Effect of "rights" and "wrongs" on concept identification.

Abstract

4 groups of 32 female Ss each were compared on a 2-category concept-identification task with 6 stimulus dimensions. At the beginning of the task, 1 group received 6 trials on which E said "right" regardless of S's responses, 1 group received 6 "wrongs," and 1 group received 6 trials on which E said nothing. Another group received no prior trials. The group given 6 "rights" was significantly inferior to the other 3 groups in terms of trial of last error and learning rate parameters of the all-or-none model. The other 3 groups did not differ significantly from one another. It was suggested that noncontingent "rights" contribute maximally to random reinforcement effects.

Effect of noncontingent “rights” and random reinforcements on concept identification as a function of the relevant dimension’s cue value

“…The results of the present experiment are consistent with previous findings that an initial series of noncontingent "rights" (Merryman et al, 1968) or random reinforcements (Holstein & Premack, 1965;Levine, 1962) retarded learning in CI. The reliability of these effects underscores the inadequacy of "no memory" models (Bower & Trabasso, 1964;Restle, 1962) of CI.…”

“…The reliability of these effects underscores the inadequacy of "no memory" models (Bower & Trabasso, 1964;Restle, 1962) of CI. Merryman et al (1968) suggested " ... that noncontingent 'rights-rather than noncontingent 'wrongs' contribute maximally to the retarding effects of RR in CI learning." From this view, the additional finding of the present experiment, that 6R and 6RR did not affect performance differentially, is compatible with earlier findings that the effects of RR reached asymptote within six trials (Holstein & Premack, 1965;Levine, 1962).…”

“…19 (4) hypotheses also occurs after "rights" (Levine,1966). Merryman, Kaufmann, Brown, & Dames (1968) made a direct comparison of the separate effects of "rights" and "wrongs" in a CI task. One group began the task with six trials on which E said "right" regardless of S's response.…”

“…The present experiment afforded a direct comparison of the effects of six initial "rights" (6R) and six initial random reinforcements (6RR) as a function of the presence or absence of the relevant dimension as a cue during noncontingent reinforcement. It was hypothesized that the detrimental effects of 6R (Merryman et al, 1968) and 6RR (Holstein & Premack, 1965;Levine, 1962) resulted from S's elimination of the correct hypothesis after misinformative "rights." In this event, neither procedure (6R, 6RR) should be detrimental to S's subsequent performance in a CI task if the relevant dimension is held constant until the initial, noncontingent phase has ended, whereas both procedures should retard leaming if the relevant dimension is varied randomly.…”

Memory for intratrial events in feature identification.

“…The third class of studies provides additional indirect evidence that subjects remem-ber hypotheses. In this group are paradigms in which either (a) misinformative feedback was given (Levine, 1962;Merryman, Kaufmann, Brown, & Dames, 1968), (b) the solution was changed during the course of the problem, (Trabasso & Bower, 1966), (c) several problems were worked on concurrently (Restle & Emmerich, 1966), (d) response latencies were obtained as an estimate of the number of hypotheses currently under consideration (Erickson, Zajkowski, and Ehmann, 1966), or (e) blank trials were used to measure the size of the subject's hypothesis sample as a function of how much information he had been given regarding solution (Chumbley, 1969;Levine, 1969). These various techniques are similar in that direct probes for trial events were absent.…”