Abstract

A dissociable set of regions was active for the executive processing associated with overcoming a prepotent response tendency and task switching. Regions associated with overcoming prepotency were primarily frontal and may be part of a system involved in top-down biasing for conflict reduction. Posterior regions were recruited for switching between tasks and likely play a role in reconfiguring stimulus–response mappings. Precuneus activity was common to both manipulations and may reflect increased visual attention due to more difficult task demands.

Introduction

Cognitive control comprises several sets of processes involved in coordinating and planning actions. The recruitment of such processes occurs under novel, difficult or complex conditions; for instance, when overcoming habitual responses, ignoring irrelevant stimuli or transforming representations. Under such demands, cognitive control serves to direct the processing of goal-relevant information and to schedule actions in order to minimize conflict between potential responses (Schneider and Detwiler, 1987, 1988; Botvinick et al., 2001). Theoretical formulations of component processes include maintenance of goal representations, updating goals, top-down guidance of information-processing, and monitoring performance. Although the frontal lobes and regions of posterior parietal cortex have been implicated in a range of such control processes, the specific contributions of these regions is not clear.

Electrophysiology studies find that neurons in the dorso-lateral prefrontal cortex (DLPFC) fire when relevant information must be maintained across a delay (Goldman-Rakic, 1987; Levy and Goldman-Rakic, 1999). Similarly, functional neuroimaging studies show DLPFC activity increases parametrically with working memory load during the preparatory period of the task when information must be maintained (Cohen et al., 1997). However, this activity is not a reflection of maintenance per se. DLPFC activity also occurs during performance of tasks without a delay period (Merriam et al., 2001; Bunge et al., 2002). This suggests that activity in this region not only increases with maintenance demands, but also as a function of other control operations. For instance, other studies implicate DLPFC activity in allocating attentional resources (MacDonald, et al., 2000), selecting between competing responses (Rowe et al., 2000; Bunge et al., 2002) and overcoming residual inhibition (Dreher and Berman, 2002).

Another region of the prefrontal cortex, the anterior cingulate cortex (ACC), has also been implicated in cognitive control. MacDonald et al. (2000) found that this region is active during the target phase of a task when competing responses induce conflict. It was suggested that activity in the ACC might interact with the DLPFC to adjust cognitive control based on the amount of response conflict induced by the task. Models of a dorsolateral prefrontal–anterior cingulate cortical loop (Botvinick et al., 2001) propose that when response conflict occurs, the ACC monitors response output and signals the DLPFC for increased allocation of attentional resources. This model hypothesizes that the monitoring and signaling function of the ACC allows for modification of attention based on task demands.

While this DLPFC–ACC system has been implicated in tasks requiring that a prepotent response tendency be overcome, its role in task switching has not been established. Various task switching paradigms employ manipulations, such as varying the degree of overlap in stimulus and/or response attributes (Allport et al., 1994; Meiran, 2000), which might induce conflict and, therefore, call for the recruitment of this system. Some studies have found higher DLPFC activity for switch trials compared with repeat trials (Dove et al., 2000; Sohn et al., 2000), while others do not find DLPFC activity for task switching manipulations (Konishi et al., 1998, 1999; Nakahara et al., 2002). Task switching often activates the posterior parietal cortex to a greater extent in switch than repeat trials (Dove et al., 2000; Kimberg et al., 2000; Sohn et al., 2000) and this region has been implicated in stimulus–response (S–R) associations (Bunge et al., 2002).

The behavioral switch cost incurred from task switching paradigms is often used as a measure of cognitive control. However, it is not clear whether control processes are necessarily recruited during task switching. Allport and Wylie (2000) hypothesized that the switch cost may be accounted for by the carryover of previous task processing into the current trial. This view holds that an extra control process is not necessary to account for the switch cost. Meiran et al. (2000) suggest that there are at least three component processes that contribute to the switch cost. They found that increasing the response–cue interval (RCI) reduced the switch cost, which supports the Allport and Wylie (2000) task carryover effect. Further, they confirmed the formulation of Rogers and Monsell (1995) of two contributions to the switch cost, an anticipatory component and a stimulus-triggered component. Meiran et al. (2000) found that although increasing the cue–target interval (CTI) decreased the switch cost by reducing the anticipatory component, CTI increases beyond 600 ms did not lead to an additional reduction in the switch cost. This residual switch cost reflects a stimulus-driven component, which Meiran et al. (2000) implicate in reconfiguring S–R mappings.

The current study employs the task-cueing paradigm used by Meiran et al. (2000), which allows for the independent examination of the three task switching components. The current task design will focus on the anticipatory and stimulus-triggered component processes. During the CTI, the anticipatory or preparatory effect of the switch cost will be examined. This effect will be indexed in the functional magnetic resonance imaging (fMRI) results as additional activity during the CTI for switch trials as compared with repeat trials. The CTI used in the current study (7.5 s) is sufficiently long for preparation for task switching to complete before the target stimulus appears. For this reason, the stimulus-triggered contribution to task switching may be considered independent of any anticipatory effects.

The third component of task switching discussed by Meiran et al. (2000) was the dissipating processing of the previous trial's task set. According to their results, this component should not affect current trial processing when a sufficiently long RCI is used. Meiran et al. (2000) showed that for RCIs >1 s the switch cost reduction becomes very slow (∼1.7 s reduction per 100 ms of RCI increase). While these results suggest that for the current paradigm the previous trial's task set [Rogers and Monsell (1995) used the term ‘task set’ to refer to ‘an effective intention to perform a particular task’. In their words, ‘to adopt a task-set is to select, link, and configure the elements of a chain of processes that will accomplish a task.’ For the purposes of this paper, task set refers to the repertoire of stimuli, responses and processing stages employed by a particular task condition] processing should minimally affect the current task's processing (since the RCI of the current task is 12 s), Allport and Wylie (2000) find that long-term, retrieved task bindings contribute to the switch cost. This retrieval of the inappropriate task's S–R mappings should be present on all switch trials since the same stimulus and response sets are relevant for both tasks.

For the current study, a cued S–R incompatibility task, maximizing the S–R overlap between the two conditions, was chosen in order to increase control demands. Subjects' tendency to perform the prepotent task was set up in several ways. First, the S–R mapping was more intuitive (i.e. the stimulus ‘l’ corresponded to a response with the left finger); second, trials of this condition were more frequent (70% of trials); and third, the first three trials of every block were prepotent trials.

The present study aimed at dissociating frontal and parietal regions involved in overcoming a prepotent response tendency from those involved in task switching. In particular, we expected to replicate the MacDonald et al. (2000) results in an attempt to validate the role of the DLPFC–ACC system in overcoming a prepotent response tendency and conflict reduction. Further, we aimed to determine whether this system is also recruited for task switching. In order to distinguish control processes associated with these two cognitive constructs, two manipulations were performed: (i) prepotency and (ii) task switching. The first manipulation compared differences between prepotent trials and nonprepotent trials for both behavioral and fMRI measures. The second manipulation compared differences between repeat and switch trials for both of these measures.

Each manipulation was performed twice for each trial phase, once for the preparation phase and once for the target (response) phase of the trial. This allowed for the temporal distinction of processes associated with different stages of the trial (e.g. the distinction of processes associated with attentional allocation from those recruited in the monitoring of response conflict for the prepotency manipulation, and the distinction of anticipatory and stimulus-triggered processing for the task switching manipulation).

Further, this task design enabled comparison of processes associated with each phase of the task for the two task manipulations to determine whether similar processing stages are recruited for these two cognitive constructs. In this way, preparatory processes associated with overcoming prepotency could be compared with preparatory processes associated with task switching, and likewise, response-phase regional activity for overcoming prepotency and task switching could be compared.

Methods and Materials

Subjects and Task

Seven male and seven female right-handed subjects, between the ages of 20 and 35 years, participated in the study. One male subject was excluded for excessive movement. All subjects performed a cued S–R incompatibility task.

Stimuli were presented with PsyScope software on a Macintosh computer and viewed through a mirror reflecting a back-projection screen mounted inside the scanner bore. Subjects foveated a centrally located fixation point throughout the task. Figure 1 depicts the S–R incompatibility task. At the beginning of a trial, a colored cue (green or red) appeared in place of the fixation for 500 ms. Seven seconds later a target stimulus (either the letter ‘l’ or ‘r’) replaced the fixation cross, also for 500 ms. The inter-trial interval was fixed at 12 s. Subjects responded to the target by pressing with their index or middle finger on a glove affixed to their right hand.

Figure 1.

Stimulus–response incompatibility task. The timeline across the bottom represents the task timing for each trial. Each tick represents one scan (1500 ms). Cue onset occurs at trial start (0 s). Target onset occurs 7.5 s after trial start.

Figure 1.

Stimulus–response incompatibility task. The timeline across the bottom represents the task timing for each trial. Each tick represents one scan (1500 ms). Cue onset occurs at trial start (0 s). Target onset occurs 7.5 s after trial start.

The colored cue instructed subjects to perform one of the two task conditions. A green cue signaled the subject to perform the prepotent condition, or to make the intuitive S–R mapping. For this condition, subjects responded to the target with their right (middle) finger when it was an ‘r’ and their left (index) finger when an ‘l’ appeared. A red block instructed subjects that the current trial is of the nonprepotent condition, in which the S–R mapping is reversed. In this condition, subjects were to respond with their right (middle) finger to an ‘l’ and their left (index) finger to the appearance of an ‘r’. During the preparatory phase, subjects could prepare the S–R mapping for the relevant task, but not the particular motor response. The relevant response was unknown until the presentation of the target stimulus.

For the switching analysis, a switch trial was defined as one in which the previous trial type was different from the current one (i.e. a prepotent trial preceded by a nonprepotent trial). A repeat trial is one in which the previous and the current trial types are the same. The prepotency of the conditions was further enforced by the frequency of trial type as well as the inclusion of three prepotent trials at the beginning of each block. Seventy percent of trials were of the prepotent condition and 30% were of the nonprepotent condition. After presentation of the first three prepotent trials of each block, trials appeared in a random task order. Each subject performed four blocks of 20 trials.

For each of the analyses, a trial was broken into two phases: the preparatory phase and the response phase. Each trial lasted for 13 scans (see Image Acquisition and Analysis below). The preparatory phase lasted for the first five scans (0–7.5 s after trial start), during which the cue was either presented or maintained for further processing. The target phase lasted from scan 5 through scan 13 (7.5–19.5 s after trial start), during which the target was presented and a response was made to the target.

Image Acquisition and Analysis

Functional and structural scanning was performed in a 3.0 T GE scanner with standard head-coil. Twenty-eight 3.2 mm slices with anterior commissure–posterior commissure (AC–PC) in-line resolution were acquired. On each trial, thirteen 1.5 s functional scans were obtained, using a one-shot spiral T2*-weighted sequence with TR = 1500 ms, TE = 18 ms, flip angle = 70° and a field of view of 20 cm. Structural images were obtained with a standard T1-weighted pulse sequence.

Movement correction was performed using an automated algorithm (AIR) in which the first functional image volume is compared with each subsequent image volume and each image is moved closer to the original along six dimensions. A within-block linear detrend was performed with subtractive mean normalization between blocks. Each subject's structural volume was cross-registered to a reference volume using a 12-parameter automated algorithm (AIR). Volumes were then normalized for image intensity and smoothed (8 mm full width-half maximum Gaussian filter). Talairach coordinates were obtained with AFNI.

Statistical analysis was performed using voxel-wise analysis of variance (ANOVA). First, two planned contrasts were examined: prepotency-by-scan and switch-by-scan ANOVAs. Prepotency-by-scan ANOVAs included subject as a random factor and prepotency and scan as within subject factors. Switch-by-scan ANOVAs were performed with subject as a random factor and switch and scan as within subject factors. For each contrast, scans 1–5 (the preparatory phase) were analyzed separately from scans 5–13 (the response phase). Incorrect trials were discarded before the analyses were performed.

Once ROIs were identified for each manipulation, sensitivity analyses were performed to determine that regional activity associated with one of the manipulations (e.g. overcoming prepotency) did not also occur for the other manipulation (e.g. task switching). Within each ROI discovered for one of the two manipulation types (switch and prepotency), the other analysis was performed on a voxel-by-voxel basis to ensure that no voxels were active for both analyses. This analysis was performed separately for each trial phase. For example, for an ROI found in the preparatory prepotency-by-scan ANOVA, a voxel-wise preparatory switch-by-scan ANOVA was performed. Second, switch-by-scan ANOVAs were performed with prepotent-only trials (excluding nonprepotent trials). While the design and aims of the current study did not allow for a full interaction analysis between switch and prepotency, this sensitivity analyses ensured that regional activity for the switch analysis was not driven by activity on nonprepotent, switch trials.

Results

Behavioral Analyses

On 2.45% of all trials subjects made incorrect responses and on 0.88% of trials no response was made. Error rates were between two and three percent for each condition (prepotent, nonprepotent, switch, and repeat) and the no-response trial rates did not exceed one percent for any condition. Because overall error rates were low and error rates for each condition were comparable, only correct trials were included for both the behavioral and the functional imaging analyses.

The mean reponse time (RT) for the prepotent condition (702, SD = 267) was significantly different from the mean RT for the nonprepotent condition (787, SD = 300 ms, t = 2.79, df = 12, P < 0.05). The conflict effect, a measure of the response conflict that is induced by the nonprepotent S–R mapping (mean nonprepotent RT–mean prepotent RT), was 86 ms.

Subjects also performed significantly worse in the switch trials as compared with the repeat trials (t = 5.00, df = 12, P < 0.0003). The mean RT for trials on which subjects switched conditions was 796 ms (SD = 310) whereas on trials in which subjects performed the same condition in the preceding trial, the mean reaction time was 681 ms (SD = 246). The switch effect, a reflection of extra processing involved in switching between conditions, was 115 ms.

For the green trial only switch analysis, RTs were similar to the overall switch RTs. Green-trial-only repeat RTs were 670 ms (SD = 240) and switch RTs were 786 (SD = 314 ms, t = 4.46, df = 12, P < 0.0008). The switch effect was 116 ms.

For the RT bin analyses, mean RTs and behavioral costs are displayed in Table 1. A 2×2 ANOVA, with subject as a random factor and RT bin and task switch as fixed factors, was performed to determine the effects of preparation on task switching. Results find that the switch cost for the fast RT bin is substantially reduced from that of the slow RT bin. It has been hypothesized that preparation for switch is all-or-none and occurs only on these fast RT trials (De Jong, 2000). The current results support this hypothesis. Fast RT trials have smaller switch costs. An interaction of RT bin and task switch (F = 33.87, df = 1, 12, P < 0.0001) reveal that task switching is greater for the slow RT bin than for the fast RT bin. Further, there were main effects of both RT bin (F = 100.78, df = 1, 12, P < 0.0001) and task switch (F = 33.61, df = 1, 12, P < 0.0001).

Table 1

Behavioral data for RT bin and task switching


 
Repeat
 
Switch
 
Total
 
Switch effect
 
Fast 463 (135) 524 (154) 488 (146) 62 
Slow 1031 (323) 1272 (340) 1132 (350) 241 
Total
 
681 (246)
 
796 (310)
 

 

 

 
Repeat
 
Switch
 
Total
 
Switch effect
 
Fast 463 (135) 524 (154) 488 (146) 62 
Slow 1031 (323) 1272 (340) 1132 (350) 241 
Total
 
681 (246)
 
796 (310)
 

 

 

The interaction of task switch and response switch was examined to determine whether the effects could be attributed to the selection of a particular response. The mean RTs and the behavioral costs are displayed in Table 2. A 2 × 2 ANOVA, with subject as a random factor and task switch and response switch as fixed factors reveals that there was a significant main effect of task switch (F = 24.788, df = 1, 12, P < 0.0001). However, the effect of response switch (F = 0.025, df = 1, 12, P = 0.877) and the interaction of task switch by response switch (F = 0.251, df = 1, 12, P = 0.625) were insignificant.

Table 2

Behavioral data for task switching and response switching

  Response switch
 
   

 

 
Repeat
 
Switch
 
Total
 
Switch effect
 
Task switch Repeat 684 (226) 678 (262) 680 (246) −6 
 Switch 802 (296) 788 (310) 796 (310) −14 
 Total 736 (266) 720 (292)   

 
Switch effect
 
118
 
110
 

 

 
  Response switch
 
   

 

 
Repeat
 
Switch
 
Total
 
Switch effect
 
Task switch Repeat 684 (226) 678 (262) 680 (246) −6 
 Switch 802 (296) 788 (310) 796 (310) −14 
 Total 736 (266) 720 (292)   

 
Switch effect
 
118
 
110
 

 

 

Functional Imaging Analyses

Regions of interest (ROIs) were selected based on either a prepotency-by-scan or a switch-by-scan ANOVA at P < 0.005 with a clustering threshold of eight contiguous voxels in order to reduce the likelihood of type 1 error. This threshold ensured 0.05 probability protection for the full 28 slice volume, using the correction method described by Forman et al. (1995). Only ROIs showing greater activation in the prepotent or switch conditions for their respective analyses are reported. Further, only ROIs with time-series peaks within the hemodynamic range (3–9 s) above baseline are reported. Table 3 displays all ROIs that met these criteria for the prepotency-by-scan analyses.

Table 3

Regions involved in overcoming a prepotent response tendency

Trial phase
 
Brain region
 
BA
 
X
 
Y
 
Z
 
Size (mm3)
 
Size (voxels)
 
Condition F(1,12)
 
Scan F(4,48)
 
Condition × Scan F(4,48)
 
Preparation (scans 1–5) Precuneus −3 −78 40 374.52 12 4.889* 10.338*** 4.841** 
 Left dorsolateral PFC −42 16 36 686.62 22 10.429** 3.382* 7.239*** 
 Right dorsolateral PFC 25 15 32 436.94 14 2.354 1.797 8.151*** 
 Left medial PFC −3 42 28 1154.77 37 25.901*** 2.894* 6.032*** 
 Left caudate  −13 19 312.10 10 11.425** 3.170* 5.425** 
 Left anterior frontal cortex 10 −28 51 842.67 27 6.803* 1.416 7.884*** 
 Insula 44/45 −45 15 1061.14 34 6.242* 1.288 7.833*** 
        Condition F(1,12) Scan F(8,96) Condition × Scan F(8,96) 
Target (scans 5–13) Superior frontal gyrus 10 −1 66 686.62 22 3.605 2.693* 4.299*** 
 Precentral gyrus −27 −13 70 280.89 0.049 12.032* 3.707*** 
 Paracentral lobule −29 71 1092.35 35 0.305 1.459 4.284*** 
 Medial frontal gyrus 6/32 −16 −1 48 1279.61 41 0.414 5.238*** 5.576*** 
 Superior parietal lobule −25 −76 43 624.20 20 6.107* 3.526** 4.875*** 
 Postcentral gyrus 43/40 −45 −11 18 436.94 14 1.660 3.465** 3.856*** 
 Superior temporal gyrus 42 55 −19 12 2278.33 73 3.902 0.986 5.946*** 

 
Lingual gyrus
 
18
 
−13
 
−64
 
5
 
780.25
 
25
 
3.869
 
2.305*
 
4.701***
 
Trial phase
 
Brain region
 
BA
 
X
 
Y
 
Z
 
Size (mm3)
 
Size (voxels)
 
Condition F(1,12)
 
Scan F(4,48)
 
Condition × Scan F(4,48)
 
Preparation (scans 1–5) Precuneus −3 −78 40 374.52 12 4.889* 10.338*** 4.841** 
 Left dorsolateral PFC −42 16 36 686.62 22 10.429** 3.382* 7.239*** 
 Right dorsolateral PFC 25 15 32 436.94 14 2.354 1.797 8.151*** 
 Left medial PFC −3 42 28 1154.77 37 25.901*** 2.894* 6.032*** 
 Left caudate  −13 19 312.10 10 11.425** 3.170* 5.425** 
 Left anterior frontal cortex 10 −28 51 842.67 27 6.803* 1.416 7.884*** 
 Insula 44/45 −45 15 1061.14 34 6.242* 1.288 7.833*** 
        Condition F(1,12) Scan F(8,96) Condition × Scan F(8,96) 
Target (scans 5–13) Superior frontal gyrus 10 −1 66 686.62 22 3.605 2.693* 4.299*** 
 Precentral gyrus −27 −13 70 280.89 0.049 12.032* 3.707*** 
 Paracentral lobule −29 71 1092.35 35 0.305 1.459 4.284*** 
 Medial frontal gyrus 6/32 −16 −1 48 1279.61 41 0.414 5.238*** 5.576*** 
 Superior parietal lobule −25 −76 43 624.20 20 6.107* 3.526** 4.875*** 
 Postcentral gyrus 43/40 −45 −11 18 436.94 14 1.660 3.465** 3.856*** 
 Superior temporal gyrus 42 55 −19 12 2278.33 73 3.902 0.986 5.946*** 

 
Lingual gyrus
 
18
 
−13
 
−64
 
5
 
780.25
 
25
 
3.869
 
2.305*
 
4.701***
 

BA = Brodmann's area.

*

Statistically significant at P < 0.05.

**

Statistically significant at P < 0.01.

***

Statistically significant at P < 0.005.

The preparatory (first five scans) prepotency-by-scan interaction identified mostly frontal ROIs. The bilateral DLPFC (25, 15, 32; −42, 16, 36; BA 9), left medial PFC (−3, 42, 28; BA 32), left anterior frontal cortex (−28, 51, 9; BA 10) and left insula (−45, 15, 7) showed higher activity for the nonprepotent condition than the prepotent condition (see Fig. 2).

Figure 2.

Regions active for preparing to overcome a prepotent response tendency.

Figure 2.

Regions active for preparing to overcome a prepotent response tendency.

Figure 3 displays activity found for the response period prepotency-by-scan manipulation. As expected, a medial frontal region (−16, −1, 48; BA 6/32) showed greater activity for the nonprepotent condition than for the prepotent condition during the target period (last eight scans). This region extended into the left caudal ACC and supplementary motor area, which supports the role of the ACC in the monitoring of response conflict. Regions of the superior frontal cortex (6, −29, 71; −27, −13, 70; 10, −1, 66; BA 6) and a region of the superior parietal lobule (−25, −76, 43; BA 7) were also differentially active for the target period.

Figure 3.

Regions showing activity for overcoming a prepotent response tendency for the target phase of the task.

Figure 3.

Regions showing activity for overcoming a prepotent response tendency for the target phase of the task.

For the switch-by-scan analyses, mostly posterior regions were identified (see Table 4). A precuneus region (−6, −78, 38; BA 7) was more active to switch than repeat trials during the preparatory period (Fig. 4). For the response phase of the task, several posterior ROIs were more active to switch than repeat trials: a left inferior parietal lobule ROI (−45, −58, 33; BA 39/40) and a left precuneus ROI (−23, −76, 43; BA 7) (see Fig. 5). This precuneus region overlapped with the precuneus activity in the prepotency-by-scan response phase analysis. An inferior frontal region (33, 28, 19; BA 45/46) also showed greater response-related switch activity.

Figure 4.

Preparation related task switch activity.

Figure 4.

Preparation related task switch activity.

Figure 5.

Target related task switch activity.

Figure 5.

Target related task switch activity.

Table 4

Regions involved in switching between tasks

Trial phase
 
Brain region
 
BA
 
X
 
Y
 
Z
 
Size (mm3)
 
Size (voxels)
 
Condition F(1,12)
 
Scan F(4,48)
 
Condition × Scan F(4,48)
 
Preparation (scans 1–5) Precuneus −6 −78 38 592.99 19 6.060* 8.995*** 7.131*** 
        Condition F(1,12) Scan F(8,96) Condition × Scan F(8,96) 
Target (scans 5–13) Precentral sulcus −39 −9 51 249.68 3.098 11.875*** 3.918*** 
 Precuneus −23 −76 43 499.36 16 2.792 3.159** 3.790*** 
 Supramarginal gyrus 40 −45 −58 33 280.89 10.533** 0.632 3.915*** 

 
Inferior frontal sulcus
 
45/46
 
33
 
28
 
19
 
436.94
 
14
 
0.161
 
1.24
 
4.318***
 
Trial phase
 
Brain region
 
BA
 
X
 
Y
 
Z
 
Size (mm3)
 
Size (voxels)
 
Condition F(1,12)
 
Scan F(4,48)
 
Condition × Scan F(4,48)
 
Preparation (scans 1–5) Precuneus −6 −78 38 592.99 19 6.060* 8.995*** 7.131*** 
        Condition F(1,12) Scan F(8,96) Condition × Scan F(8,96) 
Target (scans 5–13) Precentral sulcus −39 −9 51 249.68 3.098 11.875*** 3.918*** 
 Precuneus −23 −76 43 499.36 16 2.792 3.159** 3.790*** 
 Supramarginal gyrus 40 −45 −58 33 280.89 10.533** 0.632 3.915*** 

 
Inferior frontal sulcus
 
45/46
 
33
 
28
 
19
 
436.94
 
14
 
0.161
 
1.24
 
4.318***
 

BA = Brodmann's area.

*

Statistically significant at P < 0.05.

**

Statistically significant at P < 0.01.

***

Statistically significant at P < 0.005.

Confirmatory and Exploratory Analyses

In order to dissociate control processes involved in task switching from those involved in overcoming a prepotent response tendency, a voxel-wise, switch-by-scan ANOVA was performed within each ROI found in the prepotency-by-scan ANOVA, and vice versa. This ensured that no voxels that were differentially active for the prepotent and the nonprepotent conditions were also involved in switch related processes. The precuneus activation was the only ROI with voxels significant at the P < 0.05 level in the dissociation analyses. Within this ROI, there were eight voxels that were significant for both analyses.

Activity within a region of interest (ROI) that appears in one analysis but not the other could be driven by a difference in the relative difficulty for the two conditions of each manipulation. However, a t-test of the behavioral difference between the switch effect and the conflict effect, found no detectable difference between these two measures (t = 1.14, df = 12, P > 0.27). In as far as the control process performed by a region is evidenced by these behavioral measures, we can exclude the possibility that the dissociation of regions involved in switching and overcoming prepotency is due to differential recruitment of the same process.

Three additional post-hoc analyses were performed to better determine the nature of switch-related processing and to address possible reasons for the lack of significant prefrontal cortical activity for the task switching analyses. First, a sensitivity analysis was performed. This analysis was a switch-by-scan ANOVA including prepotent trials only. All nonprepotent trials were excluded from the analysis and only current-trial-prepotent repeat trials (prepotent → prepotent) and current-trial-prepotent switch trials (nonprepotent → prepotent trials) were compared. This analysis ensured that regional activity due to overcoming prepotency (activity on nonprepotent trials) did not interact with regional activity due to task switching to produce spurious or null results.

Second, switch-by-scan analyses were performed separately for high-performance and low-performance trials. According to De Jong (2000), switch-related preparation is engaged only on a proportion of trials and switch costs are the average of high-performance, prepared trials and low-performance, unprepared trials. The bin analyses were performed in order to detect switch-related processes that occur earlier or show greater activity on those trials in which participants enter a prepared state (Braver et al., 2003). For this analysis, a three-way RT bin-by-switch-by-scan ANOVA was performed. Activity for trials with the fastest reaction times (the top decile of the reaction time distribution for switch trials and for repeat trials) was compared with activity for the slowest reaction time trials (the bottom decile of the reaction time distributions for switch and for repeat trials).

Third, in order to further examine switching mechanisms, a task switch-by-response switch-by-scan analysis was performed. This response–switch analysis addressed the possibility that prefrontal regions may be involved in the selection of particular responses, and therefore, may show switch-related activity for particular responses rather than for task sets. This analysis was only performed for the response phase of a trial. The particular response is not known during the preparatory period and therefore, should not differentially influence processing for the two conditions.

For the green-trial-only switch analysis, only one ROI was more active in the switch than repeat condition. This ROI was active in the response phase of the task in the inferior parietal lobule (−39 −54 42, BA 40). This ROI overlaps with the ROI found in the all-trial switch analysis however, the center of this region is superior to that of the original parietal region.

For the bin analysis, RT bin-by-switch-by-scan analyses were performed and activity for the two trial phases was examined separately. The superior frontal gyrus (−20, −6, 64; BA 6) showed a significant preparatory interaction with the greatest activity for the fast RT bin, switch condition. For the response phase, only the superior frontal gyrus showed the greatest activity for the fast RT bin, switch condition (17, −8, 60; BA 6). For the slow RT bin, switch condition, regions of precuneus (16, −69, 54; BA 7), middle frontal sulcus (51, 17, 31; BA 9), the middle occipital gyrus (−24, −85, 20; BA 18/19), and the anterior frontal cortex (31, 58, 5; BA 10) showed the greatest activity.

Regions involved in switching between individual responses were also determined. For the 3 way, task switch-by-response switch-by-scan ANOVA, regions of superior frontal cortex (4 27 47; BA 8) and precuneus (−16, −70, 44; BA 7), and the post-central gyrus (−44, −22, 43; BA ¾) showed an interaction of activity for task switching and response switching.

Discussion

The current study aimed at dissociating processes recruited during overcoming a prepotent response tendency and task switching. Both manipulations have been associated with competitive processing and, therefore, may require processes devoted to conflict reduction. For this reason, the role of the DLPFC–ACC system, which has been implicated in conflict reduction, was examined for both of these manipulations. Planned contrasts for prepotency-by-scan and switch-by-scan ANOVAs were analyzed separately. The current study was not designed and did not provide power to examine the interaction of prepotency and scan. Two sensitivity analyses were performed post-hoc to ensure that task switching activity was not merely a reflection of nonprepotent-trial activity.

Task cueing paradigms have proven useful in temporally isolating processes contributing to both prepotency effects (MacDonald et al., 2000) and task switching effects (Meiran et al., 2000). The current study employed this type of task design in order to temporally separate control processes in each of these manipulations. On each trial, the preparatory phase of the task occurred after the appearance of the cue and lasted until the appearance of the target. The response phase began with the appearance of the target and lasted until the end of the trial. In this manner, preparatory processes and response-related processes were examined independently for both the prepotency and task switching manipulations.

The current study found regional activation differences related to component cognitive control processes for overcoming a prepotent response tendency and task switching. The processes recruited during performance of the nonprepotent task set reflect the need for strong top-down biasing and conflict reduction when the tendency to perform the more automatic task set must be overcome. Processes necessary for switching between tasks include changing task representations and reconfiguring S–R mappings for performance of the new, now relevant task condition. The current study also revealed regions active in both manipulations, reflecting general processes recruited commonly while overcoming a prepotent response tendency and task switching.

Overcoming a Prepotent Response Tendency: Top-down Biasing and Conflict Reduction

In the current study, greater DLPFC activation occurred in preparation for the nonprepotent condition. This activity is analogous to the DLPFC activity reported by MacDonald et al. (2000) in that activity occurs for overcoming the more habitual, prepotent task set. However, this activity differed from that of the MacDonald et al. (2000) study in two ways. First, the activation was bilateral, whereas MacDonald et al. (2000) found unilateral left DLPFC activity. Second, the time course of this activity peaked at 3–4.5 s after stimulus presentation and returned toward baseline by the end of the preparatory period. MacDonald et al. (2000) found DLPFC activity maintained throughout the preparatory period. Because DLPFC activity is not sustained throughout the inter-stimulus interval, we exclude the interpretation that this activation is simply a reflection of greater maintenance demands during performance of the nonprepotent condition. This account is consistent with other studies implicating the DLPFC in cognitive control processing other than maintenance (Bunge et al., 2002; Dreher and Berman, 2002; Rowe et al., 2000).

One proposed role for the DLPFC in cognitive control is the selection of a target response (Rowe et al., 2000; Bunge et al., 2002). Bunge et al. (2002) suggested that a repertoire of potential S–R associations is activated within posterior parietal cortex and the DLPFC selects from amongst competing responses. In the current task, DLPFC activity occurred during the preparatory phase of the task. Since the particular response was unknown during this period, the target response could not be prepared at this time. Miller and Cohen propose that the DLPFC performs top-down biasing necessary for overcoming the activation of the dominant task set. Since the dominant task set is primed by more frequent use and since the stimuli in this task set also inherently prime the responses, strong attentional biasing is necessary to activate the nonprepotent task set. Similarly, MacDonald et al. (2000) suggest that the DLPFC is more active in preparation to overcome a prepotent response tendency, when the demands for attentional allocation are increased.

Other regions that displayed greater preparatory activation for the nonprepotent task set were the rostral cingulate, anterior frontal cortex and left inferior frontal cortex. The rostral cingulate activation may be due to the violation of stimulus expectation. The P3b event-related potential component shows amplitude variation as a function of stimulus expectancy and has been implicated in evaluation of infrequent target events in determination of behavioral relevance. In the oddball paradigm, the less frequent the target stimulus, the greater the amplitude of the P300. Studies employing fMRI and event-related potential have implicated regions of the cingulate cortex in a system involved in rare target/novel stimulus detection (Friedman et al., 2001).

Anterior frontal activation has been implicated in the monitoring of internally generated information (Christoff and Gabrieli, 2000; Christoff et al., 2001), cognitive branching (Koechlin et al., 1999) and subgoal activation during working memory processing (Braver and Bongiolatti, 2002). In the nonprepotent condition, this region may be involved in integrating the contents of the subgoal with the contents of working memory in preparation for the upcoming S–R reversal.

Two regions showed greater activity for the nonprepotent condition during the response phase of the trial, the left ACC and the precuneus. The ACC region shows similar activity to the ACC region found in the MacDonald et al (2000) study. The activity was implicated in response conflict detection and the current results support the role of the ACC in monitoring competing response tendencies in order to gauge the amount of control necessary for task performance (Botvinick et al., 2001).

The response-period precuneus activity may be a reflection of greater visual attention to stimulus features during more difficult task demands. Activity in this region was also found in the task switching analyses suggesting that the precuneus performs a task general function.

Task Switching: Cognitive Flexibility

Prefrontal Cortex Activity and Task Switching

A dissociable set of regions was active for the switching analyses from those active for overcoming prepotency. No prefrontal regions were active for the a priori switch preparation analysis. A region of the right inferior frontal cortex did show target-related switch activity, however, activation was overlapping for switch and repeat trials until the last scan (scan 13) of the trial. This suggests that target and response-related switch activity increases for both switch and repeat trials.

Of particular interest was the lack of DLPFC activity for preparation to switch. In the literature, DLPFC recruitment during an endogenously cued task switch is variable, with some studies finding DLPFC activity (Sohn et al., 2000; Konishi et al., 2002; Braver et al., 2003) and others finding no DLPFC activity for such manipulations (Konishi et al., 1999; Nakahara et al., 2002). Many studies use exogenous cueing or predictable task order and therefore, do not distinguish activity due to preparation for a task switch and activity for stimulus-triggered task switching (Dove et al. 2000; Kimberg et al., 2000; Rogers et al. 2000; Dreher et al., 2002; Dreher and Grafman, 2003).

This discrepancy of findings may be due to the recruitment of a variable set of component processes associated with task switching. For example, the Wisconsin Card Sorting Task (WCST), a paradigm commonly employed for set shifting manipulations, involves processes that may not be related to set shifting per se, such as feedback and trial-and-error guessing, but which are recruited during dimensional set shifting manipulations (Rogers et al., 2000; O'Reilly et al., 2002). In order to isolate set shifting processes, Konishi et al. (1999) had subjects perform a cued version of the WCST, which emulated task switching paradigms. They found no DLPFC activity for the set shifting manipulation.

The variability in findings in the task switching literature may also be because the umbrella terms ‘task switching’ and ‘set shifting’ include a variety of shift types (e.g. decision, dimensional shifts, reversal learning, visuomotor transformations). Tables 5 and 6 illustrate the variability in the literature in both the types of task switching paradigms used and in the PFC activity that is elicited during performance of these paradigms. Table 5 is organized separately by frontal and parietal coordinates to facilitate comparison of regions activated in the current study with those activated in prior task switching manipulations.

Table 5

Frontal and parietal regions implicated in previous task switching studies

Study
 
Task
 
Manipulation
 
BA reported
 
Talairach
 
  Terminology
 
Braver et al. (2003) semantic classification task switch/repeat (transient) −28 −66 45 task switch 
Dreher and Grafman (2003) 2 letter discrimination tasks switch/baseline 7/40 −36 −56 52 task switch 
    36 −60 48  
  switch/dual-task 7/40 −36 −60 56  
    48 −48 52  
   −12 −56 68  
  Dual-task>switch −4 −68 24  
Cools et al. (2002) probabilistic reversal learning final reversal err/correct baseline 40 42 −42 40 reversal learning 
Dreher et al. (2002) 2 letter discrimination tasks unpredictable task order 7/40 −32 −56 48 order predictability 
    40 −48 48  
  switch/baseline 7/40 −36 −56 52 task switch 
    36 −60 44  
Konishi et al. (2002) WCST negative feedback (A–B) −74 40 feedback 
   40 −56 −48 42  
  cognitive set shifting (B–C) 12 −58 62 set shift 
    −10 −70 60  
   7/40 −52 −36 42  
Nakahara et al. (2002)  set-shift activation 40 −54 −34 46 set shift 
Rushworth et al. (2001) visuomotor intentional shift task (RS) SR switch vs stay  10 −73 55 SR map switch 
    −29 −69 55  
    −33 −60 54  
    17 −64 52  
    −61 52  
    −11 −63 51  
    −23 −71 50  
    −6 −71 48  
    −57 48  
    −48 −41 48  
    29 −40 45  
    −4 −77 44  
    −38 −44 39  
    −33 −72 38  
   40 50 −35 30  
 visual attentional shift task (VS) stim switch/repeat −10 62 50 attentional 
    −23 −70 49 task switch 
    16 −66 48  
    −33 −54 46  
    −19 −72 42  
    −45 −34 38  
    −34 −59 37  
   15 −66 34  
  RS–VS  −17 −78 53  
    −26 −79 47  
    −29 −75 46  
   40 −53 −38 25  
  VS–RS  −30 −45 54  
    −29 −48 44  
Konishi et al. (2001) intentional encoding (words/faces) block transition effect (transient) 7/40 33 −57 48 restart cost 
   31/7 −43 40  
   40/19 −31 −53 40  
   19/7 11 −71 30  
Dove et al. (2000) SR mapping switch switch/repeat −8 −68 57 task switch 
   40 −32 −50 45  
   −75 42  
Kimberg et al. (2000) Rogers and Monsell (1995) task switch/repeat −30 −52 50 task switch 
   40 26 −64 45  
Rogers et al. (2000) ID/ED shift learning task discrimination performance −32 −64 44 general learning 
    34 −66 44  
  reversal–ID shift 39 −54 −70 34 reversal learning 
Sohn et al. (2000) Rogers and Monsell (1995) task foreknowledge/no foreknowledge 40 −36 −51 41 preparation 
  switch/repeat (no foreknow) 39/40 −37 −72 30 task switch 
Konishi et al. (1998) WCST switch/repeat 40 −45 −36 48 set shift 
    38 −33 39  
Braver et al. (2003)a semantic classification task switch/repeat (transient) −16 63 task switch 
   44/9 −46 15 21  
   45/47 −40 30  
  mixed/single task (sustained) 9/10 22 39 18  
   46/10 34 48 18  
Dreher ans Grafman (2003) 2 letter discrimination tasks switch/baseline −28 60 task switch 
   36 60  
   8/9/44 −52 12 36  
   8/9/46 48 16 40  
    48 44 28  
   28 52  
  switch/dual-task 8/9/44 −60 12 32  
   44/45 60 12  
   10 −28 60 −4  
    48 52  
  Dual-task>switch 32/24 56 28  
    −8 −48 24  
    28 20  
    36 12  
Cools et al. (2002) probabilistic reversal learning final reversal err/correct baseline 32 52 reversal learning 
   47 38 24 −2  
Dreher and Berman (2002) 3 letter discrimination tasks restart cost 32 15 42 restart cost 
   −49 30  
    42 23  
  overcoming residual inhibition 9/46 27 27 27 overcoming inh 
   45 39 19 11  
Dreher et al. (2002) 2 letter discrimination tasks regular timing −28 72 fixed task timing 
   8/9/44 −60 12 32  
   10/46 −36 60 −4  
  predictable task order 10 64 order predictability 
  unpredictable task order 8/9/44 −44 16 32  
   44/45 36 24 20  
  switch/baseline −28 56 task switch 
   8/9/44 −40 20 32  
   8/9/46 48 16 40  
    60 28 32  
  task order × timing −4 12 56 order and timing 
   4/6 −44 −12 32 predictability 
   9/46 40 28 28  
    −36 28 24  
   10 −28 48  
   24 28 16  
Konishi et al. (2002) WCST negative feedback (A–B) 6/8 30 58 feedback 
   38 10 58  
   8/6 18 50  
   36 40  
   6/44 38 40  
   −40 38  
    36 18 36  
   9/8 48 34  
   45/44 46 10 24  
   47/12 36 24 −6  
  cognitive set shifting (B–C) 6/44 −38 36 set shift 
   45/44 −48 12 20  
   46 −40 42 14  
  Hemisphere × effect 45/44 46 10 20  
  L hem dominance for (B–C) 6/44 40 36  
  R Hem Dominance for (A–B) 46 38 38 14  
Nakahara et al. (2002)  Set-shift activation −6 66 set shift 
    −10 12 50  
    −48 −4 50  
   28 52  
   6/44 34 38  
   45/44 46 16 24  
    −52 14 20  
   47/12 −32 20  
    30 26  
    42 12 −2  
Konishi et al. (2001) intentional encoding block transition effect (transient) 6/32 48 restart cost 
   6/44 41 38  
   32 23 32  
   32/24 −9 23 28  
   −29 39 36  
   9/46 29 45 26  
Dove et al. (2000) SR mapping switch switch/repeat 6/32 −8 11 47 task switch 
   9/6/44 −44 37  
   9/6 40 36  
   44/45 −36 20 13  
    28 23  
Kimberg et al. (2000) Rogers and Monsell (1995) task switch/repeat −16 −10 51 task switch 
    −56 −7 33  
   44/45 −30 26 −5  
Rogers et al. (2000) ID/ED shift learning task discrimination-performance −10 14 56 general learning 
    20 46  
    42 10 44  
   −48 14 38  
   45 52 20 22  
   10/46 −42 50 10  
   10 38 54  
   10/11 −24 52 −10  
  reversal–ID shift 56 20 reversal learning 
   24/32 −6 34  
  ED–ID shift −54 44 set shift 
   9/46 16 46 26  
   10 −8 60  
  ED shift-reversal 28 62 set shift 
   9/46 18 44 24  
   47 36 40 −8  
Sohn et al. (2000) Rogers and Monsell (1995) task switch/repeat (no foreknow) 26 23 43 task switch 
  foreknowledge/no foreknowledge 46/45 53 27 preparation 
Konishi et al. (1999) WCST switch/repeat (w/cue) 44/45 inf frontal   task switch 
Konishi et al. (1998) WCST switch/repeat 24/32 −3 25 35 set shift 
   44/45 −40 18 22  

 

 

 

 
39
 
15
 
22
 

 
Study
 
Task
 
Manipulation
 
BA reported
 
Talairach
 
  Terminology
 
Braver et al. (2003) semantic classification task switch/repeat (transient) −28 −66 45 task switch 
Dreher and Grafman (2003) 2 letter discrimination tasks switch/baseline 7/40 −36 −56 52 task switch 
    36 −60 48  
  switch/dual-task 7/40 −36 −60 56  
    48 −48 52  
   −12 −56 68  
  Dual-task>switch −4 −68 24  
Cools et al. (2002) probabilistic reversal learning final reversal err/correct baseline 40 42 −42 40 reversal learning 
Dreher et al. (2002) 2 letter discrimination tasks unpredictable task order 7/40 −32 −56 48 order predictability 
    40 −48 48  
  switch/baseline 7/40 −36 −56 52 task switch 
    36 −60 44  
Konishi et al. (2002) WCST negative feedback (A–B) −74 40 feedback 
   40 −56 −48 42  
  cognitive set shifting (B–C) 12 −58 62 set shift 
    −10 −70 60  
   7/40 −52 −36 42  
Nakahara et al. (2002)  set-shift activation 40 −54 −34 46 set shift 
Rushworth et al. (2001) visuomotor intentional shift task (RS) SR switch vs stay  10 −73 55 SR map switch 
    −29 −69 55  
    −33 −60 54  
    17 −64 52  
    −61 52  
    −11 −63 51  
    −23 −71 50  
    −6 −71 48  
    −57 48  
    −48 −41 48  
    29 −40 45  
    −4 −77 44  
    −38 −44 39  
    −33 −72 38  
   40 50 −35 30  
 visual attentional shift task (VS) stim switch/repeat −10 62 50 attentional 
    −23 −70 49 task switch 
    16 −66 48  
    −33 −54 46  
    −19 −72 42  
    −45 −34 38  
    −34 −59 37  
   15 −66 34  
  RS–VS  −17 −78 53  
    −26 −79 47  
    −29 −75 46  
   40 −53 −38 25  
  VS–RS  −30 −45 54  
    −29 −48 44  
Konishi et al. (2001) intentional encoding (words/faces) block transition effect (transient) 7/40 33 −57 48 restart cost 
   31/7 −43 40  
   40/19 −31 −53 40  
   19/7 11 −71 30  
Dove et al. (2000) SR mapping switch switch/repeat −8 −68 57 task switch 
   40 −32 −50 45  
   −75 42  
Kimberg et al. (2000) Rogers and Monsell (1995) task switch/repeat −30 −52 50 task switch 
   40 26 −64 45  
Rogers et al. (2000) ID/ED shift learning task discrimination performance −32 −64 44 general learning 
    34 −66 44  
  reversal–ID shift 39 −54 −70 34 reversal learning 
Sohn et al. (2000) Rogers and Monsell (1995) task foreknowledge/no foreknowledge 40 −36 −51 41 preparation 
  switch/repeat (no foreknow) 39/40 −37 −72 30 task switch 
Konishi et al. (1998) WCST switch/repeat 40 −45 −36 48 set shift 
    38 −33 39  
Braver et al. (2003)a semantic classification task switch/repeat (transient) −16 63 task switch 
   44/9 −46 15 21  
   45/47 −40 30  
  mixed/single task (sustained) 9/10 22 39 18  
   46/10 34 48 18  
Dreher ans Grafman (2003) 2 letter discrimination tasks switch/baseline −28 60 task switch 
   36 60  
   8/9/44 −52 12 36  
   8/9/46 48 16 40  
    48 44 28  
   28 52  
  switch/dual-task 8/9/44 −60 12 32  
   44/45 60 12  
   10 −28 60 −4  
    48 52  
  Dual-task>switch 32/24 56 28  
    −8 −48 24  
    28 20  
    36 12  
Cools et al. (2002) probabilistic reversal learning final reversal err/correct baseline 32 52 reversal learning 
   47 38 24 −2  
Dreher and Berman (2002) 3 letter discrimination tasks restart cost 32 15 42 restart cost 
   −49 30  
    42 23  
  overcoming residual inhibition 9/46 27 27 27 overcoming inh 
   45 39 19 11  
Dreher et al. (2002) 2 letter discrimination tasks regular timing −28 72 fixed task timing 
   8/9/44 −60 12 32  
   10/46 −36 60 −4  
  predictable task order 10 64 order predictability 
  unpredictable task order 8/9/44 −44 16 32  
   44/45 36 24 20  
  switch/baseline −28 56 task switch 
   8/9/44 −40 20 32  
   8/9/46 48 16 40  
    60 28 32  
  task order × timing −4 12 56 order and timing 
   4/6 −44 −12 32 predictability 
   9/46 40 28 28  
    −36 28 24  
   10 −28 48  
   24 28 16  
Konishi et al. (2002) WCST negative feedback (A–B) 6/8 30 58 feedback 
   38 10 58  
   8/6 18 50  
   36 40  
   6/44 38 40  
   −40 38  
    36 18 36  
   9/8 48 34  
   45/44 46 10 24  
   47/12 36 24 −6  
  cognitive set shifting (B–C) 6/44 −38 36 set shift 
   45/44 −48 12 20  
   46 −40 42 14  
  Hemisphere × effect 45/44 46 10 20  
  L hem dominance for (B–C) 6/44 40 36  
  R Hem Dominance for (A–B) 46 38 38 14  
Nakahara et al. (2002)  Set-shift activation −6 66 set shift 
    −10 12 50  
    −48 −4 50  
   28 52  
   6/44 34 38  
   45/44 46 16 24  
    −52 14 20  
   47/12 −32 20  
    30 26  
    42 12 −2  
Konishi et al. (2001) intentional encoding block transition effect (transient) 6/32 48 restart cost 
   6/44 41 38  
   32 23 32  
   32/24 −9 23 28  
   −29 39 36  
   9/46 29 45 26  
Dove et al. (2000) SR mapping switch switch/repeat 6/32 −8 11 47 task switch 
   9/6/44 −44 37  
   9/6 40 36  
   44/45 −36 20 13  
    28 23  
Kimberg et al. (2000) Rogers and Monsell (1995) task switch/repeat −16 −10 51 task switch 
    −56 −7 33  
   44/45 −30 26 −5  
Rogers et al. (2000) ID/ED shift learning task discrimination-performance −10 14 56 general learning 
    20 46  
    42 10 44  
   −48 14 38  
   45 52 20 22  
   10/46 −42 50 10  
   10 38 54  
   10/11 −24 52 −10  
  reversal–ID shift 56 20 reversal learning 
   24/32 −6 34  
  ED–ID shift −54 44 set shift 
   9/46 16 46 26  
   10 −8 60  
  ED shift-reversal 28 62 set shift 
   9/46 18 44 24  
   47 36 40 −8  
Sohn et al. (2000) Rogers and Monsell (1995) task switch/repeat (no foreknow) 26 23 43 task switch 
  foreknowledge/no foreknowledge 46/45 53 27 preparation 
Konishi et al. (1999) WCST switch/repeat (w/cue) 44/45 inf frontal   task switch 
Konishi et al. (1998) WCST switch/repeat 24/32 −3 25 35 set shift 
   44/45 −40 18 22  

 

 

 

 
39
 
15
 
22
 

 
Table 6

Type of shift for various task switching paradigms

Study Task Manipulation Type of shift
 
  Terminology 

 

 

 
Dimension
 
Stimulus
 
S–R
 

 
Dreher and Grafman (2003) 2 letter discrimination task switch/baseline, switch/dual-task   task switch 
Cools et al. (2002) probabilistic reversal learning task final reversal/baseline   reversal learning 
Dreher and Berman (2002) 3 letter discrimination task switch/repeat, ABA/ABC   task switch, overcoming residual inhibition 
Dreher et al. (2002) 2 letter discrimination task switch/repeat, fixed/random timing, pred/unpred order   task switch, timing and task order predictability 
Gurd et al. (2002) verbal fluency categorical switching    task switch 
O'Reilly et al. (2002) ID/ED shift learning task ED shift   set shift/switch 
  ID shift   set shift/switch 
  reversal shift    
Ravizza et al. (2002) variant of Rogers and Monsell (1995) task   task switch/set shift 
Rogers et al. (2002) ID/ED shift learning task ED shift   set shift 
  ID shift   set shift 
  ED or ID reversal shift   reversal learning 
Rushworth et al. (2002) visuomotor intentional set shift task (RS) SR switch   reversal learning 
 visual attentional set shift task (VS) selective attention switch   visual attention switch 
Cools et al. (2001) variant of Rogers and Monsell (1995) task switch/repeat, crosstalk/no crosstalk   set shift/switch 
Allport and Wylie (2000) standard Stroop Stroop task    
Goschke (2000) variant of Rogers and Monsell (1995) task    task switch 
Meiran (2000) Meiran (2000) task mixed/pure task   task switch 
Sohn et al. (2000)
 
Rogers and Monsell (1995) task (w/cue manip)
 
switch/repeat
 

 
x
 

 
task switch
 
Study Task Manipulation Type of shift
 
  Terminology 

 

 

 
Dimension
 
Stimulus
 
S–R
 

 
Dreher and Grafman (2003) 2 letter discrimination task switch/baseline, switch/dual-task   task switch 
Cools et al. (2002) probabilistic reversal learning task final reversal/baseline   reversal learning 
Dreher and Berman (2002) 3 letter discrimination task switch/repeat, ABA/ABC   task switch, overcoming residual inhibition 
Dreher et al. (2002) 2 letter discrimination task switch/repeat, fixed/random timing, pred/unpred order   task switch, timing and task order predictability 
Gurd et al. (2002) verbal fluency categorical switching    task switch 
O'Reilly et al. (2002) ID/ED shift learning task ED shift   set shift/switch 
  ID shift   set shift/switch 
  reversal shift    
Ravizza et al. (2002) variant of Rogers and Monsell (1995) task   task switch/set shift 
Rogers et al. (2002) ID/ED shift learning task ED shift   set shift 
  ID shift   set shift 
  ED or ID reversal shift   reversal learning 
Rushworth et al. (2002) visuomotor intentional set shift task (RS) SR switch   reversal learning 
 visual attentional set shift task (VS) selective attention switch   visual attention switch 
Cools et al. (2001) variant of Rogers and Monsell (1995) task switch/repeat, crosstalk/no crosstalk   set shift/switch 
Allport and Wylie (2000) standard Stroop Stroop task    
Goschke (2000) variant of Rogers and Monsell (1995) task    task switch 
Meiran (2000) Meiran (2000) task mixed/pure task   task switch 
Sohn et al. (2000)
 
Rogers and Monsell (1995) task (w/cue manip)
 
switch/repeat
 

 
x
 

 
task switch
 

The current study employs a low-level S–R reversal, like that of Dove et al. (2000). However, the current study, failed to produce DLPFC activity for preparation to switch. The absence of higher order dimensional or decisional shifts in the current paradigm may account for the lack of such dorsal PFC activity. The current study did find switch-related inferior frontal activity for the response phase of the task. Sohn et al. (2000) also found inferior frontal activity for task switching; however, this activity was greater for the preparation phase of the task and was related to endogenous preparation for task switching. The discrepancy in results between this study and the current study may be due to the type of shift required. Sohn et al. (2000) used a similar paradigm to that of Rogers and Monsell (1995), which requires a different decision to be made upon a task switch trial.

DLPFC activity has been implicated in response selection (Bunge et al., 2000; Rowe et al., 2000) and may play a role in overcoming prepotency at the level of specific responses rather than at the level of task set. This suggests that DLPFC activity should be reflected in the switching of individual responses rather than the switching of task sets. Three pieces of evidence suggest that this is not the case. First, the current results find DLPFC activity for preparing to overcome a prepotent response tendency. The target response was unknown during this phase of the task. Second, as evidenced by the current response switch analysis, response switching does not elicit DLPFC activity. Third, Dreher and Berman (2002) found that the DLPFC activity, which increased for overcoming residual inhibition, did not interact with motor priming. This led the authors to suggest that lateral PFC is involved with overcoming cognitive-level rather than motor-level prepotency.

Another possible reason for the lack of DLPFC activity in the task switching manipulations is that DLPFC is recruited for both switch and repeat trials, and therefore, the current manipulation is not sensitive to the difference in activity between these two trials types. Studies have suggested that associative retrieval of previous tasks utilizing the current task-stimuli may increase the switch cost for several min after performance of the previous task set (for review, see Allport and Wylie, 2000; Monsell, 2003). Since these long-term priming effects last over several trials, they are likely affecting both the switch and repeat trials in the current study. Allport and Wylie (2000) found that processing associated with the competing task set affected overall RTs as well as switch costs. As with these behavioral results, DLPFC activity may increase for both switch and repeat trials, and while activity may not be differentially greater for switch compared with repeat trials, such activity could be detected relative to a baseline condition of pure task performance. Because the aim of the current study was to detect activity related to task switching per se, and not dual-tasking, the current paradigm was not sensitive to such differential activity.

While dorso-lateral regions of the PFC did not show switch activity for the overall preparatory switch analyses, regions of PFC did show greater switch activity for the response phase of the overall switch analysis and the post-hoc fast and slow bin overall analyses. For the overall switch phase analysis, a region of the inferior frontal cortex (33, 28, 19; BA 44/45), the middle frontal sulcus (51, 17, 31; BA 9), and the anterior frontal cortex (31, 58, 5; BA9) were more active for task switching in the slow bin, response phase analysis. These results differ from Braver et al. (2003). They found inferior PFC activity, which peaked earlier for prepared, switch trials. However, this activity was similar in magnitude for switch and repeat trials in the slow bin analysis, which suggests that this activity may increase for dual-task conditions, and not specifically for switch trials. The current study found prefrontal activity that was superior to that found by Braver et al. (2003). Activity was greater for the slow bin analysis in the response period of the task, which suggests that this activity may be compensating for lack of switch preparation.

The discrepancy in results between the current and Braver et al. (2003) study's inferior frontal activation may be due to the different task demands. The Braver et al. (2003) results required the maintenance of a verbal cue over the cue period of the task, which may have required subvocal rehearsal in both the switch and the repeat conditions. The current study however, required the maintenance and reconfiguration of S–R mappings, which may result in frontal activation only during the retrieval of the S–R mapping. This interpretation is consistent with the pattern of current results.

Activation of Posterior Regions for Task Switching

While no regions of the frontal cortex were more active for preparation to switch in the a priori analyses, posterior regions produced greater activity for these analyses. The precuneus region displayed greater activity for preparation to switch and this ROI overlaps with precuneus activity for the switch analysis in the response period of the task. Activity was greater during the switch than the repeat trials for both phases of the task. Likewise, this activity overlaps with an ROI found in the cue-by-scan response-related manipulation and showed greater activity during performance of the nonprepotent than the prepotent task trials.

Two points of interest may be made about the precuneus activations. First, though activation did not reach threshold in the cue-by-scan preparatory analysis, this region showed subthreshold activation in the dissociation analysis (F > 2.56, df = 1,4, P > 0.05). This suggests that subthreshold increased activity was present in preparation for the nonprepotent condition as well as in the other manipulations. Second, the time course of activity for the response-related switch and overcoming-prepotency manipulations are very similar, suggesting that this region is performing a similar process in both the switch and the nonprepotent task conditions.

These results suggest that precuneus activity increases with the attentional demands for detection of stimuli or stimulus features. Astafiev et al. (2003) found that the precuneus was more active when subjects were required to make a saccade to a region of space than when they were required to simply attend to a target location, and the activation was even greater when subjects were required to point to the target. As in the present study, the authors found that, for both the preparatory and the target-related phases of the task and activity, this pattern of activity was greater during the target-phase. Further, Astafiev et al. (2003) found a small area within this precuneus region that was more active during attention than eye movement preparation. The authors suggest that this may be a region recruited exclusively for shifting attention. This region is near a superior posterior parietal area (Gurd et al., 2002) implicated in the supramodal control of attentional switching, and switching attention between visual attributes (Le et al., 1998). Although these studies suggest that the precuneus may be specific to attentional shifting, the current results and those of other studies suggest that this activity is not exclusively related to attentional shifting.

Several other studies employing S–R incompatibility paradigms have also found superior parietal and precuneus activations for compatibility manipulations (Iacoboni et al., 1996; Dassonville et al., 2001; Merriam et al., 2001; Schumacher and D'Esposito, 2002), many of which overlap with the regions implicated in attentional switching. These results, as well as the current finding of general precuneus activation for the task switching and incompatibility manipulations, suggest that this region is active under increased attentional demands, perhaps for the detection of stimulus features necessary for S–R associations.

Rushworth et al. (2001) found several precuneus and superior parietal activations for two different switching paradigms: visual attentional set shifts (VS), in which the rule for attending to one of two stimuli switched, and visuomotor intentional set shifts (RS), in which the rules for the S–R mapping switched. This activity was greater and more extensive in the RS than the VS paradigm, especially for a region that falls very close to the precuneus activity found in the current study. These results suggest that the processing demands on some regions of the precuneus may be greater during visuomotor transformations, such as in the current study, than simply for the selective attention to stimulus features.

The left inferior parietal lobule activation also showed greater switch-trial activity for the response portion of the task. This activity is consistent with posterior parietal activity in other studies for task switching (Dove et al., 2000; Kimberg et al., 2000; Sohn et al., 2000) as well as parietal activation that has been implicated in S–R associations (Bunge et al., 2002). These results suggest that this region reflects facilitation of S–R reversals during task switching.

The inferior parietal activation found in the current task switching manipulation falls near a posterior region of the lateral intraparietal area (LIP) active for attending, looking and pointing to a target location (Astafiev et al., 2003). Regions of posterior LIP are also activated for switch manipulations (Rogers et al., 2000; Sohn et al., 2000; Rushworth et al., 2001) and have been implicated in visuomotor transformations (Rushworth et al., 2001).

The current findings support the role of the posterior parietal lobe in both the anticipatory and stimulus-triggered components of task switching. The precuneus may contribute to the anticipatory component of task switching by playing a general role in readying the cognitive system for task performance under high attentional demands, while the inferior parietal activation found during the target-phase of the task may reflect the stimulus-triggered component of task switching associated with the reconfiguration of task set.

References

Allport DA, Wylie G (
2000
) Task switching, stimulus–response bindings, and negative priming. In: Attention and performance XVIII (Monsell S, Driver J, eds), pp. 35–70. Cambridge, MA: MIT Press.
Allport DA, Styles EA, Hsieh S (
1994
) Shifting intentional set: exploring the dynamic control of tasks. In: Attention and performance: XV. Conscious and nonconscious information processing (Umilta C, Moscovitch M, eds), pp. 421–452. Cambridge, MA: MIT Press.
Astafiev SV, Shulman GL, Stanley CM, Snyder AZ, Van Essen DC, Corbetta M (
2003
) Functional organization of human intraparietal and frontal cortex for attending, looking, and pointing.
J Neurosci
 
23
:
4689
–4699.
Botvinick, MM, Braver TS, Barch DM, Carter CS, Cohen JD (
2001
) Conflict monitoring and cognitive control.
Psychol Rev
 
108
:
624
–652.
Braver TS, Reynolds JR, Donaldson DI (
2003
) Neural mechanisms of transient and sustained cognitive control during task switching.
Neuron
 
39
:
713
–726.
Braver TS, Bongiolatti SR (
2002
) The role of the frontopolar cortex in subgoal processsing during working memory.
Neuroimage
 
15
:
523
–536.
Bunge SA, Hazeltine E, Scanlon MD, Rosen AC, Gabrieli JDE (
2002
) Dissociable contributions of prefrontal and parietal cortices to response selection.
Neuroimage
 
17
:
1562
–1571.
Christoff K, Gabrieli JDE (
2000
) The frontopolar cortex and human cognition: evidence for a rostrolateral hierarchical organization within the human prefrontal cortex.
Psychobiology
 
28
:
168
–186.
Christoff K, Prabhakaran V, Dorfman J, Zhao Z, Kroger JK, Holyoak KJ, Gabrieli JDE (
2001
) Rostrolateral prefrontal cortex involvement in relational integration during reasoning.
Neuroimage
 
14
:
1136
–1149.
Cools R, Barker RA, Sahakian BJ, Robbins TW (
2001
) Mechanisms of cognitive set flexibility in Parkinson's disease.
Brain
 
124
:
2503
–2512.
Cools R, Clark L, Owen A, Robbins TW (
2002
) Defining the neural mechanisms of probablistic reversal learning using event-related functional magnetic resonance imaging.
J Neurosci
 
22
:
4563
–4567.
Cohen JD, Perlstein WM, Braver TS, Nystrom LE, Noll DC, Jonides J, Smith EE (
1997
) Temporal dynamics of brain activation during a working memory task.
Nature
 
386
:
604
–608.
Dassonville P, Lewis SM, Zhu X-H, Ugurbil K, Kim S-G, Ashe J (
2001
) The effect of stimulus–response compatibility on cortical motor activation.
Neuroimage
 
13
:
1
–14.
De Jong R (
2000
) An intention-activation account of residual switch costs. In: Control of cognitive processes: attention and performance XVIII (Monsell S, Driver J, eds), pp. 357–376. Cambridge, MA: MIT Press.
Dove A, Pollmann S, Schubert T, Wiggins CJ, von Cramon, DY (
2000
). Prefrontal cortex activation in task switching: an event- related fMRI study.
Cogn Brain Res
 
9
:
103
–109.
Dreher J-C, Berman KF (
2002
) Fractionating the neural substrate of cognitive control processes.
Proc Natl Acad Sci
 
99
:
14595
–14600.
Dreher J-C, Grafman J (
2003
) Dissociating the roles of the rostral anterior cingulate and the lateral prefrontal cortices in performing two tasks simultaneously or successively.
Cereb Cortex
 
13
:
329
–339.
Dreher J-C, Koechlin E, Ali SO, Grafman J (
2002
) The roles of timing and task order during task switching.
Neuroimage
 
17
:
95
–109.
Goldman-Rakic PS (
1987
) Motor control function of the prefrontal cortex.
CIBA Found Symp
 
132
:
187
–200.
Goschke T (
2000
) Intentional configuration and involuntary persistence in task-set switching. In: Attention and performance XVIII (Monsell S, Driver J, eds), pp. 331–355. Cambridge, MA: MIT Press.
Gurd MJ, Amunts K, Weiss PH, Zafiris, O, Zilles K, Marshall, JC, Fink GR (
2002
) Posterior parietal cortex is implicated in continuous switching between verbal fluency tasks: an fMRI study with clinical implications.
Brain
 
125
:
1024
–1038.
Forman SD, Cohen JD, Fitzgerald M, Eddy WF, Mintun MA, Noll DC (
1995
) Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold.
Magn Reson Med
 
33
:
636
–647.
Friedman D, Cycowicz YM, Gaeta H (
2001
) The novelty P3: an event-related potential (ERP) sign of the brain's evaluation of novelty.
Neurosci Biobehavioral Rev
 
25
:
355
–373.
Iacoboni M, Woods RP, Mazziotta JC (
1996
) Brain-behavior relationships: evidence from practice effects in spatial stimulus–response compatibility.
J Neurophysiol
 
71
:
321
–331.
Kimberg DY, Aguirre GK, D'Esposito M (
2000
) Modulation task-related neural activity in task switching: an fMRI study.
Cogn Brain Res
 
10
:
189
–196.
Koechlin E, Basso G, Pietrini P, Panzer S, Grafman J (
1999
) The role of the anterior prefrontal cortex in human cognition.
Nature
 
399
:
148
–151.
Konishi S, Nakajima D, Uchida I, Kameyama S, Nakaharra K, Sekihara D, Miyashita Y (
1998
) Transient activation of inferior prefrontal cortex during cognitive set shifting.
Nat Neurosci
 
1
:
80
–84.
Konishi S, Kawazu M, Uchida I, Kikyo H, Asakura I, Miyashita Y (
1999
) Contribution of working memory to transient activation in human inferior prefrontal cortex during performance of the Wisconsin Card Sorting Test.
Cereb Cortex
 
9
:
745
–753.
Konishi S, Donaldson DI, Buckner RL (
2001
) Transient activation during block transition.
Neuroimage
 
13
:
364
–374.
Konishi S, Hayashi T, Uchida I, Kikyo H, Takahashi E, Miyashita Y (
2002
) Hemisheric asymmetry in human lateral prefrontal cortex during cognitive set shifting.
Proc Natl Acad Sci
 
99
:
7803
–7808.
Le TH, Pardo JV, Hu X (
1998
) 4 T-fMRI study of nonspatial shifting of selective attention: cerebellar and parietal contributions.
J Neurophysiol
 
79
:
1535
–1548.
Levy R, Goldman-Rakic PS (
1999
) Association of storage and processing functions in the dorsolateral prefrontal cortex of the nonhuman primate.
J Neurosci
 
19
:
5149
–5158.
MacDonald AW, Cohen JD, Stenger VA, Carter CS (
2000
) Dissociating the role of the dorsolateral prefrontal and anterior cingulate cortex in cognitive ontrol.
Science
 
288
:
1835
–1838.
Meiran N, Chorev Z, Sapir A (
2000
) Component processes in task switching.
Cogn Psychol
 
41
:
211
–253.
Merriam EP, Colby CL, Thulborn KR, Luna B, Olson CR, Sweeney JA (
2001
) Stimulus–response incompatibility activates cortex proximate to three eye fields.
Neuroimage
 
13
:
794
–800.
Miller EK, Cohen JD (
2001
) An integrative theory of prefrontal cortex function.
Annu Rev Neurosci
 
24
:
167
–202.
Monsell S (
2003
) Task switching.
Trends Cogn Sci
 
7
:
134
–140.
Nakahara K, Hiyashi T, Konishi S, Miyashita Y (
2002
) Functional MRI of macaque monkeys performing a cognitive set shifting task.
Science
 
295
:
1532
–1536.
O'Reilly RC, Noelle DC, Braver TS, Cohen JD (
2002
) Prefrontal cortex and dynamic categorization tasks: representational organization and neuromodulatory control.
Cereb Cortex
 
12
:
246
–257.
Ravizza SM, Ciranni MA (
2002
) Contributions of the prefrontal cortex and basal ganglia to set shifting.
J Cogn Neurosci
 
14
:
472
–483.
Rogers RD, Monsell S (
1995
) Costs of a predictable switch between simple cognitive tasks.
J Exp Psychol Gen
 
124
:
207
–231.
Rogers RD, Andrews TC, Grasby PM, Brooks DJ, Robbins TW (
2000
) Contrasting cortical and subcortical activations produced by attentional-set shifting and reversal learning in humans.
J Cogn Neurosci
 
12
:
142
–162.
Rowe JB, Toni I, Josephs O, Frackowiak RS, Passingham RE (
2000
) The prefrontal cortex: response selection or maintenance within working memory?
Science
 
288
:
1656
–1660.
Rushworth, MFS, Paus T, Sipila PK (
2001
) Attention systems and the organization of the human parietal cortex.
J Neurosci
 
21
:
5262
–5271.
Schneider W, Detwiler M (
1987
) A connectionist/control architecture for working memory.
Psychol Learn Motiv
 
21
:
53
–119.
Schneider W, Detwiler M (
1988
) The role of practice in dual-task performance: toward workload modeling in a connectionist/control architecture.
Human Factors
 
30
:
539
–566.
Schumacher EH, D'Esposito M (
2002
) Neural implementation of response selection in humans as revealed by localized effects of stimulus–response compatibility on brain activation
Hum Brain Mapp
 
17
:
193
–201.
Sohn M-H, Ursu S, Anderson JR, Stenger VA, Carter CS (
2000
) The role of prefrontal cortex and posterior parietal cortex in task switching.
Proc Natl Acad Sci
 ,
97
:
13448
–13453.

Author notes

1University of Pittsburgh, Psychology, Pittsburgh, PA 15213, USA, 2Center for the Neural Basis of Cognition, Pittsburgh, PA 15213, USA and 3University of California at Davis, Psychiatry and Psychology, Davis, CA 95817, USA