Abstract

Neuropsychological tests that require shifting an attentional set, such as the Wisconsin Card Sorting Test, are sensitive to frontal lobe damage. Although little information is available for humans, an animal experiment suggested that different regions of the prefrontal cortex may contribute to set shifting behavior at different levels of processing. Behavioral studies also suggest that set shifting trials are more time consuming than non-set shifting trials (i.e. switch cost) and that this may be underpinned by differences at the neural level. We determined whether there were differential neural responses associated with two different levels of shifting behavior, that of reversal of stimulus–response associations within a perceptual dimension or that of shifting an attentional set between different perceptual dimensions. Neural activity in the antero-dorsal prefrontal cortex increased only in attentional set shifting, in which switch costs were significant. Activity in the postero-ventral prefrontal cortex increased not only in set shifting but also in reversing stimulus–response associations, in which switch costs were absent. We conclude that these distinct regions in the human prefrontal cortex provide different levels of attention control in response selection. Thus, the antero-dorsal prefrontal cortex may be critical for higher order control of attention, i.e. attentional set shifting, whereas the postero-ventral area may be related to a lower level of shift, i.e. reorganizing stimulus–response associations.

Introduction

The Wisconsin Card Sorting Test (WCST) has been used to assess the ability to shift a cognitive set from one perceptual attribute of a complex visual stimulus to another (i.e. set shifting ability). Typically, patients with frontal lobe damage show perseverative, or ‘stuck in set’, errors in the WCST (Milner, 1963). These patients tend to continue to sort the stimulus cards by the previously correct but currently incorrect rule, and this phenomenon has been explained by a deficit in shifting attention away from the previously relevant stimulus dimension towards the previously irrelevant stimulus dimension. It is important to distinguish this deficit from failures in reversal of stimulus–reward associations. Since the subjects were required to shift attention between stimulus dimensions in the WCST, rather than merely between different exemplars of the same dimension, perseverative errors in the WCST presumably reflect a relatively high level disturbance in ‘attentional set shift’ as opposed to an impairment in the reversal of a specific stimulus– response habit (i.e. ‘reversal shift’). Although prefrontal lesions impair both forms of shifts (Milner, 1963; Owen et al., 1991; Rolls et al., 1994), the variability of frontal lobe lesions in the human has made it difficult to assign the deficits to particular prefrontal regions. Neuroimaging studies demonstrated activation in the prefrontal regions during the performance of tasks with attentional set shifting (Berman et al., 1995; Nagahama et al., 1996, 1997), but these activities also distribute over different subregions of the prefrontal cortex, such as Brodmann areas (BA) 9, 46, 44 and 10. It remains unclear whether these different regions of the human prefrontal cortex carry out different subcomponent processes of the WCST.

In addition, behavioral studies in humans have reported an increase in reaction time in set shifting trials compared with that in task repetition, or set maintenance, trials, termed ‘switch cost’ (Allport et al., 1994; Rogers and Monsell, 1995; Nagahama et al., 1998b), and one can predict that this difference may be underpinned by differences at the neural level. We examined the neural activity in shifting an attentional set between different perceptual dimensions and in reversing stimulus–response associations within a particular perceptual dimension, in order to compare neural substrates for the two different levels of shifts in response selection. As the increase in neural activity related to shift responses should be transient, we employed event-related functional magnetic resonance imaging (fMRI) in the present study. Behavioral performance was also compared between the tasks to explore the specificity of switch costs in set shifting responses and their relationship to signals in particular cerebral regions.

Materials and Methods

Subjects

Ten right-handed male subjects (27.4 ± 8.1 years old, mean ± SD) took part in this study. Six of them participated in the fMRI experiment and the rest took part only in the behavioral study. All subjects were fully informed as to the nature of the study and written consents were obtained. Approval for the experiment was given by the Committee of Medical Ethics, Faculty of Medicine, Kyoto University.

Behavioral tasks

We evaluated the subjects' ability to shift their responding at two different levels of response selection, that of stimulus–response associations (visual discrimination reversal) and that of attentional selection for specific perceptual dimensions (set shifting). The stimulus (a set of cards with color figures) was presented to the subjects using Macprobe (Aristometrics, Castro Valley, CA) on a Macintosh computer (Apple Computers, Cupertino, CA).

In the set shifting task the subjects were instructed to match a test card to one of the two target cards according to color or shape (Fig. 1). The stimulus was displayed until the subject responded (or 2.5 s maximum) and the subject selected an answer by pushing the corresponding buttons with his right index or middle finger. A green circle or red × appeared at the center of the screen for 560 ms to indicate whether each response was correct or incorrect, then all cards disappeared and, after a blank of 890 ms, the next set of cards was presented. After a prescribed number of successive correct responses (7–9 cards) to one dimension (e.g. color), the previously irrelevant dimension (i.e. shape) became relevant and the subject had to shift the sorting criterion to the other one. Consequently, the inter-stimulus interval was around 2.5 s and set shifting occurred every ~20 s.

The reversal task was different from the set shifting task only in terms of instructions. The subjects were instructed to match a test card to one of two targets according to whether the color was the same or different. After successful serial selections of ‘same color’ targets, the stimulus– response associations were reversed and ‘different color’ targets became correct and vice versa. The parameters of stimulus presentation and response selection were strictly identical to those used during the set shifting task.

MRI Procedures

The change in the blood oxygenation level-dependent (BOLD) signal in fMRI was examined in the six subjects while they performed both the set shifting and the reversal tasks.

A time series of 100 fMRI volumes was acquired using T2*-weighted, gradient echo, EPI sequences (TR 2500 ms, TE 43 ms, flip angle 90°) with a 1.5 Tesla GE Horizon MR scanner (GE, Milwaukee, WI). The in-plane resolution was 64 × 64 pixels with a pixel dimension of 3.75 × 3.75 mm. Each volume consisted of 14 slices and the slice thickness was 4 mm, with a 0.5 mm gap, covering the area from the top of the brain to the horizontal plane through the anterior and posterior commissure. The scans did not cover the orbitofrontal cortex or cerebellum. Head motion was minimized by placing tight but comfortable foam padding around the subject's head. Each subject had six fMRI sessions, three each for the set shifting and the reversal tasks. Seventy-eight sets of visual stimuli were presented in each session. The stimulus cards were presented to the subjects with a projector through a mirror that was attached to a head coil. The stimuli were arranged approximately within the 10° × 10° of the visual angle. The order of task presentation was counterbalanced among subjects. To minimize the effect of rule learning and differences in problem solving strategy among the subjects, we fully instructed the subjects as to the rules of the task and, before the fMRI scan, we trained them until they could perform the task without difficulty. Subjects were required to fixate the central point of the screen throughout the task to restrict their eye motion.

A 3-dimensional structural image was also obtained with spoiled gradient echo (TE 2.2 ms, TR 150 ms, flip angle 60°, slice thickness 3 mm) after all the fMRI sessions.

Data Analysis

The data were analyzed with Statistical Parametric Mapping (SPM99b; Wellcome Department of Cognitive Neurology, London, UK) (Friston et al., 1995).

The first three volumes of each fMRI scan were discarded because of unsteady magnetization. To correct for the different sampling times of different slices, the signals in each slice were shifted relative to acquisition of the first slice using a sinc interpolation, then the fourth and the last volumes were discarded. The remaining 95 volume images in each session were realigned to the first for motion correction. Following realignment, all images were co-registered to the subject's T1-weighted anatomical image. Thereafter, fMRI images were transformed into a standard stereotaxic space (based on the Montreal Neurological Institute reference brain) (Mazziotta et al., 1995) using the parameter that was obtained by transformation of the subject's anatomical image. The data were then spatially smoothed with an 8 mm isotropic Gaussian kernel and temporally smoothed with a 4 s Gaussian kernel to increase the signal-to-noise ratio.

The data were analyzed by modeling the evoked hemodynamic responses to the stimulus onset for each event with a canonical, synthetic hemodynamic response function in the context of the general linear model as employed by SPM99b (Friston et al., 1998). In this analysis the ‘wrong’ feedback in the tasks was used as the cue of the event to be analyzed, because set shifting or reversing stimulus–response association was assumed to follow the ‘wrong’ feedback and to be completed before appearance of the next target. The effect of differences in global activity across scans was removed using proportional scaling. Low frequency physiological noise and baseline drifts were cut off using a high pass filter (0.7 cycles/min). Thereafter, specific effects of the events were tested by applying appropriate linear contrasts to the parameter estimates. The significance of the least mean squares fit of the model to the data was calculated for each voxel and a statistical parametric map (SPM{t}) was constructed. SPM{t} was then transformed into normally distributed Z statistics, yielding a statistical map (SPM{Z}).

To reveal the significantly activated brain areas during each task, the subject's data were modeled separately and group results are presented as the conjunction of activations across all six subjects (Friston et al., 1999). This means that we explored the data only for those changes that were common to all subjects. Group results presented are those that survived a threshold of Pcorr < 0.05.

To compare signal changes between the two conditions more directly, the contrasts of the parameter estimates for the height of the canonical response for each condition (i.e. estimated activation parameter for the set shifting responses and that for the reversal responses) resulting from the least mean squares fit of the model to the data were stored as separate images. Direct comparison between the two conditions was performed using these subject-specific contrast images as ‘raw’ data for a paired t-test (effecting a random effects model) (Holmes and Friston, 1998). Regionally specific differences were considered significant when the cluster consisted of five or more contiguous voxels surviving a threshold of Puncorr < 0.001.

To strengthen the group results, we also explored the activated foci in each subject. In the single subject analysis the threshold for significance was set at Puncorr < 0.001. The stereotactic coordinates of Talairach and Tournoux (Talairach and Tournoux, 1988) were used to report the activation foci, but descriptions of the anatomical localization of the foci were determined based on individual structural MRIs and the atlas of Duvernoy (Duvernoy, 1999).

To confirm the SPM results, regions of interest analyses were also performed using the time series fMRI data at the voxels of local maximal activation during both tasks in each brain area. For each subject the data from three fMRI runs on each task were averaged into a single time series data set. The effects of task (set shifting versus reversal) and time were analyzed by repeated two-way analysis of variance (ANOVA).

Results

Behavioral Data

All subjects performed both tasks very well and there were no significant differences in the numbers of errors made between the two tasks (0.94 ± 1.30% in the set shifting and 1.39 ± 1.43% in the reversal task, respectively; F1,5 = 2.49, P = 0.17). The subjects' responses were classified as follows: correct responses during successive correct matching were classified as non-shift responses, those on the set shifting or reversing (i.e. those following the ‘wrong’ feedback) as shift responses and incorrect responses as errors. Repeated two-way ANOVA showed the significant main effects of tasks (set shifting versus reversal, F1,9 = 13.9, P = 0.0047) and responses (non-shift versus shift, F1,9 = 12.2, P = 0.0068) on the reaction times, with a significant task × response interaction (F = 16.6, P = 0.0028) (Fig. 2). Paired comparison of reaction times in each task revealed significant switch costs in the set shifting task (34.8 ± 2.3 ms; F1,9 = 39.6, P < 0.0001) but no switch costs in the reversal task (5.5 ± 7.6 ms; F1,9 < 1).

Neuroimaging Data

Prefrontal Cortex

In the set shifting task a significant transient increase in neural activity was observed following the ‘wrong’ feedback in both the right antero-dorsal part and bilateral postero-ventral parts of the prefrontal cortex (Fig. 3). Activity in the antero-dorsal part of the prefrontal cortex (ADPF) was on the surface of the middle frontal gyrus and the local maximum of activation peak (Table 1) was situated within BA 46 (Rajkowska and Goldman-Rakic, 1995). The activation peak in the postero-ventral part of the prefrontal cortex (PVPF) was situated in the inferior frontal sulcus near the junction with the precentral sulcus and extended onto the adjacent parts of the inferior and middle frontal gyrus. This activity was right-dominant and corresponded to BA 8/44 (Amunts et al., 1999).

In the reversal task significant PVPF activation was observed bilaterally, as in the set shifting task (Fig. 3). The activity was also right-dominant. In contrast to that in the set shifting task, ADPF activation was not detected in the reversal task. Individual subject analysis confirmed the group results, which showed distinct activation in the ADPF and PVPF in the set shifting task but activation only in the PVPF in the reversal task (Table 2).

When neural activation during the reversal task was subtracted from that during the set shifting task directly, two peaks were detected within the right ADPF (Fig. 5). One peak was located at coordinates similar to the peak detected within the ADPF during the set shifting task and the other was observed slightly more dorsally. These peaks were situated on the middle frontal gyrus within BA 46. No significant peak was detected within the PVPF in this direct comparison.

Regions of interest analysis also showed that the task × time interaction was significant in the right ADPF (F = 2.51, P = 0.026), but not significant in the PVPF (right, F = 1.14, P = 0.36; left, F = 1.17, P = 0.34) (Fig. 3, graphs). The activity in the ADPF was significantly increased following the ‘wrong’ feedback in the set shifting task (F5,8 = 6.80, P < 0.0001, repeated ANOVA) but not in the reversal task (F5,8 < 1).

Other Areas

Except for the ADPF, very similar brain areas were activated at the time of shifting responses in both tasks (Fig. 4 and Table 1). In the prerolandic area, the right presupplementary motor area, anterior cingulate gyrus and dorsal premotor area around the frontal eye field were activated in both tasks. In the postrolandic area, bilateral intraparietal sulci, angular/supramarginal gyri (Ga/Gsm), medial and dorsal superior parietal lobuli and the right retrosplenial gyrus were activated. Direct comparison between the two tasks showed that the signal increase in the medial superior parietal lobule during the set shifting task was larger than that during the reversal task.

Regions of interest analyses confirmed that time courses of fMRI signals in both tasks were similar in most areas. There was a significant task × time interaction only in the right anterior cingulate gyrus (F = 2.44, P = 0.029) and right Ga/Gsm (F = 2.51, P = 0.026). The signal intensities in these two areas were significantly increased in both tasks (cingulate, F5,8 = 10.5, P < 0.0001 and F5,8 = 12.6, P < 0.0001; Ga/Gsm, F5,8 = 14.8, P < 0.0001 and F5,8 = 16.5, P < 0.0001; in the set shifting and reversal tasks, respectively) and that activation in the reversal task tended to be prolonged and higher than that in the set shifting task at the later stage of hemodynamic response (Fig. 4, graphs).

Discussion

The present study showed that although the neural activity in the PVPF increased both in reversing stimulus–response associations within a perceptual dimension (color) and in shifting an attentional set between dimensions (color and shape), the activity in the ADPF increased only in shifting an attentional set from one perceptual dimension to another. This finding and the fact that switch costs were apparent only in the set shifting responses but not in the reversal responses highlight possible dissociable mechanisms of attention control within the lateral prefrontal cortex in humans.

Antero-dorsal Prefrontal Cortex

The ADPF was activated only in attentional set shifting, in which switch costs were significant. Direct subtraction of signal change on the reversal responses from that on the set shifting responses demonstrated again that a significant difference in neural activity was detected only within the ADPF and not within the PVPF. The only difference between the two tasks was the instructions, the other perceptual and motor components being exactly the same. Thus, it seems apparent that the ADPF is a critical area in shifting an attentional set between different perceptual dimensions. This result is consistent with the proposal in previous studies that the dorsolateral prefrontal cortex may be important in the set shifting process (Milner, 1963; Weinberger et al., 1986) and it is likely that disruption of this process is at least partly responsible for the deficits repeatedly seen with the WCST in patients with frontal lobe damage (Milner, 1963; Drewe, 1974; Robinson et al., 1980; Arnett et al., 1994).

The activated area in the ADPF was situated on the middle frontal gyrus near the frontal pole. Since the local peak of activation is within the conservative limit of BA 46 according to a recent cytoarchitechtonic mapping (Rajkowska and Goldman-Rakic, 1995), we believe that ADPF activation is in the rostral part of BA 46. Single subject analysis confirmed this conclusion. Four of the six ADPF activations were at the conservative coordinates of BA 46 and all the activations were within the limit of variation for BA 46. Interestingly, activations in the ADPF are more variable in location than those in the PVPF. This was also consistent with the intersubject variability of BA 46, which was mentioned in the morphological study (Rajkowska and Goldman-Rakic, 1995).

It can be argued that the differential activity in the ADPF may be related to the effect of task difficulty. We could not reject this possibility completely because although the numbers of errors that the subjects made were comparable between the two tasks, the reaction times in the set shifting task were longer than those in the reversal task. It is, however, unlikely for two reasons. First, a psychological study showed that switch costs and task difficulty are independent factors and that manipulation of task difficulty did not affect task switch costs (Dove et al., 2000). Second, neural activity levels were comparable between the tasks in other brain regions, or even greater in the reversal task in the anterior cingulate gyrus and Ga/Gsm. Previous studies showed that task difficulty modulates neural activity positively in the cingulate gyrus (Barch et al., 1997; Paus et al., 1998) and in the parietal lobe (D'Esposito et al., 1995). Therefore, the fact that the activity differed only in the ADPF and not in the cingulate and Ga/Gsm suggests that the differential activity in the ADPF is not confounded much by task difficulty. However, this point should be clarified in future studies by varying the attentional demands of the reversal task.

Postero-ventral Prefrontal Cortex

Several previous studies with event-related fMRI have shown that transient neural activity was detected in the PVPF time locked to set shifting (Konishi et al., 1998; Nagahama et al., 1998a) or task switching (Dove et al., 2000) and argued that the PVPF might be a critical area in set shifting. Our results demonstrated, however, that comparable activity was also observed time locked to shift responses in the reversal task, in which switch costs were absent. This finding suggests that the role of the PVPF in selection extends beyond those situations in which the attentional set shifting was required. What is the cognitive component related to PVPF activity? There are several pieces of evidence indicating that the PVPF plays a role in inhibiting responses to irrelevant stimuli in situations in which the subject is required to reorganize stimulus–response associations. Rushworth and co-workers examined the function of BA 45A (Rushworth et al., 1997), which is the most postero-lateral part of the inferior prefrontal cortex in monkeys and corresponds to the PVPF in our study (Petrides and Pandya, 1994). They revealed that lesions in this area impaired relearning of a simultaneous color matching task. Lesions in the PVPF also impair the reversal learning of visual stimulus–reward associations (Mishkin et al., 1969; Iversen and Mishkin, 1970). These studies indicate that the PVPF is important in learning or reorganizing the stimulus–response associations to gain a reward. Several reports suggest that the physiological function of this area has an inhibitory nature. In fMRI experiments in humans the same region of the PVPF was activated by no-go trials during the go-no-go task and by set shifting responses during the WCST (Konishi et al., 1999). Using the go/no-go task with moving colored random dots as stimuli, Sakagami and co-workers (Sakagami and Niki, 1994; Sakagami et al., 1999) recently revealed that 84% of go/no-go-related visual cells in the PVPF showed differential activity only when color (not direction of motion) was relevant in selecting go/no-go responses and that this modulatory effect of attention on neuronal activity was observed only for no-go indicating color. Since such reorganization of visual–motor associations is a major component of the reversal trials and is also involved in the process of set shifting trials, we speculate that the activity in the PVPF in both tasks is at least partly related to control of previously acquired stimulus–response associations.

Functional Organization within the Prefrontal Cortex

It is widely accepted that the prefrontal cortex is critically involved in several important cognitive processes, such as novelty detection (Berns et al., 1997; Knight and Nakada, 1998), inhibitory functions (Verfaellie and Heilman, 1987; Rolls et al., 1994) and, especially, working memory (Courtney et al., 1996; Goldman-Rakic, 1996; McCarthy et al., 1996). Therefore, the activity in the prefrontal cortex, especially in the PVPF, might be accounted for by the working memory component that underlies both tasks. Goldman-Rakic has suggested that impairment on tests such as the WCST may reflect an underlying deficit in the working memory process (Goldman-Rakic, 1987). Although there is no delay component in response selection in both tasks and the working memory load is minimal in the present study, the activity in the PVPF during both tasks might be confounded with the cognitive process of holding information on line to solve the problems.

Our findings suggest that distinct regions in the human lateral prefrontal cortex carry out different, possibly hierarchical, levels of processing in shifting behavior. Thus, reorganization of stimulus–response associations for specific visual stimuli is supported by the PVPF, whereas the higher order shifting of attention between perceptual dimensions of visual stimuli is mediated by the ADPF. Essentially, the WCST and the set shifting task in the present study are series of visual discriminations over multidimensional stimuli in which different aspects of the stimuli are relevant to reinforcement at different times. A succession of correct sorts through a sequence of trials has the effect of promoting the development of an attentional bias towards the stimulus dimension which is currently relevant. This bias is then challenged on trials that require the subject's attention to be directed away from that dimension towards the newly relevant dimension. Thus, it seems appropriate to attribute the signal changes within the ADPF in the present study to the control of attentional set. Using a task analogous to the WCST in non-human primates, Dias and co-workers revealed that lesions in the orbital prefrontal cortex impaired the ability to change stimulus–reward association (‘affective’ shift), whereas lesions in the lateral prefrontal cortex impaired higher order ‘attentional’ shifts (Dias et al., 1996). The present study indicates that such ‘attentional’ and ‘affective’ shifts may be controlled within the lateral prefrontal cortex in humans.

On the other hand, the right ADPF, which was differentially activated in the set shifting task, has also been activated in studies involving a variety of spatial and non-spatial working memory tasks [for reviews see D'Esposito et al. (D'Esposito et al., 1998) and Smith and Jonides (Smith and Jonides, 1999)]. Thus, this activation may reflect some sort of ‘executive function’ or ‘active representation’ in a working memory system needed for the performance of the set shifting task. According to the ‘two-stage’ model of prefrontal function proposed by Petrides (Petrides, 1995; Owen et al., 1996), different areas of the prefrontal cortex are hypothesized to mediate different kinds of cognitive processes implicated in working memory function. Within this framework, the mid-ventrolateral prefrontal area is principally involved in organization of sequences of responses based on conscious, explicit retrieval of information from posterior association systems, while the mid-dorsolateral prefrontal cortex is recruited when active ‘manipulation’, ‘updating’ or ‘monitoring’ of such information is required within working memory. Although the ADPF in the present study is located more rostral to the mid-dorsolateral area and the PVPF is located superior to the mid-ventrolateral area described by him, the increased neural activity seen here around the middle frontal gyrus in set shifting relative to reversal shift may reflect some additional manipulation of information or updating of the task representation to include attending to and applying the matching rule to a different set of dimensions. Although it is unclear which of these theoretical approaches (‘modulation of attentional set’ or ‘updating representation’) provides the best account of the present data, our findings suggest that a similar dichotomy into higher and lower levels of cognitive processing can be applied to functional organization of shifting behavior.

Finally, the present findings may be partly inconsistent with another possibility suggested by Dias and co-workers, of parallel and independent, rather than hierarchical, processing of affective and higher order information in distinct prefrontal regions (Dias et al., 1997). The precise nature of this organization (parallel or hierarchical) remains open for future research.

Notes

We thank Professor Trevor W. Robbins (Department of Experimental Psychology, University of Cambridge, Cambridge, UK) and Prof. Shintaro Funahashi (Division of Human and Environmental Studies, Department of Environmental Information Processing, Kyoto University, Kyoto, Japan) for their thoughtful comments on this manuscript. We also thank Professor Patricia S. Goldman-Rakic and two anonymous reviewers for their helpful suggestions. This study was partly supported by grants from Research for the Future Program (JSPS-RFTF97L00201) and a General Research Grant for Aging and Health ‘Analysis of aged brain function with neuroimaging techniques’ from the Japan Ministry of Health and Welfare.

Address correspondence to H. Fukuyama, Department of Brain Pathophysiology, Kyoto University Hospital, 54 Shogoin Kawahara-cho, Sakyo, Kyoto 606-8507, Japan. Email: fukuyama@kuhp.kyoto-u.ac.jp.

Table 1

Location of transient activation in the set shifting and reversal tasks (group results)

Area Side Set shifting task Reversal task 
  x y z Z value x y z Z value 
ADPF, antero-dorsal prefrontal cortex; PVPF, postero-ventral prefrontal cortex; preSMA, presupplementary motor area; GC, cingulate gyrus; FEF/PM, frontal eye field/premotor; IPS, intraparietal sulcus; Ga/Gsm, angular/supramarginal gyrus; LPs, superior parietal lobule; RS, retrosplenial gyrus; R, right; L, left. (x, y, z), coordinates (mm) in stereotactic space. Z score, significance of signal change. All activations survived the threshold of Pcorr < 0.05, except for RS. Activation in the RS was significant at Puncorr < 0.001. There was no significant activation in the ADPF in the reversal task (non-significant Z score shown in italics). 
ADPF  34  50 16 5.71    1.88 
PVPF  46   6 32 8.98  54   8 36 7.22 
 –54  10 36 7.12 –42   2 28 6.30 
preSMA   2   2 52 7.52   8   2 56 5.13 
GC   6   6 48 6.80   6   4 48 7.67 
FEF/PM  36 –12 64 4.72  28 –14 60 5.28 
  22  –6 60 6.55  26  –4 60 6.18 
IPS  36 –64 36 8.30  30 –64 40 6.89 
 –28 –58 44 7.44 –26 –58 44 5.07 
Ga/Gsm  40 –48 48 8.61  46 –44 52 7.94 
 –44 –44 52 6.11 –46 –42 52 6.00 
LPs  18 –70 36 7.70  12 –70 36 6.09 
 –16 –74 44 7.76 –16 –76 44 6.56 
RS   2 –34 28 4.03   4 –36 28 4.47 
Area Side Set shifting task Reversal task 
  x y z Z value x y z Z value 
ADPF, antero-dorsal prefrontal cortex; PVPF, postero-ventral prefrontal cortex; preSMA, presupplementary motor area; GC, cingulate gyrus; FEF/PM, frontal eye field/premotor; IPS, intraparietal sulcus; Ga/Gsm, angular/supramarginal gyrus; LPs, superior parietal lobule; RS, retrosplenial gyrus; R, right; L, left. (x, y, z), coordinates (mm) in stereotactic space. Z score, significance of signal change. All activations survived the threshold of Pcorr < 0.05, except for RS. Activation in the RS was significant at Puncorr < 0.001. There was no significant activation in the ADPF in the reversal task (non-significant Z score shown in italics). 
ADPF  34  50 16 5.71    1.88 
PVPF  46   6 32 8.98  54   8 36 7.22 
 –54  10 36 7.12 –42   2 28 6.30 
preSMA   2   2 52 7.52   8   2 56 5.13 
GC   6   6 48 6.80   6   4 48 7.67 
FEF/PM  36 –12 64 4.72  28 –14 60 5.28 
  22  –6 60 6.55  26  –4 60 6.18 
IPS  36 –64 36 8.30  30 –64 40 6.89 
 –28 –58 44 7.44 –26 –58 44 5.07 
Ga/Gsm  40 –48 48 8.61  46 –44 52 7.94 
 –44 –44 52 6.11 –46 –42 52 6.00 
LPs  18 –70 36 7.70  12 –70 36 6.09 
 –16 –74 44 7.76 –16 –76 44 6.56 
RS   2 –34 28 4.03   4 –36 28 4.47 
Table 2

Activation peaks within the right prefrontal cortex in each subject

Area Subject Set shifting task Reversal task 
  x y z Z score x y z Z score 
PVPF, postero-ventral prefrontal cortex; ADPF, antero-dorsal prefrontal cortex; NS, not significant. (x, y, z), coordinates (mm) in stereotactic space. Z score, significance of signal change. 
PVPF S1 56 20 28 5.08 50 24 32 3.96 
 S2 58 36 7.56 52 40 3.24 
 S3 38 32 7.86 38 32 6.77 
 S4 42 32 5.13 56 32 4.63 
 S5 50 32 5.22 50 12 32 5.62 
 S6    NS    NS 
ADPF S1 34 50 3.79    NS 
 S2 38 42 32 6.79    NS 
 S3 40 46 20 5.25    NS 
 S4 26 42 20 4.07    NS 
 S5 34 50 12 3.49 34 50 16 6.36 
 S6 38 34 36 4.09    NS 
Area Subject Set shifting task Reversal task 
  x y z Z score x y z Z score 
PVPF, postero-ventral prefrontal cortex; ADPF, antero-dorsal prefrontal cortex; NS, not significant. (x, y, z), coordinates (mm) in stereotactic space. Z score, significance of signal change. 
PVPF S1 56 20 28 5.08 50 24 32 3.96 
 S2 58 36 7.56 52 40 3.24 
 S3 38 32 7.86 38 32 6.77 
 S4 42 32 5.13 56 32 4.63 
 S5 50 32 5.22 50 12 32 5.62 
 S6    NS    NS 
ADPF S1 34 50 3.79    NS 
 S2 38 42 32 6.79    NS 
 S3 40 46 20 5.25    NS 
 S4 26 42 20 4.07    NS 
 S5 34 50 12 3.49 34 50 16 6.36 
 S6 38 34 36 4.09    NS 
Figure 1.

Scheme of the tasks. T1–T5 represent the five serial presentations on the screen. In each presentation the two upper cards were target cards and the lower one was a test card. Thick arrows indicate the subject's answers. During the set shifting task, the subject matched the test card to the target by color in T1 and it was correct. After a prescribed number of cards (three cards in this scheme, T1T3) was sorted by color correctly, the feedback response became incorrect (T4). Then the subject had to shift the sorting rule from color to shape to yield correct answers (T5). During the reversal task the subject matched the test card to the targets by whether color was the same or different.

Figure 1.

Scheme of the tasks. T1–T5 represent the five serial presentations on the screen. In each presentation the two upper cards were target cards and the lower one was a test card. Thick arrows indicate the subject's answers. During the set shifting task, the subject matched the test card to the target by color in T1 and it was correct. After a prescribed number of cards (three cards in this scheme, T1T3) was sorted by color correctly, the feedback response became incorrect (T4). Then the subject had to shift the sorting rule from color to shape to yield correct answers (T5). During the reversal task the subject matched the test card to the targets by whether color was the same or different.

Figure 2.

Reaction times during the set shifting and reversal tasks. SFT, set shifting task; RVS, reversal task; NS, non-shift responses; S, shift responses. *The reaction times in the shift responses are significantly longer than those in the non-shift responses during the set shifting task (switch cost, P < 0.05, paired t-test). Error bars show standard errors of means.

Figure 2.

Reaction times during the set shifting and reversal tasks. SFT, set shifting task; RVS, reversal task; NS, non-shift responses; S, shift responses. *The reaction times in the shift responses are significantly longer than those in the non-shift responses during the set shifting task (switch cost, P < 0.05, paired t-test). Error bars show standard errors of means.

Figure 3.

Transient brain activity in the lateral prefrontal area in the set shifting and reversal trials. (Left) Surface rendering of prefrontal activation during the set shifting (SFT, upper row) and reversal (RVS, lower row) tasks. Note that no activation is detected in the right anterior part of the middle frontal gyrus (antero-dorsal prefrontal cortex, ADPF) in the reversal trials. Coordinates of each activity are presented in Table 1. The activation is rendered on the normalized SPGR image from a subject. (Right) Time courses of fMRI signals in the right ADPF and inferior frontal sulcus (PVPF), respectively. SFT, set shifting task; RVS, reversal task; SI, signal intensity. Zero seconds represents the appearance of ‘wrong’ feedback. Error bars show standard errors of means. In the ADPF the signals in the reversal trials were obtained from the voxel whose coordinates were the same as the activity peak in the set shifting trials, because no significant activation was seen in the location in the reversal task.

Transient brain activity in the lateral prefrontal area in the set shifting and reversal trials. (Left) Surface rendering of prefrontal activation during the set shifting (SFT, upper row) and reversal (RVS, lower row) tasks. Note that no activation is detected in the right anterior part of the middle frontal gyrus (antero-dorsal prefrontal cortex, ADPF) in the reversal trials. Coordinates of each activity are presented in Table 1. The activation is rendered on the normalized SPGR image from a subject. (Right) Time courses of fMRI signals in the right ADPF and inferior frontal sulcus (PVPF), respectively. SFT, set shifting task; RVS, reversal task; SI, signal intensity. Zero seconds represents the appearance of ‘wrong’ feedback. Error bars show standard errors of means. In the ADPF the signals in the reversal trials were obtained from the voxel whose coordinates were the same as the activity peak in the set shifting trials, because no significant activation was seen in the location in the reversal task.

Figure 4.

Transient brain activity in the other areas in the set shifting and reversal trials. (Left) Surface rendering of non-prefrontal activation during the set shifting (SFT, upper row) and reversal (RVS, lower row) tasks. Coordinates of the activation are presented in Table 1. (Right) Time course of fMRI signals in regions other than the lateral prefrontal cortex. GC, right cingulate gyrus; Ga/Gsm, right angular/supramarginal gyrus; SFT, set shifting task; RVS, reversal task; SI, signal intensity. Error bars show standard errors of means.

Transient brain activity in the other areas in the set shifting and reversal trials. (Left) Surface rendering of non-prefrontal activation during the set shifting (SFT, upper row) and reversal (RVS, lower row) tasks. Coordinates of the activation are presented in Table 1. (Right) Time course of fMRI signals in regions other than the lateral prefrontal cortex. GC, right cingulate gyrus; Ga/Gsm, right angular/supramarginal gyrus; SFT, set shifting task; RVS, reversal task; SI, signal intensity. Error bars show standard errors of means.

Figure 5.

Significant signal increase in the set shifting responses compared with the reversal responses: direct comparison. Two significant peaks were detected in the right middle frontal gyrus (x = 42, y = 46, z = 12, Z = 3.35; x = 46, y = 24, z = 28, Z = 3.88) within Brodmann's area 46. An additional peak was observed on the right medial superior parietal lobule (x = 16, y = -66, z = 44, Z = 3.29). The threshold for significance was set at P < 0.001. R, right.

Figure 5.

Significant signal increase in the set shifting responses compared with the reversal responses: direct comparison. Two significant peaks were detected in the right middle frontal gyrus (x = 42, y = 46, z = 12, Z = 3.35; x = 46, y = 24, z = 28, Z = 3.88) within Brodmann's area 46. An additional peak was observed on the right medial superior parietal lobule (x = 16, y = -66, z = 44, Z = 3.29). The threshold for significance was set at P < 0.001. R, right.

References

Allport DA, Styles EA, Hsieh S (1994) Shifting intentional set: exploring the dynamic control of tasks. In: Attention and performance XV (Umilta C, Moscovitch M, eds), pp. 421–452. Hillsdale: Erlbaum.
Amunts K, Schleicher A, Burgel U, Mohlberg H, Uylings HB, Zilles K (
1999
) Broca's region revisited: cytoarchitecture and intersubject variability.
J Comp Neurol
 
412
:
319
–341.
Arnett PA, Rao SM, Bernardin MS, Grafman J, Yetkin FZ, Lobeck L (
1994
) Relationship between frontal lobe lesions and Wisconsin Card Sorting Test performance in patients with multiple sclerosis.
Neurology
 
44
:
420
–425.
Barch DM, Braver TS, Nystrom LE, Forman SD, Noll DC, Cohen JD (
1997
) Dissociating working memory from task difficulty in human prefrontal cortex.
Neuropsychologia
 
35
:
1373
–1380.
Berman KF, Ostrem JL, Randolph C, Gold J, Goldberg TE, Coppola R, Carson RE, Herscovitch P, Weinberger DR (
1995
) Physiological activation of a cortical network during performance of the Wisconsin Card Sorting Test: a positron emission tomography study.
Neuropsychologia
 
33
:
1027
–1046.
Berns GS, Cohen JD, Mintun MA (
1997
) Brain regions responsive to novelty in the absence of awareness.
Science
 
276
:
1272
–1275.
Courtney SM, Ungerleider LG, Keil K, Haxby JV (
1996
) Object and spatial visual working memory activate separate neural systems in human cortex.
Cereb Cortex
 
6
:
39
–49.
D'Esposito M, Aguirre GK, Zarahn E, Ballard D, Shin RK, Lease J (
1998
) Functional MRI studies of spatial and nonspatial working memory.
Cogn Brain Res
 
7
:
1
–13.
D'Esposito M, Detre JA, Alsop DC, Shin RK, Atlas S, Grossman M (
1995
) The neural basis of the central executive system of working memory.
Nature
 
378
:
279
–281.
Dias R, Robbins TW, Roberts AC (
1996
) Dissociation in prefrontal cortex of affective and attentional shifts.
Nature
 
380
:
69
–72.
Dias R, Robbins TW, Roberts AC (
1997
) Dissociable forms of inhibitory control within prefrontal cortex with an analog of the Wisconsin Card Sort Test: restriction to novel situations and independence from “on-line” processing.
J Neurosci
 
17
:
9285
–9297.
Dove A, Pollmann S, Schubert T, Wiggins CJ, von Cramon Y (
2000
) Prefrontal cortex activation in task switching: an event-related fMRI study.
Cogn Brain Res
 
9
:
103
–109.
Drewe EA (
1974
) The effect of type and area of brain lesion on Wisconsin card sorting test performance.
Cortex
 
10
:
159
–170.
Duvernoy HM (1999) The human brain. Surface, blood supply, and three-dimensional sectional anatomy. New York: Springer.
Friston KJ, Holmes AP, Worsley KJ, Poline JB, Frith CD, Frackowiak RSJ (
1995
) Statistical parametric maps in functional imaging: a general linear approach.
Hum Brain Map
 
2
:
189
–210.
Friston KJ, Fletcher P, Josephs O, Holmes A, Rugg MD, Turner R (
1998
) Event-related fMRI: characterizing differential responses.
NeuroImage
 
7
:
30
–40.
Friston KJ, Holmes AP, Price CJ, Buchel C, Worsley KJ (
1999
) Multisubject fMRI studies and conjunction analyses.
NeuroImage
 
10
:
385
–396.
Goldman-Rakic PS (1987) Circuitry of primate prefrontal cortex and regulation of behaviour by representational memory. In: Handbook of physiology, the nervous system, higher functions of the brain, Vol. V (Plum F, Mountcastle V, eds), pp. 373–417. Bethesda: American Physiological Society.
Goldman-Rakic PS (
1996
) Regional and cellular fractionation of working memory.
Proc Natl Acad Sci USA
 
93
:
13473
–13480.
Holmes AP, Friston KJ (
1998
) Generalisability, random effects and population inference.
NeuroImage
 
7
:
S754
.
Iversen SD, Mishkin M (
1970
) Perseverative interference in monkeys following selective lesions of the inferior prefrontal convexity.
Exp Brain Res
 
11
:
376
–386.
Knight RT, Nakada T (
1998
) Cortico-limbic circuits and novelty: a review of EEG and blood flow data.
Rev Neurosci
 
9
:
57
–70.
Konishi S, Nakajima K, Uchida I, Kameyama M, Nakahara K, Sekihara K, Miyashita Y (
1998
) Transient activation of inferior prefrontal cortex during cognitive set shifting.
Nature Neurosci
 
1
:
80
–84.
Konishi S, Nakajima K, Uchida I, Kikyo H, Kameyama M, Miyashita Y (
1999
) Common inhibitory mechanism in human inferior prefrontal cortex revealed by event-related functional MRI.
Brain
 
122
:
981
–991.
Mazziotta JC, Toga AW, Evans A, Fox P, Lancaster J (
1995
) A probabilistic atlas of the human brain: theory and rationale for its development. The International Consortium for Brain Mapping (ICBM).
NeuroImage
 
2
:
89
–101.
McCarthy G, Puce A, Constable RT, Krystal JH, Gore JC, Goldman-Rakic P (
1996
) Activation of human prefrontal cortex during spatial and nonspatial working memory tasks measured by functional MRI.
Cereb Cortex
 
6
:
600
–611.
Milner B (
1963
) Effects of different brain lesions on card sorting.
Arch Neurol
 
9
:
90
–100.
Mishkin M, Vest B, Waxler M, Rosvold HE (
1969
) A re-examination of the effects of frontal lesions on object alternation.
Neuropsychologia
 
7
:
357
–363.
Nagahama Y, Fukuyama H, Yamauchi H, Matsuzaki S, Konishi J, Shibasaki H, Kimura J (
1996
) Cerebral activation during performance of a Card Sorting Test.
Brain
 
119
:
1667
–1675.
Nagahama Y, Fukuyama H, Yamauchi H, Katsumi Y, Magata Y, Shibasaki H, Kimura J (
1997
) Age-related changes in cerebral blood flow activation during a Card Sorting Test.
Exp Brain Res
 
114
:
571
–577.
Nagahama Y, Okada T, Katsumi Y, Hayashi T, Yamauchi H, Sawamoto N, Hanakawa T, Konishi J, Fukuyama H, Shibasaki H (
1998
) Neural activity in attention shift between object features: comparison of activated areas between the state-dependent and event-related functional MRI study.
Soc Neurosci Abstr
 
24
:
1681
.
Nagahama Y, Sadato N, Yamauchi H, Katsumi Y, Hayashi T, Fukuyama H, Kimura J, Shibasaki H, Yonekura Y (
1998
) Neural activity during attention shifts between object features.
NeuroReport
 
9
:
2633
–2638.
Owen AM, Roberts AC, Polkey CE, Sahakian BJ, Robbins TW (
1991
) Extra-dimensional versus intra-dimensional set shifting performance following frontal lobe excisions, temporal lobe excisions or amygdalo-hippocampectomy in man.
Neuropsychologia
 
29
:
993
–1006.
Owen AM, Evans AC, Petrides M (
1996
) Evidence for a two-stage model of spatial working memory processing within the lateral frontal cortex: a positron emission tomography study.
Cereb Cortex
 
6
:
31
–38.
Paus T, Koski L, Caramanos Z, Westbury C (
1998
) Regional differences in the effects of task difficulty and motor output on blood flow response in the human anterior cingulate cortex: a review of 107 PET activation studies.
NeuroReport
 
9
:
R37
–R47.
Petrides M (
1995
) Impairments on nonspatial self-ordered and externally ordered working memory tasks after lesions of the mid-dorsal part of the lateral frontal cortex in the monkey.
J Neurosci
 
15
:
359
–375.
Petrides M, Pandya DN (1994) Comparative architectonic analysis of the human and the macaque frontal cortex. In: Handbook of neuropsychology, Vol. 9 (Boller F, Grafman J, eds), pp. 17–58. Amsterdam: Elsevier.
Rajkowska G, Goldman-Rakic PS (
1995
) Cytoarchitectonic definition of prefrontal areas in the normal human cortex: II. Variability in locations of areas 9 and 46 and relationship to the Talairach Coordinate System.
Cereb Cortex
 
5
:
323
–327.
Robinson AL, Heaton RK, Lehman RAW, Stilson DW (
1980
) The utility of the Wisconsin Card Sorting Test in detecting and localizing frontal lobe lesions.
J Consult Clin Psychol
 
48
:
605
–614.
Rogers RD, Monsell S (
1995
) Costs of a predictable switch between simple cognitive tasks.
J Exp Psychol Gen
 
124
:
207
–231.
Rolls ET, Hornak J, Wade D, McGrath J (
1994
) Emotion-related learning in patients with social and emotional changes associated with frontal lobe damage.
J Neurol Neurosurg Psychiat
 
57
:
1518
–1524.
Rushworth MF, Nixon PD, Eacott MJ, Passingham RE (
1997
) Ventral prefrontal cortex is not essential for working memory.
J Neurosci
 
17
:
4829
–4838.
Sakagami M, Niki H (
1994
) Encoding of behavioral significance of visual stimuli by primate prefrontal neurons: relation to relevant task conditions.
Exp Brain Res
 
97
:
423
–436.
Sakagami M, Lauwereyns J, Kobayashi S, Koizumi M, Tsutsui K, Hikosaka O (
1999
) Selective attention to color modulates neuronal activity only for no-go signals in macaque ventrolateral prefrontal cortex.
Soc Neurosci Abstr
 
25
:
1550
.
Smith EE, Jonides J (
1999
) Storage and executive processes in the frontal lobes.
Science
 
283
:
1657
–1661.
Talairach J, Tournoux P (1988) Co-planar stereotaxic atlas of the human brain: 3-dimensional proportional system: an approach to cerebral imaging. Stuttgart: Georg Thieme Verlag.
Verfaellie M, Heilman KM (
1987
) Response preparation and response inhibition after lesions of the medial frontal lobe.
Arch Neurol
 
44
:
1265
–1271.
Weinberger DR, Berman KF, Zec RF (
1986
) Physiologic dysfunction of dorsolateral prefrontal cortex in schizophrenia. I. Regional cerebral blood flow evidence.
Arch Gen Psychiat
 
43
:
114
–124.