The blocked-random effect in recall and recognition

The blocked-random effect in recall and recognition

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 8, 8 1 5 - 8 2 0 (1969) The Blocked-Random Effect in Recall and Recognition 1 PAUL R. D'AGOSTINO 2 Un...

508KB Sizes 4 Downloads 44 Views

JOURNAL OF VERBAL LEARNING AND VERBAL BEHAVIOR 8, 8 1 5 - 8 2 0

(1969)

The Blocked-Random Effect in Recall and Recognition 1 PAUL R. D'AGOSTINO 2 University of Virginia, Charlottesville, Virginia 22904 Two experiments are reported concerning the nature of item storage in free recall, In Experiment I, Ss received blocked or random presentation of a categorized list with or without prior knowle~tgeof list strficture. Knowledge of list structure did not reduce the blockedrandom effect. In Experiment II a recognition test was given following a blocked-random manipulation. Again, a significant blocked-random effect was found. These results were viewed as inconsistent with an independent storage hypothesis. An explanation of the blocked-random effect in terms of differences in effectivepresentation time was offered.

It has been repeatedly demonstrated in single-trial free-recall experiments that blocked presentation of a list of categorized words results in greater word recall than does random presentation (e.g., Cofer, Bruce, & Reicher, 1966; Dallett, 1964). This finding is compatible with two quite different conceptions of the nature of item storage in free recall. According to the unitization hypothesis (Mandler, 1967; Tulving, 1968), item traces are stored dependently on the basis of relationships detected among list members. The process of coding or "chunking" items at input is said to increase the accessibility of list members, since they can be retrieved on their own merits or as a result of the retrieval of other members of a unit (Tulving & Pearlstone, 1966). Presentation method is important in determing the amount of organization that occurs during list presentation. With blocked presentation, prior members of a category are readily available in recent memory. This facilitates the detection of item relationships

as well as the subsequent coding of items in terms of such relationships. Under the random condition, not only is detection more difficult but prior items are less available to be organized. Slamecka (1968) has recently questioned the assumption of dependent storage. He suggests that at input Ss form a general representation of list structure while at the same time storing the individual list members independently of one another. When recall is requested, the representation of list structure provides a basis for a retrieval plan which then guides the search for the independently-stored items. The efficiency of the retrieval plan is a function of the adequacy with which list structure is detected. Since t h e detection of list structure is more complete with blocked presentation than with random presentation, a more efficient retrieval plan is formulated. It is important to note that for Slamecka the blocked-random effect results from inadequate detection of list structure u n d e r the random condition. Therefore, if information ' This research is based on part of a dissertation concerning list structure were provided prior submitted in partial fulfillment of the requirements for the Ph.D. degree at the University of Virginia. This to a blocked-random manipulation, comparresearch was supported by National Science Founda- able r~call performance should be obtained tion research grant GB-6756 to L. Starling Reid. The under the two presentation methods. Howauthor is indebted to Professors Starling Reid and ever, according to the unitization hypothesis, Eugene Lbvelace for their advice and assistance. 2 Now at Gettysburg College, Gettysburg, Penn- prior knowledge of list structure should not sylvania 17325. reduce the blocked-random effect, since such 27 815

816

D'AGOSTINO

i n f o r m a t i o n does n o t alter the conditions of differential availability of prior items u n d e r the blocked a n d r a n d o m conditions. E x p e r i m e n t I was designed to test the above predictions. G r o u p s of Ss received blocked or r a n d o m p r e s e n t a t i o n of a categorized list u n d e r instructed a n d n o n i n s t r u c t e d conditions. In Experiment II, a recognition procedure was employed to further test t h e independent-storage hypothesis.

it would simplify their task if they learned and remembered the names of these categories. Following a t-min. study period, the Ss were required to write down the names of the categories. The category sheets were collected and answer booklets distributed prior to reading a typical set of free-recall instructions. The Ss in the noninstructed condition received only the second set of instructions. Following the instructions the words were presented at a 2.2-sec. rate with each word remaining on the screen for 1.5 sec. Prior to the first word three blank slides were presented to allow the Ss to adjust to the presentation rate. A 2-min. recall period followed the presentation of the last word in the list.

EXPERIMENT I

Method Subjects. The 120 Ss in the experiment were male undergraduates at the University of Virginia. The Ss were either paid for their services or received credit for a course requirement. All Ss were naive with respect to free-recall experiments. The Ss were tested individuallyor in groups of two to eight. Apparatus. All words were photographed on 35-ram. film and were mounted on 2 x 2-in. slides. The words were projected onto a large screen by a Kodak Carousel automatic slide projector. A Lafayette timer was used to control presentation rate. Stimulus materials. Five instances from each of six conceptual categories were selected from the Cohen, Bousfield, and Whitmarsh (1957) norms. Category members were chosen from among the third to the tenth most frequent male response to the category name. An effort was made to select items that had unambiguous category membership. Three different list orders were constructed for both blocked and random input conditions. For the blocked condition each order was determined by a random ordering of both categories and words within categories. For each list order under the random condition, the sequence in which the words appeared was random with the restriction that each successive block of six items contain no more than one item from a given category. Also, two items from the same category could not occupy adjacent input positions. Design andprocedure, Presentation method (blocked vs. random) was combined with type of instruction (instructed vs. noninstructed) and list order in a 2 x 2 x 3 factorial design. Groups of 10 Ss were assigned to each condition according to a predetermined random order. Under the instructed condition, Ss were given a sheet containing the names of the six categories that were to be represented on the list. A set of instructions was read which in essence told the Ss that the test words were members of the six categories and that

Results and Discussion Recall performance was measured in terms of the n u m b e r of correct responses (CR), the n u m b e r of categories recalled (NC), a n d the n u m b e r o f items recalled per category (IPC). The N C measure is defined in terms of the n u m b e r of categories from which at least one word is recalled, a n d the I P C measure is obtained by dividing the n u m b e r of correct responses by the n u m b e r of categories represented in recall. It should be n o t e d that the C R measure is the p r o d u c t of N C a n d I P C recall. The data of each S were scored for all three response measures, a n d separate analysis of variance were performed for each dependent measure. The m e a n values of t h e t h r e e scores for the various experimental conditions are presented in Table 1. The m a i n effect of p r e s e n t a t i o n m e t h o d was significant for b o t h the C R measure, F ( I , 108) = 9.98, p < .01, a n d I P C recall, F(1,108) = 15.30,p < .001. There was no tendency toward a reduction in the b l o c k e d - r a n d o m effect u n d e r the instructed condition, F < 1. The only other significant source of variance was the effect of instruction o n N C recall, F(1,108) = 10.20, p < .01. Since i n f o r m a t i o n c o n c e r n i n g list structure did n o t lead to more c o m p a r a b l e performance u n d e r the blocked a n d r a n d o m conditions, these results lend little s u p p o r t to the independent-storage hypothesis. It appears u n likely, therefore, that superior recall u n d e r the blocked c o n d i t i o n can be a t t r i b u t e d to

PRESENTATION METHOD AND ITEM STORAGE

817

TABLE 1 MEAN CORRECT RESPONSES(CR), MEAN NUMBEROF CATEGORIES(NC), AND MEAN ITEMSPER CATEGORY (IPC) FOR THE EXPERIMENTALCONDITIONS Blocked

Random

Dependent measure

Instructed

Noninstructed

Instructed

Noninstructed

CR NC IPC

19.23 5.80 3.32

17.47 5.30 3.31

16.73 5.73 2.92

16.13 5.57 2.92

more adequate detection of list structure as Slamecka suggests. The general pattern of results is consistent with the unitization hypothesis. Instructions failed to reduce the blocked-random effect because' such a manipulation does not alter the conditions of differential availability of prior items under the two presentation methods. Furthermore, since related words are more easily organized with blocked presentation than with random presentation, it follows that I P C recall under the blocked condition should be greater than under the random condition. Instructions increased the number of categories recalled under both blocked and random conditions. It has been demonstrated that category recall is increased when retrieval cues in the form of category names are provided at recall (Dong & Kintsch, 1968; Tulving & Pearlstone, 1966). In a similar manner, Ss under the instructed condition learned the names of the categories, and quite likely used this information as a source for retrieval cues. The fact that instructions increased N C recall but had no effect on IPC recall supports the Tulving and Pearlstone (1966) conclusion that N C and IPC measures reflect independent recall processes. The N C measure represents the accessibility of higherorder memory units, and the IPC measure reflects the accessibility of the particular components of these higher-order units. Thus, the results of this experiment are consistent with the notion that Ss organize related words at input into higher-order units and that such 27*

organization makes more accessible the members of these units. It can be argued that learning the names of the categories represented in the list does not provide sufficient information about the list structure to formulate an adequate retrieval plan. If the effectiveness of such a plan is dependent on a finer analysis of item relationships, then instructions may not have entirely eliminated the detection problem under the random condition. Experiment II, therefore, was designed to further test the independentstorage hypothesis. In this study the detection problem was reduced by presenting the category name with each instance. In addition, a recognition task was used to measure performance under both blocked and random conditions. If the difference between blocked and random presentation is due to more adequate detection of list structure and hence a more efficient retrieval plan under the blocked condition, then the effects of presentation method should be eliminated by use of a recognition task where retrieval of responses is not required. EXPERIMENT I[

Method Subjects. The Ss were 128 undergraduates at the

University of Virginia and were either paid for their services or received credit for a course requirement. None of the Ss had served in previous free-recall experiments. The Ss were tested individually or in groups of 2 to 12. Apparatus and stimulus materials. The apparatus was the same as that used in the previous experiment. The words, with the category name underlined above

818

D'AGOSTINO

each instance, were photographed and mounted on slides. The stimulus materials were 12 instances from each of eight conceptual categories drawn from the Cohen et al. (1957) norms. The category instances were highmedium frequency responses to the category name and ranged from the second to the seventeenth most frequent male response. An effort was made to select items that had unambiguous category membership. The 12 members of each category were ranked according to their frequency and divided into two separate lists. List 1 consisted of the first, fourth, fifth, eighth, ninth, and twelfth ranked members of each category, while List 2 contained the remaining members of the categories. Thus each list was composed of eight categories with six members in each category. Under the blocked condition, a single list order was determined for each list by a random ordering of both categories and words within categories. For the random condition, a single random order for each list was determined with the same restrictions as in the previous study. The recognition test sheet consisted of all 96 category members, rTwo random sequences of the test items were constructed with the restriction that no more than two instances from the same category could appear in succession. For the recall task, the eight category names appeared in two rows of four each on the answer sheet with spaceprovided to write under each category name. The order in which the category names appeared on the answer sheet was determined in the following manner. First, the ordinal input positions one through eight were randomly ordered. The particular categories corresponding to those ordinal input positions for Lists 1 and 2 under the blocked condition were then identified. This determined the first output order for each list. Although the position in which category names appeared on the answer sheet relative to their input order was the same for the two lists, the specific category names differed, since the two lists had different input orders. A second output order for each list was obtained by simply using the answer sheet for the other list. Unlike the first output order, therefore, the position in which the category names appeared on the answer sheet relative to their input order was different for the two lists. It will later be seen that this fact lead to an interesting finding. Design. The experiment compared blocked and random presentation methods under recall and recognition conditions. Groups of 32 Ss were assigned to each treatment combination; e.g., 32 Ss received blocked presentation followed by a recognition task. Half of the Ss under each condition received List 1, and the remaining half received List 2. These subgroups were further divided with half o f t h e Ss

receiving Order 1 and h a l f receiving Order 2. This was true for both the recall and recognition tasks. Subjects were assigned to conditions according to a predetermined random order. Procedure.Both recall and recognition answer sheets were placed face down in front of the Ss, and a set of free-recall instructions was given. All Ss were told that it was not necessary to remember the category names, since this information was present on the answer sheet. Following the instructions the words .were presented at a 2.2-sec. rate with each word remaining on the screen for 1.5 sec. Prior to the first word, three blank slides were presented to allow the Ss to adjust to the presentation rate. Written recall began immediately following the presentation of the last Word in the list. Prior to the recognition test, however, Ss received an additional brief set of instructions concerning the nature of the task. All Ss were allowed 3 min. to complete their respective tasks.

Results S e p a r a t e a n a l y s e s o f v a r i a n c e w e r e perf o r m e d o n t h e r e c a l l a n d r e c o g n i t i o n data. I n b o t h analyses, lists a n d o r d e r s w e r e i n c l u d e d as v a r i a b l e s . T h e d e p e n d e n t m e a s u r e f o r the r e c o g n i t i o n analysis was a difference score (number correct minus number incorrect) c o m p u t e d f o r e a c h S. F o r t h e recall analysis, t h e d e p e n d e n t m e a s u r e was the n u m b e r o f c o r r e c t responses. F o r the r e c o g n i t i o n c o n d i t i o n , the o n l y significant s o u r c e o f v a r i a n c e was t h e m a i n effect o f p r e s e n t a t i o n m e t h o d , F(1, 56) = 5.66, p < .05. A s c a n be s e e n in T a b l e 2, r e c o g n i t i o n p e r f o r m a n c e u n d e r t h e b l o c k e d c o n d i t i o n was s u p e r i o r t o p e r f o r m a n c e u n d e r the r a n d o m condition. T h e recall analysis i n d i c a t e d t h a t recall u n d e r the b l o c k e d c o n d i t i o n (27.53) was significantly g r e a t e r t h a n u n d e r the r a n d o m c o n d i TABLE 2 MEAN CORRECT RESPONSES, MEAN INCORRECT RESPONSES, AND DIFFERENCE SCORES FOR THE RECOGNITION TASK UNDER BLOCKED AND RANDOM CONDITIONS

Presentation method

Correct responses

Incorrect responses

Difference scores

Blocked Random

36.81 32.28

• 3.56 3.09

33.25 29.19

PRESENTATION METHOD AND ITEM STORAGE

tion (22.34), F(1,56) = 24.92, p < .001. The main effect of order was also significant, F(1,56) = 25.56, p < .001. The nature of this effect is better understood by considering the significant Presentation Method x Order interaction, F(1, 56) = 6.55, p < .05. Although output order had little effect under the random condition, it can be seen from Table 3 that output order was an important determiner of recall under the blocked condition. For both Lists 1 and 2, recall performance under Order 1 was considerably poorer than performance under Order 2. TABLE 3 MEAN WORD RECALL FOR EXPERIMENT II UNDER BLOCKED AND RANDOM CONDITIONS AS A FUNCTION OF LISTS A N D O U T P U T O R D E R

Presentation method Blocked Random

Order

List 1

List 2

1 2 1 2

24.37 31.87 21.12 21.87

23.87 30.00 21.12 25.25

DISCUSSION

It is clear that regardless of whether performance is measured by a recall or recognition task, blocked presentation results in a higher level of performance than does random presentation. These findings provide no support for the position that item traces in free recall are stored independently of one another at input. The data of Experiment II are particularly damaging in that performance differences attributable to inadequate detection of list structure and inefficient retrieval plans should have been eliminated with a recognition procedure. The fact that with blocked presentation prior items are more available for organization than with random presentation suggests a possible explanation for the blocked-random effect. With random presentation, considerable time may be necessary to retrieve prior information essential for the organization process. Under the blocked condition, where this

819

information is readily available, the total presentation interval could be devoted to the actual processing and organization of information. Thus, effective presentation time may be greater under the blocked condition than under the random condition. Such an interpretation would account for the blockedrandom effect in both recall and recognition tasks. There are several findings reported in the literature that support such an argument. Cofer, Bruce, and Reicher (1966) varied actual presentation time in a free-recall task under blocked and random conditions. While performance under the blocked condition was superior to performance under the random condition at each presentation interval, it is interesting to note that recall for the random condition with a 2-sec. rate was equivalent to recall for the blocked condition at a 1-sec. rate. This is also true when random presentation at a 4-sec. rate is compared to blocked presentation at a 2-sec. rate. Thus, the blocked-random effect is eliminated when presentation time is increased under the random condition. It has also been demonstrated (Cohen, 1966) that with random presentation IPC recall increases as a function of presentation time when presentation time is a between-Ss variable. It will be remembered that in Experiment I lhe locus of the blockedrandom effect was on IPC recall. In Experiment II the magnitude of the blocked-random effect was determined by the output order under the blocked condition. In fact, with Order 1 the blocked-random difference was 3.0 items, while with Order 2 this difference increased to 7.4 items. On the basis of these data, it seems obvious that the order in which input categories are recalled is an important determiner of performance. Since category recall was essentially perfect, the effect of output order was necessarily on IPC recall. The most apparent difference between Order 1 and Order 2 was the input positions of the first three categories appearing on the

820

D'AGOSTINO

answer sheet. For Order 1 the first three categories occupied middle list input positions (i.e., input positions 4, 5, and 6). For Order 2 the first three categories appeared in input positions 5, 7, and 8 for List 1 and positions 7, 2, and 4 for List 2. Thus, in Order 2 either early or late input categories were among the first three categories recalled. There is some suggestion from Experiment I that Order 2 is similar to the typical recall order of Ss in free recall. In Experiment I, 96 ~ of the Ss under the blocked condition recalled first either the initial or terminal input categories. For 70~o of these Ss, the second category recalled occupied either the first two or last two input positions. The fact that Ss rarely begin their recall with middle input categories suggests that Order 1 may have disrupted the Ss' preferred output order. There is direct evidence recently collected in the Virginia laboratory and elsewhere (Dong, 1969) which indicates that IPC recall decreases as a function of the output position of a category. It may well be the case that the detrimental effect of output order is greater for those categories typically recalled first. Finally, it should be pointed out that the recognition data of Experiment It appear to contradict a finding recently reported by Kintsch (1968). In that study Ss were presented a 40-item list consisting of four categories with 10 instances per category. Following a blocked-random manipulation, a recognition test was given. Difference scores revealed no significant difference in recognition performance as a function of presentation method. However, for reasons of no concern here, high-frequency responses to the category name were used under the blocked condition while low-frequency responses were used in the random condition. The recognition task also differed for the two conditions. The alternatives for blocked recognition were other high-frequency responses to the category name, while low-frequency responses were alternatives in random recognition. Since it is apparently more difficult to discriminate high-

frequency responses in the context of other high-frequency responses than to discriminate low-frequency responses in the context of other low-frequency responses (Schulman, 1967), such a procedure may tend to mask out the effect of presentation method. In the present study, identical test and alternative responses were used for the blocked and random conditions. REFERENCES CORER, C. N., BRUCE, D. R., & REICHER, G. M. Clustering in free recall as a function of certain methodological variations. Journal of ExperimentalPsychology, 1966, 71,858-866. COHEN,B. H. Some-or-none characteristics of coding behavior. Journal of Verbal Learning and Verbal Behavior, 1966, 5, 182-187. COHEN,B. H., BOUSFIELD,W. A., & WHITMARSH,G. A. Cultural norms for verbal items in 43 categories. Tech. Rep. No. 22, 1957, Univer. of Connecticut, Contract Nonr-631 (00), Office of Naval Research. DALLETT,K. M. Number of categories and category information in free recall. Journal of Experimental Psychology, 1964, 68, 1-12. DONG, T. Cued partial recall. Paper read at the meeting of the Eastern Psychological Association, Philadelphia, April 1969. DONO, T., & KXNTSCH,W. Subjective retrieval cues in free recall Journal of VerbalLearning and Verbal Behavior, 1968, 7, 813-816. KINTSCH,W. Recognition and free recall of organized lists. Journal of Experimental Psychology, 1968, 78, 481-487. MANDLER, G. Organization and memory. In K. W. Spence & J. T. Spence (Eds.), The psychology of learning and motivation. New York: Academic Press, 1967. Pp. 327-372. SCHULMAN,A. I. Word length and rarity in recognition memory. Psychonomic Science, 1967, 9, 211-212. SLAMECKA,N. J. Trace storage in free recall. ~urnal of Experimental Psychology, 1968, 76, 504-513. TULVIN~,E. Theoretical issues in free recall. In T. R. Dixon & D. L. Horton (Eds.), Verbal behavior and general behavior theory. Englewood Cliffs, N.J. : Prentice-Hall, 1968. Pp. 2-36. TULVING, E., & PEARLSTONE,Z. Availability versus accessibility of information in memory for words. Journal of Verbal Learning and Verbal Behavior, 1966, 5, 381-391. (Received July 18, 1969)