Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Nov 26.
Published in final edited form as: Psychiatry Res Commun. 2025 Jul 12;5(3):100220. doi: 10.1016/j.psycom.2025.100220

Physiological and cognitive reactivity to social exclusion among emerging adults with versus without a suicide attempt history

Beverlin Rosario-Williams a,b,c,1, Jorge Valderrama b,d,1, Evan Gilmer a,e, Florissell Rosales a, Regina Miranda a,b,*
PMCID: PMC12646608  NIHMSID: NIHMS2111931  PMID: 41306189

Abstract

Background:

Understanding cognitive responses to stress among individuals at risk for suicide attempts may help identify intervention targets to decrease the risk of future attempts. We examined differences between individuals with and without a suicide attempt history in physiological and cognitive responses to social exclusion.

Methods:

Emerging adults with (n = 37) and without (n = 39) a suicide attempt history were assigned to a social exclusion or control (inclusion) condition. Saliva samples were taken before and after the stressor to measure salivary cortisol. Participants then completed behavioral measures of impulsivity, problem solving, and semantic interference from suicide-related words.

Results:

There were no differences in cortisol reactivity trajectories by suicide attempt history, irrespective of stress condition. There was a trend for individuals with a suicide attempt history to show less semantic interference from suicide-related stimuli, compared to those without a suicide attempt history, regardless of stress condition. Furthermore, there was a trend for individuals who experienced social exclusion to perform better on the Tower of London test (a measure of problem solving) if they had a suicide attempt history than if they had no prior suicide attempt history. There were no other group differences on cognitive measures.

Conclusions:

Emerging adults with a suicide attempt history who are not in an acutely suicidal state appear to demonstrate similar physiological and cognitive responses to social exclusion as do emerging adults without a suicide attempt history, and in some cases, may show improved problem solving. Findings are contextualized within the broader literature.

Keywords: Suicide attempt, Stress, Cortisol, Cognitive bias

1. Introduction

Emerging adulthood is the time of adulthood when individuals are at the highest risk of suicide attempts (SAMHSA, 2024). Interpersonal stressors are common precipitants of suicide attempts (SAs) among emerging adults from diverse backgrounds (Rosario-Williams et al., 2022). Research examining the role of stress reactivity in the risk of SAs tends to focus on physiological reactivity (Miller and Prinstein, 2019). However, emerging research has highlighted the need to examine reactivity across multiple levels of analysis, including cognitive systems (Bendezú et al., 2022). Several theories of suicide (e.g., Cha et al., 2010; Miller and Prinstein, 2019) draw upon the stress-diathesis model of suicide, which considers the role of stress in the development of suicidal behavior as resulting from dynamic interactions between dispositional factors and acute stressors (Mann et al., 1999). For instance, underlying genetic, medical, or neuroendocrinological vulnerabilities may interact with recent stressful life events—particularly acute interpersonal stress (e.g., peer victimization, interpersonal loss)—increasing the likelihood of engaging in suicidal behavior (Auerbach et al., 2021; Mann et al., 1999; Wang et al., 2012). Therefore, using a biopsychosocial framework and drawing from the stress-diathesis model of suicide (Mann et al., 1999), we investigated how individuals with different histories of suicide attempts differ on physiological response to stress, executive functioning, and cognitive processing after experiencing an interpersonal stressor.

1.1. Stress reactivity, executive functioning, and suicidal behavior

Stress reactivity is implicated in both physical and psychological health (Reynolds, 2013). Cortisol secretion has been the gold standard measure of stress reactivity, because cortisol is secreted when the hypothalamic-pituitary-adrenal axis (HPA), a primary stress pathway, is activated. Reactivity to psychosocial stress is implicated in risk for suicide attempts (Eisenlohr-Moul et al., 2018). However, research examining physiological stress reactivity and suicidal behavior has been mixed. Some studies suggest that a history of SA is associated with blunted salivary cortisol following a physiological stressor (Melhem et al., 2016; O’Connor et al., 2018), other research suggests a link between hyperresponsiveness to cortisol and suicide ideation (Giletta et al., 2015), and still other research suggests no relationship between SA history and cortisol reactivity (Stanley et al., 2019). Among individuals with previous suicidal behaviors, cortisol hypersecretion has been associated with deficits in executive functioning, including poor working and long-term memory, deficits in cognitive set shifting/mental flexibility, and selective attention (Hinkelmann et al., 2009; Otte et al., 2007; Wolf, 2003). At the same time, meta-analyses suggest that adults with histories of SA show poorer decision-making, attentional control, and response inhibition than adults with suicide ideation but no history of SA (Escobar et al., 2024; Saffer and Klonsky, 2018). However, less information is known about the role that reactivity to social exclusion plays on executive functioning among emerging adults with (vs. without) a previous SA.

1.2. Disrupted cognitive processing and suicidal behavior

Understanding how specific cognitive processes relate to suicidal behavior has been important in assessing future suicide-related risk. Individuals with a history of SA may have biased information processing that can activate suicide-related schemas. Schemas are cognitive structures that give meaning to internal or external stimuli (Bartlett, 1932). For example, a suicidal schema may involve feelings of hopelessness when a person is experiencing a stressful life event or when a depressed mood is present, and suicidal schemas may trigger suicide ideation and eventually a suicide attempt (Wenzel and Beck, 2008).

Previous research has assessed suicide-related schemas by examining semantic interference from suicide-related cognitions via modified versions of cognitive tasks such as the Stroop task (Stroop, 1935), which measures the latency with which individuals name the color of a given word. The Suicide Stroop, an adaptation of the Emotional Stroop Task (Williams and Broadbent, 1986), found greater interference in indicating the color of a word among patients with a previous SA when the word was suicide-related than when it was neutral (Becker et al., 1999; Cha et al., 2010). Furthermore, interference from suicide-related words predicted SAs in a 6-month follow-up even after adjusting for history of mood disorder, previous SAs, and severity of suicidal thoughts (Cha et al., 2010). By contrast, interference from suicide-related words did not vary by SA history in a racially diverse sample of 736 college students (Chung and Jeglic, 2016). Additional research has also failed to replicate previous Suicide Stroop findings (Mandel et al., 2025; Wilson et al., 2019). These mixed results may be due to differences in study methods or to the absence of an acute negative mood among participants. The cognitive model of suicide implies that vulnerable individuals currently experiencing distress may exhibit cognitive biases (Wenzel and Beck, 2008), suggesting that an acute stressor and negative affect are necessary to detect biased cognitive processing. Given that stress reactivity is associated with executive functioning and cognitive processing, we examined the association between stress reactivity and cognitive biases after experiencing an implicit psychosocial stressor (i.e., social exclusion) for individuals with and without a history of SA.

1.3. The present study

Previous research and theory suggest that interpersonal stressors are common precipitants of SAs among emerging adults (Rosario-Williams et al., 2022), that the circumstances of previous SAs become part of individuals’ suicide-related schemas (Rudd, 2000; Wenzel and Beck, 2008), and that activation of these schemas might be ascertained by implicit measures such as the Suicide Stroop (Cha et al., 2010). Drawing from the stress-diathesis model, which suggests that external stressors (e.g., social conflict) interact with predisposing vulnerabilities to increase vulnerability for suicidal behaviors (Mann et al., 1999), we would expect that emerging adults with a history of SA would experience greater physiological and cognitive reactivity to a social/interpersonal stressor, compared to those without a history of SA. Thus, the present study had two main objectives. First, we investigated the effect of social exclusion – as a social/interpersonal stressor – on physiological stress reactivity among emerging adults with and without a history of SA. We hypothesized that emerging adults with a history of SA who experienced social exclusion (vs. inclusion) would have higher levels of cortisol reactivity over time relative to adults with a history of SA who did not experience social exclusion or adults without a previous SA. Second, we examined differences in executive functioning and activation of suicide-related cognitions between emerging adults with and without a history of SA following a psychosocial stressor. We hypothesized that emerging adults with a history of SA would perform more poorly on executive functioning tasks (i.e., problem solving, impulsivity) following a psychosocial stressor than their peers who did not experience the stressor; we also predicted that emerging adults with a history of SA (vs. no SA) who experienced social exclusion (vs. no exclusion) would display more suicide-specific semantic interference, reflecting the activation of a suicide-related schema. Finally, we explored the relation between physiological reactivity, executive functioning, and cognitive processing between emerging adults with and without a history of SA.

2. Method

This study consisted of a 2 × 2 (SA history/No SA History × Social Inclusion/Exclusion) experimental design. Participants with and without a history of SA were randomly assigned to a social stressor condition or to a control (inclusion) condition. Saliva samples were taken before and after the stressor task, and self-reported affect was measured before and after the stressor. Behavioral measures of impulsivity, problem solving, and semantic interference from suicide-related words were assessed after post-stressor assessment of affect.

2.1. Participants

Seventy-six emerging adults (58 females), aged 18–29 (M = 20.8, SD = 2.05), recruited primarily from a public commuter college campus in the Northeastern United States, took part in this study for monetary compensation. Ethnoracial composition of the sample was 34 % Asian, 16 % Black (non-Hispanic), 16 % White (non-Hispanic), 12 % Hispanic, and 22 % of other races/ethnicities. Participants were recruited from two larger samples of individuals from a study of cognitive risk factors associated with suicide ideation and attempts. Forty-four participants in the current study were recruited from a sample of 285 individuals, and the remaining participants (n = 32) were recruited from a sample of 2423 individuals, both screened for a history of SAs. Individuals were recruited based on whether they reported a history of one or more previous SAs (n = 37) and whether they had no history of an attempt (n = 39). Demographic characteristics of the sample stratified by SA history and stressor condition are presented in Table 1. Of note, no differences emerged in demographic variables by SA history or stressor condition, except for age, such that participants with a history of SA were slightly older.

Table 1.

Demographic Characteristics of the Sample by SA history and Stressor Condition.

Total Sample (n = 76) n (%) No SA (n = 39) n (%) SA (n = 37) n (%) X 2 Inclusion (n = 38) n (%) Exclusion (n = 38) n (%) X 2
Sex 0.02 1.82
 Female 58 (76) 29 (50) 29 (50) 26 (45) 32 (55)
 Male 18 (24) 10 (56) 8 (44) 12 (67) 6 (33)
Race/Ethnicity 3.81 4.58
 Asian 26 (34) 11 (42) 15 (58) 10 (39) 16 (62)
 Black (non-Hispanic) 12 (16) 8 (67) 4 (33) 7 (58) 5 (42)
 Hispanic 9 (12) 4 (44) 5 (56) 3 (33) 6 (67)
 White (non-Hispanic) 12 (16) 5 (42) 7 (58) 8 (67) 4 (33)
 Other 17 (22) 11 (65) 6 (35) 10 (59) 7 (41)
Medication 2.98 3.67
 No Meds 56 (74) 31 (55) 25 (45) 29 (52) 27 (48)
 Birth Control 9 (12) 5 (56) 4 (44) 2 (22) 7 (78)
 Other Meds 11 (15) 3 (27) 8 (73) 7 (64) 4 (36)
M (SD) M (SD) M (SD) t M (SD) M (SD)
Age 20.84 (2.05) 20.33 (1.42) 21.39 (2.46) −2.25* 21.03 (2.43) 20.66 (1.60) 0.78
Baseline Positive Affect 25.37 (7.56) 25.66 (7.45) 25.08 (7.77) 0.33 26.47 (7.65) 24.24 (7.40) 1.28
Baseline Negative Affect 16.09 (7.37) 13.74 (5.59) 18.51 (8.22) −2.94** 18.63 (8.68) 13.49 (4.51) 3.23**
Baseline Cortisol 0.23 (0.09) 0.25 (0.10) 0.22 (0.08) 1.11 0.23 (0.10) 0.23 (0.08) 0.41
Open in a new tab
*

p < .05;

**

p < .01.

2.2. Measures

2.2.1. Experimental manipulation

Mood Check.

The Positive and Negative Affect Scale (PANAS; Watson et al., 1988) includes 20 adjectives – 10 reflecting positive affect (e.g., interested, excited), and 10 reflecting negative affect (e.g., irritable, scared). Individuals rated each item for the extent to which they felt that way, on a 1 (“very slightly or not at all”) to 5 (“extremely”) Likert scale. Cronbach’s alpha for the first administration of the PANAS was 0.86 and 0.91 for positive affect and negative affect, respectively, and for the second administration, it was 0.89 and 0.79, respectively.

Social Stressor Task.

Participants were randomly assigned to experience a laboratory-based social stressor using the Cyberball task (Williams and Jarvis, 2006). Participants were told that they were going to play an interactive ball-tossing game used for mental visualization with three other individuals. Each player, including the participant, was portrayed with avatars. The participant was unaware that the other players were part of the computer program and not individuals playing the game with them. In the non-stress (social inclusion) condition (n = 38), a total of twenty ball tosses by three computer programmed players occurred during the task, seven of which were thrown to the participant. In the stress (social exclusion) condition (n = 38), a total of twenty-six ball tosses by the three computer programmed players occurred during the task, three of which were thrown to the participant (three tosses thrown to the participant occurred during the first nine tosses of the game).

Excluded participants in Cyberball tend to self-report higher stress and lower mood, compared to included participants (Zadro et al., 2004). The Cyberball program includes self-report measures completed following the task, assessing feelings of exclusion, sadness, and anger. In the present study, data from the Cyberball task was available from 73 of 76 participants (three participants did not have data on affect recorded due to a computer error).

2.2.2. Cognitive tasks

Biased cognitive processing.

The Suicide Stroop Test (SST; Cha et al., 2010) measures semantic interference from emotionally salient linguistic stimuli, including suicide-related words. Participants are presented with words one at a time on a computer screen and asked to name the color of the word (either red or blue). Words are suicide-related (e.g., dead), neutral (e.g., paper), or negative (e.g., alone). Increased response latency to name the color of a word indicates greater interference from the meaning of the word. The SST was administered using Empirisoft DirectRT v2004 software (Jarvis, 2004). Following Cha et al.’s (2010) procedure, trials with incorrect responses were excluded from analyses. Four participants were excluded due to having an error rate 2 SD above the error rate for the sample. Trials with response latencies ±2 SD from each participant’s mean response latency were eliminated. Data from three further participants were excluded for having a mean response latency 2 SD above the mean response latency for the sample. Interference from suicide-related stimuli was calculated by subtracting latencies for neutral words from latencies for suicide-related words. Interference from negatively-valenced stimuli was calculated by subtracting latencies for neutral words from latencies for negatively-valenced words.

Behavioral Impulsivity.

The Go-Stop Impulsivity Paradigm (Dougherty et al., 2004) assesses inhibition of an already initiated behavior. Participants are instructed to respond to a series of 5-digit numbers when a “go” signal appears and to not respond when a “stop” signal appears. Timing of the “stop” signal occurs either 50, 150, 250, or 350-ms following the “go” stimulus onset. Disinhibition scores, reflecting difficulty inhibiting a response that has already been initiated, were calculated as the percentage of disinhibited responses out of total responses found in the 50, 150, and 250 ms trials.

Problem Solving.

The Tower of London (TOL) Test (Shallice, 1982) measures planning, problem solving, and attention. Participants were shown discs of varying sizes distributed among three pegs on a computer screen and were asked to organize a stack of discs of increasing size on one of the pegs. There is a predetermined number of minimum moves to complete the 36 trial sessions. Extra moves indicate a lack of planning. The TOL Test has adequate test-retest reliability (r = 0.70) and has been concurrently validated on brain regions that are functionally important in planning and strategy use (Sullivan et al., 2009). In the present study, the TOL task was administered via the Psychology Experiment Building Language (PEBL; Mueller, 2014).

2.2.3. Measures of stress reactivity

Saliva samples were collected via passive drool. Participants were instructed to allow saliva to pool in the mouth and then drool down a saliva collection aid into a 2 ml cryovial until a minimum of 1 ml of saliva was collected. Saliva samples were collected between 12:30pm and 2:30pm to control for diurnal fluctuations in cortisol concentrations. Cortisol levels peak approximately 20 min after exposure to an acute stressor (Kirschbaum and Hellhammer, 2003). Difference scores between saliva sample 3 and saliva sample 2 (saliva sample 3 was taken 20 min after saliva sample 2) represented cortisol response in the multivariate analysis.

2.2.4. Self-report measures

Sociodemographic information.

A Participant Information Questionnaire was used to collect demographic data. Furthermore, data that are typically collected in experiments that utilize salivary cortisol samples were also gathered: participants were asked whether they had been previously diagnosed with Cushing’s syndrome, to list any current medications being taken, and if female, whether they were currently menstruating. These factors are known to affect neuroendocrine reactivity (Golden et al., 2011).

2.3. Procedure

After providing informed consent, participants completed demographic information (and self-report questionnaires not part of the present study), followed by a 10-min rest period. After the rest period, the first saliva sample was taken (SS1; taken 10 min prior to the administration of the Cyberball task). Participants then completed the PANAS, followed by the Cyberball Task (exclusion or inclusion condition). After Cyberball, the second saliva sample was taken (SS2; taken 5 min after completion of Cyberball). The PANAS was administered again before participants completed either the Go-Stop or the TOL (counter-balanced condition). After these tasks, the third saliva sample was collected (SS3; taken 20 min after the completion of Cyberball). Next, participants completed the SST. Then, the 4th and final saliva sample was collected (SS4; taken 45 min after the completion of Cyberball). Procedures were approved by the Institutional Review Board of the City University of New York.

2.4. Data preparation

Saliva samples were maintained at 4 °C for no longer than 1 h before freezing them at or below −20 °C. On the day samples were assayed, the samples were brought to room temperature and then centrifuged for 15 min at approximately 3000 rpm. Standard enzyme-linked immunosorbent assays were performed to detect cortisol.

2.5. Data analytic plan

We compared demographic characteristics by SA history and stressor condition (see Table 1). Additionally, we conducted a repeated measures analysis of variance (ANOVA) among female participants to assess the effects of birth control medication on cortisol reactivity, given previous findings noting an effect (Granger et al., 2009). A significant effect of medication on cortisol reactivity emerged, such that participants taking birth control displayed heightened cortisol reactivity compared to participants taking other medications or no medications. Thus, we included this variable as a covariate in subsequent analyses.

We conducted repeated measures analysis of variance (ANOVA) to examine if young adults in the social stressor condition reported decreased positive affect and increased negative affect. Next, we applied independent samples t-tests to compare ratings of disconnectedness, rejection, self-esteem, anger, sadness, and general “bad” feeling between adults in the inclusion vs. exclusion condition. These latter measures were only administered after the Cyberball task. We also calculated area under the curve scores with respect to the ground (AUCG; measuring total cortisol output) and with respect to increase (AUCI; measuring total cortisol response) to our stressors to determine whether participants experienced changes in cortisol response before and after the social stressor. AUCG and AUCI scores were calculated using Pruessner et al. (2003)’s formulas.

To test our first aim, we used multilevel regression modeling to examine differences in stress reactivity trajectories between adults with and without a SA following a social stressor vs. no stressor. Given that participants completed various stress reactivity assessments—thereby violating the independence of observations assumption—we used multilevel modeling to account for the data’s nested structure. We used two-level models to account for change in cortisol over time (level 1), nested within emerging adults with different SA status and stressor conditions (level 2). First, cortisol raw data were log-transformed due to their positively skewed distribution. In the multilevel models, we used both linear and quadratic trends of time to examine cortisol trajectory. However, because the quadratic trend was non-significant in all models, we did not include it in the final analyses. We began the analysis by testing the model fit for each set of analyses (See supplemental material for details on fit indices). First, we ran an unconditional means model consisting of a random intercept only; we compared this baseline model to a random intercept model with a fixed effect of time; lastly, we included a random intercept and random slope model. We conducted a multilevel analysis with SA, stressor condition, and attempt × stressor as predictors of cortisol reactivity over time. We used full information maximum likelihood to estimate the model, because it is more robust in handling missing data relative to other procedures (Baraldi and Enders, 2010). We established the significance level at p < .05.

To test the hypotheses related to aim 2, we used a series of 2 (attempt/no attempt) × 2 (inclusion/exclusion) factorial ANOVAs. First, we tested the main effects of SA status and stressor conditions on problem solving and impulsivity. Next, we tested the interaction of SA status and stressor conditions. For the second hypothesis in aim 2, we replicated the factorial ANOVA with suicide-specific semantic interference as the outcome. An a priori power analysis indicated that a total sample size of 76 would provide at least 80 % power for an expected large effect size (f = 0.40). However, a post hoc power analysis conducted using G*Power (Erdfelder et al., 1996) indicated that the achieved power for detecting a medium effect size (f = 0.25) in the 2 × 2 ANOVA was 0.58 with a total sample of 76 participants. Thus, while our study was adequately powered to detect large effects, it was underpowered to detect medium-sized effects with sufficient confidence. Therefore, results are presented as exploratory rather than confirmatory.

Finally, we computed bivariate correlations to test the association between stress reactivity, executive functioning, and semantic interference. Given the research suggesting differences in stress reactivity by SA history, we computed the correlations separately for individuals with and without a history of SA.

3. Results

3.1. Experimental manipulation check

To assess the effectiveness of social exclusion induction, we conducted repeated measures ANOVA to identify differences in positive and negative affect prior to and after stress induction. The interaction between social exclusion/inclusion condition and positive affect was statistically significant, F(1, 71) = 15.93, p < .01. Specifically, participants assigned to the social exclusion condition had a significant reduction in positive affect after social exclusion (M = 1.64, SD = 0.38) compared to before exclusion (M = 2.43, SD = 0.75). In contrast, participants assigned to the social inclusion (control) condition exhibited no statistically significant difference in positive affect before (M = 2.66, SD = 0.77) vs. after the Cyberball task (M = 2.46, SD = 0.90). Similarly, the interaction between social exclusion/inclusion condition and negative affect was also statistically significant, F(1, 71) = 46.34, p < .01. Notably, adults in the social exclusion condition reported a significantly greater increase in negative affect after exposure to social exclusion (M = 1.94, SD = 0.41) than at baseline (M = 1.43, SD = 0.45). In contrast, their peers in the social inclusion condition exhibited no difference in negative affect before (M = 1.93, SD = 0.81) vs. after (M = 1.56, SD = 0.55) completing the task. The increase in negative affect and reduction in positive affect for participants who were assigned to the social exclusion condition suggested that the experimental manipulation was effective at altering affect. In terms of ratings made following the Cyberball task, participants in the social exclusion condition reported significantly higher levels of rejection (M = 3.32), sadness (M = 3.65), and anger (M = 1.89) than did those in the control condition (M = 1.69, 1.57, and 1.28, respectively, p < .01). See Table 2 for differences in affect between the inclusion and exclusion conditions.

Table 2.

Post-cyberball self-report measures of affect.

Stressor (n = 37)a No Stressor (n = 36)a
M (SD) M (SD)
p d
Disconnected 3.38 (1.32) 2.47 (1.13) <0.01 0.74
Rejected 3.32 (1.31) 1.69 (1.06) <0.01 1.37
Self-Esteem 1.73 (0.96) 2.81 (1.10) <0.01 1.05
Bad 2.46 (1.12) 1.53 (0.77) <0.01 0.97
Angry 1.89 (0.81) 1.28 (0.66) <0.01 0.83
Sad 3.65 (1.34) 1.57 (0.89) <0.01 1.83
Open in a new tab
a

Affect ratings for 3 participants were not recorded due to a computer error.

3.2. Cortisol response over time by suicide attempt history and stress condition

We hypothesized that emerging adults with a history of SA would have a higher cortisol response over time to a social stressor than their peers without a suicide attempt history. Average trajectories over time suggested that participants with a previous SA who were in the social stress (exclusion) condition exhibited blunted cortisol reactivity by the 45 min mark, relative to their peers in the control (inclusion) condition (see Fig. 1). However, findings from the multilevel model revealed no statistically significant difference in the trajectories of cortisol reactivity over time between participants with and without a suicide attempt history, controlling for medication. Furthermore, the SA by stressor interaction was not statistically significant, indicating that irrespective of stress condition, cortisol reactivity trajectories did not differ during the span of the study. Notably, the intra-class correlation (0.82) indicated that 82 % of the variability in cortisol reactivity was due to individual differences. See Table 3 for details.

Fig. 1.

Fig. 1.

Open in a new tab

Average cortisol level over time by attempt status and stressor condition mean±1 standard error.

Table 3.

Multilevel model examining effect of suicide attempt history and social stressor on cortisol reactivity.

Model 1 Model 2
Variable b se t 95 % CI b se t 95 % CI
Fixed effects
Level 1
 Time −0.002** 0.0003 −7.44 −0.003, −0.002 −0.002* 0.0003 −7.39 −0.003, −0.002
Level 2
 Attempt −0.04 0.04 −1.07 −0.11, 0.03 −0.04 0.05 −0.74 −0.14, 0.06
 Stressor 0.03 0.04 0.89 −0.04, 0.10 0.02 0.05 0.39 −0.08, 0.12
 Attempt*Stressor – – – – 0.004 0.07 0.05 −0.14, 0.14
 Birth control – – – – 0.09 0.06 1.58 −0.02, 0.20
 Other meds – – – – −0.01 0.05 −0.21 −0.11, 0.09
Random Effects
 Intercept 0.15 0.13, 0.18 0.15 0.13, 0.18
 Slope 0.002 0.002, 0.003 0.002 0.002, 0.003
 Level 1 residual 0.07 0.07, 0.08 0.07 0.06, 0.08
Open in a new tab

Note: Model 1 includes suicide attempt history and stressor condition as level 2 predictors. Model 2 includes both predictors and their interaction; 95 % confidence intervals are for unstandardized estimates.

*

p < .05;

**

p < .01.

3.3. Differences in executive functioning and suicide-specific cognitive bias by suicide attempt history and stressor condition

We tested the main effects of SA status and exclusion/inclusion conditions on suicide-related semantic interference and executive functioning—i.e., impulsivity and problem solving (see Table 4). We then tested the interaction of SA status and exclusion/inclusion conditions on these factors. For suicide-specific semantic interference, there was a trend for attempt status, F(1, 68) = 3.86, p = .05, ηP2 = 0.06. Participants with (vs. without) a history of SA had faster reaction time to respond to suicide-related compared to neutral stimuli (see Fig. 2). However, the main effect of exclusion condition was non-significant, F (1, 68) = 0.34, p = .56, ηP2 = 0.01. The interaction between SA status and exclusion/inclusion condition was also non-significant, F(1, 68) = 1.01, p = .32, ηP2 = 0.02.

Table 4.

Mean differences in cognitive bias and executive functioning by suicide attempt history and stress condition.

Suicide Bias Impulsivity (150 msec)a Problem Solving
Effect Condition M (SE) F p M (SE) F p M (SE) F p
SA History No SA 17.88 (8.00) 3.86 0.05 67.17 (3.33) 0.94 0.34 100.93 (6.11) 1.42 0.24
SA −4.19 (7.88) 62.54 (3.42) 90.42 (6.36)
Condition Inclusion 10.12 (8.11) 0.34 0.56 65.79 (3.38) 0.15 0.70 93.27 (6.28) 0.30 0.59
Exclusion 3.57 (7.76) 63.92 (3.38) 98.07 (6.20)
Interaction No SA & Inclusion 26.80 (11.65) 1.01 0.32 71.46 (4.77) 1.97 0.17 90.16 (8.75) 3.60 0.06
No SA & Exclusion 8.96 (10.98) 62.89 (4.65) 111.70 (8.53)
SA & Inclusion −6.56 (11.30) 60.12 (4.77) 96.39 (8.99)
SA & Exclusion −1.82 (10.98) 64.96 (4.91) 84.44 (8.99)
Open in a new tab

SA = Suicide Attempt.

a

For ease of interpretation, we show response inhibition percentages at 150 msec on the Go-Stop Impulsivity Paradigm.

Fig. 2.

Fig. 2.

Open in a new tab

Suicide-Related Semantic Interference on the Suicide Stroop Task, by Suicide Attempt History

Note. Suicide-related interference = Mean reaction time (msec) to suicide-related words minus mean reaction time to neutral words. Standard error bars are shown.

Regarding the behavioral marker of impulsivity, neither the main effect of attempt status, F(1, 75) = 0.94, p = .34, ηP2 = 0.01, or exclusion/inclusion condition, F(1, 75) = 0.15, p = .70, ηP2 = 0.002, were statistically significant, nor was the interaction significant, F(1, 75) = 1.97, p = .17, ηP2 = 0.03. For problem solving, there was no significant main effect for either SA history, F(1, 74) = 1.42, p = .24, ηP2 = 0.02, or exclusion condition, F(1, 74) = 0.30, p = .59, ηP2 = 0.004. However, there was a trend for the interactive effect, F(1, 74) = 3.60, p = .06, ηP2 = 0.05 (see Fig. 3), such that among participants in the social exclusion condition, those with a history of SA showed better problem-solving (i.e., fewer extra moves) on the Tower of London task than did those without a history of SA, t(71) = 2.20, p < .05, Cohen’s d = 0.71.

Fig. 3.

Fig. 3.

Open in a new tab

Interaction between suicide attempt history and social exclusion on problem solving (tower of London).

3.4. Stress response and its relationship to executive functioning and cognitive bias

Cortisol reactivity was inversely associated with semantic interference from negative stimuli relative to neural stimuli, r(31) = −0.39, p < .05, for adults without SAs. No other executive functioning or cognitive processing variables were related to cortisol reactivity (see Table 5).

Table 5.

Correlation between cortisol with cognitive reactivity variables for individuals with and without a suicide attempt history.

No SA History SA History
Suicide bias −0.32 0.04
Negative bias −0.39* 0.20
Positive bias −0.18 0.07
Impulsivity 150 msec 0.04 0.05
Problem solving −0.13 −0.23
Open in a new tab

Note: Cortisol was log transformed prior to computation of correlation analyses. Cortisol reactivity was measured as the difference between cortisol response 20 min after Cyberball and 5 min after Cyberball.

*

= p < .05.

4. Discussion

Findings from the present study suggest that experiencing social exclusion evoked subjective feelings of rejection, sadness, and anger, irrespective of SA history. While cortisol reactivity decreased over time across all participants, those with a previous SA in the social exclusion condition exhibited more hypo-reactivity by the 45-min mark. We note, however, that the psychosocial stressor did not appear to elicit a physiological stress response, and there was no difference in trajectories of cortisol by SA history, irrespective of social exclusion/inclusion. Finally, individuals with a history of SA did not demonstrate differences in behavioral impulsivity compared to those without a history of SA, regardless of stress condition. However, there was a trend for individuals with a history of SA to demonstrate better problem solving when they experienced social exclusion. In addition, there was a trend for participants with a history of SA to demonstrate less semantic interference from suicide-related stimuli (opposite of what we predicted), regardless of social exclusion/inclusion. We contextualize these findings within the broader literature.

4.1. Cyberball and stress reactivity

Consistent with previous work, social exclusion during a Cyberball task elicited subjective negative affect, including low self-esteem, rejection, anger, and sadness (Bastian and Haslam, 2010; Peterson et al., 2011). Our finding that Cyberball did not elicit a physiological stress response among participants in a social exclusion condition is consistent with prior mixed findings for cortisol response following Cyberball exclusion (Geniole et al., 2011; Helpman et al., 2017; Zöller et al., 2010). The lack of cortisol response could reflect a defensive response described as emotional analgesia (Bass et al., 2014), which has been hypothesized to occur after social pain and to work similarly to the adaptive physiological analgesic response that occurs after experiencing physical pain (Bass et al., 2014; DeWall and Baumeister, 2006). Consistent with previous research with undergraduates (Bass et al., 2014), we found that participants excluded in Cyberball exhibited a decrease in cortisol concentrations. However, contrary to those findings, our participants in the inclusion condition did not exhibit an increase in cortisol concentrations. Bass and colleagues (2014) speculated that participants in their inclusion condition might have used greater attention and effort, resulting in a cortisol response.

An alternative explanation could be that sex differences account for the variability in cortisol response following social exclusion. For example, Helpman and colleagues (2017) found that cortisol reactivity declined over time in female participants, while male participants exhibited no change in cortisol response following social exclusion using Cyberball. A more recent study with an entirely female sample found that cortisol response decreased over time among participants in both inclusion and exclusion conditions of Cyberball (Stout et al., 2023), and that loneliness moderated the relation between social exclusion and cortisol reactivity. Other studies show a decline in cortisol response between both male and female participants (Radke et al., 2018), with female participants showing a sharper decline. Collectively, these studies point to female participants exhibiting a gradual decline in cortisol reactivity following social exclusion via Cyberball, and given that our sample was predominantly female, this could explain why we found no cortisol reactivity in the sample.

Another consideration is that Cyberball may be less effective at eliciting a physiological response compared to other stress-induction paradigms, such as the Trier Social Stress Test, which more reliably induces physiological stress (Allen et al., 2014; Man et al., 2023). Compared to other biomarkers, salivary cortisol appears to be most sensitive to stress inductions captured via the Trier Social Stress Test (Man et al., 2023). In contrast, less robust findings are available for Cyberball, suggesting that while Cyberball may not reliably activate the HPA-axis, it is associated with self-reported perceptions of social exclusion and negative affect (Kulakova et al., 2024). For example, a recent meta-analytic study comparing stress-induction paradigms considered Cyberball an “outlier” because it does not include a social judgment component as part of the task, and at a neurological level, Cyberball is associated with neurological activation of brain regions that do not overlap with other paradigms (Berretz et al., 2021). In fact, a separate neuroimaging meta-analysis found that Cyberball reliably activated brain regions associated with rejection sensitivity, emotional and visual processing, and cognitive processing (Mwilambwe-Tshilobo and Spreng, 2021). Given that a more robust body of research documents the neuropsychological effects of social exclusion via Cyberball, perhaps the effects of Cyberball might be best captured through neuroimaging and self-reported affective measures, rather than neuroendocrinological assessments. Future experimental designs comparing these stress-induction paradigms might help clarify the role that social exclusion plays in HPA-axis reactivity.

4.2. Stress reactivity and suicidal behavior

We found no differences in cortisol reactivity over time among participants with and without a history of SA, irrespective of stress condition, nor did we find an interaction between SA history and stress conditions on negative affect. These findings stand in contrast to previous studies demonstrating blunted salivary cortisol reactivity following a psychophysiological stressor (Melhem et al., 2016; O’Connor et al., 2018). A recent study found that among patients with depression, those classified as having high intent to engage in suicidal behaviors demonstrated blunted cortisol reactivity following a social stressor (Herzog et al., 2023). Similar patterns of salivary cortisol reactivity following a social stressor were observed among adolescents with depression who engaged in self-injurious behaviors (Klimes-Dougan et al., 2019). More critically, individuals who attempted suicide for the first time were found to exhibit lower hair cortisol concentrations, providing evidence for a biological marker of cortisol reactivity prior to suicidal behaviors (Melhem et al., 2016). In light of these findings, our results are puzzling, given that participants in the social exclusion condition did not exhibit blunted cortisol reactivity relative to peers in the control condition. As described above, we suspect that our null findings are due to limitations with the Cyberball task in reliably activating the HPA-axis and modulating cortisol reactivity. Indeed, the studies finding a difference in salivary cortisol reactivity among individuals with a suicide attempt history relied on the Trier Social Stress Test, which is known to activate the HPA-axis (Labuschagne et al., 2019).

Regardless of SA history, individuals in the social exclusion condition reported significantly greater negative affect than individuals in the inclusion condition. This is consistent with previous research linking social exclusion via Cyberball with increased negative affect and reduced positive affect (Zhang et al., 2017). Perhaps when individuals vulnerable to suicidal behaviors are not in an acute suicidal crisis, they may respond to minor stressors similarly to peers without previous suicidal behaviors. The severity and recency of suicidal behaviors may play a more important role in stress reactivity than does the history of SAs. Further research is needed to assess these possibilities.

4.3. Social exclusion, executive functioning, and suicidal behavior

Contrary to our hypothesis, there were no significant differences between adults with and without previous SAs on behavioral impulsivity, regardless of stress condition. Furthermore, our hypothesized interaction between SA history and stress condition in problem solving was present as a non-significant trend – but in the opposite direction to what we expected (i.e., improved performance among participants in the social exclusion condition who had a history of SA compared to those without a history of SA, but no differences compared to the inclusion condition). These findings align with previous research demonstrating that individuals with depression and a history of SA did not differ from a control group in the TOL Test, suggesting that their planning capacity was intact (Moniz et al., 2017). A more recent study demonstrated that individuals with a history of SA outperformed peers without a history of SA in the TOL Test and another planning task, suggesting that their ability to plan and initiate goals are intact (Sánchez-Sansegundo et al., 2020). Indeed, the degree of lethality in a clinical sample differentiated the number of extra moves needed to complete the TOL Test in a study examining executive functioning differences between adults with high-lethality vs. low-lethality attempts (Williams et al., 2015). However, due to our small sample being underpowered to detect significant interaction effects, these findings should be interpreted cautiously, and results should be considered exploratory and best viewed as hypothesis generating.

An additional consideration is that results from the TOL Test depend on current mood, versus whether an individual has a history of SA. Godard et al. (2012) found that performance on the TOL Test improved significantly over a 12-month period as depressive symptoms decreased in a clinical sample of individuals diagnosed with major depressive or bipolar disorder. Other studies have found that participants induced to a positive or negative, but not neutral mood state, displayed more difficulty in planning, as measured via the TOL task (Robinson and Sahakian, 2009). However, the effects may not reliably apply to non-clinical, undergraduate samples, as recent studies showed no impact of mood inductions on performance on the TOL Test (Monno et al., 2024). Thus, problem-solving, as assessed by the TOL Test, may be state-dependent.

4.4. Activation of suicide-related cognitive bias

We found no evidence that experiencing social exclusion activated suicide-related cognitive biases among individuals with a history of SA and compared to those experiencing social inclusion. However, there was a trend for participants without a history of SA to show greater interference from suicide-related words, compared to those with a previous suicide attempt, regardless of stress condition. Like the results discussed above, we exercise caution in interpreting these trend-level results due to our small sample size.

When considering the broader literature, the findings in this study contrast with Cha et al.’s (2010) finding of interference from suicide-related words by individuals with a previous SA. Importantly, participants (including controls) from the aforementioned study were recruited from a psychiatric emergency department, where SAs would have been more recent. Our findings are more consistent with those of a study of 736 college students that found no interference from suicide-related words among students with a previous history of SA, compared to those without a history of SA. However, those with a previous attempt did take significantly longer to react to the word “suicide” than did those without an attempt history (Chung and Jeglic, 2016). Furthermore, in a cross-sectional case-control study comparing patients with and without a history of attempted suicide, Richard-Devantoy et al. found no significant differences by SA history in interference scores for suicide-related, negatively valenced, or positively valenced words (Richard-Devantoy et al., 2016). Notably, reaction times to the word “suicide” were no longer significant in the Chung and Jeglic (2016) sample after adjusting for depressive symptoms, suggesting that the severity of symptoms may play a role in biased suicide-related cognitive processing. The absence of an effect on the SST is also consistent with a mega-analysis of the SST using 7 separate samples, which found that Cha et al.’s (2010) original Suicide Stroop effect did not replicate across samples of adults and adolescents (Wilson et al., 2019).

One reason for the lack of interference bias among participants with a suicide attempt history could be that individuals at risk for suicidal behaviors must be experiencing acute distressing affect, as proposed by the cognitive model of suicidal behavior (Wenzel & Beck). While the Cyberball task in our study induced negative affect for participants in the exclusion condition, the effect may not have been strong enough to activate suicide-related schemas. Additionally, since the SST was the last task that participants completed, the social exclusion effect of Cyberball may have waned by the time participants completed the task (Hartgerink et al., 2015). Another consideration is the sensitivity of the SST to capture cognitive bias across samples with varying histories of psychopathology and risk for suicidal behaviors. For example, in a systematic review and meta-analysis, Richard-Devantoy et al. (2025) found that individuals with a history of SA performed worse on the SST than patient controls and non-patient controls. Likewise, patient controls performed worse on the task than non-patient controls. Therefore, interference on the SST appears to be driven by psychopathology in general, rather than solely by a history of suicidal behaviors. Although in this study we did not collect data on clinical measures, such as severity of depressive symptoms, the recruitment strategy focused on recruiting participants at risk for suicidal thoughts and behaviors. Thus, while the sample was not recruited from clinical settings, perhaps some underlying psychological symptoms may have also contributed to the findings in this study.

Considering attentional processes more specifically, emerging research challenges our conventional ideas of suicide-specific attentional biases. For example, Baik and colleagues (2018) found that patients with Major Depression and a high score on a measure of suicide-related risk disengaged more quickly from an attentional bias task than patients with Major Depression and a low score on a measure of suicide-related risk. A more recent study found that in a non-clinical sample, young adults with recent suicidal thoughts disengaged more quickly from suicide-related stimuli than peers with more distal suicidal thoughts (Rosario-Williams et al., 2023) and negative affect moderated this relation, such that young adults with distal suicidal thoughts displayed more difficulty disengaging from suicide-related content relative to peers with recent suicidal thoughts (Rosario-Williams and Miranda, 2024). Overall, the findings point to an automatic processing of suicide-related content among individuals perceived to be at higher risk for suicidal behaviors. While we still advise caution in interpreting the findings of the current study, considering the small sample size, the findings are worth replicating to generate a comprehensive understanding of the underlying processes that differentiate suicidal thoughts and behaviors in clinical and non-clinical samples.

4.5. Implications for clinical assessment

The present study suggests that implicit cognition tasks, such as the SST, may not be helpful, in and of themselves, for assessment of suicide-related risk among young adults without acute suicidal thoughts and/or behaviors. This conclusion is consistent with findings from numerous other studies, which have suggested no relationship between suicide-related semantic interference and either suicide attempt history or future suicidal behavior, and this is the case even with individuals recruited following recent suicide ideation or attempts (Mandel et al., 2025; Shikh et al., 2025; Wilson et al., 2019). Indeed, a recent study with adolescents recruited soon after presenting for clinical care with suicide ideation or attempts suggested that a self-report measure of attention fixation was a better predictor of both future suicide ideation severity and attempt than were implicit measures like the SST (Shikh et al., 2025). Instead, behavioral measures like the SST may be more helpful in understanding underlying psychological processes (e.g., processing of suicide-related information) present even in the absence of an acute suicidal episode.

4.6. Limitations and future directions

Several limitations should be noted. Although our experimental manipulation affected participants’ self-reported affect, it did not induce a salivary cortisol response. This limitation impaired our ability to assess associations between physiological stress reactivity and other measures. Cyberball may be more appropriate for studies desiring to impact self-rated mood rather than eliciting strong physiological responses. Future studies should replicate these findings with a different stress induction paradigm (e.g., Trier Social Stress Test) in a non-clinical sample. Additionally, previous studies have demonstrated differential effects of interpersonal stress on HPA-axis activation between men and women (Helpman et al., 2017). Our study was underpowered to detect a medium effect in a 2 × 2 factorial design. This further limited our ability to examine the moderating effects of gender in the relation between SA history and stress reactivity. Thus, future studies are encouraged to expand this line of research with larger sample sizes to assess whether gender moderates the relation between interpersonal stress and stress reactivity among emerging adults with and without a history of SA. Furthermore, even though we used an experimental design, the cross-sectional nature of the study limits causal inferences about future SA risk. Prospective designs examining cognitive trajectories over time, including those using person-centered approaches to determine risk profiles (Bendezú et al., 2022), are needed. Furthermore, while this study compared participants with and without a suicide attempt history, data on recency and lethality of suicide attempts or on concurrent depressive symptom severity were not collected. These are important limitations, as these factors may moderate cortisol reactivity (Lindqvist et al., 2008; McGirr et al., 2011). Finally, it should be noted that we did not collect information on income or socioeconomic status, although the college from which participants were recruited serves lower income populations and tends to be more representative of the New York City population than participants recruited from private colleges.

4.7. Conclusion

We examined the effect of social exclusion on physiological stress reactivity, executive functioning, and suicide-related cognitive bias among emerging adults with and without a history of SA. Findings suggested overall higher negative affect as a result of social exclusion but no physiological stress response, with no significant variability in SA history. Emerging adults with previous SAs who are not in an acutely suicidal crisis did not differ from those without a history of SA on measures of executive functioning and suicide-related cognitive bias. Alternative methods, such as inclusion of a social stressor task known to produce greater physiological stress response, or prospective designs that capture responses in real-time, may be needed. Integrated research using a biopsychosocial framework may clarify how distinct components of physiology and information processing are interrelated and increase risk for suicidal behaviors.

Supplementary Material

1

Acknowledgments

This study was part of Jorge Valderrama’s dissertation, “Stress and Suicidal Behavior: A Cognitive, Behavioral, and Biological Integrative Approach,” at The Graduate Center, City University of New York (CUNY), filed in 2016. Thanks to Joel Erblich, Elizabeth Jeglic, Victoria Luine, and Mariann Weierich for comments on a previous version of this manuscript.

Funding sources

This research was funded by NIH Grant GM060665 and by a Doctoral Student Research Grant from the CUNY Graduate Center. The funding sources had no role in the study design, data collection, interpretation, writing of the manuscript, or decision to submit the manuscript for publication.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Jorge Valderrama reports financial support was provided by National Institutes of Health. Regina Miranda reports a relationship with National Institute of Mental Health that includes: funding grants. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Footnotes

CRediT authorship contribution statement

Beverlin Rosario-Williams: Writing – review & editing, Writing – original draft, Formal analysis, Conceptualization. Jorge Valderrama: Writing – review & editing, Writing – original draft, Software, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Evan Gilmer: Writing – review & editing. Florissell Rosales: Writing – review & editing. Regina Miranda: Writing – review & editing, Formal analysis, Conceptualization.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.psycom.2025.100220.

Data availability statement

De-identified data are available, upon request, from the corresponding author (with a data use agreement).

References

  1. Allen AP, Kennedy PJ, Cryan JF, Dinan TG, Clarke G, 2014. Biological and psychological markers of stress in humans: focus on the trier social stress test. Neurosci. Biobehav. Rev 38, 94–124. [DOI] [PubMed] [Google Scholar]
  2. Auerbach RP, Pagliaccio D, Allison GO, Alqueza KL, Alonso MF, 2021. Neural correlates associated with suicide and nonsuicidal self-injury in youth. Biol. Psychiatry 89, 119–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baik SY, Jeong M, Kim HS, Lee SH, 2018. ERP investigation of attentional disengagement from suicide-relevant information in patients with major depressive disorder. J. Affect. Disord 225, 357–364. [DOI] [PubMed] [Google Scholar]
  4. Baraldi AN, Enders CK, 2010. An introduction to modern missing data analyses. J. Sch. Psychol 48, 5–37. [DOI] [PubMed] [Google Scholar]
  5. Bartlett FC, 1932. Remembering: a Study in Experimental and Social Psychology. Cambridge University Press, Cambridge, UK. [Google Scholar]
  6. Bass EC, Stednitz SJ, Simonson K, Shen T, Gahtan E, 2014. Physiological stress reactivity and empathy following social exclusion: a test of the defensive emotional analgesia hypothesis. Soc. Neurosci 9, 504–513. [DOI] [PubMed] [Google Scholar]
  7. Bastian B, Haslam N, 2010. Excluded from humanity: the dehumanizing effects of social ostracism. J. Exp. Soc. Psychol 46, 107–113. [Google Scholar]
  8. Becker ES, Strohbach D, Rinck M, 1999. A specific attentional bias in suicide attempters. J. Nerv. Ment. Dis 187, 730–735. [DOI] [PubMed] [Google Scholar]
  9. Bendezú JJ, Calhoun CD, Patterson MW, Findley A, Rudoph KD, Hastings P, Nock MK, Prinstein MJ, 2022. Adolescent girls’ stress responses as prospective predictors of self-injurious thoughts and behaviors: a person-centered, multilevel study. Dev. Psychopathol 34, 1447–1467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cha CB, Najmi S, Park JM, Finn CT, Nock MK, 2010. Attentional bias toward suicide-related stimuli predicts suicidal behavior. J. Abnorm. Psychol 119, 616–622. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Chung Y, Jeglic EL, 2016. Use of the modified emotional stroop task to detect suicidality in college population. Suicide life threat. Beyond Behav. 46, 55–66. [Google Scholar]
  12. DeWall CN, Baumeister RF, 2006. Alone but feeling no pain: effects of social exclusion on physical pain tolerance and pain threshold, affective forecasting, and interpersonal empathy. J. Pers. Soc. Psychol 91, 1–15. [DOI] [PubMed] [Google Scholar]
  13. Dougherty DM, Mathias CW, Marsh DM, Papageorgiou TD, Swann AC, Moeller FG, 2004. Laboratory measured behavioral impulsivity relates to suicide attempt history. Suicide life threat. Beyond Behav. 34, 374–385. [Google Scholar]
  14. Eisenlohr-Moul TA, Miller AB, Giletta M, Hastings PD, Rudolph KD, Nock MK, Prinstein MJ, 2018. HPA axis response and psychosocial stress as interactive predictors of suicidal ideation and behavior in adolescent females: a multilevel diathesis-stress framework. Neuropsychopharmacology 43, 2564–2571. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Erdfelder E, Faul F, Buchner A, 1996. GPOWER: a general power analysis program. Behav. Res. Methods Instrum. Comput 28, 1–11. [Google Scholar]
  16. Escobar LE, Liew M, Yirdong F, Mandelos KP, Ferraro-Diglio SR, Abraham BM, Polanco-Roman L, Benau EM, 2024. Reduced attentional control in individuals with a history of suicide attempts compared to those with suicidal ideation: results from a systematic review and meta-analysis. J. Affect. Disord 249, 8–20. [Google Scholar]
  17. Geniole SN, Carŕe JM, McCormick CM, 2011. State, not trait, neuroendocrine function predicts costly reactive aggression in men after social exclusion and inclusion. Biol. Psychol 87, 137–145. [DOI] [PubMed] [Google Scholar]
  18. Giletta M, Calhoun CD, Hastings PD, Rudolph KD, Nock MK, Prinstein MJ, 2015. Multi-level risk factors for suicidal ideation among at-risk adolescent females: the role of hypothalamic-pituitary-adrenal axis responses to stress. J. Abnorm. Child Psychol 43, 807–820. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Godard J, Baruch P, Grondin S, Lafleur MF, 2012. Psychosocial and neurocognitive functioning in unipolar and bipolar depression: a 12-month prospective study. Psychiatry Res. 196, 145–153. [DOI] [PubMed] [Google Scholar]
  20. Golden SH, Wand GS, Malhotra S, Kamel I, Horton K, 2011. Reliability of hypothalamic-pituitary-adrenal axis assessment methods for use in population-based studies. Eur. J. Epidemiol 26, 511–525. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Granger DA, Hibel LC, Fortunato CK, Kapelewski CH, 2009. Medication effects on salivary cortisol: tactics and strategy to minimize impact in behavioral and developmental science. Psychoneuroendocrinology 34, 1437–1448. [DOI] [PubMed] [Google Scholar]
  22. Hartgerink CH, Van Beest I, Wicherts JM, Williams KD, 2015. The ordinal effects of ostracism: a meta-analysis of 120 cyberball studies. PLoS One 10 (5), e0127002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Helpman L, Penso J, Zagoory-Sharon O, Feldman R, Gilboa-Schechtman E, 2017. Endocrine and emotional response to exclusion among women and men; cortisol, salivary alpha amylase, and mood. Anxiety Stress Coping 30, 253–263. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Herzog S, Galfalvy H, Keilp JG, Mann JJ, Sublette ME, Burke A, Oquendo MA, Stanley BH, 2023. Relationship of stress-reactive cortisol to suicidal intent of prior attempts in major depression. Psychiatry Res. 327, 115315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hinkelmann K, Moritz S, Botzenhardt J, Riedesel K, Wiedemann K, Kellner M, Otte C, 2009. Cognitive impairment in major depression: association with salivary cortisol. Biol. Psychiatry 66, 879–885. [DOI] [PubMed] [Google Scholar]
  26. Jarvis B, 2004. Directrt. Empirisoft Corp, New York, NY: [Computer software]. [Google Scholar]
  27. Kulakova E, Graumann L, Wingenfeld K, 2024. The hypothalamus-pituitary-adrenal axis and social cognition in borderline personality disorder. Curr. Neuropharmacol 22, 378–394. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kirschbaum C, Hellhammer DH, 2003. Salivary cortisol. Encyclopedia of Stress 3, 379–383. [Google Scholar]
  29. Klimes-Dougan B, Begnel E, Almy B, Thai M, Schreiner MW, Cullen KR, 2019. Hypothalamic-pituitary-adrenal axis dysregulation in depressed adolescents with non-suicidal self-injury. PNEC 102, 216–224. [Google Scholar]
  30. Labuschagne I, Grace C, Rendell P, Terrett G, Heinrichs M, 2019. An introductory guide to conducting the trier social stress test. Neurosci. Biobehav. Rev 107, 686–695. [DOI] [PubMed] [Google Scholar]
  31. Lindqvist D, Isaksson A, Brundin L, 2008. Salivary cortisol and suicidal Behavior—a follow-up study. PNEC 33, 1061–1068. [Google Scholar]
  32. Man ISC, Shao R, Hou WK, Xin Li S, Liu FY, Lee M, Wing YK, Yau SY, Lee TMC, 2023. Multi-systemic evaluation of biological and emotional responses to the trier social stress test: a meta-analysis and systematic review. Front. Neuroendocrinol 68, 101050. [DOI] [PubMed] [Google Scholar]
  33. Mandel AA, Mitchell E, Krush C, Revzina O, 2025. Differences in suicide-specific attentional bias based on stimuli across the suicide stroop and disengagement tasks. J. Affect. Disord 369, 87–94. [DOI] [PubMed] [Google Scholar]
  34. Mann J, Waternaux C, Malone KM, 1999. Toward a clinical model of suicidal behavior in psychiatric patients. Am. J. Psychiatr 4, 25–30. [Google Scholar]
  35. McGirr A, Diaconu G, Berlim MT, Turecki G, 2011. Personal and family history of suicidal behaviour is associated with lower peripheral cortisol in depressed outpatients. J. Affect. Disord 131, 368–373. [DOI] [PubMed] [Google Scholar]
  36. Melhem NM, Keilp JG, Porta G, Oquendo MA, Burke A, Stanley B, Cooper TB, Mann JJ, Brent DA, 2016. Blunted HPA axis activity in suicide attempters compared to those at high risk for suicidal behavior. Neuropsychopharmacology 41, 1447–1456. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Miller AB, Prinstein MJ, 2019. Adolescent suicide as a failure of acute stress-response systems. Annu. Rev. Clin. Psychol 15, 425–450. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Moniz M, de Jesus SN, Pacheco A, Gonçalves E, Viseu J, Bŕas M, Silva D, Batista S, 2017. The influence of planning and response inhibition on cognitive functioning of non-psychotic unipolar depressed suicide attempters. Eur. J. Psychol 13, 717. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Monno Y, Nawa NE, Yamagishi N, 2024. Duration of mood effects following a Japanese version of the mood induction task. PLoS One 19, e0293871. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Mueller ST, 2014. PEBL: the psychology experiment building language (Version 0.14). [Computer experiment programming language]. http://pebl.sourceforge.net. [Google Scholar]
  41. Mwilambwe-Tshilobo L, Spreng RN, 2021. Social exclusion reliably engages the default network: a meta-analysis of cyberball. Neuroimage 227, 117666. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. O’Connor DB, Green JA, Ferguson E, O’Carroll RE, O’Connor RC, 2018. Effects of childhood trauma on cortisol levels in suicide attempters and ideators. Psychoneuroendocrinology 88, 9–16. [DOI] [PubMed] [Google Scholar]
  43. Otte C, Moritz S, Yassouridis A, Koop M, Madrischewski AM, Wiedemann K, Kellner M, 2007. Blockade of the mineralocorticoid receptor in healthy men: effects on experimentally induced panic symptoms, stress hormones, and cognition. Neuropsychopharmacology 32, 232–238. [DOI] [PubMed] [Google Scholar]
  44. Peterson CK, Gravens LC, Harmon-Jones E, 2011. Asymmetric frontal cortical activity and negative affective responses to ostracism. Soc. Cog. Affect. Neurosci 6, 277–285. [Google Scholar]
  45. Pruessner JC, Kirschbaum C, Meinlschmid G, Hellhammer DH, 2003. Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time-dependent change. Psychoneuroendocrinology 28, 916–931. [DOI] [PubMed] [Google Scholar]
  46. Radke S, Seidel EM, Boubela RN, Thaler H, Metzler H, Kryspin-Exner I, Moser E, Habel U, Derntl B, 2018. Immediate and delayed neuroendocrine responses to social exclusion in males and females. Psychoneuroendocrinology 93, 56–64. [DOI] [PubMed] [Google Scholar]
  47. Reynolds RM, 2013. Glucocorticoid excess and the developmental origins of disease: two decades of testing the hypothesis. Psychoneuroendocrinology 38, 1–11. [DOI] [PubMed] [Google Scholar]
  48. Richard-Devantoy S, Ding Y, Turecki G, Jollant F, 2016. Attentional bias toward suicide-relevant information in suicide attempters: a cross-sectional study and a meta-analysis. J. Affect. Disord 196, 101–108. [DOI] [PubMed] [Google Scholar]
  49. Richard-Devantoy S, Inja A, Dicker M, Bertrand JA, Turecki G, Orri M, Keilp JG, 2025. Cognitive control impairment in suicide behaviors: what do we know? A systematic review and meta-analysis of stroop in suicide behaviors. J. Affect. Disord 372, 358–369. [DOI] [PubMed] [Google Scholar]
  50. Rosario-Williams B, Akter S, Kaur S, Miranda R, 2023. Suicide-related construct accessibility and attention disengagement bias in suicide ideation. J. Psychopathol. Clin. Sci 132, 173. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Rosario-Williams B, Miranda R, 2024. Negative affect influences suicide-specific attentional biases. J. Anxiety Disord 7, 100081. [Google Scholar]
  52. Rosario-Williams B, Rowe-Harriott S, Ray M, Jeglic E, Miranda R, 2022. Factors precipitating suicide attempts vary across race. J. Am. Coll. Health 70, 568–574. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Rudd MD, 2000. The suicidal mode: a cognitive-behavioral model of suicidality. Suic. Life Threat. Behav 30, 18–33. [Google Scholar]
  54. Saffer BY, Klonsky ED, 2018. Do neurocognitive abilities distinguish suicide attempters from suicide ideators? A systematic review of an emerging research area. Clin. Psychol. Sci. Pract 25, e12227. [Google Scholar]
  55. Sánchez-Sansegundo M, Portilla-Tamarit I, Rubio-Aparicio M, Albaladejo-Blazquez N, Ruiz-Robledillo N, Ferrer-Cascales R, Zaragoza-Martí A, 2020. Neurocognitive functioning and suicidal behavior in violent offenders with schizophrenia spectrum disorders. Diagnostics 10, 1091. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Shallice T, 1982. Specific impairments of planning. Philos. Trans. R. Soc. Lond. B Biol. Sci 298, 199–209. [DOI] [PubMed] [Google Scholar]
  57. Shikh A, Mañaná J, Myruski S, Rombola C, Weierich M, Ortin-Peralta A, Miranda R, 2025. Lower- versus higher-order suicide-related attentional processing measures as predictors of adolescent suicide ideation and attempt. Psychiatry Res. 349, 116524. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Stanley B, Michel CA, Galfalvy HC, Keilp JG, Rizk MM, Richardson-Vejlgaard R, Oquendo MA, Mann JJ, 2019. Suicidal subtypes, stress responsivity and impulsive aggression. Psychiatry Res. 280, 112486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Stout ME, Keirns BH, Hawkins MAW, 2023. Hypocortisolemic reactivity to acute social stress among lonely young women. Soc. Neurosci 18, 171–182. [DOI] [PubMed] [Google Scholar]
  60. Stroop JR, 1935. Studies of interference in serial verbal reactions. J. Exp. Psychol 18, 643–662. [Google Scholar]
  61. Substance Abuse and Mental Health Services Administration (SAMHSA), 2024. Key substance use and mental health indicators in the United States: results from the 2023 national survey on drug use and health. https://www.samhsa.gov/data/report/2023-nsduh-annual-national-report.
  62. Sullivan JR, Riccio CA, Castillo CL, 2009. Concurrent validity of the tower tasks as measures of executive function in adults: a meta-analysis. Appl. Neuropsychol 16, 62–75. [DOI] [PubMed] [Google Scholar]
  63. Wang Y, Sareen J, Afifi TO, Bolton S-L, Johnson EA, Bolton JM, 2012. Recent stressful life events and suicide attempt. Psychiatr. Ann 42, 101–108. [Google Scholar]
  64. Watson D, Clark LA, Tellegen A, 1988. Development and validation of brief measures of positive and negative affect: the PANAS scales. J. Pers. Soc. Psychol 54, 1063–1070. [DOI] [PubMed] [Google Scholar]
  65. Wenzel A, Beck AT, 2008. A cognitive model of suicidal behavior: theory and treatment. Appl. Prev. Psychol 12, 189–201. [Google Scholar]
  66. Williams J, Broadbent K, 1986. Distraction by emotional stimuli: use of a stroop task with suicide attempters. Br. J. Clin. Psychol 25, 101–110. [DOI] [PubMed] [Google Scholar]
  67. Williams GE, Daros AR, Graves B, McMain SF, Links PS, Ruocco AC, 2015. Executive functions and social cognition in highly lethal self-injuring patients with borderline personality disorder. Pers. Disord 6, 107–116. [Google Scholar]
  68. Williams KD, Jarvis B, 2006. Cyberball: a program for use in research on interpersonal ostracism and acceptance. Behav. Res. Methods 38, 174–180. [DOI] [PubMed] [Google Scholar]
  69. Wilson KM, Milner AJ, Auerbach RP, Glenn CR, Kearns JC, Kirtley OJ, Najmi S, O’Connor RC, Stewart JG, Cha CB, 2019. Investigating the psychometric properties of the suicide stroop task. Psychol. Assess 31, 1052–1061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Wolf OT, 2003. HPA axis and memory. Best Pract. Res. Clin. Endocrinol. Metabol 17, 287–299. [Google Scholar]
  71. Zadro L, Williams KD, Richardson R, 2004. How low can you go? Ostracism by a computer is sufficient to lower self-reported levels of belonging, control, self-esteem, and meaningful existence. J. Exp. Soc. Psychol 40, 560–567. [Google Scholar]
  72. Zhang Q, Li X, Wang K, Zhou X, Dong Y, Zhang L, Xie W, Mu J, Li H, Zhu C, Yu F, 2017. Dull to social acceptance rather than sensitivity to social ostracism in interpersonal interaction for depression: behavioral and electrophysiological evidence from cyberball tasks. Front. Hum. Neurosci 11, 162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Zöller C, Maroof P, Weik U, Deinzer R, 2010. No effect of social exclusion on salivary cortisol secretion in women in a randomized controlled study. Psychoneuroendocrinology 35, 1294–1298. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2111931-supplement-1.docx (14.8KB, docx)

Data Availability Statement

De-identified data are available, upon request, from the corresponding author (with a data use agreement).

RESOURCES