Download Inferring a dual-stream model of mentalizing from associative white

doi:10.1093/brain/awt370
Brain 2014: 137; 944–959
| 944
BRAIN
A JOURNAL OF NEUROLOGY
Inferring a dual-stream model of mentalizing from
associative white matter fibres disconnection
Guillaume Herbet,1,2,3 Gilles Lafargue,4,5 François Bonnetblanc,6,7,8 Sylvie Moritz-Gasser,1,9
Nicolas Menjot de Champfleur10 and Hugues Duffau1,2
1
2
3
4
5
6
7
8
9
10
Department of Neurosurgery, Gui de Chauliac hospital, F-34295 Montpellier, France
Institute for Neuroscience of Montpellier, INSERM 1051, Hôpital Saint Eloi, F-34091 Montpellier, France
University of Montpellier 1, F-34967 Montpellier, France
Functional Neuroscience and Pathologies Laboratory, EA-4559, Lille Nord de France University, F-59120 Loos, France
Department of Psychology, Lille Nord de France University (Lille3), F-59653 Villeneuve d’Ascq, France
Cognition, Action and Sensorimotor Plasticity Lab, INSERM U-1093, UFR STAPS, F-27877 Dijon, France
University of Montpellier 2, LIRMM, DEMAR team, CNRS, INRIA, F-34095 Montpellier, France
University Institute of France, F-75005 Paris, France
Department of Neurology, Gui de Chauliac hospital, F-34295 Montpellier, France
Department of Neuroradiology, Gui de Chauliac hospital, F-34295 Montpellier, France
Correspondence to: Prof Hugues Duffau
Department of Neurosurgery,
Gui de Chauliac Hospital,
Montpellier University Medical Centre,
80 avenue Augustin Fliche,
34295 Montpellier,
France
E-mail: [email protected]
In the field of cognitive neuroscience, it is increasingly accepted that mentalizing is subserved by a complex frontotemporoparietal cortical network. Some researchers consider that this network can be divided into two distinct but interacting subsystems
(the mirror system and the mentalizing system per se), which respectively process low-level, perceptive-based aspects and highlevel, inference-based aspects of this sociocognitive function. However, evidence for this type of functional dissociation in a
given neuropsychological population is currently lacking and the structural connectivities of the two mentalizing subnetworks
have not been established. Here, we studied mentalizing in a large sample of patients (n = 93; 46 females; age range: 18–65
years) who had been resected for diffuse low-grade glioma—a rare tumour that migrates preferentially along associative white
matter pathways. This neurological disorder constitutes an ideal pathophysiological model in which to study the functional
anatomy of associative pathways. We mapped the location of each patient’s resection cavity and residual lesion infiltration onto
the Montreal Neurological Institute template brain and then performed multilevel lesion analyses (including conventional voxelbased lesion-symptom mapping and subtraction lesion analyses). Importantly, we estimated each associative pathway’s degree
of disconnection (i.e. the degree of lesion infiltration) and built specific hypotheses concerning the connective anatomy of the
mentalizing subnetworks. As expected, we found that impairments in mentalizing were mainly related to the disruption of right
frontoparietal connectivity. More specifically, low-level and high-level mentalizing accuracy were correlated with the degree of
disconnection in the arcuate fasciculus and the cingulum, respectively. To the best of our knowledge, our findings constitute the
first experimental data on the structural connectivity of the mentalizing network and suggest the existence of a dual-stream
hodological system. Our results may lead to a better understanding of disorders that affect social cognition, especially in
neuropathological conditions characterized by atypical/aberrant structural connectivity, such as autism spectrum disorders.
Received August 3, 2013. Revised November 2, 2013. Accepted November 11, 2013
ß The Author (2014). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.
For Permissions, please email: [email protected]
A dual-stream model of mentalizing
Brain 2014: 137; 944–959
| 945
Keywords: mentalizing system; mirror system; structural connectivity; arcuate fasciculus; cingulum
Abbreviations: RME = Reading the Mind in the Eyes; VLSM = voxel-based lesion-symptom mapping
Introduction
It is now widely acknowledged that mentalizing (a key function in
understanding and successfully performing complex social interactions) is subserved by a brain-wide neural network (Amodio and
Frith, 2006; Carrington and Bailey, 2009; van Overwalle, 2009; Mar,
2011). Although the involvement of some brain areas is still subject
to debate, it is generally accepted that this highly distributed neurocognitive network is formed by the temporoparietal junction, the
precuneus and the medial and inferior-lateral areas of the prefrontal
cortex. Group-based analyses in neuropsychological populations
have repeatedly provided additional evidence for this type of frontotemporoparietal system, with some degree of right-hemisphere
dominance (Winner et al., 1998; Happé et al., 2001; ShamayTsoory et al., 2005).
Mentalizing is not necessarily an ‘all-in-one‘ cerebral process,
but may emerge from the coherent orchestrated activity of at
least two distinct neural subnetworks (Coricelli, 2005; Frith and
Frith, 2006; Keysers and Gazzola, 2007; Uddin et al., 2007;
Lieberman, 2007; Bohl and van den Bos, 2012; Spunt and
Lieberman, 2012). The frontoparietal mirror system may subserve
the low-level embodied processes (i.e. sensorimotor-based intersubjective resonance) that are involved (for example) in emotional
empathy and the decoding of proximal (motor) intentions
(Rizzolati and Craighero, 2004; Shamay-Tsoory et al., 2009;
Rizzolatti and Sinigaglia, 2010). The mirror system may thus provide a neural route for the rapid, intuitive, pre-reflective understanding of another person’s internal affective or intentional states.
Conversely, the mentalizing system per se may be concerned with
sustaining the processes that underlie higher-level reflective inferences, such as the attribution of complex distal intentions or motives. Although a recent qualitative meta-analysis of functional
MRI data suggested that these two mentalizing subsystems are
relatively independent from a functional standpoint (van
Overwalle and Baetens, 2009), data from a number of activation
or effective connectivity functional MRI studies (Zaki et al., 2009;
Lombardo et al., 2010a; Schippers et al., 2010; Spunt and
Lieberman, 2012) do not support this view. The two subsystems
may cooperate during the ascription of psychological states.
Voxel-based lesion-symptom mapping (VLSM) (Bates et al.,
2003) is an increasingly popular method for studying correlations
between behavioural data and anatomic (lesion) data and thus
drawing conclusions on the underlying neurocognitive organization. Although this inferential statistical method is free of the inherent weaknesses of group-based analyses of patients with
defined lesions (because there is no need for a priori knowledge
of a given brain process’s anatomical substrate), VLSM itself has
several limitations. When white matter association pathways are
damaged (and almost all VLSM studies show that the lesion peaks
are located on these fascicles’ trajectories), behavioural impairments could also be due to disconnection mechanisms; this
makes it difficult to draw reliable neuropsychological inferences
in terms of brain location (Rorden and Karnath, 2004; Catani
and Ffytche, 2005; Rudrauf et al., 2008a; van Overwalle, 2009;
Duffau, 2011; Catani et al., 2012). Moreover, it is increasingly
thought that high-level cognitive functions are subserved by complex, distributed neural networks (Mesulam, 1998; Bressler and
Menon, 2010), and that cognitive disorders may result from the
general dysfunction of a neurocognitive network induced by a loss
of structural connectivity (He et al., 2007a).
In the present study, we collected mentalizing data from a large
sample of patients (n = 93) having undergone surgical resection of
diffuse low-grade glioma, a rare lesion of the CNS that migrates
preferentially along the associative white matter fibres. All patients
had been operated on under local anaesthesia with a cortical and
subcortical mapping through direct electrical stimulation. Electrical
stimulation causes a transient, ‘virtual lesion’ and thus enables
identification of the structures that are functionally essential at
each stage of the resection (Duffau et al., 2002; Duffau, 2005).
In fact, most of the associative white matter connectivity required
for basic cognitive processes is never surgically removed, despite
lesion invasion (Ius et al., 2011). Consequently, analysis of patients with diffuse low-grade glioma constitutes a unique opportunity to gain a better understanding of the functional counterpart
of surgical excisions (at the cortical level) and the functional
impact of lesion invasion on white matter bundles (at the subcortical level). In this latter case, the progressive disconnection of
white matter pathway connectivity is likely to induce functional
disruption of the underlying neurocognitive networks.
Although our study addressed basic hypotheses on the involvement of specific cortical areas in each mentalizing subsystem (e.g.
the involvement of the posterior inferior frontal gyrus and the
dorsomedial prefrontal cortex in low-level and high-level mentalizing processes, respectively), we focused on predicting the structural connectivity of these networks. The mirror system a minima
is composed of the pars opercularis of the inferior frontal gyrus,
the ventral part of the premotor cortex and the inferior parietal
lobule (Rizzolatti and Craighero, 2004). This cortical frontoparietal
network is probably interconnected through the arcuate fasciculus
and the lateral superior longitudinal fasciculus (Iacoboni and
Dapretto, 2006), as both of the latter have cortical terminations
in the anterior part of the mirror system. The arcuate fasciculus is
known to connect the pars opercularis to posterior temporal and
parietal regions, whereas the lateral superior longitudinal fasciculus
connects the ventral premotor cortex and the pars opercularis to
the supramarginal gyrus (Catani and Ffytche, 2005; Makris et al.,
2005; Martino et al., 2013). Thus, if low-level aspects of mentalizing are indeed processed by the mirror network, we should be
able to find correlations between decreases in mentalizing accuracy on one hand and the degree of damage to these fasciculi on
the other (i.e. the volume of lesion infiltration).
Conversely, the mentalizing system per se broadly overlaps with
the brain’s default mode network (Schilbach et al., 2008, 2012;
Spreng et al., 2009) that is reportedly involved in self-referential
946
| Brain 2014: 137; 944–959
processes, metacognition and reflexive awareness (Raichle et al.,
2001; Greicius et al., 2003; Cavanna and Trimble, 2006). The cingulum is likely to be the structural ‘skeleton’ of the default network
(van den Heuvel et al., 2008; Greicius et al., 2009) and interconnects
the midline structures of the brain. In particular, the cingulum connects the medial prefrontal cortex with the medial posterior parietal
cortex (including the precuneus and posterior cingulate cortices).
Hence, one can legitimately hypothesize that this associative white
matter fasciculus may play an important role in inference-based
mentalizing. If so, a decrease in mentalizing accuracy should be
correlated with the degree of damage to this tract.
As detailed below, our results confirmed this hypothesis and
provide a parsimonious anatomical basis for an integrated, interactive, dual-stream model of mentalizing.
Materials and methods
Participants
A total of 93 native French speakers having undergone surgical resection for a diffuse low-grade glioma (as confirmed by postoperative
neuropathological analyses) were recruited from Montpellier
University Hospital’s Department of Neurosurgery over 42 years
(see Supplementary Table 1 for a comprehensive overview of the
sociodemographic and clinical data). All patients were operated on
by the same experienced neurosurgeon (H.D.) under local anaesthesia
(i.e. ‘awake’ surgery) with a cortical and subcortical brain mapping
achieved by direct electrical stimulation. This validated surgical technique (Ojemann and Mateer, 1979; Duffau et al., 2002) spares the
cortical and subcortical structures that remain eloquent for certain
basic cerebral processes (e.g. visuospatial and language processes). It
should be noted, however, that none of the patients were mapped for
mentalizing processes during neurosurgery.
The exclusion criteria were as follows: previous radiotherapy (which
can impair cognition); any neurological impairment (i.e. hemianopia or
contralateral superior motor disorders) or cognitive impairment (i.e.
spatial neglect) that would prevent objective behavioural testing; a
history of neurologic or psychiatric disorders; low or abnormal premorbid IQ (590, as assessed with the French version of the National
Adult Reading Test) (Mackinnon and Mulligan, 2005). Ten patients
were excluded by the latter criteria.
The study population consisted of 45 females and 48 males, with a
mean SD age of 38.35 10.31 years (range: 18–65 years), a mean
educational level (years in full-time education) of 14 3.22 (range:
9–22 years) and a mean premorbid IQ of 107.52 6.91 (range:
90–123). The behavioural assessment was always performed by the
same trained clinician neuropsychologist (G.H.) during the chronic
phase (i.e. at least 3 months after surgery).
A group of 60 healthy subjects was also enrolled in this study, to
provide control data for low-level (n = 42) and high-level (n = 18)
mentalizing tasks.
All participants gave their written, informed consent to participation
in the study. Additionally, patients agreed to the retrospective extraction of clinical and neuropsychological data from their medical records.
Behavioural tasks
We used two different behavioural tasks [the ‘Reading the Mind in the
Eyes‘ (RME) task and the ‘comic strip‘ task] to assess primarily low-
G. Herbet et al.
level, perceptive-based aspects and high-level, inference-based aspects
of mentalizing, respectively. Ninety and 85 patients were evaluated in
the RME and comic strip tasks, respectively. These tasks have been
extensively characterized and often used in lesion and imaging studies.
The RME perception-based identification task (Baron-Cohen et al.,
2001) consists of the presentation of 36 photographs of the eye region
of human faces. Participants are asked to state which of four mental
states best describes what the character ‘is feeling or thinking’ (BaronCohen et al., 1999). In our study, each photograph was presented
separately on a PowerPointÕ slide. The four possible answers were
shown at the bottom of the slide.
Inference-based mentalizing was probed using the Brunet’s modified
version of the comic strip task, which is known to engage the classical
cortical mentalizing network (including the dorsomedial prefrontal
cortex) (Sarfati et al., 1997; Brunet et al., 2000, 2003; Atique et al.,
2011). In this behavioural paradigm, comic strip scenarios with three
images are presented. The participant is then asked to select the most
logical ending from among three possible answers (two distracters and
the correct response). There are two experimental conditions, which
require different types of causal inference: (i) the ‘attribution of intentions’ condition (corresponding to mentalizing); and (ii) the ‘physical
causality‘ (control) condition. The comic strip scenarios were presented
on a laptop computer (Windows 7; quad-core 2.8 i7; 16 Gb RAM) in a
MATLABÕ environment (2008b, version 7.7, The Mathworks Inc.).
Cogent 2000 toolbox (http://www.vislab.ucl.ac.uk) was used to generate the script. Comic strips were displayed at the top of the screen
for 5 s. Next, the three possible answers (numbered from one to three,
from left to right) were displayed at the bottom of the screen.
Participants were asked to respond as quickly and accurately as possible by pressing keys 1 to 3 of the numeric pad with their right index
finger. The answers were automatically recorded by the software. The
attribution of intention and physical causality conditions each consisted
of 28 scenarios. The order of the experimental conditions was counterbalanced. In each experimental condition, the scenarios were presented in a pseudo-random order. Before the experimental phase, a
training session with 12 items (including attribution of intention and
physical causality items) was performed.
It should be noted that each of the two tasks preferentially recruits
one mentalizing subsystem or the other. Given that the RME task
requires explicit judgement (i.e. conscious processes), it necessarily engages the inference-based mentalizing network. Nevertheless, activation functional MRI studies have shown that the posterior part of the
inferior frontal gyrus (including the pars opercularis) and the ventral
premotor cortex, i.e. the anterior part of the mirror system, are
involved in this task (Baron-Cohen et al., 1999; Russell et al., 2000;
Baron-Cohen, 2006; Adams et al., 2010; Castelli et al., 2010; Moor
et al., 2012). This involvement was formally demonstrated by lesion
studies showing that RME task performance was specifically impaired
by pars opercularis damage but not by medial prefrontal damage
(Shamay-Tsoory et al., 2009; Herbet et al., 2013). Although the
comic strip task is known to predominantly activate the mentalizing
network (Brunet et al., 2000, 2003; Vollm et al., 2006; Atique et al.,
2011) and be particularly affected by medial prefrontal damage
(Herbet et al., 2013), a magnetoencephalography study has demonstrated the early activation of posterior brain regions belonging the
mirror network (Vistoli et al., 2011) during this task. Furthermore,
surgical resection of the right pars opercularis is associated with
slight slowing of the inferential process associated with intentional attribution (Herbet et al., 2013). Based on these findings, we assumed
that both tasks can simultaneously engage the mirror and mentalizing
systems to some extent, but that the intactness of each subsystem is
more important for success in one behavioural task or the other.
A dual-stream model of mentalizing
Neuroanatomical data acquisition
In all patients, structural MRI data sets were acquired at the time of
the behavioural assessment in the same medical centre as part of their
standard care. The images had been acquired using conventional axial
fluid-attenuated inversion recovery (FLAIR) or high-resolution 3D T1weighted sequences on a 1.5 T Siemens Avento or a 3 T Siemens
Skyrya scanner (Siemens Medical Systems).
Lesion mapping
Resection cavities and residual lesion infiltrations were reconstructed in
standardized MNI space by one of the investigators (G.H.). First, each
reconstruction was compared with the initial non-normalized scans and
the corresponding, detailed surgical report produced by the neurosurgeon (H.D.). Next, the work was carefully inspected by an independent, board-certified neuroradiologist (N.M.C.) who was blinded to the
behavioural data. It should be noted that six patients were excluded at
this point because structural abnormalities [abnormal ventricle size
(n = 1) and brain hygroma (n = 5)] prevented accurate normalization.
To spatially normalize individual data sets, we used SPM8 (implemented in a MATLABÕ environment) with cost function masking (Brett
et al., 2001). This method has been extensively characterized and is
proven to be the best option for normalization when the lesion is large
(Andersen et al., 2010). During the registration process, cost function
masking avoids bias caused by abnormal lesion-induced radiologic signals. Briefly, the lesion is contoured by hand and transformed into a
binarized image. This image is used as a mask for the normalization
process. The lesion is then drawn on the normalized scan to yield a
volume of interest.
We used high-resolution whole-brain 3D T1 images (resolution:
1 1 1 mm) to map the resection cavities. However, we preferred
to use FLAIR images (23 axial slices, with a resolution of
0.898 0.898 6 mm) for the lesion infiltration maps because this
sequence is known to yield the best contrast between normal brain
tissue and infiltrated brain tissue.
Voxel-based lesion-symptom analyses
We used whole-brain VLSM analyses (Bates et al., 2003) to explore
the putative relationships between mentalizing data on one hand and
the spatial locations of the surgical resections or residual lesion infiltrations on the other. To this end, we used NPN software (version:
December 2012) provided in MRIcron package (www.mricro.com/
mricron). We excluded voxels that were resected or infiltrated in
fewer than three patients. Although this threshold may appear to be
low, it has often been used in VLSM studies and, in view of the
inhomogeneous spatial distribution of the surgical resections, was the
best option here (Fig. 1). Diffuse low-grade glioma preferentially infiltrates the insular, frontal and anterior temporal cortices (Duffau and
Capelle, 2004; Parisot et al., 2012). Given that right posterior temporal
and parietal areas are of interest in mentalizing (Saxe and Kanwisher,
2003) and to not discount these regions of interest, it made sense to
take account of voxels damaged at least in three patients. However, it
should be noted that all the VLSM analyses were repeated with a
minimum exclusion threshold of six and 10 patients for resection
cavity maps (therefore excluding right posterior areas) and five patients for residual infiltration maps (to increase statistical power in
more anterior areas). The overall results did not differ significantly as
a function of the exclusion threshold, and so the latter analyses are not
presented here.
Brain 2014: 137; 944–959
| 947
The parametric t-test statistic was chosen to compute the Z-score
statistical maps; a pairwise comparison is used to test for a statistically
significant difference between behavioural scores for patients with or
without damage in a given voxel (a resected or infiltrated voxel, in our
case).
We chose not to correct the type I error for multiple comparisons
[with a Bonferroni correction or a false discovery rate (FDR) control
method, for example]. In fact, this conservative approach can lead to a
high false negative rate, especially for brain structures with low lesion
coverage (Rudrauf et al., 2008b). Hence, we decided to threshold the
resulting statistical maps at Z 4 3.09 (corresponding to P = 0.001 uncorrected), with an extent threshold of k = 70 contiguous voxels. This
decision was justified by the fact that (i) we had strong prior assumptions (at least at the cortical level); and (ii) patients with diffuse lowgrade glioma show high levels of functional plasticity. Consequently, it
was the best trade-off between sensibility and specificity (Rudrauf
et al., 2008b). However, applying a FDR correction to the statistical
maps with a threshold of q = 0.05 did not alter the main results; there
were fewer significant voxels but the spatial location of significant
peaks did not change.
The VLSM analyses were performed on each behavioural accuracy
score, expressed initially as the percentage of correct answers.
However, we corrected the raw data before subsequent analyses.
We specifically sought to remove the variance associated with certain
sociodemographic and clinical variables of no interest—variance that
otherwise may significantly distort the output results. To this end, we
performed separate multiple regression analyses on the row score from
the low-level task and the three row scores from the high-level task
(the attribution of intention and physical causality conditions and
the differential error rate when comparing the two conditions).
Standardized residuals from each model (referred to as standardized
residual–RME, standardized residual–attribution of intention, standardized residual–physical causality and standardized residual–differential
error rate) were used as inputs in subsequent lesion-behaviour analyses. This is an acceptable approach when the design is substantially
complicated by the entry of several covariates (Kimberg et al., 2007).
Indeed, this procedure has been used by some authors to partial out
unwanted (i.e. non-interest) factors in VLSM analyses (Schwartz et al.,
2011; Gläscher et al., 2012). These residual scores were also used to
study the correlation between the degree of disconnection of the associative pathways on one hand and mentalizing accuracy scores on
the other (see below).
For the residual lesion infiltration maps, VLSM analyses were performed with and without the lesion infiltration size as a covariate. This
enabled us to check whether the degree of lesion infiltration in white
matter associative pathways could account for worse mentalizing
performance.
Subtraction plots
Resection cavity maps and residual lesion infiltration maps from patients with or without poor mentalizing accuracy were contrasted by
means of MRIcron’s ‘subtraction plot’ function. This kind of imaging
analysis is especially useful for detecting brain structures that are more
frequently damaged in a set of patients with a given functional alteration (when compared with a set of control patients without this functional alteration) (Karnath et al., 2002). However, use of a subtraction
plot is only appropriate when the two groups contain approximately
the same number of individuals (Rorden and Karnath, 2004). Firstly,
lesion maps from the two subgroups (impaired and unimpaired patients, in the present study) are overlaid. Secondly, these images are
subtracted (the impaired image minus the unimpaired image) to create
948
| Brain 2014: 137; 944–959
G. Herbet et al.
Figure 1 Lesion overlap maps for all patients. (A) Resection cavity overlaps (n = 93). The pars triangularis of the right frontal gyrus was
the region with the greatest overlap (n = 26). (B) Residual lesion infiltration overlaps (n = 70). As expected, the residual infiltrations were
located on the trajectory of the associative white matter fasciculi. The greatest degree of overlap occurred in the white matter fibres of the
right inferior occipito-frontal fasciculus (n = 21). Histograms represent lesion density plots (i.e. the number of overlapping voxels).
a new image that shows the specifically damaged cortical or subcortical regions. A colour bar is used to indicate the percentage overlap
between the lesions.
Infiltration estimation
To estimate the degree to which white matter associative fasciculi
had been infiltrated, each individual residual lesion map was first
overlaid with the diffusion tensor imaging-based white matter
fasciculus atlas from Catani and Thiebaut de Schotten (2008).
Next, MRIcron’s ‘batch descriptive’ function was used to automatically compute the number of voxels that overlapped with each associative fasciculus (the inferior fronto-occipital, uncinate, inferior
longitudinal, arcuate and lateral superior longitudinal fasciculi and
the cingulum).
The correlation between infiltration volumes (expressed as the
number of infiltrated voxels) and mentalizing scores (raw scores and
standardized residuals) were analysed using a non-parametric
Spearman test. We adopted a constraining statistical context by
including only patients showing infiltrated brain tissues.
These analyses were supplemented by a group analysis approach
based on the patient subdivisions established for subtraction lesion
analyses. Using non-parametric Mann-Whitney U-test, we compared
impaired patients and unimpaired patients in terms of the infiltration
volumes in each fasciculus, the Z-score in the RME task and the differential error rate in the comic strip tasks.
Results
Sociodemographic and clinical data
The patients’ sociodemographic and clinical data are summarized
in Supplementary Table 1. The average volume of resected brain
tissue was 72.25 61.49 cm3 (range 4.1–289.9 cm3); corresponding to 94.29 6.04% of the initial lesion size (range 70–100%).
As shown in Fig. 1A, the spatial distribution of the surgical resections was inhomogeneous. This fits well with anteroposterior
gradient typically observed for diffuse low-grade gliomas, where
lesions infiltrating posterior areas are less frequent (Duffau and
Capelle, 2004; Parisot et al., 2012). It is noteworthy that there
were more patients with right-side lesions than patients with leftside lesions. This difference was expected because we excluded
patients with lesions in the left middle/posterior temporal and
A dual-stream model of mentalizing
parietal regions, to avoid possible bias because of receptive language disorders.
As expected, the remaining lesion infiltrations were mainly
located on the anteroposterior trajectories of the associative
white matter bundles (Fig. 1B). Our lesion mapping therefore
replicated the probabilistic atlas of functional resectability published by Ius et al. (2011), which showed that most of the associative white-matter connectivity cannot be removed for
functional reasons.
Behavioural data analysis
Raw mentalizing scores for each task are given in Supplementary
Table 2. In the low-level mentalizing task, the mean standard
deviation (SD) score in the patient group (n = 90) was
61.67 11% of correct responses (36.11–83.33). In the highlevel mentalizing task, the mean score in the patient group
(n = 84) was 84.69 13.34% (39.90–100) of correct responses
in the attribution of intention condition and 93.75% [ 7.29
(42.90–100)] of correct responses in the physical causality (control) condition. Low-level mentalizing scores were found to be
moderately correlated with the attribution of intention scores
(r84 = 0.254, P = 0.022), but not at all correlated with physical
causality scores (r84 = 0.081, P = 0.470).
All the accuracy scores were fed into separate multiple regression models with age, premorbid IQ, resection volume and time
since surgery as potential predictors. For the low-level task, the
overall model was not significant (R2 = 0.082, P = 0.11) but age
(unlike all the other variables; P 4 0.16) was found to be a significant predictor (b = 0.24, P = 0.03). For the high-level task,
the model was significant for all dependent variables (attribution
of intention: R2 = 0.23, P = 0.0003; physical causality: R2 = 0.225,
P = 0.0004; differential error rate: R2 = 0.11, P = 0.04). Age and
premorbid IQ made significant contributions to these models contrary to clinical variables (the complete results of these analyses
are reported in Supplementary Table 3). Standardized residuals
from these multiple regressions were used instead of raw behavioural scores as inputs for subsequent VLSM analyses. This is
enabled us to remove the unwanted variance associated with
the covariates of non-interest mentioned above.
The patient subgroups (unimpaired and impaired) for group and
subtraction plot analyses (see below) were established by applying
a clinical criterion for normality (4 1 SD versus 4 1 SD) based
on control data from healthy participants. In clinical practice, patients performing below this cut-off are usually considered to be
poor performers. Moreover, this cut-off was particularly convenient for dividing the patients into two equally sized subgroups. For
the RME low-level mentalizing task, 42 healthy participants were
divided into three age classes (Group 1: 18–34; Group 2: 35–49;
Group 3: 50–65) and their behavioural scores were used to transform the patients’ raw scores into age-adjusted Z-scores. In contrast to the comic strip task, the RME task is known to be affected
by age in French populations (Duval et al., 2011). For the comic
strip task, 18 additional control participants provided normative
data that enabled the patient’s performance levels to be transformed in Z-scores. The subtraction plot analysis was performed
Brain 2014: 137; 944–959
| 949
on the differential error rate data only to identify patients with a
specific decrease in inference-based mentalizing accuracy.
For correlation analyses between mentalizing accuracy and the
degree of lesion infiltration into associative white-matter fasciculi
(see below), age-adjusted Z-scores (for the low-level RME task)
and raw scores (for the high-level comic strip task) as well as
standardized residuals from multiple regression models (see
above) were used. Note that raw scores were directly selected
for the comic strip task because scores were not adjusted for
age, as earlier justified. Consequently, transformation into
Z-scores for this task only served to segregate patients for group
and subtraction plot analyses.
Resections of the supplementary motor
area and the dorsal premotor cortex are
associated with poor inference-based
mentalizing accuracy
We performed VLSM analyses on standardized residuals (first for
the resection cavity maps and then for the residual lesion infiltration maps; Fig. 1A and B). As explained in the ‘Materials and
methods’ section, we chose to apply a liberal threshold for statistical significance (Z 4 3.09; P 5 0.001 uncorrected). For the sake
of completeness, the summary table also reports the clusters that
survived a correction for multiple comparisons (using a FDR control procedure with q = 0.05).
For the low-level mentalizing task, our analyses of resection
cavity maps (90 maps, with 589 001 voxels tested) were inconclusive and none of the voxels exceeded the critical value. The
same was true for our analyses of residual lesion infiltration maps
(68 maps, 33 126 voxels tested), regardless of whether or not the
lesion infiltration volume was included as a covariate.
For the high-level mentalizing task, VLSM analyses of the attribution of intention data and the resection cavity maps (84 maps,
572 302 voxels tested) revealed a cluster of significant voxels in
the right angular gyrus (cluster size = 565 voxels, Zmax = 3.76) and
the right supplementary motor area (cluster size = 211 voxels,
Zmax = 4.11). However, the posterior parietal regions were not
specific to intention mentalizing, as analyses performed on the
physical causality data also highlighted a cluster of significant
voxels in the same location (cluster size = 573 voxels,
Zmax = 3.3). Accordingly, VLSM of the differential error rate data
did not identity significant voxels in this region. In contrast, the
latter analysis revealed a large cluster of significant voxels in the
right supplementary motor area, which extended to the dorsal
premotor cortex (cluster size = 2728 voxels, Zmax = 5.32) and—
albeit to a lesser extent—the junction between the pars opercularis
of the right inferior frontal gyrus, the pars triangularis and the
posterior part of the right middle frontal gyrus (cluster size = 176
voxels, Zmax = 5.32). These results are reported in Table 1 and
illustrated in Fig. 2.
At the subcortical level, VLSM analysis of the residual lesion
infiltration maps revealed a cluster of significant voxels in the
white matter underlying the right occipito-temporoparietal junction when considering both attribution of intention data (cluster
950
| Brain 2014: 137; 944–959
G. Herbet et al.
Table 1 Overall VLSM results
Peak locations
(all right-sided)
BA
Z
P uncorrected
Cluster size
(at P 5 0.001)
Cluster size
(at P 5 0.05
FDR-corrected)
VLSM results for resection cavity maps: standardized residual–attribution of intention
SMA
6
4.1
0.000021
211
0
AG
39
3.76
0.000084
565
0
VLSM results for resection cavity maps: standardized residual–physical causality
AG
39
3.3
0.00048
573
0
VLSM results for the resection cavity maps: standardized residual–differential error rate
SMA
6
5.32
0.00000005
2728
2024
mid. FG
44
3.98
0.000034
176
0
VLSM results for residual lesion infiltration: standardized residual–attribution of intention
WM TPOj
19
3.81
0.000069
210
177
VLSM results for residual lesion infiltration: standardized residual–physical causality
WM TPOj
0
4.05
0.000026
257
863
WM AG
0
3.46
0.00027
88
VLSM results for residual lesion infiltration: standardized residual–differential error rate
VLSM results in relation to RI maps + residual infiltration volumes as a covariate
MNI coordinates
x
y
z
9
49
12
60
70
43
49
60
43
21
48
8
28
64
41
31
59
22
30
25
58
48
22
46
-
-
-
Analyses operated on standardized residual–attribution of intention, standardized residual–physical causality, and standardized residual–differential error ratewere inconclusive, and revealed systematically a volume effect (values in bold: P 5 0.00001 uncorrected; P 5 0.05 FDR-corrected).
SMA = supplementary motor area; AG = angular gyrus; mid; FG = middle frontal gyrus; WM = white matter; TPOj = temporo-parieto-occipital junction; BA = Brodmann
area; RL = residual lesion.
Figure 2 Damage to the supplementary motor area and dorsal premotor cortex is associated with impaired accuracy in inference-based
mentalizing. The statistical map is thresholded at P 5 0.001 uncorrected and rendered in three dimensions. Only significant voxels are
shown. This lesion-symptom analysis was performed on the difference in error rate between the mentalizing condition (intention attribution) and the control condition (physical causality), after removing unwanted variance associated with nuisance variables (age, premorbid IQ, lesion volume and time since surgery).
size = 210, Z = 3.81) and physical causality data (cluster
size = 257, Z: 4.05). In the physical causality analysis, an additional
cluster was observed in the white matter underlying the right angular gyrus (cluster size = 88, Z = 3.46). However, analysis of the
differential error rate data did not identify any significant lesionbehaviour relationships, demonstrating that infiltration of these
white matter fibres was not specifically associated with impaired
mentalizing accuracy. Interestingly, the clusters identified for attribution of intention and physical causality disappeared when the
analyses were repeated with the volume of residual infiltration as a
covariate (Table 1).
Brain regions most frequently damaged
in patients with low mentalizing
accuracy
To visualize the brain areas the most frequently damaged in patients with mentalizing impairments, we generated maps by overlapping individual residual lesion infiltration maps of patients with
poor accuracy and maps from patients with normal accuracy. In
view of the difference in the numbers of patients with left-side
lesions and those with right-side lesions, subtraction analyses were
only performed on the latter group of patients.
A dual-stream model of mentalizing
Brain 2014: 137; 944–959
| 951
Figure 3 The most frequently damaged brain regions in patients with impaired (i) mentalizing accuracy. (A) Low-level mentalizing task,
all right-lesion patients. (B) Low-level mentalizing task, only frontal right-lesion patients. (C) High-level mentalizing task, all right-lesion
patients. The colour bar indicates increasing frequency. Only brain regions that overlapped in at least 10% of patients are shown. The lefthand plots depict accuracy scores for the two patient subgroups. ***P 5 10 6. DER = differential error rate; ui = unimpaired.
For the low-level mentalizing task, the pars opercularis of
the right inferior frontal gyrus (224 overlapping voxels in more
than 25% of the patients, Fig. 3A) was more likely to be
damaged in patients with impaired accuracy (n = 36, 53.7%)
than in patients with unimpaired accuracy (n = 31, 46.3%). This
was also true when considering only patients with a resection
in the right frontal lobe (n = 20 patients with impaired accuracy
versus n = 23 with unimpaired accuracy; 148 overlapping
voxels in the right pars opercularis in 4 45% of the patients;
Fig. 3B).
For the high-level mentalizing task, the pars orbitalis of the right
inferior frontal gyrus (42 voxels), the posterior part of the right
middle frontal gyrus (142 voxels) and the supplementary motor
area (52 voxels) were more likely to have been resected in patients
with unimpaired performance (525%, Fig. 3C).
Damage to the right arcuate fasciculus
is specifically associated with impaired
low-level mentalizing accuracy
We estimated the degree of disconnection (i.e. the volume of
infiltration) for each associative white matter fascicule and for
each patient. These volumes were then related to the mentalizing
data. We considered both age-adjusted Z-scores and standardized
residuals from multiple regressions to test the robustness of the
952
| Brain 2014: 137; 944–959
G. Herbet et al.
results after controlling for nuisance variables. Only patients with
residual infiltrations were considered. We also restricted these
analyses to right-sided lesion patients; there were not enough
patients with left-side lesions to draw reliable and definitive conclusions on the possible involvement of left white matter pathways in mentalizing.
As expected, we found significant negative correlations between
low-level mentalizing performance and infiltration volumes in the
right arcuate fasciculus (Z-scores: r47 = 0.38, P = 0.009;
residuals: r47 = 0.35, P = 0.016) and the lateral superior
longitudinal fasciculus (Z-scores: r47 = 0.35, P = 0.017; residuals:
r47 = 0.35, P = 0.048). Performance was also significantly
and negatively correlated with the total volume of infiltration
(r47 = 0.29, P = 0.047; residuals: r47 = 0.69, P = 0.047). None
of the other correlations were statistically significant (Table 2).
In support of these results, impaired patients were significantly
more infiltrated in the right arcuate fasciculus than unimpaired
patients (Z = 1.95, P = 0.05, Fig. 4A), confirming the qualitative
result from the subtraction lesion analyses (Fig. 4B).
Damage to the right cingulum is
specifically associated with impaired
high-level mentalizing accuracy
For the high-level mentalizing task, the infiltration volumes in the
cingulum were negatively correlated with the differential error rate
Table 2 Correlation analyses between infiltration volumes in right white matter associative pathways and mentalizing
accuracy scores
Age
IQ
Low-level perceptive-based mentalizing
Z-scores
0.16
0.12
SR
0.016
0.02
High-level inference-based mentalizing
AI
-0.39**
0.19
SR AI
0.02
0.06
PhC
-0.39**
0.25
SR PhC
0.06
0.12
DER
0.23
0.15
SR DER
0.02
0.002
RIV
-0.29*
-0.29*
0.14
0.17
0.003
0.004
0.11
0.12
IFOF
ILF
UF
0.18
0.20
0.2
0.21
0.025
0.002
0.032
0.012
0.062
0.03
0.06
0.04
0.04
0.05
0.036
0.03
0.02
0.03
0.07
0.005
0.02
0.07
0.13
0.08
AF
-0.38**
-0.35*
0.22
0.16
0.11
0.08
0.18
0.15
lSLF
Cing.
-0.35*
-0.29*
0.19
0.19
0.13
0.03
0.16
0.086
0.07
0.027
-0.28*
-0.27
0.03
0.042
-0.33*
-0.31*
Analyses were performed with a non-parametric correlation test. Numeric values in the table correspond to the Spearman’s rank correlation coefficient. RCV = resection
cavity volume; RIV=residual infiltration volume; IFOF=inferior fronto-occipital fascicle, ILF=inferior longitudinal fascicle; UF=uncinate fascicle; AF=arcuate fascicle;
lSLF=lateral longitudinal fascicle; Cing.=cingulum. *P 5 0.05; **P 5 0.01; P 5 0.1.
Figure 4 Damage to the right arcuate fasciculus is specifically associated with decreased low-level mentalizing accuracy. (A) Group
analyses: the residual lesion volumes in each associative pathway were compared in patients with impaired (i) and without impaired (ui)
mentalizing performance. Only the arcuate fasciculus showed a group effect (Z = 1.95, P = 0.05). All other comparisons were nonsignificant (P 4 0.1, see also Table 2). (B) Subtraction plots: patients with impaired performance were more likely to have infiltration in
the arcuate fasciculus. The bar indicates increasing frequency. Only brain regions that overlapped in at least 10% of patients are shown.
The left arcuate fasciculus is plotted as a visual point of reference. IFOF = inferior fronto-occipital fasciculus; ILF = inferior longitudinal
fasciculus; UF = uncinate fasciculus; AF = arcuate fasciculus; lSLF = lateral superior longitudinal fasciculus.
A dual-stream model of mentalizing
Brain 2014: 137; 944–959
| 953
Figure 5 Damage to the right cingulum is specifically associated with decreased high-level mentalizing accuracy. (A) Patients with an
impaired differential error rate were more likely to have infiltration in the right cingulum than patients with a normal differential error rate
(P = 0.03, see also Table 2). Subtraction plot: patients with impaired performance were more likely to have infiltration in the cingulum and
in white matters fibres constituting the temporo-parieto-occipital junction. The bar indicates increasing frequency. The left cingulum is
plotted as a visual point of reference. IFOF = inferior fronto-occipital fasciculus; ILF = inferior longitudinal fasciculus; UF = uncinate
fasciculus; AF = arcuate fasciculus; lSLF = lateral superior longitudinal fasciculus.
(r47 = 0.33, P = 0.024), even after controlling for nuisance variables (standardized residual–differential error rate: r47 = 0.33,
P = 0.036). None of the other correlations were statistically significant. These analyses were also performed for the attribution of
intention and physical causality data sets (Table 2).
These findings were in accordance with group analyses showing
that lesion infiltrations in the cingulum were significantly larger in
impaired patients (n = 21) than in unimpaired patients (n = 26)
(differential error rate, Z = 2.075, P = 0.038). None of the other
correlations achieved statistical significance (Fig. 5A).
Subtraction lesion analyses confirmed the above results
(Fig. 5B). Impaired patients were likely to have infiltration in the
cingulum and in the posterior part of the lateral frontoparietal
network (including the white matter deep in the temporoparietal
and temporo-parieto-occipital junctions). These findings are in line
with data (Fig. 5A) showing that impaired and unimpaired patients
also differed (albeit not significantly) in terms of infiltration volumes in the arcuate fasciculus and the lateral superior longitudinal
fasciculus (but not in terms of the ventral associative fasciculi,
including the inferior fronto-occipital fasciculus and inferior lateral
fasciculus.
Discussion
Diffuse low-grade glioma migrates preferentially along white matter
associative pathways. Accordingly, we used this neurological disorder as a physiopathological model of the functional anatomy of
the associative fasciculi. To date, methods for relating cognitive or
neurological impairments to the disconnection of long-range white
matter fibres have been qualitative (Bartolomeo, 2012) and involved
establishing whether a given lesion is located within associative fasciculi (Bartolomeo et al., 2007; Thiebaut de Schotten et al., 2008;
Karnath et al., 2011). Quantitative methods are now being developed, in an attempt to gauge the respective roles of cortical
and subcortical damage in the occurrence of cognitive disorders
(Rudrauf et al., 2008a; Philippi et al., 2009). Here, we adopted
the same type of approach by estimating the degree of disconnection of each associative fascicule to make specific predictions about
the connectional anatomy of mentalizing. However, the main
strength of the present study was our ability to identify critical structures for mentalizing processes at both the cortical and subcortical
levels (by studying cortical resections and infiltrated deep connectivity, respectively) (Fig. 1). To the best of our knowledge, the present
study is the first to provide data on the structural connectivity of
mentalizing subnetworks. Our results suggest that this complex
sociocognitive function is performed by a dual-stream hodological
system.
Disconnection of the right arcuate
fasciculus/lateral superior longitudinal
fasciculus and impaired
perceptive-based mentalizing accuracy
Dissection studies and diffusion tensor imaging-based studies have
shown that the anterior part of the mirror neuron system
(including the posterior inferior frontal gyrus and the ventral premotor cortex) is connected to posterior temporoparietal areas by
the arcuate fasciculus and the lateral/superficial superior longitudinal fasciculus (Catani and Ffytche, 2005; Makris et al., 2005;
Martino et al., 2013). As previously stated by some authors
(Iacoboni and Dapretto, 2006), this perisylvian network might
therefore constitute the main pathway for the frontoparietal
mirror network (given its cortical terminations in putative mirror
areas). Consequently, we hypothesized that damage to this tract
954
| Brain 2014: 137; 944–959
could account for impairments in low-level mentalizing performance. Our present results argue in favour of this hypothesis.
Accuracy scores were specifically correlated with the degree of
infiltration of the arcuate fasciculus and the left superior longitudinal fasciculus (Table 2), and patients with impaired versus unimpaired mentalizing performance differed significantly in terms of
infiltration volumes in the arcuate fasciculus (Fig. 4A and B). This
finding was also consistent with the fact that the junction between
the right pars opercularis and the right ventral premotor cortex
was the most frequently damaged brain region in impaired patients (Fig. 3A and B).
Our data fit well with the results of two recent studies in which
the right arcuate fasciculus/lateral superior longitudinal fasciculus
was suggested to have an important role in social cognition. In a
patient population, voxel-based lesion-symptom mapping showed
that damage to these pathways was associated with poor social
and emotional intelligence (Barbey et al., 2012). In healthy subjects, interindividual variability in emotional empathy (a brain process known to recruit the mirror system; Shamay-Tsoory et al.,
2009) was positively correlated with the fractional anisotropy in
certain parts of these associative fasciculi (and especially in the
right hemisphere), suggesting that structural hyperconnectivity of
the arcuate fasciculus/lateral superior longitudinal fasciculus is
associated with greater empathic abilities (Parkinson and
Wheatley, 2012).
Disconnection of the right cingulum and
impaired inference-based mentalizing
accuracy
In activation functional MRI studies, the medial prefrontal cortex
(including the more rostral part of the anterior cingulate cortex)
and the precuneus have always been systematically linked to inference-based mentalizing (Amodio and Frtih, 2006; Carrington
and Bailey, 2009; van Overwalle and Baetens, 2009).
Furthermore, connectivity studies have also showed strong functional interactions between the medial prefrontal cortex and the
precuneus during sociocognitive tasks in general and inferencebased mentalizing in particular (Atique et al., 2011).
Interestingly, these two cortical nodes also belong to the socalled default mode network, which is thought to be involved in
self-referential cognition and overlaps greatly with the social cognition network (Schilbach et al., 2008, 2012; Spreng et al., 2009;
Mars et al., 2012). Given that the cingulum provides strong connections between the rostral medial prefrontal cortex/anterior cingulate and the medial posterior parietal cortex (including the
posterior cingulate cortex and ventral precuneus), the cingulum
is likely to shape the structural connectivity of the default network,
as recently suggested by two multimodal imaging studies (van den
Heuvel et al., 2008; Greicius et al., 2009). Based on these findings, we hypothesized that disconnection of the cingulum was
related to behavioural changes in inference-based mentalizing.
Our present data provide direct support for this hypothesis.
Impaired mentalizing accuracy was specifically correlated with
the extent to which the right cingulum was infiltrated (Table 2).
This result was corroborated by group analyses, which showed
G. Herbet et al.
that infiltration volumes in the right cingulum were significantly
larger in patients showing specific impairments in inference-based
mentalizing; this was not the case for the other fasciculi, and especially the ventral associative pathways (Fig. 5A and B).
Understanding other’s intentions is
impaired by damage to the supplementary motor area/dorsal premotor cortex
Both subtraction lesion analyses and VLSM analyses highlighted
the large number of significantly involved voxels in the right supplementary motor area (extending to the dorsal premotor regions)
(Fig. 2). These brain regions are known to harbour neurons with
mirror properties (Mukamel et al., 2010) and have a crucial role in
understanding intentions and actions (Iacoboni et al. 2005; Spunt
and Lieberman, 2013). However, a growing body of evidence
suggests that the mirror system is involved in computing proximal
(motor) intentions but not distal intentions necessitating (which,
by their very nature, require more reflective processes) (de Lange
et al., 2008; Rizzolatti et al., 2010). Nevertheless, the mirror
system is likely to facilitate psychological inference by automatically decoding low-level intentional cues during the course of complex action plans. Our results indicate that damage to the cortical
structures thought to subserve automatic processes in intention
decoding can also alter the ability to understand intention (i.e. a
higher-level process). Our findings suggest that both the mirror
and the mentalizing system are engaged during the conscious attribution of intentions (as previously demonstrated in a functional
connectivity study using the same task (Kana et al., 2014) and,
more generally, during the ascription of psychological states
(Spunt and Lieberman, 2012). In this respect, it is noteworthy
that decreasing the threshold to P 5 0.005 (uncorrected) in the
VLSM analyses revealed an additional cluster of voxels in the pars
opercularis of the right posterior inferior frontal gyrus. Taken as a
whole, our results fit well with the interactive (cooperative) dualprocess theory of mentalizing, in which it is posited that low-level
embodied processes and high-level inferential processes are
engaged during the ascription of psychological states (Uddin
et al., 2007; Keysers and Gazzola, 2007; Lombardo et al.,
2010a; Spunt and Lieberman, 2012).
A plastic hodological mentalizing
network
One can legitimately wonder why our conventional voxel-wise
analysis of a relatively large sample of patients failed to establish
comprehensive statistical maps of the relationship between impaired mentalizing processes and the location of brain damage.
One limitation of the current study relates to the fact that few
patients had lesions in the right temporoparietal junction (as a
result of natural spatial distribution of diffuse low-grade gliomas).
However, this limitation did not apply to more anterior temporal
and frontal regions. On the basis of data from lesion and multimodal imaging studies that featured the behavioural tasks used
here (Brunet et al., 2000, 2003; Adams et al., 2010; Castelli
et al., 2010; Moor et al., 2012; Herbet et al., 2013), one would
A dual-stream model of mentalizing
expect injury to the posterior inferolateral or medial frontal areas
to impact task performance (at least to some extent) in conventional, quantitative lesion-symptom analyses. The absence of significant findings cannot be reasonably attributed to low statistical
power. Firstly, we also performed VLSM analyses with higher
voxel cut-offs (5 and 10 for resection cavity maps, and 5 for residual lesion infiltration maps) to improve the statistical power.
Secondly, we sought to avoid false negative results by using a
liberal threshold for statistical significance.
One possible explanation relates to the lesions’ neuropathological. In contrast to acute lesions (such as vascular stroke), diffuse low-grade glioma slowly infiltrates brain tissues (Mandonnet
et al., 2003) and can sometimes induce impressive functional
remodelling (Desmurget et al., 2007) (as demonstrating by functional MRI and electrostimulation studies; Krainik et al., 2004;
Duffau, 2005; van Geemen et al., 2013). This is why it is possible
to surgically remove large brain regions, without inducing severe
cognitive disorders (Duffau, 2012). That is not to say, however,
that structures of the brain are equipotent. Functional plasticity in
diffuse low-grade glioma is limited and some structures are clearly
refractory to functional compensation. This is especially true for (i)
white matter fibres (including large portions of the arcuate, superior longitudinal, inferior occipito-frontal and posterior inferior longitudinal fasciculi) that cannot be re-sected; and (ii) certain cortical
epicentres with poor potential for functional compensation (Ius
et al., 2011).
One striking finding in the present study was that the degree of
frontoparietal disconnection was correlated with poor mentalizing
performance. This fully agrees with Ius and colleagues’ (2011)
proposal of critical subcortical brain systems and, more generally,
with the view that long-range connectivity is essential for cognition (Mesulam, 1998; Bressler and Menon, 2010; de Benedictis
and Duffau, 2011; Knosche and Tittgemeyer, 2011). The underlying pathophysiological mechanism of ‘progressive disconnection
syndrome‘ is probably related to disruption of cortical synchrony
throughout the neurocognitive network, which is safeguarded
under normal circumstances by the connectivity of subcortical
white matter. In this respect, some studies have shown that
damage to structural connectivity harms cognitive functions
(Shinoura et al., 2009; Karnath et al., 2011; Kummerer et al.,
2013), and induces major functional changes throughout the network (He et al., 2007a). These results have prompted some researchers to argue that structural alterations in white matter
connectivity constitute the main obstacle to functional compensation, especially in the case of diffuse low-grade glioma (Duffau,
2013). We consider that our findings have potential clinical value
and may enable prediction of the functional outcome of surgery
(based on the preoperative degree of axonal connectivity infiltration) and the adjustment of individual cognitive rehabilitation
programmes.
Conceivably, VLSM analysis of residual lesion infiltration maps
could be criticized if it does not confirm significant results generated with other methods. We do not share this scepticism. By
definition, an interpretation based on a disconnection syndrome
does not fit well with location-based models. In theory, a given
neuropsychological manifestation can be induced by injury of any
part of a fasciculus (Catani and Ffytche, 2005; Bartolomeo, 2012).
Brain 2014: 137; 944–959
| 955
This may be one reason why VLSM analyses have been inconclusive (especially regarding the low-level mentalizing task). It may
also be the case that impaired mentalizing accuracy is related to
incremental damage to associative pathways. This would fit with
the results of correlation analyses and also with the fact that our
significant VLSM associations for the high-level mentalizing accuracy scores (attribution of intention and physical causality) with the
white matter underlying the right temporoparietal region disappeared when residual lesion volumes was included as a covariate. Nevertheless, it is important to note that progressive damage
to associative pathways does not completely explain the behavioural results. Although our study failed to fully describe mentalizing subnetworks at the cortical level, our qualitative and
quantitative analyses demonstrate the importance of certain cortical epicentres (the pars opercularis of the right inferior frontal
gyrus for the low-level mentalizing task and the supplementary
motor area for the high-level mentalizing task).
At a broader (conceptual) level, our findings support the network view of high-level cerebral processes. Cognitive functions
should always be considered as the endpoint of sophisticated
functional interactions within whole-brain neurocognitive networks
and not as in classical neuropsychology (i.e. the byproduct of the
neural activity of discrete, highly specialized cortical areas)
(Bressler, 1995; Mesulam, 1998; McIntosh, 2000; Hickok and
Poepple, 2004; Fuster, 2006; Just and Varma, 2007; Bressler and
Menon, 2010; Meehan and Bressler, 2012). This opens up new
opportunities for understanding neuropsychological impairments in
non-focal disorders (i.e. neurodegenerative disease and neuropsychiatric disorders) as well as focal disorders (He et al., 2007b;
Honey et al., 2010). In the latter case, the heuristic value of the
networking hypothesis appears to be much higher, especially
when impairments following neurological disorders are sometimes
poorly explained by the lesion’s location. In this respect, researchers have started to use biomathematical approaches to
model the effect of a focal lesion on the topology of wholebrain networks (Honey and Sporns, 2008; Alstott et al., 2009).
It has been suggested that virtual lesions in certain cortical hubs
can disrupt functional connectivity between distant cortical areas;
this is in good agreement with studies of patients with focal
damage (He et al., 2007a; Gratton et al., 2012; Tuladhar et al.,
2013). Our findings fit well with a network paradigm of functional
disturbances, by demonstrating that the associative axonal connectivity underlying large-scale networks is essential for cognitive
functioning. This type of approach should prompt neuropsychological researchers to consider connectivity (rather than an oversimplistic, static, location-centred paradigm) when interpreting
cognitive impairments, especially in the field of social cognition
(Kennedy and Adolphs, 2012).
In view of the above framework, our results must be considered
in the broader context of neurodevelopmental disorders and especially autism spectrum disorders, for which impairments in social
cognition (including mentalizing and empathy) are quintessential
neuropsychological markers (Baron-Cohen et al., 1985, 2001;
Lombardo et al., 2007, 2010b). It is now well established that
autism spectrum disorders are accompanied by structural alterations through the brain (Cauda et al., 2010) and, in particular,
to long-range white matter pathways (Ke et al., 2009; Fletcher
956
| Brain 2014: 137; 944–959
et al., 2010; Cauda et al., 2013; Aoki et al., 2013). It was recently
reported that global white matter abnormalities were related to
clinical measures of social and communicative abilities in patients
with autism spectrum disorder (Gibbard et al., 2013). In view of
the interconnections between key brain areas in social perception
and cognition, structural disruption of the right arcuate/lateral superior longitudinal fasciculus complex has been posited as a possible substrate for the patients’ sociocognitive impairments (Cauda
et al., 2013; Kana et al., 2014). Our present findings show how
certain sociocognitive impairments can arise from structural alterations in specific associative pathways and shed light on the possible structural basis of related impairments in ASD populations.
According to a prominent (Dapretto et al., 2006; Rizzolatti and
Fabbri-Destro, 2010; Gallese et al., 2013) but controversial hypothesis (Hamilton, 2013), dysfunction of mirror mechanisms
may account for the emotional empathy, imitation and basic emotion recognition disorders encountered in autism spectrum disorders. On the basis of our results, one can legitimately
speculate that impaired low-level embodied representations in
autism spectrum disorders are related to abnormalities in the arcuate/lateral superior longitudinal fasciculus.
The current study has some potential limitations. We excluded
patients with left posterior temporal/parietal lesions to minimize
potential bias because of language impairments. Consequently, we
were unable to study the potential impact of disconnection of left
associative pathways in the occurrence of mentalizing impairments. Indeed, the low numbers of infiltrated patients prevented
us from drawing unambiguous conclusions in this respect. On the
methodological level, our use of a probabilistic diffusion tensor
imaging-based atlas to estimate the degree of disconnection of
each tract constitutes another possible limitation. Given that we
used the full range of spatial coordinates of each fasciculus, the
degree of the colinearity (Philippi et al., 2009) may have been
greater than that induced by natural crossing (e.g. the inferior
occipito-frontal fasciculus intersects with the arcuate fasciculus
deep within the inferior frontal gyrus).
In conclusion, the present study provided several new insights
into the neural bases of mentalizing. As was expected from the
literature data, mentalizing abilities may be functionally segregated
into two neurocognitive systems devoted to the identification and
attribution of mental states, respectively (Sabbagh et al., 2004;
Spunt and Lombardo, 2012). A right dorsal stream may process
the perceptual cues required for the rapid, pre-reflective identification of psychological states. This stream might connect the posterior part of the inferior frontal gyrus cortex and the inferior
parietal lobule through two parallel pathways (the arcuate fasciculus and the lateral superior longitudinal fasciculus). A right midline stream may process more self-reflective and inferential aspects
of mentalizing. The latter stream may connect the medial prefrontal cortex to the medial posterior parietal cortex through the
cingulum. Mentalizing abilities may thus arise from these two
physically distinct but functionally interdependent networks.
Certain brain areas belonging to each network do indeed show
effective connectivity during mentalizing (Lombardo et al., 2010a;
Spunt and Lieberman, 2012). Taken as a whole and when combined with literature data, our results provide evidence of the
previously hypothesized integrated, dynamic, dual-stream system
G. Herbet et al.
of mentalizing processes (Keysers and Gazzola, 2007; Uddin et al.,
2007; Lombardo et al., 2010a; Spunt and Lieberman, 2012). This
type of anatomic and functional model may offer new perspectives for understanding neuropathological/neurodevelopmental
conditions characterized by sociocognitive impairments and structural changes in associative pathways.
Funding
G.H received a fellowship from the Association pour la Recherche
sur le Cancer (grant number: DOC20120605069). The authors
declare that they have no conflicts of interest.
Supplementary material
Supplementary material is available at Brain online.
References
Adams RB Jr, Rule NO, Franklin RG Jr, Wang E, Stevenson MT,
Yoshikawa S, et al. Cross-cultural reading the mind in the eyes: an
fMRI investigation. J Cogn Neurosci 2010; 22: 97–108.
Alstott J, Breakspear M, Hagmann P, Cammoun L, Sporns O. Modeling
the impact of lesions in the human brain. PLoS Comput Biol 2009; 5:
e1000408.
Amodio DM, Frith CD. Meeting of minds: the medial frontal cortex and
social cognition. Nat Rev Neurosci 2006; 7: 268–77.
Andersen SM, Rapcsak SZ, Beeson PM. Cost function masking during
normalization of brains with focal lesions: still a necessity? Neuroimage
2010; 53: 78–84.
Aoki Y, Abe O, Nippashi Y, Yamasue H. Comparison of white matter
integrity between autism spectrum disorder subjects and typically developing individuals: a meta-analysis of diffusion tensor imaging tractography studies. Mol Autism 2013; 4: 2040–392.
Atique B, Erb M, Gharabaghi A, Grodd W, Anders S. Task-specific activity and connectivity within the mentalizing network during emotion
and intention mentalizing. Neuroimage 2011; 55: 1899–911.
Barbey AK, Colom R, Grafman J. Distributed neural system for emotional
intelligence revealed by lesion mapping. Soc Cogn Affect Neurosci
2012; 6: 6.
Baron-Cohen S, Klin A. What’s so special about Asperger Syndrome?
Brain Cogn 2006; 61: 1–4.
Baron-Cohen S, Leslie AM, Frith U. Does the autistic child have a “theory of mind"? Cognition 1985; 21: 37–46.
Baron-Cohen S, Ring HA, Wheelwright S, Bullmore ET, Brammer MJ,
Simmons A, et al. Social intelligence in the normal and autistic brain:
an fMRI study. Eur J Neurosci 1999; 11: 1891–8.
Baron-Cohen S, Wheelwright S, Hill J, Raste Y, Plumb I. The “Reading
the Mind in the Eyes” Test revised version: a study with normal adults,
and adults with Asperger syndrome or high-functioning autism. J Child
Psychol Psychiatry 2001; 42: 241–51.
Bartolomeo P. The elusive nature of white matter damage in anatomoclinical correlations. Front Hum Neurosci 2012; 6: 3.
Bartolomeo P, Thiebaut de Schotten M, Doricchi F. Left unilateral neglect
as a disconnection syndrome. Cereb Cortex 2007; 17: 2479–90.
Bates E, Wilson SM, Saygin AP, Dick F, Sereno MI, Knight RT, et al.
Voxel-based lesion-symptom mapping. Nat Neurosci 2003; 6: 448–50.
Bohl V, van den Bos W. Toward an integrative account of social
cognition: marrying theory of mind and interactionism to study the
interplay of Type 1 and Type 2 processes. Front Hum Neurosci
2012; 6: 11.
A dual-stream model of mentalizing
Bressler SL. Large-scale cortical networks and cognition. Brain Res Brain
Res Rev 1995; 20: 288–304.
Bressler SL, Menon V. Large-scale brain networks in cognition: emerging
methods and principles. Trends Cogn Sci 2010; 14: 277–90.
Brett M, Leff AP, Rorden C, Ashburner J. Spatial normalization of brain
images with focal lesions using cost function masking. Neuroimage
2001; 14: 486–500.
Brunet E, Sarfati Y, Hardy-Bayle MC, Decety J. A PET investigation of the
attribution of intentions with a nonverbal task. Neuroimage 2000; 11:
157–66.
Brunet E, Sarfati Y, Hardy-Bayle MC, Decety J. Abnormalities of brain
function during a nonverbal theory of mind task in schizophrenia.
Neuropsychologia 2003; 41: 1574–82.
Carrington SJ, Bailey AJ. Are there theory of mind regions in the brain? A
review of the neuroimaging literature. Hum Brain Mapp 2009; 30:
2313–35.
Castelli I, Baglio F, Blasi V, Alberoni M, Falini A, Liverta-Sempio O, et al.
Effects of aging on mindreading ability through the eyes: an fMRI
study. Neuropsychologia 2010; 48: 2586–94.
Catani M, Dell’acqua F, Bizzi A, Forkel SJ, Williams SC, Simmons A, et al.
Beyond cortical localization in clinico-anatomical correlation. Cortex
2012; 48: 1262–87.
Catani M, Ffytche DH. The rises and falls of disconnection syndromes.
Brain 2005; 128(Pt 10): 2224–39.
Catani M, Thiebaut de Schotten M. A diffusion tensor imaging tractography atlas for virtual in vivo dissections. Cortex 2008; 44: 1105–32.
Cauda F, Costa T, Palermo S, D’Agata F, Diano M, Bianco F, et al.
Concordance of white matter and gray matter abnormalities in
autism spectrum disorders: a voxel-based meta–analysis study. Hum
Brain Mapp 2013, in press.
Cauda F, Geminiani G, D’Agata F, Sacco K, Duca S, Bagshaw AP, et al.
Functional connectivity of the posteromedial cortex. PLoS One 2010;
5: e13107.
Cavanna AE, Trimble MR. The precuneus: a review of its functional
anatomy and behavioural correlates. Brain 2006; 129(Pt 3): 564–83.
Coricelli G. Two-levels of mental states attribution: from automaticity to
voluntariness. Neuropsychologia 2005; 43: 294–300.
Dapretto M, Davies MS, Pfeifer JH, Scott AA, Sigman M,
Bookheimer SY, et al. Understanding emotions in others: mirror
neuron dysfunction in children with autism spectrum disorders. Nat
Neurosci 2006; 9: 28–30.
De Benedictis A, Duffau H. Brain hodotopy: from esoteric concept to
practical surgical applications. Neurosurgery 2011; 68: 1709–23.
de Lange FP, Spronk M, Willems RM, Toni I, Bekkering H.
Complementary systems for understanding action intentions. Curr
Biol 2008; 18: 454–7.
Desmurget M, Bonnetblanc F, Duffau H. Contrasting acute and slowgrowing lesions: a new door to brain plasticity. Brain 2007; 130(Pt 4):
898–914.
Duffau H. The huge plastic potential of adult brain and the role of
connectomics: new insights provided by serial mappings in glioma surgery. Cortex 2013; S0010-9452(13)00207–4.
Duffau H. Lessons from brain mapping in surgery for low-grade glioma:
insights into associations between tumour and brain plasticity. Lancet
Neurol 2005; 4: 476–86.
Duffau H. Do brain tumours allow valid conclusions on the localisation of
human brain functions? Cortex 2011; 47: 1016–7.
Duffau H. The “frontal syndrome” revisited: lessons from electrostimulation mapping studies. Cortex 2012; 48: 120–31.
Duffau H, Capelle L. Preferential brain locations of low-grade gliomas.
Cancer 2004; 100: 2622–6.
Duffau H, Capelle L, Sichez N, Denvil D, Lopes M, Sichez JP, et al.
Intraoperative mapping of the subcortical language pathways using
direct stimulations. An anatomo-functional study. Brain 2002; 125
(Pt 1): 199–214.
Duval C, Piolino P, Bejanin A, Eustache F, Desgranges B. Age effects on
different components of theory of mind. Conscious Cogn 2011; 20:
627–42.
Brain 2014: 137; 944–959
| 957
Fletcher PT, Whitaker RT, Tao R, DuBray MB, Froehlich A,
Ravichandran C, et al. Microstructural connectivity of the arcuate fasciculus in adolescents with high-functioning autism. Neuroimage 2010;
51: 1117–25.
Frith CD, Frith U. The neural basis of mentalizing. Neuron 2006; 50:
531–4.
Fuster JM. The cognit: a network model of cortical representation. Int J
Psychophysiol 2006; 60: 125–32.
Gallese V, Rochat MJ, Berchio C. The mirror mechanism and its potential
role in autism spectrum disorder. Dev Med Child Neurol 2013; 55:
15–22.
Gibbard CR, Ren J, Seunarine KK, Clayden JD, Skuse DH, Clark CA.
White matter microstructure correlates with autism trait severity in a
combined clinical–control sample of high-functioning adults.
Neuroimage Clin 2013; 3: 106–14.
Gläscher J, Adolphs R, Damasio H, Bechara A, Rudrauf D, Calamia M,
et al. Lesion mapping of cognitive control and value-based decision
making in the prefrontal cortex. Proc Natl Acad Sci USA 2012; 109:
14681–6.
Gratton C, Nomura EM, Perez F, D’Esposito M. Focal brain lesions to
critical locations cause widespread disruption of the modular organization of the brain. J Cogn Neurosci 2012; 24: 1275–85.
Greicius MD, Krasnow B, Reiss AL, Menon V. Functional connectivity in
the resting brain: a network analysis of the default mode hypothesis.
Proc Natl Acad Sci USA 2003; 100: 253–8.
Greicius MD, Supekar K, Menon V, Dougherty RF. Resting-state functional connectivity reflects structural connectivity in the default mode
network. Cereb Cortex 2009; 19: 72–8.
Hamilton AF. Reflecting on the mirror neuron system in autism: a systematic review of current theories. Dev Cogn Neurosci 2013; 3:
91–105.
Happé F, Malhi GS, Checkley S. Acquired mind-blindness following
frontal lobe surgery? A single case study of impaired ‘theory of
mind’ in a patient treated with stereotactic anterior capsulotomy.
Neuropsychologia 2001; 39: 83–90.
He BJ, Shulman GL, Snyder AZ, Corbetta M. The role of impaired neuronal communication in neurological disorders. Curr Opin Neurol
2007b; 20: 655–60.
He BJ, Snyder AZ, Vincent JL, Epstein A, Shulman GL, Corbetta M.
Breakdown of functional connectivity in frontoparietal networks
underlies behavioral deficits in spatial neglect. Neuron 2007a; 53:
905–18.
Herbet G, Lafargue G, Bonnetblanc F, Moritz-Gasser S, Duffau H. Is the
right frontal cortex really crucial in the mentalizing network? A longitudinal study in patients with a slow-growing lesion. Cortex 2013; 49:
2711–27.
Hickok G, Poeppel D. Dorsal and ventral streams: a framework for
understanding aspects of the functional anatomy of language.
Cognition 2004; 92: 67–99.
Honey CJ, Sporns O. Dynamical consequences of lesions in cortical networks. Hum Brain Mapp 2008; 29: 802–9.
Honey CJ, Thivierge JP, Sporns O. Can structure predict function in the
human brain? Neuroimage 2010; 52: 766–76.
Iacoboni M, Dapretto M. The mirror neuron system and the consequences of its dysfunction. Nat Rev Neurosci 2006; 7: 942–51.
Iacoboni M, Molnar-Szakacs I, Gallese V, Buccino G, Mazziotta JC,
Rizzolatti G. Grasping the intentions of others with one’s own mirror
neuron system. PLoS Biol 2005; 3: e79.
Ius T, Angelini E, Thiebaut de Schotten M, Mandonnet E, Duffau H.
Evidence for potentials and limitations of brain plasticity using an
atlas of functional resectability of WHO grade II gliomas: towards a
“minimal common brain". Neuroimage 2011; 56: 992–1000.
Just MA, Varma S. The organization of thinking: what functional brain
imaging reveals about the neuroarchitecture of complex cognition.
Cogn Affect Behav Neurosci 2007; 7: 153–91.
Kana RK, Libero LE, Hu CP, Deshpande HD, Colburn JS. Functional Brain
Networks and White Matter Underlying Theory-of-Mind in Autism.
Soc Cogn Affect Neurosci 2014; 9: 98–105.
958
| Brain 2014: 137; 944–959
Karnath HO, Himmelbach M, Rorden C. The subcortical anatomy of
human spatial neglect: putamen, caudate nucleus and pulvinar. Brain
2002; 125(Pt 2): 350–60.
Karnath HO, Rennig J, Johannsen L, Rorden C. The anatomy underlying
acute versus chronic spatial neglect: a longitudinal study. Brain 2011;
134(Pt 3): 903–12.
Ke X, Tang T, Hong S, Hang Y, Zou B, Li H, et al. White matter impairments in autism, evidence from voxel-based morphometry and diffusion tensor imaging. Brain Res 2009; 1265: 171–7.
Kennedy DP, Adolphs R. The social brain in psychiatric and neurological
disorders. Trends Cogn Sci 2012; 16: 559–72.
Keysers C, Gazzola V. Integrating simulation and theory of mind: from
self to social cognition. Trends Cogn Sci 2007; 11: 194–6.
Kimberg DY, Coslett HB, Schwartz MF. Power in Voxel-based lesionsymptom mapping. J Cogn Neurosci 2007; 19: 1067–80.
Knosche TR, Tittgemeyer M. The role of long-range connectivity for the
characterization of the functional-anatomical organization of the
cortex. Front Syst Neurosci 2011; 5: 00058.
Krainik A, Duffau H, Capelle L, Cornu P, Boch AL, Mangin JF, et al. Role
of the healthy hemisphere in recovery after resection of the supplementary motor area. Neurology 2004; 62: 1323–32.
Kummerer D, Hartwigsen G, Kellmeyer P, Glauche V, Mader I, Kloppel S,
et al. Damage to ventral and dorsal language pathways in acute aphasia. Brain 2013; 136(Pt 2): 619–29.
Lieberman MD. Social cognitive neuroscience: a review of core processes. Annu Rev Psychol 2007; 58: 259–89.
Lombardo MV, Barnes JL, Wheelwright SJ, Baron-Cohen S. Self-referential cognition and empathy in autism. PLoS One 2007; 2: e883.
Lombardo MV, Chakrabarti B, Bullmore ET, Wheelwright SJ, Sadek SA,
Suckling J, et al. Shared neural circuits for mentalizing about the self
and others. J Cogn Neurosci 2010a; 22: 1623–35.
Lombardo MV, Chakrabarti B, Bullmore ET, Sadek SA, Pasco G,
Wheelwright SJ, et al. Atypical neural self-representation in autism.
Brain 2010b; 133: 611–24.
Mackinnon A, Mulligan R. The estimation of premorbid intelligence levels
in French speakers [in French]. Encephale 2005; 31 (1 Pt 1): 31–43.
Makris N, Kennedy DN, McInerney S, Sorensen AG, Wang R,
Caviness VS Jr. et al. Segmentation of subcomponents within the superior longitudinal fascicle in humans: a quantitative, in vivo, DT-MRI
study. Cereb Cortex 2005; 15: 854–69.
Mandonnet E, Delattre JY, Tanguy ML, Swanson KR, Carpentier AF,
Duffau H, et al. Continuous growth of mean tumor diameter in a
subset of grade II gliomas. Ann Neurol 2003; 53: 524–8.
Mar RA. The neural bases of social cognition and story comprehension.
Annu Rev Psychol 2011; 62: 103–34.
Mars RB, Neubert FX, Noonan MP, Sallet J, Toni I, Rushworth MF.
On the relationship between the “default mode network” and the
“social brain”. Front Hum Neurosci 2012; 6 (189): 00189.
Martino J, De Witt Hamer PC, Berger MS, Lawton MT, Arnold CM, de
Lucas EM, et al. Analysis of the subcomponents and cortical terminations of the perisylvian superior longitudinal fasciculus: a fibre dissection and DTI tractography study. Brain Struct Funct 2013; 218:
105–21.
McIntosh AR. Towards a network theory of cognition. Neural Netw
2000; 13: 861–70.
Meehan TP, Bressler SL. Neurocognitive networks: findings, models, and
theory. Neurosci Biobehav Rev 2012; 36: 2232–47.
Mesulam MM. From sensation to cognition. Brain 1998; 121(Pt 6):
1013–52.
Moor BG, Macks ZA, Güroğlu B, Rombouts SA, Molen MW, Crone EA.
Neurodevelopmental changes of reading the mind in the eyes. Soc
Cogn Affect Neurosci 2012; 7: 44–52.
Mukamel R, Ekstrom AD, Kaplan J, Iacoboni M, Fried I. Single-neuron
responses in humans during execution and observation of actions. Curr
Biol 2010; 20: 750–6.
Ojemann G, Mateer C. Human language cortex: localization of memory,
syntax, and sequential motor-phoneme identification systems. Science
1979; 205: 1401–3.
G. Herbet et al.
Parisot S, Duffau H, Chemouny S, Paragios N. Joint tumor segmentation
and dense deformable registration of brain MR images. Med Image
Comput Comput Assist Interv 2012; 15(Pt 2): 651–8.
Parkinson C, Wheatley T. Relating anatomical and social connectivity:
white matter microstructure predicts emotional empathy. Cereb
Cortex 2012, in press.
Philippi CL, Mehta S, Grabowski T, Adolphs R, Rudrauf D. Damage to
association fibre tracts impairs recognition of the facial expression of
emotion. J Neurosci 2009; 29: 15089–99.
Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA,
Shulman GL. A default mode of brain function. Proc Natl Acad Sci
USA 2001; 98: 676–82.
Rizzolatti G, Craighero L. The mirror-neuron system. Annu Rev Neurosci
2004; 27: 169–92.
Rizzolatti G, Fabbri-Destro M. Mirror neurons: from discovery to autism.
Exp Brain Res 2010; 200: 223–37.
Rizzolatti G, Sinigaglia C. The functional role of the parieto-frontal mirror
circuit: interpretations and misinterpretations. Nat Rev Neurosci 2010;
11: 264–74.
Rorden C, Karnath HO. Using human brain lesions to infer function: a
relic from a past era in the fMRI age? Nat Rev Neurosci 2004; 5:
813–9.
Rudrauf D, Mehta S, Grabowski TJ. Disconnection’s renaissance takes
shape: formal incorporation in group-level lesion studies. Cortex
2008a; 44: 1084–96.
Rudrauf D, Mehta S, Bruss J, Tranel D, Damasio H, Grabowski TJ.
Thresholding lesion overlap difference maps: application to categoryrelated naming and recognition deficits. Neuroimage 2008b; 41:
970–84.
Russell TA, Rubia K, Bullmore ET, Soni W, Suckling J, Brammer MJ, et al.
Exploring the social brain in schizophrenia: left prefrontal underactivation during mental state attribution. Am J Psychiatry 2000; 157:
2040–2.
Sabbagh MA. Understanding orbitofrontal contributions to theory-ofmind reasoning: implications for autism. Brain Cogn 2004; 55: 209–19.
Sarfati Y, Hardy-Bayle MC, Besche C, Widlocher D. Attribution of intentions to others in people with schizophrenia: a non-verbal exploration
with comic strips. Schizophr Res 1997; 25: 199–209.
Saxe R, Kanwisher N. People thinking about thinking people. The role of
the temporo-parietal junction in “theory of mind". Neuroimage 2003;
19: 1835–42.
Schilbach L, Bzdok D, Timmermans B, Fox PT, Laird AR, Vogeley K, et al.
Introspective minds: using ALE meta–analyses to study commonalities
in the neural correlates of emotional processing, social & unconstrained cognition. PLoS One 2012; 7: e30920.
Schilbach L, Eickhoff SB, Rotarska-Jagiela A, Fink GR, Vogeley K. Minds
at rest? Social cognition as the default mode of cognizing and its
putative relationship to the “default system” of the brain. Conscious
Cogn 2008; 17: 457–67.
Schippers MB, Roebroeck A, Renken R, Nanetti L, Keysers C. Mapping
the information flow from one brain to another during gestural communication. Proc Natl Acad Sci USA 2010; 107: 9388–93.
Schwartz MF, Kimberg DY, Walker GM, Brecher A, Faseyitan OK,
Dell GS, et al. Neuroanatomical dissociation for taxonomic and thematic knowledge in the human brain. Proc Natl Acad Sci USA 2011;
108: 8520–4.
Shamay-Tsoory SG, Aharon-Peretz J, Perry D. Two systems for empathy:
a double dissociation between emotional and cognitive empathy in
inferior frontal gyrus versus ventromedial prefrontal lesions. Brain
2009; 132(Pt 3): 617–27.
Shamay-Tsoory SG, Tomer R, Berger BD, Goldsher D, Aharon-Peretz J.
Impaired “affective theory of mind” is associated with right ventromedial prefrontal damage. Cogn Behav Neurol 2005; 18: 55–67.
Shinoura N, Suzuki Y, Yamada R, Tabei Y, Saito K, Yagi K. Damage to
the right superior longitudinal fasciculus in the inferior parietal lobe
plays a role in spatial neglect. Neuropsychologia 2009; 47: 2600–3.
Spreng RN, Mar RA, Kim AS. The common neural basis of autobiographical memory, prospection, navigation, theory of mind, and the default
A dual-stream model of mentalizing
mode: a quantitative meta-analysis. J Cogn Neurosci 2009; 21:
489–510.
Spunt RP, Lieberman MD. An integrative model of the neural systems
supporting the comprehension of observed emotional behavior.
Neuroimage 2012; 59: 3050–9.
Spunt RP, Lieberman MD. The Busy Social Brain: evidence for automaticity and control in the neural systems supporting social cognition and
action understanding. Psychol Sci 2013; 24: 80–6.
Thiebaut de Schotten M, Kinkingnehun S, Delmaire C, Lehericy S,
Duffau H, Thivard L, et al. Visualization of disconnection syndromes
in humans. Cortex 2008; 44: 1097–103.
Tuladhar AM, Snaphaan L, Shumskaya E, Rijpkema M, Fernandez G,
Norris DG, et al. Default Mode Network Connectivity in Stroke
Patients. PLoS One 2013; 8: e66556.
Uddin LQ, Iacoboni M, Lange C, Keenan JP. The self and social cognition: the role of cortical midline structures and mirror neurons. Trends
Cogn Sci 2007; 11: 153–7.
van den Heuvel M, Mandl R, Luigjes J, Hulshoff Pol H. Microstructural
organization of the cingulum tract and the level of default mode functional connectivity. J Neurosci 2008; 28: 10844–51.
van Geemen K, Herbet G, Moritz-Gasser S, Duffau H. Limited plastic
potential of the left ventral premotor cortex in speech articulation:
Brain 2014: 137; 944–959
| 959
evidence From intraoperative awake mapping in glioma patients.
Hum Brain Mapp 2013, in press.
Van Overwalle F. Social cognition and the brain: a meta-analysis. Hum
Brain Mapp 2009; 30: 829–58.
Van Overwalle F, Baetens K. Understanding others’ actions and goals by
mirror and mentalizing systems: a meta-analysis. Neuroimage 2009;
48: 564–84.
Vistoli D, Brunet-Gouet E, Baup-Bobin E, Hardy-Bayle MC, Passerieux C.
Anatomical and Temporal architecture of theory of mind: a MEG insight the early stages. Neuroimage 2011; 54: 1406–14.
Vollm BA, Taylor AN, Richardson P, Corcoran R, Stirling J, McKie S, et al.
Neuronal correlates of theory of mind and empathy: a functional magnetic resonance imaging study in a nonverbal task. Neuroimage 2006;
29: 90–8.
Winner E, Brownell H, Happe F, Blum A, Pincus D. Distinguishing
lies from jokes: theory of mind deficits and discourse interpretation
in right hemisphere brain-damaged patients. Brain Lang 1998; 62:
89–106.
Zaki J, Weber J, Bolger N, Ochsner K. The neural bases of empathic
accuracy. Proc Natl Acad Sci USA 2009; 106: 11382–7.