Response preference and choice-sequence preferences: I. Regression to alternation

Abstract

Five numerical binary-choice tasks, of varying difficulty, were administered to normal fourth-through eighth-grade children and adolescent educable retardates. Retardates alternated more than normals in all tasks. Alternation was greatest for the most difficult task. This nonadaptive behavior that is lower in the developmental hierarchy may be called "regression to alternation."Normal children and adolescent educable retardates tend to use alternation as their primary strategy in a wide variety of tasks from binary-choice guessing games to binary-choice learning tasks. Regardless of the reinforcement schedule these populations utilize alternation more than any other response sequence. When retardates do try some other strategy and receive nonreinforcement, they will switch to alternation even if it does not improve their percentage of reinforcements.In order to test the hypothesis that retardates and normal children choose this lower form of behavior in an objectively soluble binary-choice task that is too difficult for them to solve, first-through fifth-grade normal children and adolescent educable retardates were administered a spatial orientation task (Gerjuoy, Winters, & Hoats, 1966) in which they were required to respond "left" or "right" at each choice point of the Road Map Test (Money, 1965). Only the fourth-and fifth-graders and the retardates had significantly fewer errors than chance. All groups alternated significantly above chance with retardates producing the most alternations. The younger normals tended to alternate more than the older normals.If a soluble task has severallevels of difficulty, we can assume from these results that as the task becomes more difficult, more Ss would exhibit alternation, and if it then becomes easier, fewer Ss would alternate above chance. In order to test this hypothesis, the present experiment was designed to include tasks with several levels of difficulty. Subjects and ProcedureSs were 28 adolescent educable retardates (CA = 15.8 years, (J' = 1.5 years; IQ = 63, (J' = 9 .7) and 263 fourththrough eighth-grade normal children.Ss were presented five numerical binary-choice tasks with the same format. Each task had a number centered at the top followed by pairs of two-or three-digit numbers. Task 2, for example, was headed by the number "70." Each pair of tWO-digit numbers included one with a zero in the units column. Tasks I, 2,4, and 5 had 10 trials, whereas Task 3 included 20 trials. The 10-trial tasks were presented consecutively, two to a page; the 20-trial task was presented separately after the first Psychon. Sci., 1967, Vol. 7 (12) pair of tasks. Each task was judged for level of difficulty; the first and last tasks were judged easiest, the second and fourth more difficult, and the third most difficult. Ss, who were tested in their classrooms, were instructed to circle the number in each pair that was more similar to the top number. Examples with easier tasks were placed on the blackboard, and E worked through the examples with each class to assure the instru...

An Assessment of Hypothesis Testing in Mentally Retarded Adolescents

“…A possible reason for the educable group performing differently than in previous studies is that less complex stimuli were employed in the present study. The educable subjects were not forced to "regress" to a Response-set hypothesis (Gerjuoy & Winters, 1967). Apparently, trainable subjects are quick to try a position hypothesis not only when complex stimuli are used, such as the Japanese symbols of the Gerjuoy and Winters (1965) study, but also when the very simple stimuli of the present study are the discriminanda.…”