Preserved Recognition in aCase of Developmental : Implications for the Acquisition of Semantic Memory?

Alan Baddeley 1,Faraneh Vargha-Khadem 2,andMortimer Mishkin 3

Abstract

& We report the performance on recognition memory tests of normally associated withepisodic memory. Hisrecall of Jon, who,despite amnesia fromearly childhood, has developed previously unfamiliarnewsreel events was impaired,but gained normal levels ofperformance on tests ofintelligence, language, substantially fromrepetition over a2-day period.Our results and general knowledge. Despite impairedrecall, he performed are consistent withthe hypothesis that the recollective process withinthe normal range on each ofsix recognition tests, but he ofepisodic memory isnot necessary either forrecognition or appears to lack the recollective phenomenological experience forthe acquisition ofsemantic knowledge. &

INTRODUCTION thecapacity tostore and recollect experienced events; Inthe lastfew decades, there hasbeen extensive and semantic memory,a systemthat stores and retrieves research intothe cognitive neuropsychologyof the knowledge about theworld (Tulving,1972). The term amnesic syndrome,ranging from Milner’s(1966) early episodicmemory is used generally torefer tothe research onpatient HM,throughthe attempts to relate capacity torecollect pastexperience, either byrecalling thefunctional deficits of amnesic patientsto laboratory theevent, or inthe case of recognition, byrecollecting studiesof normalsubjects (Squire, Knowlton, &Musen, information associated withthe learning experience 1993; Baddeley &Warrington, 1970), upto the present (Tulving,1972). Thisphenomenological aspect of epi- situationin which it is widely accepted thatdata from sodicmemory is essential to Tulving’ suse of theterm, studyof theamnesic syndromemake acrucial contribu- althoughothers use theterm somewhatmore broadly. tionto our current concepts of normal humanmemory There hasin recent years been considerable interest in (Schacter &Tulving,1994). Whilethere have been thephenomenology of recognition, withconsiderable numerousattempts to argue for different typesof evidence thatsubjects are successfullyable tocategorize amnesia usuallyrelated todifferential lesions(Parkin, itemsthey have recognized on thebasis of whether they Leng, &Hunkin,1990; Huppert& Piercy,1979), such have been remembered withassociated recollective subdivisionshave proved difficult tosustain,leading toa experience, or whether subjectssimply ’ ’know’’that somewhatnegative, thoughby no means universalcon- theitem had previously been presented (Gardiner & sensusof aunitaryamnesic syndrome(Squire, 1992; Java,1993). Baddeley, 1997). Semantic memoryis assumed to reflect accumulated Thismodal view distinguishes between explicitor world knowledge suchas themeaning of theword declarative memorythat is grossly impaired in amnesia, ’’fish,’’thecolor and of abanana, thenumber of and arange of implicit,procedural, or nondeclarative yardsin a mile,or thenormal procedure for eating ina memorysystems that are intact (Baddeley,1997; restaurant. Whileamnesic patientsusually retain seman- Squire, 1992). These implicitmemory systems have in tic knowledge acquired before theonset of amnesia, commonthe fact thatthey operate inthe absence of theytypically have great difficultyin adding further consciousrecollection. information totheir semantic memorysystems, for Explicitor declarative memorycan itselfbe splitinto example, amnesic patientswould generally be unable two components:episodic memory, which represents toprovide the names of current politicalfigures suchas theU.S. President, and wouldbe unaware of recent world eventsand of themeaning of newlycoined words 1 University ofBristol, 2 University College London and Great (Squire et al.,1993). The associationof impairedepiso- Ormond Street Hospitalfor Children, London, 3 National dic memorywith the failure toupdate semantic memory Institute ofMental Health,Bethesda, MD isconsistent with the view that semantic memoryre-

© 2001Massachusetts Institute ofTechnology Journal ofCognitive Neuroscience 13:3, pp. 357–369 flects theresidue of manyepisodes (Baddeley, 1997; witha similarprofile, who were alsoable to absent Squire, 1992), althoughthis view is not universally held themselvesfrom workfor theseveral daysnecessary to (Mishkin,Vargha-Khadem ,&Gadian, 1998; Tulving, complete thefull testbattery, proved difficult. We were, 1972; Tulving& Markovitch,1998). however,able to locate two suchcontrol subjects. Althoughthis overall viewhas, for manyyears, been Their resultswere used,together withresults from able toaccount for thebulk of evidence, ithas recently thesomewhat larger butless well matched groups been challenged bythe description of whatappears to from thepublished literature, asa baseline for inter- be anew amnesic syndrome.The patientsconcerned preting Jon’sperformance. appear tohavedeveloped amnesia toa greater orlesser degree at, or within,a few years of birth,a situation Case Description whichshould, according tothe modal view, have led to grosslyimpaired semantic memory.Nevertheless, three Jon,who is now 23, was bornprematurely at 26 weeks suchcases reported byVargha-Khadem et al.(1997) ofgestation, weighing lessthan 1 kgand suffering from havedemonstrated an apparently normal acquisitionof breathing problems(Gadian et al.,2000). For prolonged language, together witha remarkably well developed periodsduring hisfirst six weeks of life, he suffered knowledge of theworld. The implicationof suchcases from severe apnea, requiring intubationand positive for theconcept of semantic memoryis discussed else- pressure ventilation.He subsequentlyshowed steady where (Mishkinet al.,1998; Squire &Zola, 1998; Tulving improvementand normal development.At 3years 10 &Markovitch,1998); thepurpose of thepresent studyis months,he hadan unconfirmed convulsiveepisode in to investigate further thenature of thememory deficit, associationwith cold and cough. Hismemory problems and inparticular, toprovide a more robusttest of the were firstnoted when he was about 5years old,and suggestionby Vargha-Khadem et al.that this type of havesince continued to be prominent.More specifi- amnesia maybe associated withpreserved recognition cally,Jon’ sparents report thathe hasdifficulty in memorydespite clear evidence of impairedrecall. Such reliablyfinding hisway, tends to forget where objects adissociationis not typical of theamnesic syndrome: and belongings are usuallykept or where theyhave just Patientsnot only fail tosuccessfully recollect prior been put,and makesmany prospective memory errors, experiences, butthey also fail todistinguish between finding itdifficult tokeep even regularly scheduled novelwords, pictures or events,and itemsthat have appointmentsunless reminded. Jon also has difficulty already been presented (Squire et al.,1993). remembering detailsof everyday activitiessuch as con- We begin byrecapitulating thecase of Jon,one of the versationsor televisionprograms watched. The severity three described byVargha-Khadem et al.,illustrating the ofthedisability is such as tomake itdifficult for Jonto evidence for hisamnesia, together withthe comparative cope independently,or tosucceed infinding and keep- preservation of hissemantic memory.This is then ing a job. followed bya summaryof hisneuropathology as deter- Table 1showsthe performance of Jonand thetwo minedby quantitative magnetic resonance techniques. control subjects,Andrew and Max, on intelligence, scho- We thengo on toa brief account of earlier claimsof lastic,and language performance tests,all carried out preserved recognition inpatients who have acquired between theages of 19 and 22. Jonhas above-average theiramnesic deficit asa resultof braininjury during intelligence witha full-scale IQof 114, withsubtest adulthood.We discusssome of the methodological scores ranging from scaled scores of 10 (the normal problemsfacing anyattempt to compare and average) oninformation,to 16 (two standard deviations recognition, before describing astudyin which we above themean) on blockdesign. His number skills as compare Jon’srecall and recognition for bothvisual measured ontheWechsler Objective Numerical Dimen- and verbal material. Inorder to studythe generality of sions(WOND) testare above average, whilehis reading our findings,we thencompare Jon’sperformance to performance iswithin the normal range for basicreading thatof normal control subjectsacross several other (55th percentile) and comprehension(42nd percentile) recognition paradigms.We thendiscuss the implications coupled withpoor spelling (14th percentile). Jon’s of our resultsfor thedistinction between recall and language isalso within the normal range asmeasured recognition memory,and for thedevelopment of se- bythe Token Test,which requires thefollowing of mantic memory. instructionsof increasing grammatical complexity,and The testsused comprisedboth standardized mea- hissemantic judgements appear tobe normal,as mea- suresfor whichappropriate normsare available, and sured bythe Pyramids and PalmTrees Test,a testof lesswidely used experimental tests.For thelatter, the semantic access from wordsand pictures(Howard & questionarises as tothe appropriate control compar- Patterson,1992). Jon’svocabulary developmentis also ison.When tested at age 19, Jonshowed a slightly withinthe normal range, withthe WAIS-R vocabulary unusualWAIS-R intelligence profile,with his perfor- subtestproducing ascaled score of 14, whilehis com- mance IQ(120) being somewhatgreater thanhis verbal prehensionsubtest score was 11. Finally,his perfor- IQ(108). Findingage- and education-matched subjects mance on theSpot theWord Test (Baddeley, Emslie,

358Journal ofCognitive Neuroscience Volume 13,Number 3 Table 1. IQ, Academic Attainments and Language Test Jon Andrew Max

WAIS-R Standard score Range Standard score Range Standard scoreRange Full-scale IQ 114 Highaverage 114 Highaverage 101 Average Verbal IQ 108 Average 107 Average 97 Average Performance IQ 120 High 120 High 108 Average

Years ofeducation 13 13 13

Academicattainments Standard scorePercentile Standard scorePercentile Standard scorePercentile WOND Mathematics reasoning 103 58 118 88 118 88 Numericaloperations 115 84 108 70 119 90 WORD Basic reading 102 55 108 70 102 55 Spelling 84 14 111 77 102 55 Reading comprehension 97 42 102 55 86 18

Language Raw score Range Raw score Range Rawscore Range Token Test 60/62 Normal n/aa n/a 56/62 Normal Pyramidsand PalmTrees 49/52 Normal 52/52 Normal 51/52 Normal

Scaledscore Range Scaledscore Range Scaledscore Range WAIS-Rvocabulary 12 Average 12 Average 9 Average WAIS-Rcomprehension 10 Average 12 Average 10 Average Spot the Word 10 Average 11 Average 9 Average aNot administered.

&Nimmo-Smith,1993) was alsowithin the normal of subtestssensitive to different aspects of executive range. Thisis a lexical decisiontask in which pairs of processing.He scored 21/24, an entirely satisfactory itemsare presented, one comprisinga word and one a performance. Hewas noted tobethinkingahead, plan- nonword;Jon’ sscore of 47 outof 60 places himat the ning for thefuture, buteasily distracted, and withsome 50th percentile. Inconclusion, apart from poorspelling, trouble withdecision-making. In general, Jon’sexecutive Jonhas above-average intelligence and performs nor- capacities appear tobe withinthe normal range. mallyon scholastic and language tests,including eight Table 2showsthe performance of Jon,Andrew, and different testsof semantic or language processing. Max on standardized memorytests. On the recently In1991, Jonwas tested ontheWisconsin Card Sorting developed Extended Rivermead BehaviouralMemory Test,using the original Milner (1963) procedure. He Test (Wilsonet al.,1999), hisprofile score of 10 indicates obtained five categories, makinga total of 25 errors of alevel of performance thatis well withinthe impaired which24 were perseverative. Thisindicates performance range (0–18). Jon’svisualmemory, as indicated bydelayed withinthe normal range. Hiscategory fluency tested at recall of thecomplex Rey Figure, was alsodramatically thistime was alsonormal (20 animalnames and 14 fruit impairedwith no subsequentscoreable reproduction. types).In 1997, he was tested on theFAS initialletter The Children’sAuditoryVerbal Learning Test-2 (with fluency task,producing 50 words,which places him normsranging from 6to18 years) involvespresenting a between the70th and 80th percentile. Inthe same year, sequence of 16 wordsfive times,after whicha second he was tested on theBehavioural Assessment of the listis presented and tested (Delis,Kramer, Kaplan, & Dysexecutive Syndrome (BADS) (Wilson,Alderman, Bur- Ober, 1987). The initiallist is then presented again and gess,Emslie, & Evans,1996), whichcomprises a number tested for recall and finallyrecognition. Jondoes well on

Baddeley,Vargha-Khadem, and Mishkin359 Table 2. Memory Performance Test Jon Andrew Max

Rivermead Behavioural Memory Profilescore Range Profilescore Range Profilescore Range Test–Extended 10 Impaired 33 Average 40 Good

CAVLT-2 Standard scorePercentile Standard scorePercentile Standard scorePercentile Trial 1 108 70 98 45 125 95 Trial 5 77 6 117 87 108 70 Post interference <60 <1 117 87 106 66 Delayed recall 63 1 118 88 118 88

Rawscore Percentile Raw score Percentile Rawscore Percentile Recognition accuracy 29 <16 32 Normal 32 Normal Intrusions 17 <16 0 Normal 0 Normal

Warrington RecognitionMemory Rawscore Percentile Raw score Percentile Rawscore Percentile Test Words 45 25 48 75 50 95 Faces 41 25 49 95 42 25

theinitial test (68th percentile) relyingheavily on recal- withthe targets, or impairedit, since bothtargets and lingthe last few words,the recency effect, whichreflects distractorswould have some degree of familiarity.It is short-termmemory and hence, tendsto be preserved in therefore clearly desirable to attemptto make amore amnesia (Baddeley &Warrington, 1970). Hissubsequent direct assessmentof recognition memory,and, if possi- scores of8, 7, 8, and 10 showlittle evidence of learning, ble,to compare itwitha comparable testof recall. continuing toreflect areliance on recency thatis atypical insubjects with normal memory(Greene, Baddeley, & Neuropathological findings Hodges,1996). Following theinterfering list,a recall of four correct itemswith seven intrusionsplaces Jon As reported previously(Gadian et al.,2000; Vargha- below thefirst percentile. Incontrast, afinal recognition Khadem et al.,1997), direct measurements performed score of 29 outof 32 places himjust outside the normal on Jon’smagnetic resonance scansrevealed thatthe cutoff of 30. Asimilarpattern of extremely poorword hippocampalvolume in each hemispherewas approxi- recall was shownwhen he was tested some5 years mately50% lessthan the normal mean volume,a degree earlier on apreviousversion of thetest, whereas his ofatrophythat was clearly visibleon thescans. This recognition scores of29/32 and 30/32 on two different abnormalitywas alsothe only one visible,although occasions were bothwithin the normal range for this quantitativeMR techniquessuggested thepresence of older versionof thetest. more subtlepathology in other regions,including the The Warrington Recognition Memory Test involves putamen bilaterally,the ventral part ofthethalamus, and presenting 50 wordsand 50 faces, and ineach case themidbrain (Gadian et al.,2000). These additional requiring subsequenttwo-alternative, forced-choice re- abnormalities,like the bilateral hippocampalatrophy, (Warrington, 1984). Inboth cases, Jonscored are consistentwith the known effects of hypoxia–ische- at or above the25th percentile, lesswell thanone might mia,which is the presumed cause oftheneuropathology expect from hisgeneral intelligence, butsubstantially inJon. It should be noted thatnone of these MR better thansuggested byhis impaired performance on techniquesyielded evidence of bilateral pathologyin thevarioustests of recall described earlier. However,this themedial temporal tissueoutside the . level of performance shouldbe treated withcaution Inconsidering theunusual pattern of Jon’simpairedand since Jonhad performed thistest on anumber of spared memoryabilities, it isimportant to keep inmind occasions inthe past, albeit separated byintervals of thisapparent selectivityof hisventral temporal damage, manymonths. Repetition mighteither haveimproved his as ithasraised theinteresting possibilitythat just as his performance, as aresultof hisbecoming more familiar memorylosses are attributable tothe bilateral hippo-

360Journal ofCognitive Neuroscience Volume 13,Number 3 campal atrophy,his preserved memoryabilities are due combinedwith apparent sparingof recognition memory at least inpart tothe preservation ofthesubjacent cortex mightthus be claimed tostem from areduction inthe inthe parahippocampalregion (Mishkin,Suzuki, Gadian, development ofthehippocampus, together withappar- &Vargha-Khadem, 1997; Vargha-Khadem et al.,1997). ent sparingof other braintissue associated withmemory function (Gadian et al.,2000; Mishkinet al.,1998; Vargha-Khadem et al.,1997), hence, tobe consistent Neuropsychological Evidence for a withthe hypothesis proposed by Aggleton and Brown. Recall–Recognition Dissociation However,the principal function of thepresent studyis The claimthat recall and recognition are differentially to explore thefunctional rather thanthe anatomical affected bymemory deficit hasa long,if somewhat characteristics of Jon’smemorydeficit and todiscuss its checkered history.It has, for example, been suggested, relevance tothe apparent preservation of hissemantic at least since Schonfield and Robertson (1966), that memorydevelopment. aging hasa greater impacton recall thanon recogni- tion,and for awhile,it was suggested thatthe classic Problems inComparing Recalland Recognition amnesic syndromemight show such a dissociation (Huppert& Piercy,1979). Unfortunately, suchearly As demonstrated byMandler, Pearlstone, and Koopmans studiesare open tothe objection thatthe recognition (1969), itis clearly unsatisfactory toassume that tasks tasksused were simplyless demanding thanthe recall, based on recall and recognition of thesame material will and itis now generally accepted thatrecognition be equivalent indifficulty, as measured bypercentage of memoryis not preserved inmost patients suffering correct responses.One wayof attempting tocircumvent from theamnesic syndrome(Aggleton &Brown,1999; thisis to attempt toadjust thedifficulty of one taskso Reed &Squire, 1997). thatit appears togive ascore equivalent tothat on the Anumber of atypicalcases have,however, been other.Calev (1984), for example, adjusted thelevel of reported inwhichrecognition does appear tohavebeen recall and recognition byusing categorized wordsfor preserved. These include patientswith bilateral recall and unrelated wordsfor recognition. Thisof damage (McMackin, Cockburn,Anslow, & Gaffan, 1995), course introduces potentiallyimportant differences be- patientswith damage inthe region of themammillary tween thetwo tasks. This procedure isalso open tothe bodies(Parkin & Hunkin,1993; DuSoir, Kapur, Byrnes, objection thatit assumes that having found one pointat McKinstry,& Hoare, 1990), patientswho sustained whichthe distribution of scores from therecall and anoxia (Volpe,Holtzman, & Hirst,1986), cases with recognition methodsgive asimilarpercentage, thenthe alcohol-induced Wernicke’sencephalopathy(Parkin, rest ofthedistribution of scores willalso be equivalent. Dunn,Lee, O’Hara, &Nussbaum,1993), and acase with Thisis not necessarily thecase. Shallice (1988) makesa anterior communicating artery aneurysm(Hanley, Da- similarpoint in challenging theclaim by Hirst et al. vies,Downes, & Mayes,1994). (1986) ofhavingdemonstrated arecall–recognition dis- Ina recent review oftheamnesia literature, Aggleton sociationin amnesic patients,pointing out the way in and Brown (1999) and Aggleton and Shaw (1996) have whicha difference invariance between two measures argued for apreservation of recognition memoryin could alsogive riseto a spuriousapparent dissociation. thosecases of amnesia where thedamage islimited to The absence of anydifferential effects of amnesia on thehippocampus or itsdiencephalic targets. Aggleton recall and recognition inmost instances was further and Brown argue thatthis is entirely consistent with the supportedby a thoroughstudy by Haist, Shimamura, literature from experimental lesionsin animals, most of &Squire (1992). whichindicates thatrecognition memoryis unimpaired Toavoidthe earlier problemsin comparing recall and bylesions limited to the hippocampus, but is impaired recognition, Baddeley, Emslie,and Nimmo-Smith(1994) whenextrahippocampal regions are damaged, suchas attempted toproduce testsof visualand verbal recall theperirhinal cortex and related regions.This view is and recognition of equivalent difficultyby using scaled stronglyopposed by Reed and Squire (1997), who scores. Such scores express apatient’sperformance on report aseriesof patientsfor whomthe damage is each testin termsof Z scores based on thedistribution ’’limitedprimarily to the hippocampus or thehippo- of normal populationscores on thattest. Using this campal formation,’’and whoshow clear evidence of method,separate measures of visualand verbal mem- impairedrecognition. As Aggleton and Brown pointout, ory,and of recall and recognition, maybe obtained however,the crucial questionis that of whether areas together withmeasures of learning and forgetting. The beyondthe hippocampus were indeed completelypre- testhas proved sensitive to a range of factors knownto served;something that is difficult toestablish, particu- influence episodicmemory, including Alzheimer’ sDis- larlyin view of thefact thatbrain areas thatappear tobe ease, normal aging, and schizophrenia(Baddeley, 1996); structurallyintact when imaged usingMRI mayshow thetest is also able todifferentiate between theeffects reduced functional activitywhen studiedusing PET of rightand left temporal lobectomyon visualand verbal (Aggleton &Brown,1999). Jon’ssyndromeof amnesia memory(Morris, Abrahams, Baddeley, &Polkey,1995).

Baddeley,Vargha-Khadem, and Mishkin361 Lessdramatic visual–verbal memorydifferences were none of thethree showa clear difference inperfor- alsofound inother groups,with the elderly tending to mance between pictorialand verbal material. Inthe case bemore impairedon names thanon visualstimuli, while of therecall– recognition distinction,however, the dis- the oppositepattern occurred for schizophrenicpa- crepancy givesa scaled recall–recognition difference tients(Baddeley, 1996). On theother hand,there has score for Jonof below 3, whichrepresents thebottom sofar been no evidence reported for amarked recall– ofthescale on thisparticular testmeasure. Incontrast, recognition difference for either etiological groupsor Andrew and Max bothhave scaled scores of 9, placing individualswithin the various groups. Since thesubjects themwithin one-third of astandard deviationof the assessedusing the test have varied markedlyin memory populationnorm. ability,from elderly Alzheimer patients(Greene et al., The resultsof theDoors and People Test indicate a 1996) toyoung Ph.D. students (Baddeley, 1993), this marked discrepancy inJon between clearly impaired suggests thatany recall– recognition discrepancy ob- recall and recognition performance thatis entirelywith- served isunlikely to bean artefact resultingfrom ceiling inthe normal range. However,in discussing the claims or floor effects. for arecall–recognition discrepancy inpatients with damage limitedto thehippocampus, Reed and Squire (1997) criticize Aggleton and Shaw for relying too heavily RECOGNITIONTEST PERFORMANCE on alimitedrange oftests,and on relativelyshort tests Doors and PeopleTest withimmediate recognition. For thatreason, we ex- We began bytesting Jon on theDoorsand People Test,a tended our investigationto cover three further tests, relativelypure testof visualmemory using recognition of one of whichuses auditory presentation and rapid setsof photographsof doors,and recall bydrawing of paced recognition (Andrade, Munglani, Jones,& Badde- four different versionsof across(Baddeley et al.,1994). ley,1994; Andrade, 1996; Shepard &Teghtsoonian, These visualmemory tests are thencontrasted with 1961), whilethe others are testsdeveloped bySquire verbal recognition and recall based on people’snames. and hiscollaborators withthe explicit purpose of invol- The resultsare shownin Figure 1, where performance vinga reasonably long sequence of words,together with on thefour subtestsis expressed interms of scaled an interpolated delay (Hamann &Squire, 1997). None of scores; ascore of 10 represents thepopulation mean for these testshave standardized norms,but in all cases, subjectsof Jon’sage, witheach scaled pointrepresent- data from normal control subjectsare available, asare ing one-thirdof astandard deviation.For bothvisual and data from Andrew and Max. verbal recall memory,Jon scores below the5th percen- tile,while scoring between the50th and 75th percentile Continuous Recognition Test for recognition. Neither of thecontrol subjectsshow a clear recall–recognition discrepancy.The variousscores Shepard and Teghtsoonian (1961) developed atestin can becombinedto estimate thelikelihood of finding a whicha sequence of wordsis presented, and thesubject visual–verbal or recall–recognition discrepancy of this isrequired toperform arunning recognition task, magnitude withinthe population. In the case of the deciding when each word ispresented, whether ithas visual–verbal distinction,Jon has a scaled score of 10, been shownpreviously. Repetitions can occur immedi- Andrew ascaled score of 9, and Max ascore of9; hence, ately,or after 1, 2, 4, 8, or 16 intervening words. Despitethe rapidly paced presentation,Jon detected all repetitionswith the exception of one at adelay of four, and two at an eight-worddelay. Jon’ sscore of 97.3% isslightly higher than the scores of Andrew (86.5%) or Max (83.0%) and exceeds themean scores of 92.9% and 94.4% reported for normalsubjects (an- esthetists)in two studiesby Andrade et al.(1994) and Andrade (1996). However,Jon made an average of 0.5 false alarms per listcompared with0.31 byAndrew, none byMax, and an average of 0.20 reported by Andrade et al.,although Jon’ sscore was withinthe range observed inthat study. Because of thenature of thetest, the probability of arepetitionincreases asthe listprogresses, making a bias-free estimate of memory problematical.The next twotests avoid this problem in two different ways,one byusing two-alternative forced Figure 1. The performance ofJon and twocontrol subjects onthe Doorsand People Test ofVisualand Verbal Recall and Recognition. Jon choice, where asingleresponse is always required, and scoresat the average level onboth recognition tests, but isimpaired theother byholdingthe probability of atarget constant onthe tworecall measures. at 0.5 and notingboth detection and false alarm rates.

362Journal ofCognitive Neuroscience Volume 13,Number 3 Delayed Recognition Tests choice and yes–no recognition, tested bothimmediately and after a5-minfilled delay.Finally, recognition has The next recognition testscomprised the yes– no recog- been tested under bothunpaced and rapidlypaced nitionand two-alternativeforced-choice recognition conditions.We therefore feel reasonably confident in measures used byHamann and Squire (1997), both claimingthat despite clear evidence ofhisamnesia, Jon tested after a5-minfilled delay. hasa well preserved recognition memory.As Reed and Figure 2a showsJon’ sperformance on forced-choice Squire (1997) pointout, such a dissociationis certainly recognition, together withdata from Andrew and Max nottypical of theamnesic syndrome.Jon is therefore and from Hamannand Squire, illustratingthe overall atypicalin two respects; first,in his recall– recognition level of performance from theiramnesic and control disparity,and second, inhis spared capacity toincre- groups,based onsix separate testsgiven over 2days. ment semantic memorydespite his episodic memory Jon’sperformance isvery close tothat of thetwo age- deficit, whichagain isnot characteristic of theamnesic matched control subjectsand of Hamannand Squire’s syndrome(Squire et al.,1993). control subjects,and substantiallyabove thatof their At thispoint, we encounter an apparent paradox. amnesic patients. Jon’ssemantic memoryis not limited to his ability to Figure 2b showsthe equivalent data for yes–no re- recognize facts about theworld. He isquite capable of cognition performance. Inthe case of forced-choice recalling theinformation necessary for defining words, recognition, performance ismeasured bysimplepercen- can recall thenames of historicfigures, and discussat tage correct. Inthe case of yes–no recognition, hitsand considerable length issuessuch as current politics.All false alarms are combinedto create d0,ameasure of these suggest thatit is not the capacity torecall per se discriminationaccuracy; Jon’saverage detection prob- thatis impaired, but rather theability to use episodic abilityis .813, whilehis average false alarm rate is.201 memoryin order tofacilitate therecall ofnewlyacquired giving him a d0 score of 1.76. As isclear from Figure 2b, material suchas lists of unrelated wordspresented once. Jon’sperformance broadlyresembles theaverage of The fact thatJon has acquired an extensive amountof Hamannand Squire’scontrol subjects (mean d0 = semantic knowledge suggeststhat learning, whether 1.89) and falls between thatof Andrew ( d0 = 1.19) and tested byrecall or recognition, shouldbe possibleunder Max (d0 =2.12). Performance on thistask was highly conditionswhere naturalisticmaterial ispresented on variable,with Andrew and one of Hamannand Squire’s multipleoccasions. While it is clearly impractical to control subjectsscoring below thelevel of one of their simulateadequately alifetime of richexperience, we amnesic patients.This suggests that subjects may have went someway in this direction bypresenting Jonwith a difficultyin setting an appropriate criterion, since per- seriesof videosbased on newsreels of eventsfrom the formance under forced-choice conditions,which re- years 1937 and 1957. One of these,1937, was presented movethe problemof criterion, does notshow this four timesdistributed over a2-day period,while the variability. other was presented onlyonce. Bothrecall and recogni- To thispoint, we havereported Jon’sscores on a tionwere tested immediatelyafter thefinal presentation range ofrecognition tests,both visual and verbal.On all ineach case, and were retested thefollowing morning of these,Jon has performed at, or even somewhat after an intervalof approximately18 hr. above,the average level tobe expected from anormal The resultsshown in Figure 3indicate firstof all that subject. Method of testinghas included bothforced- after asinglepresentation, Jon’ simmediaterecognition performance iscomparable to thatof thecontrol sub- jects, and even hisovernight recognition isonly slightly below theirs.His immediate recall score, bycontrast, is clearly muchlower thantheirs. However, after four presentations,Jon’ simmediaterecall score hasrisen substantially,and indeed, slightlyexceeds thatof An- drew, whileall three showa modestincrease inrecogni- tionscores. It is clearly thecase, therefore, thatJon’ s recall deficit can be minimizedgiven sufficient practice. The overnightrecall data are somewhatmore difficult to interpret, since there isevidence to indicate that successfullyperforming arecognition testcan enhance subsequentrecall. Thisprobably occurs because the presentation of recognition alternatives mayprovide a reminder, whichin turn acts as arelearning trial.For Figure 2. Delayed verbal recognition memory asmeasured by two- alternative forced-choice (a) and yes–no (b) recognition. On both example, one of the1937 questionsconcerned aZep- measures, Jon resembles the normal controlsubjects rather than the pelindisaster, asking for thename of thecraft involved, amnesic patients ofHamann &Squire (1997). and subsequentlyproviding four alternatives, theBis-

Baddeley,Vargha-Khadem, and Mishkin363 includingage, typeof processing,and degree of atten- tioninfluencing ’’remember’’judgements whilehaving littleinfluence on ’’know’’responses(Baddeley, 1997; Gardiner &Java,1993). Apossibleinterpretation of our resultsis to assume thatwhile Jon retains the capacity toacquire new informationas reflected byagreater offamiliarity and general accessibility,he lacks thecapacity torecol- lect thelearning experience, leading tothe absence of associated sensoryand contextual informationthat en- ables subjectsto say they ’ ’remember’’an event, rather thansimply ’ ’know’’thatit happened. We attempted totest this interpretation, spendinga considerable timetrying to explain toJon what it means to’’remember’’something,as opposed to simplyknow- ing thatit hadbeen presented earlier. We explained that the’ ’remember’’judgement isto be given whenever a recognized word remindsthe subject of thelearning experience, e.g.,the word ’’dachsund’’mightbe remem- bered because ofrecalling thatwhen itwas presented it reminded thesubject ofafriend’spet seen lastweek, or perhapsbecause itseemed tobe associated withan earlier itemsuch as ’ ’sausage.’’We tried to get Jonto recollect eventssuch as hisjourney tothe Institute that morning.He gave arelativelydetailed account butwas Figure 3. Recall and recognition ofevents fromunfamiliar news unable to provideany specific detail thatmight differ- videos viewed onone orfour occasions and tested immediately orafter entiate thisfrom previousjourneys. He indicated, for anovernight delay. instance, thathe hadattempted torun upthe final flight of stairsin one rush,and hadgot justover halfway. However,when asked howhe knew this,he replied ’’I mark,the Mecklenburg, theDeutschland, and theHin- alwaysget about halfway.’’We nevertheless decided to denberg. Havingbeen reminded of acorrect answer that proceed, usingmaterial and testingprocedures taken could notbe recalled, thesubject maythen retain the from Parkinand Walter (1992), inwhich words are information.The mostimportant feature of these re- presented for learning and thentested byyes– no re- sults,however, is the fact thatJon is able toretain the cognition,withsubjects categorizingthe’ ’yes’’re- informationhe hasacquired across theovernight delay sponsesas being ’’remembered’’or ’’known.’’This of 18 hr,whether tested byrecall or recognition; long- methodhas the safeguard of subsequentlychecking a term retention of newlyacquired knowledge is,of subsampleof ’’remember’’judgements and requiring course, anecessary feature for theincrementation of thesubject togive thereason for hisjudgement. In semantic memory. termsof overall performance, Jonhad a hitrate of.917 Whilevariables that influence recall typicallyalso and afalse alarm rate of .194, givinghim a d0 score of recognition, thequestion of whether theyreflect 2.25. Thisis slightly lower thanthe d0 obtained bythe equivalent processes hasremained controversial,at least youngcontrol subjects(mean =2.70, SD = 0.7), but since Mandler (1980) proposeda two-componentinter- clearly higherthan that obtained bythe normal elderly pretation ofrecognition. Heproposedthat a target word subjects (d0 = 1.90, SD =0.8). At firstsight, Jon mightbe subsequentlyrecognized on thebasis of either appeared tobe usingthe ’ ’remember’’judgement even therecollection of havingencountered thatword, e.g., more readily thanthe young, and substantiallymore because itreminded one of theearlier learning experi- often thanthe elderly subjects.However, in each case, ence, oralternatively,simply on thebasis of afeeling of whensubsequently asked whyhe gave the’ ’remember’’ familiarity.This line of discussionwas further developed judgement, he providedthe same account, namelythat byTulving (1985), whodefines episodicmemory in helooked at theword, tried toimagine itprinted on the termsof thesubjective experience of recollection. He stimuluscard, and if thisresulted inan immediateand proposedthat requiring subjectsto judge whether they clear image, gave a’’remember’’response.Such a truly’ ’remember’’aword,or simply’ ’know’’thatit was methodis much closer tothe use of perceptual fluency, previouslypresented mayprove to be aproductiveway amethodcommonly associated withimplicit rather than of investigating episodicmemory. This has indeed recollective memory(Whittlesea, 1993). The fact thatwe provedto be thecase, witha wholerange of variables were unable to train Jonto use the’ ’remember/know’’

364Journal ofCognitive Neuroscience Volume 13,Number 3 distinctionappropriately is entirely consistent with the impairedrecall suggeststhat they reflect twoseparate hypothesisthat he lacks thecapacity torecollect the memorysystems. This would be premature for two contextual detail necessary for an appropriate ’’remem- reasons.First, our resultsare based on asingledissocia- ber’’response.However, a conclusionthat relies on a tionrather thanthe much more powerful double dis- failure toteach thesubject touse arelativelysubtle sociation,for whichwe wouldneed acase inwhich concept clearly does notprovide adequately strong recall ispreserved and recognition impaired.Second, it evidence concerning thehypothesis. Further investiga- isimportant to bear inmind that recall and recognition tionis ongoing (see below), based on thequestion of are experimental methodsof tappingmemory, not whether Jonexhibits the electrophysiological pattern themselvesdirect measures of underlyingmemory sys- thathas been shownto be associated with’ ’remember’’ tems,although they may differ intheir relative reliance judgements (Rugg, Schloerscheidt, &Mark, 1998). onunderlying systems. Finally, however, Jon clearly is capable ofrecalling agood deal of informationof botha semantic and personal semantic nature, suggesting that DISCUSSION recall per se isnot necessarily defective. DespiteJon’ simpairedmemory when measured on Itseemsmore productiveto consider interpreting our teststhat rely on recall, hisrecognition memorywas findingsin termsof theprocesses underlyingrecall and found tobe normal across arange of testsinvolving recognition. Itis now widelyaccepted thathuman long- different materials,speeds of presentation,and recogni- term memoryis not unitary, reflecting at least two tionparadigms. The singleexception was hisslightly underlyingprocesses orsystems.One involvesthe con- lowered score for overnightrecognition ofitemson the sciousrecognition and recall of previouslyexperienced newsreel presented onlyonce. On thebasis of this,we eventsor facts, termed bySquire and Zola (1998) as wouldclaim thatJon’ srecognition memoryperfor- declarative memory.This system, in which the subject is mance falls withinthe normal range. More specifically, aware ofthefact thathe or sheis remembering, maybe hisrecall and recognition scores on theDoors and contrasted witha range of nondeclarative or implicit People Test explicitlydesigned toallow arecall–recog- memoryprocesses inwhich prior learning isreflected nitioncontrast, demonstrate veryclearly impairedrecall indirectlythrough enhanced performance. Nondeclara- and preserved recognition. Such adissociationis not tivememory may, for instance, occur inthe acquisition attributable tothe differential sensitivityof the two of skills,classical conditioning,and arange of perceptual measures,since itis not shown by Alzheimer’ sdisease primingeffects. Ingeneral, amnesic patientsshow pre- patients(Greene et al.,1996), byschizophrenicpatients served performance on these implicitmemory tasks or normal elderly subjects(Baddeley, 1996), or bya together withimpaired performance on explicitor group of patientsrecently reported byManns and declarative measures (Squire et al.,1993). Squire (1999), selected ashaving lesions that are located Itisnow generally accepted thatrecognition memory principallyin the hippocampus. itselfreflects at least two underlyingprocesses, one Withthe single exception noted above,Jon’ srecogni- recollective, and based on thecapacity to associate the tionperformance isconsistentl ywithinthe normal recognized itemwith some aspect of itsoriginal pre- range; can we therefore conclude thathis performance sentation (’’remembered’’items),while the other is isunimpaired? This would be unwise;Jon has above- more automatic (’’known’’items),possibly based on a average intelligence, ishighly motivated, and isan general feeling of familiarity(Tulving, 1985; Mandler, experienced and sophisticatedsubject. He does,for 1980). As discussedearlier, one methodof accessing the example, attemptto optimize his guessing bykeeping recollective component isthroughasking the subject to count of theproportion of ’’yes’’responseshe makes.It decide whether theitem they have recognized is’ ’re- isthus possible that, in the absence of neurological membered,’’or whether itis simply ’ ’known’’tohave damage, he wouldhave performed above average. The been presented (Gardiner &Java,1993). Asecond importantfeature of Jon’scase, therefore, isthe clear approach isbased onrequiring thesubject tomake a dissociationbetween recall and recognition rather than judgement based on listmembership, with instructions the seeminglycomplete preservation of recognition and interfering listsso arranged thatunder certain performance. Itis important to bear thisin mind, both circumstances, aknowledge of listmembership will when seeking toreplicate our findingsand inconsider- help,while in others, it will hinder performance– the ing possibletheoretical explanations. so-called process dissociation method developed by If we accept thatthe dissociation is indeed genuine, Jacoby (1994). There issome disagree ment as to whatare itsimplications for therecall– recognition dis- whether thetwo methods are equivalent (see Baddeley, tinctionand for understanding thespecific associationof 1997, chapter 20 for adiscussion).However, recent symptomsfound inJon,namely impaired recall memory workusing electrophysiological methods has claimed coupled withpreserved language, semantic memory, thatthe two approaches are bothways of accessing and intelligence? Itis initially tempting to argue that contextual or source memory,and ultimatelyrely on thepreservation ofrecognition inthepresence of clearly equivalent underlyingprocesses (Rugg et al.,1998).

Baddeley,Vargha-Khadem, and Mishkin365 Our failure toteach Jonto make theremember/ isalso reported ina few cases of acquired amnesia knowdistinction suggests that Jon may lack thecapa- (Aggleton &Brown,1999), althoughsuch cases are cityto utilize this recollective source of evidence. Yet highlyatypical (Squire et al.,1993) and theclaims of he appears tohave no difficultyin judging whether or selective preservation can be criticized on thegrounds notan itemwas presented before, presumablyon the of uncertain comparabilitybetween therecall and basisof thepresence or absence of afeeling of recognition measures (Manns &Squire, 1999; Hamann familiarity.Evidence consistentwith these notionswas &Squire, 1997). However,Holdstock et al.(2000) obtained inan electrophysiological studythat was recently reported thecase of apatient withacquired recently carried outon Jon(Du ¨zel, Vargha-Khadem, hippocampaldamage who,like Jon, shows preserved Heinze, &Mishkin,1999). Thispreliminary report recognition relative torecall performance on theDoors indicated thatduring recognition performance, Jon and People Test. exhibitsthe electrophysiological index thathas been Jon’scapacity to acquire semantic memoryis also associated withstimulus familiarity, but he appears to atypicalof acquired amnesic patients(Gabrieli, Cohen, lack theone thatnormally accompanies episodicre- &Corkin,1988; Rozin, 1976). Althoughthere does collection of thestimulus item. The latter processmay appear tobe limitedevidence for preserved semantic normallyaid inrecognition, and soits possible absence acquisitionin some patients, the extent ofsuchlearning or impairmentin Jon could well be reducing his isfar from normal (Verfaellie, Koseff, &Alexander, 2000; recognition scores below thosehe wouldotherwise Kitchener, Hodges,& McCarthy, 1998). Itis therefore have been able to achieve. Itmight account, for conceivable thatJon’ spreserved abilitiesare based,not example, for hisrelatively low score on overnight onlyon theselectivity of hishippocampal damage, but recognition of thenewsreel itemspresented onlyonce, alsoon thedevelopmental nature ofhisdeficit reflecting the singleinstance inwhich he mayhave shown a thegreater plasticityof theinfant brain,together with recognition deficit. Yet theabsence or impairmentof the possibledevelopment of arange of alternative episodicrecollection clearly didnot prevent Jonfrom learning strategies assuggested byManns and Squire attaining scores thatfell well withinthe normal range (1999). Thisquestion will not be settled untilit becomes on every other measure of recognition memorywith clear whether thepattern of preserved recognition in whichhe was assessedhere. associationwith spared acquisitionof semantic memory The recall involvedin well-established semantic mem- islimitedto cases of developmental amnesia,or whether ory,which may be conceived asdecontextualized, like- thispattern can alsooccur incases where theamnesic wisedoes notrequire, and, indeed, maynot greatly deficit isacquired during adulthood. benefit from,the recollective process.By contrast, epi- sodicrecollection isclearly of major importance for the METHODS recall ofnewlypresented material. Thus,an impairment inepisodicrecollection could explain Jon’spoorperfor- The following testswere administered.They required mance on standard recall tests,which are typically anumberof testsessions distributed over aperiod assumedto relyheavily on thecapacity for recollection. of weeks. Similarly,inasmuch as episodicrecollection presumably facilitates and enhances therecall of newlyacquired TheDoors and PeopleTest semantic information,an impairmentin recollection could alsobe the reason thatJon had difficulty in The visualrecognition testinvolves presenting colored recalling theitems of informationcontained inthe photographsof 12 doors,one at atime,each accom- newsreel thatwas shownto him only once. However, panied byan appropriate butultimately unhelpful label, onthenewsreel thatwas presented four timesover a2- for example ’’church door’’or ’’barn door’’(Baddeley dayperiod, a procedure thatdecontextualized thelearn- et al.,1994). Recognition istested immediatelyby pre- ing experience at least withrespect totime,and thereby senting 12 successivearrays of four doors,in each case presumablydecreased theimportance of episodicre- requiring thesubject toidentify the door already pre- collection for recall, Jon’srecall scores relative tothose sented. Byensuring thatthe items within each array all of thecontrol subjectsimproved substantially. There is have thesame label (e.g., church doors),the role of therefore good reason tosuppose that just such repeti- verbal labeling can be minimized.Two different setsof tion,even when limitedin frequency, iswhat has 12 doorsare presented and tested insuccession. The enabled Jonto acquire, and recall, hisvery large store visualrecall taskinvolves presenting and requiring the ofsemantic information. subject tocopy four crosses,each distinguishedby the We shouldend byconsidering thegenerality of the features comprisingthe overall shapeand thepresence findingswe haveobtained inJon. It seems likely that or absence of elaboration at thecrux and at theend of thepreservation of recognition relative torecall ofnew thearms of thecross. Subjects are shownthe four material maypotentially occur more widelyamong crosseson three successivetrials, each immediately developmental cases (Vargha-Khadem et al.,1997). It followed bya recall test.A final recall testfollows after

366Journal ofCognitive Neuroscience Volume 13,Number 3 afilled delay of approximately10 min.Verbal recogni- approximately35 minwas presented. Testing tookplace tionis tested bypresenting 12 names suchas John over three successivedays. The 1937 newsreel was pre- Wilkins,which are subsequentlytested byrecognition sented on themorning and early afternoon of thefirst day, from four-item sets(e.g., JohnWilby, John Wilkie, John and themorning and late afternoon of thesecond. The Wilkins,John Willis). Again, twosets of 12 itemsare 1957 newsreel was presented at theend of theafternoon presented and tested immediatelyafter each set.Finally, of thefirst day. The 1957 newsreel was tested immediately verbal recall istested byshowingthe subject picturesof after thissingle presentation and at thebeginning of the four people:a ministerof thechurch, a newspaper boy, next morningsession, approximately 18 hrlater. The 1937 adoctor, and apostman,and ineach case, providingthe newsreel was tested after itsfourth presentation and firstand second names.Again, there are three presenta- retested on themorning of thethird day, again after an tions,each followed byrecall testsof thefour names, 18-hr delay.The two newsreel testseach involveda total cued bytheir occupations, then followed bydelayed of 45 questions,each comprisinga recall test,followed by recall after a10-minfilled interval. afour-alternative forced-choice recognition test.Ques- tionsranged from specific detail suchas thenames of people shownand numbersof people killedin a parti- Continuous Recognition Test cular accident, tomore general issuesmentioned, such as Ina versionof thetest developed tostudy memory thestate oftheeconomy at aparticular timeand thecause under lightanesthesia, each of 16 listscomprised 23 of particular accidents. Virtuallyall theinformation ap- words,with two wordsrepeated immediately,and one peared tobe novelfor all subjects. each repeating at delaysof 1, 2, 4, 8, and 16 inter- polated words(Andrade et al.,1994; Andrade, 1996). In Acknowledgments order toincrease thelevel of difficulty,the distractor wordsused were semanticallyassociated withthe We thank Jon, hisparents, and hisfriends Andrew and Max, for target items.The wordswere read outat arate of 2 theirunfailing cooperation. We are grateful to Jackie Andrade, LarrySquire, and Alan Parkinfor generously providing details sec per word,with the subject respondingverbally ofmaterials and procedures, and to Kate Watkins,Ingrid whenever he detected arepetition. Johnsrude, and Louise Parryfor their technical assistance. Alan Baddeley acknowledges the support ofgrant G9423916 from the Medical Research Council. Delayed Recognition Tests Reprintrequests should besent to: Alan Baddeley, Department For each of these tests,24 target wordswere first ofExperimental , University ofBristol, 8 Woodland presented for studyon acomputer visualdisplay at a Road, Bristol BS8 1TNUK. Tel.:+44-117-928-8541; Fax: +44- rate of 3sec per item(Hamann &Squire, 1997). Each list 117-928-8562; e-mail:[email protected]. was preceded and followed bythree subsequentlyun- tested buffer wordsin order to minimizethe effects of primacyand recency. At study,the subject read each REFERENCES word aloud,and after 5minfilled byconversation, Aggleton, J.P.,& Brown,M. W.(1999). Episodic memory, recognition was tested. Inthe case of two-alternative amnesia, and the hippocampal–anterior thalamic axis. Be- forced-choice recognition, thesubject was tested by havioral and Brain Sciences,2, 425–490. being shown24 word pairs,each comprisingone item Aggleton, J.P.,& Shaw, C.(1996). Amnesia and recognition memory: Areanalysis ofpsychometric data. Neuropsycho- from thestudy list and one new item.He was asked to logia, 34, 51–62. sayaloud theword thathe hadbeen shownpreviously. Andrade, J.(1996). Investigations ofhypesthesia: Using Inthe case of theyes– no recognition test,a total of 24 anaesthetics to explore relationships between conscious- old(studied) words and 24 new (unstudied)words were ness, learning and memory. Consciousness and Cognition, presented one at atimein mixed order; ineach case, the 5, 562–580. Andrade, J.,Munglani, R.,Jones, J.G.,& Baddeley, A.D.(1994). subject was asked tosay whether he hadbeen shown Cognitive performance during anaesthesia. Consciousness thatword previouslyor not.A total of sixtest lists were and Cognition, 3, 148–165. given for each of thetwo conditions,beginning with Baddeley, A.D.(1993). Short-term phonological memory and three forced-choice lists,followed bythree yes–no lists, long-term learning: Asingle case study. European Journal of followed bythe remaining three forced-choice lists,and CognitivePsychology, 5, 129–148. Baddeley, A.D.(1996). Applying the psychology ofmemory to finally,the lastthree yes–no lists.The appropriate clinicalproblems. In D.Herrmann,C. McEvoy, C.Hertzog, recognition instructionswere given before each testlist. P.Hertel,& M.K.Johnson (Eds.), Basicand applied memory research: I (pp.195– 220). Mahwah, NJ:Erlbaum. Baddeley, A.D.(1997). Human memory: Theory and Practice. NewsreelRecognition and RecallTask (Revised Ed.).Hove, UK:Psychology Press. Baddeley, A.D.,Emslie, H.,&Nimmo-Smith,I. (1993). The The testwas based on presenting videoclips based on Spot-the-Word Test: Arobust estimate ofverbal intelligence excerpts from thePathe Movie newsreels from 1937 and based on lexical decision. BritishJournal ofClinical Psy- 1957, respectively.In each case, an initialsection of chology,32, 55–65.

Baddeley,Vargha-Khadem, and Mishkin367 Baddeley, A.D.,Emslie, H.,&Nimmo-Smith,I. (1994). The Mandler, G.(1980). Recognizing: Thejudgement ofprevious Doors and PeopleTest. Bury St. Edmunds, UK:Thames occurrence. PsychologicalReview, 87, 252–271. Valley Test Company. Mandler, G.,Pearlstone, Z.,& Koopmans, H.S.(1969). Effect of Baddeley, A.D.,& Warrington,E. K.(1970). Amnesia and the organization and semantic similarityon recall and recogni- distinction between long- and short-term memory. Journal tion. Journal ofVerbal Learning and Verbal Behavior,8, ofVerbal Learning and Verbal Behavior,9, 176–189. 410–423. Calev, A.(1984). Recall and recognition inchronic nonde- Manns, J.R.,& Squire, L.R.(1999). Impaired recognition mented schizophrenics. Journal ofAbnormal Psychology, memory on the Doors and People Test after damage 93, 172–177. limited to the hippocampal region. Hippoccampus, 9, Delis,D. C.,Kramer, J. H.,Kaplan, E.,& Ober,B. A.(1987). 495–499. California Verbal Learning Test:Form II. San Antonio, TX: McMackin, D.,Cockburn, J., Anslow, P., &Gaffan,D. (1995). Psychological Corporation. Correlationof fornix damage withmemory impairmentin six DuSoir,H., Kapur,N., Byrnes, D.P.,McKinstry, S.,& Hoare, cases ofcolloid cyst removal. Acta Neurochirugica,135, R.D.(1990). Therole ofdiencephalic pathology inhuman 12–18. memory disorder. Brain, 113, 1695–1706. Milner,B. (1963). Effect ofdifferent brain lesions on card Dy¨zel, E.,Vargha-Khadem, F.,Heinze, H.J.,& Mishkin,M. sorting. Archivesof Neurology,9, 90–100. (1999). ERPevidence forrecognition without episodic re- Milner,B. (1966). Amnesia followingoperation on the tem- collection inapatient withearly hippocampal pathology. porallobes. In C.W.M.Whitty& O.L.Zangwill (Eds.), Societyfor NeurosciencesAbstracts, 25, 259.11. Amnesia (pp.109– 133). London: Butterworth. Gabrieli,J. D.E.,Cohen, N.J.,& Corkin,S. (1988). Theim- Mishkin,M., Suzuki, W.,Gadian, D.G.,& Vargha-Khadem, F. paired learning ofsemantic knowledge followingbilateral (1997). Hierarchicalorganization ofcognitive memory. Phi- medial temporal-lobe resection. Brain and Cognition, 7, losophicalTransactions ofthe RoyalSociety of London, 157–177. SeriesB, 352, 1461–1467. Gadian, D.G.,Aicardi, J., Watkins,K. E.,Porter,D. A.,Mishkin, Mishkin,M., Vargha-Khadem, F.,& Gadian, D.G.(1998). Am- M.,& Vargha-Khadem, F. (2000). Developmental amnesia nesia and the organization ofthe hippocampal system. associated withearly hypoxic–ischaemic injury. Brain, 123, Hippocampus, 8, 212–216. 499–507 Morris,R. G.,Abrahams, S., Baddeley, A.D.,& Polkey, C.E. Gardiner,J. M.,&Java, R.I.(1993). Recognising and remem- (1995). Doors and People: Visualand verbal memory fol- bering.In A.F. Collins,S. E.Gathercole, M.A.Conway, &P. lowing unilateral temporal lobectomy. Neuropsychology,9 , E.Morris(Eds.), Theoriesof memory (pp.163– 188). Hove, 464–469. UK:Erlbaum. Parkin,A. J.,Dunn, J.C.,Lee, C.,O’ Hara,P. F.,& Nussbaum, L. Greene, J.D.W.,Baddeley, A.D.,& Hodges, J.R.(1996). (1993). Neuropsychological sequeli ofWernicke’ sence- Analysis ofthe episodic memory deficit inearly Alzheimer’s phalopathy ina twenty-year-old woman: Selective impair- Disease: Evidence fromthe Doors and People Test. Neu- ment ofa frontalmemory system. Brain and Cognition, ropsychologia,34, 537–551. 21, 1–19. Haist,F., Shimamura, A. P.,& Squire, L.R.(1992). On the Parkin,A. J.,& Hunkin,N. M.(1993). Impaired temporal relationship between recall and recogition memory. Journal context memory on anterograde but not retrograde ofExperimental Psychology:Learning ,Memory,and Cog- tests inthe absence offrontal pathology. Cortex, 29, nition, 18, 691–702. 267–280. Hamann, S.B.,& Squire, L.R.(1997). Intact perceptual mem- Parkin,A. J.,Leng, N.R.C.,&Hunkin,N. M.(1990). Differential ory inthe absence ofconscious memory. Behavioural sensitivity to context indiencephalic and Neuroscience,111, 850–854. amnesia. Cortex, 26, 373–380. Hanley, J.R.,Davies, A.D.M.,Downes, J.J.,& Mayes, A.R. Parkin,A. J.,& Walter,B. M.(1992). Recollective experience, (1994). Impaired recall ofverbal material followingrupture normal aging and frontaldysfunction. Psychologyand Aging, and repairof an anterior communicating artery aneurysm. 7, 290–298. CognitiveNeuropsychology, 11, 543–578. Reed, J.M.,&Squire, L.R.(1997). Impaired recognition Hirst,W., Johnson, M.K.,Kim, J. K.,Phelps,E. A.,Risse,G., & memory inpatients withlesions limited to the hippo- Volpe, B.T.(1986). Recognition and recall inamnesics. campal formation. Behavioral Neuroscience,111, Journal ofExperimental Psychology,12, 445–451. 667–675. Holdstock, J.S.,Mayes, A.R.,Cezayirli,E., Isaac, C.L.,Aggleton, Rozin,M. (1976). Thepsychobiological approach to human J.P.,&Roberts, N.(2000). Acomparison ofegocentric and memory. In M.R.Rosenzweig &E.L.Bennet (Eds.), Neural allocentric spatial memory ina patient withselective hip- mechanisms oflearning and memory (pp.3– 46). Cam- pocampal damage. Neuropsychologia,38, 410–425. bridge: MIT Press. Howard,D., & Patterson, K.(1992). Pyramid and palm trees: Rugg, M.D.,Schloerscheidt, A.M.,& Mark,R. E.(1998). An Atest ofsemantic accessfrom pictures and from words. electrophysiological comparison oftwo indices ofrecollec- Bury St. Edmunds, UK:Thames Valley Test Company. tion. Journal ofMemory and Language, 39, 47–69. Huppert,F. A.,&Piercy,M. (1979). Normaland abnormal Schacter, D.L.,& Tulving,E. (1994). Memory systems. Cam- forgetting inamnesia: Effect oflocus oflesion. Cortex, 15, bridge: MIT Press. 385–390. Schonfield, D.,& Robertson, B.A.(1966). Memory storage and Jacoby, L.L.(1994). Measuring recollection: Strategic versus aging. Canadian Journal ofPsychology, 20, 228–231. automatic influences ofassociative context. In C.Umilta& Shallice, T.(1988). From neuropsychology to mental struc- M.Moscovitch (Eds.), Attention and performance: XV. ture. Cambridge: Cambridge University Press. Conscious and non-conscious information processing (pp. Shepard, R.N.,&Teghtsoonian, M.(1961). Retention ofin- 661–680). Cambridge: MIT Press. formationunder conditions approaching asteady state. Kitchener, E.G.,Hodges, J.R.,&McCarthy, R.(1998). Journal ofExperimental Psychology,62, 302–309. Acquisition ofpost-morbid vocabulary and semantic Squire, L.R.(1992). Declarative and non-declarative memory: facts inthe absence ofepisodic memory. Brain, 121, Multiple brainsystems supporting learning and memory. 1313–1327. Journal ofCognitiveNeuroscience, 4, 232–243.

368Journal ofCognitive Neuroscience Volume 13,Number 3 Squire, L.R.,Knowlton, B.,& Musen, G.(1993). Thestructure ofnovel semantic informationin amnesia: Effects oflesion and organisation ofmemory. Annual Reviewof Psychology, location. Neuropsychologia,38, 484–492. 44, 453–495. Volpe, B.T.,Holtzman,J. D.,&Hirst,W. (1986). Further Squire, L.R.,&Zola, S.M.(1998). Episodic memory, semantic characterization ofpatients withamnesia after cardiac arrest: memory and amnesia. Hippocampus, 8, 205–211. Preserved recognition memory. Neurology,36, 408–411. Tulving,E. (1972). Episodic and semantic memory. In E.Tul- Warrington,E. K.(1984). Recognitionmemory test. Windsor, ving &W.Donaldson (Eds.), Organization ofmemory England: Nelson. (pp.381– 403). New York:Academic Press. Whittlesea, B.W.A.(1993). Illusions offamiliarity. Journal of Tulving,E. (1985). Memory and consciousness. Canadian Experimental Psychology:Learning, Memory,and Cogni- Psychologist,26, 1–12. tion, 19, 1235–1253. Tulving,E., & Markovitch, H.J.(1998). Episodic and declarative Wilson,B. A.,Alderman,N., Burgess, P.,Emslie, H.,&Evans, memory: Therole ofthe hippocampus. Hippocampus, 8, J. J. (1996) Behavioural assessment ofthe dysexecutive 198–204. syndrome. Bury St. Edmunds, UK:Thames Valley Test Vargha-Khadem, F.,Gadian, D.G.,Watkins, K. E.,Connelly, A., Company. Van Paesschen, W.,&Mishkin,M. (1997). Differentialeffects Wilson,B. A.,Clare,L., Cockburn, J. M.,Baddeley, A.D.,Tate, ofearly hippocampal pathology on episodic and semantic R.M.,& Watson, P.(1999). TheRivermead Behavioural memory. Science,277, 376–380. Memory Test–Extended Version. Bury St. Edmunds, UK: Verfaellie,M., Koseff,P., &Alexander, M.P.(2000). Acquisition Thames Valley Test Company.

Baddeley,Vargha-Khadem, and Mishkin369