Harold J. Fletcher

The Delayed-Response Problem

Publisher Summary This chapter discusses the delayed response problem. It presents an analysis of the delayed-response problem followed by an examination, in terms of this analysis, of the major determining variables consistently demonstrated in research with both normal and brain-damaged primates. There have been many explanations of response deficits and many interpretations of the delayed-response problem. The more global interpretations were offered before the enumeration of the many variables that influence performance. Hyperactivity and distractibility, and incentive value and deprivation are variables that influence delayed-response performance because they affect these initial reactions. A number of studies, in diverse research areas, consistently correlate one observation with decreased delayed-response performance. At the time of baiting, the tester can only try to get the animal to observe the baiting operation or the bait. While experience and skill certainly help in deciding when the animal has observed, any tester will admit that there are instances when this decision is difficult to make.

Monkeys' Performance on Ambiguous-Cue Problems

9 experimentally naive rhesus monkeys were tested on ambiguous-cue problems involving three stimuli: P, the positive or rewarded stimulus; N, the negative or nonrewarded stimulus; and A, the ambiguous stimulus which is nonrewarded when paired with P, but rewarded when paired with N. Either the NA pair or the PA pair was presented on every trial. All Ss were tested with two types of stimuli, planometric plaques and stereometric objects. The results corroborate apparently conflicting previous reports of performance by primates on ambiguous-cue problems with these two types of stimuli. That is, performance on PA trials was superior to performance on NA trials with plaque stimuli, but the converse was obtained with objects. A differential S-R spatial discontiguity inherent in only the plaque version of the problem was identified as one cause of the discrepant results.

Monkeys' solution of an ambiguous-cue problem.

In a WGTA situation 12 monkeys were trained on a 2-choice discrimination task involving 3 stimuli: P, the consistently rewarded stimulus; N, the consistently nonrewarded stimulus; and A, the stimulus ambiguously rewarded depending upon whether it was paired with either P or N on a particular trial. Trials were either of the PA or NA type, and 6 Ss received a random schedule of such trials throughout 12 sessions while 6 Ss received a schedule designed to gradually introduce the two types of trials. The results unambiguously demonstrated that regardless of training method Ss were correct more often on the PA trial. Confirming and extending previous research, these data were interpreted in terms of parallel development of excitatory and inhibitory processes.

Evidence Supporting an Intratrial Interpretation of Delayed Response Performance of Monkeys

Eight 4-yr.-old lab-reared Ss received delayed response problems which were factorial combinations of 3 intertrial intervals (8, 16, and 24 sec.) and 3 intratrial delay intervals (0, 10, and 20 sec.). Errors were directly related to delay duration, but no effect of intertrial interval was measured. Errors were not correlated with general activity during delay intervals, but errors were related to bodily positioning. In a second experiment 6 of the same Ss were given 18-sec. delay trials interrupted by lowering the opaque screen at 1, 6, and 11 sec. for durations of 1, 4, and 7 sec. While both onset time and duration were significant dimensions of screen lowering, only the combination of earliest and longest screen interruption increased errors significantly. Activity increased directly with interruption duration. These results were interpreted as supporting an intratrial performance analysis of the delayed response.

ACTIVITY DURING DELAY INTERVAL AND DELAYED RESPONSE ERRORS IN MONKEYS

Six naive rhesus monkeys were tested on a standard delayed response problem. For the 0-, 5-, and 20-sec. delay intervals employed, 100, 91, and 84% correct responses, respectively, were obtained. Activity scores, defined by photobeam interruptions during the actual delay interval, for all 5- and 20-sec. delay trials correlated .975 with errors on those trials. These data were interpreted as supporting the position that in the adapted S on any given trial an orienting response to the baited foodwell is completely and accurately elicited, and that errors reflect interference with orientation by competing response tendencies, in this case the idiosyncratic tendency for generalized activity.

Discrimination learning by retardates as a function of number of implicit response trials

In each of 8 sessions 10 severely retarded males received 4 new 15-trial object discrimination problems. A problem began with either 0, 3, 6, or 9 initial trials during which a prompt (light cue) merely indicated the object which would subsequently be rewarded. No overt instrumental response was permitted during these prompted trials, but conventional trial-and-error procedures were followed on all remaining trials. Discriminative performance on conventional trials was directly related to the number of previously prompted trials. Confirming and extending previous research, these results forced the assumption of an implicit response which, elicited reliably by the prompt on each prompted trial, in creased the probability of a subsequent correct overt response.

PROMPTED VS TRIAL-AND-ERROR CONCEPT DISCRIMINATION LEARNING IN MONKEYS

8 test-sophisticated rhesus monkeys were tested on concept discrimination problems under conventional trial-and-error (TE) procedures and two forms of prompting, i.e., conditions in which an established cue was temporarily available for solution of the problem. Positive Prompting (PP) included a cue attached to the correct stimulus of the discrimination pair; Negative Prompting (NP), a cue attached to the incorrect stimulus. The data revealed significantly better intraproblem learning following NP than following equivalent amounts of either PP or TE.

Prompted versus trial-and-error initial object discrimination learning by monkeys, retardates, and preschool children

Following pretraining to a prompt (a light cue) naive monkeys, retardates,. and preschool children were tested on 2 36-trial object discrimination problems under either a conventional trial-and-error condition or one involving prompting for the first 6 trials. While monkeys’ performance on trials 7–36 was equivalent for both conditions, human Ss learned more during 6 prompted trials than during either the same number of conventional trials or during as many conventional trials as were needed to make 6 correct responses.

Implicit responses in discrimination learning of retardates

Abstract15 severely retarded Ss received 12 sessions of 4 12-trial object discrimination problems differentiated by the following procedures enforced during the first 6 trials only: a prompt (light cue in front of correct stimulus) appeared or did not, and an instrumental response was permitted or was not. Subsequent intraproblem performance indicated (1) significant learning when only the prompt appeared and no response was permitted, and (2) more learning with prompt plus response than with response alone (conventional trial-and-error procedure). MA was correlated significantly with performance on prompt-only problems but not with trial-and-error performance. These data suggested a differentiation between implicit and explicit response learning processes.

Instructions and observational learning in preschool children

Preschool Ss were given object discrimination problems to solve in the absence of overt instrumental responses. A pre-established prompt indicated the rewarded object on all observation trials (no instrumental response permitted). A Look Group was instructed during, these trials to “Look,” a Point Group to “Point to the toy that has the candy under it.” One, two, or four such prompted trials preceded one nonprompted test trial (instrumental response allowed). Only the Point Group learned to solve the problems. Results were discussed in terms of implicit responses differentially elicited by the instructions.