Little is known about the neural basis of elite performers and their optimal performance in extreme environments. The purpose of this study was to examine brain processing differences between elite warfighters and comparison subjects in brain structures that are important for emotion processing and interoception.
Navy Sea, Air, and Land Forces (SEALs) while off duty (n = 11) were compared with n = 23 healthy male volunteers while performing a simple emotion face-processing task during functional magnetic resonance imaging. Irrespective of the target emotion, elite warfighters relative to comparison subjects showed relatively greater right-sided insula, but attenuated left-sided insula, activation. Navy SEALs showed selectively greater activation to angry target faces relative to fearful or happy target faces bilaterally in the insula. This was not accounted for by contrasting positive versus negative emotions. Finally, these individuals also showed slower response latencies to fearful and happy target faces than did comparison subjects.
These findings support the hypothesis that elite warfighters deploy greater processing resources toward potential threat-related facial expressions and reduced processing resources to non-threat-related facial expressions. Moreover, rather than expending more effort in general, elite warfighters show more focused neural and performance tuning. In other words, greater neural processing resources are directed toward threat stimuli and processing resources are conserved when facing a nonthreat stimulus situation.
Citation: Paulus MP, Simmons AN, Fitzpatrick SN, Potterat EG, Van Orden KF, Bauman J, et al. (2010) Differential Brain Activation to Angry Faces by Elite Warfighters: Neural Processing Evidence for Enhanced Threat Detection. PLoS ONE 5(4): e10096. https://doi.org/10.1371/journal.pone.0010096
Editor: André Aleman, University of Groningen, Netherlands
Received: October 1, 2009; Accepted: February 12, 2010; Published: April 14, 2010
This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration which stipulates that, once placed in the public domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.
Funding: Some of this work was supported by the VA Center of Excellence for Stress and Mental Health (CESAMH), which is supported by Department of Veterans Affairs. CESAMH is affiliated with VISN 22, the Veterans Affairs San Diego Healthcare System, and the University of California, San Diego. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Extreme environments are characterized as those situations that place a high demand on the physiological, affective, cognitive, and/or social processing resources of the individual. Optimal performance in extreme environments is a complex process and its neural basis is poorly understood. There is a surging interest in the use of neuroscience approaches to examine and possibly improve performance in military personnel . Several investigators have examined warfighters in extreme environments to better understand impairments of optimal performance. For example, Lieberman and colleagues  examined the effects of sleep deprivation and environmental stress on performance and mood in Navy Sea, Air, and Land Forces (SEALs) and found significant behavioral decrements. More recently, Morgan and collaborators  proposed a specific mechanism that may contribute to maintenance of optimal performance in extreme environments. Specifically, these authors suggested that vagal suppression, which is modulated by the right insular cortex , is associated with enhanced performance under high-stress conditions.
Optimal performance in extreme situations is a complex problem that is affected by multiple factors , ranging from genetic differences to interpersonal variables. One approach to examining the factors contributing to optimal performance is to compare groups of individuals who are considered “optimal performers” based on special skill sets or training with healthy volunteers. Although, there are currently no experimental probes that have been studied extensively to examine “optimal” performance per se, one can begin to delineate the neural processes that differentiate these groups. The development of a neural signature of elite performers is a first step towards understanding the brain processing characteristics of these indivudals. In particular, there may not be a simple increase or decrease in neural response or behavioral performance, but a capacity to adjust neural processing and behavioral performance during a task to most efficaciously match the environmental demands.
We recently proposed that maintaining an interoceptive balance in the presence of significant perturbations may be a neural marker of optimal performance . Interoception can be defined as the sense of the internal body state and includes a range of sensations, such as pain , temperature , itch , tickle , sensual touch , , muscle tension , air hunger , stomach pH , and intestinal tension . Taken together, these sensations provide an integrated sense of the body's physiological condition . Thus, the interoceptive system plays a crucial role in maintaining a homeostatic state under extreme perturbations. It provides body-related information to other brain areas that monitor value or salience, is important for evaluating reward, and provides critical input to cognitive control processes. This approach is based on extensive work by Craig , Critchley , and others ,  that has provided new insights into how the interoceptive system modulates self-monitoring and creates urges to act to maintain homeostasis. In particular, several neural substrates are thought to mediate these processes, which include the insular cortex in processing emotion-related tasks and the anterior cingulate as a link to cognitive control processes.
In this study, we sought to determine whether elite warfighters (i.e., SEALs), who can be considered considered an example of optimal performers in extreme environments, exhibit distinct neural processing patterns that are consistent with the notion of altered interoceptive processing. To that end, we examined off-duty Navy SEALs while performing a simple emotion face-processing task during functional magnetic resonance imaging (fMRI), and compared them with healthy male volunteers. We examined whether these elite warfighters respond distinctly to target faces exhibiting a variety of emotions. The results demonstrated that active-duty Navy SEALs exhibit a distinct pattern of brain activation during an emotion face-processing task within neural substrates that are important for interoception, indicating that elite warfighters show measurable processing differences compared with normal volunteers.
The latency to respond to a target varied by the type of the target face, F(3,29) = 3.94, p = 0.018 (Figure 1). Although Navy SEALs did not differ from healthy male comparison subjects on the overall response latency, F(1,31) = 2.87, p = 0.10, there was a significant group-by-face interaction, F(3,29) = 6.21, p = 0.002. Specifically, Navy SEALs were relatively slower to respond to happy, t(32) = 3.43, p = 0.002 and fearful, t(32) = 2.74, p = 0.01, faces. There were no significant differences across task conditions on accuracy of responding, F(3,29) = 0.539, p = 0.659. Moreover, Navy SEALs did not differ from healthy male comparison subjects on response accuracy, F(1,31) = 0.714, p = 0.405, or on accuracy as a function of target face, F(3,29) = 1.14, p = 0.349. Taken together, although there were significant latency differences, which were primarily due to longer latencies when matching to happy or fearful target faces by the SEAL group, there were no accuracy differences across groups.
Activation during the emotion face assessment task involved both limbic and paralimbic structures including bilateral insula, amygdala, and the fusiform gyrus (see Figure 2). There were no significant differences across groups in the left amygdala, F(1,32) = 1.42, p = 0.242 or in the group by target face interaction, F(2,63) = 2.71, p = 0.074. Moreover, both groups showed similar activation in the right amygdala, F(1,31) = 0.40, p = 0.529 and did not differ across target faces, F(2,63) = 0.16, p = 0.851.
Task-related activation differed significantly across groups in three areas. Healthy volunteers, relative to SEALs, showed greater face emotion processing related activation in the left anterior insula, F(1,32) = 8.82, p = 0.005. In comparison, SEALs showed greater right mid-insula activation to faces, F(1,32) = 6.55, p = 0.015 (see Figure 3). Finally, whereas comparison subjects showed significant activation in dorsal anterior cingulate, SEALs showed relative deactivation in this area, F(1,32) = 11.21, p = 0.002. Thus, although there were no differences in task-related activation in the amygdala, SEALs showed relatively stronger right insular versus left insular activation, whereas normal volunteers showed the opposite pattern.
There were two areas within the left, F(2,63) = 5.51, p = 0.006, and a trend in the right, F(2,63) = 2.453, p = 0.094, insular cortex that showed group-by-face interactions (see Figure 4). Interestingly, in the right insular cortex, SEALs showed significant activation to angry target faces relative to fearful targets, t(10) = 2.21, p = 0.05, and happy targets, t(10) = 2.51, p = 0.03, whereas the direct comparison was not significant in the left insular cortex. In comparison, normal volunteers did not show this differential effect. Two additional analyses were conducted to determine whether this was simply a positive versus negative emotional valence effect or an effect specific to anger as a target emotion (see Text S1). A reduced mixed model was computed separately for positive versus negative valenced target emotion and for anger versus fear/happiness target emotion, respectively. The group-by-emotion type interaction showed a larger area on bilateral posterior insula for anger versus fear/happiness (Figure S1) but not for positive versus negatively valenced target emotions (Figure S2).
There are three main findings in this study. First, elite warfighters relative to comparison subjects showed relatively greater right-sided insula, but attenuated left-sided insula, activation. Second, these individuals showed selectively greater activation to angry target faces relative to fearful or happy target faces bilaterally in the insula. Third, these individuals also showed slower response latencies to fearful and happy target faces. Taken together, these findings support the notion that elite warfighters, when examined cross-sectionally, deploy greater neural processing resources toward potential threat-related facial expressions and reduced processing resources to non-threat-related facial expressions. This finding suggests that rather than expending more effort in general, elite warfighters show more focused neural and performance tuning, such that greater neural processing resources are directed toward threat stimuli and processing resources are conserved when facing a nonthreat stimulus situation. Moreover, the suggestion of relatively greater right-sided insula activation is consistent with the lateralization of feelings hypothesis, which suggests that right-sided processing is a more energy-consuming (sympathetic) condition .
Navy SEALs are a unique group of elite warfighters. A recently conducted systematic review  and qualitative assessment  revealed several factors that influence the degree to which individuals successfully complete Basic Underwater Demolition/SEAL (BUD/S) training, which is considered one of the most challenging military training programs. Individuals who are likely to complete this training program are characterized by an attitude of mental toughness, achievement motivation, physical strength, physical endurance, emotional stability, and team orientation. These factors are clearly multidimensional but support the critical importance of optimal monitoring and balancing of the relationship between brain processing and body functioning. These factors are reminiscent of Damasio's somatic marker model , which extends the James Lang theory of emotion, and involves the insular cortex that can instantiate body sensation without necessarily receiving peripheral inputs. Specifically, the somatic marker model proposes that “body states” that have been experienced during the past are instantiated in decision-making situations with uncertain outcomes, and provide weights in favor or against choosing specific options. This model has been extended by Craig  who suggested that body states undergo a complex integration within the insular cortex, which is critical for the process of awareness itself. Therefore, the relative neural activation differences between SEALs and comparison subjects may reflect somatic marker differences that are instantiated when presented with specific emotional faces in general and angry faces in particular.
The insula (reviewed elsewhere , ) is a paralimbic structure that constitutes the invaginated portion of the cerebral cortex, forming the base of the sylvian fissure, and is considered limbic sensory cortex by some . Activation of the insular cortex has been reported in a number of processes, including pain , interoceptive , emotion-related , , cognitive , and social processes . Moreover, we have shown that the insular cortex is an important structure for processing the anticipation of aversive emotional states –, risk-taking , and decision-making . In reward-related processes, the insular cortex is important for subjective feeling states and interoceptive awareness ,  and together with middle and inferior frontal gyri, frontal limbic areas, and the inferior parietal lobe plays an important role in inhibitory processing . Thus, differential activations in the insular cortex when assessing an emotional face could be attributed to the degree to which individuals integrate the presentation of a facial expression with the experience of other processes, such as interoception, pain, and social interactions.
Several investigators have proposed that different types of emotions are lateralized to the left- or right-sided hemisphere , , . In particular, these researchers have argued that aversive, negative, or energy-consuming emotions are more right-lateralized, whereas approach, positive, or energy-saving emotions are left-lateralized. Although this assumption has been called into question or has been refined , , this notion still provides a useful heuristic for the current findings. Thus, the relatively greater right insular cortex activation by Navy SEALs supports the idea that these individuals deploy more processing resources to the potential aversive or negative affective associations with facial expressions. Moreover, together with the selectively increased activation to angry target faces in Navy SEALs, these individuals may selectively processing facial features that are critical for potentially aversive or negative consequences. It is important to point out, however, that this cross-sectional study cannot be used to differentiate what could be a trait characteristic or whether this is anger-related processing difference is a consequence of training.
We have proposed a neuroanatomical processing model as a heuristic guide to understand how one can link optimal performance to how the individual “feels inside.” This model focuses on the notion of a body prediction error (i.e., the difference between the value of the anticipated/predicted state and the value of the current interoceptive state) and consists of four components. First, information from peripheral receptors ascends via two different pathways, the A-beta-fiber discriminative pathway that conveys precise information about the “what” and “where” of the stimulus impinging on the body, and the C-fiber pathway that conveys spatially and time-integrated affective information . These afferents converge via several way stations to the sensory cortex and the posterior insular cortex to provide a sense of the current body state. Second, centrally generated interoceptive states (e.g., via contextual associations from memory) reach the insular cortex via temporal and parietal cortex to generate body states based on conditioned associations , . Third, in the insular cortex there is a dorsal-posterior to inferior-anterior organization from granular to agranular, which provides an increasingly “contextualized” representation of the interoceptive state , irrespective of whether it is generated internally or via the periphery. These interoceptive states are made available to the orbitofrontal cortex for context-dependent valuation ,  and to the anterior cingulate cortex for error processing ,  and action valuation , . Fourth, bidirectional connections to the basolateral amygdala , ,  and the striatum , particularly ventral striatum , provide the circuitry to calculate a body prediction error (similar to reward prediction error –), and provide a neural signal for salience and learning. The insular cortex relays information to other brain systems to initiate motivated action to achieve a steady state  by minimizing the body state prediction error. Thus, the greater activation to angry faces in SEALs may represent a relatively stronger body prediction error signal, which would help guide individuals to deploy cognitive and behavioral resources to adjust to anticipated aversive outcomes.
This investigation had several limitations. First, the group of elite warfighters we studied was relatively small and thus there could have been a significant lack of power to detect additional behavioral/functional relationships. With larger number of subjects and different tasks, other important relationships may become apparent. Second, there were no significant correlations between performance on the task and brain activation. This is not surprising, however, because this task is not design to probe emotional or cognitive processes in a performance-related manner. Future investigations will need to use performance-based paradigms (e.g., the detection of mild threat using morphed faces). Third, and most importantly, this cross-sectional study could not address the question whether the observed processing differences were part of the preexisting characteristics of individuals who were selected and then trained to become elite warfighters, or whether these neural processing differences were a consequence of training. Thus, future studies will need to examine, in a within-subjects study design, individuals prior to and again after elite warfighter training.
This study is a first step in elucidating the neural processes that characterize optimal performers. A key difference between optimal performers and comparison subjects revealed in this study is that both neural response and behavioral response are adapted such that greater resources are expended in threat-relevant conditions and conserved in nonrelevant conditions. Thus, the capacity of optimal performers to deploy resources effectively may ensure that they can perform better in extreme situations. However, more studies are needed to examine how modulation of brain resource deployment when engaging in different cognitive and affective processes contributes to optimal performance. Moreover, there is a need to examine the link between behavioral performance during a challenging cognitive or emotion-processing task, and brain-related activation, to more conclusively determine whether differential brain processing patterns directly relate to measurable behavioral performance differences. Nevertheless, this study shows that with a relatively small group of subjects one can begin to delineate the neural circuitry that contributes to performance differences. The ultimate goal of these studies is to better understand the role of these circuits in determining performance, and then to develop more targeted training interventions that will further improve individual and team performance in extreme and complex environments.
Materials and Methods
This study was approved by the University of California San Diego (UCSD) Institutional Review Board and all subjects signed informed consent. Subjects were recruited as healthy volunteers or as comparison subjects for studies with Afghan and Iraqi war veterans as part of the research effort supported by the Center of Excellence for Stress and Mental Health (CESAMH). All subjects were interviewed with a structured diagnostic interview (SCID) , modified to enable us to document the presence of posttraumatic stress disorder. Only subjects who did not have a Diagnostic and Statistical Manual of Mental Disorders DSM-IV  diagnosis were included in this study. Thirty-four male subjects completed the study. Specifically, 11 Navy SEALs aged 26.8 years (SD = 3.7), who all had been deployed an average of 2.8 times (range 1–5) and 23 healthy male volunteers aged 24.6 years (SD = 7.4) with 13.7 (SD = 1.6) and 12.5 (SD = 0.7) years of education participated in the study. The groups did not differ in age, t(32) = 0.95, p = 0.35, but the healthy volunteers had more years of education, t(32) = 2.58, p = 0.014. Thus, all analyses were covaried for years of education. All subjects were trained to perform the emotion face-processing task prior to testing during fMRI scanning and received $50 for participation. No restrictions were placed on the consumption of caffeinated beverages; none of the subjects were smokers.
During fMRI, each subject was tested on a slightly modified  version of the emotion face-processing task , . During each 5-second trial, a subject was presented with a target face (on the top of the computer screen) and two probe faces (on the bottom of the screen) and was instructed to match the probe with the same emotional expression to the target by pressing the left or right key on a button box. A block consists of 6 consecutive trials where the target face is angry, fearful, or happy. During the sensorimotor control task, subjects were presented with 5-second trials of either vertical or horizontal ovals or circles in an analogous configuration and instructed to match the shape of the probe to the target. Each block of faces and of the sensorimotor control task was presented three times in a pseudo-randomized order. A fixation cross lasting 8 seconds was interspersed between each block presented at the beginning and end of the task (resulting in 14 fixation periods). For each trial, response accuracy and reaction time data were obtained. There were 18 trials (3 blocks of 6 trials) for each face set as well as for shapes, and the whole task lasted 512 seconds.
Acquisition of images.
All scans were performed on a 3T GE CXK4 Magnet (General Electric Medical. Systems, Milwaukee, WI) at the UCSD Keck Imaging Center, which is equipped with 8 high-bandwidth receivers that allow for shorter readout times and reduced signal distortions and ventromedial signal dropout. Each 1-hour session consisted of a 3-plane scout scan (10 seconds), a standard anatomical protocol (i.e., a sagittally acquired spoiled gradient recalled sequence) (FOV = 25 cm, matrix = 192×256, 172 sagittally acquired slices 1-mm thick, TR = 8 ms, TE = 3 ms, flip angle = 12°). We used an 8-channel brain array coil to axially acquire T2*-weighted echo-planar images (EPIs) with the following parameters: FOV = 23 cm, matrix = 64×64, 30 slices 2.6-mm thick, gap = 1.4 mm, TR = 2000 ms, TE = 32 ms, flip angle = 90°.
Image analysis pathway.
The basic structural and functional image processing were conducted with the Analysis of Functional NeuroImages (AFNI) software package . A multivariate regressor approach detailed below was used to relate changes in EPI intensity to differences in task characteristics . Echoplanar images were coregistered using a 3D-coregistration algorithm  that has been developed to minimize the amount of image translation and rotation relative to all other images. Six motion parameters were obtained for each subject. Three of these motion parameters were used as regressors to adjust for EPI intensity changes due to motion artifacts. All slices of the EPI scans were temporally aligned following registration to ensure that different relationships with the regressors were not due to the acquisition of different slices at different times during the repetition interval.
Multiple regressor analyses.
The four orthogonal regressors of interest were (1) happy, (2) angry, (3) fearful, and (4) circle/oval (i.e., shape) sensorimotor condition. These 0–1 regressors were convolved with a gamma variate function  modeling a prototypical hemodynamic response (6–8 second delay ) and to account for the temporal dynamics of the hemodynamic response (typically 12–16 seconds) . The convolved time series was normalized and used as a regressor of interest. A series of regressors of interest and the motion regressors were entered into the AFNI program 3DDeconvolve to determine the height of each regressor for each subject. The main dependent measure was the voxel-wise normalized relative signal change (or percent signal change for short), which was obtained by dividing the regressor coefficient by the zero-order regressor. Spatially smoothed (4-mm full-width half-maximum Gaussian filter) percent signal change data were transformed into Talairach coordinates based on the anatomical magnetic resonance images, which was transformed manually in AFNI.
Anatomically constrained functional regions of interest .
For the amygdala region of interest, a priori regions of interest were defined by the Talairach Daemon atlas  and functional neuroimage analyses were constrained to the a priori defined regions of interest. For the insular cortex we extended this approach to use a probability mask. Briefly, to extract a mask for the insular cortex, we used Individual Brain Atlases using Statistical Parametric Mapping software (IBASPM, http://www.thomaskoenig.ch/Lester/ibaspm.htm), a toolbox for segmenting structural MR images. All programs in this toolbox are developed in MATLAB (http://www.mathworks.com), based on a widely used neuroimaging software package, SPM (Wellcome Trust Centre for Neuroimaging, London, UK). This package uses the nonlinear registration and gray matter segmentation processes performed through SPM5 subroutines. Three principal elements for the labeling process are used: gray matter segmentation, normalization transform matrix, which maps voxels from individual space to standardized space, and MaxPro MNI Atlas. Data from a set of an existing set of 39 individuals, with similar sociodemographic characteristics as the target population, were processed using the SPM-based voxel-based morphometry approach . These data were subsequently processed using the IBASPM toolbox to obtain estimates of each individual's insula. The group insula mask was obtained by averaging across the individual insular masks and requiring that the insula voxels covered at least 50% of all subjects' gray matter.
All second-level analyses were conducted using the statistical programming language R (http://cran.r-project.org/) and with SPSS software, version 10.0 . Specifically, a mixed-model analysis was conducted with the R program lme, which is part of the nlme library. The fixed effects were emotion type, group, education, and response latency; the random effects were subjects (i.e., an individual intercept was fitted for each subject). Moreover, we conducted voxel-wise multiple linear regression analyses with performance on the emotion-processing task (latency to respond to angry, fearful, or happy faces) as independent measures, and the percent signal change between faces and the sensorimotor control condition as the dependent measure using the lm program of R.
This file provides supporting information.
(0.04 MB DOC)
A reduced linear mixed effects model focusing on anger-related processing revealed significant group differences in bilateral posterior insula.
(1.93 MB TIF)
Conceived and designed the experiments: MPP ANS EGP KFVO JB JLS. Performed the experiments: MPP ANS SNF. Analyzed the data: MPP ANS. Contributed reagents/materials/analysis tools: MPP ANS SNF EGP. Wrote the paper: MPP ANS EGP KFVO JB JLS.
- 1. Committee on Opportunities in Neuroscience for Future Army Applications Board on Army Science and Technology Division on Engineering and Physical Sciences (2009) Opportunities in neuroscience for future Army applications. Washington, DC: National Academies Press.. 136 p.
- 2. Lieberman HR, Tharion WJ, Shukitt-Hale B, Speckman KL, Tulley R (2002) Effects of caffeine, sleep loss, and stress on cognitive performance and mood during U.S. Navy SEAL training. Psychopharmacology (Berl) 164: 250–261.
- 3. Morgan CA , Aikins DE, Steffian G, Coric V, Southwick S (2007) Relation between cardiac vagal tone and performance in male military personnel exposed to high stress: three prospective studies. Psychophysiology 44: 120–127.
- 4. Oppenheimer S (2006) Cerebrogenic cardiac arrhythmias: cortical lateralization and clinical significance. Clin Auton Res 16: 6–11.
- 5. Paulus MP, Potterat EG, Taylor MK, Van Orden KF, Bauman J, et al. (2009) A neuroscience approach to optimizing brain resources for human performance in extreme environments. Neurosci Biobehav Rev 33: 1080–1088.
- 6. LaMotte RH, Thalhammer JG, Torebjork HE, Robinson CJ (1982) Peripheral neural mechanisms of cutaneous hyperalgesia following mild injury by heat. J Neurosci 2: 765–781.
- 7. Craig AD, Bushnell MC (1994) The thermal grill illusion: unmasking the burn of cold pain. Science 265: 252–255.
- 8. Schmelz M, Schmidt R, Bickel A, Handwerker HO, Torebjork HE (1997) Specific C-receptors for itch in human skin. J Neurosci 17: 8003–8008.
- 9. Lahuerta J, Bowsher D, Campbell J, Lipton S (1990) Clinical and instrumental evaluation of sensory function before and after percutaneous anterolateral cordotomy at cervical level in man. Pain 42: 23–30.
- 10. Vallbo AB, Olausson H, Wessberg J, Kakuda N (1995) Receptive field characteristics of tactile units with myelinated afferents in hairy skin of human subjects. J Physiol 483: 783–795.
- 11. Olausson H, Lamarre Y, Backlund H, Morin C, Wallin BG, et al. (2002) Unmyelinated tactile afferents signal touch and project to insular cortex. Nat Neurosci 5: 900–904.
- 12. Light AR, Perl ER (2003) Unmyelinated afferent fibers are not only for pain anymore. J Comp Neurol 461: 137–139.
- 13. Banzett RB, Mulnier HE, Murphy K, Rosen SD, Wise RJ, et al. (2000) Breathlessness in humans activates insular cortex. Neuroreport 11: 2117–2120.
- 14. Feinle C (1998) Role of intestinal chemoreception in the induction of gastrointestinal sensations. Dtsch Tierarztl Wochenschr 105: 441–444.
- 15. Robinson SK, Viirre ES, Bailey KA, Gerke MA, Harris JP, et al. (2005) Randomized placebo-controlled trial of a selective serotonin reuptake inhibitor in the treatment of nondepressed tinnitus subjects. Psychosom Med 67: 981–988.
- 16. Craig AD (2002) How do you feel? Interoception: the sense of the physiological condition of the body. Nat Rev Neurosci 3: 655–666.
- 17. Craig AD (2009) How do you feel—now? The anterior insula and human awareness. Nat Rev Neurosci 10: 59–70.
- 18. Critchley HD (2009) Psychophysiology of neural, cognitive and affective integration: fMRI and autonomic indicants. Int J Psychophysiol 73: 88–94.
- 19. Vaitl D (1996) Interoception. Biol Psychol 42: 1–27.
- 20. Davenport PW, Vovk A (2008) Cortical and subcortical central neural pathways in respiratory sensations. Respir Physiol Neurobiol.
- 21. Craig AD (2005) Forebrain emotional asymmetry: a neuroanatomical basis? Trends Cogn Sci 9: 566–571.
- 22. Taylor M, Miller A, Mills L, Potterat E, Padilla G, et al. (2006) Predictors of success in Basic Underwater Demolition/SEAL (BUD/S) training—Part I: what do we know and where do we go from here? San Diego, CA: Naval Health Research Center, Technical Report No.. pp. 06–37.
- 23. Taylor M, Larson G, Miller A, Mills L, Potterat E, et al. (2007) Predictors of success in Basic Underwater Demolition/SEAL (BUD/S) training—Part II: a mixed quantitative and qualitative study. Health Research Center Technical Report No.. pp. 07–10.
- 24. Damasio AR (1994) Descartes' error and the future of human life. Sci Am 271: 144.
- 25. Augustine JR (1996) Circuitry and functional aspects of the insular lobe in primates including humans. Brain Res Brain Res Rev 22: 229–244.
- 26. Augustine JR (1985) The insular lobe in primates including humans. Neurol Res 7: 2–10.
- 27. Craig AD (2003) A new view of pain as a homeostatic emotion. Trends Neurosci 26: 303–307.
- 28. Tracey I, Becerra L, Chang I, Breiter H, Jenkins L, et al. (2000) Noxious hot and cold stimulation produce common patterns of brain activation in humans: a functional magnetic resonance imaging study. Neurosci Lett 288: 159–162.
- 29. Critchley HD, Wiens S, Rotshtein P, Ohman A, Dolan RJ (2004) Neural systems supporting interoceptive awareness. Nat Neurosci 7: 189–195.
- 30. Phan KL, Wager T, Taylor SF, Liberzon I (2002) Functional neuroanatomy of emotion: a meta-analysis of emotion activation studies in PET and fMRI. NeuroImage 16: 331–348.
- 31. Wager TD, Phan KL, Liberzon I, Taylor SF (2003) Valence, gender, and lateralization of functional brain anatomy in emotion: a meta-analysis of findings from neuroimaging. NeuroImage 19: 513–531.
- 32. Huettel SA, Misiurek J, Jurkowski AJ, McCarthy G (2004) Dynamic and strategic aspects of executive processing. Brain Res 1000: 78–84.
- 33. Eisenberger NI, Lieberman MD, Williams KD (2003) Does rejection hurt? An FMRI study of social exclusion. Science 302: 290–292.
- 34. Simmons A, Matthews SC, Stein MB, Paulus MP (2004) Anticipation of emotionally aversive visual stimuli activates right insula. Neuroreport 15: 2261–2265.
- 35. Simmons A, Strigo I, Matthews SC, Paulus MP, Stein MB (2006) Anticipation of aversive visual stimuli is associated with increased insula activation in anxiety-prone subjects. Biol Psychiatry 60: 402–409.
- 36. Simmons A, Matthews SC, Paulus MP, Stein MB (2008) Intolerance of uncertainty correlates with insula activation during affective ambiguity. Neurosci Lett 430: 92–97.
- 37. Paulus MP, Rogalsky C, Simmons A, Feinstein JS, Stein MB (2003) Increased activation in the right insula during risk-taking decision making is related to harm avoidance and neuroticism. NeuroImage 19: 1439–1448.
- 38. Paulus MP (2007) Decision-making dysfunctions in psychiatry—altered homeostatic processing? Science 318: 602–606.
- 39. Garavan H, Ross TJ, Stein EA (1999) Right hemispheric dominance of inhibitory control: an event-related functional MRI study. Proc Natl Acad Sci U S A 96: 8301–8306.
- 40. Davidson RJ, Irwin W (1999) The functional neuroanatomy of emotion and affective style. Trends Cogn Sci 3: 11–21.
- 41. Critchley HD, Taggart P, Sutton PM, Holdright DR, Batchvarov V, et al. (2005) Mental stress and sudden cardiac death: asymmetric midbrain activity as a linking mechanism. Brain 128: 75–85.
- 42. Wager TD, Phan KL, Liberzon I, Taylor SF (2003) Valence, gender, and lateralization of functional brain anatomy in emotion: a meta-analysis of findings from neuroimaging. NeuroImage 19: 513–531.
- 43. Craig AD (2007) Interoception and emotion: a neuroanatomical perspective. In: Lewis M, Haviland-Jones JM, Feldman Barrett L, editors. Handbook of emotions. New York: Guilford Press. pp. 272–290.
- 44. Gray MA, Critchley HD (2007) Interoceptive basis to craving. Neuron 54: 183–186.
- 45. Yaguez L, Coen S, Gregory LJ, Amaro E Jr, Altman C, et al. (2005) Brain response to visceral aversive conditioning: a functional magnetic resonance imaging study. Gastroenterology 128: 1819–1829.
- 46. Shipp S (2005) The importance of being agranular: a comparative account of visual and motor cortex. Philos Trans R Soc Lond B Biol Sci 360: 797–814.
- 47. Rolls ET (2004) The functions of the orbitofrontal cortex. Brain Cogn 55: 11–29.
- 48. Kringelbach ML (2005) The human orbitofrontal cortex: linking reward to hedonic experience. Nat Rev Neurosci 6: 691–702.
- 49. Critchley HD, Tang J, Glaser D, Butterworth B, Dolan RJ (2005) Anterior cingulate activity during error and autonomic response. NeuroImage 27: 885–895.
- 50. Carter CS, Braver TS, Barch DM, Botvinick MM, Noll D, et al. (1998) Anterior cingulate cortex, error detection, and the online monitoring of performance. Science 280: 747–749.
- 51. Rushworth MF, Behrens TE (2008) Choice, uncertainty and value in prefrontal and cingulate cortex. Nat Neurosci 11: 389–397.
- 52. Goldstein RZ, Tomasi D, Rajaram S, Cottone LA, Zhang L, et al. (2007) Role of the anterior cingulate and medial orbitofrontal cortex in processing drug cues in cocaine addiction. Neuroscience 144: 1153–1159.
- 53. Jasmin L, Burkey AR, Granato A, Ohara PT (2004) Rostral agranular insular cortex and pain areas of the central nervous system: a tract-tracing study in the rat. J Comp Neurol 468: 425–440.
- 54. Reynolds SM, Zahm DS (2005) Specificity in the projections of prefrontal and insular cortex to ventral striatopallidum and the extended amygdala. J Neurosci 25: 11757–11767.
- 55. Chikama M, McFarland NR, Amaral DG, Haber SN (1997) Insular cortical projections to functional regions of the striatum correlate with cortical cytoarchitectonic organization in the primate. J Neurosci 17: 9686–9705.
- 56. Fudge JL, Breitbart MA, Danish M, Pannoni V (2005) Insular and gustatory inputs to the caudal ventral striatum in primates. J Comp Neurol 490: 101–118.
- 57. Pessiglione M, Seymour B, Flandin G, Dolan RJ, Frith CD (2006) Dopamine-dependent prediction errors underpin reward-seeking behaviour in humans. Nature 442: 1042–1045.
- 58. Preuschoff K, Quartz SR, Bossaerts P (2008) Human insula activation reflects risk prediction errors as well as risk. J Neurosci 28: 2745–2752.
- 59. Schultz W, Dickinson A (2000) Neuronal coding of prediction errors. Annu Rev Neurosci 23: 473–500.
- 60. First MB, Spitzer RL, Gibbon M, Williams JB (1995) Structured clinical interview for DSM-IV Axis I disorders–patient edition (SCID-I/P, Version 2.0). New York: New York State Psychiatric Institute, Biometrics Research Department.
- 61. American Psychiatric Association (1994) Diagnostic and statistical manual of mental disorders, Fourth Edition (DSM-IV). Washington, DC: Author.
- 62. Paulus MP, Feinstein JS, Castillo G, Simmons AN, Stein MB (2005) Dose-dependent decrease of activation in bilateral amygdala and insula by lorazepam during emotion processing. Arch Gen Psychiatry 62: 282–288.
- 63. Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, et al. (2002) Serotonin transporter genetic variation and the response of the human amygdala. Science 297: 401–403.
- 64. Hariri AR, Drabant EM, Munoz KE, Kolachana BS, Mattay VS, et al. (2005) A susceptibility gene for affective disorders and the response of the human amygdala. Arch Gen Psychiatry 62: 146–152.
- 65. Cox RW (1996) Software for analysis and visualization of functional magnetic neuroimages. Comput Biomed Res 29: 162–173.
- 66. Haxby JV, Hoffman EA, Ida Gobbini M (2000) The distributed human neural system for face perception. Trends Cogn Sci. In press.
- 67. Eddy WF, Fitzgerald M, Noll DC (1996) Improved image registration by using Fourier interpolation. Magn Reson Med 36: 923–931.
- 68. Boynton GM, Engel SA, Glover GH, Heeger DJ (1996) Linear systems analysis of functional magnetic resonance imaging in human V1. J Neurosci 16: 4207–4221.
- 69. Friston KJ, Frith CD, Turner R, Frackowiak RS (1995) Characterizing evoked hemodynamics with fMRI. NeuroImage 2: 157–165.
- 70. Cohen MS (1997) Parametric analysis of fMRI data using linear systems methods. NeuroImage 6: 93–103.
- 71. Johnstone T, Somerville LH, Alexander AL, Oakes TR, Davidson RJ, et al. (2005) Stability of amygdala BOLD response to fearful faces over multiple scan sessions. NeuroImage 25: 1112–1123.
- 72. Lancaster JL, Woldorff MG, Parsons LM, Liotti M, Freitas CS, et al. (2000) Automated Talairach atlas for functional brain mapping. Human Brain Mapping 10: 120–131.
- 73. Ashburner J, Friston KJ (2000) Voxel-based morphometry—the methods. NeuroImage 11: 805–821.
- 74. Norusis MJ (1990) SPSS base system user's guide. Chicago: SPSS Inc.