Dismissing Attachment Characteristics Dynamically Modulate Brain Networks Subserving Social Aversion

Abstract

Attachment patterns influence actions, thoughts and feeling through a person's "inner working model". Speech charged with attachment-dependent content was proposed to modulate the activation of cognitive-emotional schemata in listeners. We performed a 7 Tesla rest-task-rest functional magnetic resonance imaging (fMRI)-experiment, presenting auditory narratives prototypical of dismissing attachment representations to investigate their effect on 23 healthy males. We then examined effects of participants' attachment style and childhood trauma on brain state changes using seed-based functional connectivity (FC) analyses, and finally tested whether subjective differences in responsivity to narratives could be predicted by baseline network states. In comparison to a baseline state, we observed increased FC in a previously described "social aversion network" including dorsal anterior cingulated cortex (dACC) and left anterior middle temporal gyrus (aMTG) specifically after exposure to insecure-dismissing attachment narratives. Increased dACC-seeded FC within the social aversion network was positively related to the participants' avoidant attachment style and presence of a history of childhood trauma. Anxious attachment style on the other hand was positively correlated with FC between the dACC and a region outside of the "social aversion network", namely the dorsolateral prefrontal cortex, which suggests decreased network segregation as a function of anxious attachment. Finally, the extent of subjective experience of friendliness towards the dismissing narrative was predicted by low baseline FC-values between hippocampus and inferior parietal lobule (IPL). Taken together, our study demonstrates an activation of networks related to social aversion in terms of increased connectivity after listening to insecure-dismissing attachment narratives. A causal interrelation of brain state changes and subsequent changes in social reactivity was further supported by our observation of direct prediction of neuronal responses by individual attachment and trauma characteristics and reversely prediction of subjective experience by intrinsic functional connections. We consider these findings of activation of within-network and between-network connectivity modulated by inter-individual differences as substantial for the understanding of interpersonal processes, particularly in clinical settings.

Keywords: attachment; cognitive schema; functional connectivity; individual differences; resting-state.

Figure 1

Study design. First, a 10 min baseline resting-state was measured. After a distractor (90 s), participants listened to the first narrative. Then a 10 min post-task resting-state was measured. Afterwards participants rated the narrative and their feelings. The block distractor–narrative–resting-state–rating was conducted for a second and third time, so every subject listened to all three narratives (secure, insecure-preoccupied, insecure-dismissing) in a randomized order over subjects.

Figure 2

Behavioral effects of narratives. (A) Friendliness. A significant effect of narrative on countertransference reactions was observed (n = 23, *p < 0.03, F_(2,19) = 11.73). The dismissing narrative was rated as the least friendly. The bars represent the standard deviation between subjects. (B) Tendency for social interaction. A significant effect of narrative on countertransference reactions was observed (n = 23, *p < 0.004, F_(2,19) = 11.30). The participants showed the lowest tendency for potential social interaction with the dismissing narrative. The bars represent the standard deviation between participants.

Figure 3

Correlation of attachment characteristics with childhood trauma score. Childhood Trauma Questionnaire (CTQ) score correlates positively with attachment anxiety (r = 0.66, p < 0.002; A) and attachment avoidance (r = 0.59, p < 0.006; B).

Figure 4

Main effect of dismissing narrative. (A) Depicted is the increase in functional connectivity (FC) between left dorsal anterior cingulated cortex (dACC; seed region) and left anterior middle temporal gyrus (aMTG; x = −50, y = −6, z = −24) after the dismissing narrative compared to baseline (p < 0.05, corrected, Z = 5.1, k = 8). The same peak coordinates were found for right dACC (p < 0.05, corrected, Z = 4.9, k = 6). (B) Comparison of FC between the seed dACC and anterior MTG between all narrative conditions. The bars represent the standard deviation between subjects. The black star indicates a significant difference (p < 0.01) between dismissing and secure conditions.

Figure 5

Inter-individual differences. (A) FC between right dACC and aMTG (x = −56, y = −8, z = −24) after the dismissing narrative showed a positive correlation with attachment avoidance (Z = 3.7, k = 18, p < 0.001, uncorrected, r = 0.34). (B) FC between right dACC and MPFC (x = −18, y = 38, z = 52) after the dismissing narrative showed a positive correlation with attachment avoidance (Z = 3.8, k = 21, p < 0.001, uncorrected, r = 0.34). (C) FC between left dACC and right DLPFC (x = 26, y = 62, z = 18) after the dismissing narrative showed a positive correlation with attachment anxiety (p < 0.001, uncorrected, Z = 4.1, k = 40, r = 0.54). (D) FC between right dACC and right hippocampus (x = 30, y = −10, z = −24) after the dismissing narrative showed a positive correlation with attachment avoidance (p < 0.001, uncorrected, Z = 3.6, k = 12, r = 0.69).

Figure 6

Childhood trauma related influences. FC between right amygdala and right aMTG (x = 52, y = −4, z = −18) at baseline showed a positive correlation with CTQ total score (r = 0.79, p = 0.008, k = 43, small volume corrected for an AAL-derived MTG ROI on a cluster level).

Figure 7

Baseline prediction by friendliness scores. FC between left hippocampus and left inferior parietal lobule (IPL; x = 38, y = −40, z = 50) correlated negatively with friendliness ratings (p < 0.05, corrected, Z = 5.17, k = 9, r = −0.86).