Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A meta-analysis

Institute of Cognitive Neuroscience and Division of Psychology and Language Sciences, University College London, London, UK.
NeuroImage (Impact Factor: 6.36). 12/2010; 53(4):1359-67. DOI: 10.1016/j.neuroimage.2010.07.032
Source: PubMed
ABSTRACT
Recent studies have shown that functional connectivity in the human brain may be detected by analyzing the likelihood with which different brain regions are simultaneously activated, or "co-activated", across multiple neuroimaging experiments. We applied this technique to investigate whether distinct subregions within rostral prefrontal cortex (RoPFC) tend to co-activate with distinct sets of brain regions outside RoPFC, in a meta-analysis of 200 activation peaks within RoPFC (approximating Brodmann Area 10) and 1712 co-activations outside this region, drawn from 162 studies. There was little evidence for distinct connectivity between hemispheres or along rostral/caudal or superior/inferior axes. However, there was a clear difference between lateral and medial RoPFC: activation in lateral RoPFC was particularly associated with co-activation in dorsal anterior cingulate, dorsolateral PFC, anterior insula and lateral parietal cortex; medial RoPFC activation was particularly associated with co-activation in posterior cingulate, posterior superior temporal sulcus and temporal pole. These findings are consistent with anatomical studies of connectivity in non-human primates, despite strong cross-species differences in RoPFC. Furthermore, associations between brain regions inside and outside RoPFC were in some cases strongly influenced by the type of task being performed. For example, dorsolateral PFC, anterior cingulate and lateral parietal cortex tended to co-activate with lateral RoPFC in most tasks but with medial RoPFC in tasks involving mentalizing. These results suggest the importance of changes in effective connectivity in the performance of cognitive tasks.

Full-text

Available from: Emmanuelle Volle
Distinct functional connectivity associated with lateral versus medial rostral
prefrontal cortex: A meta-analysis
Sam J. Gilbert
, Gil Gonen-Yaacovi, Roland G. Benoit, Emmanuelle Volle, Paul W. Burgess
Institute of Cognitive Neuroscience and Division of Psychology and Language Sciences, University College London, London, UK
abstractarticle info
Article history:
Received 25 August 2009
Revised 9 July 2010
Accepted 14 July 2010
Available online xxxx
Recent studies have shown that functional connectivity in the human brain may be detected by analyzing the
likelihood with which different brain regions are simultaneously activated, or co-activated, across multiple
neuroimaging experiments. We applied this technique to investigate whether distinct subregions within
rostral prefrontal cortex (RoPFC) tend to co-activate with distinct sets of brain regions outside RoPFC, in a
meta-analysis of 200 activation peaks within RoPFC (approximating Brodmann Area 10) and 1712 co-
activations outside this region, drawn from 162 studies. There was little evidence for distinct connectivity
between hemispheres or along rostral/caudal or superior/inferior axes. However, there was a clear difference
between lateral and medial RoPFC: activation in lateral RoPFC was particularly associated with co-activation
in dorsal anterior cingulate, dorsolateral PFC, anterior insula and lateral parietal cortex; medial RoPFC
activation was particularly associated with co-activation in posterior cingulate, posterior superior temporal
sulcus and temporal pole. These ndings are consistent with anato mical studies of connectivity in non-
human primates, despite strong cross-species differences in RoPFC. Furthermore, associations between brain
regions inside and outside RoPFC were in some cases strongly inuenced by the type of task being
performed. For example, dorsolateral PFC, anterior cingulate and lateral parietal cortex tended to co-activate
with lateral RoPFC in most tasks but with medial RoPFC in tasks involving mentalizing. These results suggest
the importance of changes in effective connectivity in the performance of cognitive tasks.
© 2010 Elsevier Inc. All rights reserved.
Two important aims of cognitive neuroscience are to understand
the specialized functions of individual brain region s, and the
interactions between them that support competent behavior. There
are a variety of methods for investigating such interactions, both in
terms of the direct anatomical connectivity between distinct brain
areas (most commonly studied in the non-human brain) and in terms
of functional connectivity (typically studied in the human brain). The
term functional connectivity refers to the correlation between
remote neurophysiological events (Friston et al., 1996), in the sense
that the signal in one region predicts the signal in another. This can
potentially be detected in moment-by-moment uctuations in brain
activity (e.g. by showing a correlated signal in two regions across the
timepoints of an fMRI timecourse), or across entire studies (e.g. by
showing that studies that report activations in region 1 also tend to
report activations in region 2). Here, we use the term in the latter
sense. Functional connectivity between two regions should not be
taken to imply the existence of direct anatomical (i.e. monosynaptic)
links, although it would be consistent with such links.
Recently, Toro et al. (2008) showed that consistent patterns of
functional connectivity may be detected by analyzing peak activations
across multiple functional imaging studies that were originally
performed for other purposes (see also Kober et al., 2008; Koski and
Paus, 2000, and Postuma and Dagher, 2006, for similar approaches).
They investigated a database of results from earlier studies and
calculated, for each pair of locations in the brain, the likelihood that
one region would be activated given that the other was activated. If
one region is more likely to be activated if the other is also activated,
this was taken as evidence of functional connectivity between the two
regions. Toro et al. (2008) showed that this technique can recover
patterns of functional connectivity predicted from anatomical studies
(e.g. activation in one region is associated with a greater probability of
acti vation in the homologous contralateral region), along wi th
patterns of functional connectivity predicted from frequently associ-
ated activation patterns in earlier studies (e.g. co-activations in
multiple nodes of the so-called default mode network).
In the present study we adopted a similar approach to Toro et al.
(2008) in order to investigate associations between activation within
particular subparts of rostral prefrontal cortex (RoPFC), approximat-
ing Brodmann Area (BA) 10, and activations elsewhere in the brain.
Specically, we investigated whether patterns of activation outside
RoPFC are predicted by the precise location of co-activations within
RoPFC. Were such results to be found, this would indicate that
different subregions of RoPFC are distinguished by distinct patterns of
functional connectivity.
NeuroImage xxx (2010) xxxxxx
Corresponding author. Institute of Cognitive Neuroscience, 17 Queen Square,
London WC1N 3AR, UK. Fax: + 44 20 7813 2835.
E-mail address: sam.gilbert@ucl.ac.uk (S.J. Gilbert).
YNIMG-07493; No. of pages: 9; 4C:
1053-8119/$ see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2010.07.032
Contents lists available at ScienceDirect
NeuroImage
journal homepage: www.elsevier.com/locate/ynimg
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 1
Recent anatomical studies suggest that RoPFC may be divided into
distinct architectonic regions. Carmichael and Price (1994) suggested
that in monkeys, area 10 should be divided into two distinct areas:
medial and orbital (areas 10m and 10o respectively), while in
humans, three different subregions: medial, rostral and polar (areas
10m, 10r and 10p, respectively) have been proposed (Ongur et al.,
2003).
Furthermore, recent functional neuroimaging studies in humans
have provided evidence for considerable functional specialization
within RoPFC, at a remarkably ne scale. A meta-analysis by Gilbert
et al. (2006c) found that RoPFC may be divided into at least three
functionally distinct regions, according to both a lateral/medial and an
anterior/posterior axis. An anterior polar region was associated with
multitasking, i.e. situations where participants have more than one task
to perform within a single block of trials. More posteriorly, a lateral
region was associated with episodic memory retrieval whereas a medial
region was associated with mentalizing, i.e. explicit reection on one's
own mental states or those of others. These meta-analytic ndings have
also been corroborated in a within-subjects fashion, in functional
neuroimaging studies demonstrating segregation between distinct
nearby regions of RoPFC (Gilbert et al., 2007a; Gilbert et al., 2010).
The evidence reviewed above indicates the presence of both
anatomical and functional subdivisions within RoPFC. However, it is
not clear whether these functional subdivisions reect differences in
functional connectivity between distinct subregions within RoPFC and
other brain regions outside this region. We therefore examined a
database of neuroimaging studies reporting activation both within
and outside RoPFC, to investigate whether activation peaks outside
RoPFC differed according to the location of co-activated regions
within RoPFC. Furthermore, we additionally asked whether these
ndings differed as a function of task category. This allowed us to
investigate whether activation in particular regions of RoPFC was
inevitably associated with increased likelihood of activation in
particular regions outside RoPFC (reecting xed functional connec-
tivity), or whether patterns of co-activation varied according to task
category (potentially reecting changes in effective connectivity
depending on experimental conditions).
Methods
Study selection
The present study required a database of activation peaks within
RoPFC from previous functional neuroimaging studies. Additionally, for
each contrast producing an activation peak within RoPFC we recorded
the set of activation peaks outside RoPFC that were also reported. All
studies from the earlier meta-analysis of RoPFC activations by Gilbert
et al. (2006c) were included in the present database, so long as they also
reported at least one activation peak outside RoPFC (102 studies; 124
independent contrasts). A further 60 studies (76 contrasts) were added
to the database on the basis of additional searches in November 2008 of
the PubMed database for the terms PET or fMRI along with one of the
following: anterior prefrontal, rostral prefrontal, Brodmann's area
10, BA 10,
frontal pole, fr
ontopolar. Although this xed set of
search terms may have led us to miss relevant studies that did not
include the relevant keywords in the abstract, it also allowed us to avoid
experimenter bias in the selection of studies to be included in the meta-
analysis.
Inclusion criteria were identical to Gilbert et al. (2006c), with the
exception that there was no requirement that studies report response
time for all conditions, unlike the earlier meta-analysis. We included
studies using PET or fMRI that 1) investigated unmedicated healthy
young adults; 2) reported the coordinates of activations in the space
of the MNI template brain (Collins et al., 1994) or according to the
atlas of Talairach and Tournoux (1988); and 3) reported one or more
activations with peak coordinates falling within BA 10, according to
the atlas of Talairach and Tournoux or as dened by the Brodmann
map in MNI space supplied with MRIcro (Rorden and Brett, 2000).
When activations were reported in Talairach and Tournoux coordi-
nates, they were transformed into MNI space using a nonlinear
transformation ( http://www.mrc-cbu.cam.ac.uk/Imaging; Brett et al.,
2001) so that all coordinates were in a common stereotaxic
framework. Activations were accepted as signicant according to
the criteria set by each individual study. The database therefore
included any signicant difference in signal between a pair of
conditions (using the criteria for statistical signicance set by each
study), regardless of which condition was labeled by the original
authors as experimental and which as control.
Where a contrast yielded more than one activation peak falling
within BA 10, only the most statistically signicant was included in
the list of RoPFC activations. Furthermore, only independent contrasts
were included in the meta-analysis; for example, if two contrasts
involving a shared baseline condition were reported, only the most
signicant would be included. The nal database included 200
activation peaks falling within RoPFC, drawn from 162 studies. Each
RoPFC activation peak was associated with between 1 and 46 co-
activations (mean: 8.7). We refer to these as extra-RoPFC activation
peaks, because they could fall anywhere within the brain, although in
a few contrasts (where more than one RoPFC activation was reported)
additional RoPFC activations were included in this list. In total there
were 1712 extra-RoPFC activation peaks. All studies were categorized
into one of the following eight task categories, based on the criteria
described by Gilbert et al. (2006c): Attention, Perception, Language,
Working memory, Episodic retrieval, Other memory, Mentalizing,
Multitask. The rst six of these were based on categories described by
Cabeza and Nyberg (2000); the nal two were added by Gilbert et al.
(2006c), where Mentalizing refers to any study involving explicit
re
ection on one's own or others' mental states and Multitask refers
to
any study in which there was more than one task to perform within
a single block of trials. Studies were categorized into task categories
by two raters (inter-rater reliability for studies that were not included
in Gilbert et al., 2006c: 91%; all disagreements were resolved by
discussion).
Activation likelihood estimation analyses
Data were analyzed using an approach based on Activation
Likelihood Estimation, a common technique for meta-analyses of
neuroimaging results (Laird et al., 2005; Turkeltaub et al., 2002;
Wager et al., 2007), modied here to allow statistical comparisons
between two sets of results. In all analyses, activation peaks within
RoPFC were divided into two categories (e.g. left vers us right
hemisphere; medial versus lateral etc.). Activation peaks were
classied as lateral or medial by calculating whether the x coordinate
was closer to the midpoint or lateral edge of the MNI template brain
(Collins et al., 1994), given the y and z coordinates (where x denes a
leftright axis, y denes a rostralcaudal axis, and z denes a
superiorinferior axis). Activation peaks were classied as rostral/
caudal and superior/inferior by a median split of y and z coordinates
respectively (median y coordinate: 56; median z coordinate: 10); this
ensured that the sample size in each category was approximately
equal. Extra-RoPFC a ctivation peaks associat ed with these two
categories were then compared, with the aim of nding regions that
were signicantly more likely to be associated with one category (e.g.
lateral RoPFC) than the other (e.g. medial RoPFC). The brain was
divided into 5 mm isotropic voxels and results were computed at each
voxel. This voxel size was chosen to maximize efciency given the
available computational resources, seeing as the time taken for
analyses scales with the cube of the voxel size in each dimension (e.g.
3 mm voxels would increase computation tim e by a factor of
approximately ve).
2 S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 2
Table 1
Additional contrasts included in the meta-analysis, beyond those reported in Gilbert et al. (2006c).
Study x y z Medial/lateral Superior/inferior Rostral/caudal Category
Abe, N. et al. (2007). Journal of Cognitive Neuroscience, 19, 287 36 62 2 L I R Mentalizing
Abraham, A. et al. (2008). Journal of Cognitive Neuroscience, 20, 965 259 5 M I R Episodic
Addis, D.R and Schacter, D. (2008). Hippocampus, 18, 227 20 64 14 L S R Episodic
Addis, D.R. et al. (2007). Neuropsychologia, 45, 1363 2 64 8 M I R Multitasking
Arnott, S.R. et al. (2005). Journal of Cognitive Neuroscience, 17, 819 30 51 27 L S C Working
32 51 3 L I C Working
Antonova, E. et al. (2009) Memory, 17, 125 12 52 13 M I C Episodic
Bengtsson, S and Ulle´n, F. (2006). NeuroImage, 30, 272 36 42 3 L I C Multitasking
Blondin, F. and Lepage, M. (2008). Human Brain Mapping, 29, 1159 4 42 10 M I C Episodic
Cunningham, W.A., Raye, C.L. and Johnson, M.K. (2004). Journal of Cognitive
Neuroscience, 16, 1717
8 52 20 M S C Mentalizing
Daselaar, S. et al. (2008). Cerebral Cortex, 18, 217 42 45 1 L I C Episodic
De Martino, B. et al. (2009). Cerebral Cortex, 19, 127 30 54 10 L C Attention
Deeley, Q. et al. (2008). NeuroImage, 40, 389 18 43 6 M I C Perception
14 64 32 L S R Perception
den Ouden, H.E. et al. (2005). Neuroimage, 28, 787 9 63 12 M S R Mentalizing
27 63 6 L I R Multitasking
Dobbins, I. and Han, S. (2006). Cerebral Cortex, 16, 1614 33 60 12 L S R Episodic
Dobbins, I., and Han, S. (2006). Journal of Cognitive Neuroscience, 18, 1439 33 63 21 L S R Episodic
Dudukovic, N. and Wagner, A. (2007). Neuropsychologia, 45, 2608 24 57 9 L I R Episodic
Erk, S. et al. (2006). European Journal of Neuroscience, 1227 12 57 12 M S R Working
Evan Nee, D. et al. (2007). NeuroImage, 38, 740 32 64 16 L S R Working
Flores-Gutiérrez, E.O. et al. (2007). International Journal of Psychophysiology, 65, 6984. 34 56 7L I Perception
Gilbert, S.J. et al. (2007). SCAN, 2, 217 0 58 14 M I R Attention
8 54 30 M S C Mentalizing
Goel, V. and Vartanian, O. (2005). Cerebral Cortex, 15, 1170 16 58 10 M R Multitasking
Green, A. et al. (2006). Brain Research, 1096, 125 8 60 31 M S R Multitasking
Harrison, B.J. et al. (2005). Neuroimage, 24, 181 38 61 2 L I R Attention
34 64 10 L I R Attention
Heatherton, T. et al. (2006). SCAN, 1, 18 9 60 3 M I R Mentalizing
Henson, R.N. et al. (2005). Journal of Cognitive Neuroscience, 17, 1058 24 60 12 L S R Episodic
Kikyo, H. and Miyashita, Y. (2004). Neuroimage, 23, 1348 33 60 21 L S R Episodic
King, J.A. et al. (2005). Neuroimage, 28, 256 33 54 6 L I C Episodic
30 57 6 L I R Episodic
Koch, K. et al. (2006). Brain Research, 1107, 140 32 56
2L I Working
Koenig,
P. et al. (2005). Neuroimage, 24, 369 16 57 12 M S R Other
24 57 25 L S R Other
Konishi, S. et al. (2005). PNAS, 102, 12584 30 51 22 L S C Multitasking
Konrad, K. et al. (2005). Neuroimage, 28, 429 3 57 12 M I R Attention
Kroger, J.K. (2008). Brain Research, 1243, 86 30 61 4 L I R Attention
Kulkarni, B. (2005). European Journal of Neuroscience, 21, 3133 32 68 6 L I R Attention
Lee, T.W. (2006). SCAN, 1, 122 358 5 M I R Perception
Leung, H.C. et al. (2005). Cerebral Cortex, 15, 1742 29 54 13 L S C Working
Lie, C. et al. (2006). NeuroImage, 30, 1038 32 54 14 L S C Working
Locke, H and Braver, T. (2008). Cognitive, Affective, & Behavioral Neuroscience, 8, 99 32 61 10 L R Perception
Luks, L. et al. (2008). Neuroreport, 19, 155 42 51 12 L S C Attention
Marklund, P. et al. (2007). Cortex, 43, 22 38 56 4L I Working
Milham, M.P.and Banich, M.T. (2005). Human Brain Mapping, 25, 328 30 58 5 L I R Attention
Okuda, J. et al. (2007). International Journal of Psychophysiology, 64, 233 16 48 24 M S C Multitasking
2 66 4 M I R Multitasking
Ramnani, N. et al. (2004). Neuroimage, 23, 777 10 64 6 M I R Other
12 58 16 M S R Other
Ranganath, C. et al. (2007). NeuroImage, 35, 1663 7 63 0 M I R Episodic
Reed, C.L. et al. (2005). Neuroimage, 25, 718 0 60 33 M S R Perception
Schnell ,K. et al. (2007). NeuroImage, 34, 332 24 60 9 L I R Attention
Simons, J.S. et al. (2006). NeuroImage, 32, 696 30 48 18 L S C Episodic
Simons, J.S. et al. (2006). Neuropsychologia, 44, 1388 39 54 15 L S C Multitasking
048 6 M I C Multitasking
Simons, J.S. et al. (2008). Journal of Cognitive Neuroscience, 20, 447 33 57 6 L I R Episodic
Sommer, M. et al. (2007). NeuroImage, 35, 1378 34 64 6 L I R Mentalizing
Stelzel, C. et al. (2008). Journal of Cognitive Neuroscience, 20, 613 50 36 26 L S C Multitasking
Stern, E. et al. (2007). Brain Research, 1176, 92 22 54 20 M S C Attention
Knutson, K. et al. (2007). Human Brain Mapping, 28, 915 20 52 13 M S C Other
Strangman, G. et al. (2005). Neurohabilitation and Neural Repair, 19, 93 16 60 12 M S R Other
Tanabe, J. et al. (2007). Human Brain Mapping, 28, 1276 25 58 8 L I R Working
Turner, G. and Levine, B. (2006). Neuroscience, 139, 327 37 38 24 L S C Working
37 38 24 L S C Working
37 38 24 L S C Working
37 38 24 L S C Working
van Eimeren, T. et al. (2006). Neuroimage, 29, 286 33 57 6 L I R Attention
Vartanian, O., Goel, V. (2005). Neuroimage, 27, 927 4 68 2 M I R Language
Wager, T.D. et al. (2005). Neuroimage, 27, 323 34 41 20 L S C Attention
26 49 15 L S C Attention
30 49 30 L S C Attention
(continued on next page)
3S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 3
First, all extra-RoPFC activation peaks associated with one category
(e.g. lateral RoPFC) were considered. We generated a full three-
dimensional map of activations throughout the brain associated with
this category, by rst smoothing each activation peak with a 8 mm full-
width half-maximum Gaussian kernel, then summing together the
three-dimensional volumes, each representing a single smoothed
activation peak. A second whole-brain three-dimensional map was
generated in an analogous fashion from extra-RoPFC activation peaks
associated with the other category (e.g. medial RoPFC). These two maps
were then subtracted from each other to create, at all voxels outside
RoPFC, a new map representing the difference in the likelihood of co-
activations associated with the two categories of RoPFC activations. This
difference map was compared against a null-distribution as follows.
Each RoPFC activation and its associated extra-RoPFC activations were
randomly assigned to one of two categories. The number of RoPFC
activations in these randomly-generated categories matched the two
categories under investigation (e.g. lateral versus medial RoPFC).
Difference maps between the two randomly-assigned categories were
generated as above. This process was repeated 1000 times. At each
voxel, a z score was calculated by comparing the true difference score
against the mean and standard deviation obtained by randomly
assigning activations to the two categories. This z score indicated
whether a particular voxel was signicantly more likely to be associated
with one or the other RoPFC category than would be expected by
chance. Below, we refer to this map of z scores as the difference z-map.
Predictive validity of extra-RoPFC activations
In order to mitigate the multiple-comparisons problem created by
the mass-univariate approach described above, we rst performed a test
that was sensitive to the entire difference map, rather than investigating
results at a voxel-by-voxel level. This allowed us to perform a single test
to investigate whether there were differences in extra-RoPFC activa-
tions associated with two types of RoPFC activation, lookingat the whole
brain rather than specic regions. If this test was signicant, we then
went on to examine which regions outside RoPFC had a different
likelihood of co-activation with different categories of RoPFC activations
(i.e. performing a test at each voxel). We investigated four binary
classications of RoPFC activation peaks: left versus right hemisphere;
lateral versus medial; rostral versus caudal; and superior versus inferior.
For each of these classications, we investigated whether it was possible
to predict the category of RoPFC activation peaks, just from knowing the
coordinates outside RoPFC that were activated in the same contrast.
Because these classications did not divide RoPFC activation peaks
perfectly in half, chance performance for these predictions ranged from
5259% for the various classications. Performance signicantly above
chance would indicate signicantly different patterns of functional
connectivity associated with the two categories of RoPFC activations,
licensing follow-up voxel-by-voxel analyses of the extra-RoPFC regions
differentially associated with these two categories. These follow-up
analyses investigated clusters of voxels showing differential associa-
tions with the two RoPFC regions, using a uncorrected threshold of
pb .001. This relatively liberal threshold was used to discover specic
brain regions that may have contributed to the signicant omnibus test.
We performed a separate analysis for each classication of RoPFC
activation peaks (e.g. lateral versus medial). For these analyses, we
considered each RoPFC activation peak in turn. One by one, we labeled
each RoPFC activation peak the target and attempted to predict its
location from the associated extra-RoPFC activation peaks. For each
target RoPFC activation peak, we rst created a difference z-map as
above, but excluding the target RoPFC activation itself and associated
extra-RoPFC activations so that they could not bias the results. We then
e
xamined this difference z-map at each of the extra-RoPFC coordinates
associated with the target RoPFC activation. We summed the values of
the differencez-map ateach of these extra-RoPFC coordinates and made
a prediction for which category the target RoPFC activation fell into (e.g.
lateral versus medial) based on the value of this sum (greater or less
than zero). This is a form of leave-one-out cross-validation. We repeated
this process for all RoPFC activations and then compared the percentage
assigned to the correct category (e.g. lateral versus medial) against
chance using a binomial test. Where this gure was signicantly greater
than chance for a particular categorization, we then went on to analyze
clusters of voxels signicantly associated with one or the other category
of RoPFC activation peaks, using the modied activation likelihood
estimation technique described above.
Effect of task category
In a nal set of analyses, we examined the effect of task category on
the link between extra-RoPFC activations and lateral versus medial
RoPFC activations. For each extra-RoPFC region that was differentially
associated with lateral versus medial RoPFC, we rst gathered all
extra-RoPFC activations in the database within a 15 mm radius of this
region. The value of 15 mm was chosen to ensure sufcient statistical
power (mean activation peaks per region: 29.3; range 648); smaller
search volumes were associated with relatively few activation peaks
across all task categories and both RoPFC regions. These activations
were then categorized depending on which task category they came
from (Attention, Perception, Language, etc.) and on whether they
were associated with an activation peak within lateral or medial
RoPFC. This was used to generate a 8 (Task) ×2 (lateral/medial RoPFC)
contingency table, which was submitted to a chi-square test. A
signicant chi-square test would indicate that the association of a
particular extra-RoPFC region with lateral versus medial RoPFC
activations was inuenced by the task category. These tests were
implemented in SPSS Exact Tests 7.0 for Windows (Mehta and Patel,
1996) so that results were reliable even when sample sizes were small
(e.g., expected counts below 5), as in the present data where there
were relatively few activation peaks when split into different task
categories.
Table 1 (continued)
Study x y z Medial/lateral Superior/inferior Rostral/caudal Category
Wolf, R.C. et al. (2006). Neuropsychologia, 44, 2558 39 54 21 L S C Working
Yarkoni, T. et al. (2005). Cognitive Brain Research, 23, 71 38 53 3 L I C Working
2 62 3 M I R Working
Table 2
RoPFC and extra-RoPFC activations included in the meta-analysis, divided by task
category.
Task category Activations inside
RoPFC
Co-activations
outside RoPFC
N%N %
Attention 30 15 247 14
Perception 10 5 66 4
Language 7 4 60 4
Working memory 30 15 283 17
Episodic memory 49 25 558 33
Other memory 17 9 104 6
Mentalizing 31 16 197 12
Multitasking 26 13 197 12
Total 200 100 1712 100
4 S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 4
Results
A list of all additional contrasts that were included in the present
study but not Gilbert et al. (2006c) is provided in Table 1. In addition,
Table 2 provides a breakdown of the full sample of RoPFC and extra-
RoPFC activations included in the present study, according to task
category.
Predictive validity
Is it possible to predict the location of RoPFC activation peaks just
from knowing the location of associated extra-RoPFC activation peaks?
Performance was not signicantly above chance levels when predicting
left versus right hemisphere RoPFC, inferior versus superior RoPFC, or
rostral versus caudal RoPFC (55%, 47% and 55% correct respectively
versus chance levels of 54%, 52% and 50%; p N .1). Thus, we were not able
to detect any differences in functional connectivity associated with
these classications of RoPFC. However, performance was well above
chance levels when predicting the location of lateral versus medial
RoPFC activation peaks from the location of extra-RoPFC activations
(76% correct; chance: 59%; p = .000001), in dicating signicantly
different functional connectivity associated with these subregions of
RoPFC. We therefore went on to examine functional connectivity of
lateral versus medial RoPFC. The raw data, indicating all co-activations
associated with lateral versus medial RoPFC are illustrated in Fig. 1.
Functional connectivity of lateral versus medial RoPFC
Analysis of predictive validity indicated that lateral and medial
RoPFC were associated with signicantly different patterns of co-
activation outside RoPFC. But which regions outside RoPFC were
responsible for this effect? The difference z-map was examined for
voxels showing signicantly different likelihood of co-activation with
lateral versus medial RoPFC; these results are summarized in Table 3.
Activation peaks in lateral RoPFC (versus medial RoPFC) were
signicantly associated with activity in bilateral dorsolateral PFC,
anterior cingulate, bilateral lateral parietal cortex, and bilateral anterior
insula. Activation peaks in medial RoPFC (versus lateral RoPFC) were
signicantly associated with posterior cingulate, bilateral temporal pole
and posterior superior temporal sulcus. These regions are illustrated in
Fig. 2.
Effect of task category
In ve of the twelve regions listed in Table 3, differential functional
connectivity with lateral versus medial RoPFC was signicantly
Fig. 1. Activations within lateral RoPFC (dark blue) and medial RoPFC (dark red), along with associated co-activations (light blue and light red). Results are shown on eight axial slices
of the MNI 152 template brain; each co-activation is plotted on the nearest slice. (For interpretation of the references to color in this gure legend, the reader is referred to the web
version of this article.)
Table 3
Regions with signicantly different probability of co-activation with lateral versus
medial rostral prefrontal cortex. PFC = prefrontal cortex.
Region BA x y z Zmax Cluster size/mm
3
Lateral N medial
Dorsolateral PFC 9/46 45 15 35 4.89 21,250
9/46 40 20 35 3.56 625
Anterior cingulate 32 5 20 45 4.18 8625
Anterior insula 40 20 5 3.34 625
25 20 0 3.22 375
Lateral parietal cortex 40 40 45 50 3.20 250
40 45 50 40 3.72 2625
40 35 65 35 3.71 3875
7 15 70 45 3.22 250
MedialN lateral
Temporal pole 21 50 5 30 3.33 625
Posterior cingulate 23/31 0 55 20 3.80 1875
Posterior superior
temporal sulcus
39 50 65 20 3.67 1625
5S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 5
modulated by task category (chi-square exact test, pb .05). Posterior
cingulate was generally associated with activations in medial RoPFC,
but in the episodic retrieval task category p osterior cing ulate
activations were more commonly associated with lateral RoPFC.
Conversely, anterior cingulate, lateral parietal cortex (45, 50, 40),
and bilateral dorsolateral PFC were generally associated with
activations in lateral RoPFC, but in the mentalizing task category
were associated with medial RoPFC. These results are shown in
Table 4 and Fig. 3. Thus, differential functional connectivity with
lateral versus medial RoPFC was not always constant across studies, at
least assuming that different extra-RoPFC activation peaks within a
15 mm-radius sphere may be considered to correspond with the same
region.
Activation versus deactivation
While the majority of the 200 RoPFC activation peaks entered into
the meta-analysis resulted from increased signal in an experimental
condition versus a control condition, seven resulted from a contrast of
a control versus experimental condition (i.e. a deactivation). All of
these were found in the larger sample of studies described by Gilbert
et al. (2006c). In order to assess the possible effect of including both
activations and deactivations in our sample, the analysis of co-
activations with lateral versus medial RoPFC was rerun with these
seven RoPFC activations (and associated co-activations) excluded.
Results were similar: all of the 12 activation peaks listed in Table 3
remained signicant (pb .001) even when deactivations were exclud-
ed. Furthermore, the prevalence of deactivations did not differ
signicantly according to task category (χ
2
=7.4, df=7, p =.35).
Thus the present results could not result from the inclusion of both
activations and deactivations in the meta-analysis.
Discussion
This study investigated functional connectivity associated with
different subregions within RoPFC, by investigating co-activations
between particular areas inside and outside RoPFC across a database
of neuroimaging studies. There was no evidence for distinct
connectivity associated with left versus right RoPFC, superior versus
inferior RoPFC, or rostral versus caudal RoPFC. However, there were
clear differences between lateral and medial RoPFC, such that
knowledge of the extra-RoPFC activation peaks alone was sufcient
to predict the location of the co-activated region of RoPFC with 76%
accuracy (against 59% expected by chance).
The location of extra-RoPFC regions associated with lateral versus
medial subregions of RoPFC accorded well with previous anatomical
investigations of RoPFC connectivity in non-human primates. In the
rhesus monkey, Barbas et al. (1999) have reported connections
between medial prefrontal cortex and posterior superior temporal
sulcus. In the macaque monkey, Petrides and Pandya (2007) reported
Fig. 2. Regions with signicantly different probability of co-activation with lateral versus medial rostral prefrontal cortex. Warm colors indicate greater probability of co-activation
with medial rostral prefrontal cortex; cool colors indicate greater probability of co-activation with lateral rostral prefrontal cortex. Results are shown at an uncorrected threshold of
pb .001. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)
Table 4
Contingency tables for ve regions, describing the number of activations within 15 mm of each region, divided according to task category and whether there was a co-activation in
medial or lateral RoPFC. In all ve regions, the chi-square test on these data was signicant (pb .05).
Left DLPFC
(45, 15, 35)
Right DLPFC
(40, 20, 35)
Anterior cingulate
(5, 20, 45)
Lateral parietal cortex
(45, 50, 40)
Posterior cingulate
(0, 55, 20)
Medial Lateral Medial Lateral Medial Lateral Medial Lateral Medial Lateral
Attention 0 3 0 4 2 9 2 3 2 1
Perception 0 0 2 0 1 1 0 1 1 0
Language 0 2 1 0 0 2 0 0 1 0
Working memory 0 9 0 7 0 10 0 7 1 0
Episodic retrieval 0 19 0 10 0 15 0 16 1 6
Other memory 1 2 1 1 0 0 0 1 1 1
Mentalizing 1 0 2 1 3 1 2 0 15 0
Multitask 0 5 0 7 1 3 2 5 4 0
6 S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 6
efferent connections from RoPFC to anterior and posterior cingulate,
insula, temporal pole, and superior temporal sulcus, all of which were
implicated in the present functional neuroimaging results. However,
this study did not systematically compare lateral and medial RoPFC.
Petrides and Pandya (2007) note the absence of direct connections
between RoPFC and occipital and inferotemporal visual areas, and
parietal cortex. Consistent with this observation, the present study did
not produce any evidence of functional connectivity between RoPFC
and occipital or inferotemporal areas. However, there was clear
evidence for greater functional connectivity between lateral parietal
cortex and lateral rather than medial RoPFC, despite the absence of
anatomical evidence for direct RoPFCparietal pathways. One possi-
bility is that functional connectivity between RoPFC and lateral
parietal cortex is mediated via connections of the two regions with
caudal DLPFC (BA 9/46). This is supported by studies of the macaque,
in which there are dense connections between lateral RoPFC and
caudal DLPFC (Petrides and Pandya, 2007), which in turn is strongly
connected with parietal cortex (Petrides and Pandya, 1999).
It was not inevitable that the present ndings, regarding functional
connectivity of medial versus lateral RoPFC, would be congruent with
previous anatomical studies in non-human primates. Strong differ-
ences in RoPFC anatomy have been reported between humans and
other primates, both in relative size (Semendeferi et al., 2001) and
cytoarchitecture (Carmichael and Price, 1994; Ongur et al., 2003). Of
course, the present results concerning functional connectivity do not
necessarily i mply anatomical connectivity. However, functional
connectivity between a pair of regions suggests the plausibility of
anatomical connectivity, which may be then examined in studies of
non-human primates, or non-invasively in humans with tractography
(Johansen-Berg and Rushworth, 2009). Such studies may help to
clarify anatomical subdivisions within RoPFC, which are at present
poorly dened.
Our results also t well with theoretical accounts of large-scale
brain networks derived from other functional imaging studies. In
particular, lateral PFC, anterior cingulate and superior parietal cortex
have been proposed to play a co-ordinated role in a variety of
situations involving high cognitive demand (Duncan and Owen, 2000;
Duncan, 2005). By contrast, medial PFC, posterior cingulate and
temporo-parietal junction have been implicated in low-demand
situations, including rest, and have therefore been proposed to
participate in a default mode of brain function (Raichle et al., 2001).
Some authors refer to the network of regions associated with lateral
and medial RoPFC as task positive and task negative networks
respectively (Fox et al., 2005). Although there is clear evidence that
these two networks tend to be activated in different experimental
paradigms, such as those involving slow versus fast reaction times
(Burgess et al., 2007; Gilbert et al., 2006b), their functional role is still
a matter of considerable debate. In particular, although it is generally
agreed that the task negative or default mode network supports
cognitive processes that are more common in low-demand situations,
the precise nature of these processes is not clear (e.g. involvement in
task-unrelat
ed mind-wandering, enhanced perceptual attention, or a
combination of the two; Gilbert et al., 2007b; Mason et al., 2007).
One advantage of the present technique is that it can reveal the
extent to which patterns of functional connectivity depend on task
category. There were clear effects of task category on functional
connectivity between RoPFC and posterior cingulate, anterior cingu-
late, lateral parietal cortex and bilateral dorsolateral PFC. These results
suggest that participation in large-scale networks such as those
described above does not reect immutable links between sets of
brain regions but may instead reect dynamic shifts in effective
connectivity to accomplish particular tasks. Of course, it is possible
that a larger sample size and/or a different taxonomy of tasks might
have revealed additional effects of task catego ry on functional
connectivity between RoPFC extra-RoPFC regions. Furthermore, the
studies included in the meta-analysis are probably not exhaustive of
all the different states into which the brain can organize itself. Future
studies may therefore reveal additional exceptions to the general
patterns of functional connectivity reported here.
The effects of task category on functional connectivity suggest that
pairs of regions such as medial RoPFC and posterior cingulate have
functionally dissociable roles, seeing as they tend to co-activate in
some task categories but not others. It is therefore clear that even
though it may be helpful to conceive of particular sets of brain regions
as part of a functionally connected network (e.g. the default mode
network), the component regions will have distinct roles. In order to
understand these roles it will be necessary to contrast their responses
in well-controlled experimental paradigms. It seems less likely that
unconstrained, multi-componential cognitive states such as rest,
over which there is little or no experimental control, will yield
insights that demarcate the precise roles of individual brain regions
that tend to be co-activated in low-demand conditions (see Christoff
et al., 2009a; Gilbert et al., 2006a, 2007b for further discussion).
Inspection of Fig. 2 suggests that the effect of task category on
functio nal connectivity of RoPFC was largely driven by studies
involving mentalizing. For these studies, four brain regions that
typically co-activate with lateral RoPFC (bilateral dorsolateral PFC,
anterior cingulate and lateral parietal cortex) instead tended to co-
activate with medial RoPFC. Previous studies have suggested that,
within RoPFC, mentalizing is specically associated with a posterior
Fig. 3. Percentage of co-activations in medial versus lateral RoPFC, in the ve regions whose likelihood of co-activation with medial versus lateral RoPFC was signicantly modulated
by task category. All RoPFC activation peaks underwent a binary classication into medial or lateral, so a gure of (e.g.) 80% medial indicates 20% lateral. Results are shown separately
for the ve task categories that were associated with at least one activation in every region. The rst four regions tend to co-activate with lateral RoPFC whereas the last region
(posterior cingulate) tends to co-activate with medial RoPFC. However, this regularity was signicantly affected by task category. In the mentalizing category, regions that typically
co-activated with lateral RoPFC instead co-activated with medial RoPFC. In the episodic retrieval category, posterior cingulate, which typically co-activated with medial RoPFC,
instead co-activated with lateral RoPFC.
7S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 7
medial region (Gilbert et al., 2006c, 2007a; Simons et al., 2008). The
present results suggest that mentalizing studies may promote
interactions between medial RoPFC and a variety of other brain
regions, even those that typically co-activate with lateral rather than
medial RoPFC. This would be consistent with the idea that mentalizing
is not supported by a strictly modular system (Fodor, 1983) but
instead involves gathering information from a variety of domains; in
other words mentalizing is not informationally encapsulated
(Goldman, 2006, pp. 104106).
It should be noted that the relationship between functional con-
nectivity and task category was tested in analyses in which extra-
RoPFC activations within a 15 mm sphere were denoted as a single
region. This was important to ensure sufcient statistical power for
these analyses. However, it is possible that at a ner level of analysis
the extra-RoPFC co-activations associated with different categories of
task might be s patially distinct. Indeed, s tudies of functional
specialization within RoPFC indicate considerable ne-grained segre-
gation of function (Gilbert et al., 2006c, 2010). Thus, although the
present results are valid at a relatively coarse level of analysis, it
remains to be seen whether they would also hold at a more detailed
level. In order to address this question, a larger database of studies
would be required, or additional neuroimaging experiments crossing
multiple task demands within the same study. For further discussion
of the question of functional specialization at multiple levels of
analysis, see Henson (2005) and Gilbert et al. (2010).
Although the present study revealed clear differences in functional
connectivity between lateral and medial RoPFC, there were no
signicant hemispheric effects, or effects of rostral versus caudal or
inferior versus superior RoPFC. This is perhaps surprising because
functional differences have been reported both between rostral and
caudal RoPFC (Gilbert et al., 2006c, 2007a; Simons et al., 2008) and
between inferior and superior RoPFC (Mitchell et al., 2006; Van
Overwalle, 2009). One possibility is that the present database
(containing 200 activations in RoPFC and 1712 co-activations) was
too small, or the voxel size of 5 mm
3
too coarse, for sufcient statis-
tical power to detect differences in connectivity along these axes.
There may be less variance in connectivity along these axes than the
medial/lateral axis, or the types of study that would reveal this
variance may have been under-represented in the literature to date.
For example, recent studies have suggested a relationship between
activation along the rostro-caudal axis of the RoPFC and the level of
abstraction of task materials and/or complexity of task demands
(Christoff et al., 2009b). However, many studies do not systematically
manipulate (or control for) the abstractness of task materials and
therefore may potentially yield activations in both rostral and caudal
sections of RoPFC, obscuring the ability to discern unique connectivity
patterns. In addition, the present approach, using a database simply
indicating the presence or absence of a signicant signal change based
on a single RoPFC activation peak from each contrast, may have
missed differences in functional connectivity expressed as differences
in the strength of co-activations (e.g. bilateral activation in a par-
ticular region, more signicant in the hemisphere ipsilateral to the
activated area of RoPFC). Alternatively, it may be the case that
functional differences between RoPFC subregions along these axes
exist in the absence of systematic differences in functional connec-
tivity, i.e. that different RoPFC subregions perform distinct intrinsic
computations with equivalent input and output representations.
In summary, the present study established clear evidence for
differences in functional connectivity between lateral and medial
RoPFC. Furthermore, we found that these differences in connectivity
were in some cases modulated by task category. These results
underline the importance of delineating 1) particular subregions
within RoPFC, and 2) the specic roles of brain regions that tend to co-
activate, rather than treating them as functionally homogenous. In
addition, the relationship between functional and anatomical con-
nectivity of RoPFC remains to be explored. The present results may
provide a starting point for investigations of anatomical connectivity
of RoPFC in the human brain using tractography.
Acknowledgments
SJG was supported by a Royal Societ y University R esearch
Fellowship; EV was supported by the Fondation Bettencourt Schueller.
References
Barbas, H., Ghashghaei, H., Dombrowski, S.M., Rempel-Clower, N.L., 1999. Medial
prefrontal cortices are unied by common connections with superior temporal
cortices and distinguished by input from memory-related areas in the rhesus
monkey. J. Comp. Neurol. 410, 343367.
Brett, M., Christoff, K., Cusack, R., Lancaster, J., 2001. Using the Talairach atlas with the
MNI template. Neuroimage 13, 85.
Burgess, P.W., Dumontheil, I., Gilbert, S.J., 2007. The gateway hypothesis of rostral
prefrontal cortex (area 10) function. Trends Cogn. Sci. 11, 290298.
Cabeza, R., Nyberg, L., 2000. Imaging cognition II: an empirical review of 275 PET and
fMRI studies. J. Cogn. Neurosci. 12, 147.
Carmichael, S.T., Price, J.L., 1994. Architectonic subdivision of the orbital and medial
prefrontal cortex in the macaque monkey. J. Comp. Neurol. 346, 366402.
Christoff, K., Gordon, A.M., Smallwood, J., Smith, R., Schooler, J.W., 2009a. Experience
sampling during fMRI reveals default network and executive system contributions
to mind wandering. Proc. Natl Acad. Sci. USA 106, 87198724.
Christoff, K., Keramatian, K., Gordon, A.M., Smith, R., Madler, B., 2009b. Prefrontal
organization of cognitive control according to levels of abstraction. Brain Res. 1286,
94105.
Collins, D.L., Neelin, P., Peters, T.M., Evans, A.C., 1994. Automatic 3D intersubject
registration of MR volumetric data in standardized Talairach space. J. Comput.
Assist. Tomogr. 18, 192205.
Duncan, J., Owen, A.M., 2000. Common regions of the human frontal lobe recruited by
diverse cognitive demands. Trends Neurosci. 23, 475483.
Duncan, J., 2005. Prefrontal cortex and Spearman's g. In: Duncan, J., Phillips, L., McLeod,
P. (Eds.), Measuring the Mind: Speed, Control, and Age. Oxford University Press,
Oxford, pp. 249272.
Fodor, J., 1983. The Modularity of Mind. MIT Press, Cambridge, MA.
Fox, M.D., Snyder, A.Z., Vincent, J.L., Corbetta, M., Van Essen, D.C., Raichle, M.E., 2005.
The human brain is intrinsically organized into dynamic, anticorrelated functional
networks. Proc. Natl Acad. Sci. USA 102, 96739678.
Friston, K.J., Frith, C.D., Fletcher, P., Liddle, P.F., Frackowiak, R.S.J., 1996. Functional
topography: multidimensional scaling and functional connectivity in the brain.
Cerebral Cortex 6, 156164.
Gilbert, S.J., Simons, J.S., Frith, C.D., Burgess, P.W., 2006a. Performance-related activity in
medial rostral prefrontal cortex (area 10) during low-demand tasks. J. Exp. Psychol.
Hum. Percept. Perform. 32, 4558.
Gilbert, S.J., Spengler, S., Simons, J.S., Frith, C.D., Burgess, P.W., 2006b. Differential
functions of lateral and medial rostral prefrontal cortex (area 10) revealed by
brainbehavior associations. Cereb. Cortex 16, 17831789.
Gilbert, S.J., Spengler, S., Simons, J.S., Steele, J.D., Lawrie, S.M., Frith, C.D., Burgess, P.W.,
2006c. Functional specialization within rostral prefrontal cortex (area 10): a meta-
analysis. J. Cogn. Neurosci. 18, 932948.
Gilbert, S.J., Dumontheil, I., Simons, J.S., Frith, C.D., Burgess, P.W., 2007a. Comment on
Wandering minds: the default network and stimulus-independent thought.
Science 317, 43 author reply 43.
Gilbert, S.J., Williamson, I.D., Dumontheil, I., Simons, J.S., Frith, C.D., Burgess, P.W.,
2007b. Distinct regions of medial rostral prefrontal cortex supporting social and
nonsocial functions. Soc. Cogn. Affect. Neurosci. 2, 217226.
Gilbert, S.J., Henson, R.N., Simons, J.S., 2010. The scale of functional specialization within
human prefrontal cortex. J. Neurosci. 30, 12331237.
Goldman, A.I., 2006. Simulating Minds: The Philosophy. Oxford University Press, New
York, Psychology and Neuroscience of Mindreading.
Henson, R.N., 2005. What can functional imaging tell the experimental psychologist?
Q. J. Exp. Psychol. A 58, 193233.
Johansen-Berg, H., Rushworth, M.F., 2009. Using diffusion imaging to study human
connectional anatomy. Annu. Rev. Neurosci. 32, 7594.
Kober, H., Barrett, L.F., Joseph, J., Bliss-Moreau, E., Lindquist, K., Wager, T.D., 2008.
Functional grouping and corticalsubcortical interactions in emotion: a meta-
analysis of neuroimaging studies. Neuroimage 42, 9981031.
Koski,
L., Paus, T., 2000. Functional connectivity of the anterior cingulate cortex within
the human frontal lobe: a brain-mapping meta-analysis. Exp. Brain Res. 133, 5565.
Laird, A.R., Fox, P.M., Price, C.J., Glahn, D.C., Uecker, A.M., Lancaster, J.L., Turkeltaub, P.E.,
Kochunov, P., Fox, P.T., 2005. ALE meta-analysis: controlling the false discovery rate
and performing statistical contrasts. Hum. Brain Mapp. 25, 155164.
Mason, M.F., Norton, M.I., Van Horn, J.D., Wegner, D.M., Grafton, S.T., Macrae, C.N., 2007.
Wandering minds: the default network and stimulus-independent thought.
Science 315, 393395.
Mehta, C.R., Patel, N.R., 1996. SPSS Exact Tests 7.0 for Windows. SPSS Inc, Chicago, IL.
Mitchell, J.P., Macrae, C.N., Banaji, M.R., 2006. Dissociable medial prefrontal contribu-
tions to judgments of similar and dissimilar others. Neuron 50, 655663.
Ongur, D., Ferry, A.T., Price, J.L., 2003. Architectonic subdivision of the human orbital
and medial prefrontal cortex. J. Comp. Neurol. 460, 425449.
8 S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 8
Petrides, M., Pandya, D.N., 1999. Dorsolateral prefrontal cortex: comparative cytoarch-
itectonic analysis in the human and the macaque brain and corticocortical
connection patterns. Eur. J. Neurosci. 11, 10111036.
Petrides, M., Pandya, D.N., 2007. Efferent association pathways from the rostral
prefrontal cortex in the macaque monkey. J. Neurosci. 27, 1157311586.
Postuma, R.B., Dagher, A., 2006. Basal ganglia functional connectivity based on a meta-
analysis of 126 positron emission tomography and functional magnetic resonance
imaging publications. Cereb. Cortex 16, 15081521.
Raichle, M.E., MacLeod, A.M., Snyder, A.Z., Powers, W.J., Gusnard, D.A., Shulman, G.L.,
2001. A default mode of brain function. Proc. Natl Acad. Sci. USA 98, 676682.
Rorden, C., Brett, M., 2000. Stereotaxic display of brain lesions. Behav. Neurol. 12, 191200.
Semendeferi, K., Armstrong, E., Schleicher, A., Zilles, K., Van Hoesen, G.W., 2001.
Prefrontal cortex in humans and apes: a comparative study of area 10. Am. J. Phys.
Anthropol. 114, 224241.
Simons, J.S., Henson, R.N., Gilbert, S.J., Fletcher, P.C., 2008. Separable form s of reality
monitoring supported by ante rior prefron tal cortex. J. Cogn. N eur osci. 20,
447457.
Talairach, J., Tournoux, P., 1988. Co-planar Stereotaxic Atlas of the Human Brain.
Thieme, Stussgart.
Toro, R., Fox, P.T., Paus, T., 2008. Functional coactivation map of the human brain. Cereb.
Cortex 18, 25532559.
Turkeltaub, P.E., Eden, G.F., Jones, K.M., Zefro, T.A., 2002. Meta-analysis of the
functional neuroanatomy of single-word reading: method and validation. Neuro-
image 16, 765780.
Van Overwalle, F., 2009. Social cognition and the brain: a meta-analysis. Hum. Brain
Mapp. 30, 829858.
Wager, T.D., Lindquist, M., Kaplan, L., 2007. Meta-analysis of functional neuroimaging
data: current and future directions. Soc. Cogn. Affect. Neurosci. 2, 150158.
9S.J. Gilbert et al. / NeuroImage xxx (2010) xxxxxx
Please cite this article as: Gilbert, S.J., et al., Distinct functional connectivity associated with lateral versus medial rostral prefrontal cortex: A
meta-analysis, NeuroImage (2010), doi:10.1016/j.neuroimage.2010.07.032
Page 9
  • Source
    • "We found that anatomically the ventral FP (both medial and orbital) is directly connected to visual cortex. Previously, it had been concluded that the human FP had no connections with posterior sensory cortex, as there was no evidence of such connections in monkeys [18] or in human functional co-activation studies [12]. Supporting a connection from PF to visual cortex, we found that ventral medial and orbital FP co-activate with lateral occipital cortex (BA39/ Angular Gyrus), and furthermore, the ventral orbital FP co-activates with extrastriate and fusiform cortex. "
    [Show abstract] [Hide abstract] ABSTRACT: The goal of the current study was to examine the pattern of anatomical connectivity of the human frontal pole so as to inform theories of function of the frontal pole, perhaps one of the least understood region of the human brain. Rather than simply parcellating the frontal pole into subregions, we focused on examining the brain regions to which the frontal pole is anatomically and functionally connected. While the current findings provided support for previous work suggesting the frontal pole is connected to higher-order sensory association cortex, we found novel evidence suggesting that the frontal pole in humans is connected to posterior visual cortex. Furthermore, we propose a functional framework that incorporates these anatomical connections with existing cognitive theories of the functional organization of the frontal pole. In addition to a previously discussed medial-lateral distinction, we propose a dorsal-ventral gradient based on the information the frontal pole uses to guide behavior. We propose that dorsal regions are connected to other prefrontal regions that process goals and action plans, medial regions are connected to other brain regions that monitor action outcomes and motivate behaviors, and ventral regions connect to regions that process information about stimuli, values, and emotion. By incorporating information across these different levels of information, the frontal pole can effectively guide goal-directed behavior.
    Full-text · Article · May 2015 · PLoS ONE
  • Source
    • "Importantly, the present RSFC analyses indicate that, even in the absence of tasks, caudal dmPFC clusters are preferentially connected to networks within ipsilateral hemispheres . We thus corroborate absent lateralization in the rostral dmPFC (Gilbert, Gonen-Yaacovi, et al. 2010 ) and existing hemispherical specialization in the caudal dmPFC. The present lateralization in the caudal dmPFC might therefore indicate a functional integration of task-dependent as well as task-independent networks. "
    [Show abstract] [Hide abstract] ABSTRACT: The human dorsomedial prefrontal cortex (dmPFC) has been implicated in various complex cognitive processes, particularly in social cognition. To unravel its functional organization,we assessed the dmPFC’s regional heterogeneity, connectivity patterns, and functional profiles. First, the heterogeneity of a dmPFC seed, engaged during social processing, was investigated by assessing local differences in whole-brain coactivation profiles. Second, functional connectivity of the ensuing dmPFC clusters was compared by task-constrained meta-analytic coactivation mapping and task-unconstrained resting-state correlations. Third, dmPFC clusters were functionally profiled by forward/reverse inference. The dmPFC seed was thus segregated into 4 clusters (rostroventral, rostrodorsal, caudal-right, and caudal-left). Both rostral clusters were connected to the amygdala and hippocampus and associated with memory and social cognitive tasks in functional decoding. The rostroventral cluster exhibited strongest connectivity to the default mode network. Unlike the rostral segregation, the caudal dmPFC was divided by hemispheres. The caudal-right cluster was strongly connected to a frontoparietal network (dorsal attention network), whereas the caudal-left cluster was strongly connected to the anterior midcingulate cortex and bilateral anterior insula (salience network). In conclusion, we demonstrate that a dmPFC seed reflecting social processing can be divided into 4 separate functional modules that contribute to distinct facets of advanced human cognition.
    Full-text · Article · Jan 2015 · Cerebral Cortex
  • Source
    • "This network has been associated with spontaneous cognition and mind wandering (Gilbert et al., 2007; Mason et al., 2007; Christoff et al., 2009; AndrewsHanna et al., 2010). This network can be distinguished from the set of regions functionally connected to lateral PFC (Gilbert et al., 2010). Some recent studies highlighted the role of the default network and of spontaneous cognition in creativity (Takeuchi et al., 2012; Wise and Braga, 2014). "
    [Show abstract] [Hide abstract] ABSTRACT: Concepts from cognitive neuroscience strongly suggest that the prefrontal cortex (PFC) plays a crucial role in the cognitive functions necessary for creative thinking. Functional imaging studies have repeatedly demonstrated the involvement of PFC in creativity tasks. Patient studies have demonstrated that frontal damage due to focal lesions or neurodegenerative diseases are associated with impairments in various creativity tasks. However, against all odds, a series of clinical observations has reported the facilitation of artistic production in patients with neurodegenerative diseases affecting PFC, such as frontotemporal dementia (FTD). An exacerbation of creativity in frontal diseases would challenge neuroimaging findings in controls and patients, as well as the theoretical role of prefrontal functions in creativity processes. To explore this paradox, we reported the history of a FTD patient who exhibited the emergence of visual artistic productions during the course of the disease. The patient produced a large amount of drawings, which have been evaluated by a group of professional artists who were blind to the diagnosis. We also reviewed the published clinical cases reporting a change in the artistic abilities in patients with neurological diseases. We attempted to reconcile these clinical observations to previous experimental findings by addressing several questions raised by our review. For instance, to what extent can the cognitive, conative, and affective changes following frontal damage explain changes in artistic abilities? Does artistic exacerbation truly reflect increased creative capacities? These considerations could help to clarify the place of creativity-as it has been defined and explored by cognitive neuroscience-in artistic creation and may provide leads for future lesion studies.
    Full-text · Article · Jul 2014 · Frontiers in Psychology
Show more