NeuroImage 12, 495–503 (2000) doi:10.1006/nimg.2000.0624, available online at http://www.idealibrary.com on

Prolonged Reaction Time to a Verbal Working Memory Task Predicts Increased Power of Posterior Parietal Cortical Activation 1 Garry D. Honey,* Edward T. Bullmore,† ,‡ and Tonmoy Sharma* ,2 *Section of Cognitive Psychopharmacology, †Department of Psychiatry, Institute of Psychiatry (King’s College), London, United Kingdom; and ‡Department of Psychiatry, University of Cambridge, Addenbrooke’s Hospital, Cambridge CB2 2QQ, United Kingdom Received September 28, 1999

We used multislice functional magnetic resonance imaging (fMRI) to investigate the association between behavioral and neurophysiological measures of working memory task performance in 20 right-handed male healthy volunteers. Images were acquired over a 5-min period at 1.5 Tesla. We used a periodic design, alternating 30-s blocks of the “n-back” working memory task with 30-s blocks of a sensorimotor control task to activate verbal working memory systems. The power of functional response to the task was estimated by sinusoidal regression at each voxel. The relationship between power of fMRI response and mean reaction time over all 11 working memory trials was explored by multiple regression, with age and mean reaction time to the control task as covariates, at voxel and regional levels of analysis. All subjects were able to perform the n-back task accurately. A spatially distributed network was activated, including dorsolateral prefrontal cortex, inferior frontal gyrus, lateral premotor cortex, and supplementary motor area (SMA) in the frontal lobes. More posteriorly, there were major foci of activation in parietal and occipitoparietal cortex, precuneus, lingual, and fusiform gyri of the ventral occipital lobe, inferior temporal gyrus, and cerebellum. The power of functional response was positively correlated with reaction time in bilateral posterior parietal cortex (Talairach coordinates in x, y, z (mm) 35, ⴚ44, 37 and ⴚ32, ⴚ56, 42), indicating that subjects who found the task difficult, and responded with a slower reaction time, tended to activate these regions more powerfully. One interpretation of this regionally specific relationship between prolonged reaction time and increased power of posterior parietal activation is consistent with prior studies identifying similar areas of parietal cortex as the site of the phonological storage function in verbal working memory. © 2000 Academic Press 1

This work was presented in part at the Fifth International Conference on Functional Mapping of the Human Brain, Dusseldorf, June, 1999. 2 To whom correspondence and reprint requests should be addressed. Fax: ⫹44 (0) 20 7848 0646. E-mail: [email protected].

Key Words: functional magnetic resonance imaging (fMRI); phonological storage; phonological loop; prefrontal cortex; load-response relationship; regression.

INTRODUCTION Working memory is a hypothetical construct in cognitive psychology that refers to a limited capacity system for the simultaneous storage and processing of information, to guide behavior in the absence of environmental cues (Baddeley, 1986, 1992). It has been fractionated into a central executive that organizes the allocation and coordination of processing resources to utilize stored representations and modality-specific short-term “slave stores”: a speech-based, “phonological loop” for verbal rehearsal and storage; and a “visuospatial scratchpad” for storage of nonverbal information (Baddeley, 1992). Working memory is considered to be fundamental to a broad range of cognitive processes, including reasoning, language comprehension, and problem solving (Jonides, 1995). Working memory dysfunction is considered central to several clinical conditions, including Alzheimer’s disease (Morris, 1994), Parkinson’s disease (Dalrymple et al., 1994), and schizophrenia (Green, 1996; Goldman-Rakic, 1990). Studies using both positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) have consistently supported the involvement of prefrontal and parietal regions in verbal working memory in humans (Grasby et al., 1993; Paulesu et al., 1993; Petrides et al., 1993a,b; Cohen et al., 1994, 1997; Mellers et al., 1995; Awh et al., 1996; Salmon et al., 1996; Schumacher et al., 1996; Smith et al., 1995, 1996; Barch et al., 1997; Braver et al., 1997; Jonides et al., 1997, 1998; Kammer et al., 1997; Manoach et al., 1997; Callicot et al., 1998; Fletcher et al., 1998; Ye et al., 1998; Collette et al., 1999; Postle et al., 1999; Clark et al., 2000). Functional specialization has been suggested, whereby dorsal prefrontal cortex is specialized for ex-

495

1053-8119/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

496

HONEY, BULLMORE, AND SHARMA

ecutive control, left inferior frontal gyrus (Broca’s area) is specifically responsible for subvocal rehearsal, and posterior parietal cortex is specialized for phonological storage or active maintenance of verbal memoranda (Paulesu et al., 1993; Gathercole, 1994). However, there is evidence to suggest that the functional neuroanatomy of working memory may be more complex and anatomically distributed, with the prefrontal cortex also involved in active maintenance (Cohen et al., 1997) and parietal regions playing a role in executive functions (Cohen et al., 1997; Collette et al., 1999). A number of studies have examined the relationship between working memory and brain activity by parametrically incrementing load, i.e., by increasing the number of memoranda n that must be manipulated in active storage during performance of an n-back task. Activation of prefrontal and posterior parietal cortical areas has been positively correlated with working memory load (Braver et al., 1997; Barch et al., 1997; Cohen et al., 1997; Jonides et al., 1997; Manoach, 1997; Klingberg et al., 1997; Fletcher et al., 1998). However, this approach is potentially complicated by the fact that increasing n not only increases the demand on executive processes required for conscious manipulation of memoranda, but also the delay during which the material must be maintained. Representations of verbal information within the phonological store decay with time, unless refreshed by subvocal rehearsal mechanisms. Thus activations correlated with load could identify brain regions specialized for executive processing, storage, or rehearsal functions of the phonological loop, or even conceivably for nonmnemonic functions, such as estimation of temporal delay, that may be engaged during performance of working memory tasks (Rubia et al., 1998). In this study, we have taken an alternative approach to examine the relationship between brain activity and working memory performance. Rather than increasing load experimentally and estimating the relationship between n and functional response, we have kept n constant (⫽ 2) and investigated the relationship between power of functional response and mean reaction time over 11 trials presented in the context of a blocked periodic design. We assume that increased reaction time indexes an increase in the temporal delay during which memoranda must be maintained in active storage; and we rely on natural variability in efficiency of task performance over 20 subjects to identify brain regions where functional response is related specifically to delay. METHODS Subjects and Study Design Twenty right-handed male volunteers, with no history of neurological or psychiatric disease, were re-

cruited by advertisement from the local community in South East London. Group mean age was 39.3 years (range, 19 – 64; standard deviation (SD) ⫽ 13.6 years). Mean IQ, measured using the National Adult Reading Scale (NART), was 113 (SD ⫽ 6.44). Written informed consent was obtained from all participants. The study was approved by the Bethlem Royal & Maudsley Hospital (Research) Ethical Committee. Verbal Working Memory Task We used a blocked periodic BA design to activate brain regions specialized for executive and active maintenance components of verbal working memory, as originally described by Cohen et al. (1994). Two contrasting conditions were visually presented in 30-s epochs to subjects via a prismatic mirror as they lay in the scanner. During each epoch of the baseline (B) condition, subjects viewed a series of 13 letters, which appeared one at a time with interstimulus interval (ISI ⫽ 2.3 s) and were required to press a button with their right index finger when the letter “X” appeared. During each epoch of the activation (A) condition, subjects again viewed a series of 13 letters (ISI) ⫽ 2.3 s and were required to press a button with their right index finger if the currently presented letter was the same as that presented two trials previously (e.g., GD-G, but not R-L-F-R or T-T). The two conditions were matched for number of target letters presented per epoch ⫽ 2 or 3. Five cycles of alternation between conditions were presented in the course of each 5-min experiment; the baseline condition was always presented first. Subject performance on both tasks during scanning was monitored in terms of reaction time to target letters and accuracy (number of target letters correctly identified). All subjects received identical training in task performance prior to scanning. Subjects also participated in two other 5-min experiments during the same scanning session (to be reported elsewhere); in order to minimize the potential confounding effects of experimental order, the within-session order of experiments was pseudo-randomized. Functional MRI Image acquisition. Gradient-echo echoplanar MR images were acquired using a 1.5 Tesla GE Signa System (General Electric, Milwaukee WI) fitted with Advanced NMR hardware and software (ANMR, Woburn MA) at the Maudsley Hospital (London, UK). In each of 14 noncontiguous planes parallel to the intercommissural (AC-PC) line, 100 T2*-weighted MR images, depicting BOLD contrast were acquired: TE ⫽ 40 ms, TR ⫽ 3 s, in-plane resolution ⫽ 3.1 mm, slice thickness ⫽ 7 mm, slice skip ⫽ 0.7 mm.

497

WORKING MEMORY AND FUNCTIONAL MRI

Activation Mapping Following estimation and correction of movementrelated effects in each fMRI time series (Bullmore et al., 1999a), the power of periodic signal change at the (fundamental) BA frequency of stimulation was estimated by sinusoidal regression (Bullmore et al., 1996). The standardized power of functional response at BA frequency (or fundamental power quotient, FPQ) was estimated at each voxel and represented in a parametric map. Each observed fMRI time series was then randomly permuted 10 times, and FPQ reestimated after each permutation. This resulted in 10 parametric maps (for each subject at each plane) of FPQ estimated under the null hypothesis that FPQ is not determined by experimental design. All parametric maps of FPQ were registered in the standard space of Talairach and Tournoux (1988) and smoothed with a 2-D Gaussian filter (full-width half-maximum ⫽ 7 mm). Voxels demonstrating significant power of response over all 20 subjects were then robustly identified by computing the median value of FPQ at each intracerebral voxel of the observed parametric maps (total search volume V ⫽ 19,858 voxels) and comparing it to the permutation distribution of median FPQ obtained from the permuted parametric maps (Brammer et al., 1997). The voxel-wise one-tailed probability of false positive activation was P ⫽ 0.00005; the expected number of false-positive tests was less than 1. Generically activated voxels were colored and superimposed on the grey scale Talairach template, to create generic brain activation maps (GBAMs). To examine the relationship between power of functional response and mean reaction time (RT), we fitted the following linear model at each intracerebral voxel of the observed FPQ maps: FPQ i,j ⫽ ␮ i ⫹ ␤ 1RT1 j ⫹ ␤ 2RT2 j ⫹ ␤ 2A j ⫹ ⑀ i,j

(1)

Here FPQ i,j is the power of response by the jth subject at the ith voxel; ␮ i is the overall mean at the ith voxel; RT1 j and RT2 j are, respectively, the mean reaction times for the activation (working memory) condition and baseline (“look for X”) conditions; A j is the age of subject j, and ⑀ i,j is an error term. The null hypothesis that ␤ 1 ⫽ 0 was tested by permutation at all generically activated voxels with two-tailed P ⫽ 0.005 (Bullmore et al., 1999b; Edgington, 1980). To examine the form of the relationship between functional response and reaction time in greater detail, we also estimated regional mean power, by averaging FPQ observed at an index voxel and its eight nearest neighbors in 2-D, for the following five regions (with x, y, z coordinates (mm) in standard space in parentheses): right and left posterior parietal cortices (35, ⫺44, 37 and ⫺43, ⫺53, 37), right and left dorsal prefrontal cortices (48, 8, 37 and ⫺43, 17, 26), and supplementary motor area (SMA) (3, 3, 48). Locally weighted regres-

sion lines were superimposed on scatterplots of these data against reaction time to assess possible nonlinearities in the form of the reaction time–response relationship. RESULTS Verbal Working Memory Performance Subjects performed the working memory task with a high degree of accuracy: 96% (standard deviation (SD) ⫽ 6.1%) of targets correctly identified. Reaction times ranged from 0.36 to 0.72 s (mean ⫽ 0.56 s, SD ⫽ 0.11 s). Reaction time for the control condition and the working memory condition were significantly correlated (Spearman’s rho 0.652, two-tailed P ⫽ 0.002). For both conditions, there was no correlation between age and reaction time or task accuracy (as indexed by correct identification of targets) and reaction time (P ⬎ 0.05). Functional MRI Generic brain activation map. A significant response during the working memory condition was observed in a distributed cortical network, comprising bilateral parietal and occipitoparietal cortex (extending through the supramarginal gyrus (Brodmann’s area (BA) 40), angular gyrus (BA 39) and precuneus (BA 7)), bilateral (predominantly left sided) dorsolateral prefrontal cortex (BA 9, 10, and 46), inferior frontal gyrus (BA 44, 45), lateral premotor cortex (BA 6), precentral gyrus (BA 4), ventral occipital cortex (fusiform, lingual and inferior/middle occipital gyri (BA 18, 19, 37)), and the cerebellum. Midline structures showing significant response included the SMA (BA 6) and anterior cingulate gyrus (BA 24, 32). Significant responses during the baseline condition were observed in the medial frontal lobe (BA 8, 10) and the left postcentral gyrus (BA 43). Talairach coordinates and other details for these regional foci of activation are given in Table 1; selected slices of the GBAM are shown in Fig. 1. Reaction time–response relationship. There was a significant linear effect of reaction time for the working memory condition on power of functional response at 55 voxels in total: search volume ⫽ 1060 voxels, twotailed P ⫽ 0.005, expected number of false-positive tests ⫽ 5. This relationship between reaction time and functional response was evident in posterior parietal cortex bilaterally (BA 40), (x, y, z coordinates: 35, ⫺44, 37 and ⫺43, ⫺53, 37) and SMA (0, 6, 48); (see Fig. 1). There was no evidence from this voxel level analysis for a significant reaction time–response relationship in other frontal areas. A significant linear effect of reaction time for the working memory condition on the power of functional

498

HONEY, BULLMORE, AND SHARMA

TABLE 1 Main Regional Foci of Generic Brain Activation Talairach coordinates Brain region

Brodmann region

Hemisphere

Number of voxels

x

y

z

Supramarginal gyrus, extending into angular gyrus, inferior, and superior parietal lobe and precuneus Dorsolateral prefrontal cortex and inferior frontal gyrus Dorsolateral prefrontal cortex Supplementary motor area Lateral premotor cortex Precentral gyrus

BA40, 39, and 7

Bilateral

470

⫺35

⫺53

37

BA BA BA BA BA

Extrastriate cortex

BA 18/19

Left Right Midline Left Right Left Left Right Right

48 5 54 16 23 19 40 16 7

⫺43 35 0 ⫺43 32 ⫺20 ⫺35 ⫺35 26

17 47 8 ⫺6 ⫺8 ⫺14 ⫺72 ⫺64 ⫺69

26 15 48 42 53 53 ⫺13 ⫺13 ⫺18

Cerebellum

9/46 and 44 46/10 6 6 4

—

response in bilateral posterior parietal cortex was confirmed at a regional level, whereas reaction time for the control task was excluded in a stepwise multiple regression model (Table 2). There was no significant linear effect of reaction time for either condition on regional measures of frontal fMRI response. Scatterplots of regional mean power of activation against reaction time (Fig. 2) confirm the lack of a strong relationship between prefrontal response and reaction time. The difficulty–response relationship for posterior parietal cortex bilaterally is linearly significant, but locally weighted regression highlights a degree of nonlinearity: power of parietal response is disproportionately increased in subjects with mean reaction times greater than 0.5 s. DISCUSSION The main finding of this study is that prolonged reaction time to a verbal working memory task was significantly associated with increased power of functional response in a subset of cortical regions generically activated by the task. Specifically, we found a reaction time–response relationship in bilateral posterior parietal cortex, but not in dorsolateral prefrontal or inferior frontal cortex. This reaction time–response relationship was not evident for the control task, suggesting that the observed relationship can not be attributed to a nonspecific speed of processing effect. In this study, the manipulation of verbal information, interpreted in terms of Baddeley’s working memory model, involves numerous component processes, including sensory encoding of the input, phonological transcoding and storage of the stimulus, subvocal rehearsal and maintenance of the memoranda, updating of stored representations, comparison of current and previously presented items, and the preparation and

execution of a motor response. The relationship between prolonged reaction time and power of functional response in the posterior parietal cortex could theoretically involve any of these component processes. However, there are several lines of evidence from both neuropsychological and neuroimaging studies that may implicate the phonological storage process to be of particular importance in this behavioral–physiological correlation. Patients with selective deficits of the phonological loop typically have lesions of the inferior parietal lobe, often specifically involving the supramarginal gyrus (Sagar et al., 1988; Vallar and Shallice, 1990; Della Sala and Logie, 1993). Using positron emission tomography (PET) Paulesu et al. (1993) showed that when brain activation associated with a rhyming task, which would be expected to engage articulatory rehearsal but not the phonological store (Burani et al., 1991; Vallar and Baddeley, 1984), was subtracted from that produced by a phonological working memory task, which would be expected to activate both the phonological store and the articulatory rehearsal mechanism (Warrington and Shallice, 1969; Shallice and Butterworth, 1977; Vallar and Baddeley, 1984), differential activation was observed only in the posterior parietal cortex. These findings were replicated by Salmon et al. (1996). Similarly, Jonides et al. (1998) showed parietal activation during short-term storage of nonwords, predominantly by a phonological code. This may explain the lack of parietal activation in a similar working memory study reported by Fiez et al. (1996), which used real words, which are stored using a predominantly semantic code, with little requirement for phonological representations in working memory. The localization of the bilateral parietal foci observed in our study (Talairach coordinates in x, y, z (mm) 35, ⫺44, 37; ⫺32, ⫺56, 42) are very similar to many previously published coordinates for a phonological store

499

WORKING MEMORY AND FUNCTIONAL MRI

FIG. 1. Generic brain activation maps. The four maps in the top row show the main brain regions activated by the verbal working memory task. The maps in the bottom row show the linear relationship between functional response and reaction time observed bilaterally in posterior parietal cortex (BA 40) and SMA (BA 6). The distance above the intercommissural line in the standard space of Talairach & Tournoux is given in millimeters below each map.

in parietal cortex. Becker et al. (1999) reported mean coordinates (⫺33, ⫺48, 39) averaged over six experiments, which localized the phonological store to left superior parietal lobe (Awh et al., 1996; Smith et al., 1995 (including data from two separate experiments (Smith et al., 1996; Petrides et al., 1993a, b). This

average location is very close to the left hemisphere locus in our data. Using subtraction analysis of PET image datasets to identify structures mediating phonological storage, Jonides et al. (1998) reported a right hemisphere locus (42, ⫺51, 40) close to the right hemisphere localization in the current data. These authors

TABLE 2 Multiple Regression Values for Regional Analysis of Behavioral–Functional Relationships Independent Variables 1 Twoback Twoback Twoback Twoback Twoback

2 RT RT RT‡ RT‡ RT‡

Control Control Control Control Control

condition condition condition condition condition

RT‡ RT‡ RT‡ RT‡ RT‡

Model Goodness of Fit Statistics 3

Dependent Variable

F

df

Adjusted r 2

P value

Age‡ Age‡ Age‡ Age‡ Age‡

Right posterior parietal cortex Left posterior parietal cortex Supplementary motor area Left prefrontal cortex Right prefrontal cortex

6.64 10.895

1,19 1,19

0.229 0.342

0.019* 0.004** n.s. n.s. n.s.

Note. **P ⬍ 0.01; *P ⬍ 0.05; ‡ excluded by stepwise regression; n.s., nonsignificant (all variables excluded by stepwise regression); RT, reaction time.

500

HONEY, BULLMORE, AND SHARMA

FIG. 2. Scatterplots of regional mean power of activation against working memory reaction time in bilateral posterior parietal cortex, dorsolateral prefrontal cortex, and SMA. A linear effect of reaction time on the power of functional response is observed bilaterally in posterior parietal cortex (dotted line); solid lines show results of locally weighted regression.

also reported bilateral ROI analyses, based on regions derived from the results of Awh et al. (1996) (12, ⫺64, 47; ⫺33, ⫺46, 38) and also averaged across three previous studies of working memory (37, ⫺49, 40; ⫺35, ⫺49, 38) (Smith et al., 1996; Schumacher et al., 1996; Jonides et al., 1997). The close correspondence between the localization of the phonological store in previous studies and the locus of the relationship between reaction time and parietal activation observed in this study support the suggestion that reaction time latency in this study indexes increased phonological storage. However, as noted by Becker et al. (1999), there are also several studies which consistently find a more

ventral location (mean coordinates: ⫺52, ⫺27, 22) for phonological storage (Paulesu et al., 1993, 1996; Salmon et al., 1996; Becker et al., 1996). It is currently unclear as to how these apparently discrepant findings are to be reconciled; but, as suggested by Becker et al. (1999), a reevaluation of current theoretical conceptualizations of the mechanisms of phonological storage may be necessary to account for the functional involvement of at least two (dorsal and ventral) areas of posterior parietal cortex. The close anatomical coincidence between the results of these prior studies of phonological storage and our data support the interpretation that prolonged re-

WORKING MEMORY AND FUNCTIONAL MRI

action time entails retaining the memoranda in active storage for a longer period of time, and this causes enhanced power of activation in responsible brain regions. The regional specificity of this relationship is interesting. We did not see a reaction time–response relationship in inferior frontal gyrus or supplementary motor area, which have been implicated in the subvocal rehearsal mechanism of the phonological loop. This may be because the natural variability in reaction time for these subjects was not sufficient to demand extra rehearsal of memorized material in the slowest subjects. The frequency of subvocal rehearsal depends on the number of memoranda, but it is unlikely to be more than 1 Hz, whereas the range of mean reaction times for these subjects was less than 1 s. The lack of relationship between reaction time and power of response in dorsolateral prefrontal cortex suggests that previously reported relationships between experimental increased load and prefrontal activation (Cohen et al., 1996; Braver et al., 1997; Jonides et al., 1997; Manoach et al., 1997; Callicot et al., 1999) may be determined more by the extra demands on executive processing of incrementing n than by the concomitantly increased period of time for which the n memoranda must be memorized. However, it is important also to consider alternative explanations for the relationship between prolonged RT and power of parietal response, such as increased attentional demands and use of visuospatial strategies. The correlation with reaction time and fMRI signal was observed bilaterally in the superior parietal cortex. Lesion studies have generally implicated the left hemisphere in verbal working memory deficits; however, functional imaging studies have been equivocal on the issue of laterality, with some studies finding predominantly left-sided activation (Paulesu et al., 1993; Mellers et al., 1995; Awh et al., 1996; Smith et al., 1996; Salmon et al., 1996; Barch et al., 1997) and others finding bilateral activation (Petrides et al., 1993; Schumacher et al., 1996; Cohen et al., 1997; Braver et al., 1997; Manoach et al., 1997; Jonides et al., 1997, 1998). Awh et al. (1996) suggest that the activation of right parietal cortex in a demanding task may imply functional recruitment of homologous regions in the right hemisphere to provide assistance in the processing of an essentially left-hemispheric task. Jonides et al. (1998) suggested that the greater right parietal activation in their study may reflect the shifting of attentional processing of internal representations from one item to another, which would preferentially involve the right hemisphere (Corbetta et al., 1993; Heinze et al., 1994). This seems an unlikely explanation of our data: the analysis used by Jonides et al. (1998) contrasted a verbal storage condition to a visual fixation control, thus sequential attentional shifts between items were not “subtracted” from the experimental condition in this design. However, stimulus frequency and duration

501

of presentation were both matched across the control and experimental conditions in the present study, and therefore one would not expect differences in attentional shifts between items across the two conditions. General attentional or speed-of-processing differences between subjects that might contribute to variability in their reaction times are also discounted as possible explanations by the persistence of the association between working memory RT and parietal response after including RT to the sensorimotor control task as a covariate in the multiple regression model. However, Salmon et al. (1996) contrasted a working memory updating task with a phonological working memory task and observed bilateral parietal activation, predominantly in the right hemisphere, which they attributed to the subjects’ use of a visual imagery strategy. Preferential use of visual imagery in the subjects who responded more slowly is thus a plausible alternative explanation of the relationship we observed between reaction time and right parietal activation. The main limitation of the study is that, although the blocked periodic design is highly efficient and provides a comparatively greater signal to noise ratio than an event-related design, there is limited temporal resolution in analysis of behavioral-physiological associations. For example, it would be very interesting to know if the association we have shown between RT and parietal response over subjects was also demonstrated over multiple trials within a single subject. It would also be interesting to ascertain whether RT was prolonged and parietal response was increased, specifically in trials that were preceded by phonemically similar “distractor” stimuli. These analyses would considerably strengthen the case for working memory RT as an index of phonological storage time; but to address these questions powerfully will require eventrelated experimental studies in future. In conclusion, the present study has replicated the findings of a number of functional imaging studies of verbal working memory and extended these findings to relate power of physiological response measured by fMRI to task performance efficiency, as measured using response latency. Exploiting natural variability in subjective task difficulty in this way can provide complementary insights into functional anatomy of working memory to those provided by experimentally manipulating objective task difficulty. ACKNOWLEDGMENTS We thank Dr. Andy Simmons, Mr. Chris Andrew, and the neuroimaging staff at the Maudsley Hospital for technical support. We are also grateful to the subjects for their participation in the study. E.T.B. was supported by the Wellcome Trust. This study was funded by PsychMed Ltd.

502

HONEY, BULLMORE, AND SHARMA

REFERENCES Awh, A., Jonides, J., Smith, E. E., Schumacher, E. H., Koeppe, R. A., and Katz, S. 1996. Dissociation of storage and rehearsal in verbal working memory: Evidence from Positron Emission Tomography. Psychol. Sci. 7: 25–31. Baddeley, A. D. 1986. Working Memory. Claredon Press, Oxford. Baddeley, A. D. 1992. Working memory. Science 256: 556 –559. Barch, D. M., Braver, T. S., Nystrom, L. E., Forman, S. D., Noll, D. C., and Cohen, J. D. 1997. Dissociating working memory from task difficulty in human prefrontal cortex. Neuropsychologia 35: 1373–1380. Becker, J. T., Mintun, M. A., Diehl, J. H., Dobkin, J., Martidis, A., Madoff, D. C., and DeKosky, S. T. 1996. Compensatory reallocation of brain resources supporting verbal episodic memory in Alzheimer’s disease. Neurology 46: 692–700. Becker, J. T., MacAndrew, D. K., and Fiez, J. A. 1999. A Comment on the functional localisation of the phonological storage subsystem of working memory. Brain Cogn. 41: 27–38. Brammer, M. J., Bullmore, E. T., Simmons, A., Williams, S. C. R., Grasby, P. M., Howard, R. J., Woodruff, P. W. R., and RabeHesketh, S. 1997. Generic brain activation mapping in fMRI: A non-parametric approach. Magn. Reson. Imaging 15: 763–770. Braver, T. S., Cohen, J. D., Nystrom, L. E., Jonides, J., Smith, E. E., and Noll, D. C. 1997. A parametric study of prefrontal cortex involvement in human working memory. Neuroimage 5: 49 – 62. Bullmore, E., Brammer, M., Williams, S., Rabe-Hesketh, S., Janot, N., David, A., Mellers, J., Howard, R., and Sham, P. 1996. Statistical methods of estimation and inference for functional MR image analysis. Magn. Res. Med. 35: 261–277. Bullmore, E. T., Brammer, M. J., Rabe-Hesketh, S., Curtis, V. A., Morris, R. G., Williams, S. C. R., Sharma, T., and McGuire, P. K. 1999a. Methods for diagnosis and treatment of stimulus correlated motion in generic brain activation studies using fMRI. Hum. Brain Mapp. 7: 38 – 48. Bullmore, E. T., Suckling, J., Overmeyer, S., Rabe-Hesketh, S., Taylor, E., Brammer, M. J. 1999b. Global, voxel, and cluseter tests, by theory and permutation, for a difference between two groups of structural MR images of the brain. IEEE Trans. Med. Imag. 18: 32– 42. Burani, C., Vallar, G., Bottini, G. 1991. Eur. J. Cogn. Psychol. 3: 379 –398. Callicot, J., Ramsey, N. F., Tailent, K., Bertolino, A., Knable, M. B., Coppola, R., Goldberg, T., van Gelderen, P., Mattay, V. S., Frank, J. A., Moonen, C. T. W., and Weinberger, D. R. 1998. fMRI brain mapping in psychiatry: Methodological issues illustrated in a study of working memory in schizophrenia. Neuropsychopharmacology 18: 186 –196. Callicot, J. H., Mattay, V. S., Bertolino, A., Finn, K., Coppola, R., Frank, J. A., Goldberg, T. E., and Weinberger, D. R. 1999. Physiological characteristics of capacity constraints in working memory as revealed by functional MRI. Cereb. Cortex 8: 1047. Clark, C. R., Egan, G. F., McFarlane, A. C., Morris, P., Weber, D., Sonkkilla, C., Marcina, J., and Tochon-Danguy, H. J. 2000. Updating working memory for words: A PET activation study. Hum. Brain Mapp. 9: 42–54. Cohen, J. D., Forman, S. D., Braver, T. S., Casey, B. J., ServanSchreiber, D., and Noll, D. S. 1994. Activation of the prefrontal cortex in a non-spatial working memory task with functional MRI. Hum. Brain Mapp. 1: 293–304. Cohen, J. D., Peristein, W. M., Braver, T. S., Nystrom, L. E., Noll, D. C., Jonides, J., and Smith, E. S. 1997. Temporal dynamics of brain activation during a working memory task. Nature 386: 604 – 608.

Collette, F., Salmon, E., van der Linden, M., Chicherio, C., Belleville, S., Degueldre, C., Delfiore, G., and Franck, G. 1999. Regional brain activity during tasks devoted to the central executive of working memory. Cogn. Brain Res. 7: 411– 417. Corbetta, M., Miezin, F., Shulman, G. L., and Petersen, S. E. 1993. A PET study of visuospatial attention. J. Neurosci. 13: 1202–1206. Dalrymple-Alford, J. C., Kalders, A. S., Jones, R. D., and Watson, R. W. 1994. A central executive deficit in patients with Parkinson’s disease. J. Neurol. Neurosurg. Psych. 57: 360 –367. Della Sala, S., and Logie, R. H. 1993. When working memory does not work: The role of working memory in neuropsychology. In Handbook of Neuropsychology (F. Boller and J. Grafman, Eds.), Vol. 8, pp. 1– 62. Elsevier, Amsterdam. Edginton, E. S. 1980. Randomization tests. Marcel Dekker, New York. Fiez, J. A., Raife, E. A., Balota, D. A., Schwarz, J. P., Raichle, M. E., and Petersen, S. E. 1996. A positron emission tomography study of the short-term maintenance of verbal information. J. Neurosci. 16: 808 – 822. Fletcher, P. C., McKenna, P. J., Frith, C. D., Grasby, P. M., Friston, K. J., and Dolan, R. J. 1998. Brain activations in schizophrenia during a graded memory task studied with functional imaging. Arch. Gen. Psychol. 55: 1001–1008. Gathercole, J. 1994. Neuropsychology and working memory: A Review. Neuropsychology 8: 494 –506. Goldman-Rakic, P. S. 1990. Prefrontal cortical dysfunction in schizophrenia: The relevance of working memory. In Psychopathology and the Brain (B. J. Carroll and J. E. Bartrett, Eds.), pp. 1–23. Raven Press, New York. Grasby, P. M., Frith, C. D., Friston, K. J., et al. 1993. A graded task approach to the functional mapping of brain areas implicated in auditory-verbal memory. Brain 116: 1–20. Green, M. F. 1996. What are the functional consequences of neurocognitive deficits in schizophrenia? Am. J. Psychol. 153: 321– 330. Heinze, H. J., Mangun, G. R., Burchert, W., Hinrichs, H., Scholz, M., Munte, T. F., Gos, A., Scherg, M., Johannes, S., Hundeshage, H., Gazzaniga, M. S., and Hyllyard, S. A. 1994. Combined temporal imaging of brain activity during visual selective attention in humans. Nature 372: 543–546. Jonides, J. 1995. Working memory and thinking: In Invitation to Cognitive Science: Thinking (E. E. Smith and D. N. Osherson, Eds.), Vol. 3, pp. 215–265. MIT Press, Cambridge, MA. Jonides, J., Schumacher, E. H., Smith, E. E., Lauber, E. J., and Awh, E. 1997. Verbal working memory load affects regional brain activation as measured by PET. J. Cogn. Neurosci. 9: 462– 475. Jonides, J., Schumacher, E. H., Smith, E. E., Koeppe, R. A., Awh, E., Reurter-Lorenz, P. A., Marshuetz, C., and Willis, C. R. 1998. The role of the parietal cortex in verbal working memory. J. Neurosci. 18: 5026 –5034. Kammer, T., Bellemann, M. E., Guckel, F., Brix, G., Gass, A., Schlemmer, H., and Spitzer, M. 1997. Functional MR imaging of the prefrontal cortex: Specific activation in a working memory task. Magn. Reson. Imag. 15: 879 – 889. Klingberg, T., O’Sullivan, B. T., and Roland, P. E. 1997. Bilateral activation of fronto-parietal networks by incrementing demand in a working memory task. Cereb. Cortex 7: 465– 471. Manoach, D. S., Schlaug, G., Bettina, S., Darby, D. G., Bly, B. M., Benfield, A., Edelman, R. R., and Warach, S. 1997. Prefrontal cortex fMRI signal changes are correlated with working memory load. Neuroreport 8: 545–549. Mellers, J. D. C., Bullmore, E., Brammer, M., Williams, S. C. R., Andrew, C., Sachs, N., Andrews, C., Cox, T. S., Simmons, A. S., Woodruff, P., David, A. S., and Howard, R. 1995. Neural correlates

WORKING MEMORY AND FUNCTIONAL MRI of working memory in a visual letter monitoring task: An fMRI study. Neuroreport 7: 109 –112. Morris, R. G. 1994. Working memory in Alzheimer-type dementia. Neuropsychology 8: 544 –554. Paulesu, E., Frith, C. D., and Frackowiak, R. S. J. 1993. The neural correlates of the verbal component of working memory. Nature 362: 342–345. Paulesu, E., Frith, U., Snowling, M., Gallagher, A., Morton, J., Frackowiak, R. S. J., and Frith, C. D. 1996. Is developmental dyslexia a disconnection syndrome? Brain 119: 143–157. Petrides, M., Alivastos, B., Meyer, E., Evans, A. C. 1993a. Functional activation of the human frontal cortex during the performance of verbal working memory tasks. Proc. Natl. Acad. Sci. USA 90: 878 – 882. Petrides, M., Alivastos, B., Meyer, E., and Evans, A. C. 1993b. Dissociation of human mid-dorsolateral from posterior-dorsolateral frontal cortex in memory processing. Proc. Natl. Acad. Sci. USA 90: 873– 877. Postle, B. R., Berger, J. S., and D’Esposito, M. 1999. Functional neuroanatomical double dissociation of mnemonic and executive control processes contributing to working memory performance. Proc. Natl. Acad. Sci. USA 96: 12959 –12964. Rubia, K., Overmeyer, S., Taylor, E., Brammer, M., Williams, S., Simmons, A., Andrew, C., and Bullmore, E. 1998. Prefrontal involvement in “temporal bridging” and timing movement. Neuropsychologia 36: 1283–1293. Sagar, H., Sullivan, E., Gabrielli, J., Corkin, S., and Growden, J. 1988. Temporal ordering and short-term memory deficits in Parkinson’s disease. Brain 111: 525–539.

503

Salmon, E., Van der Linden, M., Collette, F., Delfire, G., Maquet, P., Degueldfre, C., Luxen, A., and Granck, G. 1996. Regional brain activity during working memory tasks. Brain 119: 1617–1625. Schumacher, E. H., Lauber, E., Awh, E., Jonides, J., Smith, E. E., and Koeppe, R. A. 1996. PET evidence for an amodal verbal working memory system. Neuroimage 3: 79 – 88. Shallice, T., and Butterworth, B. 1977. Short-term memory impairment and spontaneous speech. Neuropsychologia 15: 729 –735. Smith, E. E., Jonides, J., Koeppe, R. A., Awh, E., Schumacher, E., and Minoshima, S. 1995. Spatial vs object working memory: PET investigations. J. Cogn. Neurosci. 7: 337–358. Smith, E. E., Jonides, J., Koeppe, R. A. 1996. Dissociating verbal and spatial working memory using PET. Cereb. Cortex 6: 11–20. Talairach, J., and Tournoux, P. 1988. A Co-planar Stereotaxic Atlas of a Human Brain. Theime Verlag, Stuttgart. Vallar, G., and Baddeley, A. D. 1984. Fractionation of working memory: Neuropsychological Evidence for a Phonological Short Term Store. J. Verbal Learn. Verbal Behav. 23: 151–161. Vallar, G., and Baddeley, A. D. 1984. Cogn. Neuropsychol. 1: 121–141. Vallar, G., and Shallice, T. 1990. Neuropsychological impairments of short-term memory. Cambridge Univ. Press, Cambridge. Warrington, E. K., and Shallice, T. 1969. The selective impairment of auditory verbal short-term memory. Brain 92: 885– 896. Ye, F. Q., Smith, A. M., Mattay, V. S., Ruttimann, U. E., Frank, J. A., Weinberger, D. R., and McLaughlin, A. C. 1998. Quantitation of regional cerebral blood flow increases in prefrontal cortex during a working memory task: A steady-state arterial spin-tagging study. Neuroimage 8: 44 – 49.