You are on page 1of 13


com/locate/ynimg NeuroImage 28 (2005) 835 – 847

Attachment-style differences in the ability to suppress negative thoughts: Exploring the neural correlatesi
Omri Gillath,a,* Silvia A. Bunge,a Phillip R. Shaver,a Carter Wendelken,a and Mario Mikulincer b
a b

Department of Psychology, University of California, Davis 95616, USA Department of Psychology, Bar-Ilan University, Ramat Gan 52900, Israel

Received 1 December 2004; revised 10 June 2005; accepted 21 June 2005 Available online 8 August 2005

Beginning in infancy, people can be characterized in terms of two dimensions of attachment insecurity: attachment anxiety (i.e., fear of rejection and abandonment) and attachment avoidance (distancing oneself from close others, shunning dependency; Bowlby, J., 1969/1982. Attachment and loss: Vol. 1. Attachment, 2nd ed., Basic Books, New York). The capacity for emotion regulation varies with attachment style, such that attachment-anxious individuals become highly emotional when threatened with social rejection or relationship loss, whereas avoidant individuals tend to distance themselves or disengage from emotional situations. In the present study, 20 women participated in an fMRI experiment in which they thought about—and were asked to stop thinking about—various relationship scenarios. When they thought about negative ones (conflict, breakup, death of partner), their level of attachment anxiety was positively correlated with activation in emotion-related areas of the brain (e.g., the anterior temporal pole, implicated in sadness) and inversely correlated with activation in a region associated with emotion regulation (orbitofrontal cortex). This suggests that anxious people react more strongly than non-anxious people to thoughts of loss while under-recruiting brain regions normally used to down-regulate negative emotions. Participants high on avoidance failed to show as much deactivation as less avoidant participants in two brain regions (subcallosal cingulate cortex; lateral prefrontal cortex). This suggests that the avoidant peoples’ suppression was less complete or less efficient, in line with results from previous behavioral experiments. These are among the first findings to identify some of the neural processes underlying adult attachment orientations and emotion regulation. D 2005 Elsevier Inc. All rights reserved.

Preparation of this article was facilitated by a grant from the UC Davis Imaging Center. The authors would like to thank Sarah Donohue and Kateri Evans for assistance with data collection and analysis, Shelley Blozis for statistical advice, and Matthew Lieberman for very helpful editorial comments. * Corresponding author. Fax: +1 530 752 2087. E-mail address: (O. Gillath). Available online on ScienceDirect ( 1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2005.06.048


People differ in the ways they form emotional bonds and regulate emotions in close relationships, and the differences have been extensively studied using behavioral methods with people of various ages from birth through early adulthood. We still know relatively little, however, about the neural underpinnings of individual differences in what researchers call ‘‘attachment style’’ (e.g., Shaver and Clark, 1994). The present article reviews some of the behavioral, psychophysiological, and neuroscientific data related to attachment and then describes an initial attempt to use fMRI methods to examine the neural basis of emotion regulation as it relates to attachment style. Attachment theory (Ainsworth et al., 1978; Bowlby, 1969/ 1982, 1973, 1980) concerns the formation of emotional bonds between people and the effects of a person’s attachment history on emotion regulation and other aspects of personality. According to Bowlby (1969/1982), proximity-seeking behavior, beginning early in infancy, is regulated by an innate attachment behavioral system , the function of which is to obtain protection and care from another person, an attachment figure. The system becomes adapted to characteristics of key attachment figures over the life course, and the resulting attachment style (e.g., secure, anxious, or avoidant) becomes relatively stable and can be measured by self-report questionnaires and structured interviews (see Mikulincer and Shaver, 2003 for a review of behavioral studies of attachment processes). These styles of attachment then affect numerous features of adults’ close relationships and emotional experiences. Attachment styles can be assessed in terms of two dimensions of insecurity: attachment-related anxiety and attachment-related avoidance (e.g., Ainsworth et al., 1978, Fig. 10; Brennan et al., 1998; Fraley and Spieker, 2003). People are roughly normally distributed within the space formed by these two orthogonal dimensions. Since 1987, hundreds of studies have shown that a person’s attachment style is a powerful predictor of various psychological and social – relational phenomena including selfand social schemas, self-regulation of stress and emotion, the quality of relations with romantic or marital partners, sexual motivation, and reactions to relationship breakups or losses (see Mikulincer and Shaver, 2003; Shaver and Clark, 1994 for reviews).

1998). some attention has been paid to psychophysiological processes as well (see Diamond. Cohen and Shaver (2004) found that avoidant individuals made more errors when judging positive attachmentrelated words presented to the right hemisphere. 2004). Dawson et al. 2004). Hofer. Drevets. orbital frontal cortex (OFC). Most such studies. 2005). For example. These are some of the areas in which we expected to see ‘‘don’t think vs. 2000).. we expected anxious individuals to exhibit greater activation in brain regions associated with emotions and less activation in emotion-regulation areas. 2002.. as if their suppression process was not reliable or complete.. 1969/ 1982). see Phan et al. ventromedial prefrontal cortex (VMPFC). 1999). In contrast. Feeney and Kirkpatrick. to vividly imagine that their partner was leaving them for someone else. for further discussion of hyperactivation and deactivation.) These two coping strategies and their effects on mood and behavior have been documented in both cross-sectional and longitudinal studies. In those experiments. Mikulincer et al. their pattern of brain activation and deactivation would be different from the pattern exhibited by non-avoidant (more secure) individuals. insula. when avoidant individuals were asked not to think about something. Rolls et al. 2004). Mikulincer. and more specifically the dorsolateral prefrontal cortex (DLPFC). first. (See Cassidy and Kobak. 2004. Subsequent experiments showed that avoidant individuals’ ability to down-regulate their emotion in this situation could be disrupted by adding a cognitive load (remembering a 7digit number). In preliminary work on adult emotion regulation (approach and withdrawal tendencies) and brain laterality. and the different regions have different functions related to experiencing emotion or evaluating emotional states. Recently. Insecure individuals (those scoring relatively high on either the anxiety or the avoidance dimension. Some of the brain regions most consistently associated with emotional arousal are the amygdala. and other attachment-related cognitive and emotional processing differences. Wyland et al.. behavioral. Hyperactivation includes a high degree of emotionality and hypersensitivity to signs of acceptance or rejection. 1997. anterior temporal pole (ATP)... Carpenter and Kirkpatrick. Schore. After imagining this stressful experience for a few minutes. the few studies that have examined brain correlates of attachment have focused on laterality differences. A mechanistic account of these laterality differences. 2003.. These findings suggest that suppression is somewhat different for avoidant and non-avoidant individuals (Mikulincer et al. 1997. either by seeking support from trusted attachment figures or by calling upon mental representations of support received in the past (Mikulincer and Shaver. but not non-avoidant... 2001. people who were involved in longterm romantic or marital relationships were asked. 2003. 1997). . (2001). whereas those labeled ‘‘avoidant’’ deactivate the system. Eugene et al. on average. however. or both).g.. / NeuroImage 28 (2005) 835 – 847 Secure individuals (those scoring relatively low on both the anxiety and avoidance dimensions) cope well with stress. found that insecurely attached infants exhibited relatively reduced left frontal brain activity. a cognitive load did not affect secure individuals’ down-regulation of their emotions. Gillath et al.. Mikulincer et al. however. people experience a rebound of suppressed material when a cognitive load is added. using EEG. Mikulincer and Shaver. Turning to the dimension of attachment anxiety. Deactivation includes suppression or down-regulation of emotion and is accomplished by a combination of conscious and unconscious processes.. Shin et al. Ochsner et al. In the study reported here. We expected that. Individuals labeled ‘‘anxious’’ by attachment researchers hyperactivate their ‘‘attachment behavioral system’’ (Bowlby. there has been a call for more direct exploration of the neurophysiological foundations of attachment-related behavior (Diamond. Several brain regions have been associated in previous studies with regulation and suppression of thoughts and emotions. 2001. To date. Hare et al. Not all of these regions are associated with every emotion. rostral anterior cingulate cortex (rACC).g. Mikulincer et al. 1988.g. 2003).1 We expected attachment anxiety to be positively correlated with activation in the following regions associated with negative 1 Many researchers associate the ATP with semantic processing. Results indicated that. and the subcallosal cingulate cortex (SCC) (e. Fraley and Shaver... 2000. In the present study we examined ATP activation in relation to sad thoughts and memories connected with relationship breakups and losses. 1996. 2003) and the retrieval of emotional memories (Dolan et al. have relied on surface indicators of autonomic arousal rather than using brain-imaging techniques (e.g. Therefore. Although most studies of adult attachment rely on self-report and observational methods.. whereas the ATP has been associated with sadness (e. are forced to decide between two alternative coping strategies. 2002. When asked to stop thinking about these negative scenarios. 2001 for a review). lateral prefrontal cortex (LPFC). For example.. participants were asked to stop thinking about it..g. These findings allowed us to make preliminary predictions to be tested in an fMRI experiment focused on the ability to cope with attachment-related thoughts and emotions. 1997. That is. do attachment-anxious individuals activate emotions more intensely? Why do they find it difficult to regulate them? Do avoidant individuals use a particular kind of mental suppression to keep attachment-related negative emotions out of awareness? The latter question is particularly interesting because several recent studies suggest that avoidant suppression strategies are not as complete or reliable as more secure approaches to emotion regulation. compared to individuals who score low on the anxiety dimension. dorsal anterior cingulate cortex (dACC). The nature of this difference might help to explain why avoidant.836 O.. 2004). who do not feel as confident about the availability and responsiveness of others. 2004). Phan et al. 2002. avoidant and secure individuals were able to down-regulate their negative emotions (as indicated by a reduction in skin conductance). Beauregard et al. Le ´ vesque et al. They also find it difficult to suppress negative thoughts and emotions even when asked to do so (Fraley and Shaver. Fraley and Shaver. a growing number of studies implicate this region in the experience of sadness.g... Examination of the neural mechanisms of attachment would help to answer questions that cannot be fully answered using cognitive. participants exhibited increased autonomic arousal (assessed with skin conductance measures) while thinking about the breakup and loss. but anxious individuals were not (Fraley and Shaver. avoidant suppression is more subject to disruption by competing mental processes (e.. However. the amygdala is strongly associated with calculating emotional significance (e. Anderson et al. These regions include the anterior cingulate cortex (ACC). we used functional magnetic resonance imaging (fMRI) to explore the neural underpinnings of emotion-regulation processes identified in previous behavioral experiments (e. 2003. 2003. will require in-depth characterization of attachment-style differences in brain function. we know that attachment-anxious individuals become more emotional than their less anxious counterparts when thinking about breakups and losses. 1996. or surface physiological measures. 2004). 2000. and ventral striatum (for reviews. think’’ differences as well as differences between avoidant and non-avoidant people during thought and emotion suppression..

1998.g. each of which included four task epochs (blocks)..03. Le ´ vesque et al. Rule et al. 4.g. (2) the ATP. 1998).64.69. Cronbach alpha for the 8-item scale was high (0. 1.00. and (3) the ACC (both its rostral and dorsal subdivisions). 1. Participants rated the extent to which an item was self-descriptive on a 5-point scale ranging from ‘‘disagree strongly’’ (1) to ‘‘agree strongly’’ (5).. All gave written informed consent following instructions approved by the university’s Institutional Review Board.. 1. Calarge et al.45. Eugene et al. Phan et al. 2004) reported a positive correlation between regulation of sadness and OFC activation in both women and young girls. particularly in relation to emotional and social stimuli (e. low scores on the two dimensions (indicating a relatively secure attachment style). several weeks before the experiment took place. 1. the two estimates were averaged. 3. 2003.19. 2004). The rACC is known to be engaged during emotion processing (e. 3. 5. especially in a social context (e. ‘‘I prefer not to show a partner how I feel deep down’’). through its anatomic projections to the limbic and paralimbic regions (specifically the ATP). participants completed the Experiences in Close Relationships (ECR) scale (Brennan et al.00. participants again completed the ECR scale.. To maximize the reliability of participants’ attachment scores. 2. 4.50. 2003).g. right-handed. Beer et al. we expected attachment anxiety to be negatively correlated with activation in the OFC. 2003. 1.. The OFC has been implicated in thought regulation. 1998). We therefore expected to find that attachment-anxious people would have higher activation in the ATP and lower activation in the OFC when thinking about relationship losses. 2002).g. respectively). 3. but also from failure to regulate it. 5.39. Benet-Martinez and John.03. Thus. 1. 1. Finally. is involved in behavioral inhibition and voluntary suppression of emotion (see also Cavada et al. 0. After the experimental and anatomical scans. 2. 2003). None had a prior history of neurological or psychiatric disorders.92..g. 2000).66. The structure of a typical condition in the upcoming experiment (control block. 2001.81. 4. In the current study.36.. they tend to activate many other associated negative memories (Mikulincer and Orbach. 1.. (2003. 2000.g. Miller and Cohen.. thinking block. which would involve greater hippocampal activation.11. Ranote et al.94. which is engaged when sadness is experienced (e.O. These authors suggested that the OFC.g. whereas others recruited OFC and medial PFC.39. and neuroticism (Eisenberger et al. 1.g. Maguire and Frith. 2 . Eighteen items on the scale measured attachment anxiety (e. 2003).22... who tend to recall previous negative experiences when thinking about a current negative event.. Davis. when asked to recall one negative memory. Therefore.28. 2.. would retrieve more memories while engaged in our task. there is considerable evidence indicating that attachment-anxious people tend to ruminate on negative thoughts and feelings and become overwhelmed by distressing memories. Gillath et al. 3. 2003 for a review).64. 1. with a mean age of 20. The correlations between the pre-study and post-study ECR scores were high (0. Beauregard et al. / NeuroImage 28 (2005) 835 – 847 837 emotionality: (1) the amygdala.89. 2.. Neuroticism and general anxiety (i. and the dACC has been associated with subjective distress caused by physical pain (Rainville et al.25. on which participants rated statements on a 4-point Likert scale ranging from ‘‘almost never’’ to ‘‘almost always.80. Their ages ranged from 18 to 25. 5. 1.. 5.92. 2004. Each participant had either (1) a relatively high score on one of the two attachment dimensions (anxiety. 2004). The hippocampus is important for the retrieval of both emotional and non-emotional memories. Trait anxiety was assessed using Spielberger et al. 4. 1997). not only from heightened emotionality per se.. as well as questionnaires relating to demographics and their relationship history.22. participants were given a sample task in which they read a hypothetical scenario similar to what they would see during the experiment and were guided through the instructions and the meanings of the different symbols (traffic lights. For example.. 2.. as explained below).2 Participants were healthy.05. free-thinking block) was explained and demonstrated using the hypothetical scenario.00. Questionnaire measures In a separate session..70. ‘‘I worry about being abandoned’’) and 18 items measured avoidance (e. 4.08. a brain structure known to be important for associative memory retrieval (e. 2. 2. suppressing block. Adolphs.. 1. 2003.. 3. Participants rated the extent to which each item was descriptive of their experiences in close relationships on a 7-point scale ranging from ‘‘not at all’’ (1) to ‘‘very much’’ (7).95. 1999.’s (1983) 20-item Trait Anxiety Inventory (STAI). Cronbach alpha for the STAI was 0. 2... 1995). The task design The scores on attachment anxiety and avoidance (means on 7-point scales) for each participant were as follows: 6. 2000. Brennan et al..03.’’ with higher scores indicating greater trait anxiety. Mesulam and Mufson. Le ´ vesque et al. in press). Cohen et al.. Behavioral tasks Materials and methods Participants Participants were 20 women students from the subject pool at the University of California. selected on the basis of their attachment scores. 1982). 1. Bush et al... trait anxiety) were also assessed because these personality traits are known to be correlated with attachment anxiety and might be legitimately viewed as offering alternative explanations of our findings (see Mikulincer and Shaver. avoidance) and a score close to the median on the other or (2) Participants underwent five MRI scans (9 min. 2003).83.89. Pelletier et al.g.92 and 0.72. 2004). we were interested in exploring the possibility that attachment anxiety would correlate with activation in the hippocampus.31. Mikulincer and Florian. something not observed in secure or avoidant individuals.87). fluent English speakers. 2. Our prediction was that anxious people. Immediately before the fMRI experiment.55. 2000. 1. Neuroticism was assessed with the 8-item neuroticism subscale from the Big Five Inventory (BFI. social rejection (Eisenberger et al..83 for anxiety.e.06. Individual differences related to activation of the ATP and the OFC have been demonstrated by Eugene et al. 1999. The heightened emotionality of attachment-anxious individuals might result. Some study participants recruited the ATP and the insula (regions associated with emotional expression) while watching film clips about bereavement. Blair et al. including autobiographical memories (e. Roberts and Wallis. 2. Cronbach alpha coefficients were high for the anxiety and avoidance scales (0. (2003) in a study examining the voluntary expression of sadness. 43 s apiece). which is engaged when fear and anxiety are experienced (e. The reliability and validity of the scales have been repeatedly demonstrated (e. 2001.82 for avoidance).53. Eichenbaum. 2.g.78.58.92..

neutral but relationship-related. Task-related responses were generally considered significant if they consisted of at least 5 contiguous voxels that exceeded an uncorrected threshold of P < 0. and the resulting contrast images. 1) instructed participants to think for 2 min about a specific scenario and press a response button whenever they had a new thought or image related to the scenario (henceforth referred to as the Think block). they pressed the button whenever they happened to think about the forbidden topic. as well as a covariate for session effects.838 O. The fifth scenario involved imagining that their partner died. A green traffic light with the number F2_ superimposed on it instructed participants to think about whatever came to mind and press the response button if they happened to think about the previously excluded topic. hippocampus. Order in which the blocks were presented. or emotionally negative and relationship-related. a number of ROIs were examined (9 total). such as shopping in a mall or supermarket. and were then submitted to rigid body motion correction. At the end of each scan. In the Don’t Think block. voxel dimensions = 1 Â 1 Â 1 mm).5. Gillath et al. Natick. just before the first Think block participants read the following instructions: ‘‘In the next 2 min (during which you will see a green traffic light with no number in it). respectively. The goal of this exploratory study was to identify candidate regions that might be differentially engaged by individuals who differed in attachment style. and SCC. and resampled the volumes to 2 Â 2 Â 2 mm cubic voxels. In the FreeThinking The instructions for each block were presented for 20 s followed by a traffic light for 120 s. image. Participants viewed the stimuli on a screen via a mirror in the head coil of the scanner. when exploring details of individual differences. They then rested and completed a series of questionnaires and were debriefed and asked about their thoughts and methods of suppression. we used anatomical ROIs provided by MarsBar for regions that were of interest but not identified in the whole-brain contrasts: OFC (BA 11). Additionally. along with a basis set of cosine functions that high-pass filtered the data. TE = 4. although in a few cases. the ATP. Templates are based on the MNI305 stereotaxic space (Cocosco et al. 1997). London..5 s. Region-of-interest (ROI) analyses were performed with the MarsBar toolbox in SPM2 (Brett et al. sourceforge. Head motion was minimized with comfortable padding around the participant’s head.. Data were preprocessed with SPM2 (Wellcome Department of Cognitive Neurology. high-resolution anatomical images were collected for each participant (TR = 12 ms. and the amygdala. UK) on the Matlab 6. unpublished manuscript). MA). Mean contrast values for each participant and condition were extracted for each ROI and submitted to a repeated measures analysis and post-hoc comparisons. participants pressed a button every time the yellow light appeared. The data were high-pass filtered with the maximal frequency of task alternation (280 s) to cut off baseline drifts and low-pass filtered with a hemodynamic response function to control for temporal autocorrelation. In the Think block. they pressed the button whenever they happened to think about the formerly forbidden topic. In the control block.. The remaining 224 volumes were used for the subsequent analyses. 1.1 platform (Mathworks Inc. Functional images were acquired with gradient-recalled echo EPI sensitive to the blood oxygen level dependent (BOLD) contrast (TR = 2. Functional volumes were spatially smoothed with an 8-mm FWHM isotropic Gaussian kernel. scene. treating participants as a random effect. P < 0. 24 contiguous 4 mm oblique axial slices parallel to the AC – PC line). Results for the Think and Don’t Think blocks are reported here. Finally. We used functional ROIs for targeted regions identified in contrasts of interest: namely. and the Fig. please imagine as vividly as you can that you are driving alone to Lake Tahoe. was based on a prior study of thought suppression (Mitchell et al.’’ Each of the five scenarios was either emotionally neutral. Structural and functional volumes were spatially normalized to T1 and EPI templates in SPM. The resulting functions were used as covariates in a general linear model. CA.. A green traffic light (see Fig.005. LPFC (BA 9). Images were corrected for differences in timing of slice acquisition. were submitted to group analyses. Each block (120 s) occurred only once per scan. they underwent a 10-min anatomical scan. Participants indicated a change in thoughts (in the ‘‘thinking’’ conditions) or thinking about the forbidden topic (in the ‘‘don’t think’’ conditions) by pressing a button with their right hand. a yellow traffic light meant simply that participants should press the response button. USA) was used to present the traffic lights and instructions synchronously with fMRI data acquisition. http://marsbar. Please press the button whenever a major shift in topic. computed on a subject-by-subject basis. they pressed the button whenever they experienced a shift in thoughts or images. The second scenario involved remembering or imagining a neutral activity engaged with the participant’s romantic partner. contrasts between conditions were computed by performing one-tailed t tests on these images. A red traffic light instructed participants to try not to think about the previous scenario but to press the response button whenever they did happen to think about it (Don’t Think block). At the group level. San Francisco. At the end of the fifth scan. The third scenario involved remembering or imagining a conflict or argument with the romantic partner. After the experimental scans. we used a more liberal threshold. Presentation software (Neurobehavioral Systems Inc.001. For example. The fourth scenario involved remembering or imagining breaking up with the partner. ACC. or activity occurs. The fMRI time series data were modeled as a series of epochs convolved with a canonical hemodynamic response function. an approximation of Talairach space based on the Talairach and Tournoux (1988) atlas.5 ms. TE = 40 ms. Statistical analyses were performed on individual participants’ data with the general linear model implemented in SPM2. The order of the scenarios was the same for all participants. / NeuroImage 28 (2005) 835 – 847 fMRI data acquisition and analysis Brain images were acquired with a 1. participants were given a 30s rest break. The least-squares parameter estimates of height of the best-fitting synthetic HRF for each condition were used in pairwise contrasts. This yellow light block served as a visual and motor control for the other blocks.5 T GE Signa scanner at the UC Davis Imaging Research Center. .. 2002. The first five brain volumes of each scan were discarded from the analysis to eliminate nonequilibrium effects of magnetization. The normalization algorithm consisted of a 12-parameter affine transformation together with a nonlinear transformation involving cosine basis functions. Thus.

Neural correlates of thought suppression revealed at the group level..001).. Activation while suppressing was also found in the MPFC (Fig. and fewer button presses for the more avoidant participants. Nevertheless. 2004). in line with our instructions. Cavada et al.002). Otherwise.. a region known to be associated with self-related processing and monitoring (e. P < 0.19) = 4. for review of relevant studies.001). and the ATP are interconnected in humans (e. 1982. Correlations with avoidance We hypothesized that avoidant participants.001). 2004 studied the reactions of 120 people). 1997 studied the reactions of 200 people. P < 0. Mesulam and Mufson.001. Gusnard et al.g. 2002).88).42. the first dealing with individual differences in attachment style and their effects on emotion regulation and cognitive control and the other dealing with neural correlates of thought suppression. we examined only relations that made sense anatomically (Buchel and Friston.19. Mora ´ n et al.46.05) and a trend toward a positive correlation between attachment anxiety and neuroticism (r = 0. and Mikulincer et al. 2002. 2004): i.52. As expected.19) = 4.32. In these group-level analyses. conflict. for example. The failure of these trends to reach statistical significance may be due to the small sample size (in contrast.. P = 0. P < 0. Specifically. P < 0. MPFC: t (1. see Botvinick et al. OFC.. Wyland et al. and Think Negative > Think Neutral. P < 0. Similar results were achieved when only attachment-related conditions were included in the comparison (ACC: t (1. Mikulincer et al. / NeuroImage 28 (2005) 835 – 847 839 resulting ROI analyses were not corrected for multiple comparisons. We conducted another set of regression analyses predicting activation in various brain regions from the average number of button presses in each block. there was a positive correlation between attachment anxiety and trait anxiety (r = 0.92.37. breakup.065). 1987. In summary. P < 0.19) = 3. the amount of button pressing did not differ significantly across the different scenarios and was not correlated with either attachment-style dimension.. P = 0. 2001. Neuroticism and trait anxiety were highly correlated (r = 0. Behavioral data Participants made fewer button presses in the not-thinking conditions than in the thinking conditions overall (t (1. we focused on four key contrasts: Don’t Think > Think. as intended.. Johnson et al. The Don’t Think > Think contrast revealed activation in the ACC (a) and MPFC (b). conditions.e. P < 0. would have a different pattern of activation related to not thinking about a particular topic because.O. neutral with partner. When assessing statistical relations between different brain regions. participants exhibited greater activation in this region when trying not to think about a given scenario (collapsing across the five conditions that involved ‘‘not thinking’’: neutral alone. 2. compatible with previous studies (Gusnard et al... 1992).. 2003). 2002. Don’t Think Negative > Think Negative. we assumed that the ACC. Fraley and Shaver. and partner’s death) than when ‘‘thinking’’ about that same scenario (t (1.19) = 4..g. Although our main interest was attachment style.81. Johnson et al.13. more button presses for more anxious participants in the rebound..36.001). Fraley and Shaver. 2a.. Neuroimaging data Whole-brain comparisons Our study combines two lines of research. on it). such people seem to retain some activation or availability of Results Questionnaire measures Attachment anxiety and avoidance were uncorrelated (r = 0.19) = 4.001.19) = 5. or Free Thought (green light with a ‘‘2’’ superimposed Fig.. 1997. P < 0. in behavioral studies. 2b) (t (1. 1997). we began our analyses at the level of the entire group of 20 participants to see whether our suppression-related findings were in line with previous findings..19) = 4.034. 2002. Kelley et al. . P = 0. thought suppression was associated with higher activation in a region known to be related to control. Morecraft et al. Gillath et al. 2001.001) or when only negative conditions were included in it (ACC: t (1... and self-regulation: the dorsal ACC (see Fig. MPFC: t (1. Kelley et al. 2000. there were small correlations within conditions compatible with previous studies (e. conflict with partner. In the whole-brain analyses. as compared with non-avoidant ones. Don’t Think Relationship > Think Relationship. Statistical models Regression analyses were used to examine associations between signal intensity in various brain regions and scores on the two attachment-style dimensions. the grouplevel analyses revealed increased activation during thought suppression in several midline PFC structures. Based on anatomical studies in non-human primates.g.

54. among participants scoring low on avoidance. The analysis revealed a significant interaction between avoidance and the ‘‘Don’t Think – Think’’ contrast ( F (1.840 O. SD = 0. We started by regressing the contrast ‘‘Don’t Think – Think’’ (collapsed across the five scenarios) on avoidance. Think: M = À0.20 for thinking. not thinking was associated with lower SCC activation than thinking (Don’t Think: M = À0.015). À2).17) = 2. Panel (c) shows the interaction between avoidant attachment and ‘‘Don’t Think Relationships vs.22. confirming that relative activation in the SCC during ‘‘not thinking’’ was positively associated with avoidance (b = 0. whereas avoidant people did not.68. t (1.g. Interaction between avoidant attachment and Don’t Think vs. Thus. the Y-axis represents the contrast between each task condition (Think relationship or Don’t think relationships) and the control condition (thinking about driving alone). while everyone activated the same regions (ACC and MPFC) to suppress Fig.03.05). P < 0. we examined the way in which attachment avoidance correlated with activation in the SCC while controlling for attachment anxiety (an unlikely confound but one we wished to rule out completely). non-avoidant participants tended to increase deactivation in the SCC when not thinking. In the first step of the regression analysis. F < 1).25 for not thinking. skin conductance) indicate that they have successfully suppressed them.05. the Y-axis in Panel (b) represents the contrast between each task condition (Think or Don’t think) and the control condition (thinking about driving alone).64. in which the ‘‘Don’t Think’’ and ‘‘Think’’ conditions were treated as a repeated measure and avoidance was treated as a predictor. F (1. SD = 0.21. we entered attachment avoidance. Think.27. A similar pattern was revealed for the suppression of relationship-related thoughts. we conducted a repeated measures analysis of variance. P = 0. In the second step.9) = 6.17) = 5.. 22. This analysis revealed a positive association between avoidance and activation in the subcallosal cingulate cortex (SCC (0. To test this prediction. 3b). Gillath et al.70. again. see Fig. Avoidant attachment was positively correlated with thought suppression (Don’t Think > Think) in the SCC (BA 25) (a). Think Relationships’’.18. Tests for simple main effects revealed that. To better understand the role of the SCC in the suppression process and to examine the differences in its activation between avoidant and non-avoidant individuals.09. P = 0. After looking at the whole-brain results. t (2. 3a). we entered attachment anxiety. .07. we first examined whether avoidance is correlated with the pattern of activation noted in the whole-brain comparisons described above. / NeuroImage 28 (2005) 835 – 847 loss-related thoughts even when some indexes (e. P < 0.002. No such relative deactivation in the SCC was evident among individuals who scored high on avoidance (M = À0. see Fig. SD = 0. which turned out not to be significantly associated with relative activation in the SCC. SD = 0.19) = 3. This suggests that. 3. M = À0.

We therefore examined possible effects of avoidance on suppression of relationship-related thoughts.26. SD = 0. Thus. 26).63 3. Think: M = À0. we conducted a repeated measures analysis of variance. t (1.46 P = 0.44 0.36. SD = 0. the more avoidant participants showed more activation (or less deactivation) in another region. We conducted hierarchical regression analyses in which activation in the ATP.75. P = 0.37 3. Table 1 Whole-brain contrasts of interest Region ¨BA Coordinates of P t peak activity value Z scores (A) Thought suppression group contrast (Don’t Think > Think) L anterior cingulate cortex 32 À10 18 26 0. activation in emotion. More avoidant participants activated the ACC and MPFC but did not show corresponding deactivation in the SCC.9) = 31.001 4. this time.001). P = 0. we entered the two attachment scores 0.50. as indicated in Table 1.001 4. The results were similar to those for the regression of the ‘‘Don’t Think – Think’’ contrast on avoidance.and conflict-related regions was correlated with attachment anxiety. Thus. No such relative LPFC deactivation was evident among individuals who scored high on avoidance (Don’t Think Relationship: M = À0.36 2.92 3. breakup. uncorrected. where activation in the ‘‘Don’t Think Relationship’’ and ‘‘Think Relationship’’ conditions was treated as a repeated measure and avoidance was the predictor. b Regression analyses were conducted using a statistical threshold of P < 0. previous studies of attachment-related cognitive and emotional processes focused mainly on relationship-related thoughts and emotions. Don’t think about-negativerelationship scenarios and Think about them.17) = 4. however. Think: M = À0. F = 2.001 4.002 3. Correlations with anxiety As expected.59. and death versus driving alone or shopping with one’s partner).01.19 3.45. we conducted a series of analyses involving only relationship-related content. we compared activation across the three negative conditions and the neutral ‘‘shopping with partner’’ condition. higher activation was found in the left ATP. that is.’’ non-avoidant participants focused their efforts in those regions while deactivating other regions. we entered neuroticism and trait anxiety scores to control for their effects. We did this by regressing the contrast ‘‘Don’t Think Relationship – Think Relationship’’ on the avoidance score.99a L Anterior cingulate cortexb 24 À4 24 20 0.53. In these ROIs. In the first step of each regression analysis.59 0. 12. without regard to the nature of the material participants were thinking about or trying not to think about. To examine this result in greater detail. hippocampus. attachment avoidance. 4).001 (B) Attachment-related thought suppression (Don’t Think Relationship > Think Relationship) L anterior cingulate cortex 32 À10 18 26 L medial prefrontal cortex 10 À8 56 2 (C) Negative thought suppression group contrast (Don’t Think Negative > Think Negative) L anterior cingulate cortex 32 À14 20 24 L medial prefrontal cortex 10 À8 56 0 To better understand this result. and the left dorsal ACC as a function of attachment anxiety (see Table 1 and Fig. 12. breakup.88 3. the left hippocampus.79.151). P < 0.O. among participants scoring low on avoidance. using the ‘‘driving alone’’ condition as a baseline for all four conditions. Tests for simple main effects revealed that.79.005). see Fig. when thinking about the negative scenarios (conflict. neuroticism. SD = 0.05. When we focused only on the emotionally negative relationship scenarios (conflict.001 4. 3c). Attachment anxiety. a similar pattern of deactivation in the LPFC was evident. F = 2. not thinking was associated with lower LPFC activation than thinking (Don’t Think Relationship: M = À1.79. but the avoidant participants did not.94 (D) Regions in which the difference in activation between Think Negative and Think Neutral was correlated with attachment anxiety L anterior temporal pole cortex 21 À38 2 À30 0. (As in the previous analysis. No such relative LPFC deactivation was evident among individuals who scored high on avoidance (Don’t Think Negative: M = À0.001). . whereas highly avoidant participants did not. SD = 0. The above analyses focused only on individual differences in attachment style. among participants scoring low on avoidance. the whole-brain comparisons revealed similar patterns of activation in the ACC and the MPFC regardless of thought content.32 3. / NeuroImage 28 (2005) 835 – 847 841 thoughts or ‘‘stop thinking. non-avoidant participants tended to increase deactivation in the LPFC when not thinking about the relationship scenarios. SD = 0.001 3. where activation in the two conditions.17) = 6.05). we found a difference in the LPFC (BA9 (À38.85. In the second step. However. F (1. SD = 0.14. although everyone activated certain brain regions during suppression (ACC and MPFC). 3c) rather than in the SCC.e. and partner’s death). When the ‘‘Don’t Think Negative – Think Negative’’ contrast was regressed on avoidance.001 4. P < 0. SD = 0.48 0. SD = 0. and trait anxiety (the latter three being potential confounds) served as the independent variables. Think: M = À0.82a a The correlation coefficients for these analyses are reported in the corresponding figures. Tests for simple main effects revealed that. we conducted a repeated measures analysis of variance.26a (E) Region in which the difference in activation between Don’t Think and Think was correlated with attachment avoidance Subcallosal cingulate cortexb 25 0 22 À2 0. greater deactivation) than thinking (Don’t Think Negative: M = À1.90 L medial prefrontal cortex 10 À8 58 2 0.002 3.9) = 25. Gillath et al.95 P = 0. This revealed a significant interaction between avoidance and the repeated measure comparing the ‘‘Don’t Think Negative’’ and ‘‘Think Negative’’ conditions ( F (1.001 3. As already noted.19.120).19) = 3.97. P = 0. Think: M = À0.0017.13 3. not thinking was associated with lower LPFC activation (i. 26). Therefore.63. or ACC served as the dependent variable. unlike the previous regression involving avoidance. t (1.19) = 3.85. non-avoidant participants tended to increase deactivation in the LPFC when not thinking. see Fig.47 2.55. P = 0. greater deactivation for the less avoidant participants occurred in the LPFC (BA9 (À38.005. This yielded a significant interaction between avoidance and the repeated measure comparing the ‘‘Don’t Think Relationship’’ and ‘‘Think Relationship’’ conditions ( F (1.58. was treated as the repeated measure and avoidance was the predictor.. we used a more liberal threshold of P < 0. We conducted region-of-interest (ROI) analyses in each of these regions to better characterize the activation patterns. 92.21 2.71a L hippocampusb À32 À24 À14 0. F (1. The relevant contrast was ‘‘Don’t Think Relationship – Think Relationship’’ (which excludes the condition in which participants thought about or tried not to think about driving alone to Lake Tahoe). Specifically.

and avoidance were respectively controlled (the b for anxiety predicting activation in the ATP was 0. the b for anxiety when predicting activation in the hippocampus was 0. 4. P = 0. and ACC. t (4.57. P < 0.05. neuroticism. and the same set of regions was identified.0001. after statistically controlling for avoidance. trait anxiety. in our initial search for correlated regions of activation. These analyses revealed that neuroticism and trait anxiety did not significantly predict activation in the regions of interest but that attachment anxiety was positively associated with relative activation during emotion-related thought (‘‘Think Negative > Think Neutral’’) in all three regions—ATP.15) = 2.002).79. the b for anxiety when predicting activation in the ACC was 0. anxiety was associated with greater activation during negative thoughts (Think Negative > Think Neutral) in the ATP. / NeuroImage 28 (2005) 835 – 847 Fig. t (4. Attachment anxiety and neural correlates of thinking negative thoughts. and ACC—even when neuroticism. . (which allowed us to control for avoidance while considering the effects of anxiety). t (4. Gillath et al. and general anxiety.05. hippocampus. Across individuals.15) = 7.842 O. hippocampus.15) = 3. We conducted similar analyses in which we used the residual attachment anxiety score.78. P < 0.55.91.

2003). revealing that more anxious participants failed to activate OFC as robustly as their less anxious counterparts. P = 0. We also expected participants high on attachment anxiety to show greater activation in memory-related regions when thinking about negative attachment-related events. 5b). The previously significant correlation became nonsignificant (r = À0. Wyland et al. breakups. providing further support for the claim that they are different..15) = À2. Finally. r = À0. attachment anxiety was associated both with higher activation in the ATP. Discussion In general. OFC activation was negatively correlated with ATP activation (b). which is known to be important for memory retrieval (for reviews. that Fig. As predicted. since our study was exploratory. Le ´ vesque et al. based on previous behavioral findings in the attachment literature.57. To explore the possibility that attachment anxiety is related to these regions.. a region associated with sadness (Le ´ vesque et al.01). In contrast. Cavada et al. we examined the neural correlates of general anxiety and neuroticism when participants were thinking negative thoughts because these correlates might be important both for understanding the underlying mechanisms of general anxiety and for distinguishing general anxiety from attachment-specific anxiety.009) (Fig.. This finding led us to consider the possibility that activation in the ATP. In general. 2004.. 1982). we found that the two regions were negatively correlated across participants. 2003. we infer. although not directly related to the goals of the present study. even when neuroticism. 2003) associating thought suppression with higher activation in the ACC and MPFC. There were null effects for all of the ROIs not reported above. Le ´ vesque et al. different brain regions were identified for the three different constructs (attachment anxiety. that people’s self-reports of high attachment anxiety are based on repeatedly experiencing high emotional arousal and low ability to down-regulate it.. nine in all. the relative degree of activation while thinking negative thoughts was negatively correlated with attachment anxiety. death of partner). 5. Based on prior behavioral research on adult attachment style. ns). In line with previous studies in which highly attachment-anxious individuals had greater access to negative memories (Mikulincer and Orbach. Gillath et al.g. 2000. This correlation was reduced to insignificance when scores on attachment anxiety were statistically controlled because attachment anxiety was associated with high ATP activation and low OFC activation.. Attachment anxiety was negatively correlated with negative thoughts (Think Negative > Think Neutral) in the OFC (BA 11) (a). as discussed in the Materials and methods section. 2003). there was a negative correlation between OFC and ATP activation (r = À0. 2004). general anxiety. The relevant correlations are shown in Table 2.95. see Eichenbaum. so it is possible that the OFC suppresses sadness through top – down modulation of activity in the ATP. we predicted that people high on attachment anxiety would show greater activation in emotion-related brain regions when thinking about negative attachment-related events (conflicts. anxiety was associated with higher activation in the hippocampus. which we interpret as indicating emotional arousal.64.. therefore. 5a). we replicated previous findings (e. the correlation remained significant when neuroticism and trait anxiety were partialed (r = À0. P = 0. a region associated with emotion regulation (Beer et al. was negatively correlated with activation in the OFC. 2000.. we computed the correlation again while partialing attachment anxiety scores.01).56. Furthermore. suggesting that the effects of attachment anxiety on activation in the ATP and relative deactivation in the OFC are specific and not due to a more generic form of anxiety.g. such that those who exhibited the lowest levels of OFC activation tended to exhibit the highest levels of ATP activation. Although we cannot know for sure that negative memories were being activated since the hippocampus plays a general role in memory retrieval.. It seems possible.g. 1995).26. and with lower activation in the OFC.. 1982). and avoidance were controlled (b = À0. P = 0. / NeuroImage 28 (2005) 835 – 847 843 In a regression analysis performed to examine the same contrast (‘‘Think Negative > Think Neutral’’) with respect to the left OFC (Fig. amygdala and LPFC). The OFC is known to regulate emotion via its anatomical connections with the ATP (e. Mesulam and Mufson. 2003. P = 0. and neuroticism).01. . we examined a relatively large number of ROIs (e. trait anxiety. In line with our predictions. Mesulam and Mufson. The ATP and OFC are known to be anatomically connected (Cavada et al.57. Le ´ vesque et al.. Consistent with this possibility.O. t (4.

Further investigations are needed to determine the significance of these two distinct regions in dACC. anxiety was associated with activation in the dorsal ACC.002 3. a region that has been associated with the subjective distress of physical pain (Rainville et al. / NeuroImage 28 (2005) 835 – 847 Table 2 Whole-brain contrasts of interest that are correlated with neuroticism or general anxiety Region ¨BA Coordinates of P peak activity value t r (A) Regions in which the difference in activation between Think Negative and Think Neutral was correlated with general anxiety R parahippocampal 37 32 À52 2 0. (2004) compared the cue period prior to successful inhibitions with the cue period prior to failures. 2002. Shulman et al.96 0.001 3. Taken together.66 a Regression analyses were conducted using a statistical threshold of P < 0.003 3. Hester et al. 24. when cues signaled a forthcoming need for cognitive control (inhibition of a response). caudate.0001 4. avoidant people failed to maintain suppression when a cognitive load was added to a suppression task.64 gyrusa R caudate (tail)a 20 À42 14 0..18 0. In other words.. successful inhibition involves both activation of task-relevant areas and deactivation of task-irrelevant ones (Hester et al.63 orbital R rostrolateral prefrontal 10 36 50 8 0. Activation in this latter region was uncorrelated with anxiety.56 0. 47 À42 40 À8 0.76 cortex L lateral orbitofrontal 11/47 À40 38 À16 0. less specifically emotional. Since the LPFC is involved in the integration of emotion and cognition. 20] was positively correlated with anxiety during negative thoughts.61 R parahippocampal gyrus.24 0.61 gyrus L anterior insula À22 26 16 0. and deactivation occurred in regions not required for cognitive control.67 cortex L medial orbitofrontal 10 À4 58 À6 0.96 0. Also based on behavioral research on adult attachment style.844 O.001 3. one region in dACC [À4. we expected avoidant people to use a somewhat different strategy for ‘‘not thinking’’ as compared with non-avoidant people. Gusnard et al.63 0. whereas more general.003 3.70 orbital L inferior frontal gyrus. parietal.. The decreases in the SCC and LPFC (BA 9) among the less avoidant participants resembled the regional decreases in blood flow noticed in other prefrontal regions in previous studies and labeled ‘‘task-induced deactivation’’ (Binder et al.001 3.06 0. uncorrected.58 cortexa (B) Regions in which the difference in activation between Think Negative and Think Neutral was correlated with neuroticism R parahippocampal gyrus 19/39 32 À52 2 0..78 0. This inference should be tested more rigorously in future studies.61 L visual association cortexa R lateral orbitofrontal 11 36 46 À12 0. Mazoyer et al. and feelings. 2000).37 0. 1997). Such deactivation occurred in the left medial frontal gyrus (but also in the left insula. Raichle et al. we found that more avoidant people maintained stable levels of activation in regions where less avoidant people showed greater deactivation when not thinking about particular scenarios.62 cortex R inferior frontal gyrus.004 2. 1999.0001 3.71 cortex L rostrolateral prefrontal 10 À24 50 6 0. 47 36 38 À6 0. 2001. more avoidant individuals showed less differential deactivation during suppression.24 0. 30 18 À42 8 0.. 2004). Gillath et al. study.. and cingulate regions (such as the dorsal ACC). (2004) suggested that deactivation in the left medial frontal gyrus (BA 9) is related to successful inhibition in a task requiring inhibition.61 0. In our study.. and precuneus/posterior cingulate). 2004).002 3. In the Hester et al. our results provide further support for the idea that anxious individuals experience negative emotions intensely. Finally.0001 4.05 0. 2001. 20.58 gyrus R lateral orbitofrontal 11 38 46 À12 0. activation in the hippocampus while thinking about negative social experiences was related to recalling negative memories. Mikulincer et al.48 0.003 3.001 3.52 0.001 3. this kind of coordinated process. suggested that cueing provided an opportunity for top – down control to decrease the resources devoted to monitoring internal states (Gusnard and Raichle.99 0. Future imaging studies could determine what happens in the brain when avoidant people become unable to .002 3. The left medial frontal region was also significantly deactivated when Hester et al. 2000. 2004.. Our finding that more avoidant people failed to fully deactivate certain regions while suppressing thoughts and emotions seems compatible with previous finding from behavioral studies (e..25 0.33 0. including emotions related to approach or withdrawal (Davidson. and find it difficult to suppress negative feelings. In our study.60 R posterior cingulate 31 22 À28 42 0. 2003). 2001).005. failure to decrease activation in the left medial frontal region in concert with ‘‘attentional’’ increases in other areas was directly related to poor performance. Hester et al.001 3.68 cortex L rostrolateral prefrontal 10 À26 50 À4 0. Eisenberger et al. whereas another region in dACC [À14. 1997).64 31/18 À20 À72 18 0. 2001). This higher activation and greater memory accessibility may be one reason for their inability to inhibit the spread of activation across negative thoughts. have greater access to a web of negative memories. and in the regulation of emotion (Gray et al. in press). Le ´ vesque et al. It has also been associated with performance monitoring and detection of mental conflict (Bush et al.76 0. In line with this expectation.002 3. was evident only among non-avoidant participants. In those studies.g. it makes sense that the contrasts between ‘‘Don’t Think Relationship – Think Relationship’’ and ‘‘Don’t Think Negative – Think Negative’’ resulted in differential avoidance-related deactivation in the LPFC. contrasts produced avoidance-related differences in the SCC..58 precuneus R caudate (tail) 28 À40 16 0. Furthermore. 47 À50 34 À4 0. while at the same time increasing attentional resources devoted to processing external stimuli. increased activation occurred in frontal.65 cortex R superior/middle frontal 6 32 28 58 0. 2001.. 2004). 1995. In both cases. something that did not happen with less avoidant people.. but the regions in which the difference occurred differed somewhat based on what was being suppressed.0001 4..16 0. and neuroticism (Eisenberger and Lieberman. 24] was activated in general when participants were asked not to think about a particular topic. social rejection (Eisenberger et al.. although occurring in different brain regions (SCC and LPFC).65 orbital L inferior frontal gyrus. memories.001 3. Harmon-Jones and Sigelman.

For example. First. pp. M. Finally. Cognitive and emotional influences in anterior cingulate cortex.. 2004). G. Valabregue.A. 539 – 546.. Blehar. M. and emotion regulation. Noppeney and Price. K. Wall. 1969/1982. J. Sci. Bellgowan.. 2005). 2004.. Brett. 883 – 893. Attachment and Loss. 3rd edR MIT Press. Modulation of connectivity in visual pathways by attention: cortical interactions evaluated with structural equation modeling and fMRI. 1091 – 1104.P. Beauregard. Moreover. In: Simpson.T. which suggests that emotional processes are modulated by top – down control from the OFC and PFC and bottom – up processes in the anterior temporal pole and hippocampus.R.... 8. Sadness and depression.005) we used for identifying the differences between avoidant and non-avoidant participants.. it is difficult to infer. Rholes. Carter. References Adolphs. 215 – 222. Vandenberghe et al.S.. J. Los cinco grandes across cultures and ethnic groups: multitrait multimethod analysis of the Big Five in Spanish and English. Bush. Psychol. NJ.. 351 – 367. M. (Ed. Blair. Although we view this study as encouraging. M. This new methodology may clarify individual differences that could previously be studied only behaviorally or via self-reports and interviews.. Conceptual processing during the conscious resting state: a functional MRI study.A.F.. E.. Ainsworth. Basic Books.S. III. K.. Attachment and Loss. Hillsdale.L. We ended up excluding many secure and anxious volunteers while working hard to attract avoidant ones. 1997. S. Abstract presented at the 8th International Conference on Functional Mapping of the Human Brain. 3rd edR MIT Press. R. regions associated with emotion.. The questions under investigation are rooted in a well-established theory. Hammeke. and the experimental procedures employed have been used in previous cognitive and behavioral experiments. June 2 – 6. J..I.).’s (2003) study.. Cereb. C. Many supposedly general mental processes documented by social cognition researchers have turned out to be moderated by attachment style (for a review. Poline. Vol. Inclusion of a non-attachment-related negative emotion condition will be important in future studies because we cannot yet be sure how many of our findings can be attributed to attachment-related negative emotions alone and how many might be replicated when we include non-attachment-related stimuli and situations..I.N. While doing this. M. Anderson. Kuhl... Relat. 2004. Kirkpatrick.. Basic Books.A. vol. The results fit well with other neuroscientific studies of emotion regulation (e.. Soc. vol. vol. 1978.. Separation: anxiety and anger. P. Self-report measurement of adult romantic attachment: an integrative overview.).. New York.. Cortex 7 (8)..D.. Brain 122. Posner.. M. 1. W. we studied only women to eliminate gender-related variance in the issues under investigation. C. Patterns of Attachment: Assessed in the Strange Situation and at Home. Beer. Dolan. Neural systems underlying the suppression of unwanted memories. New York. M.. Many psychological processes related to people’s immersion in a social environment are likely to be moderated by such individual differences. Frith. John. J. Basic Books. K. based on our findings.. Neurosci. Neural correlates of conscious self-regulation of emotion. Bowlby.S. This might have been related to our requirement that participants be involved in a serious long-term relationship. J. Morris. 2000. 2.. 232 – 235. 1998. memory. 2004. Knight... others have linked it to semantic processing (e.. Cogn. 1999. S.. Robertson. M... 3. Gabrieli. 1017 – 1031. J. M. 11. Cogn. J. Cogn. MA. Shaver.. J. 2001. 1996. 1973. Waters. our sample size was small for an individual-differences study.J.. pp. Our findings indicate that individual differences play an important role in emotion regulation. P. Cambridge. Bourgouin.S.W. 4.P. 1 – 6. Cohen. Botvinick.).. In: Gazzaniga. Cox. Neurosci. Conflict monitoring and anterior cingulate cortex: an update. Although some investigators have linked ATP activation to affective processing.B. what psychological process is associated with ATP activation.g.C. Carpenter.O.A. Neurosci. 75.. Clark.M. We were not able to include all relevant control conditions in our experiment. No. Thus.. Guilford Press.W. Luu. Benet-Martinez. J. MA. 80 – 93.. we were able to represent the anxiety dimension of attachment style more completely than the avoidance dimension. S. 1999.. another interpretation of Eugene et al. C.. Cambridge.D..D. Attachment Theory and Close Relationships. which many avoidant college students are not.A. III.. Cooper. A. Glover.S. Bowlby. 3. pp.H. Ochsner and Gross. Perrett. (Ed. Erlbaum. vol. Pers.J. as we are inclined to do here.. R.S. suggesting that future studies should include relevant individual-difference measures. Gabrieli. J. Brennan. 2002. Pers. P. T.. more representative samples. This is especially important in light of the liberal threshold ( P < 0. C. J. see Mikulincer and Shaver. V. R. the inclusion of individual differences enabled us to look beyond seemingly general processes that are likely to be clouded by such differences. further investigation of the ATP activations in relation to attachment anxiety is necessary.J. the study revealed associations between self-reported attachment style and brain activation in . 2. vol. 2004.. J. Trends Cogn. whereas their second group included a relatively high proportion of avoidant individuals. Japan. Attachment. E.S... Although preliminary. Neurosci. G.. Binder.. 2002. R. and we therefore included relatively extreme scorers on the attachment scales.R. P. B. Gillath et al. Conclusion Despite these limitations.M.D. 46 – 76. New York. J. making it possible that more or clearer findings related to avoidance will be obtained in future studies with larger.. our study yielded promising results. 2002). Frost. 2nd edR Attachment and Loss. J. Processing of emotional and social information by the human amygdala.. 1980. Ochsner. L. Region of interest analysis using an SPM toolbox. Bowlby. 768 – 778. described in the introduction. 729 – 750. it is important to consider some of its limitations. O. R. Trends Cogn. Sendai. Friston. Second. Science 303 (5655). Levesque. is that their first group of participants included a relatively high proportion of anxious individuals. (Eds. E. R. 2002. Rao. Buchel. 16. Using brain imaging techniques to study such social cognitive processes as they are moderated by attachment style should reveal where and how the differences arise. New York. / NeuroImage 28 (2005) 835 – 847 845 suppress certain mental contents while operating under a cognitive load and whether the failure to deactivate regions that are directly related to the suppression task is related to the disruptive effects of the load. Sci.L.. J. we discovered that it was easier to recruit secure and anxious participants for a study of ‘‘close relationships and the brain’’ than to recruit avoidant participants. Frontal lobe contribution to executive control of cognitive and social behavior.C. J. 21. Whatever the reasons.g. Attachment style and presence of a romantic partner as moderators of psychophysiological responses to a stressful laboratory situation.S. Anton. 1998. Dissociable neural responses to facial expressions of sadness and anger. In: Gazzaniga. J. Available on CD-ROM in NeuroImage. Shimamura. D.M.

Research. 2005.B.. Comp.. K. Gusnard. Sci. Biol. 1997.L. The emerging neurobiology of attachment and separation: how parents shape their infant’s brain and behavior.. M. Dolan. M. Mikulincer. Leroux. Evans.. M. In: Rholes. Hofer. A. Rev.. Mental representations of attachment security: theoretical foundation for a positive social psychology.. P. Mikulincer. Bourgouin.R. J. R. An integrative theory of prefrontal cortex function. Spieker. P. 1998.L... Beaudoin. Proc. Mikulincer.J.R. M.. Tejedor.. Kelley. Why rejection hurts: a common neural alarm system for physical and social pain. W. Calarge. Insula of the old world monkey: architectonics in the insulo-orbital temporal component of the paralimbic brain. M....R. S. Gray.S. 2005.and insecurely-attached infants of depressed mothers. Casey. D. 2004. S.. Adult Attachment: Theory.. September 11: Trauma and Human Bonds. J.. Le ´ vesque. Lieberman..). 2003.P. Andreasen. C.J. Brain 126. 167 – 202. Beaudoin. J. H. emotion. Romine... MA. 159 – 195. Analytic Press. C. Kollokian.. M.. S.. Natl. J. Trends Cogn. Pers. G. Panagiotides. 53 – 152..846 O. Eugene. Psychol. 255 – 270.. Mikulincer. Harmon-Jones. Neurosci. Cereb. Diamond. Schechter. Brain Res. K. Neural correlates of self-reflection. Cohen. C. 212. Kirkpatrick.A. Wyland. E. W. Soc. 260 – 273. W. NeuroImage 5.. 300 – 326. K.. Mensour.. 24. S. Etard. Pers. 509 – 523.S.. O. Eichenbaum. Cerebral asymmetry. New York. M.. J. V. N. Contributions of psychophysiology to research on adult attachment: review and recommendations. Shaver. L.J. 1988. Javitt.D. P. M. 290 – 292. 2004.. Shaver. S. Erlbaum. 2000. 8. Lieberman. Petit.K. Cohen. Neurosci. pp. 361 – 387. Ryan. 83 – 98.... Eugene. B. Hunt. Exploring individual differences in reactions to mortality salience: does attachment style regulate terror management mechanisms? J. 387 – 404.. P. L.G.. 1996. Autonomic and brain electrical activity in securely.R.I. Interpersonal Cognition.. Beauregard. (Ed. Are infant attachment patterns continuously or categorically distributed? A taxometric analysis of strange situation behavior.. Bourgouin. Joanette.. Cortex 10.M.). Cortical networks for working memory and executive functions sustain the conscious resting state in man.. D. Attachment-related strategies during thought-suppression: ironic rebounds and vulnerable selfrepresentations.C. 940 – 956. 1982. J. Clinical Implications of Attachment. Wszalek.K. In: Davidson. Beauregard. Hillsdale. Psychol. J. Woddis. Mufson. Shaver.L... Soc. 191 – 209. Feeney. S. Psychol. Sci. D. G.D. 420 – 435. MA. O. R.. Does rejection hurt? An FMRI study of social exclusion. Heiserman. J. Murphy.. J. Cogn.. Cohen. R. Pers.... M.E. State anger and prefrontal brain activity: evidence that insult-related relative left prefrontal activation is associated with experienced anger and aggression.. The impact of individual differences on the neural circuitry underlying sadness. M. NJ.. Fletcher. Eisenberger. Gillath et al. Adv. Effects of adult attachment and presence of romantic partners on physiological responses to stress. vol.R. 1995. G. 5. N. Heatherton. Avoidance and its relation to other defensive processes. Soc. R. Cogn.. 1995. NJ. Glover. Y. Foxe. in press. M.I.. T.. 73.. Pers.. The attachment behavioral system in adulthood: activation. M.. 220 – 242.. 2004. E. Simpson. L. 2000. 1). The anatomical connections of the macaque monkey orbitofrontal cortex: a review. Le ´ vesque. Kobak...W. T. S. M. and affective style..D. Exp.. Psychol. M. N. Pipe. pp... 685 – 694. Soc. 135 – 149.. 80. Psychol. 799 – 813. Hessl. III. Drevets.M. Academic Press. 679 – 690. J. Bricogne.. Foxe.. D... 1080 – 1091.N. (Eds. Dev.. Dissociable temporal lobe activations during emotional episodic memory retrieval. Neuroscience 129. J. J.. C. and Clinical Implications..S.. / NeuroImage 28 (2005) 835 – 847 Hare. J.. Psychiatry 48. Frith.J. R.. P. Cogn. Cogn.. Dev. R. Searching for a baseline: functional imaging and the resting human brain. 24. Contributions of amygdala and striatal activity in emotion regulation. Rev. In: Zanna. Lieberman..). Hillsdale. J. (Eds.J. pp. Compan ˜ y. J. Guilford Press. Pers. Soc. Psychiatry 160.I. Psychol.X. 425. Mikulincer. 2003. Neurol. R...J. Shaver. Behav.. Pers..). G. Stress and accessibility of proximity-related thoughts: exploring the normative and intraindividual components of attachment theory.. Y. U. 2001. 624 – 632. Dawson. D. B. T. O’Leary.P. NeuroImage 19 (2 Pt. Integration of emotion and cognition in the lateral prefrontal cortex. 776 – 785.. M. Tzourio-Mazoyer. Spieker. Houde.. Guilford Press. N. V..M.. extraversion. Nezworski. C. E. Sci. Biol.. Mazoyer.. San Diego. Aging affects the engagement of the hippocampus during autobiographical memory retrieval. R. In: Belsky. Acad. (Eds. Lane. 1997. R.D. Shulman.. Neural circuitry underlying voluntary suppression of sadness. 70. Beaudoin. 87. Soc.S. 294 – 300.C....E. Proc. Cavada. M. Hippocampus 9..J.. Dolev. Affect. M. J. P... P.C. C. 2004. Annu... Fraley.E. L.C.. S. J. Cambridge..A.L. 2002. and interpersonal processes. 2004.. 2001. Rev. Frey. Orbach. MIT Press. Emot. 1954 – 1964.E. M. 354 – 364. Prigatano.. S..B. 14. Psychol..... 1 – 22.. Crivello. H. B. 2004. Neurosci. 502 – 510. D. E. J.S. 98.. Caglar. Shaver. T.J.H.. Mensour..and emotion-related words. 4115 – 4120. C. 2003.M. Maguire.. Mensour. Leroux. 54... P. G. pp. Mikulincer. Nat. 203 – 209. P. Adult attachment style and affect regulation: strategic variations in self-appraisals. 78. Joanette. 2003. Psychol. Avoidant attachment and hemispheric lateralisation of the processing of attachment. Joliot. . J. O. L.M. 4259 – 4264. Chua. 287 – 298. Acad. Raichle. Attachment styles and repressive defensiveness: the accessibility and architecture of affective memories. vol.. 276 – 295. Psychiatry 57. Macrae. Eisenberger.. Neurosci.. 233 – 266. T. Leroux. (Ed. 39... Cocosco.F.A. 35.. P. Sigelman.A. 917 – 925. G. T. Hippocampal system and declarative (relational) memory: summarizing the data from functional neuroimaging studies. E. Raichle. Biol. Nachmias. 2002.). M. D. 813 – 829.. pp. F. Soc. J. 2003. Hester. Brainweb: online interface to a 3D MRI simulated brain database. Williams. Hugdahl. A. In: Gazzaniga.. Psychol.. Tottenham. 2001. Natl. N. 16. J. Medial prefrontal cortex and self-referential mental activity: relation to a default mode of brain function. 2000. Wilder. Soc. Miller. A.. B. Mellet. Cambridge. Finding the self? An event-related fMRI study. 1999.. 2004. 3rd edR MIT Press.. J. 99. 2003.A. Reinoso-Sua ´ rez. Pers. Adult attachment and the suppression of unwanted thoughts.. Davidson. J.. Brain 125. F. 75.J. Bourgouin. L. (eds. 2000. N.. 361 – 369. J. In: Baldwin.A.. Raichle. Zago. 2001. (Ed. J. M.C.. Psychol. M... pp.R. Nash. E. Security-based self-representations in adulthood: contents and processes.C. M. Paquette.... H. Inati. B. G. Visualizing how one brain understands another: a PET study of theory of mind. Davidson. M.. Johnson. P. 797 – 803. Infant Behav. Neuroimaging studies of mood disorders. 1511 – 1523. Akbudak. 2001. 2.C. J. M. 1808 – 1814. 2001.. G. 2002. Bull. Gusnard. Pers. Florian. 79.J. 2000.S. Am. Rosenthal. 785 – 794.. M. B.D. Beauregard. Mikulincer. and self-consciousness.C. Pers. Neural basis of emotional selfregulation in childhood. Birnbaum.A.S. Ashman.. Soc. Science 302. Cruz-Rizzolo1. In: Coates.. Neurosci. I. Shaver. D. pp. Braver.. Baxter. F. W. Soc.A.D... 2003.. T.M. L. Mikulincer. Mesulam. R. Psychiatry 53.).C... Kwan.R. Fraley.C. C. K. 18.. Cogn. J. Garavan. L. F. Embry. Cassidy.. Le ´ vesque. psychodynamics. 2003. New York. N. Satpute.. Neurosci. An information processing framework for memory representation by the hippocampus. 68.. 2001. V. Predicting success: patterns of cortical activation and deactivation prior to response inhibition.M. Personality from a controlled processing perspective: an fMRI study of neuroticism. J. Psychol.. Eisenberger.. NeuroImage 11.).

1344 – 1350. S. Gabrieli.. Ochsner.E. A. 1988.O. Manual for the State-Trait Anxiety Inventory.. J. Talairach.. R. Bouthillier.N.. J. Mesulam. Heatherton. Kelley.T.. Ochsner. S. Handbook of Self-regulation: Research. Thompson. R. Levesque. NeuroReport 15....N. F.M. Neural correlates of thought suppression. Gross.D. C... Deakin.L. 90 (1).A. Welsh.J.R. Phan. Regional cerebral blood flow during script-driven imagery in childhood sexual abuse-related PTSD: a PET investigation. K. J. T.J. 9. Attachment and the regulation of the right brain.. 2002..J.. Nobre. 2004. Empirical Perspectives on Object Relations Theories.P. M. R.A. Natl. Price. Functional neuroimaging studies of human emotions.M. K. 2003. American Psychological Association. M. 968 – 971.. K.... 2002. G. Gross. S.. 1997.M. Liberzon. M. J. 9 (4). B.J. Neurol. Gordon. Fiez.. Metzger..L.F. Theory. 2003. unpublished manuscript. Shaver. Shin. P. Noppeney.N. Orbitofrontal cortex and dynamic filtering of emotional stimuli. 98. M. / NeuroImage 28 (2005) 835 – 847 Mitchell. Acad. Neurol... Price. R.. Behav.M. 331 – 348..M. W..F. T. 2002.. Duncan. L. J. NeuroReport 14. Paquette. Mensour. Britton.. Washington..... Schore.E.. 2004. C. Orr... C. J. Gusnard. J. U.L. J. S. Common blood flow changes across visual tasks: II. L.L. Wager. M. S. K. I. Controlling the contents of consciousness: the neural substrates of thought suppression.. Psychiatry 156.. I. T.A. Phan. 676 – 682. R. 229 – 255.. Raichle. Spielberger. R. CNS Spectr. Proce.J. 550 – 560. 2000. Cytoarchitecture and neural afferents of orbitofrontal cortex in the brain of the monkey.E...A. 1992. Cogn. Browning. Consulting Psychologists Press. Carrier. Knight...R. New York. 2004. Rule. Cogn. In: Baumeister. Liberzon. Hum.. U. Roberts. Comp. Neuropsychopharmacology 28. Form Y.N. 23 – 47. A PET study of stimulus.. J.. Thieme..J.C.. Shulman. 14 (4).. J.. Lasko. Neurosci.L. R.F.F. Leroux...L..C. 1994.. Abel. Functional neuroanatomy of emotion: a meta-analysis of emotion activation studies in PET and fMRI.F. pp. Mitchell.M. Neurosci. A. 341 – 358. Cogn.. 1111 – 1116.L. Pitman. Rethinking feelings: an fMRI study of the cognitive regulation of emotion.M. C.P. 2002. Am. Inoue. 1999.. R.. 252 – 262. C. Macrae. P. Powers. 105 – 156. J.. The psychodynamics of adult romantic attachment. Inhibitory control and affective processing in the prefrontal cortex: neuropsychological studies in the common marmoset. Bunge... A.L. Phan..N. In: Masling. Neurophysiol. Kelley.L. M. Guilford Press.J. S.. Shimamura. .. E. Dev.. S. Kosslyn. Snyder.D. Vandenberghe. CA. Activity of primate subgenual cingulate cortex neurons is related to sleep.. Decreases in cerebral cortex.. NeuroImage 15 (4). R. T. 256. 2 (3). D... A. Shulman. Rauch.. Clark. M.J. (Eds. H.M. Science 277. B. NeuroImage 162. 575 – 584. M.A.. Bourgouin.. Beaudoin... Neuropsychologia 41 (14). A. Cereb. 2003. Breault.and task-induced semantic processing. J. Taylor. Co-planar Stereotaxic Atlas of the Human Brain. 1997. C. Corbetta. Stuttgart. pp.H. DC..). K.L.L. 1987. 927 – 935.B. Separate neural circuits for primary emotions? Brain activity during self-induced sadness and happiness in professional actors. Affect. Rainville. Pelletier. P... Pain affect encoded in human anterior cingulate but not somatosensory cortex.. Petersen. 1215 – 1229.. J. C. Wyland. 648 – 663. Vohs. Wallis. Neurosci. I. 2.. J. Alpert.. Wyland. 847 Raichle. Beauregard. The response of left temporal cortex to sentences.F. P. R. Ranote. Macrae.. A. Cogn. G. G. Mora ´ n. Bornstein. Carrier. Mesulam.. S. L. Proc. Jacobs. C.. A default mode of brain function. W.D. and Applications... Bushnell...M.. 2002. E.. 14 (8). 2003. 2000. Sci. R. G. C. C. 264 – 270. A.. Taylor... S. M. S. J. S. Gillath et al. D.. Comp. Attach.P.. Morecraft.... Neurosci.. N. K. Cortex 10.C. Thinking makes it so: a social cognitive neuroscience approach to emotion regulation.K. Buckner. W. Liberzon.L.Z. G.T..R.. J.M. K.C.F. Lushene.F. N. Wager. 2001. A.C. Tournoux.J. Geula. 258 – 266.. Rolls.). Elliott. A. J. Mufson. Heatherton. J. Gorsuch. Appleby. R.D.. 323. MacLeod. Vagg. R. 88 – 103. Habituation of rostral anterior cingulate cortex to repeated emotionally salient pictures.M.... K.. 1863 – 1867. 134 – 142. (Eds. V. 1825 – 1829. The neural basis of maternal responsiveness to infants: an fMRI study.L. Palo Alto. 1983.M. P.D. Meizin. Neural inputs into the temporopolar cortex of the rhesus monkey. W.. Taylor. McNally. R. J.