You are on page 1of 23

M EMOR Y, 1996, 4 (6), 633653

Directed Forgetting: Further Com parisons of the Item and List Methods
Barbara H. Basden and David R. Basden
California State University, USA

Three experiments contrasted the effects of directed forgetting instructions when given by item and given by list. In Experiment 1, which involved free recall of a single-category list, directed forgetting was greater with the item method than with the list method when the study format was pictures or words whose referents were imaged (these encouraged item-specific processing), but not when the study format was words not imaged (this encouraged relational processing). In Experiment 2, directed forgetting was observed with the item method but not with the list method in an indirect test of general knowledge. In Experiment 3, ``Recollect judgements showed directed forgetting with the item method but not with the list method. ``Know judgements did not show directed forgetting with either method. These experiments show that the mechanisms underlying directed forgetting differ for the item and list methods.

IN TR O D U C T IO N
In studies of directed forgetting, subjects are typically instructed to forget either the first of two lists (the list method) or individual targets within the list (the item method). W hen the list method is used (e.g. Epstein, 1970), the instruction to forget the first list is interpolated between the two successively presented lists. W hen the item method is used (e.g. W oodw ard, Bjork, & Jongew ard, 1973), the instruction to remember or forget each indiv idual target is given after each item is presented. Bjork (1970, 1972) proposed two m echanism s in directed forgetting: differential rehearsal of Remember and Forget items and segregation of Remem ber and Forget items. He argue d that different mechanisms might unde rlie these m ethods, with differential rehearsal occurring with the item method and not with the list method. Subsequently, Bjork (1989) propos ed
Requests for reprints should be sent to Barbara H. Basden, Department of Psychology, California State University, Fresno, CA 93740-0011, USA. Email: barbb@zimmer.csufresno.edu. Portions of this research were presented at the 34th annual meeting of the Psychonomic Society and at the 23rd annual Conference for Undergraduate Psychology Research at Santa Clara University. W e would like to thank Kim Maxwell, Scott Pennington, Diane Pugliese, Richard Torzynski, and John Williams for assistance in data collection.

1996 Psychology Press, an imprint of Erlbaum (UK) Taylor & Francis Ltd

634

B A S D E N AN D B A S D E N

retrieval inhibition as an additional mechanism, but did not tie it to a specific m ethod. W e have argue d (Basden et al., 1994; Basden, Basden, & Gargano, 1993) that different m echanism s are responsible for directed forgetting with the two m ethods, but we have proposed that retrieval inhibi tion unde rlies directed forgetting with the list method, and that differential processing of Rem ember and Forget targets underlies directed forgetting with the item method. In addition, we have assumed that relationa l (organisation) processing occurs to a greater extent with the list method than with the item method, and that itemspecific processing occurs to a greater extent with the item method than with the list method. In a com prehensive review of the directed forgetting literature, Johnson (1994) drew a similar conclusion. She attributed greater recall of Remember than of Forget targets with the list m ethod to segregation and selective search of each list of targets followed by inhibition of the to-beforgotten list. W ith the item method, segregation and retrieval inhibition are not required because only the set of Remember targets is stored and searched. Three lines of evidence support the po sition that retrieval inhibi tion occurs with the list method and not with the item method. First, the item m ethod typically yields greater directed forgetting than the list method on recall tests, a result that should be expec ted if Forget targets are m ore extensively processed with the list method than with the item method. Second, the item method yields directed forgetting on recognition tests, but the list method does not. If items are less well processed at storage (as we believe is the case with the item method) , recognition m emory should be affected. If a set of equally well-processed items is inhibited from retrieval (as we believe in the case of the list method), a test that de-emph asises retrieval should be unaffected. Third, the re-presentation of targets on a recognition test seems to effect a ``release from inhibition on a subsequent recall test when the list method is used, but not when the item m ethod is used 1 (e.g. Basden et al., 1993; Bjork, 1989). Re-presentation of both Remember and Forget item s would not be expected to eliminate a processing deficit for the Forget items, but might very well overcom e retrieval inhibition. According to this distinction between relational and item-specific processing (Einstein & Hunt, 1980; Hunt & Einstein, 1981), procedure s that focus the subject s attention on the list as a whole (or at least on large sections of it) encourage the ide ntification of relationships am ong the item s, whereas procedures that focus the subject s attention on individual item s emphasise inform ation that is item -specific. Thus, the list method of directed forgetting w ould encourage relational processing whereas the item m ethod wou ld encourage item-specific processing. This is not to say, of course, that either m ethod limits the subject to only one type of processing. W e are m erely
Although Geiselman and Bagheri (1985) reported a release from inhibition with the item method, additional study of forget-cued targets was required to accomplish that release. Repetition may enhance processing rather than releasing items from retrieval inhibition.
1

DIR E C T E D F O R G E T T IN G

635

suggesting that the mix of relationa l and item-specific information is influenced by the method of directed forgetting. Relational processing should facilitate listmethod directed forgetting because in this case suppression of an entire, alreadyprocessed list must occur. Item-specific processing should facilitate itemmethod directed forgetting because in this case indivi dual targets m ust be forgotten at the time they are unde rgoing proce ssing. If the item method encourages item-specific processing and the list method encourages relational processing, and if directed forgetting results from a processing deficit with the item method and from retrieval inhibi tion with the list method, then factors that enhance item-spe cific processing should increase the m agnitude of directed forgetting with the item method, and factors that enhance relationa l processing should increase directed forgetting with the list method. In Expe riment 1, we manipulated the study format as a m eans of encouraging either relationa l or item -specific processing. Our expectation was that the relative magnitude of directed forgetting with the two metho ds would vary with study form at.

EX P ER IM E N T 1
In this experiment one group of subjects attempted to mem orise a set of animal pictures and two other groups of subjects attempted to mem orise a list of names for those pictures. The subjects in one of the two latter groups were given simple learning instructions and those in the other group were told to form a m ental image of the referent of each target. A few prior studies have employe d im ages or pictures in the context of directed forgetting (Bray, Justice, & Zahm , 1983; Bugelski, 1970). Unfortunately, no firm conclusions could be drawn from these previous studies. According to Hunt and his colleagues, item-specific processing is encouraged either by forming images of the targets (Hunt & M arschark, 1989) or by studying pictures corresponding to the targets (Hunt & M cDaniel, 1993). Hunt and M cDaniel have also argued that simple presentation of a categorised list for learning encourage s relationa l processing. As mentioned earlier, item-m ethod directed forgetting is usually greater than list-method directed forgetting. According to our analysis, this difference in the magnitude of directed forgetting should be greater when subjects study pictures or study words with im agery instructions that when they study words alone, particula rly when the words belong to a single category. One interpretation of directed forgetting is that contextual information about Forget targets is inaccessible on recall tests. Geiselm an, Bjork, and Fishman (1983) used the list method, asking subjects to write their recall in separate colum ns corresponding to the two lists they had studied. Contr ol subjects remem bered list membership of Forget targets better than did directed forgetting subjects. Geiselm an et al. suggested that loss of contextual information underlies

636

B A S D E N AN D B A S D E N

retrieval inhibition. Unfortunately, a potential source of confo undin g is present if subjects base their judgem ents of list membe rship on item strength. If this were the case, judge ments of list membership might not reflect context inform ation alone. A more direct measure of memory for context inform ation is to have subjects indicate the approxim ate serial position of previously recalled targets. Tzeng, Lee, and W etzel (1979) introduced this technique. W ith the item m ethod they found that serial position inform ation was remembered better for Remember than for Forget targets. No control subjects were tested, so no absolute assessments were possible. The results of Tzeng et al. are equally consistent with the idea that Forget item s are less extensively processed with the item method. If Remem be r and Forget targets are equally well processed with the list method, then accuracy of serial position inform ation should be equivalent for the Forget and Rem ember item s with this method. How ever, if Geiselm an et al. are correct and context m emory is reduced by directed forgetting instructions, then memory for serial position of Forget targets should be affected with both the item and the list m ethods. In this experim ent we asked our subjects to indicate approximate serial positions for the item s they recalled. To determ ine the influence of directed forgetting instructions on absolute memory for contextual information, we included control subjects who did not receive directed forgetting instructions. If contextual information is lost during directed forgetting, then memory for serial position should be less accurate for Forget targets in the experim ental conditions than in the control conditions. Although control subjects were included to permit measurement of absolute loss of contextual inform ation, their inclusion also permitted us to com pare the absolute effects of directed forgetting instructions on Remember and Forget targets for each of the two methods. Past research has show n that directed forgetting instructions yield not only lower recall of Forget targets, but also greater recall of Rem ember targets (see Bjork, 1972).

M eth od
Subjects. W e tested 19 4 students enrolled in introdu ctory psychology classes at California State University, Fresno. Their participation was in partial fulfilment of course requirements. Approximately 22 subjects were tested in each of the nine treatment conditions. M aterials and Apparatu s. W e selected 24 pictures of animals from the Snodgra ss and Vanderwart (1980) norm s. All pictures show ed high levels of name agreement (M = 89% ) according to those norm s. A set of 35m m slides was m ade for the pictures and another set was made for the names. The slides were presented using a Kodak slide projector paced by a Hunter timer. Each target

DIR E C T E D F O R G E T T IN G

637

was designated Forget and Remember equally often across groups. Order was varied across groups to ensure that particula r targets did not consistently occupy favourable serial positions across m ethods. Subje cts responde d by writing their recall on lined pages included in their response booklets. Design. A 3 2 2 mixed factorial design was employe d with study form at (pictures, words with imagery instructions, or words without im agery instructions) and m ethod (list or item ) as between-subjects factors and instruction (Rem em ber or Forget) as a within-subject factor. Three outside control groups were also tested, one with each of the three study formats. Subje cts in these groups were not given directed forgetting instructions. Procedure. Groups of up to six subjects at a time were assigned to testing conditions in accordance with a block-random isation schedule. Subje cts were given instructions appropriate to their study format prior to target presentation. For the im age study format, subjects were instructed to form an im age of each word as it was presented to them. For the picture and word study formats, subjects were simply instructed to do their best to recall all targets. A blank screen or a spot of light of 1.1 second duration followed each target for all subjects. Subjects tested with the item method were given the following additional instructions:
After each item has been on the screen for a few seconds, it will be followed by a blank screen or a spot of light. A blank screen means you are to forget that item but a spot of light means you are to remember that item.

Subje cts tested by the list method were not told to differentiate among items and were told to ignore the presence or absence of lights. After subjects had read the initial instructions, they were show n the words or pictures indivi dually at a three-second rate. After the first 12 targets had been presented, subjects tested by the item method were told to rest for a few seconds. Subje cts tested by the list method were given directed forgetting instructions for the first list. Th us, the time interval between list halves was equiva lent for the two m ethods. Then the remaining 12 targets were presented. Target presentation was immedia tely followed by a distractor task intended to purge short-term mem ory. Subjects com pleted simple arithmetic problems in their response booklets for 30 seconds; then they were instructed to write as many of the targets they had studied as they could remember, including both the targets they were told to remem ber and those they were told to forget. They were given three minutes for this free recall test. Subje cts were then instructed to indicate the approxim ate serial position of each word they had recalled. This was done by asking them to write a ``1 next to a word if it occurred am ong the first four targets, a ``2 if it occurred among the second four targets, and so on

638

B A S D E N AN D B A S D E N

through to ``6 if it occurred among the last four targets (cf Tzeng et al., 1979). After the experim ent was over, the subjects were told the purpose of the forget instruction and were asked not to discuss the experiment with other potential subjects.

R esu lts
The depend ent variable in free recall was the proportion of Remember or Forget targets recalled. For control subjects, targets that occupied the same serial positions as actual Remem ber and Forget targets in the experimental conditions were designated nom inal Rem ember and Forget targets. Thus, determination of nom inal Remember and Forget targets for individual control subjects depended on whether the comparison was with list method or item method experim ental subjects. M ean recall proportions for each of the conditions are show n in Table 1. A mixed three-factor analysis of variance was perform ed with method and study form at as between-subjects factors and instruction as a repeated measure. The three-factor interaction was significant, F(2, 127) = 3.53, M Se = 0.036. Subsequent analysis revealed that the simple interaction between method and instruction was significant for pictures, F(1, 42) = 4.21, MSe = 0.032; and for words with images, F(1, 45) = 14.89, M Se = 0.027, but was not significant for words without im ages, F < 1. Thus, the m agnitude of directed forgetting differed as a function of method when pictures or words with images were studied, but not when words were studied. These results, with directed forgetting expressed as difference scores between the m ean proportion of Remember and Forget targets recalled, are show n in Fig. 1. W ith the item m ethod , the magnitude of directed forgetting was statistically sign ificant for all three study form ats: for pictures, F(1, 21) = 58.33 ,
TA BLE 1 M e an P ro p o rtio ns o f R em e m b er, F o rg et a n d C o n trol T arg ets P ro d u c ed d u rin g Free R ec all

Experimental Study Condition Item Method Pictures Images W ords List M ethod Pictures Images W ords Remember Forget Remember

Control Forget

0.783 0.848 0.721 0.754 0.757 0.716

0.364 0.417 0.462 0.545 0.587 0.417

0.596 0.745 0.644 0.562 0.741 0.625

0.621 0.732 0.587 0.654 0.737 0.606

DIR E C T E D F O R G E T T IN G

639

FIG . 1.

M agnitude of directed forgetting as a function of study format and instruction.

M Se = 0.037, for words with im ages, F(1, 22) = 95.86, M Se = 0.022, and for words without images, F(1, 19) = 15.62, M Se = 0.043. W ith the list m ethod, the magnitude of directed forgetting was also statistically significant for all three study formats: for pictures, F(1, 21) = 12.47, M Se = 0.038, for words with images, F(1, 23) = 11.12, MSe = 0.031, and for words without images, F(1, 21) = 20.45, MSe = 0.048. As mentioned earlier, we compared the list method and item method directed forgetting subjects with the same control subjects. This necessitated separate three-factor analyses of variance for each method, with study format and group (experimental vs. control) as between-subjects factors and instruction as a repeated m easures factor. For the item method analysis, the effect of study form at was significant, F(2, 120) = 7.77, MSe = 0.035. Images were recalled better than pictures or words. The interaction between instruction and group was significant, F(1, 120) = 65.21. Tests of simple effects confirm ed that recall of Remem ber targets was greater for experimental than for control subjects, F(1, 120) = 13 .04, M Se = 0.158 ; but recall of Fo rget targets w as poo rer for experim ental than for control subjects, F(1, 120) = 46.06, MSe = 0.192. 2 For
The triple interaction narrowly missed significance (P = 0.07). The magnitude of the difference between Forget and control items was greater with the picture and image formats than with the word format. These results are consistent with our argument that subjects perform only superficial processing of Forget targets with the item method.
2

640

B A S D E N AN D B A S D E N

the list method analysis, the interaction between group type and instruction was again significant, F(1, 123) = 27.87. As in the analysis of the item method, recall of Remem be r targets was greater for experimental than for control subjects, F(1, 123) = 8.03 , M Se = 0.201; bu t recall of Forget targe ts w as poo rer for experimental than for control subjects, F(1, 123) = 17.75, M Se = 0.017. An analysis conduc ted on the proportion of correct serial position placements for the item method yielded a significant main effect for instruction, F(1, 60) = 5.64, M Se = 0.054. The proportion of correct placements was 0.50 for Remember targets and 0.40 for Forget targets. A similar analysis for the list m ethod yielded no significant effects, the mean propor tion of correct placements was 0.42 for Remember targets and 0.48 for Forget targets. An analysis was conduc ted on the proportion of correct serial position placem ents of Forget items contra sting directed forgetting subjects with controls. Subje cts placed Forget targets equally accurately in the experim ental and control conditions, Fs < 1. This held true for both the item and list methods .

D isc u ssion
W e have hypoth esised that study form ats encouraging item-specific processing would increase directed forgetting m ore with the item m ethod than with the list m ethod. This hypothe sis was confirmed. The difference in magnitude of directed forgetting with the item and list methods was greater when subjects studied pictures or words for which they formed images than when they simply studied words. Directed forgetting was actually equivalent for the two methods when subjects studied words without forming images. These results are consistent with our distinction between item and list methods , i.e. that relational processing underlies directed forgetting with the list method, but that item -specific processing underlies directed forgetting with the item m ethod. The present results are not consistent with those of Bugelski (1970), who reported an absence of directed forgetting with the item method when subjects were asked to form images of the targets they studied. His report is quite sketchy with regard to procedures, but we assume that they differed from our ow n in som e important way. Our results are consistent with those of Bray and his colleagues, e.g. Bray et al. (1983). These investigators reported robus t directed forgetting when subjects studied pictures. Bugelski had argued, on the basis of his failure to find directed forgetting, that subjects could not inhibi t targets for which they had formed images. On the basis of our ow n results we would argue that subjects did not carry out extensive processing of Forget targets even when they were presented in the picture and image study formats. This interpretation is supported by our finding that control subjects recalled those same targets better than did the experim ental subjects. Based on the directed forgetting literature, we had expected the directed forgetting instruction to both reduce recall of Forget items and increase recall of

DIR E C T E D F O R G E T T IN G

641

Remem ber items. In the present study this occurred with both the list and item methods. Recall was poore r for Forget targets and better for Rem embe r targets as com pared with controls. Bjork (1970) has argued that Rem ember targets are subject to less interference in the experim ental conditions. This argum ent could be applied to either method of directed forgetting. If differential processing unde rlies item-method forgetting, then serial position information should be better for Remember than for Forget item s with this method. If differential processing is not responsible for list-method directed forgetting, then serial position information should be the same for Remem ber and Forget items with this method. As predicted, m emory for serial position information was better for Remember than for Forget items with the item method, but not with the list method. These results support our position. To test the hypoth esis expressed by Geiselman et al. (1983) that directed forgetting occurs when subjects lose access to contextual inform ation, we com pared accuracy of serial position information for Forget targets in the experim ental and control condi tions. Although Geiselman et al. reported that memory for list membe rship of Forget items was greater for control than for experim ental subjects, we found no corresponding difference in the accuracy of memory for serial position. Furthermore, our results were quite similar with the list and item methods. Thus, our results do not support the hypoth esis that directed forgetting results from loss of contextual inform ation. Of course, a quite different picture might emerge if recollection for serial position of all items had been assessed. Thus, results of Expe riment 1 were consistent with the hypothe sis that retrieval inhibition and relationa l processing under lie directed forgetting with the list m ethod, but that differential item-specific processing unde rlies directed forgetting with the item method. In Expe riment 2, we com pared the influe nce of the list- and item-m ethod directed forgetting on perform ance in an implicit memory test. Current research and reviews (e.g. Brown & M itchell, 1994; Th apar & Greene, 1994) indicate that levels of processing manipulations m ay influence priming on implicit tests. If, as we suggest, Remembe r targets are more extensively processed than Forget targets with the item method, then itemmethod directed forgetting m ay occur on im plicit tests but list-method directed forgetting should not.

EX P ER IM E N T 2
Directed forgetting on implicit tests has sometimes been observed with the item method, but has never been observed with the list m ethod (e.g. Basden et al., 1993; M acLeod, 1989; Paller, 1990). W e suggest that this difference occurs because implicit tests typically involve item-by-item cueing and thus are m ore sensitive to the effects of differential processing than to the effects of retrieval inhibition. To our know ledge, even explicit tests that provide item-by-item cues

642

B A S D E N AN D B A S D E N

(e.g. word fragm ent cued recall, stem-cued recall, recogniti on, and word association) do not produc e list-method directed forgetting (see Basden et al., 1993). List-method directed forgetting instructions appear to act on the list as a whole rather than on individual items. In previous experiments that obtained item -m ethod directed forgetting on implicit tests (e.g. M acLeod, 1989), perceptual tests such as word fragm ent com pletion were used. Such tests are typically less sensitive to differential processing (see Roediger & M cDermott, 1993) than are conce ptual tests (tests relying on meanin g). Thus, demonstrations of item -method directed forgetting should be easier to obtain on implicit conceptual tests. In research reporte d earlier we failed to observe item-m ethod directed forgetting with an implicit word association test (Basden et al., 1993, Experiments 1 & 2). How ever, this failure may have resulted from a floor effect, as priming was not statistically significant with the conceptual test we used. The present experim ent represents anothe r attem pt to obtain item -method directed forgetting with an implicit conceptual test. The list m ethod was included for purposes of com parison. After studying a list of low-frequency words with either item- or list-method directed forgetting instructions, the subjects were given an implicit test of general know ledge. Definitions of targets and distractors were provided at the tim e of the test, and subjects were instructed to respond to each of them with the first word that cam e to m ind. Low -frequency words were used because such m aterials typically yield higher leve ls of priming than do high-frequency words (Roediger & M cDermott, 1993). As this test, like other implicit tests, includes item-by-item cueing, it should be sensitive to differential processing of Remember and Forget items of the sort we believe the item method encourages. Therefore, we expected directed forgetting with the item method but not with the list m ethod. After com pleting the implicit general know ledge test, subjects were asked to free recall the Remem be r and Forget targets. W e expected directed forgetting with the list method as well as with the item method on this test, because free recall is sensitive to the retrieval inhibition we believe occurs with the list m ethod. W e expected greater directed forgetting with the item method than with the list m ethod as in our previous research.

M eth od
Subjects. W e initially tested 68 subjects and later added 69 more subjects. W e also tested 28 similar subjects in a pilot study used to assess the materials. All were of the same description as in Experim ent 1. M aterials and Apparatus. The materials comprised 192 low-frequency, m ultisyllabic words (Tulving, Schacter, & Stark, 1982). Definitions were developed for each of these words by consulting W ebster s New Collegiate

DIR E C T E D F O R G E T T IN G

643

Dictionary. The materials were then piloted to confirm that prospective subjects could match the words with their definitions. These definitions were used to prepare a test of general know ledge. All materials were presented using a microcomputer. Th e subjects typed their responses at the computer keyboa rd. Design. W e employe d a mixed two-factor design with m ethod (item or list) as a between-subjects factor and instruction (Forget or Remember) as a withinsubject factor. The depend ent measure was the num ber of experim enterdesignated term s produc ed on the test of general know ledge and on the subsequent free recall test. Procedure. Subje cts were assigned to the method of directed forgetting in accordance with a block-random isation schedule. After entering the condition code at the com puter keyboa rd, the experimenter left the test cubicle; the remainder of the procedure was automatically administered. The program me random ly selected a study list of 48 targets and 48 distractors from the corpus of 192 words. The 48 list members were random ly separated into 24 Remem ber and 24 Forget targets. Individual random presentation and testing orders were also program -generated. Prior to list presentation all subjects were inform ed that their ability to remem ber list membe rs would be tested later. Subje cts tested with the item m ethod were also given the following instructions:
After each word has been on the screen for a few seconds, a brief signal will tell you whether to REM EM BER it or FORGET it. If the signal is REMEMBER you should be prepared to recall the word later, but if the signal is FORGET you can forget the word.

After reading these initial instructions, all subjects were show n the targets individually at a six-second rate. For subjects tested with the item method, ``REM EM BER or ``FORGET joined each item during its final three seconds of exposure. After the first 24 targets had been presented, subjects tested with the item m ethod were told to rest for a few seconds and subjects tested with the list method were given the following directed forgetting instructions:
The list you have just studied was only for practice. You can forget it now. The list you will see next is the one we want you to remember, so forget the practice list and concentrate on this new list.

Th en the remaining list of 24 targets was presented. After list presentation, all subjects were given a distractor task in which a list of 24 prod ucts, services, or activities appeared on the m onitor screen simultaneously. The subject was instructed to type the name of a state he/she

644

B A S D E N AN D B A S D E N

though t was associated with each comm od ity on the screen, e.g. ``Georgia for PEACHES. Subjects continued with this task until they had responded to all com m odities or until 30 seconds had elapsed without a response. Subje cts were told that the general know ledge test, which immedia tely followed the distractor task, was part of a different experim ent. The definitions corresponding to the 48 targets and 48 distractors were displayed individually in random order. The subjects were instructed to type the first appropriate word that cam e to m ind for each definition. The next definition appeared immedia tely after the subject s response to the previous one, or after 30 seconds in any case. After the subjects had com pleted the general know ledge test, they were given a free recall test. Subjects typed their recall on the compute r keyboa rd, continuing until 30 secon ds had elapsed without additional recall.

R esu lts
A preliminary analysis revealed no differences in results for the initial and subsequently added subjects, so the two data sets were com bined for the analysis reported here. Priming Scores. The dependent variable was the proportion of designated terms (Remem ber targets, Forget targets, and distractors) that were produc ed in respons e to their definitions on the general know ledge test. M ean proportions for each group are show n in Table 2. W e performed a two-factor mixed analysis of varianc e on the proportions of target words produc ed on the general know ledge test. M ore targets were given with the list method than with the item m ethod, F(1, 135) = 11.33, MSe = 0.03; and more targets were given to definitions of Remem be r words than to definitions of Forget words, F(1, 135) = 20.33, M Se = 0.007. The interaction was also significant, F(1, 135) = 6.82. Sim ple tests established that directed fo rge tt in g w as si g ni fi ca nt w i th t he ite m m e th od , F (1, 6 7 ) = 2 2. 80 , M Se = 0.008, but not with the list method, F(1, 68) = 2.02, M Se = 0.00 7. To determine whether priming had occurred, the proportions of Rem ember and Forget targets provide d were individ ually compared with the propor tion of
TA BLE 2 M ean P ro p o rtio ns o f R em em b er, Fo rg et, a n d D istrac tor T erm s P ro d u ce d o n the G en e ra l K n o w le d g e T es t in E xp erim en t 2

Word Type M ethod List Item Remember 0.320 0.279 Forget 0.300 0.204 Distractor 0.141 0.146

DIR E C T E D F O R G E T T IN G

645

d istracto rs provid ed. R e m em ber-target p rim i ng w as si gnifican t, F (1 , 135) = 239.40, M Se = 0.007. The interaction with method was significant, F(1, 135) = 5.20, reflecting greater priming with the list method than with the item method. Sim ple tests confirm ed significant prim ing of Remember targets with both the list method, F(1, 68) = 159.18, M Se = 0.007, and the item m ethod, F(1, 67) = 86.15, MSe = 0.007. The results of the Forget-target prim ing analysis were similar to those for the Remem ber-target prim ing analysis. Both the effect of method, F(1, 135) = 8.25, M Se = 0.006, and the effect of instruction, F (1, 135) = 137 .48, were significant. Th e significant interaction, F(1, 135) = 29.47, again reflected greater priming with the list method than with the item method. Simple tests confirm ed that Forget-target priming was significant for both the list method, F(1,68) = 149.51 , M Se = 0.006, and the item method, F(1, 67) = 19.51. Thus, both Remember and Forget words show ed significant prim ing with both methods, but priming was greater with the list method than with the item method. Free R ecall. A tw o-factor m ixe d m ode l analysis of varianc e w as performed on the proportions of Remember and Forget words produc ed during the final free recall test. M ean recall proportions for Remember and Forget words were 0.14 and 0.11 with the list m ethod, and 0.17 and 0.06 with the item method. The effect of instruction was significant, F(1,134) = 60.53, M Se = 0.005. Th e interaction was also significant, F(1, 134) = 24.32, MSe = 0.005, reflecting greater directed forgetting with the item method than with the list method. Directed forgetting was significant with both the item method, F (1,67) = 75.59, M Se = 0.006, and the list m ethod, F(1, 67) = 4.36, MSe = 0.005.

D iscu ss ion
Earlier we reported our failure to obtain item-m ethod directed forgetting on an implicit word association test (Basden et al., 1993, Experim ents 1 & 2). W e attributed that null result to the low levels of priming that were observed. This experim ent was designed to provide a m ore adequate test of directed forgetting on a conceptua l implicit test; i.e. one that show ed adequate prim ing. W e were successful in that prim ing on the general know ledge test was significant for both Remem ber and Forget words and with both methods . There are several reasons for greater prim ing in the present experim ent. First, we used more targets, thus perm itting more observations. Second, the general know ledge items were of relatively low frequenc y. Th ird, m ore redintegrative inform ation is provided by a comple te definition than by a single associate. W e had predicted that, given adequate priming, directed forgetting would occur with the item m ethod but not with the list method. This prediction was confirmed. W e expected directed forgetting to occur with the list method as well as the item method on the subsequent free recall test, and this too was

646

B A S D E N AN D B A S D E N

confirm ed. Th us, item -m ethod directed forgetting occurred on bo th the conceptual test and on the explicit free recall test, but list-method directed forgetting occurred only on the latter. These results are consistent with our hypoth eses that (a) Remember targets are more extensively processed than Forget targets with the item method but not with the list method, and (b) tests of implicit memory are sensitive to the effects of differential processing but not to those of retrieval inhibition. W hen subjects are given retrieval cues on an itemby-item basis, suppression of the entire Forget list is unlikely to occur. Golding, Lo ng and M acleod (1994) have suggested that, as directed forgetting instructions may result in differential processing of Remember and Forget words, the underlying basis for directed forgetting on im plicit tests m ay be differential levels of processing, much as we have suggested here. How ever, they did not distinguish between item- and list-method directed forgetting. If differential processing unde rlies directed forgetting with both the list and item m ethod, then the list method should yield directed forgetting on implicit tests. Not only should it occur, it should be greater than with the item method. In m ixed-list designs, investigators have found that the levels of processing effect is larger when targets are blocked at presentation as in the list m ethod rather than random ly intermingled as in the item method (Challis & Brodbeck, 1992; Thapar & Greene, 1994). Th e fact that this does not occur makes it clear that differential processing of Remember and Forget items is not respons ible for listm ethod directed forgetting. It could be argued that directed forgetting on item-m ethod im plicit tests results from contam ination of a supposedly pure test of autom atic processing by intentional retrieval. Toth, Reingold, & Jacoby (1994) recently applied the process dissociation procedure to leve ls of processing findings. They reporte d that conscious recollection accounted for the advantage of semantically processed targets relative to nons emantically processed targets on implicit tests. Previously, item-method directed forgetting has been observed when subjects were told to engage in intentional retrieval (Basden et al., 1993, Experiment 3) or when the implicit test was not well camouflaged (M acLeod, 1989). Russo and Andrade (1995) recently applied the process dissociation procedure to word fragment com pletion data that show ed a directed forgetting effect with the inclusion test procedure. Their estimate of the contribution of conscious recollection was greater for Rem ember than for Forget targets. Their results are consistent with our argument (Basden et al., 1993) that intentional retrieval m ay contribute to directed forgetting on implicit tests. However, an inclusion test encourages subjects to consciously recollect the target item s, whereas a well-disguised implicit test does not. In the present experiment, we attempted to camouflage the general know ledge test; subjects were given a conceptual distractor task prior to the implicit test, and they were told that the general know ledge test was a word game. Furthermore, on the implicit test itself, the num ber of distractors equalled the num ber of targets. These measures were

DIR E C T E D F O R G E T T IN G

647

taken to reduce the likelihood of contamina tion by intention al retrieval as sugg ested by Roediger and M cD erm ott (1993) . App lication of process dissociation procedures to prope rly conduc ted im plicit tests may be gratuitous. In Experim ent 3, we provide d a further test of our contention that itemmethod directed forgetting involves differential processing but list-m ethod directed forgetting does not. If Remem ber words are m ore extensively processed than Forget words with the item m ethod but not with the list m ethod, directed forgetting should be absent when only the familiarity of Remember and Forget item s is assessed.

EX P ER IM E N T 3
In this experiment we turned to the distinction between Know and Recollect judge ments presented by Tulving (1985) and developed extensively by Gardiner 3 (e.g. Gardiner, 1988; Gardiner & Java, 1993). Such judge m ents partition recognition perform ance into two com po nents, on e reflecting conscious recollection (Recollect judgem ents), and the other reflecting familiarity (Know judge ments). Recollect and Know judge ments are obtained by asking subjects to indicate for each recognised item whether its actual occurrence in the study list is remem bered or it merely seems fam iliar. Gardiner (1988, Experim ent 1) observed a levels of processing effect with Recollect judge ments but not with Know judgem ents. Gardiner, Gawlik, and Richardson -K lave hn (1994 ) fou nd item -m ethod directed forgetting w ith Recollect but not with Know judgem ents. In the present experim ent we included both methods of directed forgetting. If, as we have proposed, Remem ber and Forget targets are differentially processed with the item method but not the list m ethod, then directed forgetting should occur with Recollect judge ments for the item m ethod but not for the list method. Neither method was expec ted to yield directed forgetting with Know judge m ents.

M etho d .
Subje cts, Materials, and Appara tus. W e tested 40 students of the same description as in our earlier experim ents. The word list and apparatus were the same as in Experim ent 2. Design. The design was a 2 2 factorial. Method (item or list) was manipulated between subjects and instructions (Rem em ber or Forget) was manipulated within subjects. All subjects were given a recognition judgem ent test.
W e have chosen to refer to judgements as ``Recollect and ``Know rather than ``Remember and ``Know as designated by Gardiner (1988), in order to avoid confusion between subjects judgements and type of instruction, ``Remember and ``Forget .
3

648

B A S D E N AN D B A S D E N

Procedure. The study instructions and method of presentation were the same as in Expe riment 2. List presentation was immedia tely followed by a recognition judge ment test. Instructions were drawn verbatim from Rajaram (1993). To paraphrase, subjects were told to respond with ``N (Not Recognised), if they did not remem ber having studied any given word; with ``K (Know ), if the word appeared familiar, i.e. if they were aware of the word in the context of the experiment, but were unable to recollect any contextual information about that word; and with ``R (Recollect) if they consciously recollected having studied that word. Each target or distractor was presented in upper case letters, centred on the m onitor screen. Definitions of the available responses appeared below each target and distractor. Subje cts pressed ``N , ``K , or ``R , on the com puter keyboa rd for each word. Each new word was presented immediately after a respons e to the previous word, or after 30 seconds in any case.

R esu lts
The m ean proportions of Remember and Forget words judge d as Recollect and Know are show n in Table 3, along with Recollect and Know judge ments for distractors. A mixed two-factor analysis of variance was perform ed on the Recollect judgem ents for Rem em ber and Forg et targets. It produced a significant interaction between method and instruction, F(1, 38) = 11.32, M Se = 0.015. Subsequent simple tests show ed directed forgetting with the item method, F(1, 19) = 26.04, M Se = 0.014, but not with the list m ethod, F < 1. A similar analysis perform ed on the Know judgem ents failed to produc e significant effects (Fs < 1). To determine whether this interaction was produc ed by (a) better recollection of Remember items with the item m ethod than with the list m ethod, (b) poore r recollection of Forget items with the item m ethod than with the list method, or (c) some com bination of the two, we performed two additional analyses. First, we performed a mixed tw o-factor analysis of varianc e on Recollect judge ments for Rem embe r and Distractor words. The interaction between m ethod and word type was not significant, F < 1. Next, we performed this analysis for Forget and Distractor words. This time the interaction between method and word type was

TA BLE 3 M ea n P ro p o rtio n s o f K n o w a n d R ec o llec t J u d g em en ts in E xp erim en t 3

Remember M ethod List Item R = Recollect, K = Know R 0.72 0.67 K 0.71 0.48 R 0.17 0.18

Forget K 0.18 0.20

False Alarm R 0.04 0.04 K 0.08 0.07

DIR E C T E D F O R G E T T IN G

649

significant, F(1, 38) = 11.54. The difference between Forget and Distractor words was significant not only with the list m ethod, F(1, 19) = 235.39 , M Se = 0.019, but also with the item m ethod, F(1, 19) = 69.17, M Se = 0.028. The difference between Forget and Distractor words was greater with the list method than with the item m ethods.

D iscu ss ion
In earlier reports (see Basden et al., 1993; Basden et al., 1994), we dem onstrated that directed forgetting occurs in recognition with the item method but not with the list m ethod when overall perform ance is examined. This experim ent confirms this conclusion, and dem onstrates that it applies to R ecollect judge ments and not to Know judge ments. This dissociation supports our hypoth esis that the processes unde rlying directed forgetting differ for the two methods. The difference in correct Recollect judge ments between Forget and Distractor items was greater with the list m ethod than with the item method, but the correspondin g difference between Remember and Distractor items was the same with the two m ethods. Forget words appear to receive less extensive processing than Remember words with the item method, but Remember and Forget words appea r to receive equivalent processing with the list method. Our results with the item m ethod replicated those of Gardiner et al. (1994). Th ey varied the delay of the Forget or Remember cue and found that increased delay was associated with an inc rease in Know judge m ents for item s accom panied by either cue. Increased cue delay was associated with an increase in Recollect judge ments for Remem be r targets but not for Forget targets. Their interpretation, with which we concur, was that m aintenance rehearsal influences familiarity, and hence, Know judge ments. There is no correspo nding increase in Recollect judge m ents with delay, because elaborative rehearsal is necessary to increase conscious recollection. The present results are consistent with evidence from Experiment 1 that subjects retain greater contextual inform ation for Rem embe r than for Forget words with the item m ethod but not with the list method. Subje cts were m ore accurate in providing serial position information for Remem be r than for Forget targets with the item method than with the list m ethod. The ability to remember the serial position of targets reflects greater conscious recollection for their context. Similarly, Recollect judgem ents reflected memory for the specific context in which particular words occurred. This is consistent with our argum ent that the item method is more encouraging to differential processing than is the list method.

G E N ER A L D IS C U S S IO N
Th e experiments reported here yielded a num ber of dissociations between itemmethod and list-method directed forgetting. In Experiment 1, which involve d

650

B A S D E N AN D B A S D E N

presentation of items from a single category, having subjects study pictures or form images increased directed forgetting more with the item method than with the list m ethod. In Ex periment 2, directed forgetting occurred with the item m ethod but not with the list m ethod when responses to general know ledge questions were analysed. In Experiment 3, Recollect judge ments show ed directed forgetting with the item method but no t with the list method. Thus, the present research further confirms that the processes underlying item- and listm ethod directed forgetting differ. The present results should be of interest to those who study inhibitory m echanism s in recall. Although directed forgetting was greater with the item m ethod than with the list method when pictures and images (which encourage item -specific processing) were studied, it was equivalent with the two methods when words alone (which encourage relationa l processing) were studied. Thus, for materials that encourage relational processing, the retrieval inhibi tion process m ay yield substantial directed forgetting. Perhaps because directed forgetting tends to be more reliably obtained with the item method than with the list method (e.g. Basden et al., 1993), m any recent experiments have been conduc ted with the item method, particularly when implicit tests were used (e.g. Gardiner et al., 1994; Golding et al., 1994; M acLeod, 1989; Russo & Andrade, 1995). This seem s an unfor tunate choi ce because, by our account, it is the list method that reflects subjects ability to lose or regain access to previously stored information. Understanding of retrieval inhib ition in the context of directed forgetting may facilitate our unde rstanding of retrieval inhibition in more exotic contexts, such as recovery of repressed memories (e.g. Cloitre et al., 1995). Interest in inhibi tory m echanism s in memory has enjoyed a recent resurgence and new theoretical approa ches have em erged. Zacks, Radvansky, and Hasher (1995) presented an approach that distinguishes between two different inhibitory m echanism s in directed forgetting. The first involves retrieval inhibition; the second involve s the stopping of rehearsal following presentation of a Forget cue. The latter m echanism is one that prevents inform ation from entering working m emory (see Hasher & Zacks, 1988). According to this formulation older subjects are deficient in inhibitory processing. Target information that is designated as to-be-forgotten may not be successfully inhibited, either because subjects cannot successfully inhibi t retrieval of previous ly stored inform ation or because items designated as to-be-forgotten are processed just as effectively as item s that are not designated as to-be-forgotten. Experiments conduc ted by Zacks et al. show ed that older subjects were less successful in meeting the requirements of directed forgetting tasks than younge r subjects with both the item and the list methods. Greater processing of Forget items with the item m ethod m ay occur when attentional inhibi tion is less efficient, and increased retrieval of fully processed Forget item s may occur with the list method when retrieval inhibition is less efficient.

DIR E C T E D F O R G E T T IN G

651

It has been proposed (Geiselman et al., 1983, p.63) that directed forgetting can be accounte d for in terms of a m echanism by which ``an F cue serves to initiate a process that inhibits the accessibility of a space in time in episodic memory. How ever, when we m easured m emory for context by asking subjects to provide serial position information for the Forget item s they had recalled, we found that experim ental subjects were as accurate in their judgem ents as were control subjects. As our measure of contextual m emory is less likely to have been contaminated by attribution al errors than is the m easure used by Geiselman et al., we conclude that there is little evidence to support an explanation of retrieval inhibition in directed forgetting that relies on loss of contextual information. In conclusion, our results clearly show the desirability of distinguishing between the two m ethods of directed forgetting, as initially propos ed by Bjork (1972). Item-method directed forgetting relies on proce ssing distinctive features of individual targets and influences retention of Remember and Forget targets in a manner analogous to levels of processing procedures. On the othe r hand, listmethod directed forgetting encourages relational proce ssing and appears to differentially influence performance only on a free recall test, a test that is sensitive to relationa l processing. W hen subjects are provide d with retrieval information in the form of word associates, word fragments, or copy cues, directed forgetting with the list m ethod disappears. Only when subjects are required to rely on their ow n organ isational system is list-method directed forgetting manifested. These observations support the idea that list-m ethod directed forgetting is dependent on organisational (relational) processing, whereas item-m ethod directed forgetting is dependent on distinctive (itemspecific) processing.
Manuscript received 25 M ay 1995 M anuscript accepted 11 December 1996

R E FER EN C E S
Basden, B.H., Basden, D.R., Coe, W .C., Decker, S., & Crutcher, K. (1994). Retrieval inhibition in directed forgetting and posthypnotic amnesia. International Journal of Clinical and Experimental Hypnosis, 42, 184203. Basden, B.H., Basden, D.R ., & Gargano, G.J. (1993). Directed forgetting in implicit and explicit memory tests: A comparison of methods. Journal of Experimental Psychology: Learning, M emory and Cognition, 19, 603616. Bjork, R.A. (1970). Positive forgetting: The noninterference of items intentionally forgotten. Journal of Verbal Learning and Verbal Behavior, 9, 255268. Bjork, R.A. (1972). Theoretical implications of directed forgetting. In A.W . Melton & E. Martin (Eds.), Coding processes in human memory (pp. 217235). W ashington, DC: Winston. Bjork, R.A. (1989). Retrieval inhibition as an adaptive mechanism in human memory. In H.L. Roediger III, & F.I.M . Craik (Eds.), V arieties of memory and consciousness; Festschrift for Endel Tulving: M emory Research (pp. 309330). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

652

B A S D E N AN D B A S D E N

Bray, N.W ., Justice, E.M ., & Zahm, D.N. (1983). Two developmental transitions in selective remembering strategies. Journal of Experimental Child Psychology, 36, 4355. Brown, A.S., & Mitchell, D.B . (1994). A re-evaluation of semantic versus nonsemantic processing in implicit memory. M emory & Cognition, 22, 533541. Bugelski, B.R. (1970). Words and things and images. American Psychologist, 25, 10021012. Challis, B.H., & Brodbeck, D.R. (1992). Level of processing affects priming in word fragment completion. Journal of Experimental Psychology: Learning, M emory, and Cognition, 18, 595 607. Cloitre, M., Cancienna, J., Brodsky, B., Zeitlin, S.B., Dulit, R., & Perry, S. (1996). M emory performance among wom en with parental abuse histories: Enhanced directed forgetting or directed remembering? Journal of Abnormal Psychology, 105, 204211. Einstein, G.O., & Hunt, R.R. (1980). Levels of processing and organization: Additive effects of individual-item and relational processing. Journal of Experimental Psychology: Human Learning and M emory, 6, 588598. Epstein, W. (1970). Facilitation of retrieval resulting from post-input exclusion of part of the input. Journal of Experimental Psychology, 86, 190195. Gardiner, J.M . (1988). Functional aspects of recollective experience. Memory & Cognition, 16, 309313. Gardiner, J.M ., Gawlik, G., & Richardson-Klavehn, A. (1994). M aintenance rehearsal affects knowing not remembering; Elaborative rehearsal affects remembering not knowing. Psychonomic Bulletin and Review, 1, 107110. Gardiner, J.M., & Java, R.I. (1993). Recognizing and remembering. In A. Collins, S. Gathercole, M. Conway, & P. M orris (Eds.), Theories of memory (pp.163188). Hillsdale, NJ: Lawrence Erlbaum Associates Inc. Geiselman, R.E., & Bagheri, B. (1985). Repetition effects in directed forgetting: Evidence for retrieval inhibition. M emory & Cognition, 13, 5762. Geiselman, R.E., Bjork, R.A., & Fishm an, D.L. (1983). Disrupted retrieval in directed forgetting: A link with posthypnotic amnesia. Journal of Experimental Psychology: General, 112, 5872. Golding, J.M ., Long, D.L., & MacLeod, C.M . (1994). You can t always forget what you want: Directed forgetting of related words. Journal of Memory and Language, 33, 493510. Hasher, L., & Zacks, R.T. (1988). Working memory, comprehension, and aging: A review and a new view. In G.H. Bower (Ed.), The psychology of learning and motivation (Vol. 22, pp.193 225). San Diego, CA: Academic Press. Hunt, R.R ., & Einstein, G.O. (1981). Relational and item-specific information in memory. Journal of Verbal Learning and Verbal Behavior, 20, 497514. Hunt, R.R ., & M arschark, M. (1989). Yet another picture of imagery: The roles of shared and distinctive information in memory. In M .A. M cDaniel & M . Pressley (Eds.), Imagery and related mnemonic processes: Theories, individual differences, and applications. New York: SpringerVerlag. Hunt, R.R ., & M cDaniel, M .A. (1993). The enigma of organization and distinctiveness. Journal of M emory and Language, 32, 421445. Johnston, H.M . (1994). Processes of successful intentional forgetting. Psychological Bulletin, 116, 274292. M acLeod, C.M . (1989). Directed forgetting affects both direct and indirect tests of memory. Journal of Experimental Psychology: Learning, M emory and Cognition, 15, 1321. Paller, K.A. (1990). Recall and stem-completion priming have different electrophysiological correlates and are modified differentially by directed forgetting. Journal of Experimental Psychology: Learning, M emory, and Cognition, 16, 10211032. Rajaram, S. (1993). Remembering and knowing: Two means of access to the personal past. M emory & Cognition, 21, 89102.

DIR E C T E D F O R G E T T IN G

653

Roediger, H.L . III, & McDermott, K.B. (1993). Implicit memory in normal human subjects. In H. Spinnler & F. Boller (Eds.), Handbook of neuropsychology, Vol. 8 (pp.63131). Am sterdam: Elsevier. Russo, R., & Andrade, J. (1995). The directed forgetting effect in word fragment completion: An application of the process dissociation procedure. The Quarterly Journal of Experimental Psychology: Section A: Human Experimental Psychology, 48A, 405423. Snodgrass, J.G., & Vanderwart, M. (1980). A standardized set of 260 pictures: Norm s for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and M emory, 6, 174215. Thapar, A., & Greene, R.L. (1994). Effects of level of processing on implicit and explicit tasks. Journal of Experimental Psychology: Learning, M emory, and Cognition, 20, 671679. Toth, J.P., Reingold, E.M., & Jacoby, L.L. (1994). Toward a redefinition of implicit memory: Proc ess dissociations following elaborative processin g and self-generation. Journal of Experimental Psychology: Learning, M emory, and Cognition, 20, 290303. Tulving, E. (1985). M emory and consciousness. Canadian Psychologist, 26, 112. Tulving, E., Schacter, D.L., & Stark, H.A. (1982). Priming effects in word-fragment completion are independent of recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 336342. Tzeng, O.J.L., Lee, A.T., & W etzel, C.D. (1979). Temporal coding in verbal information processing. Journal of Experimental Psychology: Human Learning and M emory, 5, 5264. Woodward, A.E., Bjork, R.A., & Jongeward, R.H. (1973). Recall and recognition as a function of primary rehearsal. Journal of Verbal Learning and Verbal Behavior, 12, 608617. Zacks, R.T., Radvansky, G., & Hasher, L. (1996). Studies of directed forgetting in older adults. Journal of Experimental Psychology: Learning, M emory and Cognition, 22, 143156.

You might also like