Secure attachment priming protects against relapse of fear in Young adults

The study investigates whether priming secure attachment prior to extinction learning can enhance extinction retention and reduce relapse of fear. Anxiety disorders are prevalent and burdensome, and exposure-based treatments show modest remission and frequent relapse. Experimental models of fear learning and extinction are used to understand mechanisms of relapse (spontaneous recovery, renewal, reinstatement). Prior work shows attachment priming can reduce fear acquisition and consolidation and may enhance extinction, but effects could be confounded by positive mood induction. The authors aim to test an ecologically valid, brief imagery-based attachment prime administered immediately before extinction, comparing it to a matched positive mood imagery control, to isolate the unique contribution of attachment priming to extinction learning and return of fear.

Pharmacological and neuromodulatory strategies to augment extinction have shown modest effects. Attachment priming, which activates mental representations of supportive attachment figures, has been found to attenuate fear acquisition and consolidation and to enhance extinction (e.g., images of attachment figures reduced immediate and delayed relapse versus strangers). However, positive stimuli or mood induction can also reduce fear responses and improve extinction, raising the need to dissociate attachment-specific effects from general positive affect. The present work builds on Hornstein et al. (2017) by using a brief imagery prime rather than repeated presentation of attachment figure images and by including a positive mood control to test specificity.

Design: Two-session differential fear conditioning and extinction experiment with a between-subjects manipulation of pre-extinction prime (attachment vs positive control). Session 1 included startle habituation, pre-conditioning, conditioning, and extinction. Session 2 (24 h later) included extinction recall and reinstatement (after unsignaled shocks). Primary outcomes were fear-potentiated startle (FPS) and shock expectancy ratings. Participants: 124 undergraduate students (86 females; mean age 19.82, SD 3.13) recruited for pay or course credit. Exclusions: Extremely severe DASS scores (n=4), withdrawal/nonattendance (n=6), technical issues (n=2), contingency unaware (n=7), startle non-responders (n=10), startle non-learners (n=14), yielding a final N=81 (57 females; mean age 19.79, SD 1.81): attachment group n=41; control n=40. Randomization: Computer-generated to attachment vs positive control conditions. Ethics: UNSW HREC (HC15507); informed consent obtained. Measures: DASS-21; Experiences in Close Relationships (ECR) for attachment anxiety/avoidance; Vividness of Visual Imagery Questionnaire (modified one item to avoid eliciting attachment figure); demographics/health. Stimuli and apparatus: CSs were colored squares (orange/purple; counterbalanced), presented 8 s on black screen with fixation. Startle probes were 40 ms, 100 dB white noise via headphones, delivered during CSs (at 6–7 s) and on half the inter-trial intervals (ITIs). Background pink noise at 65 dB. US was a 500 ms mild electric stimulation to left forearm via bar electrode; constant 5 V with current individually titrated (0.3–30 mA) to "uncomfortable but not painful" (actual chosen range 0.3–9.6 mA; SD 1.90). Software: Presentation (Neurobehavioral Systems). Priming procedures: Attachment prime: participants imagined a specific person who is currently very supportive, dependable, and makes them feel safe/loved; they described the relationship and a typical interaction. Positive control prime: participants imagined a specific hypothetical situation that would make them very happy, fulfilling dreams and improving life, explicitly excluding other people or social/attachment aspects (e.g., achievement, wealth, fame); details were elaborated. All participants rated happiness, excitement, and closeness (0–9 scales) and, after 3 min eyes-closed imagery with periodic prompts, rated vividness (0–5). Outcomes: Startle EMG of orbicularis oculi per standard guidelines; shock expectancy ratings (1–10) during each CS via a slider. Procedure: Startle habituation with 9 probes; US titration; instruction that shocks may/may not occur. Pre-conditioning: 4 non-reinforced CS+ and 4 CS− trials. Conditioning: 16 CS+ and 16 CS−; 10 of 16 CS+ paired with US (partial reinforcement); no CS− reinforcement; randomized with constraints (no >4 same trial type or >4 consecutive reinforced/non-reinforced CS+). After a 25-min break (neutral airplane clip, questionnaires), the prime was administered in final 5 min, immediately followed by extinction: identical to conditioning but with no USs. Session 2 (24 h later): extinction recall (12 CS+ and 12 CS−; no US), then reinstatement: 4 unsignaled USs, followed by 12 CS+ and 12 CS− (no US). Data processing and analysis: Trial scores averaged in blocks of 4 trials. Startle non-responders: ≥3 consecutive missing/zero responses during baseline phases. FPS defined as CS+ minus CS− differential startle; startle non-learners lacked FPS during acquisition and were excluded a priori. Repeated-measures ANOVAs tested Group (attachment vs control) × CS (CS+ vs CS−) × Block effects per phase; habituation examined separately. Percent recall scores computed as FPS during extinction/recall/reinstatement expressed as percentage of acquisition FPS. Moderation: Linear regressions tested ECR anxiety or avoidance, group, and their interactions predicting percent recall measures. Power analysis based on prior effect indicated need for ≥42 per group (90% power, alpha=0.05, two-tailed); oversampling conducted to allow for exclusions.

Prime ratings: Attachment and control primes were equally positive (F(1,79)=1.16, p=0.284, η²=0.014); control rated more exciting (F(1,79)=4.76, p=0.032, η²=0.057). Attachment elicited greater closeness (F(1,79)=64.55, p<0.001, η²=0.450) and higher vividness (F(1,79)=7.04, p=0.010, η²=0.082). Fear learning checks: Habituation startle decreased across trials (F(1,79)=67.47, p<0.001, η²=0.461). No CS differences at pre-conditioning (startle F(1,79)=0.73, p=0.397; expectancy F(1,79)=0.86, p=0.347). Conditioning showed robust CS effects (startle F(1,79)=55.43, p<0.001, η²=0.412; expectancy F(1,79)=856.64, p<0.001, η²=0.916) and increased learning across blocks (startle CS×Block F(1,79)=6.11, p=0.016; expectancy CS×Block F(1,79)=271.47, p<0.001). Extinction: CS main effect persisted (startle F(1,79)=37.69, p<0.001; expectancy F(1,79)=146.59, p<0.001); expectancy showed linear CS×Block decrease (F(1,79)=59.44, p<0.001); startle decreased across blocks for both CSs (F(1,79)=26.50, p<0.001). Extinction recall (Day 2): CS main effects (startle F(1,79)=21.72, p<0.001; expectancy F(1,79)=95.28, p<0.001); expectancy decreased across blocks (CS×Block F(1,79)=38.74, p<0.001); startle decreased across blocks (F(1,79)=24.47, p<0.001). CS+ remained greater than CS− in the final recall block (startle F(1,79)=9.15, p=0.003; expectancy F(1,79)=32.78, p<0.001). Reinstatement: From last recall block to first reinstatement block, fear increased (phase main effect: startle F(1,79)=11.32, p=0.001; expectancy F(1,79)=57.77, p<0.001), with a stronger increase for CS+ in expectancy (phase×CS F(1,79)=22.49, p<0.001) but not startle (F(1,79)=2.45, p=0.122). Across reinstatement, CS main effects were present (startle F(1,79)=27.71, p<0.001; expectancy F(1,79)=58.83, p<0.001); expectancy CS effect decreased across blocks (CS×Block F(1,79)=51.76, p<0.001); startle decreased across blocks for both CSs (Block F(1,79)=30.59, p<0.001). Priming effects: No group differences during habituation, pre-conditioning, conditioning, extinction, or recall (all Group×CS and Group×CS×Block ps>0.23). From recall to reinstatement, significant Phase×CS×Group interactions for startle (F(1,79)=4.30, p=0.041, η²=0.052) and expectancy (F(1,79)=4.46, p=0.038, η²=0.053); follow-up showed greater increase in fear in the control group (startle phase main effect F(1,39)=7.75, p=0.008) than the attachment group (F(1,39)=4.58, p=0.039). During reinstatement, attachment group showed lower overall startle than control (Group main effect F(1,79)=4.24, p=0.043), with no significant group difference for expectancy (F(1,79)=2.68, p=0.105). Percent recall (relative to acquisition) showed less reinstatement in the attachment group for startle (F(1,79)=8.90, p=0.004, η²=0.102). Moderation: No significant interactions of ECR attachment anxiety or avoidance with group on percent extinction, recall, or reinstatement (all ps>0.05). Overall, attachment priming reduced physiological relapse of fear but did not significantly alter subjective shock expectancy.

The findings indicate that a brief secure attachment imagery prime administered immediately before extinction did not change within-session extinction or next-day extinction recall, but it did reduce the return of fear after reinstatement at the physiological level (startle), with weaker or absent effects on subjective expectancy. This supports the hypothesis that attachment priming can strengthen long-term inhibitory safety memories, beyond effects attributable to positive mood or arousal, as the control prime matched or exceeded positive valence/arousal. The attachment prime appeared to confer broader safety learning, lowering responses to both CS+ and CS− during reinstatement, consistent with the notion of attachment figures acting as unique safety signals that enhance inhibitory learning rather than merely protecting extinction. Differences from prior work that showed immediate extinction benefits may reflect procedural differences (more extensive acquisition/extinction with partial reinforcement here, which slows extinction) and control conditions (prior studies used less positive controls). Clinically, enhancing awareness of attachment figures before exposure-based therapy may reduce relapse risk by bolstering safety memory consolidation or retrieval. The absence of moderation by attachment style suggests potential generalizability across varying attachment security levels, though larger samples may be required to confirm.

This study demonstrates that brief secure attachment priming prior to extinction learning reduces reinstatement-driven relapse of fear measured via startle but does not significantly affect within-session extinction, next-day recall, or subjective expectancy measures. The results suggest attachment priming enhances long-term inhibitory/safety learning beyond positive mood effects and has translational potential to augment exposure therapies to reduce relapse. Future research should include a no-prime control, post-manipulation mood checks, better match primes for vividness, larger samples to test moderation by attachment style, longer or adjusted extinction protocols, and clinical populations to assess generalizability and therapeutic impact.

- No post-manipulation assessment of mood following the prime, limiting confirmation that groups remained equivalent in affect after manipulation. - Primes differed in vividness (attachment imagery more vivid), which could confound effects despite analyses suggesting it did not account for outcomes. - Study may have been underpowered to detect moderation by attachment style; effect sizes for such interactions are unknown. - Non-clinical, predominantly young female convenience sample limits generalizability and may underrepresent insecure attachment styles. - Procedural choices (partial reinforcement, number of trials) may have reduced sensitivity to detect within-session extinction benefits.