JOURNAL
OF EXPERIMENTAL
CHILD
Components Analysis
PSYCHOLOGY
of Verbal
Saul
Learning
by Selective HERMAN
The
18, 488-496
(1974)
in Children-
Reminding1
BUSCHI~E”
R. Korey Depnrtment of Neurology and Rose Research in Mental Retardation and Human Albert Einstein College of Medicine
F. Kennedy Development,
Center
foi
Free recall verbal learning by 5- and S-year-old children was analyzed by selectively reminding them only of items not recalled on the preceding trial (instead of continuing to present the entire list before each recall trial) to show learning by retrieval from long-term storage without presentation Concurrent analysis of long-term storage, consistent and random retrieval from long-term storage, and recall from short-term storage indicat.es that, while 5-year-olds showed slower acquisition than 8-year-olds. lower recall by 5-year-olds also was due to less effective retrieval from lonpterm storage. Repeated retrieval, without any further presentation after an item has been recalled just once, indicates that lower recall by g-year-old children than by adults also reflects retrieval difficulty, since these children showed storage and retention of almost as many items as adults by eventual spontaneous retrieval without further presentation.
This report is intended to illustrate how selective reminding may, by simultaneously evaluating the relative effects of acquisition and retrieval (Belmont & Butterfield, 1969) in the development of free recall verbal learning (Cole, Frankel & Sharp, 1971), provide another useful approach to problems of the development of memory and learning (Flavell, 1971). Selective reminding means reminding the child only of those items which were not recalled on the immediately preceding recall trial, instead of continuing to present all items in the list before each trial throughout learning. This shows when items have been learned, by allowing recall without presentation (since those items which were just recalled are not presented). Therefore, the number of items presented can decrease over trials while t,he number of items recalled increases, until learning may be ‘This work was supported by USPHS Grants MH-17733 to H.B. from NIMH? NS-03356 from NIMS, and HD-01799-07 from NICHD. I thank Tom and Katie Buschkr, Christine Hiney and Susan Berenzweig for their experimental assistance. ‘Requests for reprints should be sent to H. Buschke, Department of Neurology, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461. Copyright All rights
@ 1974 by Academic Press, of reproduction in any form
488 Inc. reserved.
SELECTIVE
REMINDING
IN
VERBAL
LEARNING
489
shown by spontaneous recall of (all) items without any further presentation at all. Because such recall without presentat,ion (after interference due to presentation and recall of other items) indicates retrieval from long-term storage (Glanzer & Cunitz, 1966; Craik, 1968; Tulving & Colotla, 1970)) selective reminding separates retrieval from long-term st,orage (LTR) and recall from short-term storage (STR). Long-term storage (LTS) is estimated by the cumulative number of items which have been retrieved from LTS at least. once, under the defensible assumption t’hat (informatin about) items remain in LTS. More restricted reminding, without any further presentation at all after an item has been recalled just once, indicates that most recall failures represent failure to retrieve from LTS rather than loss of items from LTS, since most recall failures after storage eventually are retrieved again spontaneously without any further presentation (Buschke, 1974). In addition t’o demonstrating relevant components of memory involved in learning at the same time in the same task, selective reminding also increases the opportunity to learn items which have not yet been learned, because reminding the child of just those items which were not recalled on the preceding trial directs attention to those items which may still need to be learned. METHOD
In this study, ten g-year-old and eight 5-year-old children learned to recall a list of ten animals (goat, cat, bear, cow, mouse, sheep, rabbit, monkey, lion, turtle) verbally in any order. Each child was tested individually by the same experienced examiner (S.B. j . The entire list was read aloud once at a 2-set rate before the first recall trial. After the first trial, each child was selectively reminded only of those items which were not recalled on the immediately preceding recall trial (using the same relative order, by omitting items which were recalled on the preceding trial). On each trial the children attempted to recall all of the items in the entire list, and were encouraged to use “extended recall,” i.e., to continue searching for additional items on each recall trial. A list, consisting of items from the same category (animals) was used to make such searching for more items seem reasonable. The 5-year-olds had 8 recall trials and the 8-year-olds had 12 recall trials. Recall without presentation was used to separate retrieval from longterm storage (LTR) and recall from short-term storage (STR), and (cumulative) LTR was used to estimate (the number of items in) longterm storage (LTS). To simplify the scoring of LTR and STR, and to obtain a reasonable estimate of LTR on trial one, items are assumed to be retrieved from long-term storage (LTS) on the trial on which they
490
HERMAN
BCSCHKE
entered LTS, as well as thereafter. Once an item has entered LTS, its subsequent recall is assumed to represent retrieval from LTS, since (information about) such items apparently are retained in LTS; items are assumed to be recalled from short-term storage only until they have entered LTS. While it is not necessary to accept these conventions for scoring LTR and STR, they seem reasonable. Additional information was obtained from two other statistics, which arc not necessarily reskicted to learning by selective reminding. Consistent recall on all subsequent trials (without any further presentation at all) provides a measure of the reliability of retrieval, and separates retrieval from long-term storage (LTR) into consistent LTR and inconsistent or random LTR. Consistent recall on all subsequent trials without any further presentation also may provide a measure of list learning (Buschke, 1973), so that the extent, to which individual items have been learned for independent retrieval (LTS) can be compared with the degree to which those items have been learned as part of a list, (for consistent retrieval together with other items which are also consistently retrieved on all subsequent trials without further presentation), i.e., item learning (LTS) can be compared with list learning (consistent LTR). The use of such consistent retrieval (on all subsequent recall at,tempts without any further presentation) to evaluate list, learning is derived from the generally accepted understanding that a list has been learned when all of the items in that list can be recalled together consistently (without including any other items not in that list). Application of this conception of list learning to the evaluation of list learning before the entire list has been learned results in the reasonable conclusion that, if two Ss can both recall 75% of the items in the list, but the first S can recall the same 50% consistently on all trials while the second S can recall only 25% consistently, the first S has learned more of the entire list. (50%) than the second S (25%). Since we know that items are not recalled independently of one another in list learning (e.g., Tulving & Hast’ie, 1972)) it seems reasonable to regard the number of items which are consistently recalled (together), without any recall failure at all despite the presentation and recall of other items, as a measure of t,he degree of list learning. This measure of list learning also does not require the unnecessary and possibly unwarranted assumption that list learning must be shown by relatively constant clustering or ordering of recall, an assumption which seems especially inappropriate for the evaluation of list learning by free recall. The (cumulative) number of items which have been “recalled at least once” (first recall) provides the first empirical indication that some attention has been paid to such items. Such first recall may also provide
SELECTIVE
REMINDING
IN
VERBAL
491
LEARNING
the maximum estimate of storage that can be obtained which may be useful when recall is not very effective. RESULTS
AND
through
recall,
DISCUSSION
Figure 1 (top panels) shows the free recall verbal learning of these j-year-old and E&year-old children, The uppermost dashed lines show that it took longer for the younger children to recall all items at least once. Recall from short-term storage (STR) contributed little to recall. Although long-term storage (LTS) by 5-year-olds was less than by SFIVE-YEAR
0
2
A
6
EIGHT-YEAROLDS
OLDS
81012
0
2
A
6
8
1012
TRIALS
TRIALS
LTR
80
%
70 60
40 30
RANDOM
LfR
20
10 b 0
2
6 8 TRIALS
A
1012
2
4
6 0 TRIALS
10 12
FIG. 1. Components of verbal learning shown by selective reminding only of items not recalled on the immediately preceding trial: long-term storage (LTS), retrieval from long-term storage (LTR), recall from short-term storage (STR), and consistent LTR or list learning (LIST). Uppermost dashed lines show the number of items recalled at least once (top). Lower panels show the proportion of it,ems in LTS retrieved by consistent and random LTR as well as by total LTR.
492
HERMAN
BUSCHKE
year-olds, retrieval from long-term storage (LTR) by 5-year-olds was even more limited and less consistent, so that the discrepancy between retrieval (LTR) and storage (LTS) by the 5-year-olds increased over trials. This can also be seen in the lower panels of Figure 1, which also show the proportion of (items in’/ LTS retrieved by consistent LTR and by random LTR. Consistent retrieval, as well as total retrieval from LTS, was greater by 8-year-olds than by 5-year-olds; t’he %year-olds apparently learned more of the list as well as more of the items. Figure 2 shows the results of a second experiment in which ten thirdgrade children (mean age 8.5 years, SD 0.53) and ten young adults (mean age = 24.4 years, SD = 2.99) learned a 20-item list of animals by selective reminding. As in the first experiment, the items were iead
16 2 14 E 12 B 10 E
0
26 Z
4 CHILDREN
2
FIRST RECALL
,TOTAL
2
4
6 8 TRIALS
10
RECALL
12 14 TRIALS
FIG. 2. Long-term storage (LTS), retrieval from long-term storage (LTR), consistent LTR (list learning), and recall from short-term storage (STR) evaluated concurrently by selective reminding (top left). Proportion of items in long-term storage retrieved (top right). Number of items recalled at least, once (first recall), total recall, and number of items presented before each recall trial (lower left). Performance of the third-grade children expressed as proportion of t.he performance by young adults (lower right).
SELECTIVE
REMINDING
IX
VERBAL
LEARNING
493
aloud at a 2-set rate for free recall of all items in any order immediately after present’ation. After the first recall trial, only those items not recalled on the immediately preceding trial were presentd again before the next recall trial, in the same relative order as first presented. The top panels of Figure 2 show that retrieval by these children, especially consistent ret,rieval or list learning, was less effective than retrieval by the young adults, although long-term storage by these children also was less efficient than long-term storage by the adults. Since list learning (consistent LTR on all subsequent trials, without any furt,her presentation) presumably involves relating the items in the growing list to each other, it is reasonable that list learning by these children was relatively slower than item learning (LTS) , since they would know less than the adults about the relationship of the items to each other. The lower left panel of Figure 2 shows some basic characteristics of such verbal learning: the number of items recalled at least once (first recall), the total number of items recalled on each trial, and the number of items presented before each recall trial. In recent experiments (Buschke, in press‘), comparing selective reminding with restricted reminding, which requires complete recall without ally further presentation at all of items which have been recalled just once, almost. all items were in long-term storage by the trial of their first recall and consistent retrieval from longterm st)orage was greater in restricted reminding (without any further presentation after initial recall’) than in selective reminding (with continuing presentation of some items). Those findings suggest that the number of items recalled at least once may provide a more realistic estimate of long-term storage, at least when retrieval is poor (since demonst,ration of storage depends on retrieval), and also that continuing presentation of t’hose items not, recalled on the preceding trial during selective reminding may interfere with the subsequent, retrieval of other (previously recalled) items. This is consistent with the findings of Tulving and Hastie (1972) that facilitating the free recall of some items may inhibit the recall of ot’her items (Roediger, 1974). One reason for lower recall by children than by adults (and by younger than older children1 therefore might be t,he continuing presentation of more items throughout learning, which may interfere with the recall of other items not presented. This could account for both their lower total recall and their lower retrieval from long-term storage, as well as their less consistent retrieval from long-term st’oragc. The lower right panel of Figure 2 shows that the recall performance of these children differed from t,hat of adults more in terms of consistent retrieval and total recall than in terms of the number of items recalled at least once (first recall), as the children continued to receive more pre-
494
HERMAN
BUSCHKE
sentations throughout learning. Since first recall (the number of items recalled at least once) is the maximum empirical estimate of storage that we can obtain through recall, the relatively lower first recall by these children seems to provide the best indication of their less effective storage. If the number of items recalled at least once (first recall) does provide a more realistic estimate of storage for children, as well as young adults, when retrieval is less effective, then these data would seem to indicate that the lower recall of these children may be due more to their less effective retrieval than to their less effective storage. St.orage and retrieval during free recall learning also can be evaluated by the use of restricted reminding, by presenting each item only until it has been recalled just once, while the subject attempts to recall all items in the list on each recall trial. When items are never presented again after they have been recalled for the first time, the spontaneous retrieval of an item from long-term storage without any further presentation shows that it, must have entered storage when it was first recalled (if not. before). The eventual retrieval of items spontaneously on some subsequent recall attempt without further presentation provides the fairest estimate of the number of items retained in long-term storage (LTS), and also can show that recall failure was due to retrieval failure rather than to retention failure or loss from long-term storage. Figure 3 shows free recall learning of the same list of 20 animals by another group of ten children (mean age = 9.4 years, SD = 0.97), using such restricted reminding. Longterm storage is shown as the number of items in LTS before each recall attempt, determined by spontaneous retrieval eventually without any further presentation. Retrieval from LTS is shown as the number of items recalled without presentation on that trial. Figure 3 indicates that re,rRECALLED
ATLEASTONCE 11 STORAGE
ii RETRIEVAL
D-CONSISTENTRECALL D." WIIHOUTPRESENTATION
2
4
I
I
6
8 IO I2 TRIALS
I
I
/
/
,
I4
I6
I6
FIG. 3. Children’s free recall learning by restricted reminding. without any further presentation after each item was recalled just once. Storage is shown by eventual retrieval of items without any further presentation at all after initial recall.
SELECTIVE
REMINDING
2
IN
I 4
I 6
I 8 IO TRIALS
VERBAL
495
LEARNING
I , , 12 14 16 18
FIG. 4. Comparison of long-term storage (LTS) storage (LTR) by children and adults, using restricted presentation after initial recall of each item.
and retrieval from long-term reminding without any further
trieval from LTS appears to he relatively incomplete and inconsistent, while storage appears quite high despite the minimal number of presentations provided by such restricted reminding. The comparison of storage (LTS) and retrieval (LTR) by these children and by ten adults (mean age = 22.9 years, SD = 3.14) in Figure 4 shows that, although retrieval by the children was much less effective than retrieval by the adults, these children appear to have stored almost, as many items as the adults. The comparison of the cumulative total of all recall failures, the total lost (after initial recall), and the total lost, (after retrieval) from long-term storage by children and adults in Figure 5 shows that these children experienced almost three times as many recall failures as the adults, and that most recall failures were eventually recovered later by children as CHILDREN
;_,/I/’
ZO1816z
ld-
\
2 12-
TOTALRECALLRILURES
9 lo= 5 :6
8
42o-
2
4
6 a IO 12
TRIALS
FIG. 5. Comparison of the cumulative total initial recall), and the total lost (after retrieval) and by adults.
recall from
failures, long-term
the total storage
lost (after by children
496
HERMAN
BUSCHKE
well as adults. Thus it appears that the lower recall of children in free recall verbal learning may be due to retrieval failure rather than to storage or retention failure. Although the components of learning shown by selective reminding have been interpreted here in terms of storage and retrieval from shortterm and long-term retention processes, the results of selective reminding also lend themselves to other reasonable interpretation. For instance, the empirical points in learning to recall items consistently without presentation shown by selective reminding (e.g., first recall of an item, first recall without presentation, consistent recall without any further presentation at, all on all subsequent trials) may be regarded as indications of different degrees or levels of processing (Craik Rr.Lockhart, 1972). These empirical distinctions in learning by selective reminding seem reasonable, regardless of their particular interpretation. Because these paradigms allow recall without presentation to show that learning has occurred, selective reminding and restricted reminding seem to show some relevant components of learning which are not apparent when all items are always presented before each recall attempt. throughout learning. REFERENCES J. M., & BUTTERFIELD. E. C. The relations of short-term memory to development and intelligence. In I,. L. Lipsitt and H. W. Reese (Eds.), Advances in child development nr~l behnwior. New York: Academic Press, 1969. Vol. 4. Pp. 29-82. BUHCHKE. H. Selective reminding for analysis of memory and learning. Journal of Verbal Learning and Verbal Behavior. 1973. 12, 543-550. BUSCHKE, H. Spontaneous remembering after recall faihne. Science, 1974, 184, 579581. COLE, M., FRASKEL, F., & SHARP, D. Development of free recall learning in children. Developmental Psychology. 1971, 4, 109-123. CRAIK. F. I. M. Two components in free recall. Journal of Verbal Learning rrnfd Verbal Behavior, 1968, 7, 996-1004. CRAIK, F. I. M.. & I,OCKHART, R. S. Levels of processing: 9 framework for memory research. Journal of Verbal Learning and Verbal Behavior, 1972, 11, 671-4. FLAVELL, J. H. What is memory development the development of? Human Development, 1971, 14, 2262%. GLANZER, M.. & CUNITZ, A. R. Two storage mechanisms in free recall. Joz~~nl of Verbal Learning and Verbal Behavior, 1966. 5, 351-360. ROEDIGER, H. 1,. Inhibiting effects of recall. Memory & Cog&ion, 1974, 2, 261-269. TTLVING, E., & COLOTLA. V. A. Free recall of tri-lingual lists. Cognitive Psychology. 1970, 1, 86-98. TVLVING. E., & HASTIE, R. Inhibition effects of intralist repetition in free recall. Journal of Experimental Psychology, 1972. 92, 297-304. BELMONT,