Disrupted Retrieval in Directed Forgetting: A Link …...erate motivational set to forget than with a passive nonrehearsal or inattention strategy (Weiner & Reed, 1969). This article

Journal of Experimental Psychology: General1983, Vol. 112, No. 1,58-72

Copyright 1983 by the American Psychological Association, Inc.0096-3445/83/1201-0058S00.75

Disrupted Retrieval in Directed Forgetting: A LinkWith Posthypnotic Amnesia

Ralph E. Geiselman, Robert A. Bjork, and Deborah L. FishmanUniversity of California, Los Angeles

Certain reliable findings from research on directed forgetting seem difficult toaccommodate in terms of the theoretical processes, such as selective rehearsal orstorage differentiation, that have been put forward to account for directed-for-

x getting phenomena. Some kind of "missing mechanism" appears to be involved.In order to circumvent the methodological constraints that have limited the con-clusions investigators could draw from past experiments, a new paradigm is in-troduced herein that includes a mixture of intentional and incidental learning.With this paradigm, a midlist instruction to forget the first half of a list was foundto reduce later recall of the items learned incidentally as well as those learnedintentionally. This result suggests that a cue to forget can lead to a disruption ofretrieval processes as well as to the alteration of encoding processes postulatedin prior theories. The results also provide a link between intentional forgettingand the literature on posthypnotic amnesia, in which disrupted retrieval has beenimplicated. With each of these procedures, the information that can be remem-bered is typically recalled out of order and often with limited recollection forwhen the information had been presented. It therefore was concluded here thatretrieval inhibition plays a significant role in nonhypnotic as well as in hypnoticinstances of directed forgetting. The usefulness of retrieval inhibition as a mech-anism for memory updating was also discussed.

Although most instances of forgetting areunintentional or incidental, there are occa-sions when we try to forget, either becausethe memory is unappealing or because thememory constitutes a source of interferencein conducting routine mental operationssuch as memory updating. Consequently, se-lective forgetting has been assigned an inte-gral role in the processing of to-be-remem-bered (R) information by various scholarsand researchers. In 1882, for example, Ribotwrote in his book thatwithout the total obliteration of an immense number ofstates of consciousness, and the momentary repressionof more, recollection would be impossible. Forgetfulness,except in certain cases, is not a disease of memory, buta condition of its health and life. (p. 61)

In 1890, James wrote that "if we remem-bered everything, we should on most occa-

Portions of this research were presented by the seniorauthor as part of a symposium on directed forgetting atthe meeting of the American Psychological Association,Toronto^ August 1978.

Requests for reprints should be sent to Ralph E. Gei-selman, Department of Psychology, University of Cali-fornia, Los Angeles, California 90024.

sions be as ill off as if we remembered noth-ing" (p. 680). More recently, Bjork (1972)has stated, "That we need to update ourmemories is clear: We would degenerate toa proactive-interference-induced state of to-tal confusion otherwise" (p. 218).

The apparent importance of selective for-getting in daily experience has led to a broad-based search for the mental mechanisms un-derlying intentional forgetting. Some mem-ory theorists have contended that intentionalforgetting can be carried out simply throughselective inattention. The Roth MemoryCourse (Roth, 1918/1961), for example,teaches thatby denying any attention to your temporary mental[associations] after they have served their purpose, theywill pass out of your mind. In this way, your [codingschemes] are left free for filing other facts, (p. 287)

There is some empirical evidence, however,that subjects can influence the inaccessibilityof memories to a greater extent with a delib-erate motivational set to forget than with apassive nonrehearsal or inattention strategy(Weiner & Reed, 1969). This article attemptsto (a) evaluate the explanatory status of cer-tain mechanisms that have been proposed to

58

DISRUPTED RETRIEVAL 59

account for directed-forgetting phenomena,(b) suggest an alternative methodology to cir-cumvent some of the methodological limi-tations that characterize past research, (c)provide evidence that retrieval inhibition isan important mechanism in the intentionalforgetting of information, and (d) note anempirical link between two currently inde-pendent literatures on directed forgetting.

Directed Forgetting: Basic FindingsOne popular method for studying the

memory mechanisms involved in selectiveforgetting has been to instruct subjects atsome point during the presentation of Ritems that the items already presented arenow to be forgotten and will not be testedlater. The volume of literature that has beengenerated under the rubric of directed for-getting has been enormous with an incrediblenumber of variations in procedure (see Bjork,1972; Epstein, 1972; and Timmins, 1976, forreviews). In many of the more recent exper-iments, these paradigms have been used asinvestigative tools to address issues of mem-ory other than the nature of intentional for-getting itself. Among the topics that havebeen addressed are effects of rehearsal on re-call and recognition (Woodward, Bjork, &Jongeward, 1973) and constituent processesin the differentiation of items in memory(Bjork & Geiselman, 1978).

For the present purposes, there are twothings that we would like to know in consid-ering studies of directed forgetting. The firstis the effect of the forget instruction on therecall of the R items. Hence, the typical studycompares the level of recall of R items pre-sented after a cue to forget the prior itemsin a list with the level of recall of R items intwo control conditions—one in which onlythe R items are presented, and one in whichsubjects are not given the cue to forget theinitial items. Such comparisons typicallydemonstrate the striking power of the forgetinstruction to eliminate the interference fromthe to-be-forgotten (F) items—sometimescompletely (Bjork, 1970). Studies of directedforgetting, then, substantiate the suggestionsof Ribot (1882), James (1890), and Bjork(1972) that selective forgetting plays an in-tegral role in the successful processing of Rinformation.

The second thing one would like to knowis the fate of those items that have been cuedto be forgotten. This question is the primary

, focus of the present research. Even thoughthe recall of F items is typically very low whentested, the recognition of F items is typicallyquite high (Davis & Okada, 1971), sometimesequal to that for R items (Block, 1971; Gei-selman, 1974,1977). Because of these results,the hypothesis that a cue to forget serves toerase an F item from memory has been dis-missed. Similarly, it does not appear that acue to forget inhibits the consolidation of anF item in memory (Bjork & Geiselman,1978).

Theories of directed forgetting largely havetended to emphasize selective remembering,not selective forgetting. That is, the differencein a subject's ability to recall the R versus Fitems has usually been explained in terms ofa difference in the amount or type of re-hearsal given to the items. There are a varietyof results that are consistent with such anexplanation. Often, for example, the increasein recall of R items that accompanies a cueto forget is approximately equal in magni-tude to the corresponding decrease in recallof the F items (Geiselman, 1974; Reitman,Malin, Bjork, & Higman, 1973). Also, if theforget instruction is given at the time of recall,rather than immediately following the Fitems, the .forget instruction has little if anyeffect on recall performance (Bjork, 1970).Finally, one manipulation that appears toreduce the deleterious effect of cue to forgeton F-item recall is to increase the amount ofelaborative rehearsal given to the item beforeit is cued to be forgotten. For example, Bug-elski (1970) has shown that requiring com-plex processing of F words, such as imaging,eliminates the effect of forget cues on the re-call of the F words. Also, Roediger and Crow-der (1972) have shown that when subjects aretold to forget an item within the Brown-Pe-terson paradigm, the subjects perfornj moreoptimally on the distractor task.

Bjork (1970, 1972) has suggested that twointerrelated processes operating at encodingcan account for most of the observed di-rected-forgetting phenomena. According toBjork's theory, subjects are presumed to (a)devote all post-F-cue rehearsal and othermnemonic activities to the R items but also

60 R. GEISELMAN, R. BJORK, AND D. F1SHMAN

to (b) group R items in memory in a waythat functionally segregates or differentiatesthem from the F items. The addition of thedifferential-grouping mechanism is necessaryto account for two classes of results. One isthat when paired associates are used as thestimulus materials, subjects rarely give R re-sponses to F-item probes or vice versa, eventhough they do make intrasublist intrusionerrors (Bjork, 1970). The other is that if aseries of logically connected sentences is usedas the stimulus materials, the decrease inF-item recall and the increase in R-item re-call are less than if sentences are used thatdo not have a logical ordering (Geiselman,1974). This result implies that differentialgrouping of the items is necessary in orderto conduct selective rehearsal efficiently.

A Missing Mechanism?

Although the selective-rehearsal and dif-ferential-grouping mechanisms together givea reasonable qualitative account of directed-forgetting phenomena, there are some situ-ations where it is difficult to explain how sub-jects could accomplish either selective re-hearsal or differential grouping. Consider oneexample. In an experiment by Jongeward,Woodward, and Bjork (1975), subjects werepresented lists of 32 words, each of whichconsisted of eight four-word sets. Each set offour words was presented one word at a time(2.3 sec per word). After the last word in aset, there was a 3-sec rehearsal period, whichwas in turn followed by a 1-sec cue that in-formed the subjects as to which of the fourwords, if any, they would have to rememberat the end of the experiment. Immediatelyafter the cue, the next word set was presented.Subjects were instructed to devote all of theirrehearsal activities to the current set ofwords, something they said they were onlytoo happy to do because of the demandingnature of the task.

Jongeward et al's experimental procedurewould appear to provide little in the way ofopportunities to devote differential process-ing to R versus F words. During the' 12.2 secfrom the time the first word in a set was pre-sented to the cue following the rehearsal pe-riod, subjects did not know which, if any, ofthe four words they would need to remember.

Subjects were asked not to (rehearse wordsfrom prior sets during the presentation oflater sets, and they said that they did not. Insupport of their claim, there was no relationobtained between the probability of recallingan R word and the number of R words in thefollowing set. In spite of all that, subjects re-called approximately 35% of the R words butintruded less than 5% of the F words.

Another indication that subjects may usean F cue to do more than simply stop pro-cessing the F items can be seen in a study ofpupillary responding during a directed-for-getting task. The magnitude of task-evokedpupillary dilations is commonly taken as anindex of cognitive load. Johnson (1971) re-ported a short-lived increase (followed by asizeable decrease) in pupil dilation in subjectsimmediately following a cue to forget all prioritems in a list. Although it has been arguedthat the momentary decrease in residual pro-cessing capacity is due to the processing ofthe cue itself, this seems unlikely. It shouldnot take more processing capacity to processa cue that is well learned in the experimentand expected occasionally than to processanother list item. Further, it is probably notthe case that the pupillary response imme-diately following an F cue is due solely to thepreparation of a new rehearsal scheme for theR items because such a process is also nec-essary at the very beginning of the list whereno such response was observed. A startle-re-sponse explanation is inappropriate as wellbecause a control group who received thecues without meaning did not show a cor-responding pupil dilation. It is curious thenwhat must have occurred, either consciouslyor unconsciously, immediately following thecue to forget.

Retrieval Inhibition

The results mentioned above and someresults from additional experimental para-digms (cf. Bjork & Geiselman, 1978) suggestthat there may be a missing mechanism inthe account of directed forgetting. One pos-sibility is that a cue to forget can initiate aprocess that inhibits or blocks access routesto the episodic memory traces correspondingto the F items, making them nonretrievableat the time of recall except in the presence


of prepotent retrieval cues, such as the copycues on a recognition test. With respect tothe retention of items that are explicitly cuedto be forgotten, we have not had an adequateparadigm to evaluate the notion of retrievalinhibition. Within the framework of directedforgetting, if one were to instruct a subjectthat he should actively attempt to forget (re-press) the F items, how then should we mea-sure his retention of those items? We havecreated a dilemma for our subjects. Our re-sults may be due to nothing more than con-scious response withholding (suppression) ofretrievable items. As Weiner (1968) has noted,one is reminded (in such situations) of a story about aking who was told that any wish would be granted, aslong as he did not think of the eye of a camel whenmaking the wish. Needless to say, none of his wishes werefulfilled, (p. 217)

Similarly, Roediger (Note 1) has noted thatresponse withholding could be responsiblefor a variant of directed forgetting, posthyp-notic amnesia.

One study that has addressed the questionof whether subjects can actively forget somebits of encoded information was carried outby Timmins (1976, Experiment 6). His ex-periment was modeled on what may be re-ferred to as the "harried short-order cook"phenomenon. Once an order has been filled,the cook must actively forget that order sothat it will not interfere with his memory forunfilled orders. In Timmin's experiment,subjects were presented a list of words, someof which were repeated in the list. Only non-repeated words were to be remembered. Theinitial presentation and encoding of a wordwas viewed to be analogous to the taking ofan order, and its repetition to be analogousto the filling of that order, at which time itcould (and should) be forgotten. Nonre-peated words represented orders that wereunfilled and therefore were to be remem-bered. The subjects were told that when aword was repeated, they should "stop tryingto remember it and forget it if you can." Theywere further told that it was to their advan-tage to actively forget the repeated items dur-ing the presentation of the list so that theseitems would not interfere with their memoryfor the nonrepeated items.

The results were consistent with the hy-pothesis of active forgetting. In a condition

where no mention was made of forgetting therepeated items, the repeated items were re-membered more often than the nonrepeateditems (43% vs. 27%). However, with the ac-tive-forgetting instructions, the repeated (F)items were recalled slightly less often thanwere the nonrepeated items (21% vs. 27%),even though the subjects were told at test totry to recall all the items. This result, whichis reminiscent of the Zeigarnik effect whereuncompleted tasks are more likely to be re-membered than completed tasks (filled or-ders), provides some support for the notionthat subjects can actively block access routesto previously encoded information. However,as was noted above, we cannot be sure towhat extent the subjects were engaging inactive response withholding (suppression) attest, even though they were told to try to re-call everything.

If it is the case that subjects initiate an in-hibition process when a cue to forget is given,then a later countercommand making the Fmaterial R material may serve to free manyof the blocked access routes. This result istypical in studies of posthypnotic amnesia(Evans & Kihlstrom, 1973; Freud, 1920/1952, p. 288; Kihlstrom & Evans, 1976) inwhich hypnotized subjects are given a sug-gestion to forget certain events with an ac-companying countercommand that will latersignal the removal of this suggestion. Sucha countercommand could not negate an ini-tial encoding deficit resulting from the F cue;but the countercommand does, in large part,reverse the previous recall impairment. Ananalogous experiment on nonhypnotic for-getting was conducted by Reed (1970). In hisexperiment, some letter trigrams were ini-tially cued R, whereas others were initiallycued F. On a later repetition of the trigrams,the F commands were reversed. Consistentwith the inhibition release idea and at firstglance inconsistent with the encoding deficitnotion, subjects were able to recall just asmany trigrams that were initially F cued aswere initially R cued. Unfortunately, inde-pendent research on the repetition effect sug-gests that the magnitude of the effect may begreater if the initial presentation is more dif-ficult to remember at the time of the secondpresentation (cf. the argument in Bjork,1972, p. 228). Thus, Reed's results are not

62 R. GEISELMAN, R. BJORK, AND D. FISHMAN

entirely persuasive because an initial encod-ing deficit owing to the F cues, rather thanretrieval inhibition, could have been offset byenhanced encoding at repetition, rather thaninhibition release.

Experiment 1

It is apparent, then, that new experimentalparadigms are needed to circumvent pastmethodological limitations if retrieval inhi-bition is to be convincing as a contributor todirected-forgetting phenomena. The presentparadigm involves a typical directed-forget-ting procedure in which an unpredictable Fcue is presented in the middle of certain lists.The innovation is that some materialthroughout the list (both precue and postcue)is learned incidentally, via a pleasantness-judging task, and is therefore not explicitlysubject to the F cue. Thus, any effect of theF cue on the retention of the incidental items,for which selective rehearsal or response,withholding would be unlikely, can be com-pared with the effect of the F cue on the to-be-learned items. If the F cue were to haveno effect on the recall of the items that thesubjects had no intent to rehearse for laterrecall (the to-be-judged words), then selectiverehearsal would provide an adequate ac-counting of the results. If, on the other hand,the F cue were to affect the recall of the itemslearned incidentally as well as those learnedintentionally, then disrupted retrieval forevents occurring prior to the cue would, pro-vide a more plausible explanation of the re-sults. Response withholding should not be afactor because the subjects are not told toforget the items learned incidentally, andhence there is no conflict created.

MethodSubjects. The subjects were 64 undergraduate vol-

unteers from the introductory psychology course at theUniversity of California, Los Angeles. The subjects weretested in groups of four to seven. Course credit was givenin exchange for-participation in the experiment.

Materials and procedure. A list of 48 four-letternouns was presented to subjects auditorily, with 7-secinterword intervals. To-be-learned and to-be-judged wordswere presented in strict alternation throughout the 'list.To-be-learned words were each preceded by an instruc-tion to learn that word in preparation for a recall testat the end of the experiment (e.g., "learn hand"). Theremaining words were each preceded by an instruction

to judge that word on the basis of pleasantness (e.g.,"judge boat"). The experiment was thus presented as adual-requirement task: The subjects were told explicitlythat the judged words were not to be learned. Specifically,the instructions stated that two studies were being con-ducted simultaneously to save time, oqe to see how mem-orable certain words were and one to see how pleasantcertain words were.

At the midpoint in the list, half the Subjects were givenan instruction to forget the preceding to-be-learnedwords. They were told: "What you have done thus farhas been practice; therefore, you should forget about allof the to-be-learned words that you have heard." Theremaining subjects participated in the R-cue conditionand were told: "The first portion of the list has now beenpresented; continue to try to remember the to-be-learnedwords that you have heard."

At the end of the list, all subjects were given a 3-min.distractor task, consisting of mathematical deductive rea-soning problems. Then, half of the subjects in each cuecondition (remember or forget) were'told to try to recallall of the words that they had heard in the experiment,whereas the remaining subjects were given a word-rec-ognition test. Four columns were provided for recall: onefor the to-be judged words from the first half of the list,one for the to-be-judged words from'the second half ofthe list, one for the to-be-learned words from the firsthalf of the list, and one for the to-be-learned words fromthe second half of the list. In the F-cue condition, subjectswere told that the "practice" items comprised the firsthalf of the list and that it was important to write downthese words as well as the words from the second halfof the list. Eight min. were allowed for recall.

For the 32 subjects who were given the recognitiontest instead of the recall test, a sheet of paper was pro-vided with 96 four-letter nouns typed on it, consistingof the 48 list items plus 48 distractors'. The subject's taskwas to respond "yes" or "no" to each word dependingon whether the word had appeared in the experimentallist.

Results and Discussion

Word recall. With respect ;to the recall ofthe learn words, the typical directed-forget-ting phenomena were obtained. Recall of thelearn words from the first half of the list waspoorer with the F cue than with the R cue,whereas recall of the learn words from thesecond half of the list was greater with the Fcue. The results of the recall task are pre-sented as a function of input serial positionin Figure 1 and collapsed across serial posi-tion in Figure 2. Inspection of either figureshows that the recall of the judge words fol-lowed the same pattern as with the learnwords, just at a lower level of performanceoverall. The Cue (remember or forget) X ListHalf interaction effect was significant, F(\,30) = 47.9, MSe = .02, p < .001, and analyses


1.0

Q .8LJ

OUJ

I A

occa.

FIRST LIST HALF (Precue)

•"Learn"Words

. o "Judge"Words

SECOND LIST HALF!Postcue)

XFcue)

V^(Rcue)x*--,.y ^-^

(Fcue)

1 2 3 4 1 2 3 4

INPUT SERIAL POSITION GROUP

Figure 1. Proportion correct recall in Experiment 1 as a function of list half, type of midlist cue, andserial position group. (R = remember; F = forget.)

of the simple main effects showed that theabove-mentioned relationships were signifi-cant for the judge words (ps < .05) as well asfor the learn words (ps < .001). However, theWord-Type (learn vs. judge) X Cue X ListHalf interaction effect was also significant,F(i, 30) = 8.3, MSe = .02,;? < .01. The effectof the F cue on the recall of the judge wordswas not as great as the effect of the F cue onthe recall of the learn words-.

These results are consistent with the hy-pothesis that an F cue serves to initiate a pro-cess that inhibits the accessibility of a spaceof time in episodic memory. That is, not only

.8

Q .7nj

a .5

DBcuei)F_cue

"Learn"Words

FIRST LIST HALF

"Judge"Words

"Learn"Words

SECOND LIST HALF

"Judge"Words

Figure 2. Proportion correct recall in Experiment 1 col-lapsed across serial position. (R = remember; F = forget.)

were the to-be-learned items more poorly re-called following the F cue but so were itemsthat the subjects had no intent to rememberanyway. It is, therefore, difficult to explainthese results in terms of a selective-rehearsalmechanism or in terms of response with-holding. In a postexperimental debriefing ses-sion, none of the recall subjects and onlythree of the recognition subjects reportedhaving suspected a memory test on the judgewords.

The greater effect of the F cue on the learnwords is understandable since after the F.cue,rehearsal processes would be devoted entirelyto the second list-half learn words. The firstlist-half learn words would therefore receiveless rehearsal than in the R-cue condition,and the second list-half learn words wouldreceive more rehearsal. The judge words, onthe other hand, are not rehearsed in any con-dition so such selective rehearsal argumentscannot apply to those items. In fact, the en-hanced recall of postcue judge words in theF-cue condition is as surprising as the poorerrecall of the precue judge words. Two sourcesof evidence imply that the subjects did notrehearse the judge words along with the learnwords. First, on a postexperimental ques-tionnaire, none of the 32 subjects reportedthat they suspected a "surprise" test on thejudge words. Second, inspection of Figure 1shows that there was no second list-half pri-


macy effect for the judge items, whereas thereis a strong second list-half primacy effect forthe learn words following the F cue.

Finally, it is apparent from Figure 2 thatmore learn words were recalled than judgewords, F(l, 39) = 25.2, MSe = .03, p < .001.This outcome is intriguing because Hyde andJenkins (1969, 1973) and others have shownthat intent to remember has no greater effecton recall than certain semantic orientingtasks, such as the pleasantness judgment taskin the present experiment. However, intentto learn has been found to yield greater de-layed recall when intent is manifested in dif-ferential interitem organization (Battig &Bellezza, 1979). Given that the judge wordswere not rehearsed, it appears that they donot profit from the kind of interitem asso-ciations that are formed via cumulative re-hearsal in the learn word case. Alternatively,perhaps the interassociation network for thelearn words served to inhibit the retrieval ofthe judge words (Anderson & Bower, 1973).

Word classification and input-output orderanalyses. The obvious initial explanation ofthe above recall results would be that differ-ential grouping of the learn and judge wordsin memory was rather poor, consequently,some judge words became, functionally, learnwords for purposes of rehearsal. However, inthe subjects' recall protocols, the discrimi-nability of learn versus judge words was ex-cellent. Less than 5% of the total words re-called of either type were misclassifled withrespect to the learn-judge distinction. A cell-by-cell breakdown of these results is pre-sented in Table 1. On the basis of this finding,the hypothesis that judge words received in-advertent rehearsal because they were con-fused with learn words seems unlikely. None-theless, Experiment 2 was designed to providea further, more conclusive test of such a pos-sibility.

It was also observed that the subjects hadrelatively poor memory for the list-half mem-bership of both learn and judge words pre-sented before the F cue (see Table 1). TheCue (remember or forget) X List Half inter-action effect was significant, F( 1, 30) = 11.54,MSe = .07, p< .001. Further, the order inwhich the F learn words were recalled wasrelatively unrelated to the order in which theywere presented (see Table 2). These results

Table 1Proportion Correct List-Half Classification GivenRecall and Proportion Correct Learn-JudgeClassification Given Recall

List-half words

First Second

List- Learn- List- Learn-half judge half judge

Word type and classifi- classifi- classifi- classifi-midlist cue cation cation cation cation

Intentional learnwords

RememberForget

Incidental judgewords

RememberForget

.92

.63

.72

.36

.92

.95

1.00.86

.95

.94

.69

.75

' .951.00

.881.00

are in contrast to those observed for wordspresented after the F cue, or before or afterthe R cue. This additional indication of dis-rupted retrieval of F items provides an in-teresting parallel with results from studiesconcerning posthypnotic amnesia (see Kihls-trom, 1977, for a review). Evans and Kihls-trom (1973), for example, reported that forsubjects who are susceptible to hypnosis,events that can be remembered in spite of ahypnotic suggestion to forget typically arerecalled in an order unrelated to the orderof input. In contrast, a significant positiverelationship is usually observed between in-put order and output order in the recall pro-tocols of subjects who are relatively unsus-ceptible to hypnosis.

Table 2Average Rho Correlation Between Input andOutput Order for Learn Words Recalledas Learn Words

List-half words

First Second

Midlist cue

RememberForget

.72*

.29128

.70*

.84*108

Note. Correlation computed for those subjects who re-called three or more learn words.*p<.01.


Thus, the hypothesis that directed forget-ting involves disrupted retrieval provides alink with the literature on posthypnotic am-nesia, suggesting that at least one of the un-derlying mechanisms is similar in the twosituations in spite of the dramatic proceduraldifferences. The inability of subjects to clas-sify the F items as first list-half words is alsoconsistent with the conclusion of Stern (1981)that human amnesia can, in some cases, beattributed to a context retrieval deficit.

It is instructive to note that disorganizedretrieval in the present data was not simplya property of the items that were least mem-orable. As can be seen by comparing Figure2 with Tables 1 and 2, the words presentedin the second list half following the R cuealso were recalled at a low level, comparableto that of the pre-F-cue words, but the clas-sification performance was much more ac-curate for the post-R-cue words (.95 vs. .63for the learn words and .69 vs. .36 for thejudge words), and the input-output ordercorrelation was highly significant for the post-R-cue words (.70). These results are consis-tent with findings reported within the hyp-nosis domain. Schwartz (1980), for example,found the retention of order information tobe unrelated to the number of items recalledunder hypnosis (r = .01).

One final result of interest from Table 1is that the learn words were classified cor-rectly more often than the judge words, P(l,30) = 67.8, MSe = .03, p < .001. This dif-ference is consistent with the hypothesis thatorder information includes both temporaland relation-chaining cues (Lee & Estes,1977). With the judge words, only temporalcues would be available since cross-item in-terassociations were not intentionally devel-oped. This result provides additional indirectevidence that the subjects did not rehearsethe judge words during the presentation ofthe list.

Word recognition. The results of the yes-no word-recognition test are presented inFigure 3. An analysis of variance (ANOVA)revealed no significant differences among themeans. This outcome is consistent with thehypothesis considered above tKat the differ-ences observed in the recall results are largelyretrieval phenomena. The complete absenceof significant effects of list half, item type,

I 1.0I -9

S .8

O

CD mid-list R cue^mid-list F cue

' Learn"Words

r!

FA FA

"Judge"Words

FA

"Learn"Words

FA

"Judge"Words

FIRST LIST HALF SECOND LIST HALF

Figure 3. Proportion correct recognition in Experiment1 as a function of list half and type of midlist cue (R =remember; F = forget. FA = false-alarm rate.)

and cue type is quite remarkable in view ofthe dramatic effects of those same variablesreflected in the various measures of recallperformance reported above. The F cueclearly had its effects primarily on the re-trieval side.

Alternative hypotheses. There are twoplausible explanations for the pattern of re-call results described above other than theretrieval-inhibition hypothesis. First, perhapsthe judge words became unavoidably inter-associated with the learn words to some ex-tent. The effect of the F cue on the judgewords could then be explained as a disrupted-rehearsal phenomenon without reference toretrieval processes. Two aspects of the datafrom Experiment 1 argue against this hy-pothesis; namely, the judge words were notconfused with the learn words in the classi-fication data, and there was no second list-half primacy effect for the judge words fol-lowing the F cue. Nevertheless, Experiment2 was designed to render the learn and judgewords completely nonconfusable with eachother; they were drawn from two distinct se-mantic categories, nouns representing ani-mate entities and nouns representing inani-mate entities.

The second alternative hypothesis centeredon the possibility that the results of Experi-ment 1 can be attributed to differential out-put interference. Perhaps with an F cue, thesubjects recalled the words from the secondhalf of the list first, producing enhanced recall


of those words and depressed recall of boththe learn and judge words from the first listhalf. Experiment 3 and 4 tested whether theresults of Experiment 1 can be attributed tosuch differential patterns of output interfer-ence rather than to retrieval inhibition.

Experiment 2

As mentioned above, if some of the judgewords were inadvertently included in the sub-jects' rehearsal sets, then perhaps the effectof an F cue on the precue judge words isattributable to the termination of rehearsalof those words as well as the precue learnwords. This hypothesis was evaluated in Ex-periment 2 by selecting the judge words froma semantic category that is clearly differentfrom that of the learn words and by inform-ing the subjects of this difference before theexperiment began. Experiment 2 also pro-vided an opportunity to replicate the patternof recall results from Experiment 1 with thelearn-judge distinction more clearly defined.


unteers from the introductory psychology course at theUniversity of California, Los Angeles. The experimentwas carried out with groups of 4 to 7 subjects per session.Course credit was given in exchange for their partici-pation. Twenty-one subjects were assigned to the F-cuecondition, and 25 to the R-cue condition.

Materials. Forty words were selected for use in Ex-periment 2,20 nouns representing animate entities suchas painter and 20 nouns representing inanimate entitiessuch as magazine. Words with high meaningfulness val-ues (m = 5.04 to 7.96) were selected and the two sets of20 words were matched in a pairwise manner on mean-ingfulness using the Paivio, Yuille, and Madigan (1968)norms. As in Experiment 1, two tape recordings of the40 words were made, one with the team words as ani-mate entities and one with the judge words as animateentities.

Procedure. Only two modifications of the procedurefrom Experiment 1 were made. Just prior to list presen-tation, the subjects were told:

To help you keep the words that you are supposed tolearn separate from the words that you are supposedto judge, we have constructed the list such that all ofthe learn (judge) words are inanimate objects whereasall of the judge (learn) words are animate entities.

The second modification of the procedure was to shortenthe list to 40 words. The list was shortened in an attemptto raise recall performance for the judge words so as toevaluate more accurately the effect of the F cue on thoseitems. The method for testing recall of the words was

identical to that of Experiment 1, but recognition per-formance was not evaluated in this experiment.


Word recall. The ANOVA performed on theword-recall data showed that, as in Experi-ment 1, more learn words than judge wordswere recalled, F(l, 44) = 55,1, MSe = .06,p < .001. In addition, the Cue (remember orforget) X List Half interaction effect was sig-nificant, F(l, 44) = 53.4, MSe = .03, p <.001. Inspection of Figure 4 shows that morewords were recalled from the first list half inthe R-cue condition (p < .05), whereas morewords were recalled from the second list halfin the F-cue condition (p < .05). Most im-portant, this pattern held for the judge wordsas well as for the learn words, as the Cue XList Half X Word Type (learn vs. judge) in-teraction effect was not significant, F( 1,44) =3.59, MSe = .04, p > .05. Thus, with thejudge words distinguished from the learnwords by semantic category (animate vs. in-animate), the effect of the F cue on the recallof the precue judge words was still evident.This result suggests that the pattern of recallobtained in Experiment 1 is robust and thatthe hypothesis in question, namely, that thejudge words in Experiment 1 were inadver-tently intertwined with the learn words dur-ing rehearsal, is unlikely.

Word classifidation and iriput-output or-der. Because the learn and judge words weredistinguished by semantic category, only thelist-half classification errors were of interesthere. The classification data are summarized

Q

y_iucc

'' O mid-list Roue• E^ mid -list F_cue

"Learn"Words

"Judge"Words

"Judge"Words

FIRST LIST HALF

"Learn"Words

SECOND LIST HALF

Figure 4. Proportion correct recall in Experiment 2.(R = remember; F = forget.)


Table 3Proportion Correct List-Half ClassificationGiven Recall

List-half words

Word type and midlist cue First Second

Intentional learn wordsRememberForget

Incidental judge wordsRememberForget

.96

.89

.84

.72

.90

.97

.73

.89

in Table 3. As in Experiment 1, the learnwords that were recalled were classified cor-rectly by list half more often than were thejudge words that were recalled (93% vs. 79%correct), F(\, 44) = 29.5, MS,. = .04, p <.001. Also as in Experiment 1, the Cue (re-member or forget) X List Half interactioneffect was significant, F(\, 44) = 7.19, MS,. =.06, p < .01. With the R cue, list-half classi-fications were more often correct for first list-half words than for second list-half words(90% vs. 82% correct, p < .05); whereas withthe F cue, list-half classifications were moreoften correct for second list-half words (93%vs. 81% correct, p< .01). Thus, as in Exper-iment 1, the subjects clearly had difficultyclassifying the words that were presentedprior to the F cue. Again, this result is con-sistent with the hypothesis that a cue to forgetaffects recall, in part, through disruption ofretrieval processes.

Table 4Average Rho Correlation Between Input andOutput Order for Words Classified Correctly

List-half words

First SecondWord type

and midlist cue

Intentional learn wordsRemember .74** 23 .76** 20Forget .32* 17 .56** 20

Incidental judge wordsRemember .62** 16 .59** 11Forget .05 13 .50** 11

Note. Correlation computed for those subjects who re-called three or more words in that cell.*p<.05. **p<.001.

Additional evidence for disrupted retrievalis presented in Table 4. As in Experiment 1,the correlation between input order and out-put order was found to be depressed for thelearn words presented prior to the F cue.Given the somewhat higher level of recall inthis experiment, these correlations could alsobe examined for the judge words, and thesame pattern was observed as for the learnwords. In sum, these data replicate and ex-tend the classification and input-output or-der data from Experiment 1 and further sug-gest that an F cue leads to disrupted retrieval.

Experiment 3Experiment 3 was designed to ascertain

whether differential output order in the F-cueand R-cue conditions could be responsiblefor the foregoing results. If, following an Fcue, subjects choose to start their recall withthe R-word list half (that is, the second listhalf), then recall of the second list half mightbe enhanced, and recall of the first list halfmight be depressed compared with the R-cuecase. Presumably, subjects in the R-cue con-dition would show no list-half preference instarting their recall (or, possibly, would1 tendto start with the first list half). The design ofExperiment 3 permitted measurement of theaverage output position for any given wordin each condition.


unteers from the introductory psychology course at theUniversity of California, Los Angeles. They served ingroups of 4 to 7 and received course credit for theirparticipation. Sixteen subjects were assigned to the F-cuecondition, and 16 to the R-cue condition.

Materials. The materials were identical to those usedin Experiment 1 except the list was shortened to 40 wordsas in Experiment 2.

Procedure. The procedure was the same as that usedin Experiment 1 except for the method of testing wordrecall. In this experiment, the subjects were instructedto record the words in one long column, independent ofwhether the word was a learn word or a judge word andindependent of where in the list the word was presented.The subjects were told to write the words down in theorder in which they came to mind. To avoid any inter-ference with this recall procedure, the subjects were notrequired to classify the words in any way during recall.

Results and DiscussionWord recall. The word-recall results are

presented in Figure 5. As in the previous two


QRcueEl F cue

-S.Averageoutputpercentile

"Learn".Words

"Judge"Words

(!3>

"Learn"Words

"Judge"Words

FIRST LIST HALF SECOND LIST HALF


experiments, more learn words were remem-bered than judge words, P(l, 30) = 31.4,MSe = .07, p < .001, and the Cue (rememberor forget) X List Hajf interaction was signif-icant, F(l, 30) = 29.2, MSe = .07, p < .001.The latter result reflects the depressed recallfrom the first list half in the F-cue condition(,p<.01). Also as before, the Cue X ListHalf X Word Type (learn or judge) interac-tion effect did not reach significance, F(l,30) = 3.60, MS; = .06, p > .05, thus provid-ing a second replication of the inhibitory ef-fect of the F cue on the recall of the pre-F-cue judge words. The basic pattern of effectsin Experiments 1, 2, and 3 (Figures 2, 4, and5) is remarkably similar across studies.

Output order. The output-interferencehypothesis predicts that with the F cue, thewords from the second list half are recalledfirst, before the words from the first list halfare recalled. To evaluate this prediction, ameasure of average output position for eachof four classifications of words was computedfor each subject. The four classifications werefirst-half learn words, second-half learnwords, first-half judge words, and second-halfjudge words. Because the nominal outputposition is confounded between subjects withdifferences in the number of words recalled,the specific measure of output position wasan output-position percentile, as developedby Bjork and Whitten (1974). The output-position percentile for a given word that isrecalled was computed as nominal outputposition divided by total number of wordsrecalled by that subject then multipliedby 100.

The output-position percentiles averagedacross subjects are included in Figure 5 justabove each bar. A lower value indicates thatwords from that classification tended to beoutput earlier in recall. As can be seen, thelearn words were recalled before the judgewords, F(l, 30) = 25.4, MSe = .05, ;> < .001,and with the F cue, the words from the sec-ond list half were recalled prior to the wordsfrom the first list half. The Cue X List Halfinteraction effect was significant, F(l, 30) =8.2, MSe = -07, p < .01, and this pattern wasnot significantly different as a function ofword type (learn vs. judge). Thus, the output-interference hypothesis cannot be ruled out.

Input-output order. As was found in Ex-periments 1 and 2, the correlation betweeninput presentation order and output recallorder was found to be depressed for wordspresented prior to the F cue. The rho cor-relations from the present experiment arepresented in Table 5. These data constitutethe second replication of the effect of the Fcue on the input-output order relationship.

Experiment 4

The results of Experiment 3 provided asecond replication of the pattern of recallobtained in Experiment 1. The analysis ofthe output-position data, however, suggeststhat an output-interference explanation ofthe recall results cannot be ruled out. Thus,Experiment 4 was designed to provide a more

Table 5Average Rho Correlation Between Input andOutput Order for Words Classified Correctly

List-half words

First SecondWord type

and midlist cue

Intentional learn wordsRememberForget

Incidental judge wordsRememberForget

p

.75**

.34

.67*

.22

n

1411

86

P

.72**

.68**

.64*

.59*

n

1112

77

Note. Correlation computed for those subjects who re-called three or more words in that cell.* p < .05. **p< .01.


direct test of the output-interference hypoth-esis. In this experiment, the order of recallof the first list half versus the second list halfwas controlled between subjects. Accordingto the output-interference notion, controllingthe order of list-half output should eliminatethe effects of the F cue on the judge words.

Method

Subjects. The subjects were 92 undergraduates fromthe introductory psychology course at the University ofCalifornia, Los Angeles. Each subject was randomly as-signed to one of two list conditions (R cue vs. F cue) andto one of two test conditions (recall the first vs. the secondlist half first). The subjects were tested in groups of 4 to7, with 43 subjects participating in the R-cue conditionand 49 participating in the F-cue condition.

Materials and procedure. The materials were thesame as in Experiment 3. The procedure differed fromthat in the previous experiments by the nature of therecall test. In the present experiment, some of the sub-jects were required to write down the words from thefirst list half first, whereas the remaining subjects wererequired to write down the words from the second listhalf first. Three min. were allowed for recall from eachlist half, and these two recall tasks were carried out onseparate sheets of paper. The subjects were not asked toclassify the words by word type (learn vs. judge) in thisexperiment.


Word recall. The word-recall results arepresented in Figure 6. An ANOVA confirmedthat these data represent a third replicationof the basic pattern of recall found in Ex-periment 1. More learn words were recalledthan judge words, F(l, 88) = 53.4, MSe =.06, p < .001. With the R cue, more wordswere recalled from the first list half, whereas

PR

OP

OR

TIO

N

RE

CA

LL

ED

O

—

w

GI

'-&

ui

cn

^i

j

, 21.46

DScue^ F cue

).41 (.34)

1

251.28)

M (.21)

^

"Learn" "Judge"Words Words

ri.181.211

FIRST LIST HALF

"Learn" 'Judge"Words Words

SECOND LIST HALF

with the F cue, more words were recalledfrom the second list half; the Cue X List Halfinteraction effect was significant, F(\, 88) =86.9, MSe = -01, p < .001. This pattern heldfor both the learn words and the judge words.

Most important ,here, the forced order ofoutput (first list-half output first vs. secondlist-half output first) did not interact signifi-cantly with any other factor in the design (all.Fs < 1.07). This result rules out an output-interference explanation of the effects of theF cue in the present experiments.

Word classification. The subjects hi thisexperiment were not asked to classify thewords that they recalled on the basis of thelearn-judge distinction. However, classifica-tion errors could be examined with respectto list half as in Experiments 1 and 2. Thesedata are presented in Table 6. It will be re-called from Experiment 3 that the subjectstended to recall the second list-half items firstin the F-cue condition. As can be seen inTable 6, when the subjects were asked to re-call the second list-half words first, the wordsthat were classified correctly least often werethose presented before the F cue, both learnand judge, just as in Experiments 1 and 2.However, when the subjects were asked torecall the first list-half words first, an effectof the output-order manipulation was evi-dent, as the Cue X List Half X Output Orderinteraction effect was significant, F(l, 88) =4.01, MS, = .06, p < .05. Specifically,, withthe F cue, the first list-half words were clas-sified correctly more often when the first list-

Table 6Proportion Correct List-Half ClassificationGiven Recall

List-half words

First Second


Word typeand midlist cue

Intentional learnwords

RememberForget

Incidental judgewords

RememberForget

Outputlast

.92

.70

.91

.52

Outputfirst

.94

.81

.90

.82

Outputfirst

.95

.96

.99

.97

Outputlast

.82

.95

.69

.93

70 R. GEISELMAN, R. BJORK, AND D, FISHMAN

half words were output first; and with the Rcue, the second list-half words were misclas-sified more often when the first list-half wordswere output first.

Both of these outcomes are interpretableif subjects tended to write down words theywere unsure about on the first page of testing.It is not clear, however, why the same out-come did not obtain for the remaining twocases. With the F cue, the second list-halfwords were not misclassined more when thefirst list-half words were output first; with theR cue, the first list-half words were not clas-sified correctly more often when the first list-half words were output first. Perhaps the sub-jects in these conditions were less likely towrite down words they were unsure about onthe first page because a greater number ofwords were accessible to them for recall inthose cases. That is, there may have been areduced tendency to "pad" the recall on thefirst page. Because of these potential con-founding biases, input-output order corre-lations were not computed for Experi-ment 4.

General Discussion

In four experiments, a midlist cue to forgetthe words in the first half of a list served todepress later recall of those words. This effectwas observed for items that the subjects hadno a priori intent to remember (items learnedincidentally) as well as for R words. It doesnot seem possible to explain the present re-sults without postulating that a cue to forgethas the power to initiate a process that blocksor inhibits access routes to the F items. If theF cue served only to terminate rehearsal ofthe pre-F-cue learn words, then the forgetinstruction should have had no effect on therecall of the words learned incidentally, whichwere not rehearsed in any case. Because rec-ognition performance was unimpaired, it ap-pears that the inhibition induced via the Fcue took the form of retrieval blocking. Twoless dramatic, less interesting interpretationsof the depressed recall of the pre-F-cue in-cidental items were ruled out, one by Exper-iments 1 and 2 and the other by Experiments3 and 4.

Two other aspects of the recall data suggestthat disrupted retrieval plays a significant role

in intentional-forgetting experiments of thetype presented here. First, the -, subjects couldnot remember whether the words that werepresented prior to the F cue we?e presentedin the first half of the list or; in the secondhalf of the list. That is, they exhibited list-half source amnesia for words presentedprior to the F cue. Second, the average cor-relation between input order and output or-der for words presented prior to the F cuewas greatly depressed. These results are com-parable to those reported from studies ofposthypnotic amnesia (Evans & Kihlstrom,1973; Kihlstrom & Evans, 1976), in whichdisrupted retrieval also has been implicated.Thus, intentional forgetting and posthyp-nbtic amnesia Appear to have strong parallelssuch that future work in one disciplineshould impact on the other. For example, ithas been shown that subjects who experienceposthypnotic amnesia do not have differen-tially poor waking memories (Kihlstrom &Twersky, 1978), but future research may re-veal that these individuals are more likely toforget when the forgetting is intentional, asin the present experiments. That is, there maybe a subset of individuals who are extremelyefficient at forgetting. Consistent with thisargument is the recent finding by Geiselmanet al. (in press) that subjects who exhibit lowF-item recall in a nonhypnotic directed-for-getting experiment are the subjects who aremost likely to exhibit low recall (prior to thecountercommand) in a hypnotic amnesia ex-periment.

The comparison between paradigms mustbe tempered, however, because it appears thata hypnotic suggestion to forget can have aprofound effect on the accessibility of epi-sodic information that is well Jearned (Kihls-trom, 1980), whereas under typical labora-tory conditions, a nonhypnotic cue to forgethas little effect on memory for items that re-ceive elaborate precue processing (Bugelski,1970). Further, once a hypnotic suggestionto forget is lifted, previously hypnotized sub-jects can retrieve much of the target infor-mation from memory. No convincing pro-cedure for reversing a nonhypnotic forgetcommand has been reported. Thus, it is ap-parent that procedures that rely on hypnosisobtain greater control over the subjects' recallperformance, whereas the nonhypnotic pro-


cedures from'cognitive psychology lead todifferential encoding of the items in additionto disrupted retrieval.

Finally, a concluding speculation may bein order on the adaptive character of retrievalblocking as an updating mechanism (seeBjork, 1976). Any ongoing information-pro-cessing system needs some mechanism forupdating information, that is, some meansby which to prevent old, out-of-date infor-mation from interfering with new, currentinformation. In a sense, computers are theultimate in efficiency in updating becausestoring new information at a given memorylocation obliterates the prior entry. However,should it become important to gain access tothe out-of-date information, such an updat-ing system is not optimal. In comparison,inhibition that takes the form of retrievalblocking has some desirable properties as anupdating mechanism. On the one hand, in-formation that is not retrievable is also non-interfering. On the other hand, such"blocked" information is still readily recog-nizable as having occurred before, and it ap-pears that upon becoming pertinent again(that is, changing from F to R), such infor-mation shows a repetition effect (Reed, 1970).

Reference Note1. Roediger, H. L. Directed forgetting and post-hypnotic

amnesia. Paper presented at the meeting of the'American Psychological Association, Toronto, August1978.

ReferencesAnderson, J. R., & Bower, G. H. Human associative

memory. Washington, D.C.: Winston & Sons, 1973.Battig, W. E, & Bellezza, F. S. Organization and levels

of processing. In C. R. Puff (Ed.), Memory organi-zation and structure. New York: Academic Press,1979.

Bjork, R. A. Positive forgetting: The noninterference ofitems intentionally forgotten. Journal of Verbal Learn-ing and Verbal Behavior, 1970, 9, 255-268.

Bjork, R. A. Theoretical implications of directed for-getting. In A. W. Melton & E. Martin (Eds.), Codingprocesses in human memory. Washington, D.C.: Win-ston & Sons, 1972.

Bjork, R. A. The updating of human memory. In G. H.Bower (Ed.), The psychology of learning and motiva-tion (Vol. 12). New York: Academic Press, 1976.

Bjork, R. A., & Geiselman, R. E. Constituent processesin the differentiation of items in memory. Journal ofExperimental Psychology: Human Learning and Mem-ory, 1978, 4, 347-361.

Bjork, R. A., & Whitten, W. B. Recency-sensitive re-trieval processes in long-term free recall. CognitivePsychology,A974, 6, 173-189..

Block, R. A. Effects of instructions to forget in short-term memory. Journal of Experimental Psychology,1971, 89, 1-9.

Bugelski, B. R. Words and things and images. AmericanPsychologist, 1970, 25, 1002-1012.

Davis, J. C., & Okada, R. Recognition and recall of pos-itively forgotten items. Journal of Experimental Psy-chology, 1971, 89, 181-186.

Epstein, W. Mechanisms in directed forgetting. In G. H.Bower (Ed.), The psychology of learning and motiva-tion (Vol. 6). New York: Academic Press, 1972.

Evans, F. J., & Kihlstrom, J. Posthypnotic amnesia asdisrupted retrieval. Journal of Abnormal Psychology,1973, 82, 317-323.

Freud, S. A. A general introduction to psychoanalysis(J. Riviere, trans.). New York: Washington SquarePress, 1952. (Originally published, 1920.)

Geiselman, R. E. Positive forgetting of sentence material.Memory & Cognition, 1974, 2, 677-682.

Geiselman, R. E. Effects of sentence ordering on the-matic decisions to remember and forget prose. Mem-ory & Cognition, 1977, 5, 323-330.

Geiselman, R. E., et al. Mechanisms of hypnotic andnon-hypnotic directed forgetting. Journal of Experi-mental Psychology: Learning, Memory, and Cognition,in press.

Hyde, T. S., & Jenkins, J. J. Differential effects of inci-dental tasks on the organization of recall of a list ofhighly associated words. Journal of Experimental Psy-chology, 1969, 82, 472-481.

Hyde, T. S., & Jenkins, J. J. Recall of words as a functionof semantic, graphic, and syntactic orienting tasks.Journal of Verbal Learning and Verbal Behavior, 1973,12, 471-480.

James, W. The principles of psychology (Vol. 2). NewYork: Holt, 1890.

Johnson, D. A. Pupillary responses during a short-termmemory task: Cognitive processing, arousal, or both?Journal of Experimental Psychology, 1971, 90, 311-318.

Jongeward, R. H., Woodward, A. E., & Bjork, R. A. Therelative roles of input and output mechanisms in di-rected forgetting. Memory & Cognition, 1975, 3, 51-57.

Kihlstronv J- F. Models of posthypnotic amnesia. InW. E. Edmonston (Ed.), Conceptual and investigativeapproaches to hypnosis and hypnotic phenomena (Vol.296). New York: New York Academy of Sciences,1977.

Kihlstrom, J. F. Posthypnotic amnesia for recentlylearned material: Interactions with "episodic" and"semantic" memory. Cognitive Psychology, 1980,12,227-251.

Kihlstrom, J. F, & Evans, F. J. Recove'ry of memoryafter posthypnotic amnesia. Journal of Abnormal Psy-chology, 1976, 85, 564-569.

Kihlstrom, J. F, & Twersky, M. Relationship of post-hypnotic amnesia to waking memory performance.The International Journal of Clinical and Experimen-tal Hypnosis, 1978, 26, 292-306.

Lee, C. L., & Estes, W. K. Order and position in primary


memory for letter strings. Journal of Verbal Learningand Verbal Behavior, 1977, 16, 395-418.

Paivio, A., Yuille, J. C, & Madigan, S. A, Concreteness,imagery, and meaningfulness values for 925 nouns.Journal of Experimental Psychology Monograph, 1968,76, (1, Pt. 2).

Reed, H. Studies of the interference processes in short-term memory. Journal of Experimental Psychology,1970, 84, 452-457.

Reitman, W., Malin, J. X, Bjork, R. A., & Higman, B.Strategy control and directed forgetting. Journal ofVerbal Learning and Verbal Behavior, 1973,12, 140-149.

Ribot, T. A. Diseases of memory: An essay in the positivepsychology. New York: Appleton-Century-Crofts, 1882.

Roediger, H. L., & Crowder, R. G, Instructed forgetting:Rehearsal control or retrieval inhibition (repression)?Cognitive Psychology, 1972, 3, 244-254.

Roth, D. M. Roth memory course. Santa Monica, Calif.:Motivation, 1961. (Originally published, 1918.)

Schwartz, W. Hypnosis and episodic memory. The In-

ternational Journal of Clinical and Experimental Hyp-nosis, 1980, 28, 375-385.

Stern, L. D. A review of theories of human amnesia.Memory & Cognition, 1981, P, 247-262.

Timmins, W. K. Strategies of remembering: A study ofdirected forgetting. Unpublished doctoral dissertation,Australian National University, Canberra, 1976<

Weiner, B. Motivated forgetting and the study of repres-sion. Journal of Personality, 1968, 36, 213-234.

Weiner, B., & Reed, H. Effects of the instructional setsto remember and to forget on short-term retention:Studies of rehearsal control and retrieval inhibition(repression). Journal of Experimental Psychology,1969, 79, 226-232.

Woodward, A. E., Bjork, R. A., & Jongeward, R. H.Recall and recognition as a function of primary re-hearsal. Journal of Verbal Learning and Verbal Be-havior, 1973, 12, 608-617.

Received April 19, 1982Revision received September 13, 1982 •

Disrupted Retrieval in Directed Forgetting: A Link …...erate motivational set to forget than with a passive nonrehearsal or inattention strategy (Weiner & Reed, 1969). This article

Documents