 1978, Vol. 87, No. 5, 541-5S3 Variables Affecting Stimulus Fading and Discriminative Responding in Psychotic Children...
1MB Sizes 0 Downloads 0 Views