The neural implementation of task rule activation in the task-cuing paradigm: an event-related fMRI study.

Abstract

To isolate the neural correlates for task rule activation from those related to general task preparation, the effect of a cue explicitly specifying the S-R correspondences (rule-cue) was contrasted with the effects of a cue specifying only the task to performed (task-cue). While the task-cue provides merely information about the type of task, the rule-cue is explicit about both the task type and the task rule (i.e., the set of S-R correspondences). The rule-cue was expected to activate the task rule more efficiently in the preparation period (prior to target presentation); by contrast, in the task-cue condition, part of the task rule activation was expected to be postponed into the task execution period (following the presentation of the target). In an event-related fMRI experiment, we found the right anterior and middle parts of the middle frontal and superior frontal gyri, the right inferior frontal junction, the pre-SMA, as well as the right superior and inferior parietal lobes to show larger activation elicited by the rule-cue than by the task-cue prior to target presentation. Conversely, the results revealed larger activations in these regions in the task-cue than in the rule-cue condition during the task execution period. In summary, this study identified some of the neural correlates of task rule activation and showed that these are a subset of the general task preparation network.

Full text

Page 1

The neural implementation of task rule activation in the task-cuing paradigm:
An event-related fMRI study
Yiquan Shia,b,⁎, Xiaolin Zhoua,c, Hermann J. Müllerb,d, Torsten Schubertb
aCenter for Brain and Cognitive Sciences and Department of Psychology, Peking University, Beijing, China
bDepartment of Psychology, Ludwig-Maximilians-University, Munich, Germany
cKey Laboratory of Machine Perception and Intelligence (Ministry of Education), Peking University, Beijing, China
dSchool of Psychology, Birkbeck College, University of London, UK
a b s t r a c ta r t i c l ei n f o
Article history:
Received 10 June 2009
Revised 11 January 2010
Accepted 25 January 2010
Available online 2 February 2010
Keywords:
Task rule activation
Task preparation
Rule-cue
Task-cue
Cue-only trials
Task switching
fMRI
To isolate the neural correlates for task rule activation from those related to general task preparation, the
effect of a cue explicitly specifying the S–R correspondences (rule-cue) was contrasted with the effects of a
cue specifying only the task to performed (task-cue). While the task-cue provides merely information about
the type of task, the rule-cue is explicit about both the task type and the task rule (i.e., the set of S–R
correspondences). The rule-cue was expected to activate the task rule more efficiently in the preparation
period (prior to target presentation); by contrast, in the task-cue condition, part of the task rule activation
was expected to be postponed into the task execution period (following the presentation of the target). In an
event-related fMRI experiment, we found the right anterior and middle parts of the middle frontal and
superior frontal gyri, the right inferior frontal junction, the pre-SMA, as well as the right superior and inferior
parietal lobes to show larger activation elicited by the rule-cue than by the task-cue prior to target
presentation. Conversely, the results revealed larger activations in these regions in the task-cue than in the
rule-cue condition during the task execution period. In summary, this study identified some of the neural
correlates of task rule activation and showed that these are a subset of the general task preparation network.
© 2010 Elsevier Inc. All rights reserved.
Introduction
The ability to flexibly activate appropriate task rules in situations
with changing task contexts represents an important prerequisite for
successful goal-directed behavior. According to Miller and Cohen
(2001) rule knowledge is processed in the prefrontal working
memory and it contains knowledge about the stimuli, the behavioral
responses, and the context of the situations in which a particular rule
has to be applied. Presumably, the activation of such rule representa-
tions is part of a more general mechanism of task preparation
(Gollwitzer and Sheeran, 2006; Monsell, 2003; Rubinstein et al.,
2001), which includes the prior activation of neural modules
necessary for behavior and starts long before the manifestation of
the overt behavior (Brass and von Cramon, 2002, 2004; Gruber et al.,
2006; Luks et al., 2002; MacDonald et al., 2000; Sohn et al., 2000).
The neural mechanisms of task rule activation in changing task
contexts are still not clear, although there are a number of studies that
have investigated the neural basis of the broader mechanisms of task
preparation. These studies showed cortical regions including the
lateral prefrontal cortex(LPFC), the medialfrontalcortex (MeFC),pre-
motor regions, and parietal regions to be part of a network that comes
into play when participants prepare for an upcoming sensory-motor
task. The present study is aimed at investigating whether regions
specific to the mechanisms of task rule activation can be found.
To start with, studies concerned with understanding the neural
mechanisms of task preparation have often used the task-cuing
paradigm in combination with an event-related fMRI design (Brass
and von Cramon, 2002, 2004; Gruber et al., 2006; Luks et al., 2002;
Sohn et al., 2000). In the task-cuing paradigm, participants are
required to rapidly switch between two different tasks, which leads to
ongoing changes of the relevant task representations including the
corresponding rule knowledge. The current task can either be the
same or different to the preceding task, which is referred to as
repetition or switch condition, respectively. Prior to the onset of the
target, a task-cue is presented that indicates the upcoming task, thus
permitting preparation for the task to be performed next and making
it possible to temporally dissociate task preparation from task
execution (e.g., Meiran, 1996). Using the task-cuing paradigm,
participants' performance (reaction times (RT) and switch costs)
has been shown to benefit from a prolonged cue-target interval (CTI),
which points to their ability for efficient task preparation (Meiran,
2000; Rogers and Monsell, 1995).
Earlier neuroimaging studies investigated preparation-related
activity by analyzing the fMRI activity during very long CTIs (e.g., up
NeuroImage 51 (2010) 1253–1264
⁎ Corresponding author. Department of Psychology, University of Munich, Leo-
poldstr. 13, 80802 Munich, Germany.
E-mail address: Yiquan.shi@campus.lmu.de (Y. Shi).
1053-8119/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2010.01.097
Contents lists available at ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg

Page 2

to 12.5 s) and, therefore, their findings may have been compromised
by memory load confounds (Luks et al., 2002; MacDonald et al., 2000;
Sohn et al., 2000); in particular, the cue-related activity in these
studies may have been related to the maintenance, rather than to the
preparation of the task sets. More recent studies isolated task
preparation-related activity by measuring neural activity separately
for cue-only trials, cue-target trials, and null-events in the task-cuing
paradigm (Brass and von Cramon, 2002, 2004). On cue-only trials,
there is no target following the cue; by contrast, on cue-target trials, a
target is presented that requires the execution of the task; and null-
events represent a baseline condition without any cue and target
information. Because participants do not know in advance whether or
not a target will follow the cue, they have to prepare for task
execution on every type of trial, that is, on both cue-only and cue-
targettrials(seealsoCorbettaetal.,2000; Weissman et al.,2005). This
allows for measurement of preparation-related activity during the
processing of cue-only trials (but see Lavric et al., 2008, for effects of
cue-only trials on the degree of task preparation). Brass and von
Cramon (2002) contrasted activation on cue-only trial and null-event
trials and found a fronto-parietal network to be related to task
preparation. In particular, this network included regions in the
dorsolateral prefrontal cortex (DLPFC), for example, near the inferior
frontal junction point (IFJ), regions surrounding the intraparietal
sulcus (IPS), in the dorsal premotor cortex, and in the pre-
supplementary motor area (pre-SMA) of the medial frontal gyrus.
Although these findings provided a number of valuable insights
into the functional neuroanatomy of task preparation, they are not
unequivocal regarding the neural correlates of task rule activation.
This is so because a task-cuing paradigm such as that used by Brass
and von Cramon (2002) does not permit the mechanisms of activating
the specific task rules to be distinguished from rather general task
preparation (see also Ruge et al., 2009). The presentation of the cue
informed participants about the task they had to perform later upon
the presentation of the target. If the time was sufficient and the
participants intended to do so, they could either activate the current
task rule or, alternatively, they could wait with the activation of the
task rule until the presentation of the imperative stimulus. Thus,
depending on participants' strategy, either to prepare the task rule
early upon the presentation of the cue or only later upon the
presentation of the target, the point in time when the task rule was
activated was not sufficiently controlled.
For the present study, we used what we refer to as a rule-cue,
which differs from the task-cue in earlier neuroimaging studies
because it conveys information not only about the task (e.g., classify
color), but also about the S–R rule to be applied (e.g. red → press left
button, yellow → press right button; see also Logan and Bundesen,
2003). Thus, while a task-cue (the sort of cue used in earlier studies)
conveys only general information about which task to perform, the
rule-cue provides also specific information about task rules, that is,
the stimulus–response (S–R) mapping, on the upcoming trial (task-
and-rule information). By administering the rule-cue randomly mixed
with task-cues, we aimed to trigger processes related to the activation
of the specific task rules during task processing.
In particular, participants were presented with either a color or
a gender discrimination task, with the particular task specified by
the presentation of a cue before the target stimulus. In the rule-cue
condition (Fig. 1, left panel), we displayed the Chinese symbols
“颜色” (for color) or “性别” (for gender) to indicate the upcoming
task, and also the specific instructions of its S–R mapping rule. For
example, if the task was gender discrimination, the symbols “男”
(for male) and “女” (for female) were presented above the
corresponding response keys (e.g., “male” was shown above the
left key and “female” above the right key).
In the task-cue condition, we also used the symbols “颜色” (color)
and “性别” (gender) to indicate the next task, whereas there was no
specific information about the task rule (see Fig. 2). Instead, only non-
informative words “按 键” (press key) were presented below the task-
cues, in order to make the cue display similar to that in the rule-cue
condition.
Similar to Brass and von Cramon (2002), we presented cue-only
trials (Fig. 1, right panel), null-events, and cue-target trials (Fig. 1, left
panel). While an analysis of the cue-only trials allows for detection of
preparation-related activation that is elicited by the cues (rule-cues
and, respectively, task-cues), target-related processes are revealed by
contrasting activity between cue-target and cue-only trials (e.g., Brass
and von Cramon, 2002; Weissman et al., 2005). Because the activation
on cue-target trials consists of activation related to cue- and to target-
processing, subtracting the cue-related activation from the activation
on cue-target trials will leave the target-related activation only.
The distinction between rule-cues and task-cues permits rule-
related neural activity to be analyzed in the following manner. First of
all, we expected a significant performance benefit from the presen-
tation of rule-cues compared to task-cues and we expected rule-cues
toevokestrongercue-relatedactivationthantask-cues,specificallyon
cue-only trials. The reason for the latter hypothesis is that, in the rule-
cue condition, the cue provides explicit rule information and this
information may be activated by the cue presentation. In contrast, in
the task-cue condition, participants may postpone at least part of the
rule activation processes until later, for example, up to the time where
the target is expected to appear. And even if activation of the rule is
not postponed, it may be less effective because the cue provides no
explicit information as to the precise task rule. Consequently, rule-
related activation should be manifest during the preparation period
on cue-only trials in terms of an increased amount of activity in the
rule-cue, compared to the task-cue, condition.
The converse pattern (of activation in rule-cue and task-cue
conditions) may be expected when considering the rule-related
activation that emerges after target presentation on cue-target trials,
that is, during task execution. It is reasonable to assume that, if
participants failed to activate the (complete) task rule right upon cue
presentation, they must activate the necessary S–R mapping rule
following target presentation (Gruber et al., 2006). This would be
consistent with Gruber et al. (2006) who analyzed the neural activity
under conditions of short versus long cue-target intervals (CTIs) in a
task-cuing paradigm. While the time for preparing the upcoming task
was sufficient after cue presentation with long CTIs, it was insufficient
with short CTIs. The latter led to the postponement of (at least parts
of) the preparation processes until after target presentation, as
indicated by an increased amount of neural activity in preparation-
related brain regions under conditions of short compared to long CTIs
(Gruber et al., 2006; see also Brass and von Cramon, 2002). In analogy
to these findings, we expected postponed rule activation in the task-
cue condition compared to the rule-cue condition. This should lead to
greater activation in rule-related brain regions under task-cue,
compared to the rule-cue, conditions upon target presentation on
cue-target trials.
In summary, we expected stronger neural activity related to task
rule activation in the rule-cue compared to the task-cue condition
during the preparation period, and stronger activity in rule activation-
related regions during task execution in the task-cue compared to the
rule-cue condition. The common neural substrate in these two
comparisons thus represents those brain regions that are important
for the process of task rule activation in either the task preparation or
the task execution period; this should be revealed by means of a
conjunction analysis of the corresponding contrasts.
Method
Subjects
Fifteen right-handed, healthy students of Peking University
(recruited by advertisement in the campus Bulletin Board System)
1254
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 3

Fig. 1. Illustration of the task situation. Upper part: Left panel shows a cue-target trial (example for the gender discrimination task). Right panel shows a cue-only trial (example for
gender discrimination task). The lower part of the figure represents the cue displays and their English translation. The cue could either be a rule-cue (left) or a task-cue (right) (for
details see Fig. 2).
Fig. 2. Illustration of the rule-cue and task-cue displays (left panel) and their English translation (right panel). In the rule-cue and the task-cue conditions, the current task was
indicated by the words “颜色” (color) and “性别” (gender), respectively. In the rule-cue condition (upper row), additional information indicated the assignments of the response keys
to the stimulus categories male and female in the gender task and yellow and red in the color task.
1255
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 4

participated in the study. Six participants were female; participants'
ages ranged between 20 and 26 years, and all had normal or
corrected-to-normal vision. Prior the fMRI scanning session, they
gave informed consent about the investigation according to the
Helsinki guidelines and the approval of the Academic Committee of
the Department of Psychology, Peking University. Participants were
paid 50 yuan (about EUR 5) for their service.
One participant's response error rate was more than 20%. Hence
this participant's behavioral and fMRI data were removed from the
data set. There was also a loss of the behavioral data from one
participant, due to data recording error. Thus, ultimately, 14
participants' image data sets and thirteen participants' behavioral
data sets were available for analysis.
Design
Paradigm and procedure
The task to be performed by the participants was either color
discrimination or gender discrimination. Each trial began with the
presentation of a cue for a fixed duration of 1200 ms, which could
either be a rule-cue or a task-cue (Fig. 1). Both cues displayed an
instruction for the upcoming task; however, a precise instruction
about the required task rule was provided only in the rule-cue
condition (for more details, see Fig. 2). On cue-only trials (n=160
trials, of which 80 presented a rule-cue and 80 a task-cue), there was
no target following the cue offset, but only a black screen that lasted
for 600 ms, and there was no need for participants to make a response
(Fig. 1, right panel).
In contrast, on cue-target trials (n=280, of which 140 presented a
rule-cue and 140 a task-cue), the cue was followed by a colored face
picture that was presented for 600 ms; during this period, the task-
cue instruction remained visible on the screen (above the target
picture) by presenting the words ‘gender’ or ‘color’, so as to reduce
participants' working memory load for maintaining the task goal in
thetwoconditions.Importantly,theinformationpresentedduringthe
executionperiod concerned only thetask information andnotthe rule
information because the symbols ‘press key’ and the symbols
illustrating the rule information were not presented during the
execution period (see Fig. 1, left panel). Participants were to respond
to either the color or the gender of the face depicted in the target
display, depending on the instruction of cue. Participants made two-
alternative forced-choice responses using either their left or right
thumbs, with response sets counterbalanced across participants. After
the offset of the target picture, a black screen was presented for a
variable interval of 1000, 1200, 1400, 1600, or 1800 ms. The next trial
could then either be a cue-target or a cue-only trial, that is an ‘event
trial’, or a ‘null trial’ (n=110) in which there was neither a cue nor a
target event. Together with the duration of the null trials, which were
of the same duration as the task trials, the interval between two event
trials (the interval between the disappearance (offset) of the target in
the present trial and the appearance (onset) of the next cue) resulted
in 2200 ms on average.
Task conditions and trial types
The present study used a 2×2 event-related fMRI design. The first
factor was cue type: the cue could be either a rule-cue or a task-cue
(Figs. 1 and 2). The second factor was task transition: the task was
eitherrepeatedorswitched relative totheprecedingtrial.Basedonthe
instructioncuepresentedpriortothetarget,participantswererequired
to distinguish either the color or the gender of the face pictures. If the
current taskwasdifferentfromthe precedingone,the current trialwas
classified as a switch trial; if the current task was identical with the
previous one, the current trial was classified as a repetition trial. This
factor was examined because rule activation (or retrieval) was hypo-
thesized to differ between task repetition trials and switch trials (Mayr
and Kliegl, 2000; Rogers and Monsell, 1995; Monsell, 2003; Rubinstein
et al., 2001). That is, this factor was introduced to examine whether or
not preparation for a switched, compared to a repeated, task leads to a
modulation of the task rule activation.
Each one of the four conditions (rule / task-cue×task switch /
repetition) consisted of 40 cue-only trials and 70 cue-target trials. In
sum, there were 440 event trials, the order of which was unpredict-
able for the participants. In addition, the event trials were randomly
intermixed with 110 null trials in which only a black screen was
shown. The length of a null trial varied from 2800 ms to 3600 ms,
which was similar to the length of the other (task) trials.
For each condition, the cue-related activation can be assessed by
measuring the activation on the cue-only trials, whereas target-
related activation can be assessed by calculating the contrast between
the activation in corresponding cue-target minus cue-only trials.
Stimulus and response conditions
On cue-only trials, only a black screen (i.e., no target) was
presented after the presentation of the cue and there was need to
respond. On cue-target trials, the target stimulus was a colored face
picture. In order to create colored face pictures we merged each one
of the original black–white face pictures (two males and two
females) with same-sized, faded red rectangles (RGB 187- 124-
106) and yellow rectangles (RGB 179- 155- 111) with Photoshop
software. As a result eight colored face pictures were created, which
we used as target stimuli: two yellow male faces, two red male
faces, two yellow female faces, and two red female faces (with the
same face presented in either red or yellow on different trials).
Participants were informed by the cue to respond to either the color
or the gender of the face. The stimuli (cue and target stimuli) were
located on a black background in the center of the screen and
subtended 5 degrees of visual angle.
Participants used their left and right thumbs for response. They
were instructed to respond as fast and as accurately as possible. For
half the participants, the S–R mapping rule was male-left, female-
right and yellow-left, red-right. This was reversed for the other half:
female-left, male-right and red-left, yellow-right.
fMRI measurement
Imaging was performed with a SIEMENS TRIO 3-Tesla scanner at
the Beijing MRI Center for Brain Research. T2⁎-weighted echo-planar
images (EPI) with blood oxygenation level-dependent contrast were
acquired (TR=1500 ms, TE=30 ms, flip angle=90°, voxel
size=3.4×3.4×5 mm3, matrix size=64×64 voxels). Twenty-six
axial slices (thickness=4 mm, spacing=1 mm) were acquired
parallel to the AC-PC plane, covering the whole cortex and part of
the cerebellum. The order of acquisition of the slices was interleaved.
The first five volumes (dummy volumes) were discarded because of
possible instabilities in the magnetic field at the beginning of a scan.
Stimuli were displayed on a back-projection screen mounted in the
bore of the magnet behind the participant's head by using an LCD
projector. Participants viewed the screen by wearing mirror glasses.
fMRI data analysis
Preprocessing
Preprocessing of the functional images was carried out using SPM2
(Wellcome Department of Cognitive Neurology, London, UK). Images
were interpolated in time (temporal realignment to the middle slice).
In addition, they were spatially realigned to the first volume for head
movement correction, unwrapped, and then normalized to the
standard SPM2 EPI template in MNI space (resampled to
2×2×2 mm3isotropic resolution) with default normalization
estimation. The data were then smoothed with a Gaussian kernel of
8-mm full-width half-maximum to account for inter-subject anatom-
ical variability.
1256
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 5

Then the image data were modeled by applying a general linear
model (Friston et al., 1995). In event-related single-subject analyses,
the four cue-only and the four cue-target conditions were modeled as
separate volumes (resulting from the factorial combination of the two
cue type (rule-cue vs. task-cue) and the types of task transition (task
switch vs. task repetition). Additionally, all error trials were selected
to form an error trial volume. The resulting nine volumes were
convolved with the hemodynamic response function (HRF), and then
beta values of these regressors were estimated according to the
ordinary least-squares (OLS) method.
Whole-brain analyses
For group statistics, one-sample t-tests of contrast maps across
subjects (random-effects model treating subjects as a random
variable) were computed to indicate whether observed differences
between conditions were significantly different from zero.
In particular, two main contrasts were calculated: Contrast 1: For
cue-only trials, rule-cue minus task-cue trials, intended to isolate
extra activation for a rule-cue. Contrast 2: (cue-target trials minus
cue-only trials for task-cues) minus (cue-target trials minus cue-only
trials for rule-cues), intended to isolate the extra activation related to
the target-processing when the cue did not specify the rule. In a
subsequentconjunction analysis,SPM5 (Nicholset al.,2005) wasused
to locate the common task rule-related activation between these two
main contrasts. Only those voxels were reported as active which
proved to be significant for both contrasts, Contrast 1 AND Contrast 2;
that is we tested for a rejection of the conjunction null hypothesis:
voxel (not activated in Contrast 1) OR (not activated in Contrast 2).
The way in which the remaining statistical contrasts were
calculated is detailed in the Results section. Unless stated otherwise,
for one-sample t-tests, we used a statistical threshold of pb0.001,
uncorrected, covering at least 10 contiguous voxels. This threshold
was commonly used in the studies of rule-related processing and cue-
related processing (e g., see Bunge et al., 2003; Crone et al., 2006a,b;
Gruber et al., 2006; Wendelken et al., 2008). We also checked all
reported activation foci with a small volume correction procedure
(10 mm sphere centered at the voxel with local maximum activation).
If not otherwise noted, then the reported foci prove significant at a
threshold of pb0.05 (small volume corrected on both the voxel and
the cluster level). For the conjunction analysis, the statistical
threshold was pb0.005, uncorrected, again spanning at least 10
contiguous voxels.
Results
Behavioral results
Fig. 3 presents group means of the RTs (left panel) and error
rates (right panel) as a function of task transition, for the two types
of cue. Mean RTs and error rates were submitted to a 2×2
repeated-measures ANOVA with the factors task transition and cue
type. RTs were significantly faster in the rule-cue than in the task-
cue condition (main effect of cue type, F(1,12)=6.71, pb0.05),
which indicates that participants effectively utilized the rule-cue
information during the preparation period following cue presenta-
tion. The RT advantage for rule-cue compared to the task-cue
presentation (i.e., the ‘behavioral rule-cue effect’) was 17 ms. In
addition, RTs were significantly slower for task switch than for task
repetition trials (main effect of task transition, F(1,12)=12.96,
pb0.005), with switch costs amounting to 25 ms. With mean switch
costs of 24 and 27 ms in rule-cue and task-cues conditions,
respectively, the interaction effect between cue type and task
transition was not significant (F(1,12)=0.11, pN0.7).
The error rate ANOVA revealed a significant main effect of task
transition (F(1,12)=60.91, pb0.0001): more errors were made on
task switch than on task repetition trials. Additionally, a significant
interaction between cue type and task transition was obtained
(F(1,12)=8.84, pb0.05). Further analyses with separate t-tests
revealed elevated error rates in switch compared to repetition trials
in the rule-cue and task-cue conditions (both t's(12)N4.00, both
p'sb0.005), and larger switch costs (error rate switch–error rate
repetition) in the task-cue (error rate=6.6 %) compared to the
rule-cue condition (error rate=3.8 %) (t(12)=2.97, pb0.05). Thus,
as with the RT data, the error data indicated that participants'
performance benefited from the presentation of the rule-cue as
compared to the task-cue. This benefit was especially pronounced in
conditions in which participants had to switch between the tasks as
revealed by the increased error rate in the switch compared to the
repetition condition.
Imaging results
Cue-related activation in rule-cue and task-cue conditions
To identify the cue-related activation, we calculated the main
effect for the cue-only trials separately for the rule-cue and task-cue
conditions by fitting the empirical fMRI data to the hemodynamic
response function (HRF) described above. The resulting beta values
are presented in Fig. 4. Both the presentation of rule-cues and of
task-cues elicited neural activations in a large cortical network, with
foci in the MeFC, bilateral regions of the LPFC near the IFJ, and the
dorsal and the lateral premotor cortex. Additionally, the medial and
lateral parietal lobe, the posterior cingulate cortex (PCC), and the
thalamus showed significant activation. Finally, there was bilateral
activation in the occipital cortex. In addition to these activation foci
which were similar for the two types of cue, two small clusters
were activated by the rule-cue in the right and left anterior
prefrontal cortex (aPFC). Note that, with a more liberal statistical
threshold of pb.005, these two clusters also showed activation
under task-cue conditions.
In summary, the two types of cue activated highly overlapping
brain networks, that is, the preparation processes associated with
rule-cues and task-cues are mediated by similar brain regions.
Fig. 3. Reaction time (RT) and error rates as function of task transition and cue type.
1257
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 6

Analysis of rule-related activation during the task preparation and
execution periods
As outlined in the Introduction, we expected stronger activation
on rule-cue comparedto task-cuetrials during the preparation period,
and, additionally, we assumedthese regions to be associatedwith task
rule activation. To examine for this, we calculated a whole-brain
contrast of the activation in the rule-cue versus the task-cue condition
specifically for cue-only trials. The results of this analysis are
presented in Fig. 5 and Table 1.
Stronger activation was found for the rule-cue compared to the
task-cue condition in the anterior part of the superior frontal gyrus
(SFG), that is, the right aPFC, bilaterally in the premotor cortex, and in
regions of the MedFC; the latter regions extended from anterior
portions in the pre-SMA to posterior portions of the pre-SMA/SMA
region. In addition, we found increased activation in the right superior
parietal lobe (SPL) and the left precuneus. Finally, activation foci were
found bilaterally in the occipital cortex (e.g., in the lingual gyrus and
the fusiform gyrus) (see Fig. 5a and Table 1).
During task execution on cue-target trials, we expected stronger
activation in the task-cue, compared to the rule-cue, condition in
cortical regions that are associated with the activation of the task
rules; this is because of the expected postponement of the rule
activation under task-cue conditions. To determine the corresponding
activation foci, we contrasted the target-related activation during the
execution period in the task-cue and rule-cue conditions. For this
purpose, we calculated the contrast: cue-target–cue-only trials
separately for the task-cue and rule-cue conditions, so as to derive
the corresponding task execution-related activations in both types of
trial. Subsequently, we calculated the second-order contrast, task-cue
(cue-target minus cue-only trials)–rule-cue (cue-target minus cue-
only trials), to compare the target-related activation between the
task-cue and rule-cue conditions.
This analysis revealed stronger target-related activity in the task-
cue compared to the rule-cue condition in most regions that had
proved to be rule-related during the preparation period in the above
analysis (see Cue-related activation section). In particular, these
regions were the right anterior part of the SFG (i.e., aPFC), the right
pre-motor cortex, the MeFC (i.e., pre-SMA), the right SPL, and the
bilateral lingual and fusiformgyri. In addition to these regions, activity
was found in the LPFC, with peak activation in the right posterior MFG
that extended into the IFJ (see Fig. 5b and Table 1).
Subsequently, we performed a conjunction analysis in order to
identify the regions commonly associated with task rule activation
during the preparation and the execution period (see Fig. 5c and
Table 2). This analysis was calculated across the contrasts rule-cue
minus task-cue of the cue-related activation in the preparation
period, and task-cue minus rule-cue of the target-related activation
in the execution period (see the two analyses above).
This analysis revealed common activation foci in the right LPFC
extending from anterior to posterior portions of the LPFC regions near
the IFJ and in anterior and more posterior medial regions of the SFG
(pre-SMA/SMA) and the MeFG. Furthermore, the two contrasts
exhibited common activity in the right SPL extending into inferior
parts of the parietal cortex (IPL), as well as common activation foci in
the bilateral lingual gyrus (see Fig. 5c and Table 2). Note that there are
some regions that showed activation in the conjunction analysis but
not in both of the two single contrasts (pb0.001, for clusters of 10
contiguous voxels); e.g., the right inferior frontal junction (Fig. 5a),
and the middle part of the dorsolateral prefrontal cortex (Figs. 5a and
b). However, these regions showed significant activation foci in the
two single contrasts of 5a and 5b, when using a more liberal threshold
of pb0.005.
We propose that these regions, which proved to be activated in the
conjunction analysis, are associated with processes of task rule
Fig. 4. Illustration of the brain activation elicited by the presentation of the cue in cue-only trials in the rule-cue condition (top) and in the task-cue condition (below). The resulting
cue-related activation across the two cue conditions is associated with the general mechanism of task preparation.
1258
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 7

activation either during the preparation period or later, during the
execution period subsequent to target presentation.
Neural activation in the preparation period and the need to prepare for a
switch
Although the aim of the present study was to understand the
neural correlates of task rule activation, the adopted paradigm
allows us also to investigate the processes genuine to task-
switching situations in which participants alternate between
different task rules. Therefore, we also examined whether the rule
activation in the preparation period is modulated by the need to
prepare for a task switch, compared to a repetition. For this analysis,
the switch-related activity on cue-only trials (collapsed across cue
types) was examined by calculating the contrast: cue-only (switch–
repetition). This contrast revealed cortical activation to be increased
only in the medial SFG (pre-SMA) for the preparation for a task
switch versus a repetition, as is illustrated in Fig. 6. In an additional
ROI analysis, we aimed to test whether the need to process a rule-
cue or a task-cue leads to any additional modulations of the neural
activity in this switch-related region during the preparation period.
In order to decrease the second-order error of overlooking a
possible modulation effect of the switch-related activity, we
selected an ROI that depended on the particular switch-repetition
contrast; according to Kriegeskorte et al. (2009), this way of ROI
selection increases the probability of finding any effects related to
the depending contrast (i.e., in the present case, the switch-
repetition contrast). In more detail, we defined an ROI consisting
of 11 active voxels surrounding the local-peak voxel in the contrast:
cue-only (switch–repetition), and for this ROI, we extracted the
beta values individually for each participant in the rule-cue and
task-cue conditions dependent on the task transition (switch vs.
repetition). The data are presented in Fig. 6. A 2×2 repeated-
measures ANOVA of the beta values revealed significant main
effects of the factors task transition ( F(1,13)=5.07, pb0.05) and
cue type ( F(1,13)=17.82, pb0.001), but no significant interaction
(F(1,13)b1). The non-significance of the interaction means that the
need to prepare for a task switch (compared to a less demanding
task repetition) affects the activation in the medial SFG on cue-only
trials to the same degree in the rule-cue and the task-cue condition.
In other words, the need to process a rule-cue or a task-cue does
not modulate the switch-related activation in the medial SFG during
the preparation period.
Fig. 5. Cortical activation associated with rule activation in the task-cuing paradigm. (a) Significant activation in the comparison of rule-cue versus task-cue for cue-only trials. (b)
Brain regions which show increased target-related activation in the comparison of task-cue versus rule-cue trials (for details see text). Note that target-related activation is observed
when contrasting activation in cue-target minus cue-only trials. (c) The brain regions which are observed in the conjunction analysis across the contrasts illustrated in a and b. In the
conjunction analysis, we used a criterion of pb0.005. Note that there are some regions that showed activation in c but not in a or b (thresholded with pb0.001, for clusters of ten
contiguous voxels); e.g. the right inferior frontal junction (a), and the middle part of the dorsolateral prefrontal cortex (a and b). However, these regions showed significant
activation in the single contrasts of a and b, when using a more liberal threshold of pb0.005. For further details, see Tables 1 and 2.
1259
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 8

Discussion
The present study investigated the functional neuroanatomy of
task rule activation as a component process of general task
preparation. In order to modulate the degree of task rule activation,
we adopted the task-cueing paradigm and presented either rule-cues
or task-cues to enable participants to prepare for the upcoming task.
The rule-cues provided explicit information not only about the type of
task to be performed, but also the specific S–R rule to be applied. We
expected that, when the task was indicated by a rule-cue, the
corresponding task rule should be activated,upon cue presentation, to
a higher degree of ‘preparedness’ compared to when the task was
indicated by a task-cue, which (relative to rule-cues) presents only
nonspecific information about the type of task to be performed next.
This hypothesis was supported by a behavioral performance advan-
tage in terms of both RTs and error rates deriving from the
presentation of a rule-cue as compared to a task-cue.
The present paradigm permits task rule-related brain regions to be
identified by analyzing the effects of rule-cues and task-cues on brain
activity separately for the preparation and execution periods of the
task. In the preparation period, rule-activation-related regions should
be activated more strongly following the presentation of rule-cues as
compared to task-cues. Conversely, for the execution period, rule-
related activation would be expected specifically upon target
presentation if task rules were not activated sufficiently during the
preparation period.
In line with these predictions, the conjunction analysis revealed
similar fronto-parietal networks of activation foci in the
corresponding contrasts, that is, the contrast of rule-cue minus task-
cue for cue-only trials (preparation period) and the contrast
comparing target-related activation on task-cue versus rule-cue trials
(execution period). The common activation foci in these two contrasts
included the anterior and middle parts of the right MFG and SFG, the
posterior region of the MFG near the IFJ, regions in the medial SFG
extending from anterior to posterior portions of the pre-SMA, as well
as the right SPL and IPL. All these activations conformed to the pattern
expected for cortical regions that are correlated with the mechanisms
underlying task rule activation.
Task preparation and task rule activation
In the present paradigm, general preparation-related activation is
reflected in the activity elicited by cue presentation on cue-only trials,
for both the rule-cue and task-cue conditions. The presentation of
these cues led to the activation of a large fronto-parietal brain
network including the MeFC, the bilateral IFJ, the dorsal and lateral
premotor cortices, the medial and lateral parietal lobe, with a
relatively smaller activation in the bilateral anterior LPFC. By and
large, this network is consistent with that reported in a number of
previous studies concerned with the neural correlates of task
preparation (Brass and von Cramon, 2002, 2004; Gruber et al., 2006;
Luks et al., 2002; MacDonald et al., 2000; Sohn et al., 2000).
Importantly, we found rule-related activation foci to be a subset of
this general task preparation network. In this subset of rule-related
regions, the processing of rule (cue) information led to enhanced
activation compared to the processingof task (cue) information — as a
resultof theexplicit rule information providedby therule (but notthe
task) cues. While rule-cues are as effective as task-cues in activating
the general task goal (i.e., the type of task to be performed), they are
more powerful in activating the specific task rule. As a result, rule-
cues engender superior task preparation compared to task-cues,
which is expressed in better performance measures such as response
speed and accuracy and in increased neural computations in the
related brain regions.
There has been a long-standing debate concerning the extent to
which participants prepare in advance the whole set or only a fraction
of the relevant task parameters following cue presentation (Brass and
vonCramon,2002; Gruberet al.,2006; Lukset al.,2002; Meiran,1996;
Monsell and Mizon, 2006; Ruge et al., 2009). The present findings
Table 2
Significant activity in the conjunction analysis across the contrast rule-cue versus task-
cue (cue-only trials) and the contrast task-cue versus rule-cue (target-related
activation) (pb0.005).
RegionBA MNI coordinatesVoxel numberT max
R MFG/SFG
R MFG/SFG
R MFG
Medial SFG/MeFG (pre-SMA)
R inferior parietal lobule
R superior parietal lobule
L lingual gyrus/fusiform gyrus
R lingual gyrus/fusiform gyrus
10
46, 9
8, 6
8, 6
40
7
18
18
32, 64, 14
48, 38, 32
50, 8, 40
0, 36, 58
52, −58, 46
38, −74, 48
−8, −92, −18
12, −92, −18
137
90
64
413
40
23
157
110
3.69⁎⁎
3.47⁎⁎
2.90
3.71⁎⁎
2.94
2.95
3.55 ⁎⁎
3.74 ⁎⁎
Note. SFG=superior frontal gyrus; MFG=middle frontal gyrus; MeFG=medial frontal
gyrus; pre-SMA=supplementary motor area. Regions marked by ⁎⁎ showed significant
activation at a lower threshold of p b 0.001, uncorrected.
Table 1
Cortical activation for the comparison of rule-cue versus task-cue in cue-only trials (left) and for the comparison of task-cue versus rule-cue for the target-related activation⁎⁎
(right).
Cue-related activation (rule-cue–task-cue)Target-related activation (task-cue–rule-cue
Region BAMNI coordinates Voxel
number
T max RegionBA MNI coordinatesVoxel
number
T max
R anterior SFG
R MFG
L MFG/precentral gyrus
10
6
6
38, 62, 6
38, 2, 62
−48, 2, 48
186
97
23
6.15
7.56
4.66
R anterior SFG
R MFG
10
6
36, 64, 8
38, 4, 64
67
43
5.36
7.34
R MFG9, 8 56, 18, 38975.13
MeFG
Medial SFG (pre-SMA)
Medial SFG (pre-SMA/SMA)
R SPL
L precuneus
L fusiform gyrus/MOG
R fusiform gyrus
L MOG/IOG
L lingual gyrus/fusiform gyrus
R lingual gyrus
R fusiform gyrus/MOG/IOG
8
6
6
7
7
19
37
19, 18
17, 18
18
19, 18
0, 50, 48
0, 34, 60
−2, 8, 72
38, −58, 56
−18, −76, 48
−44, −72, −20
48, −52, −24
−44, −84, −12
−6, −92, −16
6, −86, −16
40, −66, −20
62
100
67
17
62
49
16
69
91
32
57
6.08
6.91
5.41
3.98
6.09
4.88
4.04
5.00
5.34
4.91
4.84
Medial SFG (pre-SMA)
MeFG (pre-SMA)
R SPL
6
6
7
0, 18, 62
−2, 12, 50
36, −62, 56
43
14
11
4.83
4.30
4.04
L fusiform gyrus/MOG, IOG
R fusiform gyrus
19
37, 20
−42, −68, −16
54, −58, −20
91
17
4.82
4.50
L/R lingual gyrus/fusiform gyrus18, 19
−8, −92, −181985.23
R fusiform gyrus, MOG/lingual gyrus 18, 1926, −84, −14 554.86
Note. SFG=superior frontal gyrus; MFG=middle frontal gyrus; MeFG=medial frontal gyrus; SMA=supplementary motor area; SPL=superior parietal lobe; IPL=inferior parietal
lobe; MOG=middle occipital gyrus; IOG=inferior occipital gyrus.
⁎⁎Target-related activation is observed when contrasting activation in cue-target and cue-only trials (see text).
1260
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 9

suggest that this strongly depends on the amount of explicit task
informationprovidedby thecue. If the cue containsinformationabout
both the type of task and the specific task rules (and if there is
sufficient time until the onset of the target), the level of preparedness
for the upcoming task will be superior to conditions in which the cue
does not provide any explicit rule information. Restated, even when
the time to prepare would be sufficient as such, participants may not
retrieve or pre-activate task parameters as fully or completely when
presented with a task-cue as when presented with a rule-cue. Rather,
a considerable part of the task preparation, in particular, the retrieval
and activation of the specific task rule information, appears to be
deferred until the presentation of the target.
A recent study by Verbruggen et al. (2007) revealed a pronounced
tendency of participants to defer part of the task preparation until
target presentation under conditions in which the task-cue remained
available (as was the case in the present study), as compared to
situations in which the task-cue was removed, during the preparation
interval. However, it is unlikely that the reduced amount of task rule
activation observed for the task-cue condition compared to the rule-
cue condition in the preparation period of the present study was
exclusively caused by the persisting cue presentation. The reason is
that a strategic delay in task preparation (as a result of the prolonged
cue availability) should have led to a reduced level of task rule
activation in the rule-cue condition as well as the task-cue condition,
especially since participants had to specify the task goal first before
they could activate the task rules. Given this, it remains plausible that
active task rule retrieval and rule activation were indeed facilitated by
the rule-cue (even if it involves a volitional-strategic component to
make use of the rule information) and the observed activation
changes in rule-related regions are associated with these processes in
the present paradigm.
Our interpretation receives support from a recent study by Ruge
et al. (2009), which separated task-cue-related preparatory activity
from rule-related activity in a way that differed from that in our
study. In their study, participants were presented with either task-
cue-first or target-stimulus-first trials when performing a task-
switching paradigm with an ambiguous stimulus consisting of a
digit (odd/even decision) and a letter (vowel/consonant decision).
On task-cue-first trials, a cue informed participants about the task to
be performed upon the presentation of the target stimulus; this was
similar to the task-cue in the present study. On target-stimulus first
trials, the presentation of the target stimulus preceded the
presentation of the task-cue. When comparing the preparation-
related activity in both kinds of trials, Ruge et al. found regions in
the lateral PFC, near the IFJ, in medial frontal areas, in pre-motor
and in posterior parietal brain areas to be activated more strongly
under the target-stimulus-first condition than under the task-cue-
first condition. This suggests that the preparation of task set
parameters was induced more strongly when the target stimulus
information was presented first compared to when the task-cue
information was provided first.
The findings of Ruge et al. (2009) and those of our study may be
consideredasbeingcomplementary,in thattheyrevealdifferentways
ofhowthecognitivesystemmay flexiblyadaptthetask preparation to
the specific task demands under different conditions of information
provided prior to task execution. The findings of Ruge et al.'s may be
interpreted by assuming that the presentation of umbiguous target
information, rather than of task-cue information, evokes stronger
recruitment of preparation-related brain regions because of a
relatively unselective preparation of task rule parameters. On the
other hand, our findings show that the presentation of task rule
information prior to the target stimulus encourages stronger
preparation-related activity which, however, leads to a selective and
more complete preparation of the specific task parameters of the
upcoming task.
Task rule activation and the lateral prefrontal cortex
The present findings indicate that a major region associated with
online task rule activation is the LPFC, which showed significant rule-
related activity in anterior, middle, and posterior portions. Earlier
studies with single-cell recordings in monkeys (Wallis and Miller,
2003; White and Wise, 1999) or fMRI in humans (Bunge et al., 2003;
see also Crone et al., 2006a,b) had already shown an association of the
LPFC with the retrieval of task rules from long-term memory. For
example, in Bunge et al. (2003), participants learned different rules of
how to respond to probe stimuli in a separate learning phase prior to
the fMRI scanning session. Similar rules (e.g., press left key if two
stimuli match each other) were associated with different types of
rule-cue (verbal or symbolic cues). In the fMRI scanning session,
participants had to activate the acquired rule knowledge upon
presentation of the rule-cues and then, after a delay, process two
sequentially presented probe stimuli (same, different). Bunge et al.
found ventral regions in the left LPFC to be active during the delay
after rule-cue presentation, to be sensitive to the difficulty of the rule,
and to be insensitive to the type of rule-cue. Because of the
insensitivity of these regions to the type of rule-cue, ventral LPFC
regions were assumed to be related to abstract rule knowledge; this
was in contrast to regions in the left and right DLPFC which proved to
be sensitive to the different types of rule-cue and were, therefore,
assumed to be related to the specific rule knowledge.
Whilethese earlier studies show an involvement of LPFC regionsin
the retrieval of (abstract and specific) rule knowledge from long-term
memory, the present findings show that the degree of rule-related
activity can also be modulated by the amount of rule information
provided by the current rule-cue. Prior findings were not conclusive
Fig. 6. Illustration of switch-related activation in cue-only trials. The activation in medial parts of the superior frontal gyrus (SFG; MNI coordinates −8, 14, 58) was found by
comparing the activation in switch and repetition trials independently on the cue type, i.e., rule-cue and task-cue. Beta values in the switch-related region-of-interest (ROI) in the
medial SFG as a function of cue type and task transition (cue-only trials) are presented in the right side. For details about the ANOVA results on the beta values see text.
1261
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 10

about the degree of online rule activation in the LPFC because the cues
used did not permit distinguishing between general task information
and specific rule information.
A recent study of MacDonald et al. (2000) used the Stroop
paradigm (color naming of a written (color) word or reading the
color word) in combination with a task-cueing manipulation to
investigate brain regions associated with task rule processing. In
that paradigm, a task-cue specified whether the next upcoming trial
required color naming or word reading. Interestingly, the magni-
tude of cue-related activity increased with the expected difficulty of
the next upcoming task: the left LPFC was more strongly activated
when the cue indicated to participants that they would have to
process the more difficult color naming task, as compared to the
word reading task. The present findings extend the results of
MacDonald et al., because they show that the degree of activity in a
particular LPFC region varies with the amount of the task
information provided by the cue about the upcoming task. The
more information about the specific task rule is provided by the cue
(rule-cue versus task-cue), the stronger the activity during task
preparation (i.e. during cue processing). This does not contradict
accounts assuming that different types of information provided by
external cues may be related to activation in different regions of the
LPFC (Koechlin et al., 2003; Koechlin and Summerfield, 2007).
In the present study, a considerable amount of activation was
found in posterior regions of the LPFC near the right IFJ. This region
had been shown to be associated with mechanisms involved in the
‘actualization’ of a current task representation (e.g., uploading new
task parameters or a new task representation by the bilateral
posterior regions of the LPFC) in situations with changing tasks
(Brass and von Cramon, 2002, 2004; Derrfuss et al., 2005; Gruber et
al., 2006). The present findings are consistent with this, while
additionally showing that the degree to which a new task represen-
tation is uploaded in advance depends on the specificity of the
information provided by the cue. A larger amount of presented task
information permits a more complete uploading of the task
parameters required on the upcoming trial, and this is accompanied
by an increased amount of neural activity in brain areas near the IFJ.
The observed rule-related activation in the right aPFC is consistent
with studies suggesting that this region is critical for difficult retrieval
processes on both episodic-memory (Della-Maggiore et al., 2002;
Nyberg et al., 2000) and working-memory contents (Christoff and
Gabrieli, 2000; Leung et al., 2005; MacLeod et al., 1998; Soto et al.,
2007). For example, the meta-analysis of MacLeod et al. showed that
the right aPFC is activated especially in WM tasks in which a high,
rather than a low, number of items has to be maintained in memory.
In the present paradigm, the richer rule information provided by the
rule-cues may have induced the participants to more strongly activate
and retrieve the task-relevant S–R associations from memory,
compared to the task-cue condition. This additional effort during
task preparation provides a likely explanation for the observation of
rule-related activation in the right aPFC in the present study (see also
Stern et al., 2007).
Rule-related activation outside the lateral prefrontal cortex
Further rule-related activation was found in medial frontal
regions and here specifically within the pre-SMA. Single-cell and
tracer studies suggest that regions of the pre-SMA receive direct
input from the LPFC, while the neural regions in the SMA proper
are connected to the motor areas (Picard and Strick, 2001; Tanji,
1994). The specific connections to the LPFC make the pre-SMA
most appropriate for the preparation of the specific task rule
during sensori-motor performance (Hikosaka et al., 1996). In line
with this, several authors have shown the pre-SMA to be involved
in the acquisition and control of arbitrary S–R associations in
humans (Gordon et al., 1995; Halsband and Freund 1990; Hikosaka
et al., 1996; Picard and Strick, 1996; Sakai et al., 1998, 1999) and
monkeys (Halsband and Passingham, 1985; Halsband et al., 1994).
In the present study, we found the pre-SMA to be activated in
conditions promoting task rule activation as well as conditions of
general task preparation. However, the fact that there was
additional activation in the pre-SMA in rule-cue compared to
task-cue conditions suggests that providing participants with
explicit rule information leads to an enhanced preparation of
(specific) S–R associations.
The observation of rule-related activation in the parietal cortex is
consistent with the assumption that regions along the intraparietal
sulcus are involved in the activation of motor representations which
are spatially mapped to different sensory stimuli (Andersen, 1987;
Caminiti et al., 1996; Schubert et al., 1998; Stein, 1989). A number of
neuroimaging studies have revealed parietal activation when parti-
cipants have to produce motor responses upon the presentation of
sensory stimuli in various kinds of behavioral paradigms (Brass and
Von Cramon, 2002, 2004; Bunge et al., 2002, 2003; Snyder et al., 1997;
Stelzel et al., 2008; Zysset et al., 2006). The present findings
additionally suggest that the parietal involvement is modulated by
the amount of prior information provided about the motor response
that has to be performed upon stimulus presentation: the more
information is provided about the S–R rule, the larger the amount of
neural computations in parietal areas involved in processing the
required S–R association.
A higher need of control for the task rule preparation in switch trials
A larger activation was found during the preparation period (cue-
only trials) for task switch compared to task repetition conditions in
the medial SFG. This extends findings of other studies, pointing to an
association of this region with switching between tasks (e.g., Dove et
al., 2000; Yeung et al., 2006). It is also consistent with the view that
the pre-SMA has an anticipatory role during the intentional
reconfiguration of a response set (Rushworth et al., 2002).
A related ROI analysis revealed the observed increase of switch-
related (compared to repetitions) activation in the medial SFG region
to be similar in the rule-cue and task-cue conditions. Additionally, the
ROI analysis showed that the general amount of activity in switch-
related regions was larger with rule-cues than with task-cues. In our
view, this latter finding suggests that, given sufficient task rule
information, the need to prepare for a task switch evokes efficient
processes of re-loading the task rule information already during the
preparation period of the task processing.
Rule retrieval, or activation of the currently required task set, is
presumedtobeanimportantcomponentoftask-switching(Mayrand/
INS; Kliegl,2000; Rogers and Monsell, 1995; Monsell,2003; Rubinstein
etal.,2001).Inthepresentstudy,thetaskruleinformationprovidedby
rule-cues may have evoked preparatory processes that included even
the rule representation, thus permitting a more complete task set
reconfiguration (on switch trials) compared to the presentation of
meretask-cues.Thiswouldexplainwhythetaskswitchcosts(i.e.,error
rates) were reduced in the rule-cue relative to the task-cue condition,
and whytheamountofneuralactivation wasincreasedinpreparation-
related regions with rule-cues compared to the task-cues.
Previous findings point to the possibility that activity involved in
switching task sets may be confounded by processes involved in the
perception of the different cues presented on successive trials (Logan
and Bundesen, 2003). However, in the present study, we can rule out
such confounds on the observed fMRI activity. A post-hoc analysis, in
which we compared the RTs in conditions with changing cues (rule-
cue followed by task-cue, and vice versa) to the RTs in conditions with
repeated cues, showed no additional influence of a cue change on the
size of the task switch costs; the difference between the RTs in both
conditions (i.e., RR: task repeat and cue repeat; RS: task repeat but cue
switch) amounted to a negligible 1 ms.
1262
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 11

Some previous fMRI studies had failed to find any additional
region, or even activation, to be involved in preparing for task
switches (as compared to repetitions), which was taken to cast doubt
on the assumption that switch-specific control processes are
operating during the preparation period (e.g., Brass and von Cramon,
2002, 2004; Gruber et al., 2006; Ruge et al., 2005; but see Chiu and
Yantis, 2009, Slagter et al., 2006, and ERP studies, e.g., Lavric et al.,
2008, for positive evidence). We agree that the kinds of preparatory
processes occurring after cues that indicate a task repeat and cues that
indicate a task switch are similar, but they are more intense in task
switch situations. In other words, more control is required when
preparing for switch trials, and the corresponding additional
activation can be found in either the task preparation or the execution
period, depending on the type of cue.
A possible reason for the discrepant findings concerning switch-
related activations in the preparation period may lie in the different
types of cues, or cue information, which were used in studies that
failed to and that did find such activations. Studies that failed to
find additional switch-related activation (see below) used arbitrary
cues (e.g., in Brass et al., 2002, a square or diamond indicating an
odd/even and or a number size task), whereas we used semanti-
cally unequivocal cues to indicate the upcoming task (see also
Wylie et al., 2006). It would appear plausible that, if the cue is an
arbitrary shape, a considerable amount of time needs to be spent to
decode the cue and to represent the general task goal — that is, task
rule activation may be delayed and moved to the target period. This
might be the reason why studies that used such arbitrary cues failed
to find any larger activation in the preparation for task switches
compared to repetitions (Brass and von Cramon, 2002, 2004;
Gruber et al., 2006; Ruge et al., 2005).
Inourstudy,thewordsymbol‘gender’indicatedthegendertaskand
the symbol ‘color’ the color task. This use of semantic cues is similar to
the conditions in other studies which also reported additional switch-
related activity during the preparation period (e.g., Wylie et al., 2006).
Inthecaseofsemanticcues,thecuespecifiesthetaskrelativelydirectly,
making it much easier for participants to establish the task represen-
tation(Miyakeetal.,2004;Wylieetal.,2006)—andpermittingthemto
activate the task rule already within the preparation period. As a
consequence,theneuraleffortassociatedwiththeuploadingofthetask
rule information would be increased on task switch compared to
repetition trials, and this effort may be strong enough to evoke
significantfMRIactivationinthecomparisonofswitchversusrepetition
trials during the task preparation period (Wylie et al., 2006).
Thus, these findings conform well with recent evidence from ERP
studies indicating that switch-related neural activity can indeed be
observed for processes associated with the task preparation (Lavric et
al., 2008). Perhaps the use of more elaborate paradigms permitting
further differentiation of the various components in switching activity
(in particular, cue-related and task-related activity) may add to these
findings in future studies (Monsell and Mizon, 2006).
Conclusions
The present study identified neural correlates of task rule
activation and revealed these to be subcomponents of a general task
preparation network. The processes of task rule activation are
operating in task-cuing paradigms given that the cues provide explicit
information about the task rule. In this case, part of the rule activation
may be brought forward from after to before the onset of the target
stimulus. This leads to increased neural ‘effort’ in rule activation-
related regions that are a subset of the general task preparation-
related regions, and to improved task performance. In addition, a
stronger activation was observed in a rule-related region if partici-
pants prepared for a task switch, compared to a repetition, trial and if
this preparation process included a change of the relevant task rule
information. We conclude that the extent to which participants
prepare in advance the parameters of an upcoming task depends
strongly on the specificity and the amount of information provided
prior to task processing; this in turn determines the degree of
activation in brain regions associated with task preparation.
Acknowledgments
This research was supported by grants from the Natural Science
Foundation of China (30770712, 30970895, 90920012) and from the
Ministry of Science and Technology of China (2010CB8339004) to X.Z.
The work of T.S. was supported by grants of CoTeSys (No. 439), and of
DFG (No. 1397). The work of H.M was supported by a grant of CoTeSys
(No. 134). We thank Qi Chen and Ran Hou for help in designing the
experimentalinvestigationandforhelpindataprocessing,respectively.
References
Andersen, R.A., 1987. The role of the inferior parietal lobule in spatial perception and
visual-motor integration. In: Plum, F., Mountcastle, V.B., Geiger, S.R. (Eds.), The
handbook of physiology, Sec 1, The nervous system, Vol V, Higher functions of the
brain, Ft 2. American Physiological Society, Bethesda, MD, pp. 483–518.
Brass, M., von Cramon, D.Y., 2002. The role of the frontal cortex in task preparation.
Cereb. Cortex 12, 908–914.
Brass, M., von Cramon, D.Y., 2004. Decomposing components of task preparation with
functional magnetic resonance imaging. J. Cogn. Neurosci. 16, 609–620.
Bunge, S.A., Hazeltine, E., Scanlon, M.D., Rosen, A.C., Gabrieli, J.D., 2002. Dissociable
contributions of prefrontal and parietal cortices to response selection. Neuroimage
17, 1562–1571.
Bunge, S.A., Kahn, I., Wallis, J.D., Miller, E.K., Wagner, A.D., 2003. Neural circuits
subserving the retrieval and maintenance of abstract rules. J. Neurophysiol. 90,
3419–3428.
Caminiti, R., Ferraina, S., Johnson, P.B., 1996. The sources of visual information to the
primate frontal lobe: a novel role for the superior parietal lobule. Cereb. Cortex 6,
319–328.
Chiu, Y.C., Yantis, S., 2009. A domain-independent source of cognitive control for task
sets: shifting spatial attention and switching categorization rules. J. Neuroscience.
29, 3930–3938.
Christoff, K., Gabrieli, J.D.E., 2000. The frontopolar cortex and human cognition:
evidence for a rostrocaudal hierarchical organization within the human prefrontal
cortex. Psychobiology 28, 168–186.
Corbetta, M., Kincade, J.M., Ollinger, J.M., McAvoy, M.P., Shulman, G.L., 2000. Voluntary
orienting is dissociated from target detection in human posterior parietal cortex.
Nat. Neurosci. 3, 292–297.
Crone, E.A., Donohue, S.E., Honomichl, R., Wendelken, C., Bunge, S.A., 2006a. Brain
regions mediate flexible rule use during development. J. Neurosci. 26,
11239–11247.
Crone, E.A., Wendelken, C., Donohue, S.E., Bunge, S.A., 2006b. Neural evidence for
dissociable components of task-switching. Cereb. Cortex 16, 475–486.
Della-Maggiore, V., Grady, C.L., McIntosh, A.R., 2002. Dissecting the effect of aging on
the neural substrates of memory: deterioration, preservation or functional
reorganization? Rev. Neurosci. 13 (2), 167–181.
Derrfuss, J., Brass, M., Neumann, J., von Cramon, D.Y., 2005. Invovlement of the inferior
frontal junction in cognitive control: meta-analyses of switching and stroop
studies. Hum. Brain Mapp. 25, 22–34.
Dove, A., Pollmann, S., Schubert, T., Wiggins, C.J., von Cramon, D.Y., 2000. Prefrontal
cortex activation in task-switching: an event-related fMRI study. Cogn. Brain Res. 9,
103–109.
Friston, K.J., Holmes, A.P., Worsley, K.J., Poline, J.P., Frith, C.D., Frackowiak, R.S.J., 1995.
Statistical parametric maps infunctional imaging: a general lineat approach. Hum.
Brain Mapp. 2, 189–210.
Gollwitzer, P.M., Sheeran, P., 2006. Implementation intentions and goal achievement: a
metaanalysis of effects and processes. Adv. Exp. Soc. Psych. 38, 69–119.
Gordon, A.M., Flament, D., Lee, J.H., Uguibil, K., Kim, S.G., Ebner, T.J., 1995. Functional
MRI of cortical motor areas during sequential typing movements. Abstract 1422.
Annual Meeting of the Socirty of Neuroscience. MIT Press, London.
Gruber, O., Karch, S., Schlueter, E.K., Falkai, P., Goschke, T., 2006. Neural mechanisms of
advance preparation in task switching. NeuroImage 31, 887–895.
Halsband, U., Freund, H.J., 1990. Premotor cortex and conditional motor learning in
man. Brain 113, 207–222.
Halsband, U., Passingham, R.E., 1985. Premotor cortex and the conditions for movement
in monkeys (Macaca fascicularis). Behav. Brain Res. 18 (3), 269–277.
Halsband, U., Matsuzaka, Y., Tanji, J., 1994. Neuronal activity in the primate
supplementary, pre-supplementary and premotor cortex during externally and
internally instructed sequential movements. Neurosci. Res. 20 (2), 149–155.
Hikosaka, O., Sakai, K., Miyauchi, S., Takino, R., Sasaki, Y., Putz, B., 1996. Activation of
human presupplementary motor area in learning of sequential procedures: a
functional MRI study. J. Neurophysiol. 76, 617–621.
Koechlin, E., Summerfield, C., 2007. An information theoretical approach to prefrontal
executive function. Trends Cogn. Sci. 11 (6), 229–235.
Koechlin, E., Ody, C., Kouneiher, F., 2003. The architecture of cognitive control in the
human prefrontal cortex. Science 302, 1181–1185.
1263
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264

Page 12

Kriegeskorte, N., Simmons, W.K., Bellgowan, P.S., Baker, C.I., 2009. Circular analysis in
systems neuroscience: the dangers of double dipping. Nat. Neurosci. 12, 535–540.
Lavric, A., Mizon, G.A., Monsell, S., 2008. Neurophysiological signature of effective
anticipatory task-set control: a task-switching investigation. Euro. J. Neurosci. 28
(5), 1016–1029.
Leung, H.C., Gore, J.C., Goldman-Rakic, P.S., 2005. Differential anterior prefrontal
activation during the recognition stage of a spatial working memory task. Cereb.
Cortex 15 (11), 1742–1749.
Luks, T.L., Simpson, G.V., Feiwell, R.J., Miller, W.L., 2002. Evidence for anterior cingulate
cortex involvement in monitoring preparatory attentional set. Neuroimage 17 (3),
792–802.
Logan, G.D., Bundesen, C., 2003. Clever homunculus: is there an endogenous act of
control in the explicit task-cuing procedure? J. Exp. Psychol. Hum. Percept. Perform.
29, 575–599.
MacDonald III, A.W., Cohen, J.D., Stenger, V.A., Carter, C.S., 2000. Dissociating the role of
the dorsolateral prefrontal and anterior cingulate cortex in cognitive control.
Science 288, 1835–1838.
MacLeod, K., Buckner, R.L., Miezin, F.M., Petersen, S.E., Raichle, M.E., 1998. Right anterior
prefrontal cortex activation during semantic monitoring and working memory.
Neuroimage 7, 41–48.
Mayr, U., Kliegl, R., 2000. Task-set switching and long term memory retrieval. J. Exp.
Psychol. Learn. Mem. Cogn. 26, 1124–1140.
Meiran, N., 1996. Reconfiguration of processing mode prior to task performance. J. Exp.
Psychol. Learn. Mem. Cogn. 22, 1–20.
Meiran, N., 2000. Modeling cognitive control in task-switching. Psychol. Res. 63,
234–249.
Miller, E.K., Cohen, J.D., 2001. An integrative theory of prefrontal cortex function. Annu.
Rev. Neurosci. 24, 167–202.
Miyake, A., Emerson, M.J., Padilla, F., Ahn, J., 2004. Inner speech as a retrieval aid for task
goals: the effects of cue type and articulatory suppression in the random task cuing
paradigm. Acta Psychol. 115, 123–142.
Monsell, S., 2003. Task switching. Trends Cogn. Sci. 7, 134–140.
Monsell, S., Mizon, G.A., 2006. Can the task-cuing paradigm measure an endogenous
task-set reconfiguration process? J. Exp. Psychol. Hum. Percept. Perform. 32,
493–516.
Nichols, T., Brett, M., Andersson, J., Wager, T., Polinee, J.B., 2005. Valid conjunction
inference with the minimum statistic. NeuroImage 25, 653–660.
Nyberg, L., Persson, J., Habib, R., Tulving, E., McIntosh, A.R., Cabeza, R., Houle, S., 2000.
Large scale neurocognitive networks underlying episodic memory. J. Cogn.
Neurosci. 12 (1), 163–173.
Picard, N., Strick, P.L., 1996. Motor areas of the medial wall: a review of their location
and functional activation. Cereb. Cortex 6, 342–353.
Picard, N., Strick, P.L., 2001. Imaging the premotor areas. Curr. Opin. Neurobiol. 11 (6),
663–672.
Rogers, R.D., Monsell, S., 1995. The cost of a predictable switch between simple
cognitive tasks. J. Exp. Psychol. Gen. 124, 207–231.
Rubinstein, J.S., Meyer, D.E., Evans, J.E., 2001. Executive control of cognitive processes in
task switching. J. Exp. Psychol. Hum. Percept. Perform. 27, 763–797.
Ruge, H., Brass, M., Koch, I., Rubin, O., Meiran, N., von Cramon, D.Y., 2005. Advance
preparation and stimulus-induced interference in cued task switching: further
insights from BOLD fMRI. Neuropsychologia 43, 340–355.
Ruge, H., Braver, T., Meiran, N., 2009. Attention, intention, and strategy in preparatory
control. Neuropsychologia 47, 1670–1685.
Rushworth, M.F.S., Hadland, K.A., Paus, T., Sipila, P.K., 2002. Role of the human medial
frontal cortex in task switching: a combined fMRI and TMS study. J. Neurophysiol.
87, 2577–2592.
Sakai, K., Hikosaka, O., Miyauchi, S., Takino, R., Sasaki, Y., Pütz, B., 1998. Transition of
brain activation from frontal to parietal areas in visuo-motor sequence learning.
J. Neurosci. 18, 1827–1840.
Sakai, K., Hikosaka, O., Miyauchi, S., Sasaki, Y., Fujimaki, N., Putz, B., 1999.
Presupplementary motor area activation during sequence learning reflects visuo-
motor association. J. Neurosci. 19, RC1(1-6).
Schubert, T., von Cramon, D.Y., Niendorf, T., Pollmann, S., Bublak, P., 1998. Cortical areas
and the control of self-determined finger movements: an fMRI study. NeuroReport
9 (14), 3171–3176.
Slagter, H.A., Weisman, D.H., Giesbrecht, B., Kenemans, J.L., Mangun, G.R., Kok, A.,
Woldorff, M.G., 2006. Brain regions activated by endogenous preparatory set
shifting as revealed by fMRI. Cogn. Affect. Behav. Neurosci. 6, 175–189.
Snyder, L.H., Batista, A.P., Andersen, R.A., 1997. Coding of intention in the posterior
parietal cortex. Nature 386, 167–170.
Sohn, M.H., Ursu, S., Anderson, J.R., Stenger, V.A., Carter, C.S., 2000. The role of prefrontal
cortex and posterior parietal cortex in task switching. Proc. Natl. Acad. Sci. U. S. A.
97, 13448–13453.
Soto, D., Humphreys, G.W., Rotshtein, P., 2007. Dissociating the neural mechanisms of
memory-based guidance of visual selection. Proc. Natl. Acad. Sci. U. S. A. 104 (43),
17186–17191.
Stein, J.F., 1989. The representation of egocentric space in the posterior parietal cortex.
Exp. Physiol. 74, 583–606.
Stelzel, C., Kraft, A., Brandt, S.A., Schubert, T., 2008. Dissociable effects of task order
control and task set maintenance during dual-task processing. J. Cogn. Neurosci. 20,
613–628.
Stern, E.R., Wager, T.D., Egner, T., Hirsch, J., Mangels, J.A., 2007. Preparatory neural
activity predicts performance on a conflict task. Brain Res. 11, 92–102.
Tanji, J., 1994. The supplementary motor area in the cerebral cortex. Neurosci. Res. 19
(3), 251–268.
Verbruggen, F., Liefooghe, B., Vandierendonck, A., Demanet, J., 2007. Short cue
presentations encourage advance task preparation: a recipe to diminish the
residual switch cost. J. Exp. Psychol. Learn. Mem. Cogn. 33, 342–356.
Wallis, J.D., Miller, E.K., 2003. From rule to response: neuronal processes in the
premotor and prefrontal cortex. J. Neurophysiol. 90, 1790–1806.
Weissman, D.H., Gopalakrishnan, A., Hazlett, C.J., Woldorff, M.G., 2005. Dorsal anterior
cingulate cortex resolves conflict from distracting stimuli by boosting attention
toward relevant events. Cereb. Cortex 15, 229–237.
Wendelken, C., Bunge, S.A., Carter, C.S., 2008. Maintaining structured information: an
investigation into functions of parietal and lateral prefrontal cortices. Neuropsy-
chologia 46 (2), 665–678.
White, I.M., Wise, S.P., 1999. Rule-dependent neuronal activity in the prefrontal cortex.
Exp. Brain Res. 126 (3), 315–335.
Wylie, G.R., Javitt, D.C., Foxe, J.J., 2006. Jumping the gun: is effective preparation
contingent upon anticipatory activation in task-relevant neural circuitry? Cereb.
Cortex 16, 394–404.
Yeung, N., Nystrom, L.E., Aronson, J.A., Cohen, J.D., 2006. Between-task competition and
cognitive control in task switching. J. Neurosci. 26 (5), 1429–1438.
Zysset, S., Wendt, C.S., Volz, K.G., Neumann, J., Huber, O., von Cramon, D.Y., 2006. The
neural implementation of multi-attribute decision making: a parametric fMRI
study with human subjects. Neuroimage 31, 1380–1388.
1264
Y. Shi et al. / NeuroImage 51 (2010) 1253–1264