Neuropsychologia 44 (2006) 2558–2563

Note

Differential activation of ventrolateral prefrontal cortex duringworking memory retrieval

Robert Christian Wolf a,∗, Nenad Vasic a, Henrik Walter b

a Department of Psychiatry III, University of Ulm, Leimgrubenweg 12-14, 89075 Ulm, Germanyb Department of Psychiatry, Johann Wolfgang Goethe University, Frankfurt, Germany

Received 30 November 2005; received in revised form 19 April 2006; accepted 3 May 2006

Abstract

Brain imaging studies have suggested a predominant involvement of prefrontal areas during retrieval of information from working memory(WM). This study used event-related functional magnetic resonance imaging to assess the gradual recruitment of brain areas during verbal WM-retrieval with a parametrically varied modified version of the Sternberg Item Recognition Paradigm. In particular, we were interested in activationdifferences during retrieval of negative and positive probes. Fifteen subjects performed a WM-task which required the retrieval of a probe letterfrPlnd©

K

1

iofBif2AWLsoRb

0d

rom a set of a maximum of three letters. The analysis of the retrieval period regardless of probe type revealed bilateral VLPFC activation duringetrieval from a single remembered item. These initially activated regions showed a gradual activation increase of left VLPFC (BA 47) and anteriorFC (BA 10) as well as and bilateral DLPFC (BA 9) with increasing retrieval demand, i.e. during retrieval of two and three previously remembered

etters. The comparison of negative and positive probes (non-targets versus targets) revealed greater activity in VLPFC (BA 47) in response toegative than to positive probes. These findings demonstrate that ventral areas of prefrontal cortex seem to be differentially engaged during theiscrimination of a non-target from a previously manipulated set.

2006 Elsevier Ltd. All rights reserved.

eywords: Working memory; Prefrontal cortex; Retrieval; BA47; fMRI; Target; Non-target

. Introduction

The concept of working memory (WM) refers to the abil-ty of transient storage and manipulation of information heldn-line for further usage in related cognitive processes oror goal-directed behavioral guidance (Baddeley, 1996, 2003).oth anatomical tracing and functional neuroimaging stud-

es demonstrated that the prefrontal cortex (PFC) is crucialor mediating WM-functions (D’Esposito, Postle, & Rypma,000; Funahashi, Bruce, & Goldman-Rakic, 1989; Fuster &lexander, 1971; Goldman-Rakic, 1990; Walter et al., 2003;olf & Walter, 2005); for a meta-analysis see (Owen, McMillan,

aird, & Bullmore, 2005; Wager & Smith, 2003). Previoustudies have tested several WM-models, including prefrontalrganization by material type (Funahashi, Bruce, & Goldman-akic, 1993; Wilson, Scalaidhe, & Goldman-Rakic, 1993) ory the type of cognitive process (Petrides, 1994). However,

∗ Corresponding author. Tel.: +49 731 50021499; fax: +49 731 50021549.E-mail address: christian.wolf@uni-ulm.de (R.C. Wolf).

studies using functional magnetic resonance imaging (fMRI)have challenged these concepts. Thus, lateral PFC is clearlyrecruited during a variety of different cognitive processes that areengaged during the performance of a whole range of WM-tasks,including encoding, maintenance, manipulation and updatingof information (Wager & Smith, 2003). While most studieshave analyzed PFC activation during the delay period, thereis less evidence concerning PFC function during WM-retrieval(D’Esposito et al., 2000; Habeck et al., 2005; Jonides, Smith,Marshuetz, Koeppe, & Reuter-Lorenz, 1998; Leung, Gore, &Goldman-Rakic, 2005). Baddeley (1986) considered the rela-tionship between WM and retrieval from long-term memory byacknowledging the differential role of phonological processesduring successful retrieval. The neuronal correlates of these pro-cesses may involve predominantly ventral PFC-regions (at ornear BAs 44/6, 45 and 47) in elaborating on verbal informa-tion, including remembering and retrieval (Buckner & Wheeler,2001).

Event-related fMRI studies of WM-retrieval have implieddifferential contributions of both ventro- (VLPFC) and dorso-lateral prefrontal cortex (DLPFC). For instance, DLPFC has

028-3932/$ – see front matter © 2006 Elsevier Ltd. All rights reserved.oi:10.1016/j.neuropsychologia.2006.05.015

R.C. Wolf et al. / Neuropsychologia 44 (2006) 2558–2563 2559

been associated with scanning of information, VLPFC withinhibitory functions and selection (D’Esposito et al., 2000). Atpresent, parametric variations of the amount of information dur-ing WM-retrieval have been less extensively studied (but seeHabeck et al., 2005). Sufficiently parametrized designs can yieldadditional information concerning differential PFC-activationwith increasing target discrimination load. Recently, Leung etal. (2005) have demonstrated differential activation of anteriorPFC (BA 10) during the recognition stage of a spatial WM-task, indicating that this region may show greater activity inresponse to negative probes (non-targets) than positive probes(targets). These results support the concept of differential pre-frontal engagement during decision process in WM-retrieval,possibly depending on target features.

We have recently shown prefrontostriatal recruitment withincreasing load during the delay period of a verbal WM-task(Wolf & Walter, 2005). In this report, we chose to reanalyze thisdata, addressing the issue of parametric recruitment of lateralprefrontal areas with increasing retrieval demand. Furthermore,we investigated differential activation effects during paramet-ric retrieval of targets versus non-targets. Previous event-relatedfMRI studies of WM-retrieval have shown contributions of bothventral and dorsal PFC during WM-retrieval (D’Esposito etal., 2000), VLPFC-activation being consistently found at lowWM-load levels (Lee, Robbins, & Owen, 2000). Thus, wehypothesized VLPFC-activation during low-level retrieval andiMiidTtmf

2

2

arasdC

2

(loSwewp(

Fig. 1. Activation paradigm, shown for Load2. In this example, the letters L andF turned bright, and subjects had to subsequently memorize the letters M and G(“manipulated set”). The probe-letter g is a part of the previously manipulatedset, i.e. a positive probe/target. Red rectangle: period-of-interest for fMRI. Seealso text for further details.

accepted or rejected from the “manipulated set”. Probe-selection included anequal number of positive and negative probes which had to be retrieved from aset of one to three letters (‘Load’ 1–3). The control condition consisted of threegrey X’s calling for a stereotype button press during the presentation of a small xin the probe period, thus forming a motor task without mnemonic requirements.

2.3. Data acquisition

Data were acquired using a 1.5 T Magnetom VISION (Siemens, Erlangen,Germany) whole-body MRI system equipped with a standard head volume coil.T2*-weighted images were obtained using echo-planar imaging in an axial ori-entation (TR = 2400 ms, TE = 40 ms, FoV 192 mm, 64 × 64 matrix, 24 slices,slice thickness 4 mm, gap 2 mm). Stimuli were presented via LCD video gog-gles (Resonance Technologies, Northridge, CA) and both reaction times andaccuracy indices were recorded. Head movement was minimized using paddedear phones. The fMRI-protocol was a rapid event-related design with a pseu-dorandomized time-jitter of 1.5 ± 0.5 TR inter-trial-interval. Trial duration was10 s + 2.4–4.8 s. Stimuli were pseudorandomized and counterbalanced for therelative appearance frequency of each letter per load, highlighted position ortarget. Each condition was associated with a balanced number of target and non-target probe-letters. The experimental design avoided the appearance of recentnegative trials in order to prevent proactive interference during retrieval (Jonideset al., 1998). Subjects performed three sessions, each including 28 trials (7 tri-als per condition), comprising 164 volumes (492 volumes in total). The first 8volumes of each session were discarded to allow for equilibration effects.

2.4. Data analysis

2.4.1. Behavioral data analysisPerformance was recorded as percentage of correct responses (accuracy)

drtl(a

2

oNcdsAiatoWdt

ncreasing DLPFC-activation with increasing retrieval demand.oreover, we hypothesized that VLPFC-function would show

ncreasing retrieval-related activation, since recent evidencendicates that VLPFC function may be modulated in a load-ependent manner during the probe period (Habeck et al., 2005).he second aim of the study was to show that VLPFC activa-

ion is not solely modulated by ‘retrieval load’ in general butay depend on selection and discrimination processes that dif-

er between probe features.

. Materials and methods

.1. Subjects

We studied fifteen right handed healthy subjects (8 m, 7 f; meange = 28.13 ± 4.17 years; educational level = 12.0 ± 1.5 education years),ecruited from the University of Ulm campus, after having been screened tossure only the inclusion of participants without a history of major head trauma,ignificant neurological or psychiatric disorder or substance abuse or depen-ence within the past 6 months. The project was approved by the local Ethicsommittee and all subjects gave written informed consent prior to participation.

.2. Cognitive task procedure

The cognitive activation task has been described elsewhere in full detailWolf & Walter, 2005). During a stimulus period of 1500 ms, three capital greyetters were presented on a black screen. For a brief period of 500 ms, eitherne, two or all letters could turn bright depending on randomization (Fig. 1).tarting from these letters, subjects were instructed to memorize only the letter(s)hich directly followed in the alphabet (“manipulated set”); see Fig. 1 for an

xample of a single trial. In the probe period of 2000 ms a lower-case letteras presented and subjects had to indicate whether this letter was or was notart of the “manipulated set”. Probes consisted of both targets and non-targetspositive or negative probes, respectively), i.e. of probes which either had to be

uring target and non-target trials, reaction times (RT) of the correct trials areeported in ms. Additionally, accuracy and RT were calculated for target and non-arget conditions. Significant changes in performance and RT with increasingoad were assessed separately using a repeated-measures analysis of varianceMANOVA) followed by Scheffe’s Test post hoc. Differences between targetnd non-target conditions were assessed using paired t-tests.

.4.2. fMRI data analysisFunctional data analyses were carried out with SPM2 (Wellcome Department

f Cognitive Neurology, London) implemented in MATLAB 6.0 (MathWorks,atick, MA). The functional images were first subject of slice-timing and

orrection of motion artifacts, then spatially normalized to the SPM2 EPI stan-ard template of 3 mm × 3 mm × 3 mm voxels. These images were spatiallymoothed with a 9-mm full width at half maximum isotropic Gaussian kernel.nalyses were performed within the framework of the General Linear Model

n SPM2 (Friston, Holmes, Worsley et al., 1995; Friston, Holmes, Poline etl., 1995) using the canonical-hrf-function as a predictor in order to estimatehe hemodynamic response function of each event. For single-subject analyses,nly correct trials were included after removal of incorrect and omitted probes.e modelled stimulus and delay periods as one regressor, thus obtaining a lower

egree of event-correlation relatively to the retrieval period. Omitted and falserials were pooled and used as individual regressor of no interest for each subject.

2560 R.C. Wolf et al. / Neuropsychologia 44 (2006) 2558–2563

To test for differential activation during retrieval, these intervals were modelledas one event ocurring at the beginning of the retrieval period, lasting 2000 ms.Images were entered into a fixed effects model of one single subject (Friston,Holmes, Worsley et al., 1995; Friston, Holmes, Poline et al., 1995) and adjustedfor global effects. Low-frequency drifts were removed via a highpass filter usinglow-frequency cosine functions with a cutoff of 137 s, high-frequency drifts wereremoved via a Gaussian lowpass filter of 4 s. For each subject, regionally spe-cific effects of conditions were compared using linear contrasts, resulting in at-statistic for each voxel.

On the first-level, main effects of load were calculated for the retrieval periodof each condition. To account for interindividual variance and in order to gener-alize inferences (Holmes & Friston, 1998), we conducted a between-conditionsanalysis of variance (ANOVA) on the second level. Note that we do not reportabsolute load effects, i.e. comparisons of increasing WM-load against the controlcondition, but differential effects with increasing load, i.e. the contrasts Load1(L1) > control, Load2 (L2) > L1 and Load3 (L3) > L2.

To assess within-condition (i.e. target versus non-target) effects, we per-formed paired t-test analyses on the second-level by using first level contrastsobtained for L1, L2 and L3 minus the non-mnemonic control condition, thuscontrolling for baseline differences. Because we were primarily interested incharacterizing ventro- and dorsolateral prefrontal function and anterior PFCregions (namely BA 10), we restricted our discussion to BAs 9, 10, 44, 45,46 and 47. These hypothesis-driven regions-of-interest (ROI) were identifiedafter the application of a mask comprising these prefrontal areas (Maldjian,Laurienti, Kraft, & Burdette, 2003). Functional imaging results are primarilyreported as Z-scores with a significance threshold of P < 0.005 (uncorrected)and a minimum cluster size of 25 contiguous voxels. Additionally, small vol-ume correction (s.v.c.) was applied in order to further protect against type I error.S.v.c. was performed by centering a sphere of 9 mm volume-of-interest (VOI)radius on all PFC regions described in Section 3. These complementary resultsare reported at a more conservative threshold of P < 0.001, P < 0.05 corrected att

aM

3. Results

3.1. Behavioral results

We observed an association of a decline in task-accuracy withincreasing working memory load (control condition: 99.7 ± 1.3,L1: 95.5 ± 9.4, L2: 92.6 ± 8.8, L3: 93.0 ± 8.9). While accu-racy tended to decrease with increasing load (F[3,42] = 4.2779,P = 0.01), this decline reached significance for L2 and L3 only(P = 0.05). Along with decreasing accuracy, we found increas-ing RT with increasing load (control condition: 627.7 ± 180.0,L1: 770.8 ± 116.1, L2: 882.0 ± 95.1, L3: 1034.5 ± 157.5);(F[3,42] = 42.5, P = 0.0000), being statistically significant foreach condition post hoc (P = 0.05). Subjects made signifi-cantly more errors during the retrieval of targets (mean percent7 ± 7.6) compared to non-targets (mean percent 4 ± 6.3) in L3only (P = 0.05). During non-target retrieval, RT significantlyincreased in L2 and L3 (P = 0.05) only.

3.2. Functional imaging results

3.2.1. Between-condition analysesComparing Load1 versus Control (L1 > Control) we found

activation in bilateral right inferior frontal gyrus (left BA 47:x = −36, y = 18, z = −6, Z = 3.56; right BA 47: x = 27, y = 18,zBrv

F tivatioAat

he cluster-level.Coordinates were transformed to the standard anatomical space of Talairach

nd Tournoux (1988) and given as cluster-maxima according to the standardNI-template implemented in SPM2.

ig. 2. (Left) Brain regions demonstrating significantly greater differential ac

NOVA (P < 0.005, uncorrected). (Right) Brain regions demonstrating significantly

nd Load3 only). Results of the second-level paired-t-test analyses (P < 0.005, uncoemplate implemented in SPM2.

= −9, Z = 4.10; left BA 44: x = −42, y = 3, z = 36, Z = 2.92; rightA 44: x = 45, y = 9, z = 27, Z = 3.15). These initially activated

egions were subject of further increasing activation of dorso-,entrolateral and anterior prefrontal areas: the contrast L2 > L1

n during the retrieval period between conditions. Results of the second-level

greater activation during retrieval of non-targets compared with targets (Load2rrected). Shown are colour-coded activation-maps rendered onto the standard

R.C. Wolf et al. / Neuropsychologia 44 (2006) 2558–2563 2561

revealed an additional involvement of right inferior frontal gyrus(BA 47: x = 42, y = 15, z = −3, Z = 3.06), left medial frontal gyrus(BA 9: x = −18, y = 33, z = 27, Z = 3.23) and bilateral middlefrontal gyrus (left BA 9: x = −42, y = 18, z = 27, Z = 3.73; rightBA 9: x = 42, y = 30, z = 33, Z = 3.56; left BA 10: x = −39, y = 54,z = 21, Z = 4.06). The differential contrast L3 > L2 yielded acti-vation of left inferior (BA 47: x = −45, y = 18, z = 3, Z = 3.21) andright middle frontal gyrus (BA 9: x = 45, y = 45, z = 27, Z = 4.00);see also Fig. 2). All findings were confirmed after performings.v.c.

3.2.2. Within-load analyses (targets versus non-targets)Functional imaging results obtained by the comparison of tar-

get and non-target retrieval revealed the following results: in L1,we found no differences in prefrontal activation in either direc-tion. In L2 and L3, differences in brain activation were foundonly during the comparsion non-targets > targets. In L2, activa-tion differences were found in left inferior frontal gyrus (BA 47:x = −30, y = 30, z = −15, Z = 3.84; BA 47: x = −48, y = 36, z = 0,Z = 3.40). In L3, activation differences were confined to rightinferior frontal gyrus (BA 47: x = 54, y = 33, z = −9, Z = 4.2); seealso Fig. 2. Again, these results were confirmed after performings.v.c.

4. Discussion

idsc2biTmjeitbdtrpg&dccaesWese

tive processing of information (Nolde, Johnson, & D’Esposito,1998). In WM-retrieval, DLPFC activation has been specificallylinked to memory scanning. Thus, increasing DLPFC activation,as revealed by our task, may well be associated with scanningthrough an increasing array of potentially relevant memoranda,although other interpretations may be valid as well, including theoccurrence of increasing executive processing (Wager & Smith,2003).

One potential limitation of the between-conditions analysis isthat our functional model may not thoroughly allow a clear-cutsegregation of delay-specific versus retrieval-specific activationdue to methodological restrictions, e.g. due to the modellingof the delay-period as a single regressor. However, this con-straint does not affect differential comparisons of both targetsand non-targets, since activation differences within-conditionare relatively unbiased from potential delay-related activationdue to the subtraction procedure: in our study, we could identifyventrolateral BA 47 as the only brain region being differentiallyactive during retrieval of non-targets compared with targets. LeftBA 47 was active in L2, whereas activation of right BA 47 waspresent in L3. No activation differences were found comparingtargets with non-targets.

In previous neuroimaging studies, VLPFC activation wasfound during tasks that require comparison and judgement ofstimuli held in both WM and long-term memory (Petrides,1994), stimulus selection (Rushworth, Nixon, Eacott, &Psiiao1(wrmhsoe

gUtmsr(mtagit

m

In this study, we analyzed load-dependent responses dur-ng retrieval from verbal working memory as well as activationifferences between the retrieval of targets and non-targets. Con-istent with previous studies of memory-related retrieval pro-esses (Leung et al., 2005; Ranganath, Johnson, & D’Esposito,000; Wagner, 1999), we found load-sensitive activation ofoth VLPFC and DLPFC. Bilateral VLPFC was active dur-ng the retrieval of one item from a previously manipulated set.his region is thought to be critical for a variety of first-orderemory processes including comparisons between probes or

udgements between the ocurrence of a given stimulus (Leet al., 2000). Thus, VLPFC may initiate explicit retrieval ofnformation from WM, at least at low load-demands (i.e. afterhe manipulation of a single item). Both VLPFC (BA 47) andilateral DLPFC (BA 9) were engaged with increasing retrievalemand. Findings from previous memory studies have suggestedhat ventral prefrontal regions may implement both encoding andetrieval, whereas more dorsal regions of prefrontal cortex mayreferentially act during retrieval (Wagner, 1999) or memory-uided response selection (Rowe, Toni, Josephs, Frackowiak,

Passingham, 2000). Our results show that with increasingemand, both ventral and dorsal areas may support retrieval pro-esses, being in line with the concept that VLPFC and DLPFCan be flexibly engaged during memory retrieval (D’Esposito etl., 2000; Johnson, Hashtroudi, & Lindsay, 1993; Ranganatht al., 2000). However, it remains open if ventro- and dor-olateral may simultaneously support different types during

M-retrieval. For instance, prefrontal foci during retrieval frompisodic memory have been associated with retrieval attempt anduccess (McIntosh, Nyberg, Bookstein, & Tulving, 1997; Ruggt al., 1998), monitoring functions (Petrides, 1996) and reflec-

assingham, 1997) and holding on-line of both spatial and non-patial information (Goldman-Rakic, 1990, 1996). Functionalmaging studies of memory function indeed suggest that themplementation of an intended act or plan to recall or rememberspecific information may be the lowest common denominatorf VLPFC activation (Courtney, Ungerleider, Keil, & Haxby,997; Henson, Shallice, & Dolan, 1999; Owen, 2000); see alsoOwen et al., 2005). In more recent studies, VLPFC activationas shown to be specifically related to the extent of explicit

etrieval of a given probe, and that specific attentional changesay occur after such an intention. VLPFC-function was also

ypothesized to be associated with the ocurrence of a specifictimulus presented for inspection and comparison with previ-usly relevant material, as featured e.g. by the n-back task (Owent al., 2005).

At present, only a few studies have attempted to distin-uish probe-related activation in PFC. Jiang, Haxby, Martin,ngerleider, and Parasuraman (2000), for example, have found

hat the inferior PFC may be more responsive to stimuli thatatch the sample stimuli than those that do not match. Other

tudies have shown that DLPFC and cingulate regions mayespond differentially to negative probes during a face WM-taskDruzgal & D’Esposito, 2001), and that anterior PFC may play aajor role during the recognition of non-targets compared with

argets (Leung et al., 2005). Interestingly, apart from their majorctivation focus in anterior PFC, Leung et al. (2005) also foundreater changes of activation of inferior frontal gyrus (BA 47)n response to negative probes than to positive probes, althoughhis effect failed to reach statistical significance.

However, our task and methods differ from the above-entioned studies in several ways. First, we included both non-

2562 R.C. Wolf et al. / Neuropsychologia 44 (2006) 2558–2563

targets and targets in our analysis to test for differential effects.Second, we excluded incorrect trials, accounting for the biasthat activation differences may be atrributed to response errors.Furthermore, unlike to previous imaging work, we did not ana-lyze recognition effects per se during WM-retrieval. Given thefact that the initial stimulus set had to be manipulated during thedelay and thus changes its initial character, the subjects in ourtask did not perform pure probe recognition. In this case, the spe-cific retrieval demand involved acceptance or rejection of targetsand non-targets, given the fact that subjects had to deal with anew set of stimuli after manipulation. Thus, VLPFC may medi-ate explicit retrieval processes wich extend beyond recognition,involving comparison processes with newly formed informa-tion held on-line as well as acceptance and rejection of probes.In addition, the comparison and retrieval of non-targets is asso-ciated with longer reaction times (being present in L2 and L3only), being in line with theoretical work on memory scanning(Sternberg, 1966).

Furthermore, activation of left BA 47 was found in L2,whereas right BA 47 was found to be active in L3 only. Thisfinding is likely to be due to a statistical threshold-effect, sincein L3 we found additional activation of left BA 47 when loweringthe threshold to P < 0.05. In L2, activation of right BA 47 was notpresent, even after lowering the threshold to P < 0.05. Thus, weinterpret right-lateralized activation in L3 in terms of increasingcortical processing demand, consistent with models of neuralertrae

arigmPVtrnao

R

BB

B

B

C

D’Esposito, M., Postle, B. R., & Rypma, B. (2000). Prefrontal cortical con-tributions to working memory: Evidence from event-related fMRI studies.Experimental Brain Research, 133(1), 3–11.

Druzgal, T. J., & D’Esposito, M. (2001). A neural network reflecting deci-sions about human faces. Neuron, 32(5), 947–955.

Friston, K. J., Holmes, A., Worsley, K. J., Poline, J. B., Frith, C. D., &Frackowiak, R. S. J. (1995). Statistical parametric maps in functionalimaging: A general linear approach. Human Brain Mapping, 2, 189–210.

Friston, K. J., Holmes, A. P., Poline, J. B., Grasby, P. J., Williams, S. C.,Frackowiak, R. S., et al. (1995). Analysis of fMRI time-series revisited.Neuroimage, 2(1), 45–53.

Funahashi, S., Bruce, C. J., & Goldman-Rakic, P. S. (1989). Mnemonic cod-ing of visual space in the monkey’s dorsolateral prefrontal cortex. Journalof Neurophysiology, 61(2), 331–349.

Funahashi, S., Bruce, C. J., & Goldman-Rakic, P. S. (1993). Dorsolateral pre-frontal lesions and oculomotor delayed-response performance: Evidencefor mnemonic “scotomas”. Journal of Neuroscience, 13(4), 1479–1497.

Fuster, J. M., & Alexander, G. E. (1971). Neuron activity related to short-termmemory. Science, 173(997), 652–654.

Goldman-Rakic, P. S. (1990). Cellular and circuit basis of working memoryin prefrontal cortex of nonhuman primates. Progress in Brain Research,85, 325–335 [discussion, pp. 335–326].

Goldman-Rakic, P. S. (1996). The prefrontal landscape: Implications of func-tional architecture for understanding human mentation and the centralexecutive. Philosophical Transactions of the Royal Society of London.Series B: Biological Sciences, 351(1346), 1445–1453.

Habeck, C., Rakitin, B. C., Moeller, J., Scarmeas, N., Zarahn, E., Brown, T.,et al. (2005). An event-related fMRI study of the neural networks under-lying the encoding, maintenance, and retrieval phase in a delayed-match-to-sample task. Brain Research. Cognitive Brain Research, 23(2–3),207–220.

H

H

J

J

J

L

L

M

M

N

O

O

fficiency (Rypma & D’Esposito, 2000, 2001). Alternatively,ight prefrontal engagement in L3 may be due to the fact thathis process may exceed WM, extending to episodic memoryetrieval. However, a clear lateralization effect during encodingnd/or retrieval in episodic memory has been challenged (Leet al., 2000).

As a conclusion, we have demonstrated that both DLPFCnd VLPFC are load-dependently recruited during verbal WM-etrieval. These results are consistent with theoretical and exper-mental work indicating that the PFC is a functionally hetero-eneous region which is involved in several WM and long-termemory processes to a various degree (D’Esposito et al., 2000;ostle & D’Esposito, 2000). In addition, we could show thatLPFC-function may include processes that preferentially dis-

inguish between non-targets and targets during verbal WM-etrieval. Thus, VLPFC is not solely activated by simple recog-ition of explicitly presented stimuli, and may be preferentiallyctivated during comparison processes, acceptance or rejectionf a given probe.

eferences

addeley, A. (1986). Working memory. Oxford: Oxford University Press.addeley, A. (1996). The fractionation of working memory. Proceedings of

the National Academy of Sciences of the United States of America, 93(24),13468–13472.

addeley, A. (2003). Working memory: Looking back and looking forward.Nature Reviews. Neuroscience, 4(10), 829–839.

uckner, R. L., & Wheeler, M. E. (2001). The cognitive neuroscience ofremembering. Nature Reviews. Neuroscience, 2(9), 624–634.

ourtney, S. M., Ungerleider, L. G., Keil, K., & Haxby, J. V. (1997). Transientand sustained activity in a distributed neural system for human workingmemory [see comments]. Nature, 386(6625), 608–611.

enson, R. N., Shallice, T., & Dolan, R. J. (1999). Right prefrontal cortexand episodic memory retrieval: A functional MRI test of the monitoringhypothesis. Brain, 122(Pt 7–8), 1367–1381.

olmes, A. P., & Friston, K. J. (1998). Generalisability, random effects andpopulation inference. Neuroimage, 4(4 Pt 2), 754.

iang, Y., Haxby, J. V., Martin, A., Ungerleider, L. G., & Parasuraman, R.(2000). Complementary neural mechanisms for tracking items in humanworking memory. Science, 287(5453), 643–646.

ohnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring.Psychological Bulletin, 114(1), 3–28.

onides, J., Smith, E. E., Marshuetz, C., Koeppe, R. A., & Reuter-Lorenz,P. A. (1998). Inhibition in verbal working memory revealed by brainactivation. Proceedings of the National Academy of Sciences of the UnitedStates of America, 95(14), 8410–8413.

ee, A. C. H., Robbins, T. W., & Owen, A. M. (2000). Episodic memorymeets working memory in the frontal lobe: Functional-neuroimaging stud-ies of encoding and retrieval. Critical Reviews in Neurobiology, 14(3–4),165–198.

eung, H. C., Gore, J. C., & Goldman-Rakic, P. S. (2005). Differential ante-rior prefrontal activation during the recognition stage of a spatial workingmemory task. Cerebral Cortex, 15(11), 1742–1749.

aldjian, J. A., Laurienti, P. J., Kraft, R. A., & Burdette, J. H. (2003). Anautomated method for neuroanatomic and cytoarchitectonic atlas-basedinterrogation of fMRI data sets. Neuroimage, 19(3), 1233–1239.

cIntosh, A. R., Nyberg, S., Bookstein, F. L., & Tulving, E. (1997). Dif-ferential functional connenctivity of prefrontal and medial temporal cor-tices during episodic memory retrieval. Human Brain Mapping, 5, 323–327.

olde, S. F., Johnson, M. K., & D’Esposito, M. (1998). Left prefrontalactivation during episodic remembering: An event-related fMRI study.Neuroreport, 9(15), 3509–3514.

wen, A. M. (2000). The role of the lateral frontal cortex in mnemonicprocessing: The contribution of functional neuroimaging. ExperimentalBrain Research, 133(1), 33–43.

wen, A. M., McMillan, K. M., Laird, A. R., & Bullmore, E. (2005). N-back working memory paradigm: A meta-analysis of normative functionalneuroimaging studies. Human Brain Mapping, 25(1), 46–59.

R.C. Wolf et al. / Neuropsychologia 44 (2006) 2558–2563 2563

Petrides, M. (1994). Frontal lobes and working memory: Evidence frominvestigations of the effects of cortical excisions in nonhuman primates.In F. Boller & J. Grafman (Eds.), Handbook of neuropsychology (pp.59–82). Amsterdam: Elsevier Science.

Petrides, M. (1996). Specialized systems for the processing of mnemonicinformation within the primate frontal cortex. Philosophical Transactionsof the Royal Society of London. Series B: Biological Sciences, 351(1346),1455–1461 [discussion, pp. 1461–1452].

Postle, B. R., & D’Esposito, M. (2000). Evaluating models of the topo-graphical organization of working memory function in frontal cortex withevent-related FMRI. Psychobiology, 28, 132–145.

Ranganath, C., Johnson, M. K., & D’Esposito, M. (2000). Leftanterior prefrontal activation increases with demands to recallspecific perceptual information. Journal of Neuroscience, 20(22),RC108.

Rowe, J. B., Toni, I., Josephs, O., Frackowiak, R. S., & Passingham, R. E.(2000). The prefrontal cortex: Response selection or maintenance withinworking memory? Science, 288(5471), 1656–1660.

Rugg, M. D., Fletcher, P. C., Allan, K., Frith, C. D., Frackowiak, R.S., & Dolan, R. J. (1998). Neural correlates of memory retrievalduring recognition memory and cued recall. Neuroimage, 8(3), 262–273.

Rushworth, M. F., Nixon, P. D., Eacott, M. J., & Passingham, R. E. (1997).Ventral prefrontal cortex is not essential for working memory. Journal ofNeuroscience, 17(12), 4829–4838.

Rypma, B., & D’Esposito, M. (2000). Isolating the neural mechanisms ofage-related changes in human working memory. Nature Neuroscience,3(5), 509–515.

Rypma, B., & D’Esposito, M. (2001). Age-related changes in brain-behaviorrelationships: Evidence from event-related functional MRI studies. Euro-pean Journal of Cognitive Psychology, 13, 235–256.

Sternberg, S. (1966). High-speed scanning in human memory. Science,153(736), 652–654.

Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas of the humanbrain. Stuttgart: Thieme.

Wager, T. D., & Smith, E. E. (2003). Neuroimaging studies of working mem-ory: A meta-analysis. Cognitive, Affective & Behavioral Neuroscience,3(4), 255–274.

Wagner, A. D. (1999). Working memory contributions to human learning andremembering. Neuron, 22(1), 19–22.

Walter, H., Bretschneider, V., Gron, G., Zurowski, B., Wunderlich, A. P.,Tomczak, R., et al. (2003). Evidence for quantitative domain dominancefor verbal and spatial working memory in frontal and parietal cortex.Cortex, 39(4–5), 897–911.

Wilson, F. A., Scalaidhe, S. P., & Goldman-Rakic, P. S. (1993). Dissociationof object and spatial processing domains in primate prefrontal cortex [seecomments]. Science, 260(5116), 1955–1958.

Wolf, R. C., & Walter, H. (2005). Evaluation of a novel event-related paramet-ric fMRI paradigm investigating prefrontal function. Psychiatry Research,140, 73–83.