Original Article

Spontaneous and directed applicationof verbal learning strategies in bipolar disorderand obsessive-compulsive disorder

Neuropsychological studies in bipolar disorderhave documented cognitive disturbances not onlyduring mood episodes but also when patients are

Deckersbach T, Savage CR, Dougherty DD, Bohne A, Loh R,Nierenberg A, Sachs G, Rauch SL. Spontaneous and directedapplication of verbal learning strategies in bipolar disorder andobsessive-compulsive disorder.Bipolar Disord 2005: 7: 166–175. ª Blackwell Munksgaard, 2005

Objectives: Individuals with bipolar disorder exhibitneuropsychological impairments when they are euthymic (neitherdepressed nor manic). One of the most consistently reported cognitiveproblems in euthymic individuals with bipolar disorder is impairment inverbal episodic memory. Recent findings suggest that episodic memorydifficulties in these individuals are attributable to difficulties usingorganizational strategies during encoding. The purpose of the presentstudy was (i) to investigate whether difficulties using organizationalstrategies in bipolar disorder are due to a failure in spontaneouslyinitiating verbal organization strategies or are due to difficultiesimplementing such strategies, and (ii) to compare the characteristics ofverbal organizational impairment in bipolar disorder with thoseobserved in individuals with obsessive–compulsive disorder (OCD).

Methods: Study participants were 20 individuals with bipolar Idisorder (BP-I), 20 individuals with OCD, and 20 healthy controlparticipants matched for age, gender, and education. Participantscompleted a verbal encoding paradigm that involved spontaneous anddirected use of verbal organization strategies during encoding of wordlists.

Results: Compared with control subjects, both BP-I and OCDparticipants showed impaired verbal organization in the spontaneousencoding condition. In the directed encoding condition, OCD patientsorganized the word lists as well as control participants whereas BP-Iparticipants exhibited lower verbal organization than both control andOCD participants. OCD and BP-I participants� free recall performancedid not differ from that of control participants in the spontaneousencoding condition. In the directed encoding condition, BP-Iparticipants recalled fewer words than OCD or control participants.

Conclusions: Episodic memory difficulties in OCD are associated withdifficulties spontaneously initiating verbal organization strategies duringencoding whereas the ability to implement verbal organization wheninstructed to do so is preserved. BP-I participants, on the other hand,exhibit difficulties in both spontaneously initiating verbal organizationstrategies and in the ability to implement such strategies when instructedto do so.

Thilo Deckersbacha, Cary RSavageb, Darin D Doughertya, AntjeBohnec, Rebecca Loha, AndrewNierenberga, Gary Sachsa and ScottL Raucha

aDepartment of Psychiatry, Massachusetts General

Hospital, Harvard Medical School, Charlestown,

MA, bHoglund Brain Imaging Center, Department

of Psychiatry and Behavioral Sciences, Kansas

University Medical Center, Kansas City, KS, USA,cDepartment of Psychology, University of Munster,

Munster, Germany

Key words: bipolar disorder – cognitive

dysfunction – episodic memory – euthymic –

executive functioning – obsessive–compulsive

disorder

Received 20 May 2004, revised and accepted for

publication 16 September 2004

Corresponding author: Thilo Deckersbach, PhD,

Harvard Bipolar Research Program, Psychiatric

Neuroscience Program, Department of Psychiatry,

149-2611, Massachusetts General Hospital, Bldg.

149, 13th St., Charlestown, MA 02129, USA.

Fax: (617) 726 4078;

e-mail: tdeckersbach@partners.org

The authors of this paper do not have any commercial associations

that might pose a conflict of interest in connection with this manu-

script.

Bipolar Disorders 2005: 7: 166–175Copyright ª Blackwell Munksgaard 2005

BIPOLAR DISORDERS

166

remitted and/or euthymic (1–14). One of the mostconsistently reported findings in euthymic individ-uals with bipolar disorder is impairment in verbalepisodic memory – the ability to explicitly recollectinformation encountered in a previous study epi-sode (15). Applying memory strategies duringencoding (the processes of establishing new mem-ory representations) is known to enhance learningand conscious recollection (16). For example,learning a list of words can be facilitated bygrouping items into their semantic categories (17) –an encoding strategy called semantic clustering(18). That is, if words presented for learning stemfrom different categories (e.g. clothing, tools),individuals who group these items into theircategories during presentation of the word listtypically remember more words compared withindividuals who do not.Neuroimaging studies and studies with brain

damaged patients suggest that semantic clusteringprocesses depend on the integrity of prefrontalregions in the brain (17, 19–23) and their intactinteraction with medial temporal lobe structuresinvolved in episodic memory (24–26). Individualswith frontal-lobe brain damage or with disordersaffecting fronto-striatal system function (e.g.Parkinson’sdisease,Huntington’sdisease)havebeenshown to exhibit difficulties using organizationalstrategies during encoding, which in turn wereassociated with subsequent difficulties retrievinginformation (17, 23, 27–33). Previous studies ofobsessive–compulsive disorder (OCD) conductedby our group (32, 33) using the California VerbalLearning Test (CVLT) (18) showed that OCDpatients� difficulties recalling the CVLT 16-itemword list after a long (20-min) delay were mediatedby difficulties using a semantic clustering strategyduring the learning trials. Recently, we also inves-tigated the role of semantic clustering strategies inepisodic memory performance in a cohort ofremitted, euthymic individuals with bipolar Idisorder (BP-I) using the CVLT (34). Similar toOCD patients, long-delayed recall difficulties inremitted bipolar disorder participants were alsomediated by difficulties using a semantic clusteringstrategy during learning (34). This suggests thatepisodic memory problems reported for euthymicpatients with bipolar disorder may at least in partbe secondary to difficulties using organizationalstrategies during encoding.Using a semantic clustering strategy during

encoding is a complex skill that draws upon avariety of cognitive processes and relies on avariety of brain regions and neural circuits.Semantic clustering may involve the necessity ofinhibiting other ongoing cognitive activities in

favor of initiating a semantic clustering strategy.It also involves processing semantic information(categories) as well as updating and reorganizingverbal information in working memory whengrouping the words into their respective categories(20). Among other brain structures, orbitofrontalcortex (OFC) dysfunction has been linked todifficulties inhibiting ongoing cognitive activities(e.g. 35–37). This region appears to play animportant role in the inhibition of automatic orimmediately reinforcing actions so that otheroperations can be mobilized to initiate conscious,goal-directed strategic behavior such as semanticclustering (35–37). In a recent positron emissiontomography (PET) study with healthy volunteers,regional cerebral blood flow in the OFC wasassociated with the extent to which participantsspontaneously used semantic clustering strategiesin a verbal encoding task (20). The dorsolateralprefrontal cortex (DLPFC), on the other hand,appears to be involved in the updating, manipu-lating, and reorganizing of information in work-ing memory (38–42). Functional neuroimagingstudies with healthy volunteers investigating theimplementation of clustering strategies have con-sistently shown increased DLPFC activation asso-ciated with clustering activity during encoding(20–22).To date, it is unclear which processes lead to

impairment in semantic clustering during learningin OCD and bipolar disorder. In both OCD andbipolar disorder, neuroimaging studies (using PET,single-photon emission computed tomography,and functional magnetic resonance imaging) havefound abnormalities in the ventral prefrontalcortex including the OFC (43–50). In bipolardisorder, there is a growing body of postmortemand neuroimaging evidence suggesting DLPFCpathology associated with this disorder (51–55).For example, postmortem studies indicate reduceddensity of neuronal and glial cells (51), reducedneuronal size (52), and reduced density of oligo-dendroglial cells (53) in the DLPFC. Studies usingmagnetic resonance spectroscopy have reproduc-ibly found reduced N-acetyl aspartate to creatine-phosphocreatine ratios in the DLPFC in euthymicindividuals with bipolar disorder (54, 55). Incontrast, whereas the DLPFC is implicated inimaging studies of bipolar disorder, few studieshave suggested DLPFC abnormalities in OCD(45, 56). Rather, the OCD neuroimaging literaturehas focused on the role of OFC. For instance, inOCD, OFC activity is elevated at baseline, furtheraccentuated with symptom provocation, andattenuated toward normal following successfultreatment with serotonergic reuptake inhibitors or

Verbal learning strategies in BD and OCD

167

cognitive behavioral therapy (45, 56–58). Pretreat-ment brain imaging indices of OFC activity inOCD have also been shown to predict subsequenttreatment response to pharmacotherapy andbehavior therapy (57, 59, 60).Based on these findings, we hypothesized that

OCD patients would primarily have difficultiesspontaneously initiating semantic clustering strat-egies in verbal learning paradigms (associated withOFC dysfunction). We hypothesized that BP-Iparticipants would have difficulties not only withspontaneous initiation of semantic clustering strat-egies (due to involvement of ventral prefrontalcortex pathology), but also with implementingsemantic clustering strategies (associated withDLPFC dysfunction) even when explicitly instruct-ed to do so. The purpose of this study was to use averbal encoding paradigm to investigate the extentto which difficulties using semantic clusteringstrategies in OCD and bipolar disorder are asso-ciated with (i) difficulties spontaneously using asemantic clustering strategy and/or (ii) difficultiesimplementing a semantic clustering strategy whendirected to do so.

Materials and methods

Subjects

Study participants were 20 euthymic, remittedparticipants who met DSM-IV (61) criteria for BP-I, 20 participants with DSM-IV OCD, and20 healthy control participants. BP-I participantswere recruited through the Harvard BipolarResearch Program at the Massachusetts GeneralHospital (MGH), and OCD participants wererecruited through the hospital’s OCD Clinic andResearch Unit. Healthy control participants wererecruited through bulletin board notices at MGH.All participants provided written informed consentprior to participation. The Structured ClinicalInterview for DSM-IV (SCID) (62) was used todiagnose BP-I and OCD participants, as well as toconfirm the healthy status of control participants. Inorder to meet DSM-IV remission criteria, BP-Iparticipants were only included in the sample if theydid not meet any of the SCID criteria for a DSM-IVmanic/hypomanic episode or major depressive epi-sode. This was prospectively verified for the 2 weeksprior to testing as well as retrospectively for 6 weekspreceding this prospective 2-week assessment periodvia self-report and chart review.Twelve of the BP-I participants were taking

mood stabilizing medications (lithium: n ¼ 7;anticonvulsants: n ¼ 6, including valproic acidand lamotrigine). In the OCD group, 13 patients

were taking serotonin reuptake inhibitors (includ-ing fluvoxamine, fluoxetine, and sertraline). FiveBP-I participants and eight OCD participants hadcomorbid diagnoses. For BP-I participants thisincluded panic disorder (n ¼ 3) and generalizedanxiety disorder (n ¼ 2). For OCD participantsthis included panic disorder (n ¼ 4), social phobia(n ¼ 2), and generalized anxiety disorder (n ¼ 3).In all cases either BP-I or OCD were the primarydiagnosis. All subjects were right-handed as deter-mined by the Edinburgh Handedness Inventory(63). Subjects with neurologic conditions and thosewith current and/or past drug or alcohol abuse werenot included in the sample. Non-medicated BP-Iparticipants had discontinued mood stabilizingmedication against physician advice (lithium n ¼5, anticonvulsants n ¼ 3). All non-medicated BP-Iparticipants had taken mood stabilizing medica-tions for at least 1 year and were medication-freefor a period of at least 5 weeks prior to testing.Table 1 summarizes the demographic data and

clinical characteristics of the sample. BP-I, OCD,and control participants did not differ with regardto age (F ¼ 0.096; df ¼ 2, 57; p ¼ 0.91), sex (v2 ¼0.13; df ¼ 2; p ¼ 0.94), or years of education (F ¼0.24; df ¼ 2, 57; p ¼ 0.79). Verbal IQ was esti-mated by the Information,Vocabulary, andSimilar-ities subtests from the Wechsler Adult IntelligenceScale Revised (64), and this also did not differbetween groups (F ¼ 1.21; df ¼ 2, 57; p ¼ 0.31).BP-I and OCD participants did not differ in theirage of illness onset (F ¼ 0.77; df ¼ 1, 38; p ¼ 0.39)or duration of illness (F ¼ 0.64; df ¼ 1, 38; p ¼0.43). Both BP-I and OCD participants exhibited

Table 1. Demographic and clinical characteristics of bipolar I disorder(BP-I), obsessive–compulsive disorder (OCD), and control participants

BP-I OCD Controls

GenderFemale 11 11 10Male 9 9 10

Age 32.4 (7.2) 32.9 (8.5) 31.8 (8.0)Education 15.6 (2.3) 15.1 (1.9) 15.3 (2.0)Verbal IQ 108.4 (9.3) 113.9 (11.8) 110.9 (12.5)YBOCS – 22.2 (5.8) –BDI 3.4 (2.0) 3.8 (2.3) 2.8 (2.4)HAM-D 1.8 (1.4) – –YMRS 1.3 (1.1) – –Onset 17.0 (6.4) 18.5 (4.2)Duration 17.5 (6.6) 15.9 (5.5)Depressive episodes 10.9 (4.6)Manic episodes 5.8 (3.5)

Values are presented as mean (SD).YBOCS ¼ Yale Brown Obsessive Compulsive Scale; BDI ¼Beck Depression Inventory; HAM-D ¼ Hamilton DepressionInventory; YMRS ¼ Young Mania Rating Scale.

Deckersbach et al.

168

low depressive symptoms as assessed by theBeck Depression Inventory (65) and HamiltonDepression Inventory (HAM-D) (66). In addition,for BP-I participants, Young Mania Rating Scale(67) scores indicated low residual manic symptoms.The severity of obsessive and compulsive symp-toms for OCD participants was in the moderaterange as assessed by the Yale Brown ObsessiveCompulsive Scale (68). Eight BP-I subjects and fiveOCD subjects in this study completed the encodingparadigm as part of a PET study. Data aboutregional cerebral blood flow of these subjects hasbeen presented elsewhere (69). The remainingsubjects completed the task �offline� in a quietlaboratory room in the Psychiatric NeuroscienceDivision at MGH. In addition to completing theencoding paradigm for the present study, 13 OCDparticipants also completed a dual task version ofthe serial reaction time task as part of a separatestudy (70).

Materials

Subjects completed a word encoding paradigm thatconsists of three word list encoding conditions inwhich subjects listened to prerecorded lists ofwords presented through computer speakers. Sub-jects listened to (i.e. encoded) lists of either 24categorized or uncategorized words presentedthrough computer speakers. In the Spontaneousword encoding condition, subjects were presentedwith 24 words taken from four different categories(e.g. fruits, herbs and spices, metals, furniture).There were six words per category with one wordpresented every 3.3 s. The words from each of fourcategories were presented such that a word from agiven category was never followed by a word fromthe same category. During this Spontaneous con-dition, subjects were not informed about thecategorized nature of the word list. In the Directedcondition, subjects were presented with a differentcategorized word list (24 words, six words percategory, one word presented every 3.3 s). As inthe Spontaneous encoding condition, a word froma given category was never followed by a wordfrom the same category. In the Directed condition,subjects were explicitly instructed to group thewords into their categories during encoding.Finally, in the Unrelated condition, subjects werepresented with an uncategorized word list com-prised of 24 words taken from 24 different categ-ories. Subjects were informed about theuncategorized nature of the list and were instructedto encode the words in any order. This conditiondid not permit grouping the words into semanticcategories.

The paradigm was developed and tested exten-sively in pilot studies in order to develop word liststhat promoted significant differences in semanticclustering between the spontaneous and directedconditions (20). More specifically, the word lists forthis paradigm were constructed by generating32 categories for which normal subjects (duringpilot testing) showed some clustering in the Spon-taneous condition but significantly more in theDirected condition. Words were counterbalancedacross subjects such that each word appeared oncein each condition. Thus, differences among lists didnot reflect differences in list difficulty level.

Procedure

Each encoding condition was presented twice. Thetwo Spontaneous conditions always occurred firstin order to ensure that the strategies employed bysubjects were initiated as �spontaneously� as pos-sible (i.e. to ensure that subjects were not antici-pating the presence of categorized structure in theword lists). The order of the two Directed andUnrelated conditions was counterbalanced acrosssubjects. That is, the two presentations of theDirected condition were presented either before orafter the two presentations of the Unrelatedcondition. In each word encoding condition, sub-jects were asked to recall the words immediatelyafter each list presentation, without interveningdistraction. The experimenter kept a verbatimrecord of each subject’s recall. These data wereused to calculate the semantic clustering score asan estimation of each subject’s clustering duringencoding. In addition, a cued recall test wasprovided after the second presentation of theSpontaneous and Directed conditions followingfree recall. In the cued recall test, subjects wereprovided with the categories as cues and asked torecall the words from that category. Finally,following the cued recall test (or following freerecall in the Unrelated condition), subjects com-pleted a recognition test. In this recognition test,the 24 original items of each word list wereintermixed with 24 distracter items. These itemswere read to the subject, and subjects were asked toindicate whether they thought an item was in theoriginal list.

Behavioral measures

Behavioral measures included (i) a semantic clus-tering score for each trial, (ii) a free recall score foreach trial, (iii) a cued recall score for the Sponta-neous and Directed conditions, and (iv) a recogni-tion discriminability score for each condition. The

Verbal learning strategies in BD and OCD

169

semantic clustering score in each trial of the Spon-taneous and Directed conditions was calculated asfollows: [cluster-points/(words recalled ) numberof categories recalled)] (20). Subjects received a�cluster-point� whenever they recalled two wordsfrom the same category in succession in a given trial.The free recall score was the number of wordscorrectly recalled immediately following each newpresentation of a word list. Similarly, the cued recallscore was the number of correctly recalled wordsduring the cued recall test. The ability to discrimin-ate between the original words and distracters on therecognition test (recognition discriminability) wascalculated using the following formula: [1 ) (falsepositives + false negatives)/48] · 100 (71).

Statistical analyses

Semantic clustering, free recall, cued recall, andrecognition discriminability scores were evaluatedusing mixed model ANOVA with group (controls,BP-I, OCD) as the between-subjects factor. For freerecall, condition (Unrelated versus Spontaneousversus Directed) and trial (first and second) werethe within-subjects factors. For cued recall andsemantic clustering, condition (Spontaneous versusDirected) was the within-subjects factor. For recog-nition discriminability scores, condition (Unrelatedversus Spontaneous versus Directed) was the with-in-subjects factor. Significant interactions werefollowed up by univariate ANOVAs within eachcondition followed by Scheffe tests. Associationsbetween the severity of affective symptoms, OCDsymptoms, duration of illness, age of onset, historyof affective episodes (for BP-I participants), andsemantic clustering and recall scores were investi-gated using Pearson correlations and multipleregression analysis.

Results

Table 2 shows semantic clustering scores and thenumber of words recalled during free recall, cuedrecall, and recognition for BP-I, OCD, and controlparticipants.

Semantic clustering

The mixed model ANOVA indicated main effectsfor condition (F ¼ 38.48; df ¼ 1, 57; p < 0.001)and group (F ¼ 21.44; df ¼ 2, 57; p < 0.001), butno effect for trial (F ¼ 0.48; df ¼ 1, 57; p ¼ 0.83).There was also a significant group-by-conditioninteraction (F ¼ 8.64; df ¼ 2, 57; p ¼ 0.001). Forthe Spontaneous condition, the ANOVA (controls,BP-I, OCD; collapsed over trials 1 and 2 within

each encoding condition) indicated a significantdifference between the groups (F ¼ 7.05; df ¼ 2,57; p ¼ 0.002). Follow-up Scheffe tests revealedthat both BP-I and OCD subjects clustered lessthan control participants (BP-I: F ¼6.68, df ¼ 2,57, p ¼ 0.002; OCD: F ¼ 3.32, df ¼2, 57, p ¼0.04) but did not differ from each other (F ¼ 0.58,df ¼ 2, 57, p ¼ 0.56). In the Directed condition,the ANOVA indicated a significant group differ-ence between controls, BP-I, and OCD subjects(F ¼ 37.80; df ¼ 2, 57; p ¼ 0.001). Follow-upScheffe tests revealed that BP-I subjects (F ¼31.98, df ¼ 2, 57, p < 0.001) but not OCDsubjects (F ¼ 0.35, df ¼ 2, 57, p ¼ 0.70) differedfrom control subjects in semantic clustering (BP-Iversus OCD: F ¼25.60, df ¼ 2, 57, p < 0.001; seeTable 2).

Free recall

The mixed model ANOVA indicated main effectsfor condition (F ¼ 15.99; df ¼ 2, 114; p < 0.001),trial (F ¼ 393.11; df ¼ 1, 57; p < 0.001), andgroup (F ¼ 4.46; df ¼ 2, 57; p ¼ 0.02). Therewas a significant group-by-condition interaction(F ¼ 3.81; df ¼ 4, 114; p ¼ 0.006), but no group-by-trial interaction (F ¼ 0.87; df ¼ 2, 57; p ¼0.43), no trial-by-condition interaction (F ¼ 0.46;df ¼ 2, 114; p ¼ 0.64), and no group-by-trial-by-condition interaction (F ¼ 1.52; df ¼ 4, 114; p ¼0.20). For the Unrelated condition, the ANOVA

Table 2. Summary of the recall, recognition, and semantic clustering scoresin bipolar I disorder (BP-I), obsessive–compulsive disorder (OCD), andcontrol participants

BP-I OCD Controls

Semantic clusteringSpontaneous 0.47 (0.16) 0.54 (0.19) 0.70 (0.25)Directed 0.50 (0.17) 0.83 (0.12) 0.88 (0.16)

Free recallUnrelated condition 13.05 (3.03) 12.80 (4.50) 13.60 (6.01)Spontaneouscondition 13.12 (2.89) 13.10 (4.14) 15.82 (4.37)Directed condition 13.47 (2.58) 17.97 (4.10) 18.12 (2.35)

Cued recallSpontaneouscondition 18.25 (2.63) 19.50 (2.65) 19.90 (2.45)Directed condition 18.15 (2.15) 19.60 (2.47) 19.75 (2.69)

RecognitionUnrelated rec. disc. 0.97 (0.040) 0.97 (0.042) 0.97 (0.024)Spontaneousrec. disc. 0.96 (0.037) 0.97 (0.042) 0.98 (0.023)Directed rec. disc. 0.97 (0.023) 0.98 (0.023) 0.98 (0.015)

Values are presented as mean (SD).See text for ANOVA and F-test results. Rec. disc. ¼ recognitiondiscriminability.

Deckersbach et al.

170

(controls, BP-I, OCD; collapsed over trials 1 and 2within each encoding condition) did not show asignificant difference in free recall across the threegroups (F ¼ 0.15, df ¼ 2, 57, p ¼ 0.86). For theSpontaneous condition, the ANOVA revealedsignificant differences in free recall between thethree groups (F ¼ 3.30, df ¼ 2, 57, p ¼ 0.04). Inthe follow-up Scheffe tests, however, the differencein free recall between BP-I and control participants(F ¼ 2.45, df ¼ 2, 57, p ¼ 0.09), and the differencebetween OCD and control participants (F ¼ 2.49,df ¼ 2, 57, p ¼ 0.09), failed to reach significance.For the Directed condition, the ANOVA indicatedsignificant differences in free recall between thethree groups (F ¼ 9.68, df ¼ 2, 57, p < 0.001).The follow-up Scheffe tests revealed that BP-Iparticipants recalled fewer words than controlparticipants (F ¼ 11.17, df ¼ 2, 57, p < 0.001)and OCD participants (F ¼ 10.46, df ¼ 2, 57,p < 0.001), whereas OCD participants did notdiffer from control participants (F ¼ 0.01, df ¼ 2,57, p ¼ 0.99).

Cued recall

The mixed model ANOVA indicated no main effectfor condition (F ¼ 0.28; df ¼ 1, 57; p ¼0.87), atrend towards a main effect of group (F ¼3.02;df ¼ 2, 57; p ¼ 0.06), but no significant group-by-condition interaction (F ¼ 0.07; df ¼ 2, 57; p ¼0.94). For the Spontaneous condition, the ANOVA(controls, BP-I, OCD) did not show significantdifferences in cued recall between the groups (F ¼2.23, df ¼ 2, 57, p ¼ 0.12). Likewise, there was no

significant group difference in cued recall betweencontrols, BP-I, and OCD participants in the Direct-ed condition (F ¼ 2.59, df ¼ 2, 57, p ¼ 0.08).

Recognition

The mixed model ANOVA indicated no maineffect for condition (F ¼ 1.14; df ¼ 2, 114; p ¼0.32), no main effect of group (F ¼ 1.17; df ¼ 2,57; p ¼ 0.32) and no significant group-by-condi-tion interaction (F ¼ 0.33; df ¼ 4, 114; p ¼ 0.86).

Comorbidity, medication, affective symptoms, and courseof illness

Five patients in the BP-I group and eight partici-pants in the OCD group had comorbid conditionsthat may have contributed to the group differencesin performance. Thus, we excluded data frompatients with comorbid conditions and repeatedthe previous analyses. We also repeated the analy-ses described above by introducing medication(yes ¼ 1, no ¼ 0) as a covariate where groupcomparisons had yielded significant performancedifferences between groups. This allowed us toexplore whether medication mediated the differ-ence between groups. Neither excluding patientswith comorbid conditions nor including medica-tion as a covariate changed the pattern of results.All statistically significant comparisons remainedsignificant; all non-significant comparisonsremained non-significant.The correlations between BP-I and OCD partici-

pants� illness characteristics and their performance

Table 3. Correlations between affective symptoms, onset/course of the disorder and memory variables for bipolar I disorder (BP-I) and obsessive-compulsivedisorder (OCD) participants

Semantic clustering Free recall

Spontaneous Directed Unrelated Spontaneous Directed

BP-IYMRS )0.04 )0.03 0.04 0.17 )0.05HAM-D )0.05 )0.05 0.16 0.09 0.13BDI )0.04 )0.02 0.10 )0.07 )0.05Onset )0.06 )0.05 0.05 )0.01 0.23Duration )0.44* )0.46* )0.47* )0.45* )0.51*Dep. episodes )0.46* )0.50* )0.43 )0.49* )0.45*Man. episodes )0.28 )0.34 )0.41 )0.48* )0.47*

OCDYBOCS 0.13 )0.04 )0.26 )0.28 0.08BDI 0.01 0.17 )0.08 0.07 0.07Onset 0.20 0.08 0.11 0.17 0.06Duration )0.18 )0.12 )0.11 )0.21 )0.14

Free recall ¼ number of words recalled in the Unrelated, Spontaneous, and Directed conditions; YMRS ¼ Young Mania Rating Scale;HAM-D ¼ Hamilton Depression Inventory; BDI ¼ Beck Depression Inventory; Onset ¼ onset of the disorder; Duration ¼ duration of thedisorder; Dep. episodes ¼ number of depressive episodes; Man. episodes ¼ number of manic episodes; YBOCS ¼ Yale BrownObsessive–Compulsive Scale; *p < 0.05.

Verbal learning strategies in BD and OCD

171

scores in the memory paradigm are shown inTable 3.The BP-I individuals with a longer duration of

illness clustered less in both the Spontaneouscondition and the Directed condition (see Table 3).BP-I participants� semantic clustering in both theSpontaneous and Directed conditions was alsonegatively correlated with the number of depressiveepisodes but not with manic episodes (see Table 3).For semantic clustering, the independent effects ofduration of illness and the number of depressiveepisodes for the Directed condition were investi-gated using multiple regression with semanticclustering as the dependent variable and thenumber of depressive episodes and duration ofillness as the independent variables. This indicatedthat depressive episodes (b ¼ )0.44; p ¼ 0.05), butnot duration of illness (b ¼ )0.28; p ¼ 0.18),significantly predicted semantic clustering (R2 ¼0.32; F ¼ 4.08; df ¼ 2, 17; p ¼ 0.04).In BP-I participants, the number of depressive

and manic episodes were negatively correlated withfree recall in both the Spontaneous condition andthe Directed condition (see Table 3). Because thecorrelations between free recall, duration of illness,and number of affective episodes were greatest inthe Directed condition, we investigated the inde-pendent effects of duration of illness and numberof affective episodes on free recall in the Directedcondition using multiple regression analysis. Thenumber of depressive or manic episodes werethe independent variables, and free recall was thedependent variable. This indicated that depressiveepisodes (b ¼ )0.45; p ¼ 0.04), but not durationof illness (b ¼ )0.31; p ¼ 0.15), significantly pre-dicted free recall (R2 ¼ 0.38; F ¼ 5.21; df ¼ 2, 17;p ¼ 0.02). Similarly, there was a trend for manicepisodes (b ¼ )0.42; p ¼ 0.06), but not durationof illness (b ¼ )0.29; p ¼ 0.18), to predict freerecall (R2 ¼ 0.37; F ¼ 4.95; df ¼ 2, 17; p ¼ 0.02).There were no significant correlations for OCDpatients (see Table 3).

Discussion

We compared the characteristics of semantic clus-tering in euthymic, remitted individuals with BP-I,and in individuals with OCD, with those of normalcontrol participants. Consistent with our hypothe-sis, BP-I participants in our study had difficultiesusing semantic clustering strategies compared withcontrol participants in both the Spontaneous andDirected encoding conditions, whereas OCD par-ticipants only clustered lower than control partic-ipants in the Spontaneous condition. OCD andBP-I participants� free recall performance did not

differ from that of control participants in theUnrelated and Spontaneous conditions. However,in the Directed encoding condition, BP-I partici-pants recalled fewer words than OCD and controlparticipants. Difficulties in semantic clustering andimpairment in free recall are consistent withprevious studies in patients with bipolar disorder(4–14, 34). The results of this and previous studiesfrom our group (34) suggest that difficulties insemantic clustering make it difficult to retrieveverbal information (19). However, it should benoted that some previous studies in depressedindividuals have found recognition difficulties inaddition to free recall impairment (72, 73), whichmay suggest additional difficulties with storage/retention associated with mood episodes.It is unlikely that group differences in age,

gender, or education contributed to the differencein findings between BP-I and OCD participants, asour subjects were comparable with control partici-pants on these variables. Likewise, BP-I and OCDparticipants did not differ with respect to their ageof onset or duration of illness. However, it cannotbe fully ruled out that the pattern of results foundin the present study may be attributed to the factthat some OCD and BP-I participants completedthe encoding paradigm as part of a PET study. Wealso investigated the role of depression, comorbid-ity, and medication in the observed performancedifferences between BP-I and OCD participants.BP-I, OCD, and control participants showedcomparable levels of depression at the time ofstudy, which makes it unlikely that this accountsfor the observed group differences in semanticclustering or recall. Some (8, 12) but not all (34)previous studies of verbal learning in euthymicpatients with bipolar disorder found that residualaffective symptoms account for impairments inverbal learning. This discrepancy between thepresent study and previous studies (8, 12) is likelydue to different degrees of residual symptomsacross the samples in different studies (i.e. verylow affective symptoms and small standard devi-ations in the present study). Repetition of the groupcomparisons with only comorbidity-free subjectsdid not change the pattern of results, suggestingthat comorbid conditions also did not account forgroup differences in semantic clustering or recall/recognition. Likewise, including medication as acovariate into the group comparisons did notchange the results. Multiple regression analysisfor BP-I participants in the present study indicatedthat different medication regimens did not changethe pattern of results. However, it should be notedthat these analysis strategies rely on an assumptionof linearity (e.g. medicated patients on average

Deckersbach et al.

172

perform worse than non-medicated patients), andit cannot be ruled out that more complex relation-ships exist between medications (i.e. differentmedication regimens) and performance in thisencoding task. In addition, it cannot be ruled outthat our findings may reflect prolonged exposure tomood-stabilizing medications for the BP-I partici-pants. Thus, replication of our findings is warrant-ed in non-medicated samples and in samples thatare more stringently controlled for medicationstatus and comorbidity.Overall, our results suggest that OCD patients

have difficulties spontaneously initiating semanticclustering strategies in verbal learning tasks but arewell capable of implementing such strategy wheninstructed to do so. BP-I participants, unlike OCDparticipants, continued to exhibited difficultiesusing semantic clustering strategies even whenexplicitly instructed to do so. This could reflectdifficulties mobilizing semantic clustering strategies(despite instructions to initiate them while listeningto the list) or difficulties implementing the strategydespite efforts to do so. When debriefed after thestudy, BP-I subjects reported that they attemptedto use the semantic clustering strategy but found itdifficult to reorder the words while keeping them�in mind� at the same time. This suggests that BP-Iparticipants� difficulties in the Directed encodingcondition reflect difficulties reorganizing and updat-ing information in working memory. Althoughcaution is advised when drawing conclusionsfrom neuropsychological findings with regard toinvolved brain regions and brain circuitry, ourresults are consistent with the hypothesized role ofOFC and DLPFC pathology in OCD and bipolardisorder, respectively, in semantic clustering diffi-culties and memory impairment. Among otherfunctions, the OFC may guide behavior based onanticipated consequences so that executive systemscan engage in effective strategies, particularly innovel and ambiguous situations (74). In contrast,the DLPFC, among other functions, supportsprocesses of monitoring, updating, and reorganiz-ing information in working memory in the serviceof using a semantic clustering strategy (38–42).However, in the absence of concurrent structuraland/or functional neuroimaging data for mostparticipants in the present sample, the capacity tolink neuropsychological findings with specific brainregions/circuits thought to be involved in bipolardisorder or OCD remains necessarily tentative.Finally, consistent with previous studies (9, 12,

13, 34), our study found that BP-I participantswith more lifetime affective episodes had morememory impairment, and those with more depres-sive episodes exhibited lower spontaneous and

directed semantic clustering. When consideredsimultaneously in the context of a multiple regres-sion analysis, only the number of affective episodes(but not the duration of bipolar disorder itself)predicted memory performance and semantic clus-tering. Overall, this suggests that that at least somecognitive difficulties observed in euthymic patientswith bipolar disorder may result from the cumu-lative effects of repeated affective disturbances (75).Memory impairments in individuals with bipolardisorder may be mood state dependent at ayounger age; over time they may become trait-related as repeated mood episodes may affectprefrontal cortex and/or medial temporal lobefunctioning. For example, there is evidence thatpatients experiencing their first major depressiveepisode, unlike recurrently depressed patients withunipolar depression or bipolar disorder, do not yetshow memory difficulties (76).From a clinical perspective, it is important to

recognize that episodic memory impairments mayhave an impact on daily functioning (e.g. managinghousehold duties, functioning at work) (7). Mem-ory impairments may also require appropriate andflexible adjustments of psychosocial treatments forbipolar disorder (e.g. cognitive-behavior therapy,family therapy, interpersonal therapy), which sub-stantially rely on the patient’s ability to learn andto remember. Taking these memory impairmentsinto account, patients with bipolar disorder (orOCD) may benefit from cognitive remediationinterventions that provide patients with specifictraining in, and practice with, the application ofcognitive strategies to improve functioning.

Acknowledgements

This study was supported by a Young Investigator Awardgranted to Dr Thilo Deckersbach by the National Alliance forResearch in Schizophrenia and Depression (NARSAD), andthrough funding from the Clinical Research Training Program(CRTP) at Harvard Medical School.

References

1. Martinez-Aran A, Vieta E, Colom F et al. Cognitivedysfunctions in bipolar disorder: evidence of neuropsycho-logical disturbances. Psychother Psychosom 2000; 69: 2–18.

2. Bearden CE, Hoffman KM, Cannon TD. The neuropsy-chology and neuroanatomy of bipolar affective disorder: acritical review. Bipolar Disord 2001; 3: 106–150.

3. Murphy FC, Sahakian BJ. Neuropsychology of bipolardisorder. Br J Psychiatry 2001; 178: 120–127.

4. Dupont RM, Jernigan TL, Butters N et al. Subcorticalabnormalities detected in bipolar affective disorder usingmagnetic resonance imaging. Arch Gen Psychiatry 1990;47: 55–59.

5. McKay AP, Tarbuck AF, Shapleske J, McKenna PJ.Neuropsychological function in manic-depressive psycho-

Verbal learning strategies in BD and OCD

173

sis. Evidence for persistent deficits in patients with chronic,severe illness. Br J Psychiatry 1995; 167: 51–57.

6. Tham A, Engelbrektson K, Mathe AA, Johson L, OlssonM, Aberg-Wistedt A. Impaired neuropsychological per-formance in euthymic patients with recurring mooddisorders. J Clin Psychiatry 1997; 58: 26–29.

7. Atre-Vaidya N, Taylor MA, Seidenberg M, Reed R,Perrine A, Glick-Oberwise F. Cognitive deficits, psycho-pathology, and psychosocial functioning in bipolar mooddisorder. Neuropsychiatry Neuropsychol Behav Neurol1998; 11: 120–126.

8. Ferrier IN, Stanton BR, Kelly TP, Scott J. Neuropsycho-logical function in euthymic patients with bipolar disorder.Br J Psychiatry 1999; 175: 246–251.

9. van Gorp WG, Althuler L, Theberge DC, Wilkins J, DixonW. Cognitive impairment in euthymic bipolar patients withand without prior alcohol dependence. Arch Gen Psychi-atry 1998; 55: 41–46.

10. van Gorp WG, Altshuler L, Theberge DC, Mintz J.Declarative and procedural memory in bipolar disorder.Biol Psychiatry 1999; 46: 525–531.

11. Krabbendam L, Honig A, Wiersma J et al. Cognitivedysfunctions and white matter lesions in patients withbipolar disorder. Acta Psychiatr Scand 2000; 101: 274–280.

12. Clark L, Iversen SD, Goodwin GM. Sustained attentiondeficit in bipolar disorder. Br J Psychiatry 2002; 180: 313–319.

13. Cavanagh JTO, Van Beck M, Muir W, Blackwood DHR.Case-control study of neurocognitive function in euthymicpatients with bipolar disorder: an association with mania.Br J Psychiatry 2002; 180: 320–326.

14. Martinez-Aran A, Vieta E, Reinares M et al. Cognitivefunction across manic or hypomanic, depressed, andeuthymic states in bipolar disorder. Am J Psychiatry2004; 161: 262–270.

15. Graf P, Schacter DL. Implicit and explicit memory for newassociations in normal and amnestic subjects. J ExpPsychol Learn 1985; 13: 45–53.

16. Kapur S, Tulving E, Cabeza R, McIntosh AR, Houle S,Craik FIM. The neural correlates of intentional learning ofverbal materials: a PET study in humans. Cognitive BrainRes 1996; 4: 243–249.

17. Gershberg FB, Shimamura AP. Impaired use of organiza-tional strategies in free recall following frontal lobedamage. Neuropsychologia 1995; 13: 1305–1333.

18. Delis DC, Kramer JH, Kaplan E, Ober BA. CaliforniaVerbal Learning Test Research Edition Manual AdultVersion. San Antonio, TX: The Psychological Corpora-tion, 1987.

19. Savage CR. Neuropsychology of OCD: research findingsand treatment implications. In: Jenike MA, Baer L,Minichello WE eds. Obsessive-Compulsive Disorders:Practical Management, 3rd edn. St Louis, MO: Mosby,1998: 254–275.

20. Savage CR, Deckersbach T, Heckers S et al. Prefrontalregions supporting spontaneous and directed applicationof verbal learning strategies. Evidence from PET. Brain2001; 124: 219–231.

21. Wagner AD, Maril A, Bjork RA, Schacter DL. Prefrontalcontributions to executive control: fMRI evidence forfunctional distinctions within lateral prefrontal cortex.Neuroimage 2001; 14: 1337–1347.

22. Fletcher PC, Shallice T, Dolan RJ. The functional roles ofprefrontal cortex in episodic memory. I. Encoding. Brain1998; 121: 1239–1248.

23. Baldo JV, Delis D, Kramer J, Shimamura AP. Memoryperformance on the California Verbal Learning Test – II:

findings from patients with focal frontal lesions. J IntNeuropsychol Soc 2002; 8: 539–546.

24. Squire LR. Memory and the hippocampus: a synthesisfrom findings with rats, monkeys and humans. PsycholReview 1992; 99: 195–231.

25. Hildebrandt H, Brand A, Sachsenheimer W. Profiles ofpatients with left prefrontal and left temporal lobe lesionsafter cerebrovascular infarctions on California VerbalLearning Test-like indices. J Clin Exp Neuropsychol1998; 20: 673–683.

26. Hermann BP, Wyler AR, Bush AJ, Tabatabai FR.Differential effects of left and right anterior temporallobectomy on verbal learning and memory performance.Epilepsia 1992; 33: 289–297.

27. Shimamura AP, Janowsky JS, Squire LR. What is the roleof frontal lobe damage in memory disorders? In: Levin HS,Eisenberg HM, Benton AL eds. Frontal Lobe Functionand Dysfunction. New York: Oxford University Press,1991: 173–195.

28. Bondi MW, Kaszniak AW, Bayles KA, Vance KT.Contributions of frontal system dysfunction to memoryand perceptual abilities in Parkinson’s disease. Neuropsy-chology 1993; 7: 89–102.

29. Pillon B, Deweer B, Agid Y, Dubois B. Explicit memory inAlzheimer’s, Huntington’s and Parkinson’s diseases. ArchNeurol 1993; 50: 374–379.

30. Buytenhuijs EL, Berger HJC, Van Spaendonck KPM,Horstink MWIM, Borm GE, Cools AR. Memory andlearning strategies in patients with Parkinson’s disease.Neuropsychologia 1994; 32: 335–342.

31. Grossman M, Carvell S, Peltzer L, Stern MB, Gollomp S,Hurtig HI. Visual construction impairment in Parkinson’sdisease. Neuropsychology 1993; 7: 536–547.

32. Savage CR, Deckersbach T, Wilhelm S et al. Strategicprocessing and episodic memory impairment in obsessive-compulsive disorder. Neuropsychology 2000; 14: 141–151.

33. Deckersbach T, Otto MW, Savage CR, Baer L, JenikeMA. The relationship between semantic organization andmemory in obsessive-compulsive disorder. PsychotherPsychosom 2000; 69: 101–107.

34. Deckersbach T, Savage CR, Reilly-Harrington N, Clark L,Sachs G, Rauch SL. Episodic memory impairment inbipolar disorder and obsessive-compulsive disorder: therole of memory strategies. Bipolar Disord 2004; 6: 233–244.

35. Nobre AC, Coull JT, Frith CO, Mesulam MM. Orbito-frontal cortex is activated by breaches of expectation intasks of visual attention. Nat Neurosci 1999; 2: 11–12.

36. Schoenbaum G, Chiba AA, Gallagher M. Orbitofrontalcortex and basolateral amygdala encode expected out-comes during learning. Nat Neurosci 1998; 1: 155–159.

37. Lipton PA, Alvarez P, Eichenbaum H. Crossmodal asso-ciative memory representations in rodent orbitofrontalcortex. Neuron 1999; 22: 349–359.

38. Braver TS, Cohen JD, Nystron LE, Jonides J, Smith EE,Noll DC. A parametric study of prefrontal cortex involve-ment in human working memory. Neuroimage 1997; 5:49–62.

39. Collette F, Salmon E, Van der Linden M et al. Regionalbrain activity during taks devoted to the central executiveof working memory. Cog Brain Res 1999; 7: 411–417.

40. D’Esposito M, Aguirre GK, Zarahn E, Ballard D, ShinRK, Lease J. Functional MRI studies of spatial andnonspatial working memory. Cog Brain Res 1998; 7: 1–13.

41. Owen AM, Evans AC, Petrides M. Evidence for a two-stage model of spatial working memory processing within

Deckersbach et al.

174

the lateral frontal cotex: a positron emission tomographystudy. Cereb Cortex 1996; 8: 353–364.

42. Petrides M. Functional organization of the human frontalcortex for mnemonic processing. Evidence from neuro-imaging studies. Ann N Y Acad Sci 1995; 769: 85–96.

43. Baxter LR, Phelps ME, Maziotta JC, Guze BH, SchwartzJM, Selin CE. Local cerebral glucose metabolic rates inobsessive-compulsive disorder: a comparison with rates inunipolar depression and in normal controls. Arch GenPsychiatry 1987; 44: 211–218.

44. Baxter LR, Schwartz JM, Mazziotta JC et al. Cerebralglucose metabolic rates in non-depressed patients withobsessive-compulsive disorder. Am J Psychiatry 1988; 145:1560–1563.

45. Benkelfat C, Nordahl TE, Semple WE, King C, MurphyDL, Cohen RM. Local cerebral glucose metabolic rates inobsessive-compulsive disorder. Arch Gen Psychiatry 1990;47: 840–848.

46. Nordahl TE, Benkelfat C, Semple W, Gross M, King AC,Cohen RM. Cerebral glucose metabolic rates in obsessive-compulsive disorder. Neuropsychopharmacology 1989; 2:23–28.

47. Rubin RT, Villaneuva-Meyer J, Ananth J, Trajmar PG,Mena I. Regional xenon-133 cerebral blood flow andcerebral Technetium 99m HMPAO uptake in unmedicatedpatients with obsessive-compulsive disorder and matchednormal control subjects. Determination by high-resolutionsingle-photon emission computed tomography. Arch GenPsychiatry 1992; 49: 695–702.

48. Blumberg HP, Stern E, Ricketts S et al. Rostral and orbitalprefrontal cortex dysfunction in the manic state of bipolardisorder. Am J Psychiatry 1999; 156: 1986–1988.

49. Blumberg HP, Leung HC, Skudlarski P et al. A functionalmagnetic resonance imaging study of bipolar disorder:state- and trait-related dysfunction in ventral prefrontalcortices. Arch Gen Psychiatry 2003; 60: 601–609.

50. Rubinsztein JS, Fletcher PC, Rogers RD et al. Decision-making in mania: a PET study. Brain 2001; 124: 2550–2563

51. Rajkowska G, Halaris A, Selemon LD. Reductions inneuronal and glial density characterize prefrontal cortex inbipolar disorder. Biol Psychiatry 2001; 49: 741–752.

52. Cotter D, Mackay D, Chana G, Beasley C, Landau S,Everall IP. Reduced neuronal size in area 9 of thedorsolateral prefrontal cortex in subjects with majordepressive disorder. Cereb Cortex 2002; 12: 386–394.

53. Orlovskaya DD, Vostrikov VM, Rachmanova NA,Uranova NA. Decreased numerical density of oligoden-droglial cells in postmortem prefrontal cortex in schizo-phrenia and major depression. Schizophr Res 2000; 41:105.

54. Winsberg ME, Sachs N, Tate D, Adalsteinsson E.,Spielman D, Ketter TA. Decreased dorsolateral prefrontalN-acetyl aspartate in bipolar disorder. Biol Psychiatry2000; 47: 475–481.

55. Chang K, Adleman N, Dienes K, Barnea-Goraly N, ReissA, Ketter T. Decreased N-acetyl aspartate in children withfamilial bipolar disorder. Biol Psychiatry 2000; 53: 1059–1065.

56. Swedo SE, Pietrini P, Leonard HL et al. Cerebral glucosemetabolism in childhood-onset obsessive-compulsive dis-order: revisualization during pharmacotherapy. Arch GenPsychiatry 1992; 49: 690–694.

57. Baxter LR, Schwartz JM, Bergman KS et al. Caudateglucose metabolic rate changes with both drug and

behavior therapy for obsessive-compulsive disorder. ArchGen Psychiatry 1992; 49: 681–689.

58. Hoehn-Saric R, Perlson GD, Harris GS, Machlin SR,Camargo EE. Effects of fluoxetine on regional cerebralblood flow in obsessive-compulsive patients. Am J Psychi-atry 1991; 148: 1243–1245.

59. Schwartz JM, Stoessel PW, Baxter LR, Martin KM,Phelps ME. Systematic changes in cerebral glucose meta-bolic rate after successful behavior modification. Arch GenPsychiatry 1996; 53: 109–113.

60. Brody AL, Saxena S, Schwartz JM et al. FDG-PETpredictors of response to behavioral therapy and phar-macotherapy in obsessive-compulsive disorder. PsychiatryRes 1998; 84: 1–6.

61. American Psychiatric Association. Diagnostic and Statis-tical Manual of Mental Disorders, 4th edn. Washington,DC: American Psychiatric Association Press, 1994.

62. First MB, Spitzer RL, Gibbon M, Williams JBW. Struc-tured Clinical Interview for DSM-IV Axis I Disorders,Patient Edition (SCID-I/D, Version 2.0). New York:Biometrics Research Department, New York PsychiatricInstitute, 1995.

63. Oldfield RC. The assessment and analysis of handedness:The Edinburgh Inventory. Neuropsychologia 1971; 9:97–113.

64. Wechsler D. Wechsler Adult Intelligence Scale Revised.Cleveland, OH: The Psychological Corporation, 1981.

65. Beck AT, Steer RA. Beck Depression Inventory Manual.San Antonio, TX: The Psychological Corporation, 1987.

66. Hamilton M. A rating scale for depression. J NeurolNeurosurg Psychiatry 1960; 23: 56–62.

67. Young RC, Biggs JT, Ziegler VE, Meyer DA. A ratingscale for mania: reliability, validity and sensitivity. Br JPsychiatry 1978; 133: 429–435.

68. Goodman WK, Price LH, Rasmussen SA et al. The YaleBrown Obsessive Compulsive Scale (Y-BOCS), part I:development, use, and reliability. Arch Gen Psychiatry1989; 46: 1006–1016.

69. Deckersbach T, Dougherty DD, Savage CR et al. Im-paired recruitment of the dorsolateral prefrontal cortexand hippocampus during encoding in bipolar disorder.Biological Psychiatry 2004; 55: 815.

70. Deckersbach T, Savage CR, Curran T et al. A study ofparallel implicit and explicit information processing inpatients with obsessive-compulsive disorder. Am J Psychi-atry 2002; 159: 1780–1782.

71. Underwood BJ. The role of the association in recognitionmemory. J Exp Psychol Monogr 1974; 102: 917–939.

72. Golinkoff M, Sweeney JA. Cognitive impairments indepression. J Affect Disord 1989; 17: 105–112.

73. Wolfe J, Granholm E, Butters N, Saunders E, JanowskyD. Verbal memory deficits associated with major affectivedisorders: a comparison of unipolar and bipolar patients.J Affect Disord 1987; 13: 83–92.

74. Damasio AR. The somatic marker hypothesis and thepossible functions of prefrontal cortex. Philos Trans R SocLond 1996; 351: 1413–1420.

75. Altshuler LL. Bipolar disorder: are repeated episodesassociated with neuroanatomic and cognitive changes?Biol Psychiatry 1993; 33: 563–565.

76. Fossati P, Harvey PO, Le Bastard G, Ergis AM, JouventR, Allilaire JF. Verbal memory performance of patientswith a first depressive episode and patients with unipolarand bipolar recurrent depression. J Psychiatr Res 2004; 38:137–144.

Verbal learning strategies in BD and OCD

175