Abstract

The feeling of being excluded from a social interaction triggers social pain, a sensation as intense as actual physical pain. Little is known about the neurophysiological underpinnings of social pain. We addressed this issue using intracranial electroencephalography in 15 patients performing a ball game where inclusion and exclusion blocks were alternated. Time–frequency analyses showed an increase in power of theta-band oscillations during exclusion in the anterior insula (AI) and posterior insula, the subgenual anterior cingulate cortex (sACC), and the fusiform “face area” (FFA). Interestingly, the AI showed an initial fast response to exclusion but the signal rapidly faded out. Activity in the sACC gradually increased and remained significant thereafter. This suggests that the AI may signal social pain by detecting emotional distress caused by the exclusion, whereas the sACC may be linked to the learning aspects of social pain. Theta activity in the FFA was time-locked to the observation of a player poised to exclude the participant, suggesting that the FFA encodes the social value of faces. Taken together, our findings suggest that theta activity represents the neural signature of social pain. The time course of this signal varies across regions important for processing emotional features linked to social information.

Introduction

Social pain is commonly referred to the distressing experience arising from the loss of social connections. As such, the experience of being excluded or devalued by desired relational partners or groups is considered an example of social pain (MacDonald and Leary 2005). This type of pain may be associated with a perception of rejection, social exclusion and all the physical and verbal cues that make an individual feel disconnected from others (Williams et al. 2005). Social pain has major consequences on the survival and well being of the individual. Its effects can be long lasting (Chen et al. 2008) and associated with physical diseases such as cardiovascular pathologies or cognitive decline including Alzheimer (Gow et al. 2007; Cacioppo and Hawkley 2009). The feeling of social inclusion seems to be so deeply rooted in our brain that the mere fact of imagining that we are excluded by others is often sufficient to trigger social pain (Eisenberger 2006).

Social pain has become a growing subject of interest in neuroscience. Several studies have been conducted in which feelings of social exclusion were experimentally provoked. The Cyberball paradigm is a good example of this approach. It is a tossing game, played on a computer, where the participant has to pass the ball to other subjects in an interactive way (Williams and Jarvis 2006). The partners are fictitious and their behavior (inclusion or exclusion of the participant) is controlled by the computer. However, the participant is informed that they are real and located in another room.

Using this simple paradigm, it has been possible to identify some features of social pain in humans, including an increased level of stress during exclusion periods and a decrease in self-esteem (Williams et al. 2000). Strikingly, these effects are so pervasive that participants often prefer to be included in the game, even if this means enduring a monetary loss (van Beest and Williams 2006). From a neurophysiological point of view, the brain mechanisms involved in social pain processing remain unknown. Neuroimaging studies have reported that social pain involves some regions pertaining to the so-called physical pain matrix (Eisenberger et al. 2003, 2007). This matrix is a complex neural circuit activated when people experience the unpleasant sensation caused by a nociceptive stimulus. In this circuit, 2 distinguishable pathways are present: One sensory and one cognitive-affective (Peyron et al. 2000). The sensory pathway codes information concerning the location, the duration and the intensity of the painful input. It involves sensory structures such as the thalamus, the sensorimotor cortices (S1 and S2) and the insula. The cognitive-affective pathway processes information about the unpleasantness value of the nociceptive stimulus and is coded in more anterior structures such as the anterior cingulate cortex (ACC), and the dorsolateral prefrontal cortex (dlPFC) and ventrolateral prefrontal cortex (vlPFC). As shown by the first functional magentic resonance imaging study on social pain, vlPFC, ACC, and anterior insula (AI) are activated during exclusion periods and ACC activity correlates with stress perception (Eisenberger et al. 2003). Based on this observation and the fact that ACC response in the dorsal region increases with the subjective feeling of unpleasantness, it has been suggested that this latter structure is important for regulating social pain perception. Another demonstration that the pain matrix mediates both physical and social pain is provided by the well-known “empathy for pain” studies. As shown in these studies, the pain experienced when seeing a partner suffering activates the affective-cognitive pathway of the pain matrix, and, in particular, the ACC and AI (Singer et al. 2004). Recently, a new study has shown that social pain also recruits some somatosensory regions involved in the expression of physical pain. When people experience rejection in their life, social pain activates affective and sensory pathways, including the somatosensory cortex (S1) and the posterior insula (PI) (Kross et al. 2011).

Taken together, all of these results suggest that physical and social pain share, at least partially, a common neural substrate (Panksepp 2003). However, the functional role that some nodes of the well-known physical pain matrix may play during social pain remains unclear. To address this issue, we used direct intracranial recording of brain activity in 15 epileptic patients with implanted electrodes performing a revised version of the Cyberball game (Eisenberger et al. 2003). In our task, 3 inclusion periods alternated with 3 exclusion periods (see Fig. 1A). We predicted that exclusion periods would trigger low cerebral waves, that is, waves located in the theta band frequency. Theta signals, compared with more cognitive cerebral waves such as gamma or beta (Buzsaki 2002), are known to be associated with emotional feelings and primitive sensations (Knyazev 2007), physical pain (Liu et al. 2010; Schulz et al. 2011), and empathy for pain (Jeon et al. 2010). The observation that theta oscillations play an important role in human emotional processing has been widely documented by electroencephalography studies during the last decade. A study by Krause et al. (2000) demonstrated, for instance, that theta waves are associated with prolonged visual emotional stimulation. In the same vein, a series of studies by Aftanas et al. (2001, 2003, 2004) showed that affective valence discrimination was associated with early time-locked synchronized theta activity . At a more general level, these slow frequency oscillations have also been connected to general arousal processes (Bekkedal et al. 2011), aversive noise (Crowley et al. 2009), and fear of physical shocks (Baas et al. 2002). Here, we were interested in investigating the time course of theta signal within the social pain network. Specifically, we expected differences in theta signal discharge when patients were initially experiencing (first blocks) social distress caused by exclusion compared with when they were expecting to experience (as an effect of learning) such condition (last blocks).

Figure 1.

Ball-tossing game time course (A). The game lasts 10 min and involves 186 trials. Each trial is defined by a throw of the ball. In order to maintain a higher degree of subjects' attention and involvement in the game, inclusion and exclusion blocks were alternated. First, subjects experience 2 blocks of inclusion (60 trials), followed by an exclusion period (30 trials), an inclusion period (30 trials) and 2 blocks of exclusion (30 trials for each) separated by a mini block of inclusion (6 trials). Subjects are told that they are going to play an interactive game with 2 players, placed in other rooms of the hospital. These 2 players are, fictitious. The subject is allowed as much time as he/she wants to choose to whom to send the ball. Before starting the game, the subject is informed that the experimenter is connecting him/her to the server. During that time, the subject is instructed to observe the other (fictitious) players practicing the game for 30 trials. We used this “technical exclusion” period as the baseline. Trial time course (B). Each trial lasts 2 s and consists of an uncertainty period (0–0.7 s) where the subject is waiting for the throw and a certainty period where the subject knows if he/she will receive the ball (0.7–1.6 s the ball is moving and 1.6–2 s the ball is received, static image). During the course of the trial, the subject sees the Cyberball interface (i.e. 2 cartoons players on the top of the screen and his/her hand on the bottom). The picture and the name of the fictitious players were placed near the cartoon images.

Figure 1.

Ball-tossing game time course (A). The game lasts 10 min and involves 186 trials. Each trial is defined by a throw of the ball. In order to maintain a higher degree of subjects' attention and involvement in the game, inclusion and exclusion blocks were alternated. First, subjects experience 2 blocks of inclusion (60 trials), followed by an exclusion period (30 trials), an inclusion period (30 trials) and 2 blocks of exclusion (30 trials for each) separated by a mini block of inclusion (6 trials). Subjects are told that they are going to play an interactive game with 2 players, placed in other rooms of the hospital. These 2 players are, fictitious. The subject is allowed as much time as he/she wants to choose to whom to send the ball. Before starting the game, the subject is informed that the experimenter is connecting him/her to the server. During that time, the subject is instructed to observe the other (fictitious) players practicing the game for 30 trials. We used this “technical exclusion” period as the baseline. Trial time course (B). Each trial lasts 2 s and consists of an uncertainty period (0–0.7 s) where the subject is waiting for the throw and a certainty period where the subject knows if he/she will receive the ball (0.7–1.6 s the ball is moving and 1.6–2 s the ball is received, static image). During the course of the trial, the subject sees the Cyberball interface (i.e. 2 cartoons players on the top of the screen and his/her hand on the bottom). The picture and the name of the fictitious players were placed near the cartoon images.

Materials and Methods

Subjects

Fifteen patients (males, N = 10, females, N = 5; mean age 34.8 years, range 19–51) suffering from drug-refractory epilepsy were stereotactically implanted with depth electrodes as part of a presurgical evaluation. All patients were right-handed. Neuropsychological assessment revealed normal general cognitive functioning. Subjects presented neither a depressed mood nor anxiety as measured by the profile of mood states questionnaire (McNair et al. 1971). The research protocol was approved by the local ethical committee (CPP, Lyon Sud-Est IV). Before the study, participants were given written informed consent, in which we explained that the recording had no implication in their epileptic treatment.

Experimental Paradigm

We used a modified version of the Cyberball task (Eisenberger et al. 2003) to explore the cerebral effects generated by exclusion from different partners during a ball game on the computer. The paradigm time course consisted of an alternated series of inclusion and exclusion blocks (Fig. 1A).

The patients were told that the goal of the study was to investigate neural activity during social interaction and that they would be playing a virtual ball-tossing game with 2 other players, each one placed in different rooms. The fictitious players were the same for all the participants. To avoid gender effects, men played with 2 other men and women played with 2 other women. Before playing the game, patients were shown webcam-like videos of the other 2 players waving to indicate that they were connected and ready to start the game. After the experiment, we controlled for sympathy confounds through a questionnaire that asked the patients whether they preferred one of the 2 other players. None of the patients provided a positive answer to this question. All the patients also felt that the other players and the game were real. The experimental paradigm was implemented with the software Presentation (version 11, Neurobehavioral Systems). To perform the game, subjects had to press one of 2 response mouse buttons: The left one to send the ball to the partner on the left side of the screen and the right one to send to the partner on the right side. Each recording began with a static picture of the 2 virtual players in the upper corners of the screen and an arm and a buddy icon, representing the patient in the bottom central part of the screen (Fig. 1B). Each trial included an uncertainty period (0–0.7 s), consisting of a static image, in which the player holds the ball, and a certainty period (0.7–2 s), in which the ball is sent to the chosen player. The certainty period consisted of 2 stages: Ball moving (toward the recipient) (0.7–1.6 s) and static image (when the recipient holds the ball in his/her hand) (1.6–2 s). Fictive names and the photographs of the partners were displayed below each of the 2 virtual players' animated cartoon representations. We used the animation first employed by Williams et al. in 2000, to represent the 2 virtual players, the subject and the ball. The ball-tossing game program was set for 186 throws per recording, with the computer players waiting 0.2–0.5 s before making a throw to heighten the sense that the participant was actually playing with real individuals. “The patient had unlimited time to choose to whom he/she would send the ball.” The game's time course consisted of 3 alternating inclusion and exclusion blocks of 30 trials each. More precisely, the game started with 2 blocks of inclusion (60 trials), followed by one of exclusion (30 trials), one of inclusion (30 trials), and 2 blocks of exclusion at the end (60 trials) (separated by 6 trials of inclusion to catch patient's attention). On the whole, the game lasted 10 min. During each inclusion condition the patients received the same amount of throws from the player on the left side and from the player on the right side of the screen. During the exclusion blocks the patients did not receive any throw. Before starting the game, the patient was informed that the experimenter was connecting him/her to the server. During that time, the patient was told that due to a technical problem he/she could not play and he/she was instructed to observe the other (fictitious) players practicing the game for 30 trials, until the internet connection was established. We used this “technical exclusion” period as the baseline.

Intracranial Recordings

Patients were implanted intracerebrally with depth electrodes, each bearing 5–15 recording sites (Ad-Tech Medical Instruments). The electrodes were implanted perpendicularly to the midsagittal plane using Talairach's stereotactic method (Talairach and Bancaud 1974). Recording sites were 2 mm long, 1 mm diameter cylinders, separated by a distance of 1.5 mm (DIXI Medical, Besancon, France). The structures to be explored were defined on the basis of ictal manifestations, electroencephalography (EEG), video-scalp EEG, angiography, and structural magnetic resonance imaging (MRI). The experiment started at least 8 days after the electrodes implantation and at least after 24 h after the last seizure. At that time, anticonvulsive drug treatment had been reduced for at least 1 week to record spontaneous epileptic seizures during continuous video-scalp EEG recordings performed in equipped rooms. Continuous-depth EEGs were recorded on a 128-channel device (Micromed, Treviso, Italy), white matter/screw (intracranial) referenced, amplified, filtered (0.1–200 Hz bandwidth), sampled at 512 Hz, and stored together with digital markers of specific events of the task for subsequent off-line analysis. We defined different codes to identify inclusion (I) and exclusion (E) trials, and in each of these general conditions if the participant is throwing the ball, if he/she is receiving the ball, or if the ball is thrown from one fictitious participant to the other.

Electrode Location

To maximally prevent the measurement of the same electrical signals by multiple sites, bipolar montages were calculated by subtracting the signals recorded from adjacent sites belonging to the same-depth electrode. The bipolar montage procedure is the most common method in intracranial recording (Lachaux et al. 2003). The main advantage of this technique is a more focal signal not perturbed by external artifacts and a good spatial resolution, ∼3 mm (Ossandon et al. 2012). A detailed review of bipolar recording techniques in intracranial electroencephalography has been provided in previous publications (Lachaux et al. 2003; Tallon-Baudry et al. 2005; Jerbi et al. 2009). The bipolar montage calculation resulted in a total of 827 bipolar recordings (here called for simplicity “electrodes”) across 15 patients (see Supplementary Fig. 2). For each bipolar montage, the Cartesian coordinates (x, y, z) were calculated as the medium location of the 2 adjacent recording sites, anatomically distributed as follows: Temporal lobe 47.16% (n = 390), frontal lobe 32.41% (n = 268), parietal lobe 15.96% (n = 132), and occipital lobe 4.47% (n = 37) of all the electrodes. On the whole, 45% (n = 371) of electrodes were recorded within the left hemisphere and 55% (n = 456) in the right hemisphere.

Images Reconstruction

First, electrode locations were measured from X-ray images obtained on a stereotaxic plan. For each recording site, the Cartesian coordinates (x, y, z) were calculated after normalization of the anatomical 3D spoiled gradient recalled anatomical cerebral MRI into Talairach space using SPM5 (Wellcome Department of Cognitive Neurology, London, UK; http://www.fil.ion.ucl.ac.uk/spm/).

Reconstruction of electrodes in patients' brain was obtained using the open software Anatomist/Brain Visa 3.1 software (http://www.brainvisa.info). First, we converted the dicom MR images in NIfti format, through MRI Converter program.

Reoriented and preprocessed images were then segmented via the automated segmentation pipeline of Brain Visa, to obtain meshes of the 2 hemispheres and of the head. Afterward, it was possible to import these images to Anatomist and to plot the contact electrodes in the brain reconstruction. The obtained position of electrodes was verified with the subject post implantation anatomical MRI.

Time–Frequency Analysis

Pre-processing

The EEG signal from patients' electrodes was first epoched by trials of 2000 ms (duration of a movie) for both periods of exclusion, inclusion, and the baseline (technical exclusion). Trials including motor actions (mouse clicking) were discarded from the analysis. Epoched data were detrended and then filtered with Butterworth filters: A low-pass filter with a cut-off frequency of 100 Hz and a band-stop centerd on 50 Hz. The last step of pre-processing consisted of applying automatic artifact detection. Any trial showing an absolute voltage above a 300 mV threshold was automatically rejected. This method also discarded any channel presenting more than 5% of rejected trials. Data were then visually inspected. Trials (10.3%; mean across subjects) were removed and a set of 827 recording sites across subjects was retained (5.1% of bad channels were removed). The percentage of rejected trials per condition was as follows: 9.96% for the baseline; 10.33% for the inclusion, and 10.62% for the exclusion condition. The number of rejected trials was low and very similar in each condition and it cannot explain the differences we observed between inclusion and exclusion sessions.

Wavelets Transform

Power spectra S(t, f) was obtained by convoluting the signal with complex Gaussian Morlet's wavelet (Tallon-Baudry et al. 1996) 

formula
for time t, recording site i, frequency f0, s being the preprocessed signal, and w (t, f0), the complex Morlet defined as  
formula
with forumlaforumla and the constant ratio forumla providing a good compromise between the frequency and time resolution of wavelets. Frequency f0 ranged from 3 to 100 Hz, with a step of 0.5 Hz between 3 and 10 Hz, and 1 Hz between 10 and 100 Hz. We then computed the normalized power spectra Snorm(t, f) in a z-score way so that  
formula
using the power spectra of the baseline Bl(t, f), Bi(t, f), and σ being, respectively, its mean and standard deviation across trials.

Statistics

Primary tests were done using a Mann–Whitney U test on normalized power spectra S(t, f). We performed the same test for the entire frequency-band (up to 100 Hz). Multi-tests were done by dividing the time–frequency (TF) power maps in 250 TF windows (4), whose resolution depend on the frequency band (FB): from 3 to 10 Hz, 0.5 Hz × 400 ms, and from 10 to 100 Hz, 10 Hz × 100 ms. Obtained P values were finally corrected using Bonferroni (250 TF windows × number of bipolar montages for the corresponding patient).

In order to observe the variations of TF power during all the experiments, we averaged the normalized power spectra Snorm(t, f) across the trial duration (2 s) and FBs, so that forumla for frequency band F, with the frequency band groups FB = {[3 7] [9 11], [15 20],[20 40],[40 100]}, corresponding to theta, alpha, beta, and gamma rhythms. Owing to the non-Gaussianity of the distributions of the computed forumla (high kurtosis values superior to 10), we performed a nonparametric permutation test (number of permutations = 10 000) between inclusion and exclusion trials. Corresponding P values were corrected using Bonferroni (5 FBs × number of bipolar montages). We subsequently plotted the z-score of the signal (obtained from the baseline “technical exclusion” period) during the course of the game. We compared the 3 exclusion blocks 2-by-2 with themselves, and also with all the inclusion period, using permutation tests (number of permutations = 10 000) corrected by Bonferroni (total number of tests). The difference between inclusion and exclusion trials from different sites was tested independently for each site.

A final statistical analysis was performed to estimate the average activation in the group of patients that showed significantly theta power difference during exclusion. This analysis was made after averaging for all individual trials the signal found in areas shared by a number of patients. These common areas refer to areas with the same electrodes implantation. Each brain region was defined by regrouping the electrodes belonging to the same neurological structure. The localization of each electrode in a specific region has been checked using the post-implantation structural MRI. The Talairach coordinates in mm for each electrode are indicated in Supplementary Table 1. For each region, we computed the percentage of patients presenting electrodes showing theta effect, across and within conditions. For each region identified with this approach, we performed a statistical analysis across and between blocks and across subjects that responded significantly. We used a nonparametric Bonferroni-corrected permutation test (10 000 permutations), to compare the 3 blocks of exclusion (2 by 2), and also each of the 3 blocks of exclusion with the corresponding inclusion blocks.

Phase Locking Value

Phase synchrony between pairs of electrodes is a complex phenomenon that can be unrelated to the single power of the electrodes. This phenomenon can give us important information about functional connections between remote cerebral regions (Varela et al. 2001). Synchrony between regions can reflect the existence of either common processing or information transfer. To evaluate if social pain involves synchronization between different areas of the brain, we measured phase synchrony for all pairs of electrodes, in all subjects.

The first step was to extract the instantaneous phase of signal φi(t, f, n) for each electrode i, time t, frequency f, and trial n, from the previous convolution with Morlet wavelet forumla

Then, for each pair of electrode p, the phase locking value (PLV) (Lachaux et al. 1999) was computed as  

formula
with forumla being the difference of phase between the 2 recording sites i1 and i2 of pair p: 
formula
The difference of PLV between exclusion and inclusion period ΔPLV(t, f) was simply defined as forumla with forumla and forumla corresponding to the set of exclusion and inclusion trials, respectively.

Statistics

In order to evaluate the statistical significance of ΔPLV values, a randomization test was performed. Two hundred surrogate values of ΔPLV were computed on shuffled data (4), that is, phase differences were calculated between shuffled trials, randomly permuted for one of the 2 signals without replacement. ΔPLV was thus considered significant if its value was higher (or lower) than a pre-defined proportion P of surrogate values.

Owing to the high number of electrode pairs (21 054), ΔPLV(t, f) was averaged for the 5 frequency bands group FB previously defined, and for the 2 periods of time T = {[0 0.7],[0.7 2]}, to reduce the computational complexity. The significance level of 0.05 (proportion P = 95%) was corrected by the number of frequencies and time windows, and thus set as 0.005 (P = 99.5%). ΔPLV significant values are finally plotted in a glass brain template (from SPM5) (Tallon-Baudry et al. 1996; Lachaux et al. 1999).

All these signal and statistical analyses were evaluated with a toolbox (developed in our laboratory by S.H.) working in Matlab (Matlab 7.5, MathWorks) space and using some Fieldtrip functions (2008–2009, Donders Institute for Brain, Cognition and Behaviour, The Netherlands, DCCN, DCC, DCN).

Results

Theta Waves as The Neural Code for Social Pain

Across all subjects, we analyzed 827 channels with a bipolar montage. The analysis of all TFBs (from 3 to 100 Hz) showed a main and consistent effect only in low-FBs (see Supplementary Fig. 2 for a global view of all electrodes analyzed across subjects). As predicted, we found significantly higher theta wave activity (3–7 Hz) during exclusion trials, compared with inclusion ones. The theta activity was present throughout the trial and not time-locked to any stimulus (i.e. ball thrown) suggesting that this signal may be induced by the general feeling of exclusion.

The significant theta effect was present in some major regions of the social–physical pain matrix, including, the AI (42.85%), the PI (44.44%), the subgenual anterior cingulate cortex (sACC) (BA32, only one patient had electrodes in this region), the orbitofrontal cortex (BA47, 42.86%, and BA 10, 50%), the supplementary motor area (BA6, 66.66%), the dlPFC (BA9, 50%), the fusiform face area (FFA) (BA37, 77.77%). Even thought the theta activity was similar in all of these regions, detailed analyses revealed subtle differences in activation patterns according to task characteristics.

Statistical analysis of inclusion and exclusion trials between blocks, for the AI and PI are shown in Figure 2A. For both regions, the theta signal in the exclusion blocks was significantly (Permutation test, Bonferroni-corrected P < 0.05) greater than in the inclusion blocks (except for the last block in AI, where no significant difference between inclusion and exclusion trials was found). For the AI, an analysis across the different blocks of exclusion showed a significant difference (P < 0.05) of the first and the second block compared with the last one, in which the signal rapidly decreased. In contrast, in the PI the signal was significantly greater in the second (Permutation test, Bonferroni-corrected P < 0.05) block of exclusion compared with the first one.

Figure 2.

Theta signal change in anterior and posterior insula. (A) Statistical analysis in patients who significantly respond to social exclusion in the anterior (3 patients) and posterior insula (5 patients). The total number of patient having electrodes was 7 for the anterior insula and 11 for the posterior. The vertical bars in the 2 graphs represents the theta z-score activation during inclusion (gray bar) and exclusion blocks (black bar). Asterisk indicates significant effects between exclusion and the corresponding inclusion block, P < 0.05. Dotted lines show significant differences across exclusion blocks. The panel on the right shows a 3D brain template, revealing the insular cortex segmentation. The red dots indicate the positions of the electrodes showing theta activity in the anterior insula (AI) for all patients. The fuchsia dots show the electrodes that presented theta activity in the posterior insula (PI) in all patients. The size of the dot corresponds to the number of electrodes showing significant activity in this region. (B) The top middle panel shows a magnification of the insular cortex with the 2 electrodes implanted in patient P.E. On the left and the right side the average theta power results are presented for patient P.E. across blocks during the game. The periods of inclusion (I) and exclusion (E) are shown in light gray and black, respectively. On the bottom, average time-frequency diagram of theta signal for patient P.E. for all the inclusion, (left diagram) and all the exclusion (right diagram) trials are illustrated; 0 = onset of inclusion or exclusion trial.

Figure 2.

Theta signal change in anterior and posterior insula. (A) Statistical analysis in patients who significantly respond to social exclusion in the anterior (3 patients) and posterior insula (5 patients). The total number of patient having electrodes was 7 for the anterior insula and 11 for the posterior. The vertical bars in the 2 graphs represents the theta z-score activation during inclusion (gray bar) and exclusion blocks (black bar). Asterisk indicates significant effects between exclusion and the corresponding inclusion block, P < 0.05. Dotted lines show significant differences across exclusion blocks. The panel on the right shows a 3D brain template, revealing the insular cortex segmentation. The red dots indicate the positions of the electrodes showing theta activity in the anterior insula (AI) for all patients. The fuchsia dots show the electrodes that presented theta activity in the posterior insula (PI) in all patients. The size of the dot corresponds to the number of electrodes showing significant activity in this region. (B) The top middle panel shows a magnification of the insular cortex with the 2 electrodes implanted in patient P.E. On the left and the right side the average theta power results are presented for patient P.E. across blocks during the game. The periods of inclusion (I) and exclusion (E) are shown in light gray and black, respectively. On the bottom, average time-frequency diagram of theta signal for patient P.E. for all the inclusion, (left diagram) and all the exclusion (right diagram) trials are illustrated; 0 = onset of inclusion or exclusion trial.

TF analysis showed a similar pattern of activation in the AI and PI. TF diagrams showed a clear theta band wave for the exclusion trials (Permutation test, Bonferroni-corrected, P < 0.05) (bottom panels of Fig. 2B). Average theta power over the time course of the game (upper panels of Fig. 2B) revealed clear differences between the anterior and the posterior regions of the insula. In the AI, there was increased activation from the first to the second block (maximum z-score in the first block of exclusion = 7.61; maximum z-score in the second block of exclusion = 18.33) and a drastic decrease in the third and the last block (maximum z-score in the third block of exclusion = 2.03). In the PI, by contrast, we observed increased activity from the first (maximum z-score = 0.98 to the second block of exclusion (maximum z-score = 2.08) to the third block of exclusion (maximum z-score = 2.36). Globally, the theta wave effect in the exclusion blocks was stronger for the AI compared with the PI (see Fig. 2A).

Another important region of the pain matrix that has been identified in several social pain studies (Eisenberger et al. 2003, 2007) is the cingulate cortex. The TF analysis revealed that the only sector in the cingulate cortex that responds significantly to the exclusion period in the theta band frequency is the sACC (Permutation test, Bonferroni-corrected P < 0.05). This region is well known for its involvement in emotion processing and regulation of mood disorders such as sadness (Drevets et al. 2008). It also has important cortical connections with regions such as the nucleus accumbens, the amygdala, the hypothalamus, and the orbitofrontal cortex (Johansen-Berg et al. 2008).

The results on average change in theta power during the game's time course showed that activation within the sACC (BA 32) decreased slightly from the first (maximum z-score = 7.12), to the second (maximum z-score = 6.97), and finally to the third (maximum z-score = 5.49) block of exclusion (Fig. 3A). Interestingly, it appears that the time of the occurrence of the first theta burst, decreases progressively from the first (35 s) to the second (15 s) and to the third (5 s) block of exclusion. This result suggests that the subject learns about the exclusion condition and that the sACC is directly involved in learning and predicting the effects of the exclusion event. As shown in Figure 3B,C, the other sub-regions of the cingulate cortex (pACC and dPCC) failed to show any significant variation between inclusion and exclusion trials, suggesting that the feeling of being excluded is mediated by a specific sector of the cingulate cortex, namely the subgenual portion. However, this result should be considered cautiously, given the fact that only one patient had an electrode implanted in this region. The others areas that showed significant theta activity (Permutation test, Bonferroni-corrected P < 0.05), during the exclusion period were the orbitofrontal cortex (BA10 and BA47), the dlPFC (BA9) and the supplementary motor area (SMA- BA6) (Supplementary Fig. 3AD). The primary somatosensory cortex (S1), typically involved in physical pain processing, was not found to have been affected by social pain in our experiment (Supplementary Fig. 4B) (Permutation test, Bonferroni-corrected P > 0.05).

Figure 3.

Signal changes in cingulate cortex during the ball game task. Sagittal MRI template representing the recording locations for the intracranial electrodes located in the cingulate cortex of 10 patients. Dots highlighted with the arrows represent data for patient C.T. C.T. was the only patient having an electrode in the subgenual cingulate cortex (dot in red). The remaining dots (in green) represent the electrodes recorded in other patients (N = 9), in addition to the electrodes that do not respond in patient C.T. (AC). Time course of average theta power recorded throughout the experiment during inclusion (I) and exclusion (E) blocks in the (A) dorsal posterior (dPCC), (B) pregenual anterior (pACC), and (C) subgenual anterior (sACC) cingulate cortex. Power is coded as z-score calculated from a baseline period “technical exclusion,” where subjects were excluded for technical reasons.

Figure 3.

Signal changes in cingulate cortex during the ball game task. Sagittal MRI template representing the recording locations for the intracranial electrodes located in the cingulate cortex of 10 patients. Dots highlighted with the arrows represent data for patient C.T. C.T. was the only patient having an electrode in the subgenual cingulate cortex (dot in red). The remaining dots (in green) represent the electrodes recorded in other patients (N = 9), in addition to the electrodes that do not respond in patient C.T. (AC). Time course of average theta power recorded throughout the experiment during inclusion (I) and exclusion (E) blocks in the (A) dorsal posterior (dPCC), (B) pregenual anterior (pACC), and (C) subgenual anterior (sACC) cingulate cortex. Power is coded as z-score calculated from a baseline period “technical exclusion,” where subjects were excluded for technical reasons.

Although the amygdala is thought to be important for emotion processing (Pessoa and Adolphs 2010), we did not find any significant response in this area during the task, for all patients having this region implanted. None of our patients had amygdala atrophy. In other words, for this region, there was neither a significant difference across inclusion and exclusion trials in TF analysis nor a significant effect in the z-score of theta band during the course of the game (Permutation test, Bonferroni-corrected, P > 0.05). This observation holds for all the subjects (66%) having electrodes in this region. (See Supplementary Fig. 4A).

Theta Signal in the Fusiform Cortex

Although the fusiform brain area is not a classical node of the pain matrix (either social or physical), our results showed a significant response within this area, probably due to task demands. In particular, we found that the theta band in the fusiform brain area was more active during exclusion compared with the inclusion trials. Figure 4A shows a coronal template representing, for all patients, all the electrodes that showed significant theta power activation during exclusion.

Figure 4.

Signal changes in the fusiform cortex. (A) Coronal template representing, for all patients, all the electrodes that showed significant theta power activation during exclusion in 7 patients. The total number of patients having the fusiform cortex implanted was 9. Light blue (fusiform area). (B) Statistical analysis of theta power between blocks, across subjects. Power is coded as z-score calculated from the baseline period. (C) Evolution of the theta band within a trial. The black and gray lines represent the mean theta power for exclusion trials forumla and inclusion trials forumla, respectively. forumla and forumla. are significantly different for 2 distinct periods of the trial: first from 0 to 0.7 s when the ball has not been thrown and second from 1.6 to 2 s when the player receives the ball. (D) Moreover, in the first period (0–0.7 s), which represents a period of uncertainty, there is a significant block effect for the exclusion trials with an increase in theta activity from the first to the last block (P < 0.05, Bonferroni-corrected).

Figure 4.

Signal changes in the fusiform cortex. (A) Coronal template representing, for all patients, all the electrodes that showed significant theta power activation during exclusion in 7 patients. The total number of patients having the fusiform cortex implanted was 9. Light blue (fusiform area). (B) Statistical analysis of theta power between blocks, across subjects. Power is coded as z-score calculated from the baseline period. (C) Evolution of the theta band within a trial. The black and gray lines represent the mean theta power for exclusion trials forumla and inclusion trials forumla, respectively. forumla and forumla. are significantly different for 2 distinct periods of the trial: first from 0 to 0.7 s when the ball has not been thrown and second from 1.6 to 2 s when the player receives the ball. (D) Moreover, in the first period (0–0.7 s), which represents a period of uncertainty, there is a significant block effect for the exclusion trials with an increase in theta activity from the first to the last block (P < 0.05, Bonferroni-corrected).

The statistical analysis for the FFA showed that the z-score for every block of exclusion was significantly higher than the z-score for the corresponding inclusion block (P < 0.001), see Figure 4B. All the statistics were made using Bonferroni-corrected permutation tests (n = 10 000). This activity was present in 77% of the patients' population having this region implanted and was not found in other visual regions such as the occipital cortex (Supplementary Fig. 4C). Looking at within-trial data, it appears that the significant theta activation was restricted to the first 700 ms and the last 400 ms of the exclusion trials (Fig. 4C). If we consider only the first 700 ms of each trial, we observe a significant difference between the first and the last blocks of exclusion (P < 0.05). This can be observed in the bar graph in Figure 4D.

Neural Synchrony of Theta Signal

Overall, we found increased synchrony during exclusion trials for theta and a lower effect for the others FB (alpha, beta, low, and high gamma). This is not surprising considering that the theta band was the only one to be affected during the exclusion trials in our task (see above). As a consequence, in the following section, we will focus exclusively on the theta band synchrony (detailed data are shown for all frequencies in Supplementary Fig. 5).

We conducted the analysis for 2 separate time windows: The first one from 0 to 0.7 s, corresponding to a period of “uncertainty” where the subject does not know the direction taken by the ball (inclusion or exclusion), and a second one from 0.7 to 2 s, corresponding to a period of ‘certainty’ where ball direction toward a participant was clear. In both periods (certainty and uncertainty), we found the theta phase synchrony in the exclusion trials to be higher than the inclusion trials. Overall, 81.0% and 80.6% of the significant pairs of electrodes (3% and 4.1% among all pairs) were more synchronized during exclusion blocks, irrespective of the period (uncertainty or certainty). Figure 5 shows this pattern. Also, in the uncertainty period, a substantial number of electrodes were significantly synchronized. The difference of PLV between exclusion and inclusion periods ΔPLV(t, f) was defined as forumla with forumla and forumla corresponding to the set of exclusion and inclusion trials, respectively.

Figure 5.

Phase synchrony of theta waves over the course of a trial: uncertainty (A) and certainty (B). The figure shows sagittal (top) and horizontal (bottom) orthogonal views of a transparent “glass brain.” The lines represent the significant phase-locking value differences (ΔPLVs) between exclusion and inclusion trials (exclusion minus inclusion) for all pairs of electrodes for 2 time windows: the uncertainty period A (0 –0.7 s) and the certainty period B (0.7 –2 s). Segments are colored and sized according to the difference of PLV in theta frequency band.

Figure 5.

Phase synchrony of theta waves over the course of a trial: uncertainty (A) and certainty (B). The figure shows sagittal (top) and horizontal (bottom) orthogonal views of a transparent “glass brain.” The lines represent the significant phase-locking value differences (ΔPLVs) between exclusion and inclusion trials (exclusion minus inclusion) for all pairs of electrodes for 2 time windows: the uncertainty period A (0 –0.7 s) and the certainty period B (0.7 –2 s). Segments are colored and sized according to the difference of PLV in theta frequency band.

The power of the synchrony ΔPLV in E>I (E, exclusion and I, inclusion) is stronger for specific regions. For example, during the uncertainty period (0–0.7 s) there was more synchrony between the left AI and the left PI, between the right superior frontal gyrus (BA6) and the right middle frontal gyrus (BA6), between the left PI and the left hippocampus and between the left parahyppocampal gyrus and left medial frontal gyrus (BA10) (See Supplementary Table 3 for the exact coordinates). In this period, short-range synchronies are predominant and almost all the synchronies are between the electrodes located in the same hemisphere (left or right). Concerning the second period of the trial (0.7–2 s), there was more synchrony between the left and the right AI, between the left amygdala and the left AI, between left insula and superior temporal gyrus (BA 41-22), between inferior frontal gyrus (BA9) and left insula, between right parahippocampal gyrus and right insula, and between the right and the left superior frontal gyrus (BA6). Additionally, there are regions for which the power of the synchrony (ΔPLV) is more marked in inclusion compared with exclusion trials (I>E). These are mainly frontal regions. This could reflect the involvement of these regions in the reward system. It could be that the inclusion is perceived as a form of reward by the participants. The details of these analyses are reported in Supplementary Table 3. We also plotted the phase synchrony maps of the other frequency windows: alpha, beta, low, and high gamma (see Supplementary Fig. 5) but as they clearly show the effect is lower compared with theta signal.

Discussion

Our results show that the feeling of pain provoked by social exclusion in a computer ball-tossing game has a clear “theta” signature in the human brain. This signature is visible in some of the most important regions of the so called “physical pain matrix.” The observation that social and physical pain rely, at least partially, on the same neural network is also confirmed by recent fMRI studies (Eisenberger et al. 2003, 2007). Our work goes beyond these results by showing that theta waves are the neural code through which the brain processes this kind of “painful” information. These waves go from 3 to 7 Hz and they are well known in the literature for being linked to several primary aspects of human behavior: Spatial navigation (Kahana et al. 1999), working memory (Rizzuto et al. 2003), emotion regulation (Aftanas et al. 2001, 2003, 2004), and learning (Caplan and Glaholt 2007). A recent study in mice has shown that theta waves are present in the ACC during fear learning (Jeon et al. 2010). These signals are also involved in pain physical perception (Liu et al. 2010; Schulz et al. 2011) and in empathy for pain in humans (Mu et al. 2008).

Classically, the pain matrix involves several brain regions, segmented into 2 sub-networks: A physical one that elaborates the sensory property of nociceptive stimuli and an affective one that elaborates the cognitive features of these stimuli (Peyron et al. 2000). In the present study, the brain regions that were found to present theta wave activation in response to social exclusion belong to the affective sub-network of the pain matrix, except for the PI and the FFA. The AI is classically associated with empathy, compassion, fairness, and cooperation but also with proprioceptive awareness as the representation of emotional states (Craig 2009). Here, we have shown that it is also involved in the subjective feeling of social exclusion, probably playing a role of social pain detector. Indeed, activity within this area was higher in the first 2 exclusion blocks than in the last one. In contrast with this pattern, the posterior portion of the insular cortex responded to exclusion, but no specific modulation across blocks was observed. Rather, the response within this area was constant during the different exclusion blocks, as would have been expected from a region processing the sensorial and visceral components of pain. This conclusion is in line with previous results showing that the posterior portion of the insular cortex mediates the processing of primary interoceptive representations and sensations, due to its large connections with the thalamus (Craig 2009).

In this study, the cingulate cortex presented a theta wave only in its subgenual sector (sACC). In the study of Eisenberger et al. (2003) and Eisenberger and Lieberman (2004) using a similar paradigm, the authors reported activation of the dorsal medial portion of the ACC (dACC). Previous studies have shown a positive correlation between self-reported distress and dACC activation during social exclusion (Eisenberger et al. 2003, 2007). However, subsequent studies have linked dACC with expectation violation elicited by exclusion but not with the painful feeling of exclusion itself. (Somerville et al. 2006). Within this context, it can be argued that it is the sACC that is linked to the pure feeling of social pain (Somerville et al. 2006; Bolling et al., 2011a, b; Sebastian et al. 2011). In agreement with this claim, sACC activity has been shown to be greater in adolescents experiencing distress after peer rejection (Masten et al. 2009) and reduced when people are experiencing exclusion in the presence of social emotional support (Onoda et al. 2009). The involvement of the sACC in social pain is consistent with the fact that this area is strongly connected to other limbic structures and the AI and is associated with the expression of negative emotions such as sadness and depression (Johansen-Berg et al. 2008). In our study, there was a rapid decrease in the latency of the sACC response from the first to the last block of exclusion, which suggests that this structure can learn to signal rapidly the social value of exclusion. Since sACC contribution was only observed in a single patient implanted in this region, this result should be taken cautiously. Nevertheless, one may note that our observation is consistent with the known role of sACC in pain processing (Somerville et al. 2006; Bolling et al. 2011a, b; Sebastian et al. 2011).

Our study also shows a theta signal in some frontal regions already known to be involved in physical and social pain processing (BA10, BA9, BA47, and BA6; Peyron et al. 2000; Eisenberger et al. 2003). All these regions presented a similar significant theta pattern during exclusion compared with inclusion trials, but none of them were modulated during blocks. The fMRI study from Eisenberger et al. (2003), showed a positive correlation between social pain perception and the frontal network activation suggesting that the frontal network is probably devoted to cognitive rather than emotional regulation of pain.

A critical aspect that seems to hold for all regions activated in the exclusion trials (with the exception of the fusiform area) is that the firing within the theta band is present during the totality of the trial. This suggests that the feeling of social pain is an on-going state. In other words, social pain is not directly associated to a precise stimulus, but to the understanding of being excluded and to the experience of this unpleasant feeling within a specific social context.

Other regions, important for physical pain such as somatosensory cortices (Peyron et al. 2000) and emotion processing such as amygdala (Pessoa and Adolphs 2010) were not active in our study (Supplementary Fig. 4A,B). The amygdala has an important role in fearful situations but also in social cognition without fearful experiences (Kennedy et al., 2009; Krill and Platek, 2009). However, our paradigm failed to show amygdala activation, in line with previous studies on social pain using a comparable protocol (Eisenberger et al. 2003, 2007). As argued by Eisenberger and Cole (2012), the amygdala could respond to concrete negative stimuli, rather than more complex socioemotional contexts such as the feeling of social pain .

In our study, the fusiform area was found to be more active in the exclusion than in the inclusion trials during the time when subjects were paying attention to the partner because the direction of the ball was not yet known. Interestingly, this effect increased from the first to the last block, which suggests that the subjects were learning the social value of the face of the other players. In agreement with this result, it is known that the FFA is not only involved in the coding of faces (Haxby et al. 2000) but also in learning their affective value (Kanwisher et al. 1997). Strikingly, here we show that this area responds with theta wave exclusively during exclusion trials and more important, this activity is time locked to a period where the ball is not moving (static periods). It is tempting to suggest that during this period the subject is attributing a specific social value to the partner's face, for example, unfairness because of being excluded.

As shown by the synchrony results, theta waves not only represent the neural code through which regions of the so-called pain matrix elaborate this kind of feelings, but also the neural signature through which distant brain regions communicate when social pain is experienced. These regions include several nodes that belong to the sensory (amygdala, PI) and affective (AI, the ACC, and PFC) pain matrix. The fact that we found so many cortical regions synchronized, it is not surprising, because normally the low-frequency oscillations involve large-scale synchrony (Varela et al. 2001).

Of course, our results now need to be replicated in healthy subjects using scalp EEG. With respect to this point, it may be worth mentioning a recent study of Crowley et al. (2010) reporting a slow brain wave activity in adolescents experiencing rejection distress. Another important extension of our work would be the investigation of other groups of patients, including, for instance, patients with depressive disorders. Indeed, it has been reported that patients suffering from drug-resistant major depression disorders can undergo intracranial implantation in order to reduce depressive symptoms (Mayberg et al. 2005). Of particular interest in this case is the fact that the sACC is a key implanted region in these patients. This would allow us to investigate how the pain matrix network is (under or over) activated in depressed patients and to generalize on the crucial role of the sACC in social pain as we observed in only one patient of our sample. Taken together, the results presented in this work, suggest that slow theta waves play a crucial role in processing painful feelings such as social exclusion. Specifically, our results highlight the neural dynamics of social exclusion and point to the link between theta waves and the experience of primitive feelings associated with social interaction. Finally, our findings show that theta signals are an important marker that predicts the feeling of social discomfort an individual experiences. This may have clinical relevance for detecting early stages of psychiatric disorders linked to negative social experiences and evolving toward a more complex clinical tableau of social phobia or depression.

Author Contributions

I.C. and A.S. designed research; I.C., L.M., S.H., A.P., G.D., J.I., F.M., and A.S. performed research; I.C. and S.H. analyzed data; S.H. developed analytic tool; I.C. and S.H. constructed the figures and wrote figure legends; I.C., S.H., and A.S. co-wrote the paper.

Supplementary Material

Supplementary material can be found at:http://www.cercor.oxfordjournals.org/

Funding

This work was supported by CNRS (to A.S) and Fondation pour la Recherche Médicale (to A.S. and I.C.).

Notes

We thank the patients for their kind cooperation and the clinical staff (301 Unit, Wertheimer Neurological Hospital of Lyon) for their help during testing. We are also grateful to Prof. Marc Guenot for stereotactic electrodes implantation. A special thanks to Dr Michel Desmurget for his helpful comments during the revision of the manuscript. Conflict of Interest: None declared.

References

Aftanas
LI
Reva
NV
Varlamov
AA
Pavlov
SV
Makhnev
VP
Analysis of evoked EEG synchronization and desynchronization in conditions of emotional activation in humans: temporal and topographic characteristics
Neurosci Behav Physiol
 , 
2004
, vol. 
34
 (pg. 
859
-
867
)
Aftanas
LI
Varlamov
AA
Pavlov
SV
Makhnev
VP
Reva
NV
Affective picture processing: event-related synchronization within individually defined human theta band is modulated by valence dimension
Neurosci Lett
 , 
2001
, vol. 
303
 (pg. 
115
-
118
)
Aftanas
LI
Varlamov
AA
Reva
NV
Pavlov
SV
Disruption of early event-related theta synchronization of human EEG in alexithymics viewing affective pictures
Neurosci Lett
 , 
2003
, vol. 
340
 (pg. 
57
-
60
)
Baas
JM
Kenemans
JL
Bocker
KB
Verbaten
MN
Threat-induced cortical processing and startle potentiation
Neuroreport
 , 
2002
, vol. 
13
 (pg. 
133
-
137
)
Bekkedal
MY
Rossi
J
Panksepp
J
Human brain EEG indices of emotions: delineating responses to affective vocalizations by measuring frontal theta event-related synchronization
Neurosci Biobehav Rev
 , 
2011
, vol. 
35
 (pg. 
1959
-
1970
)
Bolling
DZ
Pitskel
NB
Deen
B
Crowley
MJ
Mayes
LC
Pelphrey
KA
Development of neural systems for processing social exclusion from childhood to adolescence
Dev Sci
 , 
2011a
, vol. 
14
 (pg. 
1431
-
1444
)
Bolling
DZ
Pitskel
NB
Deen
B
Crowley
MJ
McPartland
JC
Mayes
LC
Pelphrey
KA
Dissociable brain mechanisms for processing social exclusion and rule violation
NeuroImage
 , 
2011b
, vol. 
54
 (pg. 
2462
-
2471
)
Buzsaki
G
Theta oscillations in the hippocampus
Neuron
 , 
2002
, vol. 
33
 (pg. 
325
-
340
)
Cacioppo
JT
Hawkley
LC
Perceived social isolation and cognition
Trends Cogn Sci
 , 
2009
, vol. 
13
 (pg. 
447
-
454
)
Caplan
JB
Glaholt
MG
The roles of EEG oscillations in learning relational information
Neuroimage
 , 
2007
, vol. 
38
 (pg. 
604
-
616
)
Chen
Z
Williams
KD
Fitness
J
Newton
NC
When hurt will not heal: exploring the capacity to relive social and physical pain
Psychol Sci
 , 
2008
, vol. 
19
 (pg. 
789
-
795
)
Craig
AD
How do you feel–now? The anterior insula and human awareness
Nat Rev Neurosci
 , 
2009
, vol. 
10
 (pg. 
59
-
70
)
Crowley
MJ
Wu
J
Bailey
CA
Mayes
LC
Bringing in the negative reinforcements: the avoidance feedback-related negativity
Neuroreport
 , 
2009
, vol. 
20
 (pg. 
1513
-
1517
)
Crowley
MJ
Wu
J
Molfese
PJ
Mayes
LC
Social exclusion in middle childhood: rejection events, slow-wave neural activity, and ostracism distress
Social Neurosci
 , 
2010
, vol. 
5
 (pg. 
483
-
495
)
Drevets
WC
Savitz
J
Trimble
M
The subgenual anterior cingulate cortex in mood disorders
CNS Spectrums
 , 
2008
, vol. 
13
 (pg. 
663
-
681
)
Eisenberger
NI
Identifying the neural correlates underlying social pain: implication for developmental processes
Hum Dev
 , 
2006
, vol. 
49
 (pg. 
273
-
293
)
Eisenberger
NI
Cole
SW
Social neuroscience and health: neurophysiological mechanisms linking social ties with physical health
Nat Neurosci
 , 
2012
, vol. 
15
 (pg. 
669
-
674
)
Eisenberger
NI
Lieberman
MD
Why rejection hurts: a common neural alarm system for physical and social pain
Trends Cogn Sci
 , 
2004
, vol. 
8
 (pg. 
294
-
300
)
Eisenberger
NI
Lieberman
MD
Williams
KD
Does rejection hurt? An FMRI study of social exclusion
Science
 , 
2003
, vol. 
302
 (pg. 
290
-
292
)
Eisenberger
NI
Way
BM
Taylor
SE
Welch
WT
Lieberman
MD
Understanding genetic risk for aggression: clues from the brain's response to social exclusion
Biol Psychiatry
 , 
2007
, vol. 
61
 (pg. 
1100
-
1108
)
Gow
AJ
Pattie
A
Whiteman
MC
Whalley
LJ
Deary
IJ
Social support and successful aging: Investigating the relationships between lifetime cognitive change and life satisfaction
J Indiv Differ
 , 
2007
, vol. 
28
 (pg. 
103
-
115
)
Haxby
JV
Hoffman
EA
Gobbini
MI
The distributed human neural system for face perception
Trends Cogn Sci
 , 
2000
, vol. 
4
 (pg. 
223
-
233
)
Jeon
D
Kim
S
Chetana
M
Jo
D
Ruley
HE
Lin
SY
Rabah
D
Kinet
JP
Shin
HS
Observational fear learning involves affective pain system and Cav1.2 Ca2+ channels in ACC
Nat Neurosci
 , 
2010
, vol. 
13
 (pg. 
482
-
488
)
Jerbi
K
Freyermuth
S
Dalal
S
Kahane
P
Bertrand
O
Berthoz
A
Lachaux
JP
Saccade related gamma-band activity in intracerebral EEG: dissociating neural from ocular muscle activity
Brain Topogr
 , 
2009
, vol. 
22
 (pg. 
18
-
23
)
Johansen-Berg
H
Gutman
DA
Behrens
TE
Matthews
PM
Rushworth
MF
Katz
E
Lozano
AM
Mayberg
HS
Anatomical connectivity of the subgenual cingulate region targeted with deep brain stimulation for treatment-resistant depression
Cereb Cortex
 , 
2008
, vol. 
18
 (pg. 
1374
-
1383
)
Kahana
MJ
Sekuler
R
Caplan
JB
Kirschen
M
Madsen
JR
Human theta oscillations exhibit task dependence during virtual maze navigation
Nature
 , 
1999
, vol. 
399
 (pg. 
781
-
784
)
Kanwisher
N
McDermott
J
Chun
MM
The fusiform face area: a module in human extrastriate cortex specialized for face perception
J Neurosci
 , 
1997
, vol. 
17
 (pg. 
4302
-
4311
)
Kennedy
DP
Glascher
J
Tyszka
JM
Adolphs
R
Personal space regulation by the human amygdala
Nat Neurosci
 , 
2009
, vol. 
12
 (pg. 
1226
-
1227
)
Knyazev
GG
Motivation, emotion, and their inhibitory control mirrored in brain oscillations
Neurosci Biobehav Rev
 , 
2007
, vol. 
31
 (pg. 
377
-
395
)
Krause
CM
Viemero
V
Rosenqvist
A
Sillanmaki
L
Astrom
T
Relative electroencephalographic desynchronization and synchronization in humans to emotional film content: an analysis of the 4–6, 6–8, 8–10 and 10-12 Hz frequency bands
Neurosci Lett
 , 
2000
, vol. 
286
 (pg. 
9
-
12
)
Krill
A
Platek
SM
In-group and out-group membership mediates anterior cingulate activation to social exclusion
Front. Evol. Neurosci
 , 
2009
, vol. 
1
 pg. 
1
 
Kross
E
Berman
MG
Mischel
W
Smith
EE
Wager
TD
Social rejection shares somatosensory representations with physical pain
Proc Natl Acad Sci USA
 , 
2011
, vol. 
108
 (pg. 
6270
-
6275
)
Lachaux
JP
Rodriguez
E
Martinerie
J
Varela
FJ
Measuring phase synchrony in brain signals
Hum Brain Mapp
 , 
1999
, vol. 
8
 (pg. 
194
-
208
)
Lachaux
JP
Rudrauf
D
Kahane
P
Intracranial EEG and human brain mapping
J Physiol Paris
 , 
2003
, vol. 
97
 (pg. 
613
-
628
)
Liu
CC
Ohara
S
Franaszczuk
P
Zagzoog
N
Gallagher
M
Lenz
FA
Painful stimuli evoke potentials recorded from the medial temporal lobe in humans
Neuroscience
 , 
2010
, vol. 
165
 (pg. 
1402
-
1411
)
Macdonald
G
Leary
MR
Why does social exclusion hurt? The relationship between social and physical pain
Psychol Bull
 , 
2005
, vol. 
131
 (pg. 
202
-
223
)
Masten
CL
Eisenberger
NI
Borofsky
LA
Pfeifer
JH
McNealy
K
Mazziotta
JC
Dapretto
M
Neural correlates of social exclusion during adolescence: understanding the distress of peer rejection
Soc Cogn Affect Neurosci
 , 
2009
, vol. 
4
 (pg. 
143
-
157
)
Mayberg
HS
Lozano
A
Voon
V
McNeely
H
Seminowicz
D
Hamani
C
Schwalb
J
Kennedy
S
Deep Brain Stimulation for Treatment Resistant Depression
Neuron
 , 
2005
, vol. 
45
 (pg. 
651
-
660
)
McNair
DM
Lorr
M
Droppleman
LF
Manual for the Profile of Mood States
 , 
1971
San Diego, CA
Educational and Industrial Testing Services
Mu
Y
Fan
Y
Mao
L
Han
S
Event-related theta and alpha oscillations mediate empathy for pain
Brain Res
 , 
2008
, vol. 
1234
 (pg. 
128
-
136
)
Onoda
K
Okamoto
Y
Nakashima
K
Nittono
H
Ura
M
Yamawaki
S
Decreased ventral anterior cingulate cortex activity is associated with reduced social pain during emotional support
Soc Neurosci
 , 
2009
, vol. 
4
 (pg. 
443
-
454
)
Ossandon
T
Vidal
JR
Ciumas
C
Jerbi
K
Hamame
CM
Dalal
SS
Bertrand
O
Minotti
L
Kahane
P
Lachaux
JP
Efficient “pop-out” visual search elicits sustained broadband gamma activity in the dorsal attention network
J Neurosci
 , 
2012
, vol. 
32
 (pg. 
3414
-
3421
)
Panksepp
J
Feeling the pain of social loss
Science
 , 
2003
, vol. 
302
 (pg. 
237
-
239
)
Pessoa
L
Adolphs
R
Emotion processing and the amygdala: from a ‘low road’ to ‘many roads’ of evaluating biological significance
Nat Rev Neurosci
 , 
2010
, vol. 
11
 (pg. 
773
-
783
)
Peyron
R
Laurent
B
Garcia-Larrea
L
Functional imaging of brain responses to pain. A review and meta-analysis
. Neurophysiol Clin
 , 
2000
, vol. 
30
 (pg. 
263
-
288
)
Rizzuto
DS
Madsen
JR
Bromfield
EB
Schulze-Bonhage
A
Seelig
D
Aschenbrenner-Scheibe
R
Kahana
MJ
Reset of human neocortical oscillations during a working memory task
Proc Natl Acad Sci USA
 , 
2003
, vol. 
100
 (pg. 
7931
-
7936
)
Schulz
E
Tiemann
L
Schuster
T
Gross
J
Ploner
M
Neurophysiological coding of traits and states in the perception of pain
Cereb Cortex
 , 
2011
, vol. 
21
 (pg. 
2408
-
2414
)
Sebastian
CL
Tan
GC
Roiser
JP
Viding
E
Dumontheil
I
Blakemore
SJ
Developmental influences on the neural bases of responses to social rejection: implications of social neuroscience for education
NeuroImage
 , 
2011
, vol. 
57
 (pg. 
686
-
694
)
Singer
T
Seymour
B
O'Doherty
J
Kaube
H
Dolan
RJ
Frith
CD
Empathy for pain involves the affective but not sensory components of pain
Science
 , 
2004
, vol. 
303
 (pg. 
1157
-
1162
)
Somerville
LH
Heatherton
TF
Kelley
WM
Anterior cingulate cortex responds differentially to expectancy violation and social rejection
Nat Neurosci
 , 
2006
, vol. 
9
 (pg. 
1007
-
1008
)
Talairach
J
Bancaud
J
Stereotaxic approach to epilepsy: methodology of anatomo-functional stereotaxic investigations
Progr Neurol Surg
 , 
1974
, vol. 
5
 (pg. 
297
-
354
)
Tallon-Baudry
C
Bertrand
O
Delpuech
C
Pernier
J
Stimulus specificity of phase-locked and non-phase-locked 40 Hz visual responses in human
J Neurosci
 , 
1996
, vol. 
16
 (pg. 
4240
-
4249
)
Tallon-Baudry
C
Bertrand
O
Henaff
MA
Isnard
J
Fischer
C
Attention modulates gamma-band oscillations differently in the human lateral occipital cortex and fusiform gyrus
Cereb Cortex
 , 
2005
, vol. 
15
 (pg. 
654
-
662
)
van Beest
I
Williams
KD
When inclusion costs and ostracism pays, ostracism still hurts
J Pers Soc Psychol
 , 
2006
, vol. 
91
 (pg. 
918
-
928
)
Varela
F
Lachaux
JP
Rodriguez
E
Martinerie
J
The brainweb: phase synchronization and large-scale integration
Nat Rev Neurosci
 , 
2001
, vol. 
2
 (pg. 
229
-
239
)
Williams
KD
Cheung
CKT
Choi
W
CyberOstracism: effect of being ignored over the Internet
J Personal Soc Psychol
 , 
2000
, vol. 
79
 (pg. 
748
-
762
)
Williams
KD
Forgas
JP
von Hippel
W
The social outcast Social rejection, exclusion and ostracism
 , 
2005
New York Psychological press
Williams
KD
Jarvis
B
Cyberball: a program for use in research on ostracism and interpersonal acceptance
Behav Res Methods Instrum Comput
 , 
2006
, vol. 
38
 (pg. 
174
-
180
)