Neuropsychology1996, Vol. 10, No. 1,32-41

Copyright 1996 by the American Psychological Association, Inc.0894-4105/96/S3.00

Impaired Retrieval From Remote Memory in PatientsWith Frontal Lobe Damage

Jennifer A. Mangels, Felicia B. Gershberg,and Arthur P. Shimamura

University of California, Berkeley

Robert T. KnightUniversity of California, Davis, and Veterans Affairs

Medical Center, Martinez

Patients with unilateral dorsolateral frontal lobe lesions and matched controls were given 2 tests ofremote memory for public information, the Public Events Test and the Famous Faces Test. Onboth tests, the patients with frontal lobe lesions exhibited impaired recall for remote information.Recognition memory was relatively preserved. Provision of semantic and phonemic cues in theFamous Faces Test did not completely compensate for their recall deficit. These findings suggestthat the remote memory impairment exhibited by frontal patients may be related to deficits instrategic search of memory. These deficits in retrieval from remote memory extend the array ofmemory deficits associated with damage to the frontal lobes.

Frontal lobe lesions have been associated with a variety ofmemory impairments, including deficits in short-term memory,free recall, metamemory, and memory for temporal informa-tion (for review, see Fuster, 1989; Shimamura, 1994; Stuss,Eskes, & Foster, 1994). Yet patients with frontal lobe lesionscan exhibit normal performance on standard clinical assess-ments of new learning ability (e.g., Wechsler Memory Scale—Revised; WMS-R; Wechsler, 1987), suggesting that theirmemory deficits are not due to severe anterograde amnesia(Janowsky, Shimamura, Kritchevsky, & Squire, 1989). Rather,the pattern of memory deficits suggests an impairment instrategic processes associated with organization and monitor-ing at encoding and retrieval. Indeed, the relationship betweenfrontal lobe function and strategic processing have beendeveloped in a variety of theories about the frontal lobes (seeBaddeley, 1986; Moscovitch, 1994; Norman & Shallice, 1986;Shimamura, 1994).

Organizational problems related to deficits in strategicprocesses are highlighted by the performance of frontal lobepatients on tests of free recall (Janowsky et al., 1989; Jetter,Poser, Freeman, & Markowitsch, 1986; Incisa della Rocchetta,1986; Incisa della Rocchetta, & Milner, 1993; Stuss, Alex-ander, et al., 1994). For example, patients with frontal lobelesions are less likely to engage in useful memory strategies,

Jennifer A. Mangels, Felicia B. Gershberg, and Arthur P. Shi-mamura, Department of Psychology, University of California, Berke-ley; Robert T. Knight, Department of Neurology, Center for Neurosci-ence, University of California, Davis, and Veterans Affairs MedicalCenter, Martinez, California. Jennifer A. Mangels is now at theRotman Research Centre, Toronto, Ontario, Canada; Felicia B.Gershberg is now at Boston University School of Medicine and theVeterans Affairs Medical Center, Boston, Massachusetts.

This research was supported by National Institutes of Health GrantsAG09055, MH48757 NS23115, and PO NS17778 and National ScienceFoundation graduate research fellowships. CT lesion reconstructionsare evidence of the artistic and computer talents of Clay Clayworth.

Correspondence concerning this article should be addressed toArthur P. Shimamura, Department of Psychology, University ofCalifornia, Berkeley, California 94720. Electronic mail may be sent viaInternet to aps@garnet.berkeley.edu.

such as subjective organization or category clustering (Eslinger &Grattan, 1994; Gershberg & Shimamura, 1995; Stuss, Alex-ander, et al., 1994). Such strategies are particularly importanton free recall tests because these tests provide few retrievalcues other than a general reminder of the experimentalcontext. Tests of cued recall and recognition depend less onthis type of contextual cueing, and patients with frontal lobelesions exhibit less or variable impairment on these tests(Janowsky et al., 1989; Stuss, Alexander, et al., 1994).

Recent findings indicate that the frontal lobes are involvedin organization at the time of learning. For example, on testswith categorized word lists, free recall performance of frontallobe patients can be facilitated by encoding manipulationssuch as blocking category members (Incisa della Rocchetta &Milner, 1992; Stuss, Alexander, et al., 1994) or providingcategory cues at study (Jetter et al., 1986; Gershberg &Shimamura, 1995). In addition, patients with frontal lobelesions were impaired in the ability to sort pictures intotaxonomic categories (Incisa della Rocchetta, 1986). In thatstudy, sorting performance in patients with right hemispherelesions was correlated with performance on a subsequent freerecall test. Thus for those patients, there was a direct relation-ship between organization at encoding and recall perfor-mance.

Other findings suggest that organizational problems at thetime of retrieval contribute to memory deficits in patients withfrontal lobe lesions. For example, free recall performance oncategorized word lists in frontal patients is benefited byproviding category cues at the time of test (Jetter et al., 1986;Incisa della Rocchetta & Milner, 1993; Gershberg & Shi-mamura, 1995). In some cases, retrieval cues improved theperformance of frontal lobe patients to the level of controls(Jetter et al., 1986; Incisa della Rocchetta & Milner, 1993). Inanother case, cues provided at either study or test equallybenefited free recall performance in these patients, althoughthe presentation of cues at both study and test did not fullycompensate for their deficit (Gershberg & Shimamura, 1995).These findings suggest that patients with frontal lobe lesionsmay have deficits at both encoding and retrieval stages.

Tests of remote memory may provide a better means of

32

FRONTAL LOBES AND IMPAIRED REMOTE MEMORY 33

assessing retrieval deficits than tests of new learning becauseremote memory tests generally assess memory for informationlearned prior to the onset of brain injury. Previously, tests ofremote memory for public events, famous faces, or otheraspects of general knowledge have been used to assess retro-grade amnesia following medial temporal lesions, KorsakofFssyndrome, Alzheimer's disease, and other neurological disor-ders (Albert, Butters, & Levin, 1979; Kopelman, 1989, 1991;Sanders & Warrington, 1971; Squire & Cohen, 1984; Squire,Haist, & Shimamura, 1989). Although remote memory forpublic information has not been extensively assessed in pa-tients with circumscribed frontal lobe lesions, some studieshave suggested that retrograde amnesia in patients withKorsakofFs syndrome and Alzheimer's disease is related togeneral cortical atrophy or specifically to frontal lobe damage(Mayes, 1986; Kopelman, 1991; Shimamura & Squire, 1986).In particular, Kopelman (1991) found that the performance ofpatients with KorsakofFs syndrome or Alzheimer's disease ona remote memory test was correlated with performance ontasks sensitive to frontal lobe pathology (e.g., verbal fluency,card sorting, or cognitive estimation). Yet remote memoryperformance was not significantly correlated with computer-ized tomography (CT) measures of frontal lobe atrophy inthese patient groups (Dall'Ora, Delia Sala, & Spinnler, 1989;Kopelman, 1989, 1991). Thus, although impairments in re-trieval from remote memory have been attributed to frontallobe damage in these patients, the evidence has been indirect.

One recent study reported impaired remote memory perfor-mance in patients with circumscribed frontal lobe lesions(Delia Salla, Laiacona, Spinnler, & Trivelli, 1993). In thatstudy, autobiographical memory was assessed. The retrieval ofautobiographical memories can place high demands on the useof organizational retrieval strategies because information mustbe retrieved from specific spatio-temporal contexts. In addi-tion, retrieval from autobiographical memory requires individu-als to evaluate the plausibility of the retrieved memories and toresume search if retrieved memories are not appropriate orveridical (Baddeley & Wilson, 1986; Moscovitch, 1994; Reiser,Black, & Kalamarides, 1986). If these evaluative processes(i.e., metamemory) are impaired, confabulations may occur.Confabulations have been reported in case studies of amnesiapatients with frontal lobe damage (Baddeley & Wilson, 1986).Yet, in the autobiographical memory study by Delia Salla et al.(1993), patients with circumscribed frontal lesions did notappear to make errors of confabulation, but rather demon-strated a reduction in the number of recollections and detailsabout autobiographical information. Thus, frontal lobe lesionsappeared to impair the ability to organize a productive searchof remote memory, rather than the ability to evaluate theplausibility of the retrieved information. Nonetheless, in auto-biographical studies it is sometimes difficult to ensure that theinformation being tested is verifiable (Squire & Cohen, 1984).Responses on tests of remote memory for public knowledgemay be more easily validated and discriminated from confabu-lation.

The present study assessed the role of dorsolateral prefron-tal cortex in retrieving remote memory for public events (e.g.,Cohen & Squire, 1981) and famous faces (e.g., Albert, Butters, &

Levin, 1979; Hodges, Salmon, & Butters, 1993).1 For each test,measures of free recall and recognition memory were ob-tained. To the extent that damage to dorsolateral prefrontalcortex impairs strategic retrieval processes, performance ofpatients with lesions in this brain region should be particularlyimpaired on free recall tests of remote memory relative toperformance on recognition tests. Additionally, for the Fa-mous Faces Test, we assessed the benefits of providingsemantic and phonemic cues during retrieval. If the strategicretrieval deficit of frontal patients results from a deficit inselecting appropriate search cues, then providing semantic andphonemic cues could disproportionately improve the perfor-mance of frontal patients relative to controls.

Method

Participants

Patients with frontal lobe lesions. Six patients with unilateral lesionsin the region of dorsolateral prefrontal cortex were identified by reviewof medical records and CT scans. Patients had a single cerebralinfarction in the territory of the precentral branch of the middlecerebral artery (see Figure 1). Four patients had left hemispherelesions (3 men and 1 woman), and 2 patients had right hemispherelesions (1 man and 1 woman). In 1 patient (J.D.), recent MRI scansrevealed that the lesion extended posteriorly into parietal cortex.Average lesion volume was estimated from quantitative analyses ofneuroimaging data to be 30.9 ml. Patients had no known history ofother significant medical disease such as psychiatric disorder, demen-tia, substance abuse, or additional neurological events (e.g., headinjury). For all patients, cardiovascular accident (CVA) occurredbetween 1980 and 1986, with an average of 11 years between CVA andcurrent testing. The patients averaged 67.2 years of age and had anaverage of 14.0 years of education.

On standardized neuropsychological tests (see Table 1), patientsscored within the normal range. Their mean score on the Full ScaleWechsler Adult Intelligence Scale—Revised (WAIS-R; Wechsler,1981) was 104.8 (norm = 100). Their mean scores on the five scales ofthe WMS-R were as follows: Attention = 85.7, General Memory =99.2; Delayed Memory = 86.5; Verbal Memory = 95.3; and VisualMemory = 104.3 (norm = 100). The patients were impaired, however,on tests sensitive to frontal lobe injury. Compared with controls,patients achieved fewer categories on the Wisconsin Card SortingTest, F(l, 14) = 7.9,p < .05, and made more perseverative errors, F(l,14) = 8.2,p < .05. The patients also produced fewer words on the FASverbal fluency test (Benton & Hamsher, 1978), F(l, 14) = 8.6,/? < .05,and named fewer items spontaneously on the Boston Naming Test,F(l, 14) = 7.1, p < .05. Two of the 4 patients with left hemispherelesions were mildly dysfluent and had some difficulty in word finding(J.D. and A.L.). However, none exhibited severe aphasic disorder, asall had Western Aphasia Battery scores of 85 or higher (norm = 100).

Controls. Ten healthy volunteers (6 men and 4 women) partici-pated as controls. These individuals were recruited at the outpatientclinic of the Veterans Affairs Medical Center, Martinez, California,and were matched to the patients with frontal lobe lesions with respectto age (65.7 years) and education (14.9 years). Controls performedcomparably to the patients with frontal lobe lesions on the Informationsubtest (controls = 22.9 and frontal patients = 19.7) and the Digit

1 Nelson Butters provided the stimulus materials for the FamousFaces Test and Larry Squire provided stimulus materials for the PublicEvents Test, the two tests of remote memory that were used in thisstudy.

34 MANGELS, GERSHBERG, SHIMAMURA, AND KNIGHT

cvjD.

\OC

•o

I §U '33

CDLU

FRONTAL LOBES AND IMPAIRED REMOTE MEMORY 35

Table 1Test Variables and Characteristics of Study Participants

Patient

J.D.K.K.A.L.O.A.E.B.M.M.

FrontalMenWomen

ControlMenWomen

Sex

ManManWomanManWomanMan

42

64

Lesionlocation

LeftLeftLeftLeftRightRight

LeftRight

Volume(ml)

30.817.451.217.517.351.2

30.9

Onset(year)

198619841980198419831984

1983

Age(years)

666264627668

67.2

65.7

Education(years)

201313141212

Mean

14.0

14.9

FSIQ

9310710412410794

104.8

WMS-R

871001091159193

99.2

FAS

83121291612

24.8

44.6

BNT

435650565350

51.3

56.9

\

Category

644140

3.2

5.4

VCST

Perseverative

145111313292

38.5

9.9

Note. Wechsler Memory Scale—Revised (WMS-R) scores indicate the General Memory Index. Boston Naming Test (BNT) scores indicatenumber of items spontaneously named. Category indicates number of category attained. FAS (actual letters used in test) score indicates thenumber of words correct on this verbal fluency measure. FSIQ = Full Scale IQ determined from the Wechsler Adult Intelligence Scale—Revised.WCST = Wisconsin Card Sorting Test. Blank cells indicate scores not available.aLesion locations were determined by brain scan. Indicates year when stroke originally occurred.

Span subtest (controls = 16.3 and frontal patients = 13.0) of theWAIS-R. However, controls performed marginally better than thepatients on the Vocabulary subtest (controls = 54.7 and frontalpatients = 41.8), F(\, 14) = 4.4, p = .06. Controls also performedbetter than the frontal patients on the Digit Symbol subtest (controls =50.3 and frontal patients = 37.8), F(l, 14) = 5.3,p < .05.

Remote Memory Tests

Free recall and recognition on the Public Events Test. Memory for 92public events that occurred during the 1950s (18 items), 1960s (24items), 1970s (20 items), and 1980s (30 items) was assessed. Perfor-mance was measured with a free recall test (e.g., "Who killed JohnLennon?") and a four-alternative, forced-choice recognition test(names of individuals such as John Hinkley, Sara Jane Moore, DavidRoth, or Mark Chapman). All of the recall responses required a word,name, or phrase. All questions had been used in a previous study(Squire et al., 1989), which used an updated version of the testdeveloped by Cohen and Squire (1981). The test was administered inthree successive blocks of 31 questions, each of which includedquestions from all decades. First, the free recall test for each block wasadministered orally, then the recognition test for the same questionswas presented to the participant in written form.

Free recall, cued recall, and recognition on the Famous Faces Test.Black-and-white photographs of 85 famous people were used asstimuli. The stimuli were divided into six decades according to the timeperiod in which the person achieved fame: 1940s (15 items), 1950s (15items), 1960s (15 items), 1970s (15 items), 1980s (15 items), and 1990s(10 items). This test was a version of the Boston Famous Faces Testoriginally developed by Albert et al. (1979) and most recently updatedby Hodges, Salmon, and Butters (1993). In the free recall portion ofthe test, patients were asked to recall the name of the person shown inthe photograph (e.g., Telly Savalas). Participants were given ampletime to respond. If they were unable to recall the name, participantswere given a series of cues. First, a semantic cue was given (e.g., "Hewas an actor, American, and had a popular series on TV"). If thesemantic cue failed to elicit a correct response, a phonemic cue wasgiven (i.e., the person's initials).

In a later session (range: same day to 6 months), participants weregiven a four-alternative, forced-choice recognition test for all pictures

(e.g., Yul Brynner, David Soul, Don Rickles, or Telly Savalas). Themultiple-choice recognition test was constructed specifically for thepresent experiment in order to compare recognition memory forfamous faces with recognition memory for public events. Differentnames were used as distractors for each of the faces.

Results

Public Events Test

Figure 2 displays mean free recall and recognition scores forthe frontal patients and controls on the Public Events Test. Onthe free recall test, a 2 (group) x 4 (decade) analysis ofvariance (ANOVA) revealed an overall group differencebetween frontal patients and controls, F(l, 14) = 6.53,p < .05.There was also a significant effect of decade, F(3, 42) = 15.09,p < .0001. Across decades, free recall performance was bestfor the most remote decade (the 1950s) and poorest for themost recent decade (the 1980s), as illustrated by a significantdownward linear trend in performance, F(l, 60) = 15.66,p <.001. The absence of a Group x Decade interaction, F(3,42) = .28, ns, indicates that this temporal gradient was similarin both groups.

In contrast to their performance on the free recall test,frontal patients were not impaired overall on the recognitiontest for public events, F(l, 14) = 1.74, ns. There was asignificant effect of decade, F(3, 42) = 5.75, p < .01, and asignificant interaction of Group x Decade, F(3, 42) = 3.41,p < .05. Post hoc analyses showed that frontal patients weresignificantly impaired on events from the 1960s; t(14) = 2.27,p < .05 (all t tests were two-tailed); and were marginallyimpaired on events from the 1980s, f(14) = 1.86, p = .08.Comparisons within subject group showed that for controls,performance was only marginally different between the 1970sand 1980s, t(9) = 2.08, p = .06, and between the 1970s and1950s, t(9) = 2.11, p = .07. However, the performance offrontal patients on events from the 1960s was significantly

36 MANGELS, GERSHBERG, SHIMAMURA, AND KNIGHT

Public Events

100-i

80-

60-

40-

20-

0

Free Recall

'50s '60s 70s '80s

Decade

100 -j

80-

60-

40-

20-

Recognition

'50s '60s 70s '80s

Decade

Frontal Patients

Controls Subjects

Figure 2. Free recall and recognition performance of patients with frontal lobe lesions and controls onthe Public Events Test. On the graph of recognition performance (right), the dashed line indicates chanceperformance level (25%). Error bars indicate standard error of the mean.

worse than events from either the 1950s, t(5) = 6.41, p < .01,or the 1970s, f(5) = 4.43, p < .01.

Famous Faces Test

Performance of frontal patients and controls on the freerecall, cued recall, and recognition measures of the FamousFaces Test is illustrated in Figure 3. A 2 (group) x 6 (decade)

ANOVA showed that frontal patients were impaired on thefree recall measure of the Famous Faces Test, F(l, 14) = 8.45,p < .05. There was also a significant effect of decade, F(5,70) = 5.37, p < .001. There was no Decade x Groupinteraction, F(5, 70) = .70, ns. Both groups demonstrated thebest performance on the most remote decades (1940s and1950s) and poorest performance on the middle decades (1960sand 1970s) and the most recent decade (1990s;;) > .05).

Famous Faces

100-

80-

I 60"o1 to-

°" 20-

0

Free Recall Cued Recall Recognition

•40s '50s '60s 70s '80s '90s

Decade

100n

80-

60-

40-

20-

0-

ft"""&-^^*--5-^| T "*" _ T •*-

i~~!i~T~T>41

100-

80-

60-

40-

20-

0-

»- — A-_ - - T

•̂ J.

'40s '50s '60s 70s '80s '90s

Decade

•40s '50s '60s 70s '80s '90s

Decade

Frontal Patients

Control Subjects

Figure 3. Free recall, cued recall, and recognition performance of patients with frontal lobe lesions andcontrols on the Famous Faces Test. On the graph of recognition performance (right), the dashed lineindicates chance performance level (25%). Error bars indicate standard error of the mean.

FRONTAL LOBES AND IMPAIRED REMOTE MEMORY 37

On the recognition test, a 2 (group) x 6 (decade) ANOVAindicated that frontal patients did not exhibit an overallimpairment, F(l, 14) = 1.25, ns. There was no significant effectof decade, F(5, 70) = 1.29, ns, but there was a significantinteraction of Group x Decade, F(5, 70) = 2.60, p < .05. Posthoc analyses showed that patients were impaired relative tocontrols on faces from the 1940s, f(14) = 2.19,;; < .05. Ceilingeffects in this task may have obscured additional groupdifferences, as performance on this measure approached 100%for some frontal patients and controls. Comparisons withinsubject group showed that for controls, recognition perfor-mance did not differ across decades, whereas for frontalpatients, recognition performance for faces from the 1970s wassignificantly poorer than faces from the 1960s, t(5) = 2.91,p <.05, and 1980s, t(5) = 7.01, p < .001. In addition, for frontalpatients, recognition performance for faces from the 1940s wasmarginally worse than for faces from the 1980s, ?(5) = 2.42,p = .06.

On the cued recall test, performance of frontal patients wasimpaired relative to controls F(l, 14) = 9.79, p < .01. Theoverall effect of decade, F(5, 70) = 1.84, ns, and Group xDecade interaction were not significant, F(5, 70) = 0.09, ns.For cued recall, the standard procedure was used in whichunsuccessful free recall was first followed by a semantic cueand then by a phonemic cue (e.g., Hodges et al., 1993; Squire etal., 1989). Thus, overall cued recall performance reflects thebenefit of both semantic and phonemic cueing.

The specific benefit of semantic cues and the benefit ofadding phonemic cues were also analyzed. For each partici-pant, the benefit of semantic cueing was calculated by subtract-ing the percentage correct following free recall from thepercentage correct following semantic cueing alone. Thisdifference reflects increase in performance because of seman-tic cueing. The benefit of phonemic cueing was determined bysubtracting the percentage correct with semantic cues from

total cued recall performance (i.e., semantic cueing and theaddition of a phonemic cue). This difference reflects theadditional increase in performance because of phonemiccueing.

A 2 (group) x 2 (benefit of semantic vs. phonemic cueing)ANOVA revealed a significant Group x Cue interaction, F(l,14) = 8.73,p < .01. As illustrated in Figure 4, whereas controlsbenefited equally from either semantic cues alone or theaddition of phonemic cues, t(9) = .35, ns, the performance offrontal patients improved to a greater extent from the additionof phonemic cues, as opposed to semantic cues alone, t(5) =2.62,p < .05. In addition, frontal patients and controls did notdiffer in the amount of improvement gained from semanticcueing, t(l4) = .80, ns, but phonemic cueing benefited frontalpatients to a greater extent than controls, t(l4) = 2.99,p < .01.

The interpretation of a Group x Cue interaction is compli-cated by differences in initial free recall performance betweenthe two groups. Therefore, we attempted to equate free recallperformance of the controls and frontal patients. Items wereranked by the number of controls that had correctly recalledthe name of that face on the free recall test (see Albert et al.,1979). Items that had been answered correctly by 7 or morecontrols were considered easy. The performance of controlswas lowered by excluding these easy questions from theanalysis. This procedure resulted in the elimination of 42questions for controls. The remaining 43 items were distrib-uted across decades in the following manner: 1940s = 7 items,1950s = 7 items, 1960s = 10 items, 1970s = 7 items, 1980s = 7items, and 1990s = 5 items. The performance of the frontalpatients was raised by removing those items that were notanswered correctly by any of the frontal patients and thereforewere considered hard. This procedure resulted in the elimina-tion of 16 questions for frontal patients. The remaining 69items were distributed across decades in the following manner:1940s = 14 items, 1950s = 12 items, 1960s = 9 items, 1970s =

lOO-i

80-

60-

40-

20-

Unadjusted Scores100-1

g 60H

1jj 40-

20-

Adjusted Scores

Frontal Patients Control Subjects Frontal Patients Control Subjects

Figure 4. Total percentage correct partitioned into percentage recalled spontaneously (free recall),percentage of items recalled with a semantic cue alone, and percentage of items recalled following theaddition of a phonemic cue. Unadjusted scores (all items) are shown on the left, and adjusted scores areshown on the right. The adjustment method is detailed in the Results section.

38 MANGELS, GERSHBERG, SHIMAMURA, AND KNIGHT

12 items, 1980s = 14 items, and 1990s = 8 items. Theseadjustments equated the two groups on free recall, F(l, 14) =.06, ns. Indeed, for the adjusted scores, the overall free recallperformance of the frontal patients was numerically betterthan that of the controls (controls: 44.52 and frontal patients:47.34).

A 2 (group) x 2 (type of cue) ANOVA with these adjustedscores confirmed a significant Group x Cue interaction, F(l,14) = 5.58,p < .05. However, the overall effect of cue type wasno longer significant, F(l, 14) = 2.39, ns, and there was nooverall effect of group, F(l, 14) = .96, ns. As with theunadjusted scores, semantic and phonemic cues benefitedcontrols equally, t(9) = .25, ns, but frontal patients received amarginally greater benefit from phonemic cues than semanticcues, t(5) = 2.46, p = .06. Comparisons across subject groupsindicated that controls were benefited by semantic cueing to agreater extent than the frontal patients, <(14) = 2.29, p < .05.This pattern of performance differed from that of the unad-justed scores. Specifically, when the overall difficulty of thecued items was equated, patients with frontal lobe lesionsappeared to receive less benefit from semantic cueing thancontrols, rather than more benefit from phonemic cueing.

Discussion

Patients with dorsolateral frontal lobe lesions were impairedrelative to controls on free recall tests of remote memory forboth public events and famous faces. Frontal patients were lessimpaired on recognition tests of the same material. Thedisproportionate impairment of frontal patients on tests offree recall suggests that their retrieval deficit may be related tothe development and initiation of retrieval strategies. Thesefindings extend previous reports of impaired retrieval ofremote autobiographical memories in patients with frontallobe lesions (Delia Salla et al, 1993). When demands onstrategic retrieval processes are high, impaired retrieval fromremote memory can occur in patients with frontal lobe lesionseven when opportunity for confabulation is low.

Frontal patients and controls exhibited similar patterns ofrecall performance across decades. Specifically, frontal pa-tients were equally impaired relative to controls on retrieval ofpublic knowledge from the decades before and after braindamage occurred (around 1980 for most patients). This findingsuggests that patients with dorsolateral frontal lobe lesionswere primarily impaired on retrieval of remote memoriesrather than on encoding or storage of this information.Although this conclusion appears to conflict with findings fromprevious studies of new learning, in which patients with frontallobe lesions demonstrated impairment at both encoding andretrieval stages (e.g., Gershberg & Shimamura, 1995), informa-tion about public events and famous people may require lessstrategic organization at the time of learning than lists ofwords. For example, public information is usually acquired in amore elaborate encoding context and with many more learningexposures than are word lists in free recall experiments.

Free recall performance for public events was temporallygraded, such that retrieval from remote decades was superiorto retrieval from recent decades. However, the magnitude ofthis temporal gradient did not differ between groups, suggest-

ing that this gradient resulted from differences in the initialdifficulty of items across decades. Recent public events mayhave weaker memory traces because they have had lessexposure, fewer rehearsals, or generally less time for consolida-tion (Mayes et al., 1994). It is also likely that item selectioneffects are responsible for the effects of decade found in theFamous Faces Test. Items from the decades with the poorestperformance (i.e., 1960s and 1970s) tended to focus onindividuals associated with political events.

The similar pattern of performance of frontal patients andcontrols across decades contrasts with findings from patientswith Korsakoffs syndrome. Patients with KorsakofFs syn-drome often demonstrate greater retrieval deficits for informa-tion from recent decades than from remote decades relative tocontrols (Albert et al., 1979; Sanders & Warrington, 1971;Squire et al., 1989). It has been proposed that the extensive,temporally graded retrograde amnesia found in KorsakofFssyndrome results from an anterograde amnesia of insidiousonset superimposed on a separate retrieval deficit that spansall decades (Cohen & Squire, 1981; Shimamura & Squire,1989; Squire & Cohen, 1984). Some studies have suggestedthat the pervasive retrieval deficit in patients with KorsakofFssyndrome is associated with damage to the frontal cortex(Shimamura & Squire, 1989). Findings from the present studydirectly demonstrate that deficits in retrieval of remote memorycan result from circumscribed damage to the prefrontal cortex.

The disproportionate impairment of patients with frontallobe lesions on free recall performance compared with recogni-tion suggests that their retrieval deficits are related to deficitsin strategic processing. Nonetheless, the provision of semanticand phonemic cues in the Famous Faces Test did not eliminatethe remote memory deficit of these patients. Although somestudies of new learning have demonstrated that provision ofretrieval cues can improve performance to the level of controls(letter et al., 1986; Incisa della Rocchetta & Milner, 1993), theinability of retrieval cues to mitigate completely the memorydeficits of patients with dorsolateral frontal lobe lesions isconsistent with a previous study of new learning involving thesame patient group (Gershberg & Shimamura, 1995).

Findings from the cued recall portion of the Famous FacesTest also indicate that semantic and phonemic cues are notequally effective in directing retrieval processes in patientswith dorsolateral frontal lobe lesions. Whereas controls re-ceived a comparable amount of benefit from semantic andphonemic cues, frontal patients benefited to a greater extentfrom the addition of phonemic cues. Comparisons acrossgroups revealed that frontal patients also received less benefitfrom semantic cues than controls. Several interpretations ofthis cueing pattern are possible. First, frontal patients mayderive greater benefit from phonemic cueing because strategicretrieval is more impaired at the phonemic level. Second,semantic cueing may be less effective because frontal patientsare impaired at utilizing semantic information in the searchprocess because of basic deficits in semantic processing. Third,frontal patients may not have a basic deficit in using semanticinformation to access remote memory. Rather, semantic cue-ing may be less effective for frontal patients because semanticcues (e.g., "The person is an American actor who had apopular series on TV.") are typically less specific than phone-

FRONTAL LOBES AND IMPAIRED REMOTE MEMORY 39

mic cues (e.g., "The initials of the person are T. S.") and thusmay fail to focus memory retrieval adequately. Each of thesehypotheses will be considered separately below.

The first hypothesis proposes that patients with dorsolateralfrontal lobe lesions appear to gain more from phonemic cueingbecause they have difficulty in generating phonological codes.To evaluate further this hypothesis about performance on freerecall, we correlated cued recall and recognition tests withperformance on neuropsychological tests involving word find-ing (i.e., FAS) and naming (i.e., Boston Naming; see Table 1).This correlational analysis (see Table 2) revealed significantassociations between famous faces recall performance offrontal patients and their performance on the Boston NamingTest. Free and cued recall performance of controls was notcorrelated with Boston Naming Test performance, but therewas little variability in the scores of this group. The relation-ship between performance on the Famous Faces Test and theBoston Naming Test is plausible given that the the latterrequires participants to name objects depicted by simple linedrawings and therefore involves retrieval processes that aresimilar to those used in the Famous Faces Test. Performanceon the Boston Naming Test was not correlated in either groupwith performance on the Public Events Test. Instead, freerecall performance on the Public Events Test was significantlycorrelated with performance on the Information subtest of theWAIS-R. Again, this is not surprising given that both thePublic Events Test and the Information subtest require theretrieval of remote general information.

As the Boston Naming Test requires retrieval of bothsemantic and phonological information, the finding that thefamous faces recall performance of patients with frontal lobelesions was related to their performance on this naming task

provides limited support for the hypothesis of a specificimpairment in processing at the phonological level. However,the phonological deficit hypothesis is called into question bythe lack of a correlation between recall performance andperformance on the FAS test of verbal fluency in either group.The FAS test requires the generation of items beginning with agiven letter and therefore emphasizes a phonologically basedsearch of memory. In addition, a selective phonological deficitpredicts that frontal patients and controls would benefitequally from semantic cueing and that frontal patients woulddemonstrate a greater benefit from phonological cueing. How-ever, when cued recall was adjusted for overall free recallperformance, this pattern of performance did not emerge.Instead, frontal patients and controls appeared to benefitsimilarly from phonemic cues, whereas frontal patients ben-efited less from semantic cues.

The second hypothesis—a deficit in semantic processing infrontal patients—could account for the differential pattern ofcueing effects. Several functional imaging studies have shownincreases in activation of the dorsolateral prefrontal cortexduring semantic processing such as judging the categorymembership of a word or generating a related verb (e.g.,Demonet et al., 1992; Kapur, Craik, et al., 1994; Kapur, Rose,et al., 1994; Petersen, Fox, Posner, Mintun, & Raichle, 1988,1989; Posner, Petersen, Fox, & Raichle, 1988). One interpreta-tion of findings from these studies has been that the frontallobes are fundamentally involved in semantic processing ofverbal information. Thus, in the current study, patients withfrontal lobe damage may have been impaired in their ability toprocess and use semantic cues to retrieve information fromremote memory. Yet, if patients with frontal lobe lesionsreceived little benefit from semantic cues but were benefited

Table 2Correlations of Performance With Standardized Memory Tests

WAIS-R

Test and group

Spontaneous recallFrontalControl

Cued recall (unadjusted)FrontalControl

Cued recall (adjusted)FrontalControl

RecognitionFrontalControl

Spontaneous recallFrontalControl

RecognitionFrontalControl

BNT

.889

.170

.861

.053

.905-.189

.127

.216

.559

.115

.031

.192

FAS

Famous Faces Test

* .599.036

* .626.013

* .580.173

.163

.231

Public Events Test

.071

.063

.615

.143

Information

.531

.285

.600

.513

.542

.369

.754

.379

.835*

.606

.417

.488

Vocabulary

.746-.037

.731

.513

.772

.034

.098

.096

.582

.155

.027

.140

Note. BNT = Boston Naming Test; WAIS-R = Wechsler Adult Intelligence Scale—Revised; FAS = averbal fluency test that uses the actual letters F, A, and S.*p < .05.

40 MANGELS, GERSHBERG, SHIMAMURA, AND KNIGHT

by phonemic cues in a normal way, an overall reduction ofcueing effects in frontal patients may be expected. Nonethe-less, regardless of whether cueing effects were unadjusted oradjusted for free recall performance, the amount of overallbenefit from cueing was not significantly different betweengroups. The semantic deficit hypothesis is further underminedby findings that frontal patients performed normally on im-plicit tests of memory for conceptual information, such ascategory generation priming (Gershberg, 1995).

Conflicting findings and alternative interpretations in thefunctional imaging literature have also called into question theprecise relationship between semantic processing and thefrontal lobes (e.g., Demonet, Wise, & Frackowiak, 1993;Grasby, Frith, Friston, Bench, Frackowiak, & Dolan, 1993;Kapur, Rose, et al., 1994; Raichle, 1994; Shallice et al., 1994;Tulving, Kapur, Craik, Moscovitch, & Houle, 1994). In somefunctional imaging studies, activation of prefrontal cortexduring semantic processing tasks has been interpreted asresulting from subjects engaging in strategic, organizationalprocesses rather than basic semantic analysis (Grasby et al.,1993; Kapur, Rose, et al., 1994).

The third hypothesis is that semantic cues were not effectivein organizing or focusing retrieval processes. The semanticcues developed for the Famous Faces Test mainly describe thegeneral characteristics of the famous person's occupation (e.g.,actress, political figure, comedian, or sports figure) or national-ity (e.g., German, Russian, or British). Although the semanticcues occasionally included somewhat more specific personalcharacteristics (e.g., sex symbol, Democrat, stand up comic ontelevision, or won Wimbledon), cues describing informationunique to the individual (e.g., Candice Bergen: actress, plays"Murphy Brown," Emmy Award winner; Mikhail Gorbachev:Russian, was Secretary General of USSR [former Union ofSoviet Socialist Republics], instigator of Glasnost Public Policy)were provided for only 31.8% of the total items and 33.3% ofthe adjusted items. General cues may result in the activation ofa large set of possible responses, requiring participants toselect the appropriate response from similar and plausiblecompetitors. This selection process may require the inhibitionof interference from these competitors. Phonemic cues (e.g.,the initials of the person) may be more effective retrieval cuesbecause they are more specific and help to rule out many of thecompeting responses.

Thus, it is possible that patients with dorsolateral frontallobe lesions do not benefit as greatly from semantic cuesbecause these patients are more susceptible to interference.Indeed, patients with frontal lobe lesions exhibit increasedinterference on some tests of new learning. For example,Shimamura, Jurica, Mangels, Gershberg, and Knight (in press)found that patients with frontal lobe lesions were disproportion-ately impaired on paired-associate learning paradigms inwhich the same cue word was paired with a different responseword across lists (i.e., List 1: thief-crime and List 2: thief-bandit). In addition, Incisa della Rocchetta and Milner (1993)demonstrated that part-list cueing (i.e., providing some of thewords from a study list as retrieval cues) disproportionatelyimpaired free recall performance in patients with damage tothe left frontal lobe.

Although this interference hypothesis is compelling, certain

qualifications must be made. Most notably, as in previousstudies of the Famous Faces Test (e.g., Hodges et al., 1993;Squire et al., 1989), the order of administering semantic andphonemic cues was not counterbalanced. Semantic cues werealways provided first, followed by phonemic cues. It is possiblethat semantic cues would have provided adequate cueingbenefits if they had been provided after the phonemic cues.That is, patients with dorsolateral frontal lobe lesions maysimply require more time or a greater number of cues withwhich to conduct a search of memory.

In summary, patients with lesions in the region of dorsolat-eral prefrontal cortex demonstrated deficits in retrieval ofpublic information from remote memory. The pattern of thesedeficits suggests an impairment in the ability to organizeeffective retrieval strategies. Yet, although the provision ofretrieval cues improved performance in these patients, it didnot bring their performance to the level of controls. Inaddition, unlike controls, patients with dorsolateral frontallobe damage benefited more from phonemic cues than fromsemantic cues. The present findings of impaired remotememory add to the growing list of memory disorders associatedwith frontal lobe lesions and provide further evidence for ageneral deficit in strategic memory processes in these patients.

References

Albert, M. S., Butters, N., & Levin, J. (1979). Temporal gradients inthe retrograde amnesia of patients with alcoholic Korsakoffsdisease. Archives of Neurology, 36, 211-216.

Baddeley, A. D. (1986). Working memory. Oxford, England: OxfordUniversity Press.

Baddeley, A. D., & Wilson, B. (1986). Amnesia, autobiographicalmemory, and confabulation. In D. C. Rubin (Eds.), Autobiographicalmemory (pp. 225-252). Cambridge, England: Cambridge UniversityPress.

Cohen, N. J., & Squire, L. R. (1981). Retrograde amnesia and remotememory impairment. Neuropsychologia, 19, 337-356.

Dall'Ora, P., Della Salla, S., & Spinnler, H. (1989). Autobiographicalmemory. Its impairment in amnesic syndromes. Cortex, 25, 197-217.

Della Salla, S., Laiacona, M., Spinnler, H., & Trivelli, C. (1993).Autobiographical recollection and frontal damage. Neuropsycholo-gia, 31, 823-839.

Demonet, J. F., Chollet, F., Ramsay, S., Cardebat, D., Nespoulous, J.,Wise, R., Rascol, A., & Frackowiak, R. (1992). The anatomy ofphonological and semantic processing in normal subjects. Brain, 115,1753-1768.

Demonet, J. F., Wise, R., & Frackowiak, R. S. J. (1993). Languagefunctions explored in normal subjects by positron emission tomogra-phy: A critical review. Human Brain Mapping, I, 39-47.

Eslinger, P. J., & Grattan, L. M. (1994). Altered serial positionlearning after frontal lobe lesions. Neuropsychologia, 32, 729-739.

Fuster, J. M. (1989). The prefrontal cortex: Anatomy, physiology, andneuropsychology of the frontal lobe. New York: Raven Press.

Gershberg, F. B., & Shimamura, A. P. (1995). The role of the frontallobes in the use of organizational strategies in free recall. Neuropsy-chologia. 13, 1305-1333.

Gershberg, F. B. (1995). Implicit and explicit conceptual memoryfollowing frontal lobe damage. Manuscript submitted for publication.

Grasby, P. M., Frith, C. D., Friston, K. J., Bench, C., Frackowiak,R. S. J., & Dolan, R. J. (1993). Functional mapping of brain areasimplicated in auditory-verbal memory function. Brain, 116, 1-20.

FRONTAL LOBES AND IMPAIRED REMOTE MEMORY 41

Hodges, J. R., Salmon, D. P., & Butters, N. (1993). Recognition andnaming of famous faces in Alzheimer's disease: A cognitive analysis.Neuropsychologia, 31, 775-788.

Incisa della Rocchetta, A. (1986). Classification and recall of picturesafter unilateral frontal or temporal lobectomy. Cortex, 22, 189-211.

Incisa della Rocchetta, A., & Milner, B. (1993). Strategic search andretrieval inhibition: The role of the frontal lobes. Neuropsychologia,31, 503-524.

Janowsky, J. S., Shimamura, A. P., Kritchevsky, M., & Squire, L. R.(1989). Cognitive impairment following frontal lobe damage and itsrelevance to human amnesia. BehavioralNeuroscience, 103, 548-560.

Jetter, W., Poser, U., Freeman, R. B. J., & Markowitsch, H. J. (1986).A verbal long-term memory deficit in frontal lobe damaged patients.Cortex, 22, 229-242.

Kapur, S., Craik, F. I. M., Tulving, E., Wilson, A. A., Houle, S., &Brown, G. M. (1994). Neuroanatomical correlates of encoding inepisodic memory: Levels of processing effect. Proceedings of theNational Academy of Science, USA, 91, 2008-2011.

Kapur, S., Rose, R., Liddle, P. F., Zipursky, R. B., Brown, G. M., Stuss,D., Houle, S., & Tulving, E. (1994). The role of the left prefrontalcortex in verbal processing: Semantic processing or willed action?NeuroReport, 5, 2193-2196.

Kopelman, M. D. (1989). Remote and autobiographical memory,temporal context memory and frontal atrophy in Korsakoff andAlzheimer patients. Neuropsychologia, 27, 437-460.

Kopelman, M. D. (1991). Frontal dysfunction and memory deficits inthe alcoholic Korsakoff syndrome and Alzheimer-type dementia.Brain, 114, 117-137.

Mayes, A. R. (1986). Learning and memory disorders and theirassessment. Neuropsychologia, 24, 25-39.

Mayes, A. R., Downes, J. J., McDonald, C., Poole, V., Rooke, S.,Sagar, H. J., & Meudell, P. R. (1994). Two tests for assessing remotepublic knowledge: A tool for assessing retrograde amnesia. Memory,2, 183-210.

Moscovitch, M. (1994). Models of consciousness and memory. In M. S.Gazzaniga (Ed.), The cognitive neurosciences (pp. 1341-1356). Cam-bridge, MA: MIT Press.

Norman, D. A., & Shallice, T. (1986). Attention to action. Willed andautomatic control of behaviour. In R. J. Davison, G. E. Schwardz, &D. Shapiro (Eds.), Consciousness and self-regulation: Advances inresearch and theory (pp. 1-18). New York: Plenum Press.

Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., & Raichle, M. E.(1988). Positron emission tomographic studies of the cortical anatomyof single word processing. Nature, 331, 585-589.

Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., & Raichle, M. E.(1989). Positron emission tomographic studies of the processing ofsingle words. Journal of Cognitive Neuroscience, 1, 153-170.

Posner, M. I., Petersen, S. E., Fox, P. T., & Raichle, M. E. (1988).Localization of cognitive operations in the human brain. Science,240, 1627-1631.

Raichle, M. E. (1994). Images of the mind: Studies with modernimaging techniques. Annual Review of Psychology, 45, 333-356.

Reiser, B. J., Black. J. B., & Kalamarides, P. (1986). Strategic memorysearch processes. In D. C. Rubin (Ed.), Autobiographical memory(pp. 100-121). Cambridge, England: Cambridge University Press.

Sanders, H. I., & Warrington, E. K. (1971). Memory for remote eventsin amnesic patients. Brain, 94, 601-690.

Shallice, T., Fletcher, P., Frith, C. D., Grasby, P., Frackowiak, R. S. J.,& Dolan, R. J. (1994). Brain regions associated with acquisition andretrieval of verbal episodic memory. Nature, 368, 633-635.

Shimamura, A. P. (1994). Frontal lobes and memory. In M. S.Gazzaniga (Ed.), The cognitive neurosciences (pp. 803-813). Cam-bridge, MA: MIT Press.

Shimamura, A. P., Jurica, P. J., Mangels, J. A., Gershberg, F. B., &Knight, R. T. (1995). Susceptibility to memory interference effectsfollowing frontal lobe damage: Findings from tests of paired-associate learning. Journal of Cognitive Neuroscience, 7, 144-152.

Shimamura, A. P., & Squire, L. R. (1986). KorsakofFs syndrome: Astudy of the relation between anterograde amnesia and remotememory impairment. Behavioral Neuroscience, 100, 165-170.

Squire, L. R., & Cohen, N. (1984). Remote memory, retrogradeamnesia and the neuropsychology of memory. In L. S. Cermak(Ed.), Human memory and amnesia (pp. 275-304). Hillsdale, NJ:Erlbaum.

Squire, L. R., Haist, F., & Shimamura, A. P. (1989). The neurology ofmemory: Quantitative assessment of retrograde amnesia in twogroups of amnesic patients. Journal of Neuroscience, 9, 828-839.

Stuss, D. T., Alexander, M. P., Palumbo, C. L., Buckle, L., Sayer, L., &Pogue, J. (1994). Organizational strategies of patients with unilat-eral or bilateral frontal lobe injury in word list learning tasks.Neuropsychology, 8, 355-373.

Stuss, D. T., Eskes, G. A., & Foster, J. K. (1994). Experimentalneuropsychological studies of frontal lobe functions. In F. Boiler &J. Grafman (Eds.), Handbook of neuropsychology (pp. 149-185).Amsterdam: Elsevier.

Tulving, E., Kapur, S., Craik, F. I. M., Moscovitch, M., & Houle, S.(1994). Hemispheric encoding/retrieval asymmetry in episodicmemory: Positron emission tomography findings. Proceedings of theNational Academy of Science, USA, 91, 2016-2020.

Received March 29,1995Revision received June 27,1995

Accepted June 27,1995 •