ARCHIVAL REPORTS
Volitional Control of Anterior Insula Activity Modulates the Response to Aversive Stimuli. A RealTime Functional Magnetic Resonance Imaging Study Andrea Caria, Ranganatha Sitaram, Ralf Veit, Chiara Begliomini, and Niels Birbaumer Background: A promising new approach to cognitive neuroscience based on real-time functional magnetic resonance imaging (rtfMRI) demonstrated that the learned regulation of the neurophysiological activity in circumscribed brain regions can be used as an independent variable to observe its effects on behavior. Here, for the first time, we investigated the modulatory effect of learned regulation of blood oxygenation level-dependent (BOLD) response in the left anterior insula on the perception of visual emotional stimuli. Methods: Three groups of participants (n ⫽ 27) were tested: two underwent four rtfMRI training sessions receiving either specific (n ⫽ 9) or unspecific feedback (n ⫽ 9) of the insula’s BOLD response, respectively, and one group used emotional imagery alone (n ⫽ 9) without rtfMRI feedback. During training, all groups were required to assess aversive and neutral pictures. Results: Participants able to significantly increase BOLD signal in the target region rated the aversive pictures more negatively. We measured a significant correlation between enhanced left anterior insula activity and increased negative valence ratings of the aversive stimuli. Control groups performing either rtfMRI training with unspecific feedback or an emotional imagery training alone were not able to significantly enhance activity in the left anterior insula and did not show changes in subjective emotional responses. Conclusions: This study corroborates traditional neuroimaging studies demonstrating a critical role of the anterior insula in the explicit appraisal of emotional stimuli and indicates the adopted approach as a potential tool for clinical applications in emotional disorders. Key Words: Aversive stimuli, insula, real-time fMRI, self-regulation
T
raditionally, studies in affective neuroscience investigate neural processes underlying emotions by correlating brain activity with the presentation of emotion-inducing stimuli or when participants imagined or recalled emotionally relevant scenarios. The emotional stimulus is manipulated as the independent variable and the neural response recorded as the dependent variable. This approach has generated a large amount of data showing specific relationships between brain activity and emotional behavior (1– 6). Recently, a promising noninvasive technique, real-time functional magnetic resonance imaging (rtfMRI) inspired a different approach to cognitive neuroscience and demonstrated that learned control of the local brain activity could be used as an independent variable to observe its effects on behavior, which then functions as the dependent variable (7,8). By using rtfMRI, human participants can be trained to regulate their brain activity in circumscribed regions in the presence of contingent feedback of blood oxygenation leveldependent (BOLD) response (9 –12). Recent data indicate that
From the Institute of Medical Psychology and Behavioral Neurobiology (AC, RS, RV, NB), Eberhard-Karls-University of Tübingen, Tübingen, Germany; Department of Cognitive Science (AC), University of Trento, Trento, Italy; Max Planck Institute for Biological Cybernetics (RV), Tübingen, Germany; Dipartimento di Scienze Biomediche e Terapie Avanzate, Section of Human Physiology (CB), University of Ferrara, Ferrara, Italy; and Ospedale San Camillo (NB), Istituto di Ricovero e Cura a Carattere Scientifico, Istituto di Ricovero e Cura a Carattere Scientifico, Venezia, Italy. Authors AC and RS contributed equally to this work. Address correspondence to Andrea Caria, Ph.D., Institute of Medical Psychology and Behavioural Neurobiology, Eberhard-Karls-University of Tübingen, Gartenstr. 29 D-72072, Tübingen, Germany; E-mail:
[email protected]. Received Nov 19, 2009; revised Mar 17, 2010; accepted Apr 14, 2010.
0006-3223/$36.00 doi:10.1016/j.biopsych.2010.04.020
learned modulation of neural activity leads to changes in cognitive and motor performance, pain perception, and language processing (13–16). These findings indicate that rtfMRI could complement traditional neuroimaging techniques by providing more causal insights into the functional role of circumscribed brain regions in behavior through direct manipulation of the brain system responsible for an emotional response. So far, very few studies reported an association between enhanced activity in localized brain regions and changes in emotional behavior (16,17). Moreover, our recent findings on self-regulation of BOLD activity in the anterior insula (18) prompted the question whether its learned control induces changes in the response to emotional stimuli. Several studies have gathered evidence for the involvement of the anterior insula in emotional processing (2,19 –28). Wager et al. (24) observed predominant left insula activity during tasks associated with negative or withdrawal-related emotions. Increased activity in the left insula was also associated with reported negative valence of emotional pictures (29,30). We investigated the modulatory effect of BOLD response in the left anterior insula on the perception of visual emotional stimuli. In particular, we hypothesized enhanced subjective responses to aversive pictures associated with the learned increase of insula activity through rtfMRI training. Three groups of participants were tested: two underwent four rtfMRI training sessions receiving either specific (EX group) or unspecific feedback (SH group) of the insula’s BOLD response, respectively, and one group used emotional imagery alone (MI group) without rtfMRI feedback. During training, all groups were required to assess aversive and neutral pictures. Establishing a relationship between the modulation of neural activity in anterior insula and a specific emotional behavior would not only complement the traditional neuroimaging approach but may also encourage a neurobehavioral treatment of emotional disorders. BIOL PSYCHIATRY 2010;68:425– 432 © 2010 Society of Biological Psychiatry
426 BIOL PSYCHIATRY 2010;68:425– 432
A. Caria et al.
Figure 1. Experimental design. Sessions consisted of five regulation blocks (30 sec) alternating with six baseline blocks (30 sec) both followed by an International Affective Picture System (31) picture presentation block (9 sec) and a rating block (12 sec). During rating blocks, participants were shown the Self-Assessment Manikin (34), which allows them to evaluate emotional valence and arousal. Both valence and arousal dimensions vary along a 9-point scale. Selection of the subjective rating was performed by positioning a red outline on the chosen number on each of the two scales, presented in close succession. Subjects were provided with two buttons allowing movements of the cursor in the left and right direction. IAPS, International Affective Picture System.
Methods and Materials Experimental Procedure Experimental Group. The experimental protocol consisted of four functional magnetic resonance imaging (fMRI) (Supplement 1) sessions (S1–S4) performed on 1 day. Sessions consisted of five regulation blocks alternating with six baseline blocks both followed by a picture presentation block and a rating block (Figure 1). During the regulation block (30 sec), which was indicated by a red background, participants were asked to increase insula activity, while during baseline (30 sec), indicated by a blue background, they had to return the activity to the baseline level. Subjects were guided to modulate the BOLD response in the target region of interest (ROI) (Supplement 1) by a graphical thermometer displaying changes of BOLD activity with increasing or decreasing numbers of bars (Figure 1). The number of available bars was limited. Bars above the baseline level of activation were colored in red, while those below the baseline level were colored in blue. Thermometer bars were updated every 1.5 sec when a new BOLD signal from the ROIs was available (18). Participants were informed that the colored feedback bar reflects their ongoing “brain activity” and that they should try to increase the vertical length of the red bar as much as they could during the regulation blocks. They were told that participants could use emotional imagery to drive the feedback bar but that people vary in the strategies they use: “you are free to try any other strategy to move the feedback bar except movements, tension, and facial expression changes.” After both regulation and baseline blocks, one emotionally negative or neutral picture from the International Affective Picture System (IAPS) (31–33) was presented for 9 sec and rated using a button-based control device inside the magnetic resonance imaging scanner (Figure 1). Pictures were evaluated in terms of subjective emotional valence and arousal using the Self-Assessment Manikin (SAM) (34). The Self-Assessment Manikin is a nonverbal pictorial assessment for measuring pleasure, aversion, and arousal associated with a person’s affective reaction. Valence and arousal dimensions vary along a 9-points scale (valence: from 1 ⫽ extremely negative to 9 ⫽ extremely positive; arousal: from 1 ⫽ calm to 9 ⫽ exciting). Before the experiment, participants were briefed about the experimental tasks and SAM ratings and were trained to rate the pictures. Subjects were presented with the two SAM valence (6 sec) and arousal (6 sec) scales in close succession. The final selection of each subjective rating was performed by positioning a red outline on the chosen level on the scale. Participants were provided with two buttons allowing movements of the cursor in the left and right directions. www.sobp.org/journal
The pictures presented to the participants consisted of 20 aversive and 20 neutral pictures from IAPS (Supplement 1). Picture selection was based on normative ratings (31); mean values of aversive pictures were 3.26 ⫾ .78 SD and 4.93 ⫾ .47 SD for valence and arousal, respectively. Moderately aversive/threatrelated pictures were chosen to avoid ceiling effects and allow differential ratings. Mean values of neutral pictures were 4.84 ⫾ .32 SD and 2.33 ⫾ .40 SD for valence and arousal, respectively. Pictures were pseudorandomized such that no significant difference in valence and arousal ratings was present between pictures after regulation and baseline blocks and between sessions. Additionally, presentation of the pictures was randomized across subjects. Each session lasted about 9 minutes. Control Groups. The SH group performed identical training and received the same instructions. Participants received sham feedback comparable in terms of signal magnitude and variability but not contingent on BOLD signal in the target ROI. Instead, the feedback provided information about the average activity in a large brain area not encompassing the left and right insula. The MI group was trained to assess the effects of repetitive use of emotional imagery. They were instructed to recall memories and imagery of personally relevant affective episodes during regulation blocks but no feedback about BOLD was given. The MI group was presented with the thermometer frame without rtfMRI information (no bars were shown). Apart from the missing movement of the feedback bar, conditions were identical in all three groups. Similar to the EX group, both SH and MI groups were presented with IAPS pictures following imagery blocks (30 sec) and performed SAM ratings. Data Analysis fMRI. Offline ROI and whole-brain analyses were performed using MarsBar toolbox (http://marsbar.sourceforge.net) and SPM2 (Wellcome Department of Imaging Neuroscience, London, United Kingdom), respectively. Functional echo planar imaging volumes were realigned, normalized into Montreal Neurological Institute (MNI) space, smoothed spatially (9-mm full-width halfmaximum Gaussian kernel) and temporally (cutoff period 256 sec). Hemodynamic response amplitudes were estimated using standard regressors, constructed by convolving a boxcar function representing the blocks duration with a canonical hemodynamic response function using standard SPM2 parameters. Motion parameters were also included into the general linear model as covariates to account for variance caused by head motion. As a first-level analysis, contrast images for regulation and baseline were created session by session for each subject. These images were then entered into a second-level (random effects) analysis.
A. Caria et al. The resulting statistical parametrics were thresholded at p ⬍ .001, corrected at cluster level with extent threshold of k ⫽ 10 voxels; brain regions were labeled anatomically according to TzourioMazoyer et al. (35). The averaged percentage signal change in the targeted ROI (left insula) and its homologous in the opposite hemisphere (right insula) during regulation with respect to baseline was calculated over sessions for both experimental and control groups. The influence of contingent rtfMRI feedback on learning to regulate activity in left anterior insula was compared among the groups by a repeated measures 3 ⫻ 4 analysis of variance (ANOVA) with group (EX, SH, MI) as between factor and session (S1 to S4) as within factor. Paired-samples t tests were used as post hoc tests to compare the first and the last session in each group. Additionally, a linear regression of the averaged percentage BOLD signal change over sessions was used to test for session by session increase as an index of learning. Paired t tests and linear regression were also performed on the averaged percentage signal change of the right insula to examine bilateral insula activation, which often has been reported in traditional neuroimaging studies (2,3,24,29,30). Subjective Responses and Correlation with fMRI Data. The difference in ratings of the emotional visual stimuli presented after regulation with respect to baseline were calculated individually for valence and arousal of aversive and neutral stimuli; these differences served as dependent variables and were assessed across sessions and groups. To this aim, separate 3 ⫻ 4 repeated measures ANOVAs with group (EX, SH, MI) as between factor and session (S1 to S4) as within factor, respectively, were performed. Subsequently, separate paired-samples t tests were carried out as post hoc analyses to compare the dependent variables in the first and last sessions for each group. Finally, Pearson bivariate correlation was performed to verify the association between the behavioral ratings and the averaged percentage signal change in the targeted ROI. Differences of percent BOLD signal between the regulation and baseline conditions and of the subjective ratings of the emotional pictures between the baseline and regulation conditions were considered over sessions. Statistical analysis of the percentage signal change and behavioral data analysis were accomplished using the statistical package SPSS 17.0 (SPSS, Inc., Chicago, Illinois).
Results ROI Analysis The averaged percentage signal change in the left anterior insula measured during each session in EX and the two control groups is reported in Figure 2. A significant group ⫻ session interaction effect, F (2,22) ⫽ 11.07, p ⫽ .001, emerged from the ANOVA on the percent BOLD signal change in the target ROI. Neither group nor session significant main effects were found. Paired-samples t tests highlighted a significant increase from S1 to S4 (t ⫽ ⫺3.22, df ⫽ 8, p ⫽ .012) in the EX group, no significant differences between sessions in the SH group (t ⫽ 1.89 df ⫽ 8, ns), and a significant decrease in the MI group (t ⫽ 2.62, df ⫽ 8, p ⫽ .034). Additionally, linear regression showed a progressive enhanced activity in the target area across training sessions in the EX group, indicating a learning effect (y ⫽ .28x ⫺ .29, t ⫽ 2.61, p ⫽ .014). The SH group did not show a similar learning pattern (y ⫽ ⫺.19x ⫹ .62, t ⫽ ⫺1.32, ns), whereas the MI group showed a significant decreasing activation over sessions in the target ROI (y ⫽ ⫺.20x ⫹ .77, t ⫽ ⫺3.40, p ⫽ .002). Analysis of variance of
BIOL PSYCHIATRY 2010;68:425– 432 427
Figure 2. Brain activity in the left and right anterior insula in the experimental and control groups during real-time functional magnetic resonance imaging training. Percent signal change was calculated by computing the difference of the percent BOLD signal during regulation and baseline for each participant and then averaging across all the participants. Increased percent BOLD in the target area over sessions is observed in the experimental group only. The two control groups trained with sham feedback and using mental imagery alone were not successful in learning to regulate insula activity. No percent BOLD increase was observed in the right insula in any group. *Significant changes (p ⬍ .05) in session 4 with respect to session 1. Bars represent group mean and standard error of the mean. BOLD, blood oxygenation level-dependent; EX, group receiving specific feedback; L, left; MI, group using emotional imagery alone without feedback; R, right; S1, session 1; S2, session 2; S3, session 3; S4, session 4; SH, group receiving unspecific feedback.
the right anterior insula (Figure 2) showed significant group [F (2,22) ⫽ 15.88, p ⬍ .001] and session [F (2,22) ⫽ 12.17, p ⫽ .002] main effects and group ⫻ session interaction effect [F (2,22) ⫽ 9.32, p ⫽ .001]. Paired-samples t tests between groups within first and last session revealed that whereas during S1 no differences were observed, during S4 right insula BOLD signal change in the EX group was higher than in the SH (t ⫽ 4.01, df ⫽ 16, p ⫽ .001) and the MI (t ⫽ 5.11, df ⫽ 16, p ⬍ .001) groups, which did not differ from each other. Additionally, paired-samples t tests within each group showed no significant difference in the right anterior insula between S1 and S4 for the EX group, whereas both control groups showed a significant decrease (SH: t ⫽ 2.77 df ⫽ 8, p ⫽ .040; MI: t ⫽ 3.19 df ⫽ 8, p ⫽ .015). A linear regression resulted in a no significant effect in both EX (t ⫽ .97, p ⫽ .339) and SH (t ⫽ ⫺1.95, p ⫽ .064) groups, while in the MI group, similarly as for the left insula, a significant decrease over sessions was observed (t ⫽ ⫺3.17; p ⫽ .003). Whole-Brain Analysis Whole-brain random effect analysis confirmed the results of the ROI analysis (Figure 3): training resulted in a significantly increasing activation cluster in the targeted left anterior insula over sessions (MNI ⫽ ⫺43, 16, 0; S1 and S2: no activation observed, p ⬍ .001; S3: t ⫽ 5.52, p ⬍ .001; S4: t ⫽ 7.43, p ⬍ .001); additional activations throughout the training were observed in the right insula (MNI ⫽ 43, 13, 0; t ⫽ 8.99, p ⬍ .001), the cingulate cortex (MNI ⫽ 0, ⫺13, 20; t ⫽ 7.45; MNI ⫽ 7, 26, 30; t ⫽ 7.25, p ⬍ .001), and the left dorsolateral prefrontal cortex (MNI ⫽ ⫺33, 20, www.sobp.org/journal
428 BIOL PSYCHIATRY 2010;68:425– 432
A. Caria et al.
Figure 3. Random effects analysis on the experimental group confirmed an increased blood oxygenation level-dependent magnitude in the left anterior insular cortex over time. A significant increased activation (p ⬍ .001 corrected at cluster level with extent threshold of k ⫽ 10 voxels) was observed during the third (t ⫽ 5.52) and the fourth (t ⫽ 7.43) sessions in the target area, Montreal Neurological Institute ⫽ ⫺43, 16, 0. Activation maps are projected on a single-subject T1 template at the coordinate z ⫽ 0. S1, session 1; S2, session 2; S3, session 3; S4, session 4.
10; t ⫽ 7.50, p ⬍ .001) during S1; the left middle frontal gyrus (MNI ⫽ ⫺26, 40, 20; t ⫽ 7.45, p ⬍ .001) and the right insula (MNI ⫽ ⫺26, 40, 20; t ⫽ 7.45, p ⬍ .001) during S2; the right insula (MNI ⫽ 43, 20, ⫺5; t ⫽ 7.00, p ⬍ .001) during S3; and the left superior temporal gyrus (MNI ⫽ ⫺63, ⫺40, 20; t ⫽ 7.58, p ⬍ .001) and the right insula (MNI ⫽ 46, 13, ⫺5; t ⫽ 8.13, p ⬍ .001) during S4. Separate ROI analysis of the additional active areas showed no significant training effect (Supplement 1). Behavioral Data Absolute mean ratings for all groups are reported in Figure 4. To assess changes over sessions of the emotional stimuli appraisal after regulation relative to baseline blocks, the average difference of the picture ratings (valence and arousal) between these two conditions was calculated separately for all groups. The ANOVAs revealed a significant group ⫻ session interaction effect, F (2,78) ⫽ 6.13, p ⫽ .004, for the differences in valence ratings of aversive pictures; neither group nor session significant main effects were found. Moreover, no main or interaction effects emerged for the valence of the neutral pictures and for the arousal of both aversive and neutral pictures. Paired-samples t test (two-tailed) highlighted a significant effect between S4 and S1 of the differences in valence ratings of aversive pictures (t ⫽ 2.75, p ⫽ .014) in the EX group exclusively, that is aversive stimuli were rated more negatively after regulation with respect to baseline in the last session in this group (Figure 5). Brain-Behavior Association Pursuing the relationship between brain activity and rtfMRI training associated modulation of emotional assessment and driven by the observed behavioral results, we performed a correlation analysis between the averaged percent BOLD changes in the target ROI and the differences in valence ratings. The observed significant correlation (Pearson r ⫽ .407; df ⫽ 36; p ⫽ .014 two-tailed) indicates that increased activity in the left anterior insula is associated with decreasing valence ratings—more negative— of aversive pictures (Figure 6).
Discussion The present study demonstrates that learned control of BOLD signal in the left anterior insula influences the behavioral response to emotional stimuli. Participants in the EX group, who learned to significantly increase BOLD signal in the target region, rated aversive pictures more negatively after regulation. The larger the positive difference in the BOLD activation in the anterior insula between regulation and baseline condiwww.sobp.org/journal
tions, the higher is the level of perceived negative emotion of the aversive stimuli. Conversely, the smaller or more negative this difference of the BOLD, the lower is the level of perceived negative emotion of the aversive stimuli. The SH and MI control groups, unable to successfully achieve control of the target brain region over time, showed no significant changes in the evaluation of the emotional stimuli. A reduction of BOLD response in left anterior insula over time was observed in both control groups, though a significant effect was observed in the MI group only. This effect may be due to habituation after repeated recall of emotional memories (36). Costa et al. (37) observed BOLD decrease in mesolimbic reward structures, including the amygdala, which has projection to and from insula (38,39) during unpleasant imagery and interpreted it as inhibition of appetitive circuitry. Moreover, previous studies reported a reduced BOLD response in the targeted ROI when sham or unspecific rtfMRI feedback was provided (14). Results from the control groups indicate, first, that learned self-regulation of the BOLD signal in the target region does not result from general arousal and/or global brain activation. Second, an unspecific feedback or mental imagery performed in the absence of contingent feedback is not adequate for learning to modulate the insula activity. As previously indicated (18), a combination of cognitive strategies (imagery) and real-time fMRI information drives participants to acquire successful control in a few training sessions. Whether the combination of emotional imagery and contingent feedback are sufficient to achieve voluntary control cannot be decided on the basis of these results. Nevertheless, time-contingent feedback of the physiological response (BOLD) constitutes a necessary ingredient of brain self-regulation. In the postexperimental interview, most of the participants reported to have achieved BOLD signal control in the target ROI by using mental strategies focused on recollection and reexperience of salient emotional life episodes. This is in line with the literature showing the anterior portion of the insula to be active during emotional recall and self-generated emotions and suggesting its preferential role in the evaluative, experiential, or expressive aspects of internally generated emotional and visceral responses (2,3,24,40 – 42). Training to volitionally regulate the insula leads to a decrease of the number of active clusters and to more focused brain activations. This result may represent a further indication of learning. The network involved during rtfMRI training included the dorsolateral prefrontal cortex, presumably responsible for retrieval of emotional memories (43); the cingulate cortex, implicated in emotion processing (2); the right insula as a site for
A. Caria et al.
BIOL PSYCHIATRY 2010;68:425– 432 429
the meta-representation of interoceptive states (44,45); the left inferior frontal gyrus; and the left superior temporal gyrus for language-related recall of emotional episodes (46). No significant increase over sessions was observed in these additional brain areas. Self-Regulation of Localized Brain Activity and Functional Specificity This study reinforces the rtfMRI-based approach as a useful tool to investigate brain-function relationships. Self-regulation of the anterior insula activity was associated with changes in subjective ratings of valence, but not of arousal, of emotional stimuli. Participants perceived the aversive pictures more negatively over sessions, meaning that their sensitivity increased with learned regulation. No changes in the perceived
Figure 5. The average difference of valence ratings of the aversive pictures between regulation and baseline blocks across sessions in experimental and control groups. The analyses of variance revealed a significant group ⫻ session interaction effect for the differences in valence ratings of aversive pictures presented after regulation block with respect to baseline. A significant effect between session 4 and session 1 (paired samples t test, p ⬍ .05) of the differences in valence ratings of aversive pictures was observed in the group receiving feedback only. Negative differences indicate increased negative ratings (lower values) during regulation blocks compared with baseline. Bars represent group mean and standard error of the mean. EX, group receiving specific feedback; MI, group using emotional imagery alone without feedback; S1, session 1; S4, session 4; SH, group receiving unspecific feedback.
intensity of the emotional pictures were observed over sessions, as indicated by stable arousal ratings. Though, the arousal might have been masked by an attenuation effect due to the long presentation of emotional pictures (9 sec) or to continuous emotional and cognitive effort for achieving successful regulation. In line with previous literature (29,47,48), our results indicate that different aspects of the emotional response might be mediated by different brain circuits. Emotional affect has been conceptualized along two main dimensions: valence, describing the extent of positive/pleasant/appetitive versus negative/aversive/ defensive, and arousal, which describes the intensity of activation (5,6,32,33). Neuroimaging studies (5,6,29,47) suggested a dissociation of the brain mechanisms underlying valence and arousal of emotional stimuli. However, whereas for processing
Figure 4. Self-Assessment Manikin valence (top) and arousal (bottom) ratings of the aversive pictures in the experimental and control groups in the first and in the last real-time functional magnetic resonance imaging training session. EX, group receiving specific feedback; MI, group using emotional imagery alone without feedback; S1, session 1; S4, session 4; SAM, Self-Assessment Manikin; SH, group receiving unspecific feedback.
Figure 6. A relationship between the self-regulation of the insula activity and the appraisal of the emotional stimuli was demonstrated by a significant correlation (p ⫽ .014 two-tailed) between averaged BOLD percent signal change in the target region of interest and differences in valence ratings of the aversive pictures over sessions in the group receiving feedback only. The x axis denotes the change of the average percent BOLD difference between the regulation and baseline conditions across all participants and sessions; the y axis denotes the differences in valence ratings between the baseline and regulation conditions of the aversive pictures over sessions. The larger the positive difference in the BOLD activation in the anterior insula between regulation and baseline conditions, the higher is the level of perceived negative emotion of the aversive stimuli. Conversely, the smaller or more negative this difference of the BOLD, the lower is the level of perceived negative emotion of the aversive stimuli. BOLD, blood oxygenation level-dependent.
www.sobp.org/journal
430 BIOL PSYCHIATRY 2010;68:425– 432 arousal of the emotional stimuli the amygdala seems to play a preferential role, for valence dimension, several regions including bilateral insula and mediofrontal and orbitofrontal cortex have been indicated (5,6,34,47,48). Anders et al. (29) observed that verbal reports of negative emotional valence varied with left insular activity, whereas peripheral physiologic and verbal responses, along with the arousal dimension, varied with thalamic and frontomedial activity. Moreover, Viinikainen et al. (30) showed that BOLD signal in the bilateral insula correlated positively with valence ratings of unpleasant IAPS pictures and negatively with valence ratings of pleasant pictures. On the contrary, a mutual interaction between valence and arousal has been observed. Some studies reported greater negative ratings correlating with increased arousal within structures such as the amygdala, nucleus accumbens, and fusiform and other extrastriate visual sensory areas (5,6,48). In our study, no indications about changes of the intensity dimension of emotional state of the participants, either aroused or calm, could be directly observed, as no peripheral physiological signals were measured. Future rtfMRI studies on emotional processing might clarify this point by including direct measurements of the arousal state. Interestingly, false feedback of increased heart rate—associated with changes in neural activity in the right anterior insula, bilateral mid-insula, and amygdala— has been shown to result in increases in emotional intensity ratings of neutral faces (49). However, heart rate feedback, even when false, is expected to affect emotional appraisal because of its strong sensory awareness component, while fMRI feedback acts in a more indirect and covert fashion, as subjects are not aware of brain metabolic responses. In our study, right insula activity was measured throughout the rtfMRI training but no differences in the arousal dimension were observed. Among the multitude of findings of insula involvement from the fMRI literature in emotion, empathy, and interoceptive awareness, the issue of laterality is still controversial. Craig (50) postulates a bias for the right anterior insula in negative affect and for the left insula in positive affect. In contrast, Wager et al. (24) observed predominant left insula activity associated with withdrawal and negative emotion-related beahvior. In our study, participants generated both positive and negative memories during the training, making it difficult to group them post hoc according to the valence of the imagined emotional experiences. A study disambiguating the effect of specific emotional strategies, as previously suggested (41,51), might yield insight into a possible differentiation between representations of positive and negative emotion-related bodily states in the right and left insula. Overall, our results confirm the prominent role of the anterior insula in emotional processing, specifically in the explicit appraisal of emotional and body-visceral stimuli (44,52). Previous studies demonstrated the involvement of anterior insula, along with the anterior cingulate cortex and ventral medial prefrontal cortex, in the secondary processing of emotional experience through the integration of interoceptive signals with the external context (41,52,53). On the other hand, a circuit encompassing the amygdala and the ventral striatum was implicated in primary processing of withdrawal and approach behavior (54). The anterior insula forms an explicitly accessible functional representation of interoceptive bodily states and generates regulatory signals necessary to maintain bodily homeostasis (1,25,44,45,51). It was also postulated that anterior insula activity represents a limbic substrate for the subjective awareness of emotional states (55). www.sobp.org/journal
A. Caria et al. This hypothesis is supported by studies observing increased insula activity to both unpleasant and pleasant feelings evoked by emotional stimuli (44,52). Here, the effect of the enhanced left insula activity on appraisal of negative emotional stimuli has been confirmed, though, as positive stimuli were not presented, specificity of this association has to be tested. Clinical Perspectives The rtfMRI-based approach might be relevant for clinical applications, as already indicated by deCharms et al. (14). Specifically, our study indicates a potential application of the self-regulation of insula activity for treatment of emotional disorders in light of the insula’s role in fear conditioning both in healthy subjects and patients (56 –58). Critchley et al. (56) reported a modulation of the insula activity by conscious awareness of paired conditioned stimuli during fear conditioning. The insula seems to be crucial in anticipatory anxiety induced by exposure to aversive stimuli (20,59). A recent meta-analysis of neuroimaging studies suggested a common exaggerated engagement of fear circuitry, including amygdala and insula hyperactivity, in anxiety disorders (60). Studies on anxiety-prone subjects reported increased amygdala and insula activation during both anticipation and observation of aversive pictures (61,62). It has been hypothesized that insula hyperactivity might be a common feature in persons with elevated trait anxiety and therefore might represent a marker for anxiety proneness (63). Finally, the anterior insula’s hypoactivity has been reported to be associated with lack of aversive anticipatory arousal in psychopathy (57,58), whereas its hyperactivity was observed in social phobia (57). Together, these findings indicate the insula as a candidate region to develop rtfMRI-based treatment for patients suffering from emotional disorders.
This work was supported by the Deutsche Forschungsgemeinschaft, the European Union Marie Curie Host Fellowship for Early Stage Researchers Training, Bundesministerium für Bildung und Forschung Bernstein project number 01GQ0761 and grants from the Deutsche Forschungsgemeinschaft DFG BI 195/59-1 and DFG BI 195/56-1. AC, RS, RV, and NB designed research; AC, RS, and CB performed experiments; RS programmed the real-time functional magnetic resonance imaging system; AC analyzed data; AC, RS, RV, and NB wrote the manuscript. We are grateful to Simona de Falco for helpful discussions and comments on the manuscript. The authors report no biomedical financial interests or potential conflicts of interest. Supplementary material cited in this article is available online. 1. Davidson RJ, Irwin W (1999): The functional neuroanatomy of emotion and affective style. Trends Cogn Sci 3:11–21. 2. Phan KL, Wager TD, Taylor SF, Liberzon I (2002): Functional neuroanatomy of emotion: A meta-analysis of emotion activation studies in PET and fMRI. Neuroimage 16:331–348. 3. Murphy FC, Nimmo-Smith I, Lawrence AD (2003): Functional neuroanatomy of emotion: A meta-analysis. Cogn Affect Behav Neurosci 3:207–233. 4. Feldmann-Barrett L, Wager TD (2006): The structure of emotion: Evidence from neuroimaging studies. Curr Dir Psychol Sci 15:79 – 83.
A. Caria et al. 5. Lang PJ, Davis M (2006): Emotion, motivation, and the brain: Reflex foundations in animal and human research. Prog Brain Res 156:3–29. 6. Lang PJ, Bradley PM (2009): Emotion and the motivational brain [published online ahead of print October 30]. Biol Psychol. 7. Sitaram R, Caria A, Birbaumer N (2009): Hemodynamic brain-computer interfaces for communication and rehabilitation. Neural Netw 22:1320 – 1328. 8. deCharms RC (2008): Applications of real-time fMRI. Nat Rev Neurosci 9:720 –729. 9. Weiskopf N, Mathiak K, Bock SW, Scharnowski F, Veit R, Grodd W, et al. (2004): Principles of a brain-computer interface (BCI) based on real-time functional magnetic resonance imaging (fMRI). IEEE Trans Biomed Eng 51:966 –970. 10. Weiskopf N, Scharnowski F, Veit R, Goebel R, Birbaumer N, Mathiak K (2004): Self-regulation of local brain activity using real-time functional magnetic resonance imaging (fMRI). J Physiol Paris 98:357–373. 11. Sitaram R, Caria A, Veit R, Gaber T, Rota G, Kuebler A, Birbaumer N (2007): FMRI brain-computer interface: A tool for neuroscientific research and treatment. Comput Intell Neurosci 2007:25487. 12. deCharms RC (2007): Reading and controlling human brain activation using real-time functional magnetic resonance imaging. Trends Cogn Sci 11:473– 481. 13. Weiskopf N, Veit R, Erb M, Mathiak K, Grodd W, Goebel R, Birbaumer N (2003): Physiological self-regulation of regional brain activity using realtime functional magnetic resonance imaging (fMRI): Methodology and exemplary data. Neuroimage 19:577–586. 14. deCharms RC, Maeda F, Glover GH, Ludlow D, Pauly JM, Soneji D, et al. (2005): Control over brain activation and pain learned by using real-time functional MRI. Proc Natl Acad Sci U S A 102:18626 –18631. 15. Bray S, Shimojo S, O’Doherty JP (2007): Direct instrumental conditioning of neural activity using functional magnetic resonance imaging-derived reward feedback. J Neurosci 27:7498 –7507. 16. Rota G, Sitaram R, Veit R, Erb M, Weiskopf N, Dogil G, Birbaumer N (2009): Self-regulation of regional cortical activity using real-time fMRI: The right inferior frontal gyrus and linguistic processing. Hum Brain Mapp 30:1605–1614. 17. Posse S, Fitzgerald D, Gao K, Habel U, Rosenberg D, Moore GJ, Schneider F (2003): Real-time fMRI of temporolimbic regions detects amygdala activation during single-trial self-induced sadness. Neuroimage 18:760 – 768. 18. Caria A, Veit R, Sitaram R, Lotze M, Weiskopf N, Grodd W, Birbaumer N (2007): Regulation of anterior insular cortex activity using real-time fMRI. Neuroimage 35:1238 –1246. 19. Phillips ML, Young AW, Senior C, Brammer M, Andrew C, Calder AJ, et al. (1997): A specific neural substrate for perceiving facial expressions of disgust. Nature 389:495– 498. 20. Chua P, Krans M, Toni I, Passingham R, Dolan R (1999): A functional anatomy of anticipatory anxiety. Neuroimage 9:563–571. 21. Schienle A, Stark R, Walter B, Blecker C, Ott U, Kirsch P, et al. (2002): The insula is not specifically involved in disgust processing: An fMRI study. Neuroreport 13:2023–2026. 22. Wicker B, Keysers C, Plailly J, Royet JP, Gallese V, Rizzolatti G (2003): Both of us disgusted in my insula: The common neural basis of seeing and feeling disgust. Neuron 40:655– 664. 23. Krolak-Salmon P, Hénaff MA, Isnard J, Tallon-Baudry C, Guénot M, Vighetto A, et al. (2003): An attention modulated response to disgust in human ventral anterior insula. Ann Neurol 53:446 – 453. 24. 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. 25. Critchley HD, Wiens S, Rotshtein P, Ohman A, Dolan RJ (2004): Neural systems supporting interoceptive awareness. Nat Neurosci 7:189 –195. 26. Singer T, Seymour B, O’Doherty J, Kaube H, Dolan RJ, Frith CD (2004): Empathy for pain involves the affective but not sensory components of pain. Science 303:1157–1162. 27. Wright P, He G, Shapira NA, Goodman WK, Liu Y (2004): Disgust and the insula: fMRI responses to pictures of mutilation and contamination. Neuroreport 15:2347–2351. 28. Jackson PL, Meltzoff AN, Decety J (2005): How do we perceive the pain of others: A window into the neural processes involved in empathy. Neuroimage 24:771–779.
BIOL PSYCHIATRY 2010;68:425– 432 431 29. Anders S, Lotze M, Erb M, Grodd W, Birbaumer N (2004): Brain activity underlying emotional valence and arousal: A response-related fMRI study. Hum Brain Mapp 23:200 –209. 30. Viinikainen M, Jääskeläinen IP, Alexandrov Y, Balk MH, Autti T, Sams M (2009): Nonlinear relationship between emotional valence and brain activity: Evidence of separate negative and positive valence dimensions [published online ahead of print December 2]. Hum Brain Mapp. 31. Lang PJ, Bradley MM, Cuthbert BN (2008): International Affective Picture System (IAPS): Affective Ratings of Pictures and Instruction Manual. Technical Report A-7. Gainesville, FL: University of Florida. 32. Bradley MM, Lang PJ (2007): Emotion and motivation. In: Cacioppo JT, Tassinary LG, Berntson G, editors. Handbook of Psychophysiology, 2nd ed. New York: Cambridge University Press, 581– 607. 33. Bradley MM, Lang PJ (2007): The International Affective Picture System (IAPS) in the study of emotion and attention. In: Coan JA, Allen JJB, editors. Handbook of Emotion Elicitation and Assessment. Oxford, UK: Oxford University Press, 29 – 46. 34. Bradley MM, Lang PJ (1994): Measuring emotion: The self-assessment manikin and the semantic differential. J Behav Ther Exp Psychiatry 25:49 –59. 35. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, et al. (2002): Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI singlesubject brain. Neuroimage 15:273–289. 36. Schienle A, Schäfer A (2009): In search of specificity: Functional MRI in the study of emotional experience. Int J Psychophysiol 73:22–26. 37. Costa VD, Lang PJ, Sabatinelli D, Bradley MM, Versace F (2010): Emotional imagery: Assessing pleasure and arousal in the brain’s appetitive circuitry [published online ahead of print February 2]. Hum Brain Mapp. 38. Augustine JR (1985): The insular lobe in primates including humans. Neurol Res 7:2–10. 39. Augustine JR (1996): Circuitry and functional aspects of the insular lobe in primates including humans. Brain Res Rev 22:229 –244. 40. Reiman EM, Lane RD, Ahern GL, Schwartz GE, Davidson RJ, Friston KJ, et al. (1997): Neuroanatomical correlates of externally and internally generated human emotion. Am J Psychiatry 154:918 –925. 41. Damasio AR, Grabowski TJ, Bechara A, Damasio H, Ponto LL, Parvizi J, Hichwa RD (2000): Subcortical and cortical brain activity during the feeling of self-generated emotions. Nat Neurosci 3:1049 –1056. 42. Wager TD, Feldman BL (2004): From affect to control: Functional specialization of the insula in motivation and regulation. Available at: http:// www.columbia.edu/cu/psychology/tor/Papers/Wager_Feldman_ Barrett_2004_Insula_meta-analysis.pdf. Accessed May 20, 2010. 43. Ochsner KN, Gross JJ (2005): The cognitive control of emotion. Trends Cogn Sci 9:242–249. 44. Craig AD (2002): How do you feel? Interoception: The sense of the physiological condition of the body. Nat Rev Neurosci 3:655– 666. 45. Craig AD (2003): Interoception: The sense of the physiological condition of the body. Curr Opin Neurobiol 13:500 –505. 46. Burianova H, Grady CL (2007): Common and unique neural activations in autobiographical, episodic, and semantic retrieval. J Cogn Neurosci 19:1520 –1534. 47. Cunningham WA, Raye CL, Johnson MK (2004): Implicit and explicit evaluation: fMRI correlates of valence, emotional intensity, and control in the processing of attitudes. J Cogn Neurosci 16:1717–1729. 48. Lewis PA, Critchley HD, Rotshtein P, Dolan RJ (2007): Neural correlates of processing valence and arousal in affective words. Cereb Cortex 17:742– 748. 49. Gray MA, Harrison NA, Wiens S, Critchley HD (2007): Modulation of emotional appraisal by false physiological feedback during fMRI. PLoS ONE 2:e546. 50. Craig AD (2009): How do you feel—now? The anterior insula and human awareness. Nat Rev Neurosci 10:59 –70. 51. Craig AD (2005): Forebrain emotional asymmetry: A neuroanatomical basis? Trends Cogn Sci 9:566 –571. 52. Phan KL, Taylor SF, Welsh RC, Ho S, Britton JC, Liberzon I (2004): Neural correlates of individual ratings of emotional salience: A trial-related fMRI study. Neuroimage 21:768 –780. 53. Reiman EM, Fusselman MJ, Fox PT, Raichle ME (1989): Neuroanatomical correlates of anticipatory anxiety. Science 243:1071–1074.
www.sobp.org/journal
432 BIOL PSYCHIATRY 2010;68:425– 432 54. Bechara A, Damasio H, Damasio AR, Lee GP (1999): Different contributions of the human amygdala and ventromedial prefrontal cortex to decision-making. J Neurosci 19:5473–5481. 55. Damasio AR (1999): The Feeling of What Happens: Body and Emotion in the Making of Consciousness New York: Harcourt Brace. 56. Critchley HD, Mathias CJ, Dolan RJ (2002): Fear-conditioning in humans: The influence of awareness and arousal on functional neuroanatomy. Neuron 33:653– 663. 57. Veit R, Flor H, Erb M, Hermann C, Lotze M, Grodd W, Birbaumer N (2002): Brain circuits involved in emotional learning in antisocial behavior and social phobia in humans. Neurosci Lett 328:233–236. 58. Birbaumer N, Veit R, Lotze M, Erb M, Hermann C, Grodd W, Flor H (2005): Deficient fear conditioning in psychopathy. A functional magnetic resonance imaging study. Arch Gen Psychiatry 62:799 – 805.
www.sobp.org/journal
A. Caria et al. 59. Nitschke JB, Sarinopoulos I, Mackiewicz KL, Schaefer HS, Davidson RJ (2006): Functional neuroanatomy of aversion and its anticipation. Neuroimage 29:106 –116. 60. Etkin A, Wager TD (2007): Functional neuroimaging of anxiety: A metaanalysis of emotional processing in PTSD, social anxiety disorder, and specific phobia. Am J Psychiatry 164:1476 –1488. 61. 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. 62. Stein MB, Simmons A, Feinstein JS, Paulus MP (2007): Increased amygdala and insula activation during emotion processing in anxiety prone subjects. Am J Psychiatry 164:318 –327. 63. Paulus MP, Stein MB (2006): An insular view of anxiety. Biol Psychiatry 60:383–387.
