Abstract

Frontal recruitment was characterized using functional magnetic resonance imaging (fMRI) during memory encoding in temporal lobe epilepsy (TLE) patients before and after unilateral medial temporal lobectomy. Twenty-four TLE patients and 12 healthy controls underwent a preoperative fMRI session consisting of verbal and nonverbal incidental memory-encoding tasks that typically lead to robust, lateralized frontal activity in controls. A similar postoperative fMRI session was performed in a subset of patients. Preoperatively, the verbal task resulted in significant additional recruitment of right frontal cortex in left TLE patients, compared with controls. Right TLE patients instead showed typically lateralized frontal activation. Bilateral frontal recruitment has been observed in older adults and in young adults in situations of difficult task demands. Typical right-lateralized patterns of frontal recruitment were found in both patient groups during the nonverbal task, indicating that the bilateral frontal recruitment pattern was engaged dynamically depending on the task. After surgery, left TLE patients regained more lateralized frontal activity. These results demonstrated differences in frontal recruitment in left and right TLE patients. Such differences emerged in specific task settings and were influenced by surgery, suggesting a dynamic mechanism of frontal recruitment that can be obtained in TLE patients, possibly as a response to presurgical dysfunction.

Introduction

Temporal cortex has been strongly implicated in memory function. Its role in memory became evident with the observation of profound retrograde amnesia sustained by patients undergoing bilateral medial temporal resection for the purpose of seizure control (Scoville and Milner 1957; Penfield and Milner 1958). A wealth of subsequent studies prompted by this finding, including both human and animal models, have since established medial temporal cortex as a key player in memory function (Mishkin 1978; Squire and others 1992; for recent reviews, see Simons and Spiers 2003; Frankland and Bontempi 2005).

Frontal cortex has also been implicated in memory function (Buckner and others 1999; Fletcher and Henson 2001). In primates, permanent lesions or transient cooling of frontal cortex impair association learning and recognition memory (Bauer and Fuster 1976; Voytko 1986; Bachevalier and Mishkin 1986). In humans, tasks that encourage memory encoding are associated with robust frontal activation that can be strongly lateralized depending on the material treated. Verbal encoding preferentially activates left prefrontal cortex (Kapur and others 1994; Demb and others 1995; Kelley and others 1998; Wagner, Poldrack, and others 1998), whereas “nonverbal” encoding preferentially activates right prefrontal cortex (Kelley and others 1998, Wagner, Poldrack, and others 1998; McDermott and others 1999). Although localized lesions of frontal cortex do not produce the severe amnesia observed with medial temporal damage (Milner 1964; Stuss and Benson 1986), patients with frontal lobe damage show impairment in complex memory tasks, such as ones that require access to spatiotemporal information, have high levels of interference, or require access to source information (Schacter 1987; Janowsky and others 1989; Incisa Della Rocchetta and Milner 1993). Frontal regions may thus be involved in controlled processes that contribute to memory encoding and retrieval, among other tasks (Buckner and Wheeler 2001; Fletcher and Henson 2001).

Anatomic evidence suggests that frontal and temporal regions may interact in memory function. In primates, prefrontal cortex shares extensive connections with medial temporal cortex (Kuypers and others 1965; Barbas 1988; Ungerleider and others 1989). Both lateral frontal and orbitofrontal cortex, for instance, receive strong monosynaptic connections from perirhinal, entorhinal, parahippocampal, and hippocampal regions (Rosene and Van Hoesen 1977; Van Hoesen 1982; Barbas 1993, 2000; Lavenex and others 2002). Many such frontotemporal connections are reciprocal (Lavenex and others 2002), implying a bidirectional flow of information. Other evidence suggests that frontal and temporal cortex must interact within the same hemisphere for recognition memory to take place (Fuster and others 1985; Parker and Gaffan 1998; Easton and others 2001), although the same does not hold true when the task involves the formation of new associations (Gaffan and others 2002), indicating that a direct connection between frontal and temporal cortex may not be necessary for all aspects of memory function.

The interplay between frontal and temporal regions has similarly been examined in human studies of memory (Raichle and others 1994; Dolan and Fletcher 1997), with some evidence suggesting that frontal lesions can be associated with the above-normal activity in medial temporal cortex (Levine and others 1998). Among patient populations, patients with temporal lobe epilepsy (TLE) have historically been the subject of significant interest in regard to memory function because medial temporal surgery for seizure control can result in significant memory impairment (for recent reviews, see Ojemann 1997; Rausch 2002; Wiebe 2003). Recently, functional magnetic resonance imaging (fMRI) has been applied to the study of memory function in epileptic patients, often with the purpose of obtaining a clinically useful preoperative prediction of postoperative memory deficit (Bellgowan and others 1998; Detre and others 1998; Dupont and others 2000; Billingsley and others 2001; Jokeit and others 2001; Golby and others 2002; Janszky and others 2005). As a population, however, TLE patients who undergo medial temporal surgery are uniquely poised to provide insights into the open question of interacting frontal and temporal influences on memory. One way of determining the effect of a brain region on another is, in fact, to disrupt the activity of the first one with a controlled lesion and to examine the effect of such an intervention on the second region. Following this rationale, examining memory-associated frontal activity before and after a lesion of medial temporal cortex may shed light on the effect of temporal influences on frontal cortex in human subjects.

In the present study, frontal cortex activity during encoding of verbal and nonverbal material was examined with fMRI in TLE patients before and after unilateral medial temporal resection. If activity of medial temporal cortex had a significant effect on the activity of frontal regions, we expected to observe a change in the levels of frontal activity after surgery. Alternatively, memory-associated frontal activity may be independent of temporal influences, which would have translated as no change in frontal activity levels in the postoperative state.

Methods

Subjects

Twenty-four patients with medial TLE undergoing evaluation for epilepsy surgery and 12 healthy subjects were studied preoperatively (see Table 1 for details). Seventeen of the patients agreed to participate in a second postoperative study. All subjects gave informed written consent according to the guidelines of the institutional review board of the Washington University Human Studies Committee. Control subjects were volunteers (students, hospital staff, etc.) from the Washington University community. In all TLE patients, the temporal lobe seizure focus was diagnosed by concordant scalp electroencephalography (EEG) and magnetic resonance imaging (MRI) findings of mesial temporal lobe change or by invasive electrocorticography monitoring using subdural, subtemporal strips. Patients with multiple lesions or multifocal epileptogenic sites were excluded from the analysis. Based on comparisons by unpaired t-test, left and right TLE patient groups did not differ significantly in terms of verbal, performance, or full-scale intelligence quotient (all t22 < 1, see Table 1). No significant difference was observed in age at seizure onset (t22 = 0.13, not significant [n.s.], see Table 1).

Table 1

Subjects

Subject Age/gender Handedness Age at seizure onset MRI EEG VIQ PIQ FSIQ Time to postoperative scan 
Left TLE          
    L_TLE1 30/F 30 L MTS L ant. TL 95 90 93 5.2 
    L_TLE2 46/F 40 L TL CM L TL 97 78 88 — 
    L_TLE3 37/M 2.5 L HcA L TL 93 74 84 8.7 
    L_TLE4 24/F 11 Normal L TL 107 80 95 7.9 
    L_TLE5 35/M L MTS L TL 69 73 70 5.1 
    L_TLE6 30/M 15 L MTS L ant. TL 72 69 69 6.2 
    L_TLE7 35/F L MTS L TL 97 95 96 5.7 
    L_TLE8 32/F 18 L MTS L TL 92 86 89 — 
    L_TLE9 40/F 1.5 L MTS L TL 100 87 94 6.7 
    L_TLE10 20/M Normal L TL 79 84 79 — 
    L_TLE11 41/F 14 L MTS L ant. TL 104 86 106 6.8 
    L_TLE12 35/F Normal L TL 73 75 72 — 
    L_TLE13 36/M 25 Normal L TL 84 86 84 — 
    Mean 33.9 — 13.6 — — 89.4 81.8 86.1 6.5 
Right TLE          
    R_TLE1 37/F 28 Normal R TL 90 81 85 9.1 
    R_TLE2 36/F 24 R HcA R TL 91 80 85 — 
    R_TLE3 19/F Normal R TL 75 72 71 — 
    R_TLE4 25/F 19 Normal R TL 87 84 85 4.7 
    R_TLE5 44/M R ↑ T2 R ant. TL 68 63 63 6.5 
    R_TLE6 23/M 12 Normal Bil. TL 80 92 85 6.1 
    R_TLE7 28/F Inf. R HcA R ant. TL 95 95 95 5.1 
    R_TLE8 33/F 23 Normal R TL 103 86 96 7.3 
    R_TLE9 35/F 29 Normal R TL 101 97 99 6.1 
    R_TLE10 32/F R MTS R TL 84 79 80 4.9 
    R_TLE11 41/M R MTS R TL 83 87 84 4.7 
    Mean 32.1 — 13.5 — — 87.0 83.3 84.4 6.1 
Controls          
    CTL1 25/M — — — n.t. n.t. n.t. — 
    CTL2 23/M — — — n.t. n.t. n.t. — 
    CTL3 23/F — — — n.t. n.t. n.t. — 
    CTL4 21/M — — — n.t. n.t. n.t. — 
    CTL5 21/M — — — n.t. n.t. n.t. — 
    CTL6 22/M — — — n.t. n.t. n.t. — 
    CTL7 21/F — — — n.t. n.t. n.t. — 
    CTL8 38/M — — — n.t. n.t. n.t. — 
    CTL9 34/F — — — n.t. n.t. n.t. — 
    CTL10 39/F — — — n.t. n.t. n.t. — 
    CTL11 35/M — — — n.t. n.t. n.t. — 
    CTL12 40//F — — — n.t. n.t. n.t. — 
    Mean 28.5 — — — — — — — — 
Subject Age/gender Handedness Age at seizure onset MRI EEG VIQ PIQ FSIQ Time to postoperative scan 
Left TLE          
    L_TLE1 30/F 30 L MTS L ant. TL 95 90 93 5.2 
    L_TLE2 46/F 40 L TL CM L TL 97 78 88 — 
    L_TLE3 37/M 2.5 L HcA L TL 93 74 84 8.7 
    L_TLE4 24/F 11 Normal L TL 107 80 95 7.9 
    L_TLE5 35/M L MTS L TL 69 73 70 5.1 
    L_TLE6 30/M 15 L MTS L ant. TL 72 69 69 6.2 
    L_TLE7 35/F L MTS L TL 97 95 96 5.7 
    L_TLE8 32/F 18 L MTS L TL 92 86 89 — 
    L_TLE9 40/F 1.5 L MTS L TL 100 87 94 6.7 
    L_TLE10 20/M Normal L TL 79 84 79 — 
    L_TLE11 41/F 14 L MTS L ant. TL 104 86 106 6.8 
    L_TLE12 35/F Normal L TL 73 75 72 — 
    L_TLE13 36/M 25 Normal L TL 84 86 84 — 
    Mean 33.9 — 13.6 — — 89.4 81.8 86.1 6.5 
Right TLE          
    R_TLE1 37/F 28 Normal R TL 90 81 85 9.1 
    R_TLE2 36/F 24 R HcA R TL 91 80 85 — 
    R_TLE3 19/F Normal R TL 75 72 71 — 
    R_TLE4 25/F 19 Normal R TL 87 84 85 4.7 
    R_TLE5 44/M R ↑ T2 R ant. TL 68 63 63 6.5 
    R_TLE6 23/M 12 Normal Bil. TL 80 92 85 6.1 
    R_TLE7 28/F Inf. R HcA R ant. TL 95 95 95 5.1 
    R_TLE8 33/F 23 Normal R TL 103 86 96 7.3 
    R_TLE9 35/F 29 Normal R TL 101 97 99 6.1 
    R_TLE10 32/F R MTS R TL 84 79 80 4.9 
    R_TLE11 41/M R MTS R TL 83 87 84 4.7 
    Mean 32.1 — 13.5 — — 87.0 83.3 84.4 6.1 
Controls          
    CTL1 25/M — — — n.t. n.t. n.t. — 
    CTL2 23/M — — — n.t. n.t. n.t. — 
    CTL3 23/F — — — n.t. n.t. n.t. — 
    CTL4 21/M — — — n.t. n.t. n.t. — 
    CTL5 21/M — — — n.t. n.t. n.t. — 
    CTL6 22/M — — — n.t. n.t. n.t. — 
    CTL7 21/F — — — n.t. n.t. n.t. — 
    CTL8 38/M — — — n.t. n.t. n.t. — 
    CTL9 34/F — — — n.t. n.t. n.t. — 
    CTL10 39/F — — — n.t. n.t. n.t. — 
    CTL11 35/M — — — n.t. n.t. n.t. — 
    CTL12 40//F — — — n.t. n.t. n.t. — 
    Mean 28.5 — — — — — — — — 

VIQ, verbal intelligence quotient; PIQ, performance intelligence quotient; FSIQ, full-scale intelligence quotient; time to postoperative scan, number of months between surgery and postoperative fMRI scan; M, male; F, female; R, right; L, left; MTS, medial temporal sclerosis; ant., anterior; TL, temporal; CM, cavernous malformation; HcA, hippocampal atrophy; ↑ T2, T2-weighted signal hyperintensity; Bil., bilateral; Inf., infancy (less than 1 year); n.t., not tested.

Tasks and Stimuli

In a scanned word classification task, subjects viewed blocks of words intermixed with blocks of fixation crosshair (baseline) and decided if each word represented a living object. Subjects were instructed to make a right index finger response for “living” words (e.g., RABBIT) and to withhold response for “nonliving” words (e.g., HANGER). A paradigm with a single-keypress response rather than a binary response was used to simplify the task. Stimuli were compiled from an existing set of abstract and concrete words generously provided by John Gabrieli (Demb and others 1995). Twenty-two 12-item lists were constructed, equating lists in terms of length (mean = 6.65) and frequency (mean = 23.25, based on Kucera and Francis 1967). This number of lists allowed for 2 separate sessions of stimuli with no stimulus overlap between pre- and postoperative sessions. Nine lists were used for each subject in each imaging session. Two lists were used as new items in the recognition test following each imaging session. Each list contained equal numbers of living and nonliving words. Word stimuli were presented centrally, in white Geneva font, all caps, on a black background. Twelve words (corresponding to one list) were presented in each task block. In each word trial, the stimulus was presented for 2000 ms, followed by 500 ms of fixation crosshair (white on black background). In each fixation block, a fixation crosshair appeared centrally for the duration (22.5 s) of the block. Three task blocks alternated with 4 fixation blocks in each functional run. Three such runs were acquired for the word classification task in each functional session.

A scanned face classification task had similar design and timing parameters, except that subjects viewed blocks of unfamiliar, nonfamous faces while deciding whether each face presented was male or female. Subjects made a right index finger keypress response to male faces and withheld response to female faces. The stimulus set was constructed from an existing set compiled by Neal Cohen of the University of Illinois, Urbana-Champaign. Twenty-two 12-item lists were constructed. Color images were cropped to eliminate background and clothing, standardized in size, and equated in terms of brightness and contrast. The number of face lists used for each subject mirrored that used in the word classification task (9 for each scanning session plus 2 lists as new items for each recognition task). Each list contained an equal number of male and female faces. Face stimuli were presented centrally, on a black background, and subtended a mean horizontal visual angle of 6° and a mean vertical visual angle of 6°. Fixation blocks during the face task were identical to those used in the word task.

In both word and face tasks, trial onset was time locked to the beginning of each MRI image acquisition. All 3 word runs occurred in sequence before or after all 3 face runs, with the order of task (word vs. face) counterbalanced across subjects. Subjects were unaware that memory would be tested at the end of the scanning session.

An old/new recognition test was administered outside the scanner at the end of the imaging session. Subjects viewed blocks of words and blocks of faces while sitting in front of a computer screen and decided whether each item presented had appeared in the previous imaging session. Subjects made “old” and “new” responses using a Psyscope button box (Cohen and others 1993). Half of the items presented (24 words and 24 faces) had been presented during the scan, whereas the other half were new.

Imaging Procedures

MRI scanning was performed on a 1.5-T MAGNETOM Vision scanner (Siemens, Erlangen, Germany) equipped with a standard circularly polarized head coil. Head movement was minimized with foam cushions and a thermoplastic face mask. Headphones decreased scanner noise and allowed verbal communication with subjects. Subject responses were collected with a hand-held fiber-optic keypress connected to a Psyscope button box (Cohen and others 1993) and a Power Macintosh computer (Apple, Cupertino, CA). Stimuli were projected onto a screen at the head of the magnet bore. Subjects viewed the screen via a mirror attached to the head coil.

A set of high-resolution T1-weighted structural images was acquired for each subject using a sagittal magnetization prepared rapid gradient echo (MP-RAGE) sequence (time repetition [TR] = 9.7 ms, echo time [TE] = 4 ms, flip angle = 10°, time to inversion = 20 ms, time delay [TD] = 200 ms, 1.25 × 1 × 1–mm in-plane resolution). This was followed by a functional session in which images were collected using an asymmetric spin-echo echo-planar sequence sensitive to blood oxygenation level–dependent contrast (TR = 2.5 s, TE = 37 ms, flip angle = 90°, 3.75 × 3.75–mm in-plane resolution). Whole-brain coverage was achieved by acquiring sets of 16 contiguous 8-mm-thick axial images parallel to the anterior–posterior commissural plane, with one 16-slice whole-brain sample occurring every 2.5 s. Each 16-slice image set will be referred to as an “image acquisition” and its position in time as a “time point.” The first 4 image acquisitions in each functional run were discarded to allow stabilization of longitudinal magnetization. Six functional runs were acquired in each functional session (3 during word classification and 3 during face classification), each consisting of 76 time points. Following the functional session, 2 additional sets of T1-weighted MP-RAGE structural images were acquired. Averaging of the 3 sets of structural images achieved greater signal-to-noise than individual structural images, allowing accurate definition of the surgical resection.

To determine the effect of medial temporal lobectomy on frontal activity during memory encoding, TLE patients underwent 2 functional sessions as described above, one occurring before surgery and a second one occurring more than 5 months after surgery (see Table 1 for details). Due to differential outcomes in presurgical evaluation as well as loss to follow-up, only a subset (17 of 24) of the patients who performed in the preoperative fMRI session also performed in the postoperative session (as indicated in Table 1). Analyses targeting surgery effects only included patients who took part in both preoperative and postoperative fMRI sessions. Comparisons between preoperative patients and controls employed all 36 subjects.

Surgery

Patients with medically intractable seizures were candidates for surgery based on EEG and MRI findings. In those with inconsistent data, intracranial, subdural electrodes were placed to localize seizure onset. Surgery was performed in a standard fashion. Patients underwent either a selective amygdalohippocampectomy (n = 4 left TLE, 5 right TLE), in which the temporal horn was entered through the middle temporal gyrus and lateral amygdala and medial temporal structures were resected (with the colliculus marking the approximate plane of the posterior resection), or an anterior temporal lobectomy (n = 4 left TLE, 4 right TLE), tailored to either long-term intracranial monitoring or intraoperative corticography. The extent of medial resection was similar in both groups, although the latter group had a more extensive lateral temporal resection. Resection extent was characterized in each patient by manual tracing using Analyze software (Biomedical Imaging Resource, Mayo Clinic, Rochester, MN) based on high-resolution T1-weighted whole-brain structural images obtained preoperatively and postoperatively (see Fig. 5).

Preliminary Magnetic Resonance Data Analysis

Functional data were initially processed to correct for odd–even slice intensity differences. Differences in timing of acquisition between slices were corrected with sinc interpolation. Image sets from different functional runs were registered to each other as well as to the structural images by using a 12-parameter rigid-body translation and rotation algorithm (Snyder 1996). The linear slope was removed from each functional run in voxelwise fashion (Bandettini and others 1993). Mean signal magnitude for each functional run was adjusted to an arbitrary value of 1000 to normalize global signal for between-subject comparisons. Both structural and functional data were then resampled into 2-mm isotropic voxels, warped into stereotaxic atlas space (Talairach and Tournoux 1988), and smoothed with a Gaussian filter (4-mm full-width half-maximum).

Statistical Analysis

Regional analyses were conducted to characterize and quantify the behavior of a priori–defined brain regions. Regions were defined independently based on the results of a meta-analysis of 39 healthy subjects performing word and face tasks similar to those used in the present experiment (Konishi and others 2001). These exact regions have been used previously (Logan and others 2002; Lustig and others 2003). Quantitative estimates of signal magnitude changes were determined voxelwise as the percentage of difference between task and fixation magnitude levels. Regional estimates were calculated for each subject by averaging across all voxels within a region. Individual regional estimates were entered in statistical analyses based on a random-effects model, using factorial analyses of variance (ANOVAs) followed by post hoc comparisons (t-test). Whole-brain activation maps were also constructed by comparing task blocks to fixation blocks at the voxel level using an implementation of the general linear model (similar to Logan and others 2002). Resulting t statistics were normalized, yielding whole-brain Z-score maps.

Results

Preoperative Behavioral Results

Patients Showed Impaired Performance in the Word- but Not Face-Encoding Task

Controls, preoperative left TLE patients, and preoperative right TLE patients showed relatively high accuracy in the word classification task, although both left and right TLE patients were significantly less accurate than controls (Fig. 1A; unpaired t-test—left TLE: t23 = 2.88, P < 0.01; right TLE: t21 = 2.30, P < 0.05). No significant difference was observed between the 2 patient groups (t22 = 0.90, n.s.). Left TLE patients also made significantly slower responses than controls in the word task (Fig. 1C; t23 = 2.24, P < 0.05), whereas no significant difference was observed between controls and right TLE patients (t21 = 1.23, n.s.) or between patient groups (t21 = 1.01, n.s.). Both patient groups were therefore impaired when performing the word task, with left TLE patients showing suggestion of greater impairment.

Figure 1

Preoperative behavioral results. Accuracy (percentage of correct responses) is plotted for controls, preoperative left TLE patients, and preoperative right TLE patients for word (A) and face (B) classification. Mean behavioral response time during word (C) and face (D) classification. Postscan recognition, as measured by the corrected hit rate (i.e., the difference between the percentage of hits and the percentage of false alarms), for words (E) and faces (F). Asterisks indicate significant differences (unpaired t-test). Error bars indicate the standard error of the mean here and in the following figures.

Figure 1

Preoperative behavioral results. Accuracy (percentage of correct responses) is plotted for controls, preoperative left TLE patients, and preoperative right TLE patients for word (A) and face (B) classification. Mean behavioral response time during word (C) and face (D) classification. Postscan recognition, as measured by the corrected hit rate (i.e., the difference between the percentage of hits and the percentage of false alarms), for words (E) and faces (F). Asterisks indicate significant differences (unpaired t-test). Error bars indicate the standard error of the mean here and in the following figures.

Performance was high for all groups during face classification, with no significant differences observed in accuracy (Fig. 1B; unpaired t-test: all n.s.) or response time (Fig. 1D; unpaired t-test: all n.s.), indicating that patients were performing the face task at the level of controls. Face classification may have been less demanding than word classification: in each group, accuracy was greater and response times were faster during the face task than during the word task (paired t-test: all P < 0.005).

Both Patient Groups Were Significantly Impaired in Recognition of Words and Faces

Previous reports have shown impairment of verbal memory in left TLE patients and impairment of nonverbal or visual memory in right TLE patients (Baxendale and others 1998; Martin and others 2001; for a meta-analytic review, see Lee and others 2002), but exceptions to this general pattern have been reported (Billingsley and others 2001; Lee and others 2002). Recognition of words and faces was tested in the present study after functional imaging. Word recognition was significantly impaired in both left and right TLE patients as compared with controls (Fig. 1E; unpaired t-test—left TLE: t23 = 3.98, P < 0.001; right TLE: t20 = 3.40, P < 0.005), with no significant difference observed between patient groups (t21 = 0.64, n.s.). In the face task, both patient groups showed worse recognition than controls (Fig. 1F; left TLE: t23 = 2.69, P < 0.05; right TLE: t20 = 3.99, P < 0.001), with right TLE patients demonstrating greater impairment than left TLE patients (t21 = 2.31, P < 0.05). In summary, both patient groups were impaired in word and face recognition preoperatively, with right TLE patients showing a significantly greater impairment for faces.

Preoperative fMRI Results

Left TLE Patients Overrecruited Right Frontal Cortex When Processing Verbal Material

Incidental memory-encoding tasks that require semantic-level processing of verbal stimuli often produce robust levels of activity in left frontal cortex in healthy controls (for reviews, see Tulving and others 1994; Buckner and others 1999; Fletcher and Henson 2001). In addition, such activity appears to be correlated with subsequent memory performance (Wagner, Schacter, and others 1998). A word classification task was used here to examine frontal activity in TLE patients during encoding of verbal information. Performance in the word task activated a similar network of regions in controls, preoperative left TLE patients, and preoperative right TLE patients (Fig. 2), including regions in occipital cortex, frontal cortex, parietal cortex, supplementary motor area, and motor cortex. Of note, left prefrontal cortex was robustly engaged in all 3 groups. Qualitative inspection of the activation maps also revealed salient differences among the 3 groups, including a difference in the amount of right prefrontal cortex activity (Fig. 2, arrow). Whereas controls showed little activation of this region, left TLE preoperative patients showed activity levels paralleling those observed contralaterally in left prefrontal cortex, leading to a prominently bilateral pattern of frontal activation similar to those observed in healthy older adults (Reuter-Lorenz and others 2000; Cabeza 2002; Dolcos and others 2002; Logan and others 2002). Right TLE patients also appeared to engage right prefrontal cortex more prominently than controls but to a lesser extent than left TLE patients. In summary, whereas the general pattern of activation during word classification was similar across the 3 groups, frontal activation showed differences, with left TLE patients (and to a lesser extent right TLE patients) disproportionately recruiting right frontal cortex, perhaps as a compensatory response to the increased difficulty exhibited by this patient group in this task.

Figure 2

Brain activity during word classification in controls, preoperative left TLE patients, and preoperative right TLE patients. Group activation maps representing task-related activation are overlaid on group-specific averaged structural images (colored voxels represent Z-scores). Selected axial sections parallel to the anterior–posterior commissural plane are shown. The Z stereotaxic atlas coordinate is indicated for each horizontal section (Talairach and Tournoux 1988). In all 3 groups, performance of the word classification task elicited activity in a range of brain regions, including visual cortex, left and right prefrontal cortex, supplementary motor area, motor cortex, and subcortical regions. Notably, the amount of right frontal recruitment was markedly increased in preoperative left TLE patients (arrow) as compared with controls and preoperative right TLE patients.

Figure 2

Brain activity during word classification in controls, preoperative left TLE patients, and preoperative right TLE patients. Group activation maps representing task-related activation are overlaid on group-specific averaged structural images (colored voxels represent Z-scores). Selected axial sections parallel to the anterior–posterior commissural plane are shown. The Z stereotaxic atlas coordinate is indicated for each horizontal section (Talairach and Tournoux 1988). In all 3 groups, performance of the word classification task elicited activity in a range of brain regions, including visual cortex, left and right prefrontal cortex, supplementary motor area, motor cortex, and subcortical regions. Notably, the amount of right frontal recruitment was markedly increased in preoperative left TLE patients (arrow) as compared with controls and preoperative right TLE patients.

Recruitment of Frontal Cortex Changed Dynamically Based On Task Demands

In contrast with verbal tasks, processing of face stimuli has been shown to lead to substantial activation of right frontal regions, in addition to concurrent recruitment of left frontal cortex (Wagner, Poldrack, and others 1998; McDermott and others 1999). A face classification task was used in this study to investigate frontal activity patterns associated with encoding of nonverbal information in TLE patients. Activation patterns during the face task were very similar across the 3 groups (Fig. 3), with significant task-correlated activity observed in a range of regions, including occipital cortex, left and right prefrontal cortex, parietal cortex, supplementary motor area, and motor cortex.

Figure 3

Brain activity during face classification in controls, preoperative left TLE patients, and preoperative right TLE patients. The format is similar to Figure 2. Performance of the face classification task led to significant activity in occipital cortex, prefrontal cortex, parietal cortex, supplementary motor area, and motor cortex in all groups. Frontal activity was significantly lateralized in all groups.

Figure 3

Brain activity during face classification in controls, preoperative left TLE patients, and preoperative right TLE patients. The format is similar to Figure 2. Performance of the face classification task led to significant activity in occipital cortex, prefrontal cortex, parietal cortex, supplementary motor area, and motor cortex in all groups. Frontal activity was significantly lateralized in all groups.

Qualitative inspection of prefrontal activity patterns showed right-sided lateralization in all 3 groups. Notably, preoperative left TLE patients showed lateralization during face classification, in contrast with the pronounced bilateral recruitment of frontal cortex shown during word classification. This suggests that a bilateral frontal recruitment pattern may be engaged dynamically depending on the specific demands of the task.

Regional Analyses Confirmed Atypical Bilateral Frontal Recruitment in Left TLE Patients during Verbal Encoding

Qualitative observations based on activation maps were explored quantitatively with a region of interest analysis using independently defined regions (Konishi and others 2001). In general, all statistical analyses described used factorial ANOVAs followed by post hoc comparisons (paired or unpaired t-test). Amplitude estimates were determined for 2 regions near left and right inferior frontal gyrus for both word classification (Fig. 4A) and face classification (Fig. 4B). When comparing preoperative left TLE patients with controls in the word classification task, a factorial ANOVA revealed a significant effect of region (Fig. 4A; F1,1,46 = 6.46, P < 0.05), as well as a nonsignificant but strong trend for an interaction of region by group (F1,1,46 = 3.84, P = 0.06). Post hoc tests confirmed this, showing significantly greater activity in right frontal cortex for left TLE patients than for controls (unpaired t-test: t23 = 2.40, P < 0.05), whereas no significant difference was observed in the activity of left frontal cortex (t23 = 0.48, n.s.). This confirmed the qualitative observations, indicating that left TLE patients disproportionately recruited right frontal cortex compared with controls. Similar results were obtained when comparing left TLE with right TLE patients in the same task, with a significant effect of region (F1,1,44 = 6.26, P < 0.05) and a nonsignificant trend for an interaction of region by group (F1,1,44 = 3.50, P = 0.07). Left TLE patients showed significantly greater activity in the right frontal region than right TLE patients (unpaired t-test: t22 = 2.13, P < 0.05), whereas activity in the left frontal region was similar across the 2 groups (t22 = 0.74, n.s.). Controls and right TLE patients showed similar lateralization, with a significant effect of region observed (F1,1,42 = 16.74, P < 0.001), whereas post hoc tests showed no differences between the 2 groups in either frontal region (t21: both n.s.). The regional analysis thus showed that when performing the verbal task, left TLE patients disproportionately recruited right prefrontal cortex as compared with both controls and right TLE patients.

Figure 4

Regional analysis of frontal cortex activity in controls, preoperative left TLE patients, and preoperative right TLE patients. Regions of interest are shown in red (inset). Bar graphs represent percent activity change from baseline for left (dark gray) and right (light gray) frontal regions. (A) In the word classification task, left TLE patients showed disproportionate recruitment of right frontal cortex compared with both controls and right TLE patients (asterisks represent significant differences as measured by unpaired t-test). (B) In contrast, no significant differences were observed between groups during face classification.

Figure 4

Regional analysis of frontal cortex activity in controls, preoperative left TLE patients, and preoperative right TLE patients. Regions of interest are shown in red (inset). Bar graphs represent percent activity change from baseline for left (dark gray) and right (light gray) frontal regions. (A) In the word classification task, left TLE patients showed disproportionate recruitment of right frontal cortex compared with both controls and right TLE patients (asterisks represent significant differences as measured by unpaired t-test). (B) In contrast, no significant differences were observed between groups during face classification.

A similar regional analysis was performed on individual signal amplitude estimates for the face classification task (Fig. 4B). Here both left TLE patients and controls showed significant lateralization, as evidenced by a significant effect of region (F1,1,46 = 8.70, P < 0.01), whereas no other effects were significant. Post hoc unpaired t-tests were consistent with this, showing no significant differences between the 2 groups in both left (t23 = 0.78, n.s.) and right frontal cortex (t23 = 0.20, n.s.). Thus, in contrast to the word classification task, face classification did not elicit a significant frontal overrecruitment in left TLE patients. A comparison of right TLE patients with controls similarly showed no significant effects, although a strong trend for an effect of region was observed (ANOVA: F1,1,46 = 3.67, P = 0.06). No significant differences between right TLE patients and controls were revealed by post hoc tests in either left or right frontal cortex (t20: both n.s.), indicating that, as in the case of word classification, right TLE activation patterns were similar to those of controls. Comparison of left TLE patients with right TLE patients did not yield any significant effects (F1,1,42: all n.s.), and no significant differences were observed between patient groups post hoc (t21: both n.s.). In summary, the regional analysis showed high similarity of frontal activation between controls, left TLE, and right TLE patients during face classification, with lateralized recruitment in both patient groups.

After the preoperative imaging session, a subset of 17 patients underwent medial temporal lobe surgery for the purpose of seizure control (see Table 1). High-resolution T1-weighted structural images acquired both pre- and postoperatively allowed delineation of the extent of the surgical resection in each patient by manual tracing (Fig. 5). The anterior portion of the medial temporal lobe was resected in all patients, generally including the anterior hippocampus with surrounding entorhinal and perirhinal cortex, the parahippocampal gyrus, and the amygdala. In some patients the surgery was more laterally extensive, also involving the inferior and middle temporal gyri.

Figure 5

Extent of the surgical resection in left (A) and right (B) TLE patients. Resected areas are indicated as colored pixels overlaid on coronal sections of group-specific averaged structural images. Lighter colors indicate pixels resected in a greater number of patients. The Talairach and Tournoux (1988) Y coordinate is shown for each section. In all patients, the resection included the anterior portion of the medial temporal lobe; in some patients, surgery was more extensive, including portions of lateral temporal lobe.

Figure 5

Extent of the surgical resection in left (A) and right (B) TLE patients. Resected areas are indicated as colored pixels overlaid on coronal sections of group-specific averaged structural images. Lighter colors indicate pixels resected in a greater number of patients. The Talairach and Tournoux (1988) Y coordinate is shown for each section. In all patients, the resection included the anterior portion of the medial temporal lobe; in some patients, surgery was more extensive, including portions of lateral temporal lobe.

A postoperative imaging session was administered more than 5 months after surgery, allowing the direct, within-subject comparison of preoperative and postoperative activation patterns. The same scanning procedure administered in the preoperative stage was used but with novel stimuli. Critically, a main question of interest was whether frontal activity associated with memory encoding would change as a result of medial temporal surgery. The use of 2 tasks leading to opposite lateralization patterns in controls maximized the detection of surgery-related changes affecting either cerebral hemisphere.

Behavioral Effects of Surgery

Surgery Had Little Effect on Behavioral Performance

Medial temporal resection had no significant effect on the behavioral performance of either patient group in either task (Fig. 6). Accuracy during word classification (Fig. 6A) was unchanged as a result of surgery in both left TLE (paired t-test: t7 = 1.12, n.s.) and right TLE patients (t8 = 1.09, n.s.). Similarly, responses during the face classification task were as accurate after surgery as they were before surgery in both groups (Fig. 6B; paired t-test: both n.s.). Behavioral response time was also unaffected by medial temporal surgery. Response times were not significantly different before and after surgery in either patient group for both word classification (Fig. 6C; paired t-test: both n.s.) and face classification (Fig. 6D; paired t-test: both n.s.). Postoperative recognition performance was similar to preoperative levels. Both left and right TLE patients showed no difference between preoperative and postoperative word recognition (Fig. 6E; paired t-test: both n.s.) or face recognition (Fig. 6F; paired t-test: both n.s.), although the latter result may have been due to a floor effect as preoperative face recognition was near chance in both patient groups.

Figure 6

Effect of medial temporal lobe surgery on behavioral performance. Preoperative (dark gray) and postoperative (light gray) values are shown for both left and right TLE patient groups. Word classification (A) and face classification (B) accuracy was unaffected by surgery in both patient groups. Similarly, behavioral response time during word (C) and face (D) classification did not change significantly as a result of surgery. Postscan recognition memory for words (E) and faces (F) was also similarly unaffected.

Figure 6

Effect of medial temporal lobe surgery on behavioral performance. Preoperative (dark gray) and postoperative (light gray) values are shown for both left and right TLE patient groups. Word classification (A) and face classification (B) accuracy was unaffected by surgery in both patient groups. Similarly, behavioral response time during word (C) and face (D) classification did not change significantly as a result of surgery. Postscan recognition memory for words (E) and faces (F) was also similarly unaffected.

In summary, medial temporal surgery did not have a significant effect on behavior, with performance in both word and face classification and postscan recognition largely unchanged from preoperative levels in both patient groups.

Postoperative fMRI Results

Recruitment of Frontal Cortex in Left TLE Patients Became Lateralized After Surgery

Effects of surgery on frontal activity were explored both with exploratory activation maps and by direct quantification of regional activity using a priori regions. Importantly, analysis included only patients for whom both preoperative and postoperative data were obtained, allowing within-subject comparisons.

Qualitative assessment of activation maps showed increased lateralization of frontal activity after surgery in left TLE patients (Fig. 7A). A regional analysis confirmed this observation, with a repeated measures ANOVA showing a significant interaction of surgery by region (Fig. 7B; F1,7 = 6.47, P < 0.05). Post hoc tests revealed that activity in right frontal cortex was reduced after surgery (t7 = 2.52, P < 0.05). Although this significant effect was modest, it suggests that the preoperative bilateral recruitment pattern observed in left TLE patients may be reduced by surgery, perhaps because of reduced demands due to a postoperative seizure-free status. A similar analysis performed in the same patient group for face classification yielded no significant effects of surgery, of region, or of their interaction (F1,7: all n.s.), consistent with the near-normal preoperative pattern.

Figure 7

Effect of surgery on frontal activity associated with memory encoding. Whole-brain activation maps (A, C) and regional analyses results (B, D) are shown for both left and right TLE patients. Colored voxels in activation maps indicate areas of significant task-correlated activity changes compared with baseline (Z-score). Group-specific average axial sections are used as an underlay. Region of interest analyses show activity levels in left (dark gray) and right (light gray) frontal regions as percent change from baseline. Asterisks indicate significant differences as determined by paired t-test. Surgery resulted in a more lateralized pattern of frontal activity in left TLE patients in the verbal task. Conversely, right TLE patients were generally unaffected by surgery in both tasks.

Figure 7

Effect of surgery on frontal activity associated with memory encoding. Whole-brain activation maps (A, C) and regional analyses results (B, D) are shown for both left and right TLE patients. Colored voxels in activation maps indicate areas of significant task-correlated activity changes compared with baseline (Z-score). Group-specific average axial sections are used as an underlay. Region of interest analyses show activity levels in left (dark gray) and right (light gray) frontal regions as percent change from baseline. Asterisks indicate significant differences as determined by paired t-test. Surgery resulted in a more lateralized pattern of frontal activity in left TLE patients in the verbal task. Conversely, right TLE patients were generally unaffected by surgery in both tasks.

Frontal Recruitment in Right TLE Patients Was Largely Unaffected by Surgery

Preoperative frontal activation patterns in right TLE patients were generally conserved after surgery, with little change in activity levels observed qualitatively (Fig. 7C). A regional analysis showed a significant effect of region during word classification (Fig. 7D; ANOVA: F1,8 = 6.39, P < 0.05), consistent with the significant left-sided lateralization observed in this patient group both before and after surgery when processing verbal material. Post hoc paired t-tests showed a trend for significant lateralization before surgery (t8 = 2.23, P = 0.057) and a significant difference between left and right frontal cortex activity postoperatively (t8 = 2.78, P < 0.05). No significant effect of surgery was observed (F1,8 = 0.07, n.s.), but the interaction of surgery and region was significant (F1,8 = 7.31, P < 0.05), suggesting increased lateralization after surgery, perhaps paralleling the effects seen in left TLE patients.

During face classification, frontal activity did not show significant effects of surgery or region (F1,8: both n.s.). A small but significant interaction of surgery and region was observed (F1,8 = 7.78, P < 0.05), but post hoc paired t-tests did not reveal significant differences (t8: all n.s.). Taken in the context of their largely normal recruitment preoperatively, we do not interpret this interaction, but present it as possible evidence for a modest effect that is not consistently observed in our sample. In summary, medial temporal surgery appeared to have a small effect on frontal activity in the right TLE patient group. Although the general patterns of activation observed before surgery were essentially maintained after surgery, the small, but significant, effects that were observed mirrored the effects seen in the left TLE patient group, to a lesser degree, indicating that medial temporal surgery generally led to greater lateralization of frontal activity associated with encoding of verbal information.

Discussion

Preoperative TLE patients showed robust frontal activation during verbal and nonverbal tasks known to encourage memory encoding in controls. Activity patterns varied depending on the lateralization of seizure focus and the type of task performed. When processing verbal material, left TLE patients demonstrated activation of left frontal cortex at levels similar to those observed in healthy controls and right TLE patients. Unexpectedly, left TLE patients showed “additional, robust recruitment of right frontal cortex” beyond the activation levels observed in the same region in controls or right TLE patients. Performance of the word task thus produced a bilateral frontal activation pattern in preoperative left TLE patients while leading to a left-lateralized pattern in preoperative right TLE patients and controls. Bilateral frontal activity appeared to be task specific: during face classification, frontal activity in left TLE patients was right lateralized, similar to the pattern observed in right TLE patients and controls. Activation of frontal cortex during memory encoding thus varied in preoperative TLE patients depending on the task performed and the side of seizure focus. Medial temporal surgery reduced the recruitment of right frontal cortex in left TLE patients during verbal processing, leading to a more lateralized activation pattern typical of control subjects, although with the caveat that this occurred without any significant improvement in postoperative behavioral performance.

The bilateral frontal activity pattern observed here in preoperative left TLE patients is reminiscent of the patterns observed after brain injury or degeneration, such as after stroke (for a review, see Rijntjes and Weiller 2002), brain trauma (McAllister and others 1999; Christodoulou and others 2001), or with dementia of the Alzheimer type (Becker and others 1996; Bookheimer and others 2000; Grady and others 2003). Such patterns have also been observed in healthy older adults who, compared with young controls, exhibit increased activation in prefrontal regions broadly colocalized with the ones examined here (Cabeza and others 1997; Grady and Craik 2000; Reuter-Lorenz and others 2000; Cabeza 2002; Logan and others 2002; Reuter-Lorenz 2002; Persson and others 2005). Logan and others (2002), for instance, showed that older adults performing a word classification task similar to the one used in the present study showed significantly increased activation of right prefrontal cortex, with a bilateral activation pattern similar to the one exhibited here by preoperative left TLE patients.

The significance of such increased levels of frontal activity remains the subject of debate, with speculation that they may represent a response to a pathologic state (Grady and Craik 2000; Cabeza 2002; Reuter-Lorenz 2002; Buckner 2004; Persson and others 2005). We outline a heuristic framework by which the response of a cortical region to a pathologic state can be categorized as functional, dysfunctional, or nonfunctional. In a control condition, a network of regions (referred to as the control network) is activated for the performance of a given task. This represents a normal or healthy state in which activation of the control network is sufficient to achieve the task goal. In this condition, an additional network of regions examined below remains inactive. In the presence of a pathologic state, the response of the cortical system changes with activation of the additional network. According to the functional view, the activation of the additional network represents a functionally adaptive response. In this scenario, the pathologic state adversely affects the control network so that it is unable to fulfill the task goal alone. The additional network is activated to aid in the performance of the task, in “compensatory” guise. According to an alternative view, the response to the pathologic state may be dysfunctional. Here the pathologic state leads to the activation of the additional network, but this activity interferes with the activity of the control network. In this view, the response of the system is maladaptive or dysfunctional, interfering with “task execution,” competing with the control network. Alternatively, the response to a pathologic state may be purely nonfunctional, with activation of the additional network of cortical regions neither aiding nor interfering with task execution. By such a heuristic, response of the cortical system to a pathologic state, evidenced by activation of an additional cortical network, is classified as functional, dysfunctional, or nonfunctional, based on whether such response aids, interferes with, or fails to affect task execution (see Grady and others 1994; Buckner and others 1996; Cabeza and others 1997, 2002; Reuter-Lorenz and others 2000; Konishi and others 2001; Logan and others 2002; Persson and others 2005).

Categorizing the activation response as functional, dysfunctional, or nonfunctional depends on the effect that such activity has on task execution. Here task execution refers to the set of computations performed in the healthy state by the control network. Such a heuristic is notably agnostic to what computations are performed by the network. In the case of prefrontal cortex, specifically, the nature of processing performed by these regions remains the subject of debate (Miller and D'Esposito 2005; Thompson-Schill and others 2005). The experimenter must, therefore, rely on indirect measures of performance (such as success on behavioral tasks) to assess the effectiveness of the cortical network. This is best exemplified by a recent study by Persson and others (2005), in which older adults were divided into groups based on their longitudinal performance on a battery of behavioral tests, and behavioral performance decline was directly compared with the amount of additional activity observed in frontal regions. In that study, older adults with the greatest decline in behavioral performance had the greatest increase in right prefrontal activity, when compared with older adults with stable cognitive function (Persson and others 2005).

How does such heuristics apply to the current experiment? We presented evidence of a patient group with known medial temporal pathology who showed impairment in task performance compared with controls and simultaneously demonstrated increased activity in right frontal cortex. At first glance, this would appear to be consistent with the dysfunctional view, suggesting that the additional frontal activity was detrimental to task performance. This is, however, refuted by examining the postoperative data: after surgery, left TLE patients showed no change in behavioral performance, yet the bilateral frontal activation pattern became more lateralized. It would also be unclear why, if such frontal activity is directly causing the behavioral impairment, the right TLE patients showed no significant behavioral benefit from having a more lateralized activation pattern. Instead, when taken together, the data presented here are more consistent with either the functional or nonfunctional views outlined above, with important caveats.

According to a functional view, the right frontal activity exhibited preoperatively by left TLE patients was compensatory. Postoperatively, the same group of patients may have required less processing resources, reducing the need for compensatory right frontal activity. This explanation also needs several qualifications. Given that left TLE preoperative behavioral performance was worse than control, this was clearly not a complete “compensation.” Furthermore, it remains unclear why right TLE patients, who showed no significant difference in task performance, did not exhibit the same level of bilateral recruitment, although it could be argued that they required less compensatory processing, perhaps secondary to their different clinical history. An alternative explanation is that the additional right frontal activity observed preoperatively in left TLE patients had no effect on task performance. Notably, this does not mean that the abnormal frontal activity pattern observed bore no relation to the task performed. Indeed, a clear task specificity was evident in this data set, with the bilateral activation pattern absent when the same group of patients performed a face classification task. Rather, the nonfunctional view would hold that although the aberrant frontal activation pattern is likely a response to a pathological state, it does not provide a benefit to task performance. The fact that such pattern diminished postoperatively is consistent with this view, suggesting that the cause of the aberrant response was reduced or eliminated by surgery.

The nature of the processing performed in right prefrontal cortex in the setting of the bilateral activation pattern remains unknown. According to a functional view, the processing provided is compensatory. Given the task specificity of our results, we can speculate that the right prefrontal cortex may have, for instance, contributed additional verbal information processing of the kind normally performed by left prefrontal cortex or it may have represented allocation of additional attention resources. According to an alternative, nonfunctional view, right prefrontal cortex activity represented a response to a pathologic state specific to the left TLE group, but that did not affect task performance. The nature of this additional activity may have been directly related to the type of processing required by the given task and similar in nature to that performed by left prefrontal cortex but failed to provide a performance benefit. It may also have been utterly different in nature and functionally useless. Indeed, the fact that the same region of right prefrontal cortex was significantly and comparably active across patient groups during performance of the face task, in a lateralized manner identical to that observed in healthy individuals, suggests that the type of computations performed by this brain region may be unique and complementary or unrelated to its left hemisphere counterpart. This is consistent, for instance, with the findings of Persson and others (2005), who argued that the additional right frontal activity observed in older adults with behavioral cognitive decline served as a marker of frontal dysfunction or pathology. The increased activity exhibited here by left TLE patients may itself represent a marker of accumulated disease, in the sense that, on average, our patients had a 20-year history of epilepsy at the time they participated in our study. Finally, although both the word and the face classification tasks used here are known to encourage memory encoding while leading to significant levels of prefrontal activity (Kapur and others 1994; Demb and others 1995; Kelley and others 1998; Wagner, Poldrack, and others 1998; McDermott and others 1999), it is not a given that the activity observed in frontal cortex in this data set was directly related to encoding as subsequent memory effects were not tested formally.

Although modest, the return to a more lateralized activation pattern after surgery indicates that bilateral frontal activity patterns do not reflect an irrevocable shift in the allocation of cortical resources. Rather, a more lateralized pattern can be reestablished postoperatively. Yet the mechanism by which medial temporal surgery resulted in a change in frontal activation level remains unknown. The fact that the effect of surgery was seen in contralateral frontal cortex was unexpected and does not readily support an explanation based on anatomic connectivity. Rather, it suggests an indirect effect, possibly attributable to reduced levels of seizure activity, a reduced reliance on antiepileptic drugs for seizure control, and an overall improved baseline cognitive function.

The results of this study have broad implications. First, they suggest that TLE can be associated with abnormal activation of frontal cortex, potentially as a compensatory recruitment of additional frontal resources in response to demanding task conditions (see also Dupont and others 2000; Billingsley and others 2001; Cabeza and others 2002; Buckner 2004) or alternatively as a nonfunctional response to a pathologic state. This extends the range of patient populations known to exhibit overrecruitment of frontal cortex. Although medial TLE has been associated with effects on cortical metabolism extending into frontal regions (Jokeit and others 1997), the dynamic nature of frontal activity observed in this study suggests that bilateral frontal recruitment patterns represent a response that is transiently engaged based on task conditions. If functionally related, the bilateral frontal recruitment observed in older adults and patients with dementia of the Alzheimer type may represent a similar compensatory effort. Alternatively, it may represent a broadly observed nonfunctional response to brain injury or pathology that is transient and reversible but does not contribute to task performance. Second, abnormal overrecruitment of frontal regions was exhibited in specific tasks and by specific patient groups and may have been related to difficulty in task performance. This observation should serve as a qualifier for studies that seek to utilize lateralization of frontal function as a predictor of individual postoperative memory deficits (Ojemann and Kelley 2002) because different tasks and different patient groups appear to show significantly different patterns of frontal activity.

We thank Neal J. Cohen of the Beckman Institute at the University of Illinois for helpful discussions and for graciously providing the visual stimuli used in the face task. Conflict of Interest: None declared.

References

Bachevalier
J
Mishkin
M
Visual recognition impairment follows ventromedial but not dorsolateral prefrontal lesions in monkeys
Behav Brain Res
 , 
1986
, vol. 
20
 (pg. 
249
-
261
)
Bandettini
PA
Jesmanowicz
A
Wong
EC
Hyde
JS
Processing strategies for time-course data sets in functional MRI of the human brain
Magn Reson Med
 , 
1993
, vol. 
30
 (pg. 
161
-
173
)
Barbas
H
Anatomic organization of basoventral and mediodorsal visual recipient prefrontal regions in the rhesus monkey
J Comp Neurol
 , 
1988
, vol. 
276
 (pg. 
313
-
342
)
Barbas
H
Organization of cortical afferent input to orbitofrontal areas in the rhesus monkey
Neuroscience
 , 
1993
, vol. 
56
 (pg. 
841
-
864
)
Barbas
H
Connections underlying the synthesis of cognition, memory, and emotion in primate prefrontal cortices
Brain Res Bull
 , 
2000
, vol. 
52
 (pg. 
319
-
330
)
Bauer
RH
Fuster
JM
Delayed-matching and delayed-response deficit from cooling dorsolateral prefrontal cortex in monkeys
J Comp Physiol Psychol
 , 
1976
, vol. 
90
 (pg. 
293
-
302
)
Baxendale
SA
Van Paesschen
W
Thompson
PJ
Duncan
JS
Harkness
WF
Shorvon
SD
Hippocampal cell loss and gliosis: relationship to preoperative and postoperative memory function
Neuropsychiatry Neuropsychol Behav Neurol
 , 
1998
, vol. 
11
 (pg. 
12
-
21
)
Becker
JT
Mintun
MA
Aleva
K
Wiseman
MB
Nichols
T
DeKosky
ST
Compensatory reallocation of brain resources supporting verbal episodic memory in Alzheimer's disease
Neurology
 , 
1996
, vol. 
46
 (pg. 
692
-
700
)
Bellgowan
PS
Binder
JR
Swanson
SJ
Hammeke
TA
Springer
JA
Frost
JA
Mueller
WM
Morris
GL
Side of seizure focus predicts left medial temporal lobe activation during verbal encoding
Neurology
 , 
1998
, vol. 
51
 (pg. 
479
-
484
)
Billingsley
RL
McAndrews
MP
Crawley
AP
Mikulis
DJ
Functional MRI of phonological and semantic processing in temporal lobe epilepsy
Brain
 , 
2001
, vol. 
124
 (pg. 
1218
-
1227
)
Bookheimer
SY
Strojwas
MH
Cohen
MS
Saunders
AM
Pericak-Vance
MA
Mazziotta
JC
Small
GW
Patterns of brain activation in people at risk for Alzheimer's disease
N Engl J Med
 , 
2000
, vol. 
343
 (pg. 
450
-
456
)
Buckner
RL
Memory and executive function in aging and AD: multiple factors that cause decline and reserve factors that compensate
Neuron
 , 
2004
, vol. 
44
 (pg. 
195
-
208
)
Buckner
RL
Corbetta
M
Schatz
J
Raichle
ME
Petersen
SE
Preserved speech abilities and compensation following prefrontal damage
Proc Natl Acad Sci USA
 , 
1996
, vol. 
93
 (pg. 
1249
-
1253
)
Buckner
RL
Kelley
WM
Petersen
SE
Frontal cortex contributes to human memory formation
Nat Neurosci
 , 
1999
, vol. 
2
 (pg. 
311
-
314
)
Buckner
RL
Wheeler
ME
The cognitive neuroscience of remembering
Nat Rev Neurosci
 , 
2001
, vol. 
2
 (pg. 
624
-
634
)
Cabeza
R
Hemispheric asymmetry reduction in older adults: the HAROLD model
Psychol Aging
 , 
2002
, vol. 
17
 (pg. 
85
-
100
)
Cabeza
R
Anderson
ND
Locantore
JK
McIntosh
AR
Aging gracefully: compensatory brain activity in high-performing older adults
Neuroimage
 , 
2002
, vol. 
17
 (pg. 
1394
-
1402
)
Cabeza
R
Grady
CL
Nyberg
L
McIntosh
AR
Tulving
E
Kapur
S
Jennings
JM
Houle
S
Craik
FI
Age-related differences in neural activity during memory encoding and retrieval: a positron emission tomography study
J Neurosci
 , 
1997
, vol. 
17
 (pg. 
391
-
400
)
Christodoulou
C
DeLuca
J
Ricker
JH
Madigan
NK
Bly
BM
Lange
G
Kalnin
AJ
Liu
WC
Steffener
J
Diamond
BJ
Ni
AC
Functional magnetic resonance imaging of working memory impairment after traumatic brain injury
J Neurol Neurosurg Psychiatry
 , 
2001
, vol. 
71
 (pg. 
161
-
168
)
Cohen
JD
MacWhinney
RC
Flatt
M
Provost
J
An interactive graphic system for designing and controlling experiments in the psychology laboratory using Macintosh computers
Behav Res Methods Instrum Comput
 , 
1993
, vol. 
25
 (pg. 
257
-
271
)
Demb
JB
Desmond
JE
Wagner
AD
Vaidya
CJ
Glover
GH
Gabrieli
JD
Semantic encoding and retrieval in the left inferior prefrontal cortex: a functional MRI study of task difficulty and process specificity
J Neurosci
 , 
1995
, vol. 
15
 (pg. 
5870
-
5878
)
Detre
JA
Maccotta
L
King
D
Alsop
DC
Glosser
G
D'Esposito
M
Zarahn
E
Aguirre
GK
French
JA
Functional MRI lateralization of memory in temporal lobe epilepsy
Neurology
 , 
1998
, vol. 
50
 (pg. 
926
-
932
)
Dolan
RJ
Fletcher
PC
Dissociating prefrontal and hippocampal function in episodic memory encoding
Nature
 , 
1997
, vol. 
388
 (pg. 
582
-
585
)
Dolcos
F
Rice
HJ
Cabeza
R
Hemispheric asymmetry and aging: right hemisphere decline or asymmetry reduction
Neurosci Biobehav Rev
 , 
2002
, vol. 
26
 (pg. 
819
-
825
)
Dupont
S
Van de Moortele
PF
Samson
S
Hasboun
D
Poline
JB
Adam
C
Lehericy
S
Le Bihan
D
Samson
Y
Baulac
M
Episodic memory in left temporal lobe epilepsy: a functional MRI study
Brain
 , 
2000
, vol. 
123
 (pg. 
1722
-
1732
)
Easton
A
Parker
A
Gaffan
D
Crossed unilateral lesions of medial forebrain bundle and either inferior temporal or frontal cortex impair object recognition memory in rhesus monkeys
Behav Brain Res
 , 
2001
, vol. 
121
 (pg. 
1
-
10
)
Fletcher
PC
Henson
RNA
Frontal lobes and human memory
Brain
 , 
2001
, vol. 
124
 (pg. 
849
-
881
)
Frankland
PW
Bontempi
B
The organization of recent and remote memories
Nat Rev Neurosci
 , 
2005
, vol. 
6
 (pg. 
119
-
130
)
Fuster
JM
Bauer
RH
Jervey
JP
Functional interactions between inferotemporal and prefrontal cortex in a cognitive task
Brain Res
 , 
1985
, vol. 
330
 (pg. 
299
-
307
)
Gaffan
D
Easton
A
Parker
A
Interaction of inferior temporal cortex with frontal cortex and basal forebrain: double dissociation in strategy implementation and associative learning
J Neurosci
 , 
2002
, vol. 
22
 (pg. 
7288
-
7296
)
Golby
AJ
Poldrack
RA
Illes
J
Chen
D
Desmond
JE
Gabrieli
JDE
Memory lateralization in medial temporal lobe epilepsy assessed by functional MRI
Epilepsia
 , 
2002
, vol. 
43
 (pg. 
855
-
863
)
Grady
CL
Craik
FI
Changes in memory processing with age
Curr Opin Neurobiol
 , 
2000
, vol. 
10
 (pg. 
224
-
231
)
Grady
CL
Maisog
JM
Horwitz
B
Ungerleider
LG
Mentis
MJ
Salerno
JA
Pietrini
P
Wagner
E
Haxby
JV
Age-related changes in cortical blood flow activation during visual processing of faces and location
J Neurosci
 , 
1994
, vol. 
14
 (pg. 
1450
-
1462
)
Grady
CL
McIntosh
AR
Beig
S
Keightley
ML
Burian
H
Black
SE
Evidence from functional neuroimaging of a compensatory prefrontal network in Alzheimer's disease
J Neurosci
 , 
2003
, vol. 
23
 (pg. 
986
-
993
)
Incisa Della Rocchetta
A
Milner
B
Strategic search and retrieval inhibition: the role of the frontal lobes
Neuropsychologia
 , 
1993
, vol. 
31
 (pg. 
503
-
524
)
Janowsky
JS
Shimamura
AP
Kritchevsky
M
Squire
LR
Cognitive impairment following frontal lobe damage and its relevance to human amnesia
Behav Neurosci
 , 
1989
, vol. 
103
 (pg. 
548
-
560
)
Janszky
J
Jokeit
H
Kontopoulou
K
Mertens
M
Ebner
A
Pohlmann-Eden
B
Woermann
FG
Functional MRI predicts memory performance after right mesiotemporal epilepsy surgery
Epilepsia
 , 
2005
, vol. 
46
 (pg. 
244
-
250
)
Jokeit
H
Okujava
M
Woermann
FG
Memory fMRI lateralizes temporal lobe epilepsy
Neurology
 , 
2001
, vol. 
57
 (pg. 
1786
-
1793
)
Jokeit
H
Seitz
RJ
Markowitsch
HJ
Neumann
N
Witte
OW
Ebner
A
Prefrontal asymmetric interictal glucose hypometabolism and cognitive impairment in patients with temporal lobe epilepsy
Brain
 , 
1997
, vol. 
120
 (pg. 
2283
-
2294
)
Kapur
S
Craik
FI
Tulving
E
Wilson
AA
Houle
S
Brown
GM
Neuroanatomical correlates of encoding in episodic memory: levels of processing effect
Proc Natl Acad Sci USA
 , 
1994
, vol. 
91
 (pg. 
2008
-
2011
)
Kelley
WM
Miezin
FM
McDermott
KB
Buckner
RL
Raichle
ME
Cohen
NJ
Ollinger
JM
Akbudak
E
Conturo
TE
Snyder
AZ
Petersen
SE
Hemispheric specialization in human dorsal frontal cortex and medial temporal lobe for verbal and nonverbal memory encoding
Neuron
 , 
1998
, vol. 
20
 (pg. 
927
-
936
)
Konishi
S
Donaldson
DI
Buckner
RL
Transient activation during block transition
Neuroimage
 , 
2001
, vol. 
13
 (pg. 
364
-
374
)
Kucera
H
Francis
WN
The computational analysis of present-day American English
 , 
1967
RI: Brown University Press
Providence
Kuypers
HGJM
Szwarcbart
MK
Mishkin
M
Rosvold
HE
Occipitotemporal corticocortical connections in the rhesus monkey
Exp Neurol
 , 
1965
, vol. 
11
 (pg. 
245
-
262
)
Lavenex
P
Suzuki
WA
Amaral
DG
Perirhinal and parahippocampal cortices of the macaque monkey: projections to the neocortex
J Comp Neurol
 , 
2002
, vol. 
447
 (pg. 
394
-
420
)
Lee
TMC
Yip
JTH
Jones-Gotman
M
Memory deficits after resection from left or right anterior temporal lobe in humans: a meta-analytic review
Epilepsia
 , 
2002
, vol. 
43
 (pg. 
283
-
291
)
Levine
B
Black
SE
Cabeza
R
Sinden
M
Mcintosh
AR
Toth
JP
Tulving
E
Stuss
DT
Episodic memory and the self in a case of isolated retrograde amnesia
Brain
 , 
1998
, vol. 
121
 (pg. 
1951
-
1973
)
Logan
JM
Sanders
AL
Snyder
AZ
Morris
JC
Buckner
RL
Under-recruitment and nonselective recruitment: dissociable neural mechanisms associated with aging
Neuron
 , 
2002
, vol. 
33
 (pg. 
827
-
840
)
Lustig
C
Snyder
AZ
Bhakta
M
O'Brien
KC
McAvoy
M
Raichle
ME
Morris
JC
Buckner
RL
Functional deactivations: change with age and dementia of the Alzheimer type
Proc Natl Acad Sci USA
 , 
2003
, vol. 
100
 (pg. 
14504
-
14509
)
Martin
RC
Sawrie
SM
Knowlton
RC
Bilir
E
Gilliam
F
Faught
E
Morawetz
RB
Kuzniecky
R
Bilateral hippocampal atrophy: consequences to verbal memory following temporal lobectomy
Neurology
 , 
2001
, vol. 
57
 (pg. 
597
-
604
)
McAllister
TW
Saykin
AJ
Flashman
LA
Sparling
MB
Johnson
SC
Guerin
SJ
Mamourian
AC
Weaver
JB
Yanofsky
N
Brain activation during working memory 1 month after mild traumatic brain injury: a functional MRI study
Neurology
 , 
1999
, vol. 
53
 (pg. 
1300
-
1308
)
McDermott
KB
Buckner
RL
Petersen
SE
Kelley
WM
Sanders
AL
Set- and code-specific activation in the frontal cortex: an fMRI study of encoding and retrieval of faces and words
J Cogn Neurosci
 , 
1999
, vol. 
11
 (pg. 
631
-
640
)
Miller
BT
D'Esposito
M
Searching for “the top” in top-down control
Neuron
 , 
2005
, vol. 
48
 (pg. 
535
-
538
)
Milner
B
Warren
JM
Akert
K
Some effects of frontal lobectomy in man
The frontal granular cortex and behavior
 , 
1964
New York
McGraw-Hill
(pg. 
313
-
334
)
Mishkin
M
Memory in monkeys severely impaired by combined but not by separate removal of amygdala and hippocampus
Nature
 , 
1978
, vol. 
273
 (pg. 
297
-
298
)
Ojemann
GA
Treatment of temporal lobe epilepsy
Annu Rev Med
 , 
1997
, vol. 
48
 (pg. 
317
-
328
)
Ojemann
JG
Kelley
WM
The frontal lobe role in memory: a review of convergent evidence and implications for the Wada memory test
Epilepsy Behav
 , 
2002
, vol. 
3
 (pg. 
309
-
315
)
Parker
A
Gaffan
D
Interaction of frontal and perirhinal cortices in visual object recognition memory in monkeys
Eur J Neurosci
 , 
1998
, vol. 
10
 (pg. 
3044
-
3057
)
Penfield
W
Milner
B
Memory deficit produced by bilateral lesions in the hippocampal zone
AMA Arch Neurol Psychiatry
 , 
1958
, vol. 
79
 (pg. 
475
-
497
)
Persson
J
Nyberg
L
Lind
J
Larsson
A
Nilsson
L-G
Ingvar
M
Buckner
RL
Structure-function correlates of cognitive decline in aging
Cereb Cortex
 , 
2005
 
Forthcoming.
Raichle
ME
Fiez
JA
Videen
TO
MacLeod
AM
Pardo
JV
Fox
PT
Petersen
SE
Practice-related changes in human brain functional anatomy during nonmotor learning
Cereb Cortex
 , 
1994
, vol. 
4
 (pg. 
8
-
26
)
Rausch
R
Epilepsy surgery within the temporal lobe and its short-term and long-term effects on memory
Curr Opin Neurol
 , 
2002
, vol. 
15
 (pg. 
185
-
189
)
Reuter-Lorenz
P
New visions of the aging mind and brain
Trends Cogn Sci
 , 
2002
, vol. 
6
 pg. 
394
 
Reuter-Lorenz
PA
Jonides
J
Smith
EE
Hartley
A
Miller
A
Marshuetz
C
Koeppe
RA
Age differences in the frontal lateralization of verbal and spatial working memory revealed by PET
J Cogn Neurosci
 , 
2000
, vol. 
12
 (pg. 
174
-
187
)
Rijntjes
M
Weiller
C
Recovery of motor and language abilities after stroke: the contribution of functional imaging
Prog Neurobiol
 , 
2002
, vol. 
66
 (pg. 
109
-
122
)
Rosene
DL
Van Hoesen
GW
Hippocampal efferents reach widespread areas of cerebral cortex and amygdala in the rhesus monkey
Science
 , 
1977
, vol. 
198
 (pg. 
315
-
317
)
Schacter
DL
Memory, amnesia, and frontal lobe dysfunction
Psychobiology
 , 
1987
, vol. 
15
 (pg. 
21
-
36
)
Scoville
WB
Milner
B
Loss of recent memory after bilateral hippocampal lesions
J Neurol Neurosurg Psychiatry
 , 
1957
, vol. 
20
 (pg. 
11
-
21
)
Simons
JS
Spiers
HJ
Prefrontal and medial temporal lobe interactions in long-term memory
Nat Rev Neurosci
 , 
2003
, vol. 
4
 (pg. 
637
-
648
)
Snyder
AZ
Bailey
D
Jones
T
Difference image vs. ratio image error function forms in PET-PET realignment
Quantification of brain function using PET
 , 
1996
San Diego, CA
Academic Press
(pg. 
131
-
137
)
Squire
LR
Ojemann
JG
Miezin
FM
Petersen
SE
Videen
TO
Raichle
ME
Activation of the hippocampus in normal humans: a functional anatomical study of memory
Proc Natl Acad Sci USA
 , 
1992
, vol. 
89
 (pg. 
1837
-
1841
)
Stuss
DT
Benson
DF
The frontal lobes
1986
New York
Raven Press
Talairach
J
Tournoux
P
Co-planar stereotaxic atlas of the human brain
 , 
1988
Germany: Thieme
Stuttgart
Thompson-Schill
SL
Bedny
M
Goldberg
RF
The frontal lobes and the regulation of mental activity
Curr Opin Neurobiol
 , 
2005
, vol. 
15
 (pg. 
219
-
224
)
Tulving
E
Kapur
S
Craik
FI
Moscovitch
M
Houle
S
Hemispheric encoding/retrieval asymmetry in episodic memory: positron emission tomography findings
Proc Natl Acad Sci USA
 , 
1994
, vol. 
91
 (pg. 
2016
-
2020
)
Ungerleider
LG
Gaffan
D
Pelak
VS
Projections from inferior temporal cortex to prefrontal cortex via the uncinate fascicle in rhesus monkeys
Exp Brain Res
 , 
1989
, vol. 
76
 (pg. 
473
-
484
)
Van Hoesen
GW
The parahippocampal gyrus: new observations regarding its cortical connections in the monkey
Trends Neurosci
 , 
1982
, vol. 
5
 (pg. 
345
-
350
)
Voytko
ML
Visual learning and retention examined with reversible cold lesions of the anterior temporal lobe
Behav Brain Res
 , 
1986
, vol. 
22
 (pg. 
25
-
39
)
Wagner
AD
Poldrack
RA
Eldridge
LL
Desmond
JE
Glover
GH
Gabrieli
JD
Material-specific lateralization of prefrontal activation during episodic encoding and retrieval
Neuroreport
 , 
1998
, vol. 
16
 (pg. 
3711
-
3717
)
Wagner
AD
Schacter
DL
Rotte
M
Koutstaal
W
Maril
A
Dale
AM
Rosen
BR
Buckner
RL
Building memories: remembering and forgetting of verbal experiences as predicted by brain activity
Science
 , 
1998
, vol. 
281
 (pg. 
1188
-
1191
)
Wiebe
S
Brain surgery for epilepsy
Lancet
 , 
2003
, vol. 
362
 
Suppl
(pg. 
s48
-
s49
)