new conceptual associative learning in amnesia: a …...2015/08/07 · new conceptual associative...

New Conceptual Associative Learning in Amnesia: A Case Study

Journal of Memory and Language43, 291–315 (2000)doi:10.1006/jmla.2000.2733, available online at http://www.idealibrary.com on

Suparna Rajaram

State University of New York at Stony Brook

and

H. Branch Coslett

Temple University School of Medicine

We report two experiments in this article that were designed to investigate the role of retrievalconstraints and interference in implicit learning of new verbal associations in a densely amnesicparticipant, C.V., who had presumably sustained medial temporal lobe damage secondary to ananoxic episode. In Experiment 1, repeatedly studied novel sentences produced significant primingwith Sentence1Fragment retrieval cues that provided maximal perceptual support as well as per-ceptual priming for the single-word targets. However, little learning was observed when no percep-tual cues were provided for the target itself with the Sentence1??? retrieval cues. In Experiment 2,the effects of intraexperimental interference were measured by examining new verbal learning underthe Study-Only, Study-Immediate test, Test-Study training conditions. Unlike in the findings reportedin prior studies, C.V. showed little learning with the Sentence1??? retrieval cues even under theminimal interference, Study-Only, condition. Together, these results demonstrate that implicit accessto novel verbal associations at a level more abstract than their perceptual configurations is notubiquitously observed in dense amnesia even when the learning conditions are optimized. Theseresults provide a window into the processes that mediate implicit learning of novel verbal associationswhen the explicit memory contribution is minimized.© 2000 Academic Press

In recent years, empirical studies involving(ors oionveaningw

examine the role of these variables on the ac-m-ga-

inin-

byse-as

tivetten-ce,. Aed-esia,

ndde-c-ni-

sebe

si-ssislusacan

tlyby

s repreantrant.jarak atail:

memory-intact as well as memory-impairedamnesic) participants have informed theoriememory functions and processes. Investigatinvolving memory-intact participants hademonstrated that repetition, interference,retrieval cues play a large role in modulatmemory performance. In the present article,

We are grateful to C.V. and his family for their enthuastic and sustained participation in this project. The atance of Kimberly Feldman and Yoko Yahata in stimupreparation, data collection, and scoring is gratefullyknowledged. Special thanks are due to Teresa BlaxtonEndel Tulving for sharing their stimuli. This article greabenefited from the constructive comments providedDouglas Nelson, Endel Tulving, and three anonymouviewers. Research reported in this manuscript and thearation of this manuscript was supported by an NIH grR29MH57345-01, to Suparna Rajaram and gRO1AG08870 and RO1DC02754 to H. Branch Coslet

Correspondence may be addressed to Suparna RaDepartment of Psychology, State University of New YorStony Brook, Stony Brook, NY 11794-2500. [email protected].

291

fs

d

e

quisition of new information in a densely anesic individual. The purpose of our investition is to isolate the role of implicit processesthe acquisition and retrieval of new verbalformation.

The amnesic syndrome is characterizedthe selective loss of memory for events subquent to the onset of amnesia (also knownanterograde amnesia) while other cognifunctions such as language, perception, ation, reasoning, problem solving, intelligenand short-term memory are relatively intactloss of memory for events immediately precing the onset of amnesia, or retrograde amnis also commonly reported.

The classic studies by Warrington aWeiskrantz (1968, 1970) demonstrated thatspite grossly impaired explicit memory funtions, as those involved in recall and recogtion, implicit memory functions, as thoinvolved in word fragment completion, may

-

-d

-p-,t

m,

0749-596X/00 $35.00Copyright © 2000 by Academic Press

All rights of reproduction in any form reserved.

largely intact in amnesia (Diamond & Rozin,rdarentthndomat

tingent, oitthat ree-utin

ysi&

ryaninroe

esanw

habea

d nte

notofin

s &

on-beal.forhaveietal.,9784

Vaidya et al., 1995; Warrington & Weiskrantz,h-tualsks

etciesthes ofese

cep-ednc-onof5;co-ss;hero-

acti-tep-s ofive

rch-ory

m orc thea byt alsi heri iono att tutet icle.W selya era-t hei ball

e ofi ball thea s isd intot dies,

292 RAJARAM AND COSLETT

1984; Graf & Schacter, 1985). In the wofragment completion task, participantsshown a list of words at study (e.g., elephaand are later given fragmented versions ofstudied ( _ l _ p _ a n _; solution: elephant) anonstudied words, and they are asked to cplete the fragments with the first word thcomes to mind. The advantage in complestudied fragments over nonstudied fragmconstitutes the measure of implicit memorypriming (Tulving et al., 1982). Such implicmemory tasks are largely data-driven suchchanges in modality across study and tesduce priming (Blaxton, 1989; Rajaram & Rodiger, 1993; Srinivas & Roediger, 1990), bmeaningful encoding does not necessarilycrease priming relative to surface-level analof study material (see review by RoedigerMcDermott, 1995).

Priming is also observed in implicit memotasks when the task relies on conceptual,not perceptual, processes. For example,conceptual implicit task such as category pduction, participants study exemplar nam(e.g., elephant) of various categories. At tonly the category names are provided (e.g.,imals), and participants are asked to write dothe first few exemplars of that category tcome to mind. This task is considered toconceptual in nature because the exemplarscategory names are conceptually related, anperceptual overlap exists between study andstimuli. Predictably, modality changes doimpair such priming whereas manipulationmeaningful encoding enhances such prim(e.g., Blaxton, 1989; Hamann, 1990; SrinivaRoediger, 1990).

Both perceptual and conceptual formspriming, or implicit memory, have been extesively examined in amnesia and found togenerally preserved (see Moscovitch et1993, for a review), although the evidencepreserved conceptual priming is somewmixed. Many studies have shown preserconceptual priming in amnesia across a varof conceptual implicit tasks (Gardner et1973; Hamann, 1990; Keane et al., 19Schacter, 1985; Shimamura & Squire, 19

)e

-

sr

t-

-s

da-st,-

nt

ndost

g

f

,

tdy

;;

1982; Winocur & Weiskrantz, 1976). Yet, oters have reported impairment on conceppriming in amnesia on a similar group of ta(Blaxton, 1992; Cermak et al., 1998; Keaneal., 1997). The reasons for these discrepanare not clear, but one possibility may bedifferences in procedure and different groupamnesic participants employed across thstudies.

The database on the perceptual and contual priming effects in amnesia has informand refined our theories of intact memory futions by providing a window into the operatiof implicit memory when the contributionexplicit memory is minimized (Blaxton, 199Cohen & Squire, 1980; Gabrieli, 1995; Mosvitch, 1992, 1994a; Roediger et al., in preSchacter, 1990; Tulving & Schacter, 1990). Tstudies reviewed thus far reveal both the pcessing and neural requirements for the revation of learned informationunder differenretrieval conditions. Specifically, the perctual, lexical, and conceptual representationwords such as “elephant” exist in the cognitsystem prior to the experimental context.

A key question that has concerned reseaers in recent years is the role of implicit memin mediating the learning ofnew verbal infor-

ation. It is obvious that explicit memoryonscious awareness plays a vital role incquisition of new information, as evidenced

he performance of memory-intact individun the laboratory and in the real world. Whetmplicit memory also supports the acquisitf new verbal information, and if it does, wh

he nature of such learning might be, constihe questions examined in the present art

e investigated these questions in a denmnesic individual because the minimal op

ions of explicit memory may help isolate tmplicit processes that mediate new verearning.

Recent studies have investigated the rolmplicit processes in mediating new verearning in the memory-intact as well asmnesic populations. The latter set of studieirectly relevant here, and these studies fall

hree general categories. In one set of stu

the acquisition of hitherto unknown words wasult

havofber-asre

amth

girlewingdernetcierisanm-

inovesionndfo&

l aseel

uc.g.is-els

e-licit

&

asrmtheseghwitm-nda,

ire,

1987), and name–face associations (Thoene &ar-ofentun-ch–airsirstioning

deswitheen&ing

m-lvingem-ics

tifi-&

al.,the&

dentn theears to

beui-

sicewreem--

thelemtactthewm-sesex-y

icitrtici-esicthe

293NEW LEARNING IN AMNESIA

examined in the amnesic population. The resof these studies are mixed. Some studiesshown little, if any, evidence of acquisitionnew vocabulary in the medial temporal lopatient H.M. (Gabrieli et al., 1988) or in Kosakoff’s patients (Grossman, 1987). In contrpositive evidence of new learning has beenported in amnesics in some studies. For exple, there have been reports of progress inacademic training of a young amnesic(Wood et al., 1989), normal acquisition of nFrench vocabulary (Hirst et al., 1988), learnof single-word interpretations of ambiguousscriptions (McAndrews et al., 1987), and leaing of implicit frequency judgments (Dopkinsal., 1994). The reasons for these discrepanonce again are not clear, but they may afrom differences in tasks and proceduresdifferences in the severity and etiology of anesia among the participants.

In a second set of studies, new learningamnesia has been examined for individual nitems such as nonwords and novel shapepriming tasks such as perceptual identificatword reading, and lexical decision. By alarge, there appears to be preserved primingnonwords (Cermak et al., 1991; GabrieliKeane, 1988; Musen & Squire, 1991) as welnovel shapes (e.g., Gabrieli et al., 1990; Mu& Squire, 1992a; Schacter et al., 1993; Verfalie et al., 1992). However, in some studies spriming is found to be at subnormal levels (eDiamond & Rozin, 1984; Gordon, 1988), raing the concern that evidence for normal levof priming for novel information may somtimes be contaminated by residual expmemory in some amnesics (see BowersSchacter, 1993).

In the third set of studies, new learning hbeen assessed for verbal associative infotion; these studies are directly relevant forpresent purposes because we used noveltences as the stimuli in our study. Althouassociative learning has been examineddifferent types of stimuli such as learning coputer vocabulary and programming comma(Glisky & Schacter, 1988; Glisky et al., 19861986b), simple facts (Shimamura & Squ

se

t,--e

--

sed

lin,

r

sn-h,

a-

n-

h

s

Glisky, 1995), the most extensively studied padigm in this domain involves the learningnovel word associations. In a typical experimof this sort, participants are presented withrelated word pairs (window–reason, marshave) and are later presented with intact p(window–rea_____) or recombined pa(march–rea_____) during the stem completask. The advantage in stem completion primfor intact pairs over recombined pairs provithe measure of new learning. The evidenceamnesic participants on this task has bmixed (Graf & Schacter, 1985; SchacterGraf, 1986) such that new associative learnreliably occurred only in mild to moderate anesics and not severe amnesics (but see Tuet al., 1991). However, recent studies have donstrated normal levels of priming in amnesfor new associations when rapid word idencation (Gabrieli et al., 1997; but see PallerMayes, 1994), reading time (Moscovitch et1986, but see Musen & Squire, 1992b), orlexical decision task (Goshen-GottsteinMoscovitch, 1992) were used as the depenmeasures. Once again, the discrepancies iimplicit acquisition of new associations appto be due to changes in tasks as well aseverity of amnesia.

Together, the following conclusions maydrawn from these mixed findings on the acqsition of new verbal information in the amnepopulation. One, the positive evidence of nlearning is often not purely implicit in natuand may be mediated by residual episodic mory. If the explicit memory involvement complicates the interpretation of the data inmoderately amnesic participants, this probmay become exaggerated in the memory-inpopulation. Thus, in order to understandunique contribution of implicit memory to neverbal learning in the normal cognition of meory, it is critical to examine these procesunder conditions where the operations ofplicit memory are minimized. An effective wato achieve this goal is by examining the impllearning processes in severely amnesic papants. To this end, we tested a severely amnparticipant in the present experiments. Two,

variations in tasks and procedures across labo-xedskswito enteranestRa

1).thaitsobewenterth

ativl tot, inlarvelorm-raech

atth

theenetntiartntsIn

elato

cantof&

artoue

conhe

ories have recently proposed a special role oftat-ewo-n &5;,ay

so-dialenten-aryiesas

aseal.,et

95;al.,deraveture

er-

ineporttiveheas

ra-tedit

ofedvalTheoti-Ha-92;ins ofsep-

sopos-in

andle-


ratories may be partly responsible for the mifindings. Thus, the selection of the critical taand procedures used in other laboratoriesdense amnesic participants is necessary tfectively address this problem with differepatient samples. For this reason, our expments were closely modeled after the tasksprocedures used in recent studies that invgated related issues (Hayman et al., 1992;jaram & Coslett, in press; Tulving et al., 199Three, the reviewed studies also indicateacquisition of individual, novel perceptual un(such as nonwords and novel shapes) isserved more reliably in amnesics, but nlearning that involves an associative componis often impaired (see also Curran & Schac1997). We focused on the latter outcome inpresent investigation because the associcomponent of learning may be fundamentamost verbal learning (see Rajaram & Cosletpress). For instance, learning new vocabu(Gabrieli et al., 1988) also requires the deopment of an association between a new wand its meaning. Thus, the role of implicit meory in mediating new verbal learning in genemay be best understood by examining the manisms that govern newassociativelearning.

Associative learning may occur either onlythe perceptual/lexical level, as measured byperceptual identification, lexical decision, orreading tasks (Gabrieli et al., 1997; GoshGottstein & Moscovitch, 1992; Moscovitchal., 1986), or at a higher conceptual or semalevel where some of the perceptual/lexical pof the cue are not provided to the participa(Hayman et al., 1992; Tulving et al., 1991).the present study, we focus on the latter, rtively more conceptual, level of associationsexamine whether successful new learningoccur beyond the level where the participacould simply rely on the perceptual gestaltthe stimuli (see also Goshen-GottsteinMoscovitch, 1995).

The selection of the particular amnesic pticipant (C.V.) in this study also permitted uscollect evidence pertaining to a related issthe possible neuroanatomical basis of newceptual associative learning. A number of t

hf-

i-di--

t

-

t,ee

y-d

l-

e

-

cs

-

ns

-

,-

-

the medial temporal lobe structures in faciliing the binding of various elements of ninformation (Cohen & Eichenbaum, 1993; Chen et al., 1997; Cohen et al., 1994; JohnsoChalfonte, 1994; McClelland et al., 199Moscovitch, 1994b; O’Reilly & McClelland1994). According to these proposals, one mpredict impairment in the binding of new asciations when damage is sustained to the metemporal lobe structures. Evidence from rodstudies supports this prediction (see Eichbaum et al., 1994). In humans, the preliminevidence from recent neuroimaging stud(Cohen et al., 1994; Klingberg et al., 1994)well as data from some of the amnesic cstudies support this prediction (Gabrieli et1988; Rajaram & Coslett, in press; Verfaellieal., 1995; see also Squire & Knowlton, 19but also see Hayman et al., 1992; Tulving et1991). However, the specific conditions unwhich acquisition and testing take place hoften varied across studies, and the exact naof learning impairment is not clearly undstood.

In summary, our central goal was to examwhether implicit memory processes can supthe acquisition of new, conceptual associalearning in a densely amnesic individual. Trecruitment of a densely amnesic individual wexpected to provide a window into the opetions of implicit memory that cannot be isolaeasily in memory-intact individuals in whomis often difficult to rule out the contributionsexplicit memory. We systematically examinthe effects of interference and types of retriecues on new verbal associative learning.selection of specific manipulations was mvated by the findings from previous studies (mann & Squire, 1995; Hayman et al., 19Tulving et al., 1991; Rajaram & Coslett,press). The specific goals and manipulationeach of the two experiments are presentedarately in the forthcoming sections.

In addition to this primary aim, our study alprovides suggestive data that speak to thesible role of medial temporal lobe structuresmediating the learning of new associations,by inference, in the binding of different e

ments of the associative stimuli (i.e., sentences)nottiveth

acolepting.Vashia-eer-

eeamanb

al-n-o-

tusth

cte3;5)threhetivxicip-

ome eal.ho7)

ct,e t.ledon

rde

On the Wechsler Adult Intelligence Test-Re-IQ

w larya cor-r cei theB ther icc oryS ar ralm yedr ry-l , hew icalm Att tion/c theh tentw ontA on-fi thed Qs onWnT thel thea -R( theW d-c thefin

C

ledM51 n-t leasto i-

ith

st 1o s forE


presented to the amnesic participant. Wethat findings from our study provide suggesinsights into, and not a direct assessment of,relationship.

PARTICIPANT DESCRIPTION

C.V.

C.V. was a 50-year-old college-educated mwho suffered a cardiac arrest, perhaps as asequence of a sustained seizure (status epicus), approximately 4 years prior to the tesreported here. After the cardiac arrest, Cnoted profound memory impairment but wotherwise asymptomatic. Unable to return towork in computer programming and interntional sales, he devoted his time to voluntwork and golf. C.V.’s wife stated that his pesonality and social interactions had not baltered by the cardiac arrest. Neurologic exination revealed C.V. to be alert, pleasant,cooperative. He exhibited a severe amnesiathe examination was otherwise normal.

MRI of the brain demonstrated no abnormities. Although we do not have pathologic cofirmation of the lesion site in C.V., neuropathlogic investigations of patients with staepilepticus have consistently demonstratedhippocampus to be the most severely affebrain region (e.g, Corsellis & Bruton, 198Cendes et al., 1995; Hopkins et al., 199Patchy areas of laminar neuronal loss incortex and death of Purkinje cells in the cebellum are less consistently observed. Furtmore, several studies have reported quantitaMRI analyses of the hippocampus in anopatients where significant reduction in the hpocampal region was noted in patients cpared to controls (Press et al., 1989; Squiral., 1990; Kesner et al., 1992; Hopkins et1995). In light of these and other neuropatlogic findings (e.g., Graham and Lantos, 199as well as C.V.’s strikingly preserved intellewe suggest that his amnesia is attributablhippocampal disruption and its connections

Neuropsychological examination reveathat C.V. functioned in the superior rangemany tasks despite his severe memory diso

e

is

nn-ti-

.

s

r

n-dut

ed

.e-r-e

-t

,-,

o

r.

vised (WAIS-R), C.V. obtained a full scaleof 130 (verbal5 139, performance5 113),

ith high-scaled scores of 17 on the vocabus well as arithmetic sections. These scoresoborate the clinical impression of intelligenn the superior range. C.V. scored 58/60 onoston naming test; he successfully named

emaining 2 items with the use of phonemues. In stark contrast, on the Wechsler Memcale – Revised (WMS-R), C.V. obtained

emarkably impaired score of 86 for geneemory and scored below 50 on the dela

ecall. His impairment was evident on memooaded subtests of the scale; for exampleas unable to recall any details from the logemory subtest 10 min after presentation.

he same time, his performance on the attenoncentration section of the WMS-R was inigh range (125). This last score is consisith his outstanding digit span performance

he WAIS-R (forward5 9, backward5 7).ltogether, C.V.’s profound amnesia was crmed on these neuropsychological tests asifferential between his WAIS-R full scale Icore (130) and the General Memory scoreMS-R (86) was more than 2SD apart by

ormative standards (mean5 100, SD 5 15).his pattern was further substantiated by

arge differential between C.V.’s score onttention/concentration section of the WMS125) and the delayed recall section of

MS-R (below 50). On the Warrington forcehoice recognition test, C.V. scored belowrst percentile for his age group (faces5 34/50,ames5 34/50).

ontrol Participants

Four control participants (henceforth calatched Controls), matched for age (mean50.5 years,SD5 2.38) and education (mean55.5 years,SD5 1), were tested so as to cou

erbalance materials across conditions atnce.1 Another group of nine control partic

pants (henceforth called Young Controls) w

1 Two matched control participants were tested in Lif the materials for reasons described under Resultxperiment 2.

mean age of 21.56 years (SD5 3.78) and meano

t cem ntw ticip

mr baa antK &C ol anR eirs thei uc-i ina tedp veh 80T an .gM an ea nsw as ross mat in thet bei eds nop ded( asu ncew n.T ews g ea ewl ina

ndi tiob req rela skss ing

where all of the perceptual information aboutt tici-p (Ra-j pli-c lya sicp oryl ingi en-c ndi ast ruc-t mali nc-t w-i es.M ex-t e-d ip-p eralt ofi llyi

rtic-i tod byT elya thes seda od-o m-p esw ce-d ntsp ssi-b rals ing.I esiat in-f res( al.,1 nte,1 h,1 e-d cia-t en-t d int

de-


education of 15.11 years (SD5 1.96) was alsested in Experiment 2 to fully counterbalanaterials across conditions. Control participaere treated identically to the amnesic parant except when noted otherwise.

EXPERIMENT 1

The motivation for Experiment 1 came froecent reports of the learning of new verssociations in densely amnesic particip.C. (Tulving et al., 1991) and C.C. (Rajaramoslett, in press) and from mixed evidence

earning in another densely amnesic particip.H. (Rajaram & Coslett, in press). In thtudy, Tulving et al. (1991) focused onmportance of experimental factors in prodng long-lasting learning of new associationsmnesia. Amnesic participant K.C. exhibirofound memory loss following extensiead injury from a motorcycle accident in 19ulving et al. (1991) presented K.C. withumber of novel but plausible sentences (eEDICINE cured HICCUP) at study acrossumber of sessions conducted at least one wpart. At test, implicit retrieval instructioere given to assess the retention of theseociations, and the retrieval cues varied acessions in the amount of perceptual inforion they contained. As has been reportedumerous studies, perceptual priming for

arget words (e.g., _ I _ C _ P) was found tontact in K.C. Interestingly, K.C. also showubstantial retention for the target item whenerceptual cue for the target itself was proviMEDICINE cured ???) even though he wnable to recognize any of the studied sentehen explicit memory instructions were givehis finding was taken as evidence for nemantic learning in dense amnesia. Tulvinl. (1991) concluded from their data that n

earning of this kind is slow and laboriousmnesia, but is long lasting.We became particularly interested in the fi

ngs from the Sentence1??? retrieval condiecause performance under this conditionuires the binding of new associations at atively more conceptual level than in the tauch as perceptual identification and read

s-

ls

ft

.

.,

ek

s-s-n

s

t

-n--

,

he new associations is provided to the parants. Consequently, in a recent study, we

aram and Coslett, in press) attempted to reate Tulving et al.’s findings with two densemnesic participants, C.C. and R.H. Amnearticipant C.C. developed profound mem

oss after the resection of a meningioma arisn the olfactory groove that created focalephalomalacia bilaterally in the gyri recti anferior portions of the cingulate gyri as wellhe basal forebrain and other subfrontal stures. Postoperatively, C.C. exhibited norntelligence but grossly impaired memory fuions. R.H. became profoundly amnesic follong a period of seizures in his early twenti

RI scans at the time of testing revealedensive but focal damage bilaterally in the mial temporal lobe regions, including the hocampus, and some atrophy of the left lat

emporal lobe. R.H. exhibited superior levelsntelligence, with an IQ of 121, but dramaticampaired memory functions.

The comparison of these two amnesic papants in our study provided the opportunityetermine whether the findings reportedulving et al. (1991) generalize to other densmnesic participants. To this end, we usedame set of materials that Tulving et al. und adopted the critical features of the methlogy from their study. Furthermore, the coarison of C.C.’s and R.H.’s performanchere all the materials and experimental proures were matched for the two participarovided the opportunity to delineate the pole, differential roles of the underlying neutructures involved in new associative learnn particular, based on the theories of amnhat posit a specialized function of bindingormation to the medial temporal lobe structuCohen & Eichenbaum, 1993; Cohen et997; Cohen et al., 1994; Johnson & Chalfo994; McClelland et al., 1995; Moscovitc994b; O’Reilly & McClelland, 1994), we pricted that conceptual learning of new asso

ive information, as measured by the Sence1??? retrieval cues, would be impairehe medial temporal lobe participant R.H.

For present purposes, the methodological

tails and results of Experiment 1 are relevant. This., 1trofrose,aliliaIC, 3ereE

narninkheici-enhe

axisi

s 2enR)g i

gen,

b re-b ndt elym n

diaatof

edagmeedshisith&

esne,r aopd.

However, such learning appears to be slow andnot

ngsu-

em-the

besicoraling.hathe

inuss

sec-

ewSen-rtic-beewesicbe-s toatre-

ousex-in1–3ses-

ses-

eri-per-edlett,ceds toelyoun-par-ew-

heini-icitcel-for-


experiment included many sessions (8 for C.Cfor R.H., and 5 for the three matched conparticipants). Each session was conducted1–3 weeks apart and contained a study phabrief retention interval, and a test phase. Insessions, participants studied novel and unfamsentences at study (e.g., MEDICINE cured HCUP, STAFF shot HIJACKER). In sessions 15, 7, 9, and 11, two types of retrieval cues wprovided, Sentence1Fragment (MEDICINcured _I _ C _ P) and Sentence1??? (STAFFshot ???). The Sentence1??? retrieval conditioprovides a strong test of new associative leing because participants have to develop lbetween hitherto unassociated words and tlinks have to be strong enough for the partpants to generate the target word in the absof any perceptual support for that word. TSentence1Fragment condition provided mmal cues and was expected to produce conerable priming for the target words. Session4, 6, 8, 10, and 12 presented only the fragmof all target words (_I _ C _ P, _ I J _ _ K _to assess the status of perceptual priminthese two amnesic participants.

We replicated Tulving et al.’s (1991) findinof new learning under the Sentence1Fragmas well as under the Sentence1??? conditions

ut only in the performance of our basal forain amnesic participant C.C. We failed to fi

he evidence of new learning with the relativore conceptual cues of Sentence1???, eve

after 12 sessions of study and test, in our metemporal amnesic participant R.H. These dstand in sharp contrast with the findingsTulving et al. (1991). Notably, R.H. showpreserved perceptual priming with the Frments Only cues (_ I J _ _ K _ R ) for the saset of target words (HIJACKER) that he failto produce with the Sentence1??? cue(STAFF shot ???). We further replicated tpattern of performance for C.C. and R.H. wan entirely new set of materials (RajaramCoslett, in press, Experiment 3). Together, thfindings have two important implications. Olearning of new verbal associations can occua relatively conceptual level even when theerations of explicit memory are minimize

2lm

alr

-,

-sse

ce

-d-,

ts

n

t

la

-

e

t-

laborious, depends on repeated trials, and isobserved in all amnesics. Two, these findiconstitute initial but strong evidence from hman studies for the proposal that medial tporal lobes may be specifically involved inlearning of new verbal associations. It shouldnoted, however, that Tulving et al’s amnepatient K.C. had also sustained medial templobe damage but did exhibit semantic learnThus, we need more empirical information tcan provide us with further insights into tpossible role of the medial temporal lobesmediating new associative learning. We discthis issue further in the General Discussiontion.

At the cognitive level, the absence of nassociative learning as measured by thetence1??? retrieval cues in our amnesic paipant R.H. (Rajaram & Coslett, in press) mayattributable to two different sources. One, nlearning may not occur because the amnparticipant fails to develop the associationtween different words in the sentence, or failbind the information. Two, it is possible thsome associative learning does occur withpeated trials, but this learning may be tenuand susceptible to rapid forgetting. In ourperiment with R.H. (Rajaram & Coslett,press), the study-test sessions were heldweeks apart. This extended delay betweensions may have led to forgetting betweensions.

We tested this latter hypothesis in Expment 1 here. To do so, we performed an eximent similar to the investigation describabove (Experiment 1, Rajaram and Cos1991) but with one critical change. We reduthe interval between sessions from 1–3 week10 min–17 h. We reasoned that if a relativaccelerated presentation of sessions can cteract the forgetting process, then amnesicticipant C.V. may exhibit some learning of nassociations in the Sentence1??? retrieval condition. This outcome will further delineate tmechanisms (forgetting versus failure totially bind associations) that mediate implverbal associative learning. Thus, if such aceration were not sufficient to attenuate the

getting process, then our findings would dem-s ated

edoscritce

adun

en-3

sseiedncingwel

, thtarpreseer

venco

di-endto

estich),hert ons.hingeric-heand

erasTho

ance

(not used for any other purpose in this or theom

s tothebe

sedentwasomer-kenam-

re-andtrolforudy

retudytest

10as

n acide.V.’stic-inthe

o bee re-

tn-in-

rstandses-iedted,

ents

dy-was48ook-en-


onstrate that the binding of new associationa relatively conceptual level is severely limiwhen explicit memory is suspended.

Method

Materials. Eighty three-word sentences usby Tulving et al. (1991) and Rajaram and Clett (1999, Experiment 1) were used as theical stimuli in this experiment. Each sentenconsisted of three words that together msense but also constituted a novel, hithertoknown, configuration. Of these 80 critical stences, 48 were presented for study, andwere used as nonstudied sentences to apriming. At test, 24 studied and 16 nonstudsentences were presented under the Sente1Fragment retrieval condition and the remain24 studied and 16 nonstudied sentencespresented under the Sentence1??? retrievacondition. In some sessions (specified later)fragments of all the studied and nonstudiedgets (the last word of each sentence) weresented to assess perceptual priming. Acrosssions (from 1–9), the specific sentences wnested within each condition such that a gisentence always appeared under the samedition in all sessions. In the study list, in adtion to the 48 study sentences, 12 buffer stences, 8 at the beginning of the study list anat the end of the study list, were includedeliminate primacy and recency effects. At tin addition to the 80 critical sentences (of wh48 were studied and 32 were nonstudiednonstudied buffer sentences (not used anywin the experiment) were presented at the stathe test list under different retrieval conditioThe presentation of all critical sentences witeach study and test list was randomly arranwith respect to conditions and with the resttion that the order for C.V. was identical to torder used for C.C. and R.H. in RajaramCoslett’s (1999, Experiment 1) study.

As in the Rajaram and Coslett (1991, Expiment 1) study, a recognition memory task wadministered at the end of the ninth session.recognition task in this experiment consisteda booklet containing all 48 study sentencesa new set of 48 nonstudied (buffer) senten

t

--

e-

2ss

e

re

e--s-e

n-

-4

,

8ref

d

-

efds

next experiment) arranged in a new randorder.

The selection and assignment of materialdifferent conditions as described, and allcritical details of the design and procedure todescribed shortly, were identical to those uin the Rajaram and Coslett (1991, Experim1) study. The only change in the procedurethe reduction in the intersession interval fr1–3 weeks (Rajaram & Coslett, in press, Expiment 1) to 10 min–17 h. These steps were tato ensure that the treatment of the presentnesic participant C.V. was identical in allspects to that of amnesic participants C.C.R.H., as well as to that of matched conparticipants tested in our prior study, exceptthe intended manipulation of accelerated stand test cycles.

Design and procedure.Nine sessions weconducted with each session containing a sphase, a 5-min retention interval, and aphase. Each session was conducted frommin–17 h apart. In the study phase, C.V. wpresented with one sentence at a time oMacintosh computer and was asked to dewhether the sentence made sense to him. Cresponses included both “yes” and “no,” parularly in the initial sessions. Following a 5-mbreak, the test phase was conducted. Inodd-numbered sessions 1, 3, 5, 7, and 9, tdesignated the Sentence Cues sessions, thtrieval cues consisted of Sentence1Fragmen(MEDICINE cured _ I _ C _ P) and Setence1??? (STAFF shot ???). C.V. wasstructed to provide the first solution or the fiword that came to mind. In sessions 2, 4, 6,8, to be designated Fragments Only Cuessions, the fragmented versions of all the studand nonstudied target words were presenand C.V. was asked to complete the fragmwith the first solution that came to mind.

One hour after the completion of nine stutest sessions, a recognition memory testadministered to C.V. All 48 studied andnonstudied sentences were presented in a blet, and C.V. was asked to check all the stences that looked familiar to him.

tingud-m-va,pedze

n-seudi

i tagf chr im-i riveu rivt tudi -i thao ond onu bT let( se

s rag-m g. 1i te,h be-l tive,s

asw ex-h thei tion.A .V.e dert1 ceu 9,w re-s toR tt’ss estp -s wasd ed2 d tofi .V.d ofl

TABLE 1

nt

ied


Results

New learning was measured by subtracthe proportion of correctly completed nonstied items from the proportion of correctly copleted studied items under all three retrieconditions, Sentence1Fragment, Sentence1???and Fragments Only. The mean completionformance under various conditions is presenteTable 1. The standard priming data were analyin the following way. Within each retrieval codition, the total number of correct responacross sessions was computed for each stand nonstudied item. An independentt test bytem was conducted to determine the advanor studied over nonstudied items within eaetrieval condition. In addition, an adjusted prng measure (Snodgrass, 1989) was also dender each condition for each session. To de

his measure, the standard priming score (sed 2 nonstudied) was divided by (12 nonstuded) in order to adjust for baseline differencesften complicate the comparison of different citions in case studies. This measure is commsed in priming studies and was also usedulving et al. (1991) and Rajaram and Cos1999). The adjusted priming scores across

Response Probabilities of Amn

Sentence cuessessions

Sentence1???(e.g., STAFF shot ?

Studied No

1 0.043 0.045 07 0.089 0.13

Fragments only cuessessions

Targets from sent(e.g., _IJ__K

Studied

2 0.674 0.636 0.798 0.79

l

r-ind

sed

e

de-

t-lyyts-

ions for the Sentence Cues conditions and Fent Only Cues conditions are displayed in Fi

n the left and right panels, respectively. Noowever, that the statistical analyses reported

ow were conducted on the more conservatandard, priming scores.The critical question in this experiment w

hether the amnesic participant C.V. wouldibit learning of new verbal associations when

ntersession intervals were shortened in duras Fig. 1 shows, despite this manipulation, Cxhibited little learning of new associations un

he Sentence1??? retrieval condition,t (38) 5.55,SE5 0.23. The highest level of performannder this condition was found in Sessionhere C.V. produced 3 out of 24 correctponses. This performance is quite similar.H.’s performance in Rajaram and Cosletudy (1999, Experiment 1), in which R.H.’s berformance in the Sentence1??? condition conisted of 2 out of 24 correct responses andramatically different from C.C.’s, who produc1 out of 24 correct responses. Thus, we failend evidence of new associative learning in Cespite arranging relatively optimal conditions

earning.

ic Participant C.V. in Experiment 1

)Sentence1fragment

(e.g., MEDICINE cued _I_C_P)

udied Studied Nonstudied

0.67 0.380.88 0.500.88 0.690.92 0.630.88 0.63

e1???)

Targets from sentence1fragme(e.g., _I_C_P)

nstudied Studied Nonstud

0.44 0.63 0.310.31 0.50 0.380.50 0.67 0.560.50 0.92 0.50

es

??

nst

00000

enc_R

No

onsig-tuangen

rgeth-tha

to

erye tthpen-lactersi

ingthet, iimheo

og-ctlyon-son-in-es.ingd tothe

ceF re-t inew sen-t etters itionc alsoi g. 1,b( edw r theS crosss s waso tivelym risonp stionw

lth the

the


Priming performance under the second cdition of Sentence1Fragment cues yieldednificant priming,t(38)5 2.44,SE5 0.52, demonstrating the effectiveness of percepconstraints in C.V.’s performance. Primiscores were also analyzed for the FragmOnly cues (Sessions 2, 4, 6, and 8) for the tawords of sentences that belonged toSentence1Fragment and Sentence1??? conditions. Perceptual priming scores for targetsbelonged to the sentences in the Sentence1???conditions (e.g., _ I J _ _ K _ R) were foundbe significant,t(38) 5 2.42, SE 5 0.44. Thisfinding is notable in that these were the vtargets that C.V. failed to produce in responsthe Sentence1??? cues. In other words,finding diminishes the concern that possibleculiarities in the target word in the Setence1??? items might have resulted in aof learning. These data also mimic the patreported for the medial temporal lobe amneR.H. who exhibited robust perceptual primfor targets that he failed to produce underSentence1??? condition (Rajaram & Cosletpress, Experiment 1). Finally, perceptual pring for targets that belonged to tSentence1Fragment condition was also alm

FIG. 1. The data from Experiment 1 for adisplays the adjusted priming scores for the sesstarget word missing (Sentence1??? retrieval cueversion of the target word (Sentence1Fragment rpanel displays perceptual priming scores with ttargets from both the Sentence1??? (S1? here)conditions.

--

l

tt

e

t

ois-

knc

n-

st

significant,t(38) 5 1.92,SE5 0.47 (p5 .06,two-tailed).2

On the recognition memory task, C.V. recnized 42 out of 48 studied sentences correand did not false alarm to any of the 48 nstudied sentences. Thus, C.V. exhibited reaably good recognition memory for studiedformation that was exposed to him nine timHe also admitted to the recognition task be“less threatening” than the other tasks he hado, although he could not quite describeother tasks he had been asked to do.

2 We also compared priming across the Senten1ragment retrieval condition and the Fragments Only

rieval condition for the same set of items to determhether priming levels increased as a function of the

ence cues. This comparison showed a trend for btandard priming under the Sentence1Fragment condompared to under the Fragments Only condition, asndicated by the adjusted priming scores displayed in Fiut this did not meet significance,t(38) 5 1.63, p 5 .11two-tailed),SE5 0.33. This comparison should be viewith caution because the same items served undeentence1Fragment and Fragments Only conditions aessions. It should also be noted that because our focun the measure of associative learning under the relaore conceptual, Sentence1??? condition, the comparesented here was not critical with respect to the quee addressed in this series of experiments.

esic participant C.V. are displayed. The left panes where the sentence cues were presented either widition, labeled as S1? here) or with the fragmentedval cue condition, labeled as S1F here). The right

adjusted priming measure for only the fragments ofthe Sentence1Fragment (S1F here) retrieval cue

mnionconetrieheand

Discussion

arece

henideof

tedbusa

juryssis

-enprois

tog bad

tuaos-chifi-theenag-geen-or-er-dy

).d

rib-esm-icapert re.C.e-thesem-de, iuga

simply be based on a vague sense of familiarity.pe-ibedven

ofofrch,ongtact75;par-et&k-stu-m-ofss-&al.

am-deruntat if

ivencens

al.,ener-

entandbut

sessonep-en-

itua-ailsme

sen-uc-ne-ayd-


Three main findings of this experimentnoteworthy. One, C.V. showed little evidenfor the learning of new verbal associations wno perceptual cues for the targets were provto aid the target production. This patternperformance is similar to the pattern reporfor the medial temporal lobe amnesic, R.H.,different from the pattern reported for the baforebrain amnesic, C.C., and the head inamnesic, K.C. (Rajaram & Coslett, in preTulving et al., 1991). Two, in contrast to hpoor performance under the Sentence1??? condition, C.V. showed significant priming whthe retrieval cues of Sentence1Fragmentvided maximal support to aid retrieval. Thfinding, albeit with a different task, is similarthe reports of successful associative learninamnesics in perceptual identification and reing tasks that also provide all of the percepinformation at test (Gabrieli et al., 1995; Ghen-Gottstein & Moscovitch, 1992; Moscovitet al., 1986). Three, C.V. also showed signcant perceptual priming when fragments oftargets were presented in isolation (FragmOnly cues). Perceptual priming for such frments was obtained even for the set of tarthat C.V. failed to produce under the Stence1??? condition. Overall, C.V.’s perfmance bears a striking similarity to R.H.’s pformance reported in our previous stu(Rajaram & Coslett, in press, Experiment 1

It is important to note that C.V.’s (anR.H.’s) performance cannot readily be attuted to severity of amnesia alone. Both thamnesic participants did exhibit profound anesia on both clinical and neuropsychologassessment. However, C.V.’s recognitionformance was considerably better than thaported for either R.H. (16/48 correct) or C(24/48 correct). Alternately, Tulving et al. rported that K.C. failed to recognize any ofstudied sentences after multiple study-testsions, and yet unlike C.V., Tulving et al.’s anesic K.C. did show substantial learning unthe Sentence1??? condition. Furthermorethe present experiment, C.V.’s feedback sgested that his recognition performance m

d

tl

;

-

y-l

ts

ts

e

l--

s-

rn-y

But this sense of familiarity, even though surior to that seen in the other amnesics descrabove, was clearly not sufficient to support ethe implicit production of correct targets.

In Experiment 2, we examined the roleinterference in mediating implicit learningnew verbal associations. In memory reseathe deleterious role of interference has lbeen noted both in studies of memory-in(Martin, 1971; Postman, 1971; Runquist, 19Underwood & Postman, 1973) and amnesicticipants (Cermak & Butters, 1972; Cermakal., 1974; Mayes et al., 1987; WarringtonWeiskrantz, 1973, 1974; Winocur & Weisrantz, 1976). Some theories of amnesia polate that the mechanism by which medial teporal lobe structures facilitate the bindingvarious elements in new stimuli is by suppreing interference from other stimuli (ShapiroOlton, 1994). In a recent study, Hayman et(1992) and Hamann and Squire (1995) exined new verbal, associative learning unstudy-test conditions that varied in the amoof interference they created. We reasoned thC.V.’s failure to show poor new associatlearning is attributable to possible interfereeffects in Experiment 1, then under conditiothat minimize interference (Hayman et1992), C.V. should exhibit better learning evwhen retrieval cues are perceptually impovished.

EXPERIMENT 2

The design and procedure of this experimwere adopted from Hayman et al.’s (1992)Hamann and Squire’s (1995) procedureswere modified to suit the particular hypothetested in this experiment. One possible reafor poor new learning in the absence of perctual cues may be that the test cue format (Stence1???) creates a high-interference stion. That is, when the amnesic participant fto produce the correct (studied) target, soother answer is produced to complete thetence because of the implicit retrieval instrtions given to the participant. This spontaously produced (incorrect) response minterfere with the learning of the correct (stu

ied) target, thereby resulting in little new learn-

ananucredderemprethe.Calkneouar

f ththistore

edhentatheing

sicnset th&en

ncece

??Ex-enis

re-enanfir

tlysd-ou

aftewad bthe

Sentence1??? condition in our study did notun-so

asion.en-

iondy-

ob-er-thentndi-thewodureinglly,thein-

eri-Thein-re-te-astialest-w,ing.st-

tionel ofal.,

t tovi-x-osedi-udyrn-edtiveionin-deransia.


ing.In support of this argument, recently Haym

et al. (1992) reported that amnesic participK.C. learned the correct target responses mfaster under the Study Only condition compato the Study-Test condition. Specifically, unthe Study Only condition, the entire study it(e.g., a talkative featherbrain-parakeet) wassented repeatedly without any testing untillast session. In their Study-Test condition, Kwas presented with definition cues (e.g., a tative featherbrain-???) and was asked to geate the target word. Following his spontaneresponse, the correct target (parakeet) appeon the screen. (Note here that because oorder in which the events occurred underretrieval condition, it is more appropriatelabel this condition as Test-Study, and thefore, this condition will henceforth be referrto as the Test-Study retrieval condition.) Tgeneration of a response prior to the presetion of the target within each session led toproduction of incorrect targets, thereby leadto high interference. Under our Sentence1???retrieval condition in Experiment 1, amneparticipant C.V. was asked to produce respoin every session. Thus, one might argue thafailure of R.H. in the previous study (RajaramCoslett, in press) and that of C.V. in the presstudy to learn the appropriate target sentemay be attributable to the high-interferenlearning condition used in our experiments.

It is important to note that the Sentence1retrieval condition used in our experiments (periment 1 in the present study, and Experim1 in Rajaram and Coslett’s (1999) study)somewhat different from the Test-Studytrieval condition just described. Under the Stence1??? retrieval condition, the correctcomplete sentences were always presented(at study) within a session. Subsequenwithin the same session the Sentence1??? cuewere provided for implicit retrieval of the stuied targets. Thus, the amnesic participants instudies did not first generate a responsewhich the correct target was presented, asthe case under the Test-Study condition useHayman et al. (1992). Therefore, although

th

-

.-r-sede

-

-

se

ts

?

t

-dst,

rrsy

eliminate interference to the extent possibleder the Study Only retrieval condition, it aldid not create levels of interference as highthose that occur under the Test-Study conditBecause of the order of events under our Stence1??? retrieval condition, this conditwill henceforth be referred to as the StuImmediate Test condition.

The present experiment was designed totain direct empirical evidence for these diffential levels of interference. Specifically,implicit new learning of amnesic participaC.V. was directly compared across three cotions—the Study Only, the Test-Study, andStudy-Immediate Test conditions. The first tconditions were selected to adopt the proceused by Hayman et al. (1992) for manipulat“intraexperimental” interference. Specificaintraexperimental interference arises fromconditions of the experiment that generateterference effects rather than from preexpmental associations known to participants.third, Study-Immediate Test, condition wascluded to obtain a direct replication of thesults of Experiment 1 with a new set of marials. This comparison of three conditions wexpected to demonstrate whether differenamounts of interference in the learning and ting situations lead to differential levels of neand relatively conceptual, associative learnWe predicted little new learning under the TeStudy learning condition because this condihas been shown to produce the highest levinterference for other amnesics (Hayman et1992; Hamann & Squire, 1995). With respecthe Study-Immediate Test condition, we preously found minimal new learning (present Eperiment 1), and we expected to replicate thfindings. The most interesting learning contion in the present experiment was the StOnly condition. Hayman et al. found that leaing under this condition was greatly facilitatin a severely amnesic participant K.C. Posievidence of new learning under this conditby our amnesic participant C.V. would deleate at least one experimental condition unwhich new verbal, associative learning cubiquitously occur in cases of severe amne

Method

asc wes hatt ure( N-T 20s d bR Ot imui err en-t nst dyI ucl os hea

theveudg aistsd aheingitioon

addi

ina

.s

Paurls)cecenn

d otioskcr

ical stimuli.4 The participants’ task was to com-thatsurergetrit-skedhattici-ach

e ofeencqui-ac-ay,s. Athethe

Theionrderign.was

ameiventic-trolncedteda

ndallin-

st 1f ntrop

fini-ationus an-quireletioner theeirthiselinepro-

dayster-ave

ed inof

t thisprim-


Materials. A new set of 174 sentences wonstructed for this study. These sentencesimilar to those used in Experiment 1 in they were unfamiliar and yet plausible in nate.g., MACHINE refined COPPER, COURY exported PEPPERMINT). Of these, 1entences were taken from the materials useajaram and Coslett (1999, Experiment 3).

hese 120 sentences, 72 served as critical stn this experiment. These 72 sentences wandomly divided into three sets of 24 sences, each to appear under three conditiohe experiment (Study Only, Test-Study, Stummediate Test). We constructed three sists to completely counterbalance the setsentences across conditions for both Matcnd Young Control participants.3

Of the remaining 102 sentences fromoverall set of sentences, 6 were used to serprimacy and recency buffers in the three stlists just described with one each appearinthe beginning and end of each of the three lOf the remaining 96 sentences, 24 servefillers in the test (described shortly) for tStudy-Immediate Test condition. The remain72 sentences served as fillers in a recogntask that was conducted at the end to obtainof several measures of explicit memory. Indition to these 174 sentence stimuli, 72 adtional words were selected to serve as foilsforced-choice recognition task that servedone of several measures of explicit memory

Design and procedure.Each participant watested over three days in this experiment.ticipants (amnesic participant C.V., foMatched Controls, and nine Young Controwere tested individually. The design and produre were adopted (and modified where nesary) from Hayman et al.’s (1992) and Hamaand Squire’s (1995) studies. Day 1 consistebaseline measurements for target complefor the 72 critical stimuli. The first baseline taconsisted of Sentence1??? cues for the 72

3 Two Matched Control participants were tested on Lior reasons described in footnote 5. One Matched Coarticipant each was tested in Lists 2 and 3.

re

yflie

of-hfd

asyt.s

ne

--as

r-

-s-

fn

it-

plete the sentence cues with the first wordcame to mind. The second baseline meaconsisted of fragmented versions of the tawords (presented in isolation) from the 72 cical sentences, and the participants were ato complete the fragment with the first word tcame to mind. In both baseline tasks, parpants were given a maximum of 15 s for ecompletion.

A gap of at least 1 day (6 days in the casamnesic participant C.V.) intervened betwthe baseline measurement and the next, asition, phase of the experiment. The entirequisition phase was carried out within 1 dwith a gap of 30 min to 3 h between sessiontotal of six sessions was conducted duringacquisition phase. Each session includedthree learning conditions described below.presentation of conditions within each sesswas blocked and was counterbalanced for oacross sessions with a Latin Square desAcross sessions, each set of 24 sentencesnested within each condition such that the sset of 24 sentences was presented within a gcondition in all six sessions for amnesic paripant C.V. For the Matched and Young Conparticipants, sentences were counterbalaacross learning conditions but were neswithin the condition in all six sessions forgiven participant. Following Hamann aSquire’s (1995) encoding instructions, underthree conditions, participants were asked to

l

4 Note that according to Hayman et al.’s (1992) detions, this procedure may create an interference situbecause participants are asked to produce spontaneoswers that may be incorrect. However, Hamann and S(1995) used this procedure to obtain the baseline compmeasure and subsequently found sizeable learning undminimal interference condition, i.e., Study Only, in thgroup of amnesic participants. Thus, precedence formethod already exists in this literature. Because the basmeasurement in our experiment required spontaneousduction of responses only once, and was collected sixprior to the learning phase for C.V., the influence of inference from the baseline procedure, if any, may hdissipated. The baseline intrusion rates for C.V. describfootnote 5 confirm this prediction. Finally, our selectionthis baseline measure was also guided by the fact thameasure has the obvious advantage of providing cleaning measures in our small-Ndesign.

dicate on a scale from 1–5 how much sense eachf (1) aasm

-t acs r thr teo ionw ent

-c ent firsp wea anp wap tint ai

-m cesw ndt itiou res tint dyb tionw 1?f tudi ler,n ves ncC

icita thef n 5m

of7 di-t doo eda ntea rep pa

ticipants were asked to write down for each cue

toim-andar-en-.

en-d to

beizeic

ni-ree

ow-wasatedovecue.se-eded

eri-p of

these-

of

ndtud-Thisest

n-72

hereeredied

if-oreas

osts,sk,


sentence made to them on a 5-point scale overy little, (2) a little, (3) some (average), (4fair amount, (5) a lot. A maximum of 15 s wallowed for each decision. Participants copleted the task quickly and efficiently.

—Study Onlycondition: This condition conained 24 of the 72 critical sentences. Eentence was presented one at a time foating task. No intervening test was conducn these materials until Day 3. This conditas assumed to create minimal intraexperim

al interference.—Test-Studycondition: This condition in

luded another set of 24 of the 72 critical sences. For each item, participants wereresented with the Sentence1??? cue andsked to guess the target. After the participrovided the response, the intended targetresented with the sentence cue for the ra

ask. This condition was assumed to create mmal intraexperimental interference.

—Study-Immediate Testcondition: The reaining set of 24 of the 72 critical sentenas presented under this condition. This co

ion is the same as the Sentence1??? condsed in Experiment 1. Participants were pented with the entire sentence for the raask. After the presentation of three stulocks, the 24 sentences from this condiere once again presented in the Sentence

ormat to measure new learning. These 24 sed sentences were intermixed with 24 filonstudied sentences (the same set in eession) in order to simulate the interfere.V. experienced in Experiment 1.

After the acquisition phase, tests of implnd explicit memory were conducted on

ollowing day. The participants were tested oeasures presentedin the following order.

—Implicit production:This test consisted2 studied items from all three learning con

ions described above presented in a ranrder. In addition, four buffer cues not usnywhere else in the experiment were preset the beginning of this list. These stimuli weresented in the Sentence1??? format and

)

-

he

d

-

-tre

tssgx-

i-n

-g

??-

rye

m

d

r-

the first response that came to mind.—Explicit recall: This test was identical

the implicit production test except that the stuli were presented in a new random orderdifferent retrieval instructions were given. Pticipants were asked to complete each Stence1??? cue with the studied target word

—Recognition test for studied items.Partici-pants were presented with all 72 studied stences in a new random order and were askecircle YES for the items they believed tostudied. This test was conducted to maximthe familiarity of the stimuli for the amnesparticipant.

—Two-alternative forced-choice recogtion: Each of the 72 sentences from the thstudied conditions was presented in the folling fashion. The sentence cue of the testpresented with the target as well as an unrelfoil. The two alternatives were arranged aband below, and to the right of, the sentenceThe position of the target was randomlylected across stimuli. The foils were derivfrom a different pool of items and were not usanywhere else in this experiment or in Expment 1. These foils were normed on a grou10 college undergraduates to ensure thattargets and foils were equally likely to belected as targets prior to study (selectiontargets versus foils,t(9) 5 1.14,SE5 1.67). Inthe forced-choice recognition task, C.V. acontrol participants were asked to pick the sied target that matched with the sentence.test is similar to the recognition memory tused by Hamann and Squire (1995).

—Yes/No recognition:The 72 studied setences were intermixed with a new set offiller, nonstudied sentences (not used anywelse in the experiment), and participants wasked to select the sentences they had stuearlier.

Explicit memory was tested in the four dferent ways just described in order to explvarious aspects of C.V.’s performance. It wexpected that explicit recall would be the mdifficult task for C.V. The recognition taskparticularly the forced-choice recognition ta

were considered to be easier than the otheret

ionr th

t:dinweverents,acto

thio

onhassuetsr thonduon72ea

r aerost,dntldelin

portain

than

nottedentanhe

V.,os.ed

al,ndi-ri-

esicndneledud-ed

. Inasndis

ngm-ndsico beatee tori-of

entce

ingeee-mics

TABLE 2

pantC trols(

dard

.

68

87

03


explicit tasks (Freed & Corkin, 1988; Freedal., 1987), and the forced-choice recognittask was expected to show some memory fonewly learned stimuli.

—Implicit word fragment completion tesThis test was conducted 4 h after the preceseries of tests were completed by C.V. asas the control participants. The fragmentedsions of target words from the 72 studied stences were intermixed with filler fragmenand participants were asked to complete efragment with the first solution that camemind. No sentence cues were included intest, and participants were given a maximum15 s to complete each fragment.

Results and Discussion

In the implicit production task, the proportiof correct targets produced in the baseline pserved as the nonstudied items and weretracted from the proportion of studied targproduced in the test phase on Day 3. Neitheamnesic participant C.V. nor the Matched Ctrols produced any of the correct responsesing the baseline period. Among the Young Ctrols, only two participants produced 1 out ofcorrect completion each in the baseline msure. Thus, the baseline completion rate foparticipants taken together was close to zFor the implicit word fragment completion tethe baseline completion rate was assesseDay 1 for the targets that were subsequestudied in the sentence context (on Day 2) unthe three conditions just described. The basecompletion rate was subtracted from the protions of fragments completed on Day 3 to obthe measure of standard priming. Becausesame set of items served as nonstudiedstudied items in this experiment, we didadjust for baseline differences with adjuspriming as in Experiment 1. Thus, we presthe data only for the more conservative, sdard, priming measure in this experiment. Tmean fragment completion rates for C.Matched Controls, and Young Controls acrdifferent conditions are displayed in Table 2

The implicit new learning test was design

e

gll--

h

sf

eb-

e-r--

-ll.

onyre-

ed

t-

s

to measure the extent of implicit new verbassociative learning under three learning cotions that varied in the amount of intraexpemental interference. The data from the amnparticipant C.V., the Matched Controls, aYoung Controls are displayed in the left paof Fig. 2. Amnesic participant C.V. showpoor learning under all three conditions, incling the Study Only condition that was presumto create the lowest amount of interferencefact, C.V.’s production of correct targets wequivalent under the Study Only (0.08) aTest-Study conditions (0.08). This patternquite different from the advantage in primireported by Hayman et al. (1992) for their anesic participant K.C. and by Hamann aSquire (1995) for their group of nine amneparticipants. C.V.’s performance appeared tnumerically higher under the Study-ImmediTest condition (0.17) that is most comparablthe new verbal learning condition of Expement 1 and is presumed to be the conditionmoderate (not high) interference in the presexperiment. However, this level of performanwas still quite low such that standard primscores (studied2 nonstudied) across the thrlearning conditions, Study Only, Study-Immdiate Test, and Test-Study did not differ froone another,X2 5 0.99. Predictably, amnesparticipant C.V.’s implicit new learning wa

Mean Response Probabilities of Amnesic Partici.V. and Mean Response Probabilities of Young Con

N 5 9) and Matched Controls (N5 4) in the WordFragment Completion Task in Experiment 2 (The StanDeviations Are Presented in Parentheses)

Youngcontrols

Matchedcontrols C.V

Study onlyStudied 0.29 0.26 0.4Nonstudied 0.07 0.10 0.0

Study-Immediate TestStudied 0.33 0.27 0.3Nonstudied 0.08 0.07 0.1

Test-StudyStudied 0.24 0.28 0.5Nonstudied 0.05 0.07 0.1

tednd

owthedi-2,

alim-anndallharlyskrag

tic-orre

askf 8

theereconheIn

iddy

M -

( re

antr medp theS firstg thed tegys thats pro-v thes s herp oftenc onse.T owert gherl e int tences thisp ednSs r thec par-t tici-p e re-m orec ta int

trolp ipant

ongnelning).


substantially lower than the learning exhibiby the control participants, both Matched aYoung, such that C.V.’s performance fell belthe 95% confidence intervals for each ofcontrol groups under all of the training contions in the implicit production task (see Fig.left panel).

As expected, the data from the explicit rectest (Fig. 2, right panel) revealed grosslypaired performance from amnesic participC.V. under all three learning conditions, athey fell outside the range of the explicit recperformance of Young Controls as well as tof Matched Controls. Also, it took C.V. nea15 min to perform this explicit recall tawhereas the control participants took an aveof 9 min and 40 s (Young Controls average5 9min, 20 s; Matched Controls average5 10 min)to complete this task. This difference is parularly striking in light of the duration data fthe previous, implicit production, task wheC.V. took 7 min and 30 s to complete the tand control participants took an average omin and 38 s (Young Controls average5 8 minand 26 s; Matched Controls average5 8 min50 s).

The data from the control participants forStudy Only and Test-Study conditions wcompared to determine whether these twoditions led to differential performance in timplicit production and explicit recall tasks.

FIG. 2. The correct target production in Experimcontrol groups. The left panel presents the data(studied–nonstudied) for the three learning condpresents the explicit recall data as the proportionsconditions with varying amounts of interference. (

l

t

t

e

-

the implicit production task, performance dnot differ for either the Young Controls (StuOnly 5 0.48; Test-Study5 0.46) or the

atched Controls (Study Only5 0.54, TestStudy5 0.52) across the two conditions,t , 1for both groups collapsed).5 These data a

5 The performance of one Matched Control participequires some explanation. This participant perforoorly in the implicit production task, particularly undertudy Only (0.13) and Test-Study (0.17) conditions. Atlance, these data are problematic. However, duringebriefing session, this participant explained the strahe used in the implicit production task. She explainedhe liked some other completions better than the onesided in the learning sessions. At test, when she sawentence cue, both the studied response as well areferred response came to mind simultaneously. Shehose her preferred response over the studied respherefore, her production of studied responses was l

han expected. This explanation predicts a much hievel of performance in the explicit recall task becaushis test participants were asked to complete the sentems with the studied responses. Results confirmedrediction as this Matched Control participant performear ceiling in the explicit recall task (Study Only5 0.88,tudy-Immediate Test5 0.96, and Test-Study5 0.83). Ithould be noted that none of the results reported foontrol participants, either averaged across all controlsicipants or averaged only for the Matched Control parants, change when the data from this participant aroved from the analyses. Therefore, we took the m

onservative approach of including this participant’s dahe analyses.

An additional concern pertaining to this Matched Conarticipant’s performance may be that amnesic partic

t 2 is displayed for amnesic participant C.V. and the twr the implicit production task as proportions of primis with varying amounts of interference. The right pa

corrected recall (studied–nonstudied) for the three learntrol-Y, Young Controls; Control-M, Matched Controls

enfo

itionofCo

similar to those reported by Hayman et al.hothetheandandtic-esnlyved95udli-

ci-&

al.ou

nethreInerth

theerhaToomriorder

isnslizc-ourex

pect to obtain an advantage for the Test-Studyithheet-est-dy

0 ofH t thea urei as-i ini

fort aseso thei ctp theu ory-i ks.S s,m in-v ayt icita l asi toe d ina (Ha-m 92).O hei tot icita notc oryp n-t iona

thatto

entog-rstces

rityincesasof

dusk.

esicurero-

thee eT gep %)a ingc ionf


(1992) for their four control participants walso exhibited equivalent performance inimplicit production of correct targets underStudy Only and Test-Study conditions (0.850.83, respectively). In contrast, HamannSquire (1995) reported that their control paripants produced more targets under the TStudy condition compared to the Study Ocondition (0.77 and 0.61, respectively, derifrom their graph). Hamann and Squire’s (19results showing an advantage for Test-Stcondition in memory-intact participants repcate prior findings with memory-intact partipants (e.g., Carrier & Pashler, 1992; DarleyMurdock, 1971; Runquist, 1986; Wenger et1980). The apparent discrepancy betweenfindings and Tulving et al.’s findings, on ohand, and those of Hamann and Squire’s, onother, can be resolved by considering thetrieval instructions given to all participants.Hayman et al.’s study and our present expment, participants were asked to producefirst word that came to mind, whereas inHamann and Squire study, participants wasked to “produce the target words theystudied on previous sessions” (p. 1032).gether, two implications can be derived frthis overall pattern of findings. One, supememory in memory-intact participants unthe Test-Study condition in previous reportsattributable to the explicit retrieval instructiogiven to participants, and it does not generato implicit production in memory-intact partiipants. Two, this pattern predicts that insubsequent explicit recall task, we should

C.V. also had preferred responses that blocked his protion of correct responses in the implicit production taThis concern is allayed by the examination of amnparticipant C.V.’s intrusion rates from the baseline measC.V. produced 14% baseline intrusions in the implicit pduction task (Study Only5 17%, Study-Immediate Test58%, Test-Study5 17%), and 19% baseline intrusions inxplicit recall task (Study Only5 25%, Study-Immediatest 5 21%, Test-Study5 12%). Thus, there were larroportions of items in the implicit production task (86nd the explicit recall task (81%) in which new learnould have taken place without the interference of intrusrom preexperimental associations.

t-

)y

,r

e-

i-e

ed-

e

-

items over Study Only items. Consistent wthis prediction, in the Explicit Recall task, tcontrol participants exhibited significantly bter recall for targets learned under the TStudy training condition than under the StuOnly training condition,t(12) 5 2.53, SE 5.03. This pattern is similar to the findingsamann and Squire (1995) and suggests thadvantage for the Test-Study training proced

n memory-intact participants may be more ely obtained in explicit memory tasks thanmplicit tasks.

One possible reason that the advantagehe Study Only procedure is seen in many cf amnesia (though not ours) but not under

mplicit retrieval condition for memory-intaarticipants may partially be attributable tose of explicit memory access used by mem

ntact participants even in the implicit taspecifically, with multiple training sessionemory-intact participants may more easily

oke explicit access to study materials than mhe amnesic participants. Although this explccess may only be partial and not as willfu

n the explicit recall task, it may be sufficientliminate the Study Only advantage observemnesic participants by other researchersann and Squire, 1995; Hayman et al., 19ur own results with control participants in t

mplicit production task may be susceptiblehis possibility, but we do note that such explccess in our implicit production task wasomplete because the overall level of memerformance improved significantly for the co

rol participants across the implicit productnd explicit recall tasks,t(12) 5 3.38, SE 5

0.07.On the three recognition memory tasks

followed, C.V.’s performance was expectedimprove because of the familiarity componof recognition. The data from the three recnition tasks are displayed in Fig. 3. In the firecognition task, where all 72 studied sentenwere presented to provide maximum familiafor performance (the left-most of the barsFig. 3), C.V. recognized 69% studied sentencorrectly. Although C.V.’s performance wreasonably good, it still fell outside the range

c-

.

s

olsor, itin

par6 s;

Mc

ter-n seo odr ingt hoi ogn 8;F as edp timt pef toc f 4m ;M .C al-i eso eno hef r

constructed logical sentence completions sucha ncep ”).

ofy s inF ec-o pro-dv t-i ci-p em-oaf tet arliert toc tici-p ngCm

ea-s byo ied–n con-d ub-s ts(

troltrolFig.tedvene to

oftionrtic-

barsl stim-ng

esicffer2 isognsenanedts thrninl-Y,


recognition performance of Young Contr(97% correct) and Matched Controls (99% crect), who performed at ceiling. Furthermoreis interesting to note that it took C.V. over 7 mto complete this task, whereas the controlticipants took an average of 3 min and 2(Young Controls average5 3 min 33 s

atched Controls average5 3 min 20 s) forompletion.On the second recognition task of two-al

ative forced-choice recognition (the middlef bars in Fig. 3), C.V. showed remarkably goecognition memory (91% correct), supporthe extant data that amnesic participants smproved performance on forced choice recition memory tasks (Freed & Corkin, 198reed et al., 1987). But C.V.’s performance womewhat below that of controls, who obtainerfect scores on this task. Once again, the

aken to complete the task reflected poorerormance on C.V.’s part (18 min) comparedontrol participants, who took an average oin (Young Controls average5 3 min, 40 satched Controls average5 4 min, 20 s).V.’s report following this test was also reve

ng regarding his level of conscious awarenf the study materials. C.V. remarked at thef the forced-choice recognition test that

ound this task easy becausehe had earlie

FIG. 3. The recognition memory performance of amnparticipant C.V. and two control groups across three dient measures of recognition memory in Experimentdisplayed. The left-most panel presents the correct rection of the 72 studied sentences. The middle panel prethe correct recognition of the sentences where participhad to choose between the correct target and a foil placfront of the sentence cue. The right-most panel presenhits–false alarm data collapsed across the three leaconditions from the yes/no recognition task. (ControYoung Controls; Control-M, Matched Controls).

-

-

t

w-

s

er-

sd

s “the rabbi and the turkey” (studied senteresented to C.V., “Rabbi requested turkeyOn the third recognition memory task

es/no recognition (the right-most set of barig. 3), amnesic participant C.V. correctly rgnized 70 out of 72 sentences, and heuced 7 out of 72 false alarms, yielding ad9alue of 3.16 and aB of 1.47, and demonstrang good recognition memory. Control partiants showed nearly perfect recognition mry (Young Controls, hits5 100%, falselarms5 1%; Matched Controls, hits5 100%,alse alarms5 0). The time taken to complehis task showed the same pattern as the easks such that C.V. took 8 min and 50 somplete the tasks whereas the control parants took an average of 5 min 41 s (Youontrols5 5 min 53 s; Matched Controls5 5in 30 s).The status of perceptual priming was m

ured with a word fragment completion taskbtaining the standard priming scores (studonstudied). Collapsed across three studyitions, amnesic participant C.V. produced stantial priming for the studied targestudied5 32, nonstudied5 9). C.V.’s perfor-

mance either was within the range of the conparticipants or even exceeded that of conparticipants across different conditions (see4). It is also relevant to note that C.V. exhibiperceptual priming for learned targets ewhen he failed to produce them in respons

FIG. 4. The perceptual priming data in the formadjusted priming scores from the word fragment completask in Experiment 2 are displayed for the amnesic paipant C.V. and the two control groups. The fourth set ofpresent the average adjusted priming scores across aluli from all three learning conditions. (Control-Y, Youcontrols; Control-M, Matched Controls).

-

i-tstsineg

sentence cues in the implicit production task

nodi-ndiasinn ent.’s

slyr-ais.’sth

onthce9a

ingon

hep-

ino

tive.Vfor

ievionro-ptuer tnce.Canaslv-ceitioir-ce

esiss,

Experiments 1 and 3) and quite different fromsal&

antn-for-i-thent 1in

instedtalns,st-ent

ingiont ofudy

oferitybal

ofairsso-

ndse-raf,im-

.H.the

tolityodsia.u-

vel,thanects

nglyesiaer-Foringwas


and explicit recall tasks.Taken together, C.V.’s performance did

improve under the Study Only learning contion compared to under the Test-Study cotion even though the former condition wshown to be more beneficial for learninganother densely amnesic participant (Haymaal., 1992) and a group of amnesic participa(Hamann and Squire, 1995). Predictably, C.Vexplicit recall was also found to be grosimpaired. C.V.’s recognition memory perfomance was found to be reasonably goodthough it took C.V. much longer to perform thtask compared to control participants. C.Vcontrasting pattern of performance acrossimplicit production and explicit recall tasks,one hand, and the recognition tasks, onother, is similar to the pattern of performanrecently reported by Isaac and Mayes (1991999b) in a group of amnesics for the forgettrates in free recall and cued recall tasks, onhand, and the recognition task, on the otFinally, C.V. exhibited normal levels of percetual priming for the studied targets.

GENERAL DISCUSSION

Two experiments were conducted to examthe role of different retrieval cues and typesinterference in new conceptual associalearning in a severely amnesic participant, CExperiment 1 revealed substantial primingrepeated, novel sentences when the retrcues either provided perceptual informatonly for the targets (perceptual priming) or pvided the sentence frames as well as perceinformation for the targets (Sentenc1Fragment cues). These findings are similathe priming data obtained from the performaof a basal forebrain amnesic participant, Cand a medial temporal lobe amnesic participR.H. (Rajaram & Coslett, in press), as wellfrom a densely amnesic participant, K.C. (Tuing et al., 1991). Critically, C.V.’s performanunder the Sentence1??? retrieval cue condwas grossly impaired. This pattern of impament is similar to the impaired performanobserved for the medial temporal lobe amnparticipant R.H. (Rajaram & Coslett, in pre

t

-

ts

l-

e

e

,

er.

ef

.

al

al

o

.t,

n

c

the substantial priming produced by the baforebrain amnesic participant C.C. (RajaramCoslett, in press) and the amnesic participK.C. (Tulving et al., 1991) with the Setence1??? retrieval cues. C.V.’s poor permance under the Sentence1??? retrieval condtion is particularly remarkable becauseStudy-Test sessions in the present Experimewere presented in quick succession, unlikethe previous studies described here.

In Experiment 2, the role of interferencenew conceptual associative learning was teby manipulating the levels of intraexperimeninterference across three learning conditioStudy Only, Study-Immediate test, and TeStudy. Unlike previous studies, in the presexperiment we failed to find improved learnin C.V.’s performance even under the conditthat presumably created the least amounintraexperimental interference, i.e., the StOnly learning condition.

An obvious and reasonable interpretationthe present data pertains to the role of sevof amnesia in mediating learning of new verassociations. Previous studies on primingnew associations with unrelated word-p(Graf & Schacter, 1985) showed that new asciation priming could be obtained in mildly amoderately amnesic participants but not inverely amnesic participants (Schacter & G1986). One possibility may be that grosspairments in priming with the Sentence1???retrieval cues for the amnesic participants R(Rajaram & Coslett, in press) and C.V. (present experiments) may be attributabletheir severe amnesia. Similarly, C.V.’s inabito benefit from the Study Only learning methmay also be attributable to his severe amne

Undoubtedly, new verbal learning, particlarly at the conceptual and semantic lewould be impaired more in severe amnesiain moderate amnesia. However, several aspof the data from these and other studies strosuggest that differences in severity of amncan not solely account for the patterns of pformance exhibited by amnesic participants.instance, the first evidence of positive learnunder the Sentence1??? retrieval condition

reported for a severely amnesic participant,iedulvm-arnitiotheo thana-

ri-oolonm,t inia

oun

them

iat-erof.Hammethiontic-ar-als,9avehi

n o

ateobeewpaanto

nsie-

anl.,

1991). Two, in a recent study, Vargha-Khademheretingnedne-dicandgeesen beow-oryg-ithpar-ls.rldichum-ra-tedvell in-tichat

en-ess-

ayin

resub-un-al.,ngtiond bysis

truc-ofni-

per-nsingry-the


K.C., who failed to recognize any of the studsentences after several trials of learning (Ting et al., 1991). Similarly, basal forebrain anesic participant C.C. showed substantial leing under the Sentence1??? retrieval condeven though her recognition memory forsame sentences was poor and comparable tof the medial temporal lobe amnesic participR.H., who failed to show such learning (Rjaram & Coslett, in press). Finally, in Expement 2 in the present series, C.V. exhibited plearning in response to Sentence1??? retrievacues, but his recognition memory was reasably good for the same set of stimuli. In sualthough severity of amnesia may accounlarge part for the failure of new verbal, assoctive learning, as measured by the Sentence1???retrieval cues, this explanation does not accfor all of the data in the literature.

An alternate interpretation may be thatmedial temporal lobes including the hippocapal system may be heavily involved in meding new verbal, associative learning where pceptual support at test is limited. The failuremedial temporal lobe amnesic participant Rsupports this interpretation. Because C.V.’snesia also appears to be a result of presuhippocampal damage, the results frompresent study also support this interpretatSimilarly, medial temporal lobe amnesic paripant H.M.’s failure to produce the correct tgets under the Sentence1??? conditionsfavors this notion (in Squire & Knowlton1995). Recently, Isaac and Mayes (1991999b) and Aggelton and Brown (1999) haalso proposed that damage to the extendedpocampal system disrupts the consolidatiocomplex associations.

However, three lines of evidence complicthe interpretation that the medial temporal lregions play a large role in the formation of nverbal associations. One, densely amnesicticipant, K.C., who did show successful semtic learning under conditions very similarthose tested here, reportedly sustained extebrain injury that also included some of the mdial temporal lobe regions bilaterally (Haymet al., 1992; Kohler et al., 1997; Tulving et a

-

-n

att

r

-

-

t

-

-

.-d

e.

o

,

p-f

r--

ve

et al. (1997) reported three case studies wthe participants exhibited amnesia resulfrom bilateral hippocampal pathology sustaiin early childhood. In all three cases, the amsic participants displayed debilitating episomemory deficits but had acquired speechlanguage skills, literacy, and factual knowledwithin the low-average to average range. Thfindings suggest that semantic memories caacquired despite hippocampal damage. Hever, it is also relevant to note that in laborattests of multitrial (10 trials in each task) reconition tasks of novel associations (faces wvoices, objects with places), these amnesicticipants performed poorly relative to controThus, the learning achieved in the real womay be attributable to multiple sources of rencoding that may have far exceeded the nber of repetitions typically tested in the labotory settings. Three, in Experiment 2 reporhere, we found that C.V. failed to acquire noverbal associations even under the minimaterference condition. This finding is problemaat least for the strong version of the claim tthe hippocampus and the related structuresable learning of novel associations by suppring interfering stimuli.

A possible resolution of these findings memerge with a more precise considerationhuman participants of different substructuwithin the medial temporal lobes that may sserve different cognitive components thatderlie new learning (see also Mishkin et1998). At the cognitive level, the streamliniacross laboratories of acquisition and retentasks, and the putative processes tappethese tasks, will greatly facilitate such analyand provide a better assessment of the sture–function mappings of implicit learningnovel verbal information in the human cogtive system.

Implications for Normal Cognition

The results from the present series of eximents provide a window into the contributioof the implicit memory processes in mediatnew verbal associative learning in a memointact cognitive system. As demonstrated by

extant data, new associative learning at a con-theitriar-s tom

og,uch

bim

hee

et94

erearnst

ncenc

tacor

arn-asimeithnd

f the

bain-s oo

tinsealdtsneansen-ly,onb

aysk

even though they did exhibit the predicted Test-re-

r thery-s insem-mal

lsorn-and(asant,, invalup-tionast,cep-orup-le.eenas-

s notell,

ewaitive

ry,axis.

ongriatey-

ea-r a

,

ory.ci-

velIn P.


ceptual or semantic level (as required bySentence1??? retrieval cues) requires multtraining. Multiple trials in memory-intact paticipants present the danger of explicit accesthe learned material. Such contamination cplicates our understanding of the exact roleimplicit memory in supporting new learninand the exact conditions under which slearning may occur. These problems maycircumvented by assessing the effects ofplicit learning in cases of severe amnesia.

The findings with dense amnesics from otstudies (Hamann & Squire, 1995; Haymanal., 1992; Rajaram & Coslett, 1991; Tulvingal., 1991; see also Van der Linden et al., 19Verfaellie, et al., 1995) and our findings hdemonstrate that new verbal associative leing is slow and laborious in amnesia. Intereingly, this pattern in the amnesic performafor verbal learning bears a strong resemblato the pattern observed in the memory-inpopulation in studies of implicit learning frule-governed sequences of stimuli such astificial grammar learning and probability leaing (Reber, 1989), the serial reaction time t(Stadler, 1989), and the choice reaction ttask (Lewicki et al., 1987). In these studies wthe memory-intact population, learning is fouto occur without any conscious awareness orules and typically requires several (sometimhundreds) of trials. In studies of new verlearning in the memory-intact population,cluding the control data in the present serieexperiments, learning seems to occur at a mrapid rate and with some awareness, indicathat it is much more difficult to eliminate the uof explicit memory in new learning of verbstimuli. Our comparison of new implicit anexplicit learning in memory-intact participanprovided the first direct comparison within ostudy of performance under the Test-StudyStudy Only conditions in Experiment 2. Thefindings revealed likely effects of explicit cotamination on implicit learning. Specificalthe memory-intact participants did not demstrate the Study Only advantage exhibitedother amnesics (Hamann & Squire, 1995; Hman et al., 1995) in the implicit production ta

l

o-f

e-

rt

;

--

et

r-

k

esl

freg

d

-y-

Study advantage in the explicit recall task. Psumably, the advantage amnesics show foStudy Only condition dissipates in memointact participants because of explicit accesmultitrial learning situations. Together, thefindings underscore the contribution of the anesic data toward understanding the norcognitive functions.

Our findings reported in this manuscript adelineate the differences between implicit leaing of new associations at a perceptual levela conceptual level. Amnesic participant C.V.well as another severely amnesic participR.H., we reported earlier, Rajaram & Coslettpress) failed to show any learning with retriecues that provided only partial perceptual sport and required access to learned informaat the conceptual or semantic level. In contrthese participants exhibited substantial pertual priming for single words as well as fnovel associative stimuli when perceptual sport for the to-be-retrieved item was availabThese results suggest a dissociation betwperceptual and conceptual binding of verbalsociations. Because perceptual binding doeseem to guarantee conceptual binding as wan understanding of implicit acquisition of ninformation in normal cognition will requiresystematic analysis of these separable cogncomponents and mechanisms.

REFERENCES

Aggleton, J. P., & Brown, M. W. (1999). Episodic memoamnesia, and the hippocampal anterior thalamicBehavioral and Brain Sciences,22, 425–489.

Blaxton, T. A. (1989). Investigating dissociations ammemory measures: Support for a transfer appropprocessing framework.Journal of Experimental Pschology: Learning, Memory and Cognition,15, 657–668.

Blaxton, T. A. (1992). Dissociations among memory msures in memory-impaired subjects: Evidence foprocessing account of memory.Memory & Cognition20, 549–562.

Blaxton, T. A. (1995). A process-based view of memJournal of the International Neuropsychological Soety,1, 112–114.

Bowers, J., & Schacter, D. L. (1993). Priming of noinformation in amnesic patients: Issues and data.Graf & M. E. J. Masson (Eds.),New directions in

cognition, development, and neuropsychology(pp.

sic

val

C 995toxex

C er-ts in

C meolic.

C ofrforplel-

C au,ofa-

C -:

C 97)theR.

c.

C , Jip-for

ipleb-

anso-,

gydo-rtem-n

ry:R.

c

ee

Diamond, R., & Rozin, P. (1984). Activation of existing-

re-esic

&hol-en-.woory.in

e,

ting

m-y-

heing

heof

ory

&m--

in,ired

3).f-

task.

ionor-

rn-in

).Ac-

tionms.

ion


303–326).Brooks, D. N., & Baddeley, A. (1976). What can amne

patients learn?Neuropsychologia,14, 111–122.Carrier, M., & Pashler, H. (1992). The influence of retrie

on retention.Memory & Cognition,20, 633–642.endes, F., Andermann, F., Carpenter, S., et al. (1

Temporal lobe epilepsy caused by domoic acid inication: Evidence for glutamate receptor-mediatedcitotoxicity in humans.Annals of Neurology,37, 123–126.

ermak, L. S., & Butters, N. (1972). The role of interfence and encoding in the short-term memory deficiKorsakoff’s patients.Neuropsychologia,10, 89–95.

ermak, L. S., Butters, N., & Moreines, J. (1974). Soanalysis of the verbal encoding deficit of alcohKorsakoff’s patients.Brain and Language,1, 141–150

ermak, L. S., Hill, R., & Wong, B. (1998). Effectsspacing and repetition on amnesic patients’ pemance during perceptual identification, stem comtion, and category exemplar production.Neuropsychoogy,12, 65–77.

ermak, L. S., Verfaellie, M., Milberg, W. P., LetourneL. L., & Blackford, S. (1991). A further analysisperceptual identification priming in alcoholic Korskoff patients.Neuropsychologia,29, 725–736.

ohen, N. J., & Eichenbaum, H. E. (1993).Memory, amnesia, and the hippocampal system.Cambridge, MAMIT Press.

ohen, N. J., Poldrack, R. A., & Eichenbaum, H. E. (19Memory for items and memory for relations inprocedural/declarative memory framework. In A.Mayes & J. J. Downes (Eds.),Theories of organiamnesia: A special issue of memory(pp. 131–178)UK: Psychology Press.

ohen, N. J., Ramzy, C., Hut, Z., Tomaso, H., StruppErhard, P., Anderson, P., & Ugurbil, K. (1994). Hpocampal activation in fMRI evoked by demanddeclarative-memory based bindings of multstreams of information.Society for Neuroscience Astracts,20, 1290.

Cohen, N. J., & Squire, L. R. (1980). Preserved learningretention of pattern-analyzing skill in amnesia: Disciation of knowing how and knowing that.Science210,207–210.

Corsellis, J. A. N., & Bruton, C. J. (1983). Neuropatholoof status epilepticus in humans. In A.V. DelgaEscueta, C.G. Wasterlain, D. Treiman, & R.J. Po(Eds.),Status epilepticus: Mechanisms of brain daage and treatment(pp. 129–139). New York: RavePress.

Curran, T., & Schacter, D. L. (1997). Implicit memoWhat must theories of amnesia explain? In A.Mayes & J. J. Downes (Eds.),Theories of organiamnesia(pp. 37–47). UK: Psychology Press.

Darley, C. F., & Murdock, B. B. (1971). Effects of prior frrecall testing on final recall and recognition.Journal ofExperimental Psychology,91, 66–73.

).--

--

.

.,

d

r

memories in anterograde amnesia.Journal of Abnormal Psychology,93, 98–105.

Dopkins, S., Kovner, R., & Goldmeier, E. (1994). Fquency judgments for semantic categories in amnand normal controls.Cortex,30, 127–134.

Dusoir, H., Kapur, N., Byrnes, D. P., McKinstry, S.,Hoare, R. D. (1990). The role of diencephalic patogy in human memory disorder: Evidence from a petrating paranasal brain injury.Brain, 113,1695–1706

Eichenbaum, H., Otto, T., & Cohen, N. J. (1994). Tfunctional components of the hippocampal memsystem.Behavioral and Brain Sciences,17, 449–518

Freed, D. M., & Corkin, S. (1988). Rate of forgettingH.M.: 6-month recognition.Behavioral Neuroscienc102,823–827.

Freed, D. M., Corkin, S., & Cohen, N. J. (1987). Forgetin H.M.: A second look.Neuropsychologia,25, 461–471.

Gabrieli, J. D. E. (1995). A systemic view of human meory processes.Journal of the International Neuropschological Society,1, 115–118.

Gabrieli, J. D. E., Cohen, N. J., & Corkin, S. (1988). Timpaired learning of semantic knowledge followbilateral medial temporal-lobe resection.Brain andCognition,7, 157–177.

Gabrieli, J. D. E., & Keane, M. M. (1988). Priming in tamnesic patient H.M.: New findings and a theoryintact and impaired priming in patients with memdisorders, Society for Neuroscience Abstracts,14,1290.

Gabrieli, J. D. E., Keane, M. M., Zarella, M. M.,Poldrack, R. A. (1997). Preservation of implicit meory for new associations in global amnesia.Psychological Science,8, 326–329.

Gabrieli, J. D. E., Milberg, W. P., Keane, M. M., & CorkS. (1990). Intact priming of patterns despite impamemory.Neuropsychologia,28, 417–427.

Gardner, H., Boller, F., Moreines, J., & Butters, N. (197Retrieving information from Korsakoff patients: Efects of categorical cues and reference to theCortex,9, 165–175.

Glisky, E. L., & Schacter, D. L. (1988). Long-term retentof computer learning by patients with memory disders.Neuropsychologia,26, 173–178.

Glisky, E. L., Schacter D. L., & Tulving, E. (1986a). Leaing and retention of computer-related vocabularyamnesic patients: Method of vanishing cues.Journal ofClinical and Experimental Neuropsychology,8, 292–312.

Glisky, E. L., Schacter, D. L., & Tulving, E. (1986bComputer learning by memory-impaired patients:quisition and retention of complex knowledge.Neuro-psychologia,24, 313–328.

Gordon, B (1988). Preserved learning of novel informain amnesia: Evidence for multiple memory systeBrain and Cognition,7, 257–282.

Goshen-Gottstein, Y., & Moscovitch, M. (1992). Repetit

priming effects for newly-formed and pre-existing as-Soc

citesirn-

ionia-

nd

or-

con-

tionl

2).am-e

T.

.I.-

son.

tantntsci-

emcts.

in

nd

inat

-

ofthe

s.),:

n,ndesia

Kesner, R. P., Hopkins, R. O., & Chiba, A. A. (1992).s onN.

. 2

hetive

M.,he

n-wl-g,

ent

-ion--

7).ry-

.sys-fromls of

8).sys-

m-ules-

m-andE.

:

taskThetex.

Y.calinn-

gy

).e inr’s


sociations. Paper presented at the Psychonomicety, St. Louis, MO.

Graf, P., & Schacter, D. L. (1985). Implicit and explimemory for new associations in normal and amnsubjects.Journal of Experimental Psychology: Leaing, Memory and Cognition,11, 501–518.

Goshen-Gottstein, Y., & Moscovitch, M. (1995). Repetitpriming for newly-formed and preexisting assoctions: Perceptual and conceptual influences.Journal ofExperimental Psychology: Learning, Memory aCognition,21, 1229–1248.

Grossman, M. (1987). Lexical acquisition in alcoholic Ksakoff psychosis.Cortex,23, 631–644.

Hamann, S. B. (1990). Level of processing effects inceptually driven implicit tasks.Journal of Experimental Psychology: Learning, Memory, and Cognition,16,970–977.

Hamann, S. B., & Squire, L. R. (1995). On the acquisiof new declarative knowledge in amnesia.BehavioraNeuroscience,109,1027–1044.

Hayman, C. A. G., Macdonald, C. A., & Tulving, E. (199The role of repetition and associative learning innesia: A case experiment.The Journal of CognitivNeuroscience,5, 376–389.

Hirst, W., Phelps, E., Johnson, M. K., & Volpe, B.(1988). Amnesia and second language learning.Brainand Cognition,8, 105–116.

Honavar, M., & Meldrum, B. S. (1997). Epilepsy. In DGraham, & P.L. Lantos (Eds.),Greenfield’s neuropathology (6th ed.) (pp. 931–971). London: Arnold.

Hopkins, R. O., Gale, S. D., Johnson, S. C., AnderC. V., Bigler, E. D., Blatter, D. D., & Weaver, L. K(1995). Severe anoxia with and without concomibrain atrophy and neuropsychological impairmeJournal of the International Neuropsychological Soety,5, 501–509.

Hopkins, R. O., Kesner, R. P., & Goldstein, M. (1995). Itand order recognition memory in hypoxic subjeBrain and Cognition,21, 180–201.

Isaac, C. L., & Mayes, A. R. (1999a). Rate of forgettingamnesia: I. Recall and recognition of prose.Journal ofExperimental Psychology: Learning, Memory, aCognition,25, 942–962.

Isaac, C. L., & Mayes, A. R. (1999b). Rate of forgettingamnesia: II. Recall and recognition of word listsdifferent levels of organization.Journal of Experimental Psychology: Learning, Memory, and Cognition,25,963–977.

Johnson, M. K., & Chalfonte, B. L. (1994). Bindingcomplex memories: The role of reactivation andhippocampus. In D.L. Schacter & E. Tulving (EdMemory systems 1994(pp. 311–350). Cambridge, MAMIT Press.

Keane, M. M., Gabrieli, J. D. E., Monti, L. A., FleischmaD. A., Cantor, J. M., & Noland, J. S. (1997). Intact aimpaired conceptual memory processes in amnNeuropsychology,11, 59–69.

i-

c

-

,

.

.

Learning and memory in humans with an emphasithe role of the hippocampus. In L. R. Squire &Butters (Eds.),Neuropsychology of memory, Vol(pp. 106–121). New York: Guilford Press.

Klingberg, T., Roland, P. E., & Kawashima, R. (1994). Thuman entorhinal cortex participates in associamemory.NeuroReport,6, 57–60.

Kohler, S., Habib, R., Black, S. E., Szekely, C., Sinden,& Tulving, E. (1997). Cross-modal priming in tdensely amnesic subject K.C.Brain and Cognition,35,420–423.

Lewicki, P., Czyzewska, M., & Hoffman, H. (1987). Uconscious acquisition of complex procedural knoedge.Journal of Experimental Psychology: LearninMemory, and Cognition,13, 523–530.

Martin, E. (1971). Verbal retrieval theory and independretrieval phenomena.Psychological Review,78, 314–332.

Mayes, A. R., Pickering, A., & Fairbairn, A. (1987). Amnesic sensitivity to proactive interference: Its relatship to priming and the cause of amnesia.Neuropsychologia,25, 211–220.

McAndrews, M. P., Glisky, E. L., & Schacter, D. L. (198When priming persists: Long-lasting implicit memofor a single episode in amnesic patients.Neuropsychologia, 25, 497–506.

McClelland, J. L., McNaughton, B. L., & O’Reilly, R. C(1995). Why there are complementary learningtems in the hippocampus and neocortex: Insightsthe successes and failures of connectionist modelearning and memory,Psychological Review,102,419–457.

Mishkin, M., Vargha-Khadem, F., & Gadian, D. G. (199Amnesia and the organization of the hippocampaltem.Hippocampus,8, 212–216.

Moscovitch, M. (1992). Memory and working-with-meory: A component process model based on modand central systems.Journal of Cognitive Neuroscience,4, 257–267.

Moscovitch, M. (1994a). Memory and working with meory: Evaluation of a component process modelcomparisons with other models. In D. L. Schacter &Tulving (Eds.), Memory systems.Cambridge, MAMIT Press.

Moscovitch, M. (1994b). Cognitive resources and dual-interference effects at retrieval in normal people:role of the frontal lobes and medial temporal corNeuropsychology,8, 524–534.

Moscovitch, M., Vriezen, E., & Goshen-Gottstein,(1993). Implicit tests of memory in patients with folesions and degenerative brain disorders. In H. Spler and F. Boller (Eds.),Handbook of neuropsycholo(Vol. 8) (pp. 133–174). Amsterdam: Elsevier.

Moscovitch, M., Winocur, G., & McLachlan, D. (1986Memory as assessed by recognition and reading timnormal and memory-impaired people with Alzheime

disease and other neurological disorders.Journal of

of-og-

inlied

rn-nor

alg a

ionsic

tingd

ip-ale

ndRe-

sonl

ge.

out.9).licic

itF.

9).licitH.Lylvin

mas.),

ces

Runquist, W. N. (1986). Changes in the rate of forgettingl-

dgelton,s:om

temsple

.3).ans-e,

g indirecto-

c-inge,

ciatesy-y:

ho-sia.m-

p-niga.

ag-tionral-

citso-

uraly:

ofents:

1).ning-


Experimental Psychology: General,115,331–347.Musen, G., & Squire, L. R. (1991). Normal acquisition

novel verbal information in amnesia.Journal of Experimental Psychology: Learning, Memory, and Cnition, 17, 1095–1104.

Musen, G., & Squire, L. R. (1992a). Nonverbal primingamnesia. Special issue: Memory and cognition appMemory & Cognition,20, 441–448.

Musen, G., & Squire, L. R. (1992b). On the implicit leaing of novel associations by amnesic patients andmal subjects.Neuropsychology,7, 119–135.

O’Reilly, R. C., & McClelland, J. L. (1994). Hippocampconjunctive encoding, storage, and recall: Avoidintrade-off.Hippocampus,4, 661–682.

Paller, K. A., & Mayes, A. R. (1994). New-associatpriming of word identification in normal and amnesubjects.Cortex,30, 53–73.

Postman, L. (1971). Transfer, interference, and forgetIn J.W. Kling & L.A. Riggs (Eds.),Woodworth anSchlosberg’s experimental psychology.New York:Holt, Rinehart & Winston.

Press, G. A., Amaral, D. G., & Squire, L. R. (1989). Hpocampal abnormalities in amnesic patients reveby high resolution magnetic resonance imaging.Na-ture, 342,54–57.

Rajaram, S., & Coslett, H. B. (in press). Acquisition atransfer of new verbal information in amnesia:trieval and neuroanatomical constraints.

Rajaram, S., & Roediger, H. L. (1993). Direct compariof four implicit memory tests.Journal of ExperimentaPsychology: Learning, Memory, and Cognition,19,765–776.

Reber, A. S. (1989). Implicit learning and tacit knowledJournal of Experimental Psychology: General,118,219–235.

Roediger, H. L. (1990). Implicit memory: Retention withremembering.American Psychologist,45, 1043–1056

Roediger, H. L., Buckner, R., & McDermott, K. B. (199Components of processing. In J.K. Foster & M. Je(Eds.), Memory: Systems, process or function?(pp.31–65). Oxford, UK: Oxford Univ. Press.

Roediger, H. L., & McDermott, K. B. (1993). Implicmemory in normal human subjects. In J. Grafman &Boller (Eds.),Handbook of neuropsychology(Vol. 8)(pp. 63–131). Amsterdam: Elsevier.

Roediger, H. L., Weldon, M. S., & Challis, B. H. (198Explaining dissociations between implicit and expmeasures of retention: A processing account. InRoediger & F.I.M. Craik (Eds.),Varieties of memorand consciousness: Essays in honor of Endel Tu(pp. 3–41). Hillsdale, NJ: Erlbaum.

Rozin, P. (1976). The psychobiological approach to humemory. In M.R. Rosenzweig & E.L. Bennet (EdNeural mechanisms of learning and memory.Cam-bridge, MA: MIT Press.

Runquist, W. N. (1975). Interference among memory traMemory & Cognition,3, 143–159.

.

-

.

d

.

g

n

.

produced by recall tests.Canadian Journal of Psychoogy,40, 282–289.

Schacter, D. L. (1985). Priming of old and new knowlein amnesic patients and normal subjects. In D.S. OE. Gamzu, & S. Corkin (Eds.)Memory dysfunctionAn integration of animal and human research frpreclinical and clinical perspectives(pp. 44–53), NewYork: New York Academy of Sciences.

Schacter, D. L. (1990). Perceptual representation sysand implicit memory: Towards a resolution of multimemory systems debate. In A. Diamond (Ed.),Devel-opment and neural bases of higher cognition(pp. 543–571). New York: New York Academy of Sciences

Schacter, D. L., Cooper, L. A., & Treadwell, J. (199Preserved priming of novel objects across size trformation in amnesic patients.Psychological Scienc4, 331–335.

Schacter, D. L., & Graf, P. (1986). Preserved learninamnesic patients: Perspectives from research onpriming. Journal of Clinical and Experimental Neurpsychology,8, 727–743.

Shapiro, M. L., & Olton, D. S. (1994). Hippocampal funtion and interference. In D.L. Schacter & E. Tulv(Eds.), Memory systems(pp. 87–117). CambridgMA: MIT Press.

Shimamura, A. P., & Squire, L. R. (1984). Paired-assolearning and priming effects in amnesia: A neuropchological study.Journal of Experimental PsychologGeneral,113,556–570.

Shimamura, A. P., & Squire, L. R. (1987). A neuropsyclogical study of fact memory and source amneJournal of Experimental Psychology: Learning, Meory, and Cognition,13, 464–473.

Squire, L. R., & Knowlton, B. J. (1995). Memory, hipocampus, and brain systems. In M.S. Gazza(Ed.), The cognitive neurosciences(pp. 825–837)Cambridge, MA: MIT Press.

Squire, L. R., Amaral, D. G., & Press, G. A. (1990). Mnetic resonance imaging of the hippocampal formaand mammillary nuclei distinguish medial tempolobe and diencephalic amnesia.The Journal of Neuroscience,10, 3106–3117.

Srinivas, K., & Roediger, H. L. (1990). Classifying implimemory tests: Category association and anagramlution. Journal of Memory and Language,29, 389–412.

Stadler, M. A. (1989). On learning complex procedknowledge. Journal of Experimental PsychologLearning, Memory, and Cognition,15, 1061–1069.

Thoene, A. I. T., & Glisky, E. L. (1995). Learningname–face associations in memory impaired patiA comparison of different training procedures.Journalof the International Neuropsychological Society,1,29–38.

Tulving, E., Hayman, C. A. G., & Macdonald, C. (199Long-lasting perceptual priming and semantic learin amnesia: A case experiment.Journal of Experimen

tal Psychology: Learning, Memory, and Cognition,7,

an

ngen

y-

s in

A.es

4).nts

on-7).on

olemi-m-

r-in

,

Warrington, E. K., & Weiskrantz, L. (1968). A new methode to

sic-

ofan.

m-

ofpa-

A

0).

em-

of

g-e.


595–617.Tulving, E., & Schacter, D. L. (1990). Priming and hum

memory systems.Science,247,301–306.Tulving, E., Schacter, D. L., & Stark, H. A. (1982). Primi

effects in word fragment completion are independof recognition memory.Journal of Experimental Pschology: Learning, Memory, and Cognition,8, 336–342.

Underwood, B. J., & Postman, L. (1973). Critical issueinterference theory.Memory & Cognition,1, 19–40.

Vaidya, C. J., Gabrieli, J. D. E., Keane, M. M., & Monti, L.(1995). Perceptual and conceptual memory processglobal amnesia.Neuropsychology,9, 580–591.

Van der Linden, M., Meulemans, T., & Lorrain, D. (199Acquisition of new concepts by two amnesic patieCortex,30, 305–317.

Vargha-Khadem, F., Gadian, D. G., Watkins, K. E., Cnelly, A., Van Paesschen, W., & Mishkin, M. (199Differential effects of early hippocampal pathologyepisodic and semantic memory.Science,277, 376–380.

Verfaellie, M., Croce, P., & Milberg, W. P. (1995). The rof episodic memory in semantic learning: An exanation of vocabulary acquisition in a patient with anesia due to encephalitis.Neurocase,1, 291–304.

Verfaellie, M., Milberg, W. P., Cermak, L. S., & Letouneau, L. L. (1992). Priming of spatial configurationsalcoholic Korsakoff’s amnesia.Brain and Cognition18, 34–45.

t

in

.

of testing long-term retention with special referencamnesic patients.Nature (London),217,972–974.

Warrington E. K., & Weiskrantz, L. (1970). The amnesyndrome: Consolidation or retrieval?Nature (London),228,628–630.

Warrington, E. K., & Weiskrantz, L. (1973). An analysisthe short-term and long-term memory defects in mIn J.A. Deutsch (Ed.),The physiological basis of meory (pp. 365–396). New York: Academic Press.

Warrington E. K., & Weiskrantz, L. (1974).The effectprior learning on subsequent retention in amnesictients.Neuropsychologia,16, 169–177.

Warrington, E. K., & Weiskrantz, L. (1982). Amnesia:disconnection syndrome?Neuropsychologia,20, 223–248.

Wenger, S. K., Thompson, C. P., & Bartling, C. A. (198Recall facilitates subsequent recognition.Journal ofExperimental Psychology: Human Learning and Mory, 6, 135–144.

Winocur, G., & Weiskrantz, L. (1976). An investigationpaired associate learning in amnesic patients.Neuro-psychologia,14, 97–110.

Wood, F. B., Brown, I. S., & Fulton, R. H. (1989). Lonterm follow-up of a childhood amnesic syndromBrain and Cognition,10, 76–86.

(Received March 24, 1999)(Revision received May 1, 2000)

new conceptual associative learning in amnesia: a …...2015/08/07 · new conceptual associative...

Documents