Negative social interpretation bias and suicidality: Indirect pathways through social anxiety and perceived burdensomeness

Abstract

Background

Social anxiety has been linked to suicidality through thwarted belongingness and perceived burdensomeness, consistent with the interpersonal psychological theory of suicide. Broader underlying cognitive vulnerabilities, such as biased perceptions of social situations, may also play a role. This study examined whether negative social interpretation bias predicted suicidality via social anxiety symptoms and unmet interpersonal needs (i.e., thwarted belongingness, perceived burdensomeness, and their interaction).

Methods

Undergraduate participants (N = 325) recruited from a large southern U.S. university completed online self-report measures and a social interpretation bias task. A path analysis tested indirect effects of social interpretation bias on suicidality through social anxiety and unmet interpersonal needs (i.e., thwarted belongingness, perceived burdensomeness, and their interaction). Depression symptoms were included as a covariate.

Results

Negative social interpretation bias was indirectly associated with suicidality via social anxiety symptoms and perceived burdensomeness, β = .070, p = .003, controlling for depression symptoms. The total indirect effect of interpretation bias on suicidality was significant (β = .068, p = .001), whereas the total effect was not (β = .041, p = .434). The direct effect of interpretation bias on suicidality was also not significant, β = -.028, p = .554.

Conclusions

These findings suggest that negative social interpretation bias is indirectly associated with suicidality through social anxiety and perceived burdensomeness. Together, these results are consistent with a pathway in which cognitive biases, affective distress, and interpersonal beliefs jointly contribute to suicidality. Notably, effects remained significant when depression symptoms were controlled, suggesting that social anxiety may have a unique association with suicide risk beyond depression. These results help clarify the link between social anxiety and suicidality, a relationship that remains relatively understudied.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-026-04897-5.

Introduction

Suicide is the second leading cause of death in the United States for individuals between the ages of 10 and 34 and the fourth leading contributor of years of potential life lost before 65 [21]. In addition to the gravity of suicide mortality, each year, more than 13 million adults in the United States suffer from thoughts of suicide [113]. Suicidal ideation and lower-lethality suicide-related behaviors (e.g., deliberate self-harm) are established predictors of future suicide attempts and mortality [40, 97, 99]. Thus, studying suicidal thoughts and lower-lethality suicidal behaviors has been an important component of understanding why people die by suicide [60, 67]. Despite the availability of both evidence-based interventions [59, 75, 111] and community resources [31, 44, 107] aimed at reducing suicidal ideation, suicide rates have continued to rise in recent years [43]. This underscores a need for a more comprehensive understanding of the factors that contribute to suicidal ideation and lower-lethality suicidal behaviors.

Interpersonal psychological theory of suicide

Recognizing that suicide results from multiple psychiatric, cognitive, interpersonal, and circumstantial factors, the interpersonal psychological theory of suicide (IPTS; [61, 118]) emerged as an influential and empirically supported model for understanding suicide risk. The IPTS proposes that unmet interpersonal needs, specifically thwarted belongingness and perceived burdensomeness, are two proximal predictors of suicidal desire. Within this framework, it is believed that other, distal factors can confer increased risk of suicide by exacerbating feelings of thwarted belongingness and perceived burdensomeness [118]. Although initially developed to explain lethal and near-lethal suicidal behavior, the IPTS has since been widely applied to the study of suicidal ideation [127, 32, 39, 57].

Conceptually, thwarted belongingness reflects a more intense and distressing form of social isolation than loneliness, representing a deeply felt, enduring lack of belonging, or the “sense that sustaining [social] connections are obliterated” [61], p. 120). Perceived burdensomeness is believed to emerge from unmet needs for social competence, [119], such as when one is unable to fulfill expected social roles (e.g., contributing to collective goals in one’s family or another group). If thwarted belongingness and perceived burdensomeness are regarded as global, enduring evidence of low social efficacy, a person may begin to view their existence as a liability to others. According to the IPTS, the co-occurrence of thwarted belongingness and perceived burdensomeness is considered the strongest proximal predictor of the desire to die by suicide [61, 118].

Cognitive vulnerabilities and suicidality

Importantly, interpersonal distress does not arise in isolation. Rather, it is shaped by underlying cognitive and emotional processes that increase susceptibility to distress in social contexts. One such process is negative interpretation bias, or the tendency to view ambiguous information in an excessively negative or threatening manner. Interpretation biases have been implicated in both affective disorders [23, 34, 50] and suicidal thoughts and behaviors [10, 13, 36]. These biases may be especially relevant to suicidality when they reflect distorted appraisals of social information. From this perspective, distal cognitive vulnerabilities and affective symptoms may increase suicide risk by amplifying the interpersonal distress emphasized in the IPTS.

To date, much empirical research on cognitive biases and suicidality has focused on depression [33, 34, 38, 58] and related cognitive vulnerabilities such as defeat, entrapment, and hopelessness [128, 11, 69, 89, 98, 124]. Hopelessness has been incorporated into integrated cognitive extensions of the IPTS [65, 66, 80, 115], highlighting how cognitive processes may exacerbate perceptions of thwarted belongingness and perceived burdensomeness. However, cognitive vulnerabilities extend beyond those classically associated with depression. Cognitive models of suicide [123] note that danger schemas, which are typically associated with anxiety symptoms, may contribute to suicide risk. Danger schemas refer to exaggerated beliefs about the likelihood of harm and the severity of its consequences. These schemas are thought to be activated in response to life stressors [7] and may increase negative cognitive content and bias information processing toward perceived threat [6, 73, 100].

Negative social interpretation bias and social anxiety

The content of anxiety-related danger schemas is believed to differ across disorders [73]. Among the anxiety disorders, social anxiety may be a particularly important predictor of suicidality, given its strong interpersonal focus. Social anxiety is characterized by negative self-referential schemas that reflect themes of social inadequacy, rejection, and negative evaluation [26, 95]. When activated in social-evaluative contexts, these schemas bias information processing toward signs of social threat, increasing interpersonal distress. This biased cognitive style has been associated with heightened subjective perceptions of social incompetence, with some studies indicating that individuals high in social anxiety often evaluate their social performance more negatively than independent observers do [82, 94, 112]. Other studies have found that observers also rate individuals with high social anxiety as less socially skilled or likeable [2, 46, 79], suggesting that social anxiety symptoms may also be linked with objective interpersonal impairment. Taken together, these findings suggest that biased social cognition may contribute to suicidality through its effects on social anxiety symptoms and subsequent interpersonal distress.

Both subjective self-appraisals and objective indicators of social performance are relevant to understanding how social anxiety may contribute to unmet interpersonal needs, such as feelings of thwarted belongingness or perceived burdensomeness. Negative interpretation biases, particularly the tendency to view ambiguous social cues as signs of rejection or failure, may exacerbate interpersonal distress. Internal awareness and judgments of anxious thoughts can heighten self-focused attention [62, 88], increasing vigilance for external cues of rejection. Further, biased information processing may lead one to construe minor social mishaps as catastrophic [85, 86], potentially amplifying feelings of incompetence. Empirical work has established the association between social anxiety and a particular type of interpretation bias: social interpretation bias. Experimental tasks involving sentence completion and written scenario interpretation have consistently shown that individuals with higher social anxiety symptoms are more likely to interpret ambiguous social situations in a negative or threatening manner [8, 22, 35, 48, 56]. Research suggests this effect is specific to social as opposed to non-social scenarios [120] and is especially pronounced for self-relevant versus other-relevant social information [5, 15]. This underscores the importance of social interpretation bias as a central cognitive mechanism implicated in the development and maintenance of social anxiety symptoms.

Social anxiety, interpersonal needs, and suicidality

While relatively understudied compared to mood disorders, an emerging body of research has demonstrated associations specifically between social anxiety and suicide [74, 122]. Although consistent associations have been observed between social anxiety symptoms and unmet interpersonal needs, empirical findings regarding the specific nature of these associations have been mixed, even when controlling for depression symptoms. Some studies have found that social anxiety is uniquely associated with thwarted belongingness, but not perceived burdensomeness, whereas depression has been linked to both interpersonal factors [30, 110]. In contrast, other research suggests that social anxiety is related to both thwarted belongingness and perceived burdensomeness [16, 25], thereby increasing risk for suicidal ideation. Other findings indicate that only perceived burdensomeness is associated with suicidal ideation in individuals with social anxiety disorder [32]. These discrepancies do not follow a clear pattern based on the measurement of social anxiety symptoms or the diagnostic status of the sample (e.g., nonclinical versus clinical). Taken together, these inconsistencies point to a need to refine theoretical models to account for variability in how interpersonal factors influence suicide risk.

Fewer studies have examined associations between interpretation bias and suicide. Negatively biased self-referential processing style has been associated with suicidality [10, 13]. Additionally, Everaert and colleagues [36] found that inflexible interpretations of ambiguous social scenarios involving themes of social failure and rejection were associated with increased perceived burdensomeness and, in turn, increased suicidal ideation. Other information processing biases have also been implicated in suicide risk. For instance, an enhanced late positive potential, a neural index of motivated attention, to images depicting social exclusion have been linked to higher levels of thwarted belongingness and, subsequently, to greater suicidal ideation [1].

The current study

Existing research provides preliminary support for the role of negative social interpretative bias, social anxiety, and unmet interpersonal needs in suicide risk. However, to date, no study has simultaneously examined these factors within the same model. As a result, it is still unclear how negative social information processing increases suicide risk through social anxiety and related interpersonal distress. To address this gap, the present work investigated whether negative social interpretation bias increased suicidality through social anxiety symptoms and unmet interpersonal needs. To address inconsistencies in terminology within the field [55], in the present study “suicidality” was defined as suicidal ideation and lower-lethality suicide-related behaviors (i.e., deliberate self-harm) as assessed by the Inventory of Depression and Anxiety Symptoms suicidality subscale. Using cross-sectional data, we tested a path model in which negative social interpretation bias was hypothesized to predict social anxiety symptoms, which in turn were expected to predict thwarted belongingness, perceived burdensomeness, and their interaction. These interpersonal need variables were then modeled as predictors of suicidality, such that the association between interpretation bias and suicidality was expected to occur indirectly through social anxiety and interpersonal needs. Although causal inference cannot be drawn from cross-sectional designs, cross-sectional path analysis is frequently used to investigate associations between cognitive processes and mental health outcomes (e.g., [37, 42, 101]) and can provide insight into whether the observed relationships are consistent with theoretical expectations. The current model specification was guided by theoretical accounts linking biased cognitive processing with social anxiety symptoms and suicidality, as well as by the IPTS, which emphasizes strong associations between unmet interpersonal needs and suicide risk. Accordingly, the model was specified to examine whether social anxiety and unmet interpersonal needs account for associations between negative social interpretation bias and suicidality.

Method

Participants

The sample consisted of 360 undergraduate students recruited from a large public university in the southern United States. We excluded 35 subjects (9.72%) from analyses because they did not correctly answer at least three of four attention check items, resulting in a final sample of 325. Participants reported a mean age of 19.56 years (SD = 2.18), and the majority identified as cisgender women (n = 185; 56.9%) and White (n = 278; 85.5%). Complete sample demographics are presented in Table 1.

Table 1.

Sample demographic information

Variables	M	SD
Age	19.56	2.18
Gender a	n	%
Cisgender Woman	185	56.9
Cisgender Man	134	41.2
Nonbinary	2	0.6
Agender	1	0.3
Transgender	1	0.3
Questioning	1	0.3
Race b	n	%
White	278	85.5
Black	24	7.4
Asian	14	4.3
American Indian/Alaskan Native	13	4.0
Middle Eastern/North African	5	1.5
Pacific Islander	3	0.9
Another Race Not Listed Here	8	2.5

N = 325. ^aOne missing value for gender. ^b Indicates groups are not mutually exclusive

Measures

Negative social interpretation bias

Social interpretation bias was measured using a modified version of the Brief Bodily Sensations Interpretation Questionnaire (BBSIQ; [27]). The original version of the BBSIQ presents participants with 14 items describing ambiguous events. Seven of these items describe sensations associated with autonomic symptoms of panic. The other seven items describe either general situations or social events. Participants are asked to rank order three potential explanations for each event based on the likelihood that they would consider the explanation, were they in a similar situation. One of the explanations represents a negative interpretation of the event. For instance, the negative explanation for, “You feel out of breath,” is, “You are about to suffocate or stop breathing.” The other two explanations for each event are either neutral or positive. To our knowledge, factor analysis has not been conducted on the BBSIQ, nor has overall internal consistency been reported. However, subsets of the BBSIQ based on item content have demonstrated good internal consistency in undergraduate samples [27, 29].

The current study modified the BBSIQ (see the Appendix for example items1) such that all 14 items describe ambiguous social situations. Of the three possible explanations for each event, one always represented a negative interpretation that is specific to concerns about social evaluation. For example, the negative interpretation for, “You find out that your friend did something with other people without inviting you,” is, “They did not want you to be there.” The other two explanations are benign. Participants rated the extent to which each of the three explanations would be true for them in the same situation, using an 8-point scale ranging from 0 (not at all likely) to 8 (extremely likely). A composite mean score for negative explanations was calculated, with higher scores indicating greater negative interpretations of ambiguous social information. Internal consistency of the modified BBSIQ was excellent in the present study, α = 0.94. This was comparable to the higher end of the range of internal consistencies of item subsets from the original measure, which ranged from 0.74–0.92 [27, 29].

Social anxiety

Social anxiety was assessed using the straightforward version of the Social Interaction Anxiety Scale (SIAS-S; [102]), which uses only the 17 straightforwardly worded items of the original 20-item SIAS [81]. Participants were asked to use a 5-point scale ranging from 0 (not at all characteristic or true of me) to 4 (extremely characteristic or true of me) to indicate the extent to which each statement (e.g., “I have difficulty talking with other people”) is true of them. Items were summed to create a total score, with higher scores indicating higher levels of social anxiety. The SIAS-S was scored continuously to capture the dimensional nature of social anxiety [104]. In the current study, 44% of participants (n = 143) had total scores at or above the recommended clinical cut-off (≥ 28, [102]. The SIAS-S has demonstrated good to excellent internal consistency in samples of undergraduate students [105, 109]. Internal consistency in the current study was excellent (α = 0.94).

Unmet interpersonal needs

Unmet interpersonal needs were measured using the Interpersonal Needs Questionnaire (INQ; [119]), a 15-item scale that assesses thwarted belongingness and perceived burdensomeness. The thwarted belongingness subscale consists of nine items (e.g., “These days, I feel like an outsider at social gatherings”), and the perceived burdensomeness subscale consists of six items (e.g., “These days, I think my death would be a relief to people in my life”). Participants rated the extent to which each item was true for them using a 7-point scale ranging from 1 (not at all true for me) to 7 (very true for me). Items within subscales were summed to create total scores, with higher scores indicating higher thwarted belongingness and perceived burdensomeness. The INQ has demonstrated excellent internal consistency in undergraduate samples [12, 115]. In the current study, internal consistency was excellent for perceived burdensomeness (α = 0.95) and good for thwarted belongingness (α = 0.88).

Suicidality

Suicidality was assessed using the six-item suicidality subscale of the Inventory of Depression and Anxiety Symptoms (IDAS-SS; [121]), a validated measure that is appropriate for use in non-clinical samples (e.g., [9, 16, 18]. Participants used a 5-point scale ranging from 1 (not at all) to 5 (extremely) to endorse the extent to which they experienced suicidal thoughts and self-harm behaviors (e.g., “I had thoughts of suicide” and “I cut or burned myself on purpose”) during the most recent two-week period. Therefore, suicidality in the current study is conceptually distinguished from attempt history and highly lethal suicide-related behavior (e.g., interrupted attempts). There are presently no empirically supported clinical cut-points for the IDAS-SS. In the current sample, scores were positively skewed, with a floor effect (median = 6,25th percentile = 6,75th percentile = 9). A total of 166 participants (51.08%) endorsed no suicidality (IDAS-SS total score = 6), whereas 155 participants (47.69%) reported some level of suicidality (IDAS-SS total score > 6). The distribution of IDAS-SS scores was consistent with meta-analytic findings estimating that approximately 45% of college students report past-year suicidal ideation when measures distinguish suicidal ideation from morbid ideation [84]. The IDAS-SS has demonstrated good internal consistency in undergraduate samples [121, 126]. In the current study, internal consistency was good, (α = 0.89).

Depression

All participants were administered the 9-item version of the Patient Health Questionnaire (PHQ-9; [70]). In the present sample, 50.78% of participants (n = 163) had PHQ-9 total scores at or above the recommended clinical cutoff (≥ 10, [70]. In primary analyses, depression symptoms were assessed using the 8-item version of the Patient Health Questionnaire (PHQ-8, [71]), which is identical to the PHQ-9 with the exception that it excludes the ninth item, which measures suicidal ideation. This approach was used to avoid content overlap with the IDAS-SS. Participants used a 4-point scale ranging from 0 (not at all) to 3 (nearly every day) to rate the frequency with which they experienced depression symptoms during the most recent two-week period. Items were summed to create a total score, with higher scores indicating more severe depression symptoms. The PHQ-8 has demonstrated good reliability in undergraduate samples [4], and internal consistency was excellent in the present study (α = 0.90).

Procedure

All study measures and procedures were approved by the Institutional Review Board at the participating university. Participants were recruited via the psychology department online research participation pool, which includes undergraduate students enrolled in psychology courses with a research participation requirement. To be eligible for the study, participants had to be at least 18 years old and able to read and respond in English, as all study materials were presented in English. There were no additional inclusion or exclusion criteria.

All study procedures were conducted using an online data collection platform. Participants first completed an electronic informed consent document that was saved separately from questionnaire data to deidentify responses. After providing consent, participants were redirected to the online survey. Measures were administered in the same order across participants. Responses were collected anonymously and identified only by randomly generated codes within the data collection platform. Participants received course participation credit upon completion of the study.

Given the anonymous nature of data collection, direct risk assessment and intervention were not possible. Suicidality was assessed using a measure of lower-lethality thoughts and behaviors (i.e., the IDAS-SS), as opposed to measures assessing suicidal intent, planning, or attempt history. This decision was guided by both ethical considerations and the study’s theoretical focus on lower-lethality suicidality in a nonclinical undergraduate sample. To address potential distress or elevated risk associated with survey completion, all participants were provided with referral resources for mental health services (e.g., on-campus providers and national crisis hotlines), as well as contact information for the primary investigator.

Data analysis

Descriptive statistics and model assumptions

Data were analyzed using path analysis models conducted with the lavaan package (version 0.6.19; [103]) in R (version 4.5.1, [92]). Bivariate correlations, means, and standard deviations for all study variables are reported in Table 2. The assumptions of linearity, multicollinearity, and normality were evaluated prior to model estimation. Linearity and multicollinearity were considered acceptable for all variables. The assumption of normality was violated for perceived burdensomeness, the interaction of thwarted belongingness and perceived burdensomeness (TB × PB), and suicidality. Model parameters were estimated using robust maximum likelihood estimation (MLR), which is robust to violations of normality.

Table 2.

Bivariate correlations, means, and standard deviations

Variables	1	2	3	4	5	6
1. Interpretation Bias	–
2. Social Anxiety	.58	–
3. Thwarted Belongingness	.44	.47	–
4. Perceived Burdensomeness	.47	.45	.53	–
5. Suicidality	.36	.36	.42	.77	–
6. Depression	.51	.48	.48	.58	.52	–
N	275	325	320	320	321	320
Mean	3.25	25.60	26.13	10.19	8.23	9.78
Standard Deviation	1.68	14.49	12.29	7.05	3.87	7.05

Composite scores were computed when at least 80% of scale items were completed

Correlations are Pearson correlations based on pairwise complete observations. Therefore, sample sizes vary slightly across correlations. All correlations were significant at p <.001

Missing data handling and outlier detection

Data were examined for missingness2

Composite scores were calculated only for cases with at least 80% item completion for a given scale; scores were otherwise treated as missing. Full information maximum likelihood (FIML) estimation was used to handle missing data.

Influential cases were identified using Mahalanobis distance3

Sensitivity analyses of the primary model were conducted excluding these cases. The pattern and significance of structural paths were unchanged when multivariate outliers were excluded. Global model fit and indirect effects were substantively identical. Therefore, all cases were retained for primary analyses.

Primary path model specification

A path model was specified to examine whether the association between negative social interpretation bias and suicidality was sequentially mediated by social anxiety and, in turn, by unmet interpersonal needs. Specifically, negative social interpretation bias was hypothesized to predict social anxiety, which was expected to predict thwarted belongingness, perceived burdensomeness, and their interaction (TB × PB). These three interpersonal needs variables were modeled as predictors of suicidality. The model included three indirect pathways from interpretation bias to suicidality, each via social anxiety and one of the interpersonal needs variables (i.e., thwarted belongingness, perceived burdensomeness, or TB × PB). A direct path from interpretation bias to suicidality was included to estimate the association between these variables independent of the indirect pathways through social anxiety and unmet interpersonal needs. Depression symptoms (PHQ-8 scores) were included as a covariate predicting all endogenous variables to control for depression severity. To account for shared variance among predictors, the errors terms of thwarted belongingness, perceived burdensomeness, and their interaction term were allowed to covary.

Alternative path model specifications

Two planned alternative models were also estimated. First, a model replacing the PHQ-8 with the PHQ-9, which includes a suicidal ideation item, was estimated to evaluate whether results were sensitive to the inclusion of suicide-related content in the depression covariate. Second, a model excluding depression as a covariate was estimated to examine whether the primary pattern of associations remained consistent when depression symptoms were not controlled for.

Sensitivity analyses

To evaluate the detectability of indirect effects in the specified model, sensitivity analyses using Monte Carlo simulation were conducted to evaluate the detectability of indirect effects across plausible parameter values, defined as standardized path coefficients and associated variability implied by the fitted model, with focal direct paths systematically varied across small-to-moderate values (β = 0.01 to 0.40 in 0.01 increments). Each condition was evaluated using 20,000 Monte Carlo draws per simulated study, with detectability estimated across 1,000 simulated studies.

Model fit evaluation

Model fit was evaluated using the chi-square test statistic (χ²), the Comparative Fit Index (CFI), the Tucker–Lewis Index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). Because the MLR estimation was used, robust (scaled) values are reported for each of these indices. Consistent with conventional guidelines [54], CFI and TLI values ≥ 0.95, RMSEA ≤ 0.06, and SRMR ≤ 0.08 were considered indicative of good model fit. The χ² statistic was interpreted in conjunction with these other indices, with recognition that it is sensitive to sample size and is often significant in larger samples (e.g., N ≥ 200–300, [68].

Model comparison criteria

Model comparisons were conducted using the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC), with lower values indicating better relative fit. According to [17], comparing differences in AIC values (ΔAIC) can assess relative fit: ΔAIC ≤ 2 suggests that both models have substantial support, ΔAIC ≤ 4 indicates considerably less support for the higher-scoring model, and ΔAIC > 10 suggests essentially no support for the higher-scoring model. For BIC, differences of 0–2 represent weak evidence against a competing model, values of 2–6 indicate positive evidence, 6–10 indicate strong evidence, and values greater than 10 indicate very strong evidence against a competing model [93].

Results

Primary model

The primary model tested the indirect association between negative interpretation bias and suicidality through social anxiety and unmet interpersonal needs (see Fig. 1). The model demonstrated acceptable fit to the data, χ²(4) = 11.07, p = 0.026; CFI = 0.991; TLI = 0.954; RMSEA = 0.082; SRMR = 0.027. The direct effect of negative social interpretation bias on social anxiety was significant, β = 0.445, p < 0.001. In turn, social anxiety significantly predicted thwarted belongingness (β = 0.309, p < 0.001) and perceived burdensomeness (β = 0.218, p < 0.001). Social anxiety did not predict the TB × PB interaction term, β = 0.040, p = 0.532).

Fig. 1.

Path model examining indirect pathways from interpretation bias to suicidality. IB = interpretation bias; SA = social anxiety; TB = thwarted belongingness; PB = perceived burdensomeness; TB × PB = interaction of thwarted belongingness and perceived burdensomeness; SUI = suicidality. Depression symptoms (DEP) were included as a covariate of all endogenous variables (dashed paths). Curved double-headed arrows indicate residual covariances. Values are standardized path coefficients (β). **p <.01

Among the interpersonal needs constructs, only perceived burdensomeness significantly predicted suicidality, β = 0.716, p < 0.001. Thwarted belongingness (β = −0.008, p = 0.866) and the TB × PB interaction (β = 0.003 p = 0.978) were not significant predictors of suicidality.

Standardized indirect effects and their confidence intervals are shown in Fig. 2. There was a significant indirect effect of negative social interpretation bias on suicidality through social anxiety and perceived burdensomeness β = 0.070, p = 0.003. In contrast, indirect pathways via social anxiety and thwarted belongingness (β = −0.001, p = 0.866) and via social anxiety and the TB × PB interaction (β < 0.001, p = 0.977) were not significant.

Fig. 2.

Standardized indirect effects in the primary model (PHQ-8 covariate). Points represent standardized indirect effect estimates (β), and horizontal bars indicate 95% bootstrap confidence intervals. Indirect effects are shown for pathways from interpretation bias (IB) to suicidality through social anxiety (SA) and (a) perceived burdensomeness (PB), (b) thwarted belongingness (TB), and (c) the TB × PB interaction

The direct effect of negative social interpretation bias on suicidality was not significant, β = −0.028, p = 0.554. The indirect effect through social anxiety and perceived burdensomeness was significant, β = 0.070, p = 0.003. In contrast, indirect pathways through social anxiety and thwarted belongingness (β = −0.001, p = 0.866) and through social anxiety and the TB × PB interaction (β < 0.001, p = 0.977) were not significant. The total indirect effect of interpretation bias on suicidality was significant (β = 0.068, p = 0.001), whereas the total effect was not (β = 0.041, p = 0.434). Taken together, this indicates that the association between interpretation bias and suicidality operated through the indirect pathway involving social anxiety and perceived burdensomeness.

Depression symptoms (PHQ-8) were included as a covariate and were positively associated with social anxiety, interpersonal needs variables, and suicidality. Importantly, the indirect effect of negative social interpretation bias on suicidality through social anxiety and perceived burdensomeness remained significant when controlling for depression, indicating that social anxiety contributed uniquely to suicidality beyond shared variance with depression symptoms.

Sensitivity analyses for indirect effects

Sensitivity analyses using Monte Carlo simulation were conducted to evaluate the detectability of indirect effects based on estimated model parameters. Results indicated that the indirect pathway involving perceived burdensomeness would be detectable at conventional power levels (≥ 0.80) when the perceived burdensomeness → suicidality pathway was approximately β = 0.10. Detectability of indirect effects involving thwarted belongingness or the TB × PB interaction would have required direct TB → suicidality and TB × PB → suicidality paths of approximately β = 0.04 or larger. The observed direct effects for thwarted belongingness and the TB × PB interaction were substantially smaller than these thresholds. These results suggest that the absence of indirect effects involving thwarted belongingness or the TB × PB interaction is unlikely to reflect insufficient statistical sensitivity but instead reflects the very small magnitude of the corresponding direct paths to suicidality.

Alternative models

Two planned alternative models were also estimated. Detailed results are presented in Table 3. A model replacing the PHQ-8 with the PHQ-9, which includes a suicidal ideation item, was estimated to evaluate whether results were sensitive to the inclusion of suicide-related content in the depression covariate. Fit indices were acceptable and comparable to those of the primary model, χ²(4) = 9.69, p = 0.046; CFI = 0.992; TLI = 0.959; RMSEA = 0.066; SRMR = 0.023. The direction and significance of both direct and indirect effects was consistent with that observed in the primary model. AIC and BIC favored the primary model (AIC PHQ-8 = 12,198.11, AIC PHQ-9 = 12,205.31, ΔAIC = 7.20; BIC PHQ-8 = 12,315.41, BIC PHQ-9 = 12,322.61, ΔBIC = 7.20), indicating substantially better relative support for the model including the PHQ-8 covariate.

Table 3.

Standardized direct, indirect, and total effects across primary and alternative models

	Primary Model(PHQ-8 Covariate)		Alternative Model(PHQ-9 Covariate)		Alternative Model(No Covariate)
Direct Effects	β	p	β	p	β	p
IB → SA	.445	<.001	.440	<.001	.575	<.001
SA → TB	.309	<.001	.298	<.001	.466	<.001
SA → PB	.218	<.001	.188	<.001	.447	<.001
SA → TB × PB	.040	.532	.015	.807	.190	.019
TB → SUI	-.008	.866	-.016	.731	.012	.787
PB → SUI	.716	<.001	.684	<.001	.768	<.001
TB × PB → SUI	.003	.978	.004	.967	-.003	.971
IB → SUI	-.028	.554	-.042	.379	< -.001	.993
Indirect Effects	β	p	β	p	β	p
IB → SA → TB → SUI	-.001	.866	-.002	.732	.003	.786
IB → SA → PB → SUI	.070	.003	.057	.007	.198	<.001
IB → SA → TB × PB → SUI	<.001	.977	<.001	.965	< -.001	.971
Total Indirect	β	p	β	p	β	p
	.068	.001	.054	.004	.201	<.001
Total Effect	β	p	β	p	β	p
	.041	.434	.012	.811	.200	<.001

Values are standardized path coefficients (β)

IB Interpretation bias, SA Social anxiety, TB Thwarted belongingness, PB Perceived burdensomeness, SUI Suicidality

A second alternative model excluding depression as a covariate was estimated to examine whether the observed pattern of associations depended on controlling for depression symptoms. Model fit was substantially poorer compared to the primary model, χ²(4) = 26.76, p < 0.001; CFI = 0.959; TLI = 0.845; RMSEA = 0.132; SRMR = 0.063. Although the overall pattern of direct and indirect effects was similar, path coefficients were generally larger in magnitude when depression was not included in the model, and the total effect of interpretation bias on suicidality became significant. These results suggest that depression accounts for variance shared with suicidality and that controlling for depression provides a more conservative estimate of the relationships among interpretation bias, social anxiety, unmet interpersonal needs, and suicidality.

Discussion

The purpose of this study was to explore whether negative social interpretation bias increased suicidality through social anxiety symptoms and unmet interpersonal needs (i.e., thwarted belongingness, perceived burdensomeness, and their interaction). There was a significant indirect effect of interpretation bias on suicidality through social anxiety and unmet interpersonal needs. Among the interpersonal constructs, only perceived burdensomeness predicted suicidality; thwarted belongingness and its interaction with perceived burdensomeness did not. Importantly, model modification indices supported the inclusion of direct paths from negative social interpretation bias to thwarted belongingness, perceived burdensomeness, and TB × PB. Interpretation bias exerted both direct and indirect effects on these interpersonal risk factors. This suggests that social information processing biases may increase suicide risk both by heightening social anxiety, which undermines interpersonal functioning, and by independently exacerbating more proximal interpersonal vulnerabilities.

Our findings are consistent with prior research examining the associations between interpretation bias and suicide [10, 13, 36] and support cognitive-interpersonal models of suicide risk (e.g., [65, 66, 115]. Much of the existing integrative cognitive-interpersonal research has focused on depression symptoms and depression-related cognitions. Although prior students have established associations between social anxiety, unmet interpersonal needs, and suicide-related outcomes (e.g., [16, 25, 32]), ours is the first to integrate these constructs within a model that simultaneously accounts for cognitive risk factors. The current work demonstrates the importance of considering social anxiety and associated information processing biases in the framework of the IPTS.

With respect to the interpersonal constructs, our results align with research indicating that perceived burdensomeness is the most consistent and robust predictor of suicide risk [24, 77], as only perceived burdensomeness predicted suicidality in the current sample. This suggests that perceived burdensomeness remains the more influential interpersonal predictor of suicidality, even when accounting for social anxiety symptoms, which have, in some cases, been uniquely linked to thwarted belongingness [30, 110]. The comparatively weaker effect of thwarted belongingness observed across studies has raised questions about whether the construct is adequately assessed by the Interpersonal Needs Questionnaire (INQ,e.g., [49]). Although a new scale specifically examining thwarted belongingness (i.e., the Thwarted Belongingness Scale, [78] has been developed, initial validation suggested that high perceived burdensomeness may predict suicidal ideation regardless of thwarted belongingness levels.

Meta-analytic findings (see [24]) indicate that thwarted belongingness may interact with capability for suicide (i.e., habituation to pain and fear that reduces self-preserving behavior, [61] to predict suicide attempt history, which suggests that the effects of thwarted belongingness could be context-dependent and predictive of higher lethality suicidal behavior. However, few studies of suicidal ideation, including those reviewed in the current manuscript, assess capability, potentially because it is considered a predictor of lethal attempts as opposed to ideation or lower lethality suicidal behavior. Further, research examining the association between capability for suicide and mental disorders is scant, although limited evidence suggests capability may be related to posttraumatic stress disorder, certain substance use disorders, and schizophrenia spectrum disorders [110]. It is therefore possible that unmeasured differences in capability for suicide partially account for the inconsistent relationships between unmet interpersonal needs and suicide risk observed across studies. Additional work is needed to clarify these relationships, including in the context of mental health symptoms and disorders.

Another possibility is that perceived burdensomeness may involve global self-referential cognitions that are difficult to modify, a process further complicated by social anxiety. Cognitive biases generally involve belief-consistent information processing [90]. Cognitive models propose that social anxiety is maintained by maladaptive self-beliefs [26, 47, 51]. Such beliefs resist disconfirming evidence, as individuals often prefer information consistent with preexisting self-conceptions, even when negative [114]. Consistent with this, socially anxious individuals have been shown to perceive positive social feedback as less accurate [53] and to prefer negative social feedback even when their fear of negative evaluation is high [117]. Perceived burdensomeness may better capture the global, negative self-appraisals that stem from social anxiety and biased social information processing (e.g., “These days, the people in my life would be better off if I were gone”). Behavioral outcomes such as social avoidance may further limit opportunities for social integration and contribution, compounding perceptions of burdensomeness. Importantly, social integration difficulties are known to contribute to suicidality [106] and may specifically serve as a pathway through which anxiety and depression symptoms increase suicidal ideation [87].

In contrast, thwarted belongingness may reflect more transient or context-dependent states (e.g., “These days, I feel disconnected from other people”). Supporting this distinction, Everaert et al. [36] found inflexible negative social interpretations best predicted suicidal ideation via perceived burdensomeness. This is consistent with the IPTS, which posits that the perceived intractability of burdensomeness and thwarted belongingness contributes to suicidal desire [61]. Therefore, the observed effects of interpretation bias on perceived burdensomeness may reflect the congruency of social information processing biases with burdensomeness, particularly in the context of social anxiety. Psychometric evidence also indicates that the thwarted belongingness subscale is not unidimensional [116] and may divide into the constructs of social disconnection and need for belonging [28]. Whereas disconnection may fluctuate across contexts, disrupted beliefs about acceptance may represent more stable self-referential cognitions. Distinguishing between these transient and stable components may clarify how interpretation biases, social anxiety, and thwarted belongingness contribute to suicide risk.

Our findings have several important clinical implications. The current study suggests that the relationship between social anxiety and perceived burdensomeness partially explains the association between negative social interpretation bias and suicidality. Therefore, it is important to examine how information processing biases influence proximal suicide risk factors, both within social anxiety and more broadly. Although psychiatric disorders are associated with increased risk of suicide [125], an estimated 20% of individuals who attempt suicide do not meet criteria for a mental health condition [91]. Our findings suggest that assessing both mental health and perceptions of interpersonal functioning may improve assessment of suicide risk.

With respect to treatment, there is growing interest in using cognitive bias modification (CBM) interventions to revise biased information processing in the affective disorders, but questions remain regarding their ability to meaningfully modify entrenched cognitive patterns. A study examining the effects of CBM specifically on unmet interpersonal needs [3] found that treatment reduced perceptions of burdensomeness but not thwarted belongingness. However, the active intervention included both CBM and psychoeducation, which makes it difficult to isolate the specific effect of CBM. Further, the small effect sizes reported by CBM clinical trials for affective disorders broadly [45], and social anxiety disorder specifically [76], point to a need for more reliable interventions. The dynamic interrelationship between negative affect and higher-order cognitive processing (see [14] for a review) likely impedes the revision and updating of negative social interpretation bias and interpersonal beliefs such as perceived burdensomeness. Given that the inflexibility of these beliefs has been found to increase suicidality, Everaert and colleagues [36] have suggested that CBM paradigms targeting perceived burdensomeness may be strengthened by incorporating skills from metacognitive training for psychosis (e.g., reducing overconfidence in cognitive errors and biases against disconfirming evidence, [83]. In the case of social anxiety and other affective disorders, increasing metacognitive flexibility related to disorder-specific beliefs may prove especially helpful.

Additionally, the results highlight a need to screen for suicide risk in patients with social anxiety disorder. Although less studied than depression, there is increasing evidence that social anxiety is associated with increased risk of suicide [74, 122]. The current study adds to the existing literature by confirming the link between social anxiety symptoms and suicidality, while also clarifying the cognitive and interpersonal processes that may underlie this relationship. This increased mechanistic understanding of social anxiety and suicide has the potential to refine treatment efforts beyond CBM. In addition to excessively negative self-appraisals, social anxiety is also characterized by lower observer-rated social skills [2, 46, 79], suggesting that avoidance-based behavioral strategies (e.g., limited self-disclosure and restricted emotional expression) may undermine the need for belonging and social competence. Behavioral approaches for co-occurring social anxiety disorder and suicidal ideation may benefit from identifying and modifying self-limiting social behaviors to improve interpersonal fulfillment.

This study is not without its limitations. First, inferences of causality cannot be made due to the cross-sectional nature of the design. Second, the current study did not examine the intractability of the participants’ endorsements of perceived burdensomeness and thwarted belongingness. This is conceptually important, given that the intractability of these beliefs is theorized to contribute to the transition from suicidal ideation to active suicidal desire [61]. Although this limitation does not affect the current findings, which focused on lower-lethality indicators of suicide risk, future research is necessary to assess belief inflexibility as a mechanism. Third, because the social interpretation bias measure was derived from a modified version of the Brief Bodily Sensations Interpretation Questionnaire, direct comparison with prior studies is limited. However, the scale demonstrated excellent internal consistency in the present research (α = 0.94). More research is needed to understand the factor structure of the original measure and to compare it to our revised version. Fourth, it is acknowledged that interpersonal distress may stem from multiple factors beyond social anxiety symptoms, including individual differences in social functioning and related traits that were not assessed in the current study. Finally, subjects were undergraduate students (M_age = 19.56 years) and predominately White. Although this demographic positioned us to examine lower-lethality indicators of suicidality in a non-clinical population, a needed area of research [60], the sample limits the generalizability of findings to other populations, particularly older and more diverse groups. Suicide is one cause of the so-called “deaths of despair” [19, 20], which are associated with social factors relevant to the IPTS, including social disconnection, unemployment, poverty, and perceived status loss [64, 96, 108]. Such factors have been uniquely linked to increased midlife mortality for non-Hispanic White Americans [19, 41]. This highlights the need to examine self-referential cognitive processes, affective disorders, and interpersonal vulnerabilities in this sociodemographic cohort. At the same time, suicide risk processes may differ for non-White populations. There is evidence that experiences of racial prejudice and microaggressions may increase suicide risk via perceived burdensomeness [52, 63, 72]. Thus, research is needed to clarify the specific cognitive and affective mechanisms implicated in interpersonal suicide risk in ethnically and racially diverse groups.

Conclusions

The present study identified a significant indirect association between negative social interpretation bias and suicidality through social anxiety and perceived burdensomeness. Taken together, our findings highlight the importance of considering information processing biases, social anxiety symptoms, and interpersonal vulnerabilities when examining suicide risk. By demonstrating that negative social interpretation bias contributes to suicidality through social anxiety and perceived burdensomeness, this study integrates cognitive and interpersonal models in a novel way. Although limited by its cross-sectional design and sample characteristics, the study provides preliminary evidence that interpretation biases may serve as a cognitive vulnerability that exacerbates social anxiety symptoms, undermines interpersonal functioning, and amplifies perceptions of burdensomeness. Future longitudinal and experimental work is needed to clarify temporal sequencing, assess the malleability of these processes, and extend findings to more diverse populations.

Abbreviations

IPTS

Interpersonal psychological theory of suicide

TB × PB

Interaction term of thwarted belongingness and perceived burdensomeness

Authors’ contributions

**HCH** contributed to study conceptualization and methodology and was responsible for data curation, formal analysis, and drafting the original manuscript. **AA, CGC,** and **MSM** contributed to preparation of the original manuscript. **NMH** provided substantive review and editing of the manuscript. **MRJ** contributed to conceptualization and methodology and was responsible for research supervision and manuscript review and editing. All authors read and approved the final manuscript.

Funding

Open access publication of this work was supported by the Open Access Publishing Fund administered through the University of Arkansas Libraries.

Data availability

Data analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Institutional Review Board at the University of Arkansas (Protocol # 2102318113).

Informed consent was obtained from all participants prior to beginning the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.