Knowledge of prior recall

Knowledge of prior recall

9, 84-86 (1970) JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR Knowledge of Prior Recall1 JOHN A . ROBINSON AND RICHARD A . KULP 2 University of Lo...

234KB Sizes 0 Downloads 84 Views

9, 84-86 (1970)

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR

Knowledge of Prior Recall1 JOHN A . ROBINSON AND RICHARD A . KULP 2

University of Louisville, Louisville, Kentucky 40208 After a single presentation and recall of a list of words, Ss attempted either to reproduce the words they had just recalled (RR) or to discriminate (D) them from unrecalled words when the list was represented. For one-half of the Ss a 45-second activity period was interpolated between original recall and subsequent test, while the remaining Ss were tested without any intervening delay. Performance on the D test was greater than chance, while performance on the R R test was judged to be evidence of selective retrieval of previously recalled words. Discriminative accuracy was unaffected by interpolated activity but recall was significantly reduced following the delay. These results were interpreted in terms of Ss' processing of the feedback from overt recall and of changes in item accessibility.

T h e p u r p o s e o f t h i s e x p e r i m e n t w a s t o assess the adequacy of S's knowledge of the contents of a prior recall in the absence of any pertinent f e e d b a c k f r o m E. I t is a p p a r e n t t h a t S o b t a i n s some independent information about the contents of recall from monitoring his own p e r f o r m a n c e , t h a t is, f r o m r e s p o n s e - p r o d u c e d feedback. But the accuracy and accessibility of such information

remains

a matter

for

empirical determination.

METHOD

asked to indicate for each word, in turn, whether it had or had not been recalled before. The interpolated postrecall activity consisted of working a series of simple addition problems. For one-half of the Ss no activity was interpolated between completion of first recall and subsequent test for knowledge of recall. For the remaining Ss, the interpolated period lasted 45 sec. during which 15 problems were worked at a paced rate. The words comprising the list were 15 high frequency disyllabic nouns each of which began with a different letter. Several randomizations of the list were prepared so that the order of original presentation and representation during the D test would be different for each S.

Subjects and Procedure. Sixty-eight students enrolled in the introductory psychology course at the University of Louisville, 17 in each of the four conditions, served as Ss in this experiment. Participation in experiments was a course requirement. The Ss were randomly assigned to conditions in order of appearance at the laboratory. Standard study and free-recall instructions were given to each S at the beginning of the experiment. The words of the list were then presented on a Stowe memory drum at a 2-second rate with a 1-second interitem interval. Each word was read aloud by S during the exposure interval. A 45-second free-recall period followed list presentation. Recall was oral with E recording S's responses in the order in which they were given. Instructions for the R R or D tests were given to nonactivity groups immediately after completion of the series of addition problems. These problems were presented on a memory drum, one every 3 sec., and S both read the problem and its solution aloud. The test interval was 45 sec. for both assessment procedures.

Design and Materials. After a single presentation and recall of a list of words, Ss knowledge of their recall was evaluated under four different conditions. These test conditions represented the factorial combination of two methods of assessment with two amounts of interpolated postrecall activity. The assessment methods were a re-call (RR) test and a discrimination (D) test. In the R R test, S was asked to recall the words he had just previously reproduced from the list. If other words from the list (or which he believed to be from the list) came to mind, these were to be mentioned as well but were to be identified by S as previously unrecalled words. In the D test, S was shown the list again and 1This research was sponsored by the Air Force Office of Scientific Research, Office of Aerospace Research, United States Air Force, under Grant AF-AFOSR-1008-68. : Now at Bell Telephone Laboratories, Holmdel, New Jersey. 84

PRIOR RECALL

85

without making certain ad hoc assumptions The average number of words correctly about the nature of short-term verbal memory. recalled after the first presentation of the list However, the results of a recent experiment by ranged from 7.59 to 8.00 over the four con- Tulving (1967) suggest a useful baseline, viz., ditions, but the differences between groups percentage of words common to successive were all nonsignificant. Performance on the re-recalls of the list following a single presentaR R and D tests as a function of amount of tion. A tabulation of appropriate entries for interpolated postrecall activity is summarized "Cycle" 1 in Table 1 (p. 181) of his report in Table 1. For the D test, the values reported indicated that 78 7ooof the words produced in represent the mean percentage of words the first recall period were reproduced in the correctly identified as previously recalled (PR) second recall period. The contrasting value of and previously unrecalled (PU) words, respec- 85 ~o words in common obtained in the nontively. With only two response alternatives activity condition of this experiment indicates that Ss were able to selectively retrieve P R the chance level of accuracy would be 50 ~ . The pooled average of 92 ~ correct identifica- words from the pool of items which remained tions quite obviously exceeds chance by a wide accessible following initial list recall. Several qualitative aspects of R R performargin. Two other aspects of performance on mance may be briefly mentioned: (a) the occasthe D test should also be noted: (a) PR and PU words were equally well discriminated; (b) ional P U word produced during the test accuracy of identification was unaffected by interval was, in every case, the last word the interpolation of a postrecall activity retrieved by S and was, in several instances, separated from the run of PR words by a period. The data reported for the RR test in the perceptible pause; (b) the order in which PR left-hand column of Table 1 are the mean words were reproduced closely conformed to percentages of words originally recalled which the original order of recall in the nonactivity were reproduced during the test interval. The condition but showed no systematic relation values in parentheses in the right-hand column to original order in the activity condition; represent the mean percentage of PU words (c) when omission errors, i.e., PR words not which were produced in addition to PR words reproduced during the test interval, were during the test interval. A chance level of plotted against relative ordinal position of performance for this task cannot be determined original recall a characteristic serial-oosition error curve was obtained. 3 The pattern of false-negative errors (i.e., PR words misTABLE 1 identified as PU words) on the D test suggested PERCENTAGE OF PREVIOUSLY RECALLED AND PREa similar relation between original recall VIOUSLY UNRECALLED WORDS CORRECTLY IDENTIorder and discriminative accuracy. In contrast, FIED OR RE-RECALLED AS A FUNCTION OF AMOUNT OF there was no indication of any systematic INTERPOLATED POSTRECALL ACTIVITY effect of order of item presentation on accuracy of state-discrimination. Previously Previously Group recalled words unrecalled words RESULTS

Discrimination No activity Activity Re-recall No activity Activity

DISCUSSION 91.04 90.03

95.22 93.24

The results of this experiment demonstrate that, in general, Ss' knowledge of the contents

85.46 76.42

(10.11) (4.11)

3 A relative measure of ordinal position, viz., percentage of total words recalled, was required because of the variability of initial recall.

86

ROBINSON AND KULP

of a prior recall is quite accurate. However, the two assessment procedures yielded somewhat different estimates of the extent of that knowledge, particularly after a postrecall activity period. The following interpretation attempts to account for these differences. An internal representation is generated by S for each of the words in the list as they are presented for study. These representations are maintained in a short-term "store" and are the basis of S's subsequent retrieval efforts. Hearing a word spoken aloud is evidence to S that that item has been successfully retrieved. This feedback-derived information is then used to "tag" the word's representation to indicate the fact of its recall. The words of a list are accessible for a short period of time during which they can be overtly recalled by S, they remain available for a longer period of time but require some form of cueing to evoke them (cf. Tulving & Pearlstone, 1966). So long as an item remains accessible, information about its recall history is also accessible because the two are stored together. But the recall history of inaccessible items cannot be assessed unless they are returned to a state of accessibility by presenting certain external cues. The tag hypothesis can account for the proficiency of Ss in assessing recall history, for it is the tags which make possible the selective retrieval of PR words during re-recall and are the basis of Ss' judgments on the D test. The hypothesis of changes in item accessibility can account for the difference between RR and D tests and, particularly, for the decrement in productivity associated with the former procedure following postrecall activity. It also accounts for the absence of a similar effect

with the D test because presenting the list again should serve the cue function mentioned above. Omission errors might also be explained in terms of differential item accessibility but discrimination errors could not. The relation between ordinal position in original recall and probability of errors cited previously suggests that a third hypothesis is required to fully account for our data, viz., that the feedback from S's overt recall is processed as a new input with the consequence that some items-notably those in medial output positions--are not properly tagged. Omission errors, then, may be a joint function of differential accessibility and the processing of recall, whereas, discrimination errors may be due to the latter variable alone. This account is admittedly conjectural. But the results of the present experiment, along with several others discussed by Adams (1967, 1968) emphasize the need for a more direct consideration of the processes involved in Ss" monitoring of their performance and of the ways in which such feedback-derived information is used in learning and remembering. REFERENCES ADAMS, J.

A. Human memory. New York: McGrawHill, 1967. ADAMS, J. A. Response feedback and learning. Psychological Bulletin, 1968, 70, 486-504. TULVING,E. The effects of presentation and recall of material in free-recall learning. Journal of Verbal Learning and Verbal Behavior, 1967, 6, 175-184. TULVING, E., & PEARLSTONE,Z. Availability versus accessibility of information in memory for words. Journal of Verbal Learning and Verbal Behavior,

1966, 5, 381-391. (ReceivedAugust 29, 1969)