Childhood maltreatment is linked to larger preferred interpersonal distances towards friends and strangers across the globe

The study examines whether childhood maltreatment (CM)—including abuse and neglect by caregivers—is associated with alterations in comfortable interpersonal distance (CID), an important social-regulatory behavior developed in childhood. Prior work links CM to broad social dysfunctions (e.g., isolation, peer rejection), altered processing of social stimuli (e.g., enhanced threat processing, negative bias to neutral faces), and discomfort with social touch. CID preferences vary by relationship, culture, and context and may serve protective functions. Existing CM–CID studies are scarce, based on small European samples, and focused only on strangers. This study aimed to: (1) replicate the association between CM and larger CID toward strangers; (2) test whether this extends to friends; (3) explore interaction effects by approaching individual (friend vs stranger); (4) evaluate generalizability across countries and CM subtypes; and (5) examine links between CID and social functioning (social support, strain, and attachment). Covariates included gender, social anxiety, depression, PTSD symptoms, and COVID-19-related anxiety.

Prior research documents cultural differences in interpersonal distance preferences (e.g., contact versus non-contact cultures; East Asian versus European regions) and person-related influences like trait anxiety. Limited CM–CID studies have reported larger CID in physically abused children and adults with mixed CM exposure, with mixed moderation by depressive symptoms. However, these studies were small and conducted in Croatia, Germany, and Switzerland, limiting generalizability. Moreover, they only examined CID toward strangers, leaving it unclear whether increased distances extend to familiar others. Broader literature shows CM is associated with altered social cue processing (e.g., faster recognition of negative expressions, negative interpretations of neutral faces, heightened amygdala reactivity), suggesting a generalized sensitivity to social threat that could influence CID. Attachment insecurity and lower social support—both linked to CM—have also been associated with altered CID in prior work.

Design and participants: Cross-sectional, multinational online study conducted October 2021–March 2022 via Qualtrics. As part of the Global Collaboration on Traumatic Stress, 3656 adults were recruited globally through contacts and online/social media; 2986 participants remained after predefined data-quality exclusions (69.2% female; mean age 31.27 years, SD 13.36). The study aimed for broad cultural and socioeconomic representation, with data from 43 countries and in multiple languages (Afrikaans, Arabic, Amharic, English, French, German, Hebrew, Japanese, Spanish, Swedish, Turkish, Xhosa). Ethics approvals were obtained, and procedures preregistered.

CID task: A validated computerized CID task was used. Participants viewed a circular room with a central self-avatar and an approaching stick-figure. They were instructed to press the spacebar when they first felt uncomfortable with the proximity. Trials alternated between imagining the approacher as a friend or a stranger. Eight approach angles were used; each figure appeared 48 times (six per angle). Eight speed-control trials were included to adjust for device timing. Each trial lasted 5 s to avoid speeded responding. The primary outcome was RT (minus mean control RT), where higher RTs reflect smaller preferred distances (smaller CID).

Measures: CM levels (Childhood Trauma Questionnaire; CTQ total and subscales), lifetime trauma exposure (LEC-5), PTSD symptoms (ITQ), social anxiety (MINI-SPIN), depression (PHQ-9), social support (MSPSS; significant others, family, friends), interpersonal stress (Bergen Social Relationships Scale; BSRS), attachment (ECR-S; anxiety and avoidance), and fear of COVID-19 (FCQ). Demographics included age, gender, country, SES.

Analytic strategy: Multilevel (ML) models accounted for the three-level nesting (trials within participants within countries) for Hypotheses 1–3, with L2 predictor CTQ total score, L1 predictor approaching individual (friend vs stranger), their interaction, and L3 country-level mean CTQ; covariates were gender, MINI-SPIN, ITQ, PHQ-9, and FCQ. Variables were group-mean centered appropriately. Models were fit sequentially: random intercepts (Model 1), +covariates (Model 2), +random slope for approaching individual (Model 3), +random slope for CTQ (Model 4); model selection used likelihood comparisons. For Hypothesis 4 (CTQ subscales) and Hypothesis 5 (social functioning), RTs were aggregated across trials (participants nested in countries). Hypothesis 4 tested separate ML models per CTQ subscale (given multicollinearity), including country-level subscale means and the same covariates. Hypothesis 5 used a random-intercept model with ECR-S subscales, MSPSS subscales, and BSRS as predictors. Models were fit using lme4 with REML; alpha = 0.05 two-sided.

Sample exposure and clinical characteristics: 67.31% reported any CM (emotional abuse 39.8%, physical abuse 20.0%, sexual abuse 29.3%, physical neglect 37.3%, emotional neglect 39.7%). 89.8% reported at least one potentially traumatic event. PTSD only: 7.04%; PTSD + C-PTSD: 9.76%. Mild–moderate depression: 50.5%; (moderately) severe depression: 12.4%. Social anxiety cut-off: 35.2%.

Hypotheses 1–3 (CTQ total, friend vs stranger): In multilevel models, observations were highly non-independent (ICC = 0.88). Adding covariates improved fit; adding a random slope for approaching individual further improved fit, but adding a random slope for CTQ did not (no substantial between-country variability in CTQ–CID association). Fixed effects explained 5.3% of variance in CID. Participants preferred larger distances for strangers than friends (β ≈ 0.21, p < 0.001). Higher CTQ scores were associated with larger CID (standardized β ≈ −0.08, p < 0.001; negative beta corresponds to larger CID given RT coding). The CTQ × approaching-individual interaction was not significant (β ≈ 0.003–0.00, p ≈ 0.40), with similar CTQ–CID associations for friends (simple slope B ≈ −0.0082, p < 0.001) and strangers (B ≈ −0.0077, p < 0.001). Contrary to prediction, higher social anxiety (MINI-SPIN) was linked to shorter preferred distances (smaller CID; β ≈ 0.06, p < 0.001, consistent with RT coding). Other covariates (gender, PHQ-9, ITQ, FCQ) showed no robust effects in the primary model.

Hypothesis 4 (CTQ subscales): Higher physical abuse (β ≈ −0.09, p < 0.001), sexual abuse (β ≈ −0.04, p = 0.019), emotional neglect (β ≈ −0.04, p = 0.031), and physical neglect (β ≈ −0.12, p < 0.001) were each associated with larger CID. Emotional abuse was not significant (β ≈ −0.03, p = 0.110). Effects were most pronounced for physical abuse/neglect.

Hypothesis 5 (social functioning): Less perceived support from significant others (MSPSS-SO; β ≈ 0.22, p < 0.001) and higher attachment anxiety (β ≈ −0.12, p < 0.001) and avoidance (β ≈ −0.09, p < 0.001) were associated with larger CID. Social support from family (β ≈ −0.04, p = 0.058) and friends (β ≈ −0.02, p = 0.243) and interpersonal strain (BSRS; β ≈ −0.03, p = 0.066) were not significant. Fixed effects explained 2.3% of variance.

Findings across a large, culturally diverse sample demonstrate that higher levels of childhood maltreatment are linked to larger comfortable interpersonal distances, not only toward strangers but also toward friends, suggesting a generalized elevation in preferred distance. The lack of improvement with a CTQ random slope indicates that the CTQ–CID association is broadly consistent across countries, pointing to potential universal mechanisms (e.g., CM-related alterations in threat-processing circuits such as the amygdala) rather than culture-specific factors. Larger CID among individuals with insecure attachment and with lower perceived social support suggests potential pathways to social dysfunction in CM survivors, possibly via heightened discomfort with proximity and social touch, which may undermine relationship quality and support. Contrary to several prior studies, higher social anxiety was associated with smaller distances in this virtual task, potentially reflecting greater perceived safety, monitoring/control motives, and the brevity of the MINI-SPIN; replication using clinical samples and richer measures is warranted. Subscale analyses imply that both abuse and neglect contribute to altered CID, with particularly strong links for the physical domain, while emotional abuse did not show a unique effect here. These findings highlight the importance of assessing and addressing interpersonal distance regulation in interventions for CM-affected individuals to support social functioning and well-being.

Childhood maltreatment is associated with altered regulation of comfortable interpersonal distance across cultures. Individuals with a history of CM prefer larger distances from both strangers and close others, and larger distances are also linked to insecure attachment and lower perceived social support. Given the central role of social relationships in mental health, targeting interpersonal distance regulation and perceived safety (e.g., via body-oriented approaches) may enhance preventive and therapeutic interventions for CM survivors. The cross-cultural consistency underscores the broad relevance of these mechanisms and the need for interventions adaptable across diverse settings.

• Unbalanced sample sizes across 43 countries despite multilevel modeling; future work should aim for more balanced country samples.
• Virtual/computerized CID task may limit ecological validity (no real persons, reliance on imagination, potential internet timing differences); individual differences in imagery ability may add error variance; effects may be smaller than in real-life paradigms.
• Did not manipulate or control the age/gender of the approaching figure; future VR paradigms could address this.
• No data to examine habituation across trials.
• COVID-19 fear-related measures showed low reliability; Mini-SPIN brevity may limit social anxiety assessment; results (e.g., social anxiety effect) should be replicated, including in clinical populations.
• Cross-sectional design precludes causal inference.