This article was downloaded by: [Washington University in St Louis]On: 17 September 2013, At: 19:41Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: MortimerHouse, 37-41 Mortimer Street, London W1T 3JH, UK

MemoryPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/pmem20

Successful inhibition, unsuccessful retrieval:Manipulating time and success during retrievalpracticeBenjamin C. Storm a & John F. Nestojko ba University of Illinois at Chicago, IL, USAb University of California, Los Angeles, CA, USAPublished online: 04 Aug 2009.

To cite this article: Benjamin C. Storm & John F. Nestojko (2010) Successful inhibition, unsuccessful retrieval:Manipulating time and success during retrieval practice, Memory, 18:2, 99-114, DOI: 10.1080/09658210903107853

To link to this article: http://dx.doi.org/10.1080/09658210903107853

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy, completeness, or suitabilityfor any purpose of the Content. Any opinions and views expressed in this publication are the opinionsand views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy ofthe Content should not be relied upon and should be independently verified with primary sources ofinformation. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands,costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly orindirectly in connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial orsystematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distributionin any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found athttp://www.tandfonline.com/page/terms-and-conditions

Successful inhibition, unsuccessful retrieval:Manipulating time and success during retrieval practice

Benjamin C. Storm

University of Illinois at Chicago, IL, USA

John F. Nestojko

University of California, Los Angeles, CA, USA

Retrieving an item or set of items from memory can cause the forgetting of other related items inmemory; a phenomenon known as retrieval-induced forgetting. According to the inhibitory account ofretrieval-induced forgetting, in searching for a particular item, other items that are related but incorrectcan vie for access. Inhibition functions to decrease the accessibility of such interfering items, therebyfacilitating access to the target item. Experiments 1 and 2 replicated recent work suggesting that retrievalsuccess is not a necessary condition for retrieval-induced forgetting to occur. Interfering items wereforgotten even when retrieval practice was designed to be impossible. Experiments 3 and 4 employed theimpossible retrieval practice procedure to examine the time-course of forgetting across a single retrievalpractice trial. Results support the inhibitory account of retrieval-induced forgetting and offer insight intothe dynamics of how and when inhibition plays a role in retrieval.

Keywords: Retrieval-induced forgetting; Inhibition; Interference.

Silence pervades every moment of our social andmental lives. Whether having a conversation witha friend, planning a weekend getaway, or reminis-cing in one’s own memories, we are continuouslygenerating some ideas while remaining silent toother ideas. These silences can be intentional orunintentional. We may, for example, remainintentionally silent to withhold secret or tabooinformation from discussion. Often, however,silences are unintentional and result from theselective and competitive nature of human mem-ory. Only a subset of the information that isstored in memory can and should be retrieved at agiven point in time. Although associated with thesame retrieval cues, irrelevant and inappropriateinformation must remain non-retrieved. So what

happens to information that is not retrieved? Onemight think that such information remains un-affected in memory, silently waiting for its turn tobe recalled. Research suggests otherwise.Whereas retrieval may facilitate the future acces-sibility of information that is retrieved, it appearsto impair the future accessibility of informationthat is not retrieved. In this sense, informationthat is not selected for retrieval*that remainssilent*is even more likely to remain silent in thefuture.

The immediate consequences of retrieval areobvious*items retrieved from memory becomeaccessible to consciousness. Less obvious are theconsequences of retrieval on the future accessi-bility of those and other items. Retrieving an item

# 2009 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business

Address correspondence to: Benjamin C. Storm, Department of Psychology, University of Illinois at Chicago, 1007 W Harrison

St (MC 285), Chicago, IL 60607, USA. E-mail: bstorm@uic.edu

The authors thank the following undergraduate research assistants: Dung C. Bui, Hunter Difonso, Mark Fowell, Janet Janow,

Tara A. Jobe, Shree Shah, and John A. Walker. We also greatly appreciate the constructive comments and feedback of Robert A.

Bjork, Elizabeth Ligon Bjork, Michael C. Anderson, and the members of Cogfog at UCLA and Memory Lab at UIC.

MEMORY, 2010, 18 (2), 99�114

http://www.psypress.com/memory DOI:10.1080/09658210903107853

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

from memory does not only make that itemtemporarily accessible, it permanently modifiesthe representation of that item in memory (Bjork,1975). Consequently, information retrieved frommemory becomes more recallable in the futurethan it would have been otherwise, and to anextent beyond that which would have occurredgiven extra study (see e.g., Roediger & Karpicke,2006). Perhaps most surprising is the effectretrieval has on the accessibility of other items.Whereas many models of memory predict thatnon-retrieved information should become moreaccessible following the retrieval of related in-formation (see e.g., Anderson, 1983; Loftus,1973), a burgeoning body of work on retrieval-induced forgetting suggests that non-retrievedinformation may actually become less recallable(Anderson, Bjork, & Bjork, 1994).

The majority of research on retrieval-inducedforgetting has employed the retrieval practiceparadigm, which involves three main phases:study, retrieval practice, and a final test (seeAnderson et al., 1994). In the study phaseparticipants study a list of items, often category/exemplar pairs (e.g., fruit: lemon, profession:dentist, fruit: plum, profession: accountant), pre-sented one at a time and in an interspersed order.Participants then retrieve half of the exemplarsfrom half of the categories via guided retrievalpractice (e.g., fruit: le_____). Practised exemplarsare referred to as Rp� items (e.g., lemon), non-practised exemplars from practised categories arereferred to as Rp� items (e.g., plum), andexemplars from non-practised categories arereferred to as Nrp items (e.g., dentist, accountant).After a brief delay (5�20 minutes), a memory testis administered. Not surprisingly, Rp� items arebetter recalled than are both Rp� and Nrp items.The more surprising finding is that Rp� items arerecalled less well than Nrp items. It is thisdifference between Rp� and Nrp items that isreferred to as retrieval-induced forgetting.

Retrieval-induced forgetting has been shownto be a robust and general phenomenon. Not onlyis it found using semantically associated catego-rical information, but for visual scenes and eventnarratives in eye-witness memory (e.g., MacLeod,2002; Shaw, Bjork, & Handal, 1995), visuo-spatialmaterials (e.g., Ciranni & Shimamura, 1999), factlearning (e.g., Anderson & Bell, 2001; Carroll,Campbell-Ratcliffe, Murnane, & Perfect, 2007),mathematic computations (e.g., Phenix &Campbell, 2004), autobiographical memories(e.g., Barnier, Hung, & Conway, 2004), and social

information (e.g., Storm, Bjork, & Bjork, 2005).Whereas the empirical phenomenon of retrie-val-induced forgetting is well accepted, themechanism thought to underlie retrieval-inducedforgetting remains much more controversial.

The idea that retrieval acts as a self-limitingprocess is not new to memory research (Roediger,1978). Retrieval strengthens the items that areretrieved, and such strengthening may block theretrieval of other items associated with the sameretrieval cue. However, recent evidence suggeststhat retrieval-induced forgetting may in someinstances result from active inhibitory processesthat function to resolve interference. In order tooperate adaptively, a memory system must havesome means of selecting target items in the faceof interference from competing items. For exam-ple, there may be strong yet contextually inap-propriate items that are activated by a givenretrieval cue, and it is essential that there be somemeans of bypassing or attenuating the interfer-ence caused by those competing items. Andersonand colleagues have argued that one way in whichthis type of competition can be resolved isthrough inhibition (Anderson, 2003, 2005; Ander-son, Bjork, & Bjork, 1994; Bjork, Bjork, &Anderson, 1998). In the retrieval practiceparadigm the retrieval cues presented duringretrieval practice may activate non-target items(Rp� items) in addition to target items (Rp�items), creating competition, and requiring thatthe non-target items be selected against, orinhibited. As a consequence of this inhibition,non-practised items become less accessible in thefuture than they would have been otherwise.

Researchers who doubt the role of inhibition inretrieval-induced forgetting often argue thatit can be explained more parsimoniously byprocesses of associative interference. Retrievalpractice strengthens the association between thepractised item and its associated retrieval cue, andby strengthening a subset of items associated witha common retrieval cue, non-practised items willsuffer from response competition at the time oftest, therefore becoming less accessible given thatsame retrieval cue. One problem with this ex-planation is that retrieval-induced forgetting isalso observed when independent cues are em-ployed (Anderson & Spellman, 1995). That is, notonly is the recall of Rp� items impaired whenparticipants are tested using the same categorycues, but also when they are tested using novel,independent, category cues. Cue-independentretrieval-induced forgetting has been replicated

100 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

many times (e.g., Anderson & Bell, 2001; Aslan,Bauml, & Pastotter, 2007; Levy, McVeigh, Marful,& Anderson, 2007; MacLeod & Saunders, 2005;Shivde & Anderson, 2001), providing strongevidence against interference-based accounts ofretrieval-induced forgetting. Recently, however,some researchers have questioned the validity ofthe cue-independent procedure and have failed tofind evidence of cue-independence (see, e.g.,Camp, Pecher, & Schmidt, 2005, 2007; Camp,Pecher, Schmidt, & Zeelenberg, in press; Perfectet al., 2004; Williams & Zacks, 2001), making itincreasingly important to develop other means oftesting the competing accounts of retrieval-in-duced forgetting. Two other observations thatappear to support the inhibitory account arestrength-independence and competition-depen-dence.

Strength-independence refers to the observa-tion that the extent to which Rp� items areforgotten is independent of the extent to whichRp� items are strengthened. According to inter-ference-based accounts, recall for Rp� items isimpaired because the strengthened Rp� itemsinterfere at the time of test. Thus forgettingoccurs because other items are strengthened.Research has shown, however, that strengtheningother items is not sufficient to cause retrieval-induced forgetting (e.g., Anderson, Bjork, &Bjork, 2000; Bauml, 2002). For example, whenparticipants re-study Rp� items rather thanretrieve them during what would typically bethe retrieval practice phase, the later recall of Rp�items is not impaired, even though the Rp�items are strengthened to an equivalent extent.If retrieval-induced forgetting is purely the con-sequence of strengthening the Rp� items,strengthening those items should cause forgettingeven if the strengthening occurs as a result ofadditional study.

Competition-dependence refers to the obser-vation that the extent to which an item causescompetition during retrieval practice determinesthe extent to which that item suffers fromretrieval-induced forgetting (e.g., Anderson,et al., 1994; Shivde & Anderson, 2001; Storm,Bjork, & Bjork, 2007). Anderson et al. (1994;Exp. 3), for example, found that exemplars ofhigh taxonomic strength (e.g., orange and banana)suffer from retrieval-induced forgetting to agreater extent than exemplars of low taxonomicstrength (e.g., mango and guava). Items weaklyassociated with the retrieval practice cues are lesslikely to create competition, and are therefore

less likely to be targeted by inhibitory control.Neither strength-independence nor competition-dependence can be easily accounted for bynon-inhibitory accounts of retrieval-inducedforgetting.

RETRIEVAL SUCCESS ANDRETRIEVAL-INDUCED FORGETTING

Researchers have typically assumed that retrievalmust be successful for retrieval-induced forget-ting to occur, and they have been very careful todesign retrieval-induced forgetting experimentsin such a way so as to ensure high levels ofretrieval practice success. In fact, Andersonet al.’s (1994) original study utilised an expandingschedule of trials in part to ensure participantswere successful during retrieval practice (seeLandauer & Bjork, 1978). However, accordingto the inhibitory account, high levels of successmay not be necessary. First, because inhibitory-based effects of retrieval-induced forgetting arepresumably strength-independent, failing toretrieve*and therefore failing to strengthenitems during retrieval practice*should not dimin-ish the amount of forgetting that occurs. Second,and more importantly, the crux of the inhibitoryaccount is that inhibition serves to resolvecompetition. The term retrieval-induced forget-ting is a bit misfortunate in this sense, because itseems to imply that the retrieval itself is causingforgetting. Instead, it is the competition thatarises during retrieval that creates the need forinhibition. And if inhibitory processes are elicitedto resolve competition during a retrieval attempt,the consequences of those inhibitory processesshould occur regardless of whether the retrievaleventually succeeds. Importantly, the interferenceaccount makes a very different prediction. Therate of success during retrieval practice shoulddetermine the extent to which practised items arestrengthened*which, in turn, should determinethe extent to which non-practised items sufferfrom interference on the final test. Thus, theinterference account predicts that the magnitudeof the retrieval-induced forgetting effect shouldincrease with retrieval practice success.

To date only one study has demonstrated thatretrieval success is not a necessary condition forretrieval-induced forgetting (Storm, Bjork, Bjork,& Nestojko, 2006). Storm and colleagues hadparticipants study a list of category/exemplarpairs and then engage in retrieval practice

RETRIEVAL SUCCESS AND INHIBITION 101

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

consisting of category-plus-two-letter stem cuesthat either did or did not represent the initialletters of any exemplar associated with thatcategory. For half of the cues, retrieval practicewas possible (e.g., Fruit: or____; orange); for theother half of the cues, retrieval practice wasimpossible (e.g., Weapons: wo____; no suchweapon). This manipulation effectively dictatedwhether retrieval practice could or could not besuccessful. Retrieval-induced forgetting was ob-served in both cases and, importantly, the size ofthe effect did not differ for exemplars associatedwith categories receiving possible retrieval prac-tice versus exemplars associated with categoriesreceiving impossible retrieval practice. Theseresults suggest that retrieval-induced forgettingcan be more accurately characterised as retrieval-attempt-induced forgetting.

The primary goal of the current study was toprovide additional evidence that retrieval successis not a necessary condition for retrieval-inducedforgetting. Memory researchers have cited andinterpreted this finding to be compelling evidencefavouring the inhibitory account. Furthermore,replicating the effect is especially important dueto methodological concerns regarding the finaltest employed by Storm et al. (2006). Ratherthan test categories or items independently onthe final test, participants were shown all of thecategory-plus-letter-stem cues simultaneously ona single piece of paper. It is conceivable that thisprocedure gave participants too much discretionover which items to attempt to recall. Becauseparticipants experienced difficulty in retrievingitems from the impossible categories duringretrieval practice, they may have been biasedagainst attempting to retrieve items from thosesame categories on the final test. Experiments 1and 2 attempted to replicate the retrieval-inducedforgetting effect with impossible retrieval practiceemploying final tests less susceptible to this bias.

RETRIEVAL PRACTICE TIME ANDRETRIEVAL-INDUCED FORGETTING

There is reason to believe that impossible retrievalpractice has the potential, under certain circum-stances, to lead to even more retrieval-inducedforgetting than possible retrieval practice. Whenretrieval succeeds, the retrieval attempt isfunctionally complete, and the person no longerneeds to engage in processes of retrieval. How-ever, when retrieval fails, the retrieval attempt

continues, thereby forcing the person to continueto engage in processes of retrieval. And if theretrieval attempt is extended so should the prob-ability of an item in memory causing competition,thereby increasing the probability that such anitem will be targeted by inhibition and suffer fromretrieval-induced forgetting.

In typical studies of retrieval-induced forget-ting it is impossible to examine the relationshipbetween retrieval practice time and retrieval-induced forgetting because the time participantstake to retrieve an item during retrieval practiceis determined by whether or not participants areable to retrieve that item. Given this methodolo-gical impasse, it is not surprising that the temporaldynamics of inhibition during retrieval practicehave yet to be explored. However, by employingthe impossible retrieval practice procedure it ispossible to manipulate the amount of time inwhich participants engage in retrieval. Impossibleretrieval practice guarantees that participants willfail to recall a viable item associated with thepresented retrieval cue, thereby ensuring thatparticipants will attempt to recall an item untilthe trial is complete.

In Experiments 3 and 4 participants werepresented two exemplars associated with a givencategory and then shown a retrieval practice cueassociated with that category. One-third of theretrieval cues consisted of cues associated withactual category exemplars (possible retrievalpractice); the other two-thirds of the retrievalcues consisted of cues not associated with anycategory exemplar (impossible retrieval practice).We then manipulated the amount of time parti-cipants were given to respond to each retrievalpractice cue*4, 8, or 12 seconds. On other trialsparticipants were not given any retrieval practicecue, thus providing a baseline (Nrp items). Aftera brief delay participants were given a surprisetest on all of the earlier studied items.

It is not immediately obvious what pattern ofresults to expect. A straightforward predictionmight be that retrieval-induced forgetting willincrease linearly as a function of the amount oftime participants spend during retrieval practice.That is, 4 seconds should result in a certainamount of forgetting, 8 seconds should result inmore, and 12 seconds in still more. Anotherpossibility is that there will be a non-monotonicincrease in the amount of forgetting during aretrieval attempt. For example, there is evidencethat initial retrieval practice trials lead to facil-itation and that impairment is only observed after

102 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

repeated trials (e.g., Shivde & Anderson, 2001;Levy et al., 2007). Likewise, it is possible thatrecall will be facilitated after 4 seconds, and then,as inhibition continues to exert its influence,become increasingly impaired after 8 and12 seconds. Finally, it is also possible that mostof the forgetting will take place as a consequenceof the first 4 seconds of retrieval practice, and thatlittle, if any, forgetting will take place thereafter.Regardless of the pattern of results that emerges,exploring the temporal dynamics of forgettingduring retrieval will have important implicationsfor both future investigations and theoreticalaccounts of retrieval-induced forgetting.

EXPERIMENT 1

The first experiment replicated Storm et al.’s(2006) impossible retrieval practice paradigmwith one important difference*rather than si-multaneously presenting all of the cues during thefinal test, we displayed each cue individually onthe computer screen. In doing so we sought toensure that participants spent an equivalentamount of time attempting to recall each itemon the final test. A second goal of the study was toexplore whether participants who demonstratehigh levels of retrieval-induced forgetting owingto impossible retrieval practice also demonstratehigh levels of retrieval-induced forgetting owingto possible retrieval practice. A significant corre-lation should be observed if common mechanismsunderlie the forgetting caused by both forms ofretrieval practice.

Method

Participants

A total of 56 undergraduate students (25 menand 31 women, mean age�20.6 years, all fluentEnglish speakers) from the University of Califor-nia, Los Angeles received credit towards anintroductory psychology course for their partici-pation.

Materials

Study lists. The study lists consisted of the same48 category/exemplar pairs (six members of eightcategories) used by Storm et al. (2006). Forcounterbalancing purposes, each category wasdivided into two subsets (A and B) of threeexemplars. During study, participants were

exposed to three exemplars from each of theeight categories via semi-random interleavedpresentation. Half of the participants studied setA, and half of the participants studied set B.

Retrieval practice. Four of the eight categoriesreceived retrieval practice, with the particularcategories receiving practice counterbalancedacross participants. For two of the practisedcategories, participants were guided to retrievethree exemplars that had not been presentedduring their original study phase via category-plus-two-letter-stem retrieval cues (possible re-trieval practice). For the other two categories,participants were shown a category name as wellas a realistic two-letter stem, but the two lettersdid not begin an actual exemplar associated withthat category (impossible retrieval practice).None of the category/exemplar pairs appearedin both the studied list and the retrieval practicetask.

A total of 12 category-plus-stem cues (6possible and 6 impossible) were presented to anindividual participant for practice three timeseach in a semi-random interleaved order. Theparticular categories that were associated withpossible and impossible retrieval practice werecounterbalanced across participants, and the ma-terials were controlled such that no two exem-plars shared the same initial two letters and suchthat no two exemplars in the same categoryshared the same initial letter.

Procedure

The experiment consisted of three phases:study, retrieval practice, and final test. Duringthe study phase, 24 category/exemplar word pairswere presented one at a time on the computer for5 seconds each. The order of presentation wasrandom with the constraint that no two consecu-tive pairs were from the same category. Immedi-ately following study, the retrieval practice phasebegan with category-plus-stem cues appearing onthe screen for 5 seconds each. Participants wereinstructed to write down the particular exemplarthat completed each two-letter stem in theirresponse packet. They were told that there wouldbe repetitions and that the pairs might or mightnot come from the list they had just studied.

The final test was administered after anunrelated 10-minute distractor task. Recall ofthe same 24 category/exemplar pairs that hadbeen presented during the study phase was testedusing a category-plus-one-letter-stem cued recall

RETRIEVAL SUCCESS AND INHIBITION 103

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

test (e.g., fruit: l_____ for fruit: lemon). Becausenone of the pairs had appeared during retrievalpractice, there were no Rp� items on the finaltest. The category-plus-stem cues were presentedone at a time on the computer screen for 5seconds each. The cues were presented in a semi-random interleaved order with the average posi-tion controlled across all experimental conditions.

Results

Retrieval practice responses

Participants generated responses on 50%(SD�29%) of the possible retrieval practicetrials and 16% (SD�20%) of the impossibleretrieval practice trials. As was observed byStorm et al. (2006), responses during impossibleretrieval practice often reflected incorrect re-sponses, and other times, accurate and highlycreative responses.

Final recall performance

The mean recall proportions for exemplars as afunction of their being Rp� or Nrp items and fromcategories that had received possible or impos-sible retrieval practice are presented in Table 1.When these data were analysed in a 2 (Rp� vsNrp)�2 (possible vs impossible) repeated mea-sures Analysis of Variance (ANOVA), a signifi-cant effect of retrieval-induced forgettingemerged, with Rp� items being recalled signifi-cantly less often (M�.67, SE�.02) than theirNrp counterparts (M�.73, SE�.02), F(1, 55)�4.00, MSE�.19, p�.05. The effect did notinteract with retrieval practice condition,F(1, 55)B1. Numerically, however, slightly more

retrieval-induced forgetting was found following

impossible retrieval practice (M for Rp��.67,

SE�.03; M for Nrp�.74, SE�.03) than possible

retrieval practice (M for Rp��.66, SE�.03; M

for Nrp�.71, SE�.03).A significant correlation was observed

between the retrieval-induced forgetting effects

owing to possible and impossible retrieval

practice (r�.37, pB.01). This relationship was

also explored via a median split analysis. Partici-

pants were ranked in each counterbalancing

condition according to the amount of retrieval-

induced forgetting observed in the impossible

condition and then assigned to high and low

impossible retrieval-induced forgetting groups.

Participants in the high impossible group demon-

strated an average retrieval-induced forgetting

effect of .21 (SD�.22), whereas participants in

the low impossible group demonstrated an aver-

age retrieval-induced facilitation effect of .08

(SD�.22). When final recall performance in the

possible condition was analysed, a 2 (high vs low

impossible retrieval-induced forgetting)�2 (Rp�vs Nrp) mixed design ANOVA revealed a sig-

nificant interaction, F(1, 54)�11.64, pB.001.

Participants in the high impossible group recalled

.60 (SE�.04) and .76 (SE�.04) of the Rp� and

Nrp items in the possible condition, respectively,

whereas participants in the low impossible group

recalled .73 (SE�.04) and .67 (SE�.04) of the

Rp� and Nrp items in the possible condition,

respectively. Thus, individuals who demonstrated

a larger effect of retrieval-induced forgetting

owing to impossible retrieval practice also de-

monstrated a larger effect of retrieval-induced

forgetting owing to possible retrieval practice.

EXPERIMENT 2

The results of Experiment 1 replicate those of

Storm et al. (2006), demonstrating that retrieval

success is not a necessary condition for retrieval-

induced forgetting. Impossible retrieval practice

led to just as much forgetting as possible retrieval

practice. In Experiment 2 we sought to replicate

this finding using another final test commonly

employed in studies of retrieval-induced forget-

ting: category cued recall. Rather than test each

exemplar independently, participants were pro-

vided a category cue and asked to recall the

studied exemplars associated with that category.

TABLE 1

Mean recall rates

Item type

Experiment: Type of final test

Retrieval practice Rp� Nrp

Experiment 1: Category-plus-stem cued recall

Possible .66 (.03) .71 (.03)

Impossible .67 (.03) .74 (.03)

Experiment 2: Category cued recall

Possible .50 (.02) .58 (.02)

Impossible .50 (.02) .57 (.02)

Mean recall rates for Rp� versus Nrp items from categories

receiving possible versus impossible retrieval practice in

Experiments 1 and 2 (SE in parentheses).

104 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

Method

Participants

A total of 119 undergraduate students (38 menand 81 women, mean age�20.2 years) from theUniversity of California, Los Angeles receivedcredit towards an introductory psychology coursefor their participation.

Materials and procedure

The only change made from Experiment 1 wasthe final test. Participants were given a categorycued recall test for each of the studied categories.The test involved presenting each of the eightcategory cues for 10 seconds and having theparticipants attempt to recall the three studiedexemplars associated with each category. Partici-pants were warned that they were to only recallthe exemplars that they had learned during theinitial study phase.

Results

Retrieval practice responses

Generation rates during retrieval practice werevery similar to those observed in Experiment 1.Participants generated a response on 48% (SD�31%) and 13% (SD�19%) of the possible andimpossible retrieval practice trials, respectively.

Final recall performance

The mean correct recall proportions forexemplars as a function of their being Rp� orNrp items and from categories that receivedpossible or impossible retrieval practice are pre-sented in Table 1. The data were analysed in a 2(Rp� vs Nrp)�2 (possible vs impossible)repeated measures ANOVA, and once again asignificant effect of retrieval-induced forgettingwas observed, with Rp� items being recalledsignificantly less often (M�.50, SE�.02) thanNrp items (M�.58, SE�.02), F(1, 118)�13.56,MSE�.66, pB.001. And, once again, the effectdid not interact with whether retrieval practicewas possible or impossible, F(1, 118)B1.

Although somewhat weaker than thatobserved Experiment 1, a significant correlationwas found between the effects of retrieval-induced forgetting owing to possible and impos-sible retrieval practice (r�.25, pB.01). As in

Experiment 1, we analysed these data via mediansplit within each counterbalancing condition.Participants in the high impossible retrieval-induced forgetting group, who demonstrated anaverage retrieval-induced forgetting effect of .24(SD�.16) in the impossible condition, recalled.48 (SE�.03) and .59 (SE�.03) of the Rp� andNrp items in the possible condition, respectively.Participants in the low impossible retrieval-in-duced forgetting group, who demonstrated anaverage retrieval-induced facilitation effect of.12 (SD�.21), recalled .52 (SE�.03) and .58(SE�.03) of the Rp� and Nrp items in thepossible condition, respectively. A 2 (high vs lowimpossible retrieval-induced forgetting)�2 (Rp�vs Nrp) mixed design ANOVA confirmed that theinteraction was not significant, F(1, 117)B1.

EXPERIMENT 3

The results of Experiments 1 and 2 providefurther evidence that retrieval success is not anecessary condition for retrieval-induced forget-ting. The attempt to retrieve an item frommemory, even if unsuccessful, can cause theforgetting of other items in memory. This findingfalls naturally out of the inhibitory account ofretrieval-induced forgetting and is problematicfor other accounts, such as interference or block-ing. The purpose of Experiment 3 was to explorethe relationship between the amount of timeparticipants spend attempting to retrieve anitem during retrieval practice and the amount ofretrieval-induced forgetting that results. Undernormal conditions it is impossible to manipulatethe time in which participants take to retrieve anitem from memory. Experimenters can provide aspecific amount of time for participants torespond to a cue, but there is no way tomanipulate the time in which it will take partici-pants to retrieve a given response. In fact,because the level of performance during retrievalpractice is typically very high, participants areundoubtedly retrieving answers well before theallotted time has expired. However, by employingimpossible retrieval practice we are able toeffectively manipulate the amount of time thatparticipants attempt to retrieve an item on agiven trial.

In Experiment 3 participants were first showntwo exemplars from a given category and then,

RETRIEVAL SUCCESS AND INHIBITION 105

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

immediately afterwards, a category-plus-two-let-ter-stem retrieval cue from the same category.This retrieval cue remained on the screen for 4, 8,or 12 seconds. Critically, two-thirds of the retrie-val cues were impossible, and on such trialsparticipants spent the entire allotted timeattempting to retrieve an appropriate exemplar.The other third of the retrieval cues werepossible. The possible cues were selected to bevery easy in order to guarantee that the partici-pants would be able to succeed on at least a fewof the retrieval practice trials. Because possibleretrieval practice was designed to be extremelyeasy, and thus, non-competitive, we did not expectmuch, if any, retrieval-induced forgetting tooccur. As such, our analysis will focus exclusivelyon the impossible trials and, specifically, on howmanipulating retrieval practice time influencesthe extent to which studied items suffer retrieval-induced forgetting.

It is worth pointing out how the paradigm inExperiment 3 differs from that which is typicallyused in studies of retrieval-induced forgetting. Inmost experiments participants undergo an entireretrieval practice phase during which they re-trieve several members from several categories,several times each. Rarely do experimentersmeasure retrieval-induced forgetting after a sin-gle retrieval practice trial. In consideration of thisdifference, it is unclear whether forgetting will beobserved at all, let alone whether the amount offorgetting will differ depending on the length ofthe retrieval practice trial. However, by havingparticipants engage in retrieval practice immedi-ately after studying related items we hoped toensure that the studied items would interfere withretrieval practice, and therefore be targeted byinhibition.

Method

Participants

A total of 104 undergraduate students (17 menand 87 women, mean age�18.8 years) from theUniversity of Illinois at Chicago participated forcourse credit in an introductory psychologycourse.

Materials

Studied lists. A total of 24 lists of two exemplarswere presented for participants to study. Each listconsisted of two high-frequency exemplars fromthe same category, presented vertically on the

centre of the computer screen. Eight of thecategories were designated to the possible condi-tion and were always associated with possibleretrieval practice. The other 16 categories weredesignated to the impossible condition, and werealways associated with impossible retrieval prac-tice. The possible and impossible categories weredivided into four sets of two and four categories,respectively. Each set consisted of categories forwhich participants would receive 0, 4, 8, or 12seconds of retrieval practice following study. Theparticular set associated with each retrieval prac-tice condition was counterbalanced across parti-cipants. Exemplars from categories associatedwith 0 seconds of retrieval practice will bereferred to as Nrp items, and exemplars fromcategories associated with 4, 8, or 12 seconds ofretrieval practice will be referred to as Rp� items.

Retrieval practice. A single category-plus-two-letter-stem cue was created for each of the eightcategories in the possible condition. Because wewanted to ensure that participants were able tosucceed during possible trials, each cue wasassociated with a high-frequency exemplar fromits respective category. For the 16 categories inthe impossible condition, a single category-plus-two-letter-stem cue was created that was notassociated with any exemplar from that category.None of the stems in the possible or impossiblecondition began with the same letter asan exemplar that was studied from the samecategory.

Procedure

The experiment consisted of two phases: study/retrieval practice and test. The study/retrievalpractice phase consisted of 24 trials, during whichlists of two category/exemplar pairs were pre-sented for 8 seconds. On 18 of the trials (12impossible and 6 possible), participants wereshown a category-plus-two-letter-stem cue imme-diately after study and asked to generate anassociated exemplar. Participants were warnedthat none of the answers during retrieval practicewould consist of exemplars previously studied.Participants were also warned that they would betested on the studied exemplars later in theexperiment.

The amount of time participants were given oneach retrieval practice trial was manipulatedwithin-participants. On six trials participantswere given 4 seconds, on another six trialsparticipants were given 8 seconds, and on another

106 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

six trials participants were given 12 seconds. The

other 6 of the 24 trials served as baseline, during

which no retrieval practice cue was shown. On

these trials participants immediately proceeded to

the next trial. The order of the trials was

determined via blocked randomisation, and par-

ticipants had no way to predict the amount of

retrieval practice time*or if there would be

retrieval practice*on a given trial. Once partici-

pants generated an exemplar in the possible

condition, the experimenter pressed a button to

begin the next trial. In the impossible condition

participants were always given the entire allotted

time for each retrieval practice trial. If a partici-

pant generated a response they were told that it

was not the correct response, and to continue

trying.After completing all 24 study/retrieval practice

trials, participants engaged in an unrelated 5-

minute distractor task, followed by a final test.

Half of the participants were given a category

cued recall test in which each of the 24 categories

were presented for 8 seconds, and participants

were asked to recall the two studied exemplars

associated with each category. The other half of

the participants were given a category-plus-one-

letter-stem cued recall test in which 48 cues were

presented for 4 seconds each and asked to recall

the studied exemplar associated with each cue.

The category cues and category-plus-one-letter-

stem cues were presented in a semi-random

interleaved order with the average position con-

trolled across all experimental conditions.

Results

Retrieval practice responses

Participants generated correct exemplars on90% (SD�11%) of the possible retrieval practicetrials. During impossible retrieval practice parti-cipants generated a response (regardless ofwhether it fit the appropriate cues) on 6%(SD�8%) of the trials. It should be noted that40% of the participants failed to give a singleresponse during impossible retrieval practice, andthat only 11% of the participants gave more thantwo responses.

Final recall performance

Mean recall performance for exemplars in theimpossible condition as a function of retrievalpractice time and type of final test are shown inFigure 1. These data were analysed in a 4 (0 vs 4vs 8 vs 12 seconds)�2 (category cued recall vscategory-plus-stem cued recall) mixed designANOVA, with type of test serving as a between-participants variable. A significant main effect ofretrieval practice time was observed, F(1, 102)�4.05, MSE�.13, pB.01, with a mean proportionof .48 (SE�.02), .42 (SE�.02), .40 (SE�.02),and .40 (SE�.02), exemplars being recalled inthe 0, 4, 8, and 12 second conditions, respectively.A planned t-test comparing the Nrp items withthe average of the Rp� items (4, 8, and 12 s)confirmed that significant retrieval-inducedforgetting had occurred, t(103)�3.12, pB.01.Additional t-tests confirmed that significant

Figure 1. Mean final recall performance for non-practised exemplars in Experiments 3 and 4 as a function of type of final test and

duration of retrieval practice trial.

RETRIEVAL SUCCESS AND INHIBITION 107

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

retrieval-induced forgetting occurred in eachretrieval practice time condition as well. How-ever, recall performance did not differ betweenRp� items associated with different amounts ofretrieval practice time, F(1, 102)B1. Also, as canbe seen in Figure 1, although the level of recallwas different overall, the pattern of results wasnearly identical for participants in the two typesof testing conditions. More specifically, the effectof retrieval practice time did not interact withtype of test, F(3, 375)B1.

EXPERIMENT 4

The results of Experiment 3 suggest that even 4seconds of a single impossible retrieval practicetrial is sufficient to cause retrieval-induced for-getting. However, neither 8 seconds nor 12seconds of retrieval practice increased the mag-nitude of the effect. In Experiment 4 we at-tempted to replicate this finding under slightlydifferent conditions. Participants in the previousexperiment were pre-warned that none of theanswers during retrieval practice consisted ofitems that they had previously studied*and, infact, none of the retrieval practice cues wasassociated with studied items. The procedurewas designed in that way to ensure that partici-pants would not try to actively remember thestudied items during retrieval practice. Hadparticipants believed that a retrieval practicecue could, on occasion, be completed with apreviously studied item, a very different patternof results might have emerged.

Warning participants that retrieval practicecues may be associated with earlier studied itemsmight be expected to increase the likelihood thatstudied items are explicitly rehearsed or implicitlyactivated during retrieval practice. Consequently,it is possible that final recall performance for thestudied items would be facilitated. It is unclear,however, when such facilitation would occur,should it occur. One possibility is that participantswill actively think about the studied exemplarsduring the initial seconds of retrieval practice, andthen, only after realising that neither of thestudied items fit the cue, target retrieval effortselsewhere*thereby leading to inhibition, andconsequently, retrieval-induced forgetting. An-other possibility is that the studied items will betargeted by inhibition initially, but that as parti-cipants continue to fail in their retrieval attempts,they will increasingly revisit those items as the

trial progresses. To test these possibilities, parti-cipants in Experiment 4 were warned that someof the answers during retrieval practice wouldcome from exemplars studied immediately prior,and half of the possible retrieval practice trialswere altered so as to conform to these instruc-tions.

Method

Participants

A total of 75 undergraduate students (16 menand 69 women, mean age�18.5 years) from theUniversity of Illinois at Chicago participated forcourse credit in an introductory psychologycourse.

Materials and procedure

Experiment 4 was identical to Experiment 3with two important exceptions. First, all of theparticipants were given a category cued recallfinal test at the conclusion of the experiment.Second, participants were warned that some ofthe retrieval cues during retrieval practice wouldbe associated with items that had been studiedimmediately prior. And, in fact, on three of thesix possible study/retrieval practice trials, the cueswere altered as such. All aspects of the impossibletrials were the same as in Experiment 3.

Results

Retrieval practice responses

Participants generated correct exemplars on89% (SD�15%) of the possible retrieval practicetrials. During impossible retrieval practice, parti-cipants generated a response on 11% (SD�14%)of the trials.

Final recall performance

The mean correct recall proportions for ex-emplars in the impossible condition, as a functionof retrieval practice time (0 vs 4 vs 8 vs 12 s), wasanalysed using a repeated measures ANOVA. Ascan be seen in Figure 1, a significant main effectof retrieval practice time was observed, F(1,74)�5.87, MSE�.18, pB.001. A mean propor-tion of .38 (SE�.03), .30 (SE�.02), .32 (SE�.02), and .26 (SE�.02), exemplars were recalledin the 0-, 4-, 8-, and 12-second conditions,respectively. A planned t-test comparing the Nrpitems with the average of the Rp� items (4, 8, and

108 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

12 s) confirmed that the retrieval-induced forget-ting effect was significant, t(74)�3.26, pB.01,and additional t-tests confirmed that significantretrieval-induced forgetting occurred in eachretrieval practice condition as well. Althoughthe pattern of results was similar to that whichwas observed in Experiment 3, there were someimportant differences. Specifically, recall initiallydropped after 4 seconds of retrieval practice,increased slightly after 8 seconds of retrievalpractice, and then dropped again after 12 secondsof retrieval practice. Paired samples t-tests indi-cated significant forgetting effects after the initial4 seconds, t(74)�2.53, pB.05, and after the final4 seconds, t(74)�2.20, pB.05. The facilitationeffect observed between 4 and 8 seconds ofretrieval practice was not significant, t(74)B1,p�.05. Within-participants contrasts confirmedthat both linear, F(1, 74)�11.16, MSE�.42, pB.05, and cubic, F(1, 74)�4.28, MSE�.13, pB.05,trends were statistically significant.

GENERAL DISCUSSION

The dynamics underlying human memory retrie-val are not as straightforward as they might seem.In order to retrieve a target item from memory,one must have the capacity to suppress, or silence,non-target items in memory. Researchers havesuggested that one way in which this is accom-plished is through retrieval inhibition (Bjorket al., 1998; Bjork, 1989) or the executive controlprocesses of inhibition (Anderson, 2003, 2005).Inappropriate or unwanted items in memorybecome activated by a retrieval cue, but areinhibited in order to facilitate the retrieval of atarget item. In this sense, retrieval-induced for-getting reflects inhibitory processes that act as ameans of facilitating retrieval, not as a conse-quence of retrieval. Consistent with this idea, thecurrent investigation (across four experimentsand 354 participants) clearly and convincinglydemonstrates that even unsuccessful retrievalattempts can cause retrieval-induced forgetting.

Experiment 1 sought to replicate Storm et al.’s(2006) impossible retrieval practice paradigmusing an item-by-item final cued recall test. Wefelt it was important to replicate the originaleffect using a different final test in order to ruleout a methodological concern. The final testemployed by Storm et al. involved presentingall of the test cues on a single sheet of paper, andgiving participants 3 minutes to recall studied

exemplars associated with those cues. Owing tothe prior difficulty associated with the impossiblecategories during retrieval practice, it is possiblethat participants were biased to avoid suchcategories in favour of categories that did notreceive impossible retrieval practice. Thus, it ispossible that the impaired recall observed in theimpossible condition was not due to inhibition,but merely biased retrieval at the time of test. InExperiment 1 we employed an item-by-item finalrecall test that ensured that participants spent anequivalent amount of time attempting to recallan exemplar associated with each cue. Thepattern of results was nearly identical to thatwhich was observed by Storm et al. (2006).Significant retrieval-induced forgetting wasfound, and critically, the effect did not differdepending on whether retrieval practice waspossible or impossible. The same results emergedwhen we employed a category cued recall test inExperiment 2.

Both possible and impossible retrieval practiceare believed to cause forgetting through processesof inhibition. If this is the case, and if there existdifferences in inhibitory capacity between indivi-duals, then individuals who demonstrate largereffects of retrieval-induced forgetting fromimpossible retrieval practice should also demon-strate larger effects of retrieval-induced forgettingfrom possible retrieval practice. The resultsof Experiment 1 demonstrate just such a relation-ship. Participants demonstrating larger retrieval-induced forgetting effects in the impossiblecondition also demonstrated larger effects ofretrieval-induced forgetting in the possible condi-tion. These results suggest that there are individualdifferences in inhibition, and that the same me-chanisms underlying forgetting owing to possibleretrieval practice may underlie forgetting owing toimpossible retrieval practice. Although also sig-nificant, a somewhat weaker relationship wasobserved in Experiment 2. It is possible thatindividual differences in inhibition are most read-ily observed when the effects of interference arereduced during the final test (i.e., using a category-plus-stem-cued recall test instead of a category-cued recall test). Such a possibility would beconsistent with arguments made by Andersonand Levy (2007) regarding the correlated costsand benefits of inhibition. If interference is moretightly controlled on the final test, individualdifferences in retrieval-induced forgetting may bemore likely to result from individual differences inthe executive control processes of inhibition.

RETRIEVAL SUCCESS AND INHIBITION 109

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

Experiments 3 and 4 explored the time-courseof forgetting during a single retrieval practicetrial. In most studies of retrieval-induced forget-ting participants are given between 4 and 10seconds to retrieve Rp� items during retrievalpractice. This time is often sufficient for partici-pants to retrieve the correct exemplar andrespond in writing, or vocally to the experimen-ter. To date, no studies have manipulated theamount of time in which participants engage inretrieval during a single trial. The reason for thisis relatively straightforward: it is impossible tomanipulate the amount of time participants taketo retrieve an item on a given trial. Althoughexperimenters may be able to manipulate theamount of time allotted for each trial, there is noway to dictate when and whether participants willcomplete a retrieval attempt. However, by em-ploying the impossible paradigm, we were able toeffectively manipulate the amount of time parti-cipants attempted to retrieve an item duringretrieval practice. Because participants could notsucceed during retrieval practice, the amount oftime participants spent searching for an itemassociated with each impossible retrieval cuecould be pre-determined.

The pattern of results that emerged fromExperiment 3 was quite intriguing*4 seconds ofretrieval practice led to significant retrieval-in-duced forgetting, but providing additional retrie-val practice time did not increase the magnitudeof the effect. In other words, spending 8 or 12seconds attempting to retrieve an item frommemory did not lead to more retrieval-inducedforgetting than did spending 4 seconds attemptingto retrieve an item from memory. This observa-tion is somewhat surprising given that there arereasons to expect inhibitory control to be neces-sary during the entire duration of a retrievalpractice trial. As such, one might have expectedthe amount of retrieval-induced forgetting toincrease with retrieval practice time. The factthat it does not suggests that inhibition may actprimarily during the initial seconds of a retrievalattempt in order to resolve interference, and thatonce interference is resolved, additional inhibi-tion is either not elicited, or not effective atreducing later recall.

A somewhat different pattern of results wasobserved in Experiment 4. As in Experiment 3,the majority of the retrieval-induced forgettingeffect occurred by 4 seconds. Unlike Experiment3, however, final recall performance varied be-tween the 4-, 8-, and 12-second retrieval practice

conditions*increasing slightly between 4 and 8seconds of retrieval practice, and then droppingsignificantly between 8 and 12 seconds of retrievalpractice. This trend suggests that forgetting oc-curred at two points during the 12-second retrie-val practice trials: first, during the initial 4seconds; and second, during the final 4 seconds.One possible interpretation is that after failing toretrieve an exemplar during the initial seconds ofa trial, participants attempt to revisit the studiedexemplars, and that this re-activation led to anadditional round of inhibition on those items.Such a dynamic would be more likely to occur inExperiment 4 than Experiment 3, as participantswere warned that the answers to some of theretrieval practice trials would consist of pre-viously studied items.

Another difference between the results ofExperiment 3 and Experiment 4 is that the totalamount of retrieval-induced forgetting observedafter 12 seconds of retrieval practice in Experi-ment 4 (12%) was nearly twice that which wasobserved in Experiment 3 (6%). That moreforgetting was observed in Experiment 4 mightseem counterintuitive. After all, it was in Experi-ment 4 that participants were warned that astudied item would be the correct answer on asubset of the retrieval practice trials. It seemsreasonable to assume that participants given sucha warning would explicitly or implicitly activatethe previously studied exemplars, thereby facil-itating, not inhibiting, their later recall. However,the finding that previously studied exemplars aremore impaired under this condition is consistentwith the competition-dependence assumption ofthe inhibitory account of retrieval-induced for-getting (e.g., Anderson et al., 1994; Shivde &Anderson, 2001; Storm et al., 2007). Utilising thelist-method directed forgetting paradigm, Stormet al. (2007) found that to-be-remembered itemssuffered from significantly more retrieval-inducedforgetting than to-be-forgotten items. Warningparticipants that studied items will never beanswers during retrieval practice may preventthose items from competing during retrievalpractice (making them akin to to-be-forgottenitems), and therefore protecting those items frominhibition. Likewise, warning participants thatstudied items may sometimes serve as answersduring retrieval practice may exacerbate theextent to which those items compete duringretrieval practice, and therefore increase theextent to which they are targeted by inhibition.

110 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

Finally, it is worth noting that retrieval-inducedforgetting was observed after only a singleretrieval practice trial. In contrast to our currentresults, three studies have employed a singleretrieval practice trial condition as part of theirdesign (Johnson & Anderson, 2004; Levy et al.,2007; Shivde & Anderson, 2001), and each ofthose studies observed facilitation after a singleretrieval practice trial, followed by an increas-ingly large effect of forgetting following addi-tional retrieval practice trials. Shivde andAnderson (2001) argued that an activation/inhibi-tion balance determines when items will and willnot demonstrate forgetting. Essentially, the re-trieval practice cue first activates informationrelated to the target, and inhibition initiates tocounteract the activation of related, non-targetinformation. It is possible that in some casesinhibition does not immediately outweigh initialactivation and, as a consequence, the competingresponses remain strengthened after only a singleretrieval practice trial. It is not clear why weobserved a different pattern of results here. Inboth Experiments 3 and 4, strong and reliableeffects of retrieval-induced forgetting were ob-served after only a single retrieval practice trial.Clearly multiple retrieval practice trials are notnecessary to cause retrieval-induced forgetting.

Implications for theoretical accountsof retrieval-induced forgetting

The inhibitory account of retrieval-induced for-getting contends that non-practised items areforgotten because they interfere during retrievalpractice, and are therefore targeted by inhibition.Interference-based accounts contend that non-practised items are forgotten because retrievalpractice strengthens practised items, therebycausing them to interfere with the recall of non-practised items at the time of the final test. Thefinding that unsuccessful retrieval practice leadsto the same amount of forgetting as successfulretrieval practice fits nicely with the inhibitoryaccount, and poses serious problems for theinterference account. After all, if nothing isretrieved during retrieval practice, nothing isstrengthened, and therefore nothing should inter-fere with the recall of non-practised items.According to the inhibitory account, inhibitionoccurs during retrieval practice so as to facilitatethe retrieval of practised items, and thus theforgetting that occurs as a consequence of that

inhibition should occur regardless of whetherretrieval practice eventually succeeds or fails. Inthis sense the current findings clearly support thestrength-independent assumption of the inhibi-tory account of retrieval-induced forgetting (e.g.,Anderson et al., 2000; Bauml, 2002).

One might argue that although retrieval mayfail during impossible retrieval practice, surelysomething is strengthened. For example, perhapsparticipants are failing to come up with correctresponses, but they are still coming up with sometype of responses, and those responses might beblocking recall at test. Two arguments can bemade against this possibility. First, participantswho provide more responses during impossibleretrieval practice*regardless of the appropriate-ness of their responses*do not show largereffects of retrieval-induced forgetting than parti-cipants who provide fewer responses duringimpossible retrieval practice. Such a relationshipwas not observed by Storm et al. (2006), and sucha relationship was not observed in any of the fourexperiments reported here. Second, there is everyreason to believe that if participants were toexplicitly retrieve or implicitly activate anyresponses during impossible retrieval practice, itwould be the items that they had just previouslystudied, not items that had not been studied. Assuch, the later recall of non-practised items wouldseem more likely to benefit from impossibleretrieval practice than be impaired by it.

Proponents of the inhibitory account haveoften assumed that retrieval-induced forgettingis retrieval-specific*that studying Rp� itemsduring what is normally retrieval practice, orstrengthening Rp� items in some other way, isnot sufficient to cause retrieval-induced forget-ting (see e.g., Anderson, 2003, 2005; Bjork et al.,1998; Levy & Anderson, 2002). Rather thancontradict this assumption, the current resultssuggest a slight amendment. It is not the retrievalper se that causes inhibition, it is the retrievalattempt that sets the stage for inhibition to occur.Inhibitory processes function to resolve interfer-ence, and such processes appear to be just asactive during unsuccessful retrieval as they areduring successful retrieval.

Recent research examining theoretical ac-counts of retrieval-induced forgetting hasfocused predominately on evidence for andagainst cue-independence. And although therehave been many successful demonstrations ofcue-independence (e.g., Anderson & Bell, 2001;Anderson & Spellman, 1995; Aslan et al., 2007;

RETRIEVAL SUCCESS AND INHIBITION 111

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

Levy et al., 2007; MacLeod & Saunders, 2005;Shivde & Anderson, 2001), there have also beenfailures (Camp et al., 2005, 2007; Perfect, Stark,Tree, Ahmed, & Hutter, 2004; Williams & Zacks,2001). Some have questioned whether indepen-dent cues are even truly independent (see e.g.,Camp, Pecher, Schmidt, & Zeelenberg, in press).However, we feel that evidence supporting theinhibitory account of retrieval-induced forgettingdoes not start and end with cue independence. Infact, the results of the current study*and theresults of other studies demonstrating strength-independence and competition-dependence*provide unique and compelling evidence infavour of the inhibitory account as well. Manyresearchers remain sceptical of inhibition’s rolein human memory (see e.g., MacLeod, Dodd,Sheard, Wilson, & Bibi, 2003), making it all themore important to develop multiple and com-plementary lines of evidence supporting thatrole.

Finally, the current findings have importantimplications for understanding the time-course offorgetting during a single retrieval attempt. Untilnow, researchers have been unable to manipulatethe amount of time in which participants attemptto retrieve an item during retrieval practice. Theresults of Experiment 3 suggest a negativelyaccelerated function such that inhibition mayhave a substantial impact during the initialseconds of a retrieval practice trial, and thenincreasingly less of an impact as the trial con-tinues. Several possibilities may account for thisfunction. One possibility is that participantsengage in a less and less effortful retrieval searchas a given trial goes on. In other words, they startto give up. Another possibility is that the inhibi-tion that takes place during the initial seconds ofa trial effectively inhibits the items that are goingto interfere, thereby eliminating the need forinhibition during subsequent seconds of that trial.Either way, clearly most of the forgetting that wasto take place on a given retrieval practice trialtook place during the initial seconds of that trial.

However, as suggested by the results of Ex-periment 4, there may be conditions in which themagnitude of a retrieval-induced forgetting effectdoes increase with retrieval practice time. InExperiment 4 participants were warned that theaccurate response to some of the retrieval prac-tice cues would consist of a previously studiedexemplar. It is possible that this warning exacer-bated the extent to which the studied exemplarscompeted with retrieval efforts, thereby causing

the studied exemplars to continue to interferewith retrieval practice beyond the initial 4 sec-onds of a given trial. Thus, whether the magnitudeof a retrieval-induced forgetting effect continuesto increase with retrieval practice time maydepend on the amount of competition providedby non-practised items. It should be noted,however, that the opportunity for additionalinhibition*and thus additional forgetting*isonly available when retrieval practice fails toinitially succeed. Presumably, once retrieval issuccessful, retrieval efforts cease, and inhibition isno longer necessary. In this sense, it is possiblethat retrieval failure*and, more generally, retrie-val difficulty*has the power to augment effectsof retrieval-induced forgetting.

Concluding comment

In memory research, that which is not retrieved isjust as important as that which is retrieved. Whenwe retrieve a particular item or set of items frommemory, we are effectively neglecting, or remain-ing silent, to other items in memory. And as workon retrieval-induced forgetting has shown, thissilence can have important implications for long-term memory. The current study adds an addi-tional dimension of silence to the discussion*demonstrating that neglecting items not onlyleads to forgetting when retrieval succeeds, butwhen retrieval fails. That retrieval success is not anecessary condition to cause forgetting providesstrong support for the inhibitory account ofretrieval-induced forgetting. Moreover, it pro-vides researchers with a new and powerful toolfor studying the role of inhibitory processes inretrieval. By making retrieval practice impossible,researchers have the ability to explore dynamics(such as retrieval practice time) that wouldotherwise be impossible.

REFERENCES

Anderson, J. R. (1983). The architecture of cognition.Cambridge, MA: Harvard University Press.

Anderson, M. C. (2003). Rethinking interference the-ory: Executive control and the mechanisms offorgetting. Journal of Memory and Language, 49,415�445.

Anderson, M. C. (2005). The role of inhibitory controlin forgetting unwanted memories: A consideration

112 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

of three methods. In C. MacLeod & B. Uttl (Eds.),Dynamic cognitive processes (pp. 159�190). Tokyo:Springer-Verlag.

Anderson, M. C., & Bell, T. (2001). Forgetting ourfacts: The role of inhibitory processes in the loss ofpropositional knowledge. Journal of ExperimentalPsychology: General, 130, 544�570.

Anderson, M. C., Bjork, E. L., & Bjork, R. A. (2000).Retrieval-induced forgetting: Evidence for arecall specific mechanism. Psychonomic Bulletin &Review, 7, 522�530.

Anderson, M. C., Bjork, R. A., & Bjork, E. L. (1994).Remembering can cause forgetting: Retrievaldynamics in long-term memory. Journal of Experi-mental Psychology: Learning, Memory, & Cognition,20, 1063�1087.

Anderson, M. C., & Levy, B. J. (2007). Theoreticalissues in inhibition: Insights from research on humanmemory. In D. Gorfein & C. MacLeod (Eds.),Inhibition in cognition (pp. 81�102). Washington,DC: American Psychological Association.

Anderson, M. C., & Spellman, B. A. (1995). On thestatus of inhibitory mechanisms in cognition: Mem-ory retrieval as a model case. Psychological Review,102, 68�100.

Aslan, A., Bauml, K-H, & Pastotter, B. (2007). Noinhibitory deficit in older adults’ episodic memory.Psychological Science, 18, 72�28.

Barnier, A. J., Hung, L., & Conway, M. A. (2004).Retrieval-induced forgetting of emotional andunemotional autobiographical memories. Cognitionand Emotion, 18, 457�477.

Bauml, K-H (2002). Semantic generation can causeepisodic forgetting. Psychological Science, 13,356�360.

Bjork, E. L., Bjork, R. A., & Anderson, M. C. (1998).Varieties of goal directed forgetting. In J. M.Golding & C. M. MacLeod (Eds.), Intentio-nal forgetting: Interdisciplinary approaches (pp.103�137). Hillsdale, NJ: Lawrence Erlbaum Associ-ates Inc.

Bjork, R. A. (1975). Retrieval as a memory modifier. InR. Solso (Ed.), Information processing andcognition: The Loyola Symposium (pp. 123�144).Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

Bjork, R. A. (1989). Retrieval inhibition as an adaptivemechanism in human memory. In H. L. Roediger &F. I. M. Craik (Eds.), Varieties of memory andconsciousness: Essays in honour of Endel Tulving(pp. 309�330). Hillsdale, NJ: Lawrence ErlbaumAssociates Inc.

Camp, G., Pecher, D., & Schmidt, H. G. (2005).Retrieval-induced forgetting in implicit memorytests: The role of awareness. Psychonomic Bulletin& Review, 12, 490�494.

Camp, G., Pecher, D., & Schmidt, H. G. (2007). Noretrieval-induced forgetting using item-specific cues:Evidence against a general inhibitory account.Journal of Experimental Psychology: Learning,Memory, & Cognition, 33(5), 950�958.

Camp, G., Pecher, D., Schmidt, H. G., & Zeelenberg, R.(in press). Are independent probes truly indepen-dent? Journal of Experimental Psychology: Learn-ing, Memory, & Cognition.

Carroll, M., Campbell-Ratcliffe, J., Murnane, H., &Perfect, T. (2007). Retrieval-induced forgetting ineducational contexts: Monitoring, expertise, textintegration, and test format. European Journal ofCognitive Psychology, 19, 580�606.

Ciranni, M. A., & Shimamura, A. P. (1999). Retrieval-induced forgetting in episodic memory Memory.Journal of Experimental Psychology: Learning,Memory, & Cognition, 25, 1403�1414.

Johnson, S. K., & Anderson, M. C. (2004). The role ofinhibitory control in forgetting semantic knowledge.Psychological Science, 15, 448�453.

Landauer, T. K., & Bjork, R. A. (1978). Optimumrehearsal patterns and name learning. In M. M.Gruneberg, P. E. Morris, & R. N. Sykes (Eds.),Practical aspects of memory. New York: AcademicPress.

Levy, B. J., & Anderson, M. C. (2002). Inhibitoryprocesses and the control of memory retrieval.Trends in Cognitive Sciences, 6, 299�305.

Levy, B. J., McVeigh, N. D., Marful, A., & Anderson, M.C. (2007). Inhibiting your native language: The roleof retrieval-induced forgetting during second lan-guage acquisition. Psychological Science, 18, 29�34.

Loftus, E. F. (1973). Activation of semantic memory.American Journal of Psychology, 86, 331�337.

MacLeod, C. M., Dodd, M. D., Sheard, E. D., Wilson,D. E., & Bibi, U. (2003). In opposition to inhibition.In B. H. Ross (Ed.), The psychology of learning andmotivation (Vol. 43, pp. 163�214). San Diego, CA:Academic Press.

MacLeod, M. D. (2002). Retrieval-induced forgetting ineyewitness memory: Forgetting as a consequence ofremembering. Applied Cognitive Psychology, 16,135�149.

MacLeod, M. D., & Saunders, J. (2005). The role ofinhibitory control in the production of misinforma-tion effects. Journal of Experimental Psychology:Learning, Memory, and Cognition, 31, 964�979.

Perfect, T. J., Stark, L-J., Tree, J. J., Moulin, C. J. A.,Ahmed, L., & Hutter, R. (2004). Transfer appro-priate forgetting: The cue-dependent nature ofretrieval-induced forgetting. Journal of Memoryand Language, 51, 399�417.

Phenix, T. L., & Campbell, J. I. D. (2004). Effects ofmultiplication practice on product verification:Integrated structures model or retrieval-inducedforgetting. Memory & Cognition, 32, 324�335.

Roediger, H. L. III. (1978). Recall as a self-limitingprocess. Memory & Cognition, 6, 54�63.

Roediger, H. L. III, & Karpicke, J. D. (2006). Thepower of testing memory: Basic research andimplications for educational practice. Perspectiveson Psychological Science, 1, I181�210.

Shaw, J. S., Bjork, R. A., & Handal, A. (1995).Retrieval-induced forgetting in an eyewitness-memory paradigm. Psychonomic Bulletin & Review,2, 249�253.

Shivde, G., & Anderson, M. C. (2001). The role ofinhibition in meaning selection: Insights fromretrieval-induced forgetting. In D. Gorfein (Ed.),On the consequences of meaning selection: Perspec-tives on resolving lexical ambiguity (pp. 175�190).

RETRIEVAL SUCCESS AND INHIBITION 113

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013

Washington, DC: American Psychological Associa-tion.

Storm, B. C., Bjork, E. L., & Bjork, R. A. (2005). Socialmetacognitive judgments: The role of retrieval-induced forgetting in person memory and impres-sions. Journal of Memory and Language, 52,535�550.

Storm, B. C., Bjork, E. L., & Bjork, R. A. (2007). Whenintended remembering leads to unintended forget-

ting. Quarterly Journal of Experimental Psychology,60, 909�915.

Storm, B. C., Bjork, E. L., Bjork, R. A., & Nestojko, J.F. (2006). Is retrieval success a necessary conditionfor retrieval-induced forgetting? PsychonomicBulletin & Review, 13, 1023�1027.

Williams, C. C., & Zacks, R. T. (2001). Is retrieval-induced forgetting an inhibitory process. AmericanJournal of Psychology, 114, 329�354.

114 STORM AND NESTOJKO

Dow

nloa

ded

by [

Was

hing

ton

Uni

vers

ity in

St L

ouis

] at

19:

41 1

7 Se

ptem

ber

2013