Cumulative lifetime stressor exposure assessed by the STRAIN predicts economic ambiguity aversion

Human decisions frequently involve uncertain outcomes. Economics distinguishes risk (known outcome probabilities) from ambiguity (unknown probabilities). Individual tolerance for risk and ambiguity—risk and ambiguity preferences—are largely distinct and only weakly correlated. Prior research has linked age-related changes and neuroanatomy (e.g., cortical thinning in right posterior parietal cortex) to risk preferences, with age not directly affecting risk once cortical volume is accounted for. Ambiguity preferences, by contrast, are not explained by age or cortical thickness, underscoring their distinct nature. Theoretical and empirical work suggests psychosocial stressor exposure may shape how ambiguous outcomes are appraised: acute threat/stress and negative affect increase negative appraisals of ambiguity, and higher trait anxiety biases interpretations of ambiguous stimuli. Early life stress research further argues that cumulative adversity can impart persistent cognitive biases, potentially leading to more negative appraisals and avoidance in uncertain contexts. The present studies test the hypothesis that greater cumulative lifetime stressor exposure—especially early life stress—predicts greater ambiguity aversion but not risk aversion.

• Risk vs. ambiguity: Individuals often prefer known risks over unknown risks (ambiguity aversion). Risk and ambiguity tolerances are distinct and weakly correlated.
• Age and neuroanatomy: Older individuals tend to be more risk averse; this relates to reduced gray matter in right posterior parietal cortex, explaining idiosyncratic risk preferences. Neither age nor cortical thickness accounts for ambiguity preferences.
• Stress, affect, and ambiguity appraisal: Acute threat/stress exposure and negative affect heighten negative appraisals of ambiguous stimuli. Trait anxiety increases tendency to interpret ambiguity negatively. Unpredictable threats elicit greater physiological arousal, amygdala responsivity, and avoidance, particularly among individuals exposed to trauma and adversity.
• Early life stress: The literature proposes that early adversity can produce persistent cognitive biases toward negative interpretations of uncertainty, potentially adaptive in dangerous environments but leading to pessimistic appraisals and avoidance under ambiguity.
• Measurement gap: Comprehensive, lifespan-wide assessment of stressor exposure has been challenging, limiting direct tests of links between lifetime stress and ambiguity preferences. The STRAIN provides a validated, comprehensive inventory enabling such assessment.

Design: Two studies assessed whether cumulative lifetime stressor exposure predicts economic preferences under risk and ambiguity using a well-validated lottery choice task and the STRAIN.

Participants:

• Study 1: n = 58 healthy adults (35 women), age M = 25.7 ± 7.4 (range 18–56). Recruited at NYU. Compensation: $15/hour plus a lottery bonus from a randomly selected trial.
• Study 2 (replication): Recruited via Amazon Mechanical Turk. Enrolled n = 210; 6 recording failures and 18 excluded for >15% first-order stochastic dominance violations. Final n = 188 (82 women), age M = 39.8 ± 12.14 (range 19–73). Compensation: $10 plus lottery bonus. Added covariates: IQ estimate (Raven’s Progressive Matrices, RPM), socioeconomic status (annual household income), and psychological distress (Kessler K-6).

Economic decision-making task:

• Each participant made 240 choices between a certain option ($5 for sure) and an uncertain lottery (risky or ambiguous). Monetary amounts for lotteries: 20 values from $5 to $120; outcome was win amount or $0.
• Risky lotteries: Explicit winning probabilities of 0.25, 0.50, or 0.75.
• Ambiguous lotteries: Probabilities partially occluded with a gray bar, at ambiguity levels 24%, 50%, or 74%. Objective winning probability for all ambiguous lotteries was controlled at 0.50.
• Stimuli displayed as bags of 100 poker chips: colored segments indicated known probabilities; gray occluder indicated ambiguity. Physical chip bags (Study 1) reinforced the mapping of lotteries to draws.
• Procedure: Training with visual/verbal explanations, 40 practice trials, and a comprehension quiz. Trials had open response windows, immediate 1 s choice confirmation, and jittered 1–2 s ITI. Option side and color-to-amount mappings were counterbalanced. 120 unique probability-by-amount combinations were each presented twice (total 240), in eight blocks. One randomly selected trial determined bonus payment. After the choice task, participants completed the STRAIN.

Measures of preferences:

• Risk attitude: Proportion of risky lottery choices over the certain option across the standard choice set (higher proportion indicates greater risk tolerance, rank-ordered across individuals).
• Ambiguity attitude: Proportion of ambiguous lottery choices, risk-corrected by subtracting each participant’s proportion of risky lottery choices to control for individual risk attitude.

Stressor exposure (STRAIN):

• Online structured inventory assessing 55 major acute and chronic stressors across 12 life domains and 5 social-psychological characteristics. For endorsed stressors, branching questions assessed severity, frequency, timing (including early life <18 vs. adulthood), and duration. Primary indices computed: total lifetime stressor count (objective exposure) and total lifetime stressor severity (subjective severity). STRAIN scores were z-transformed before regressions.

Analytic approach:

• Descriptives of STRAIN indices.
• Nonparametric correlations (Spearman’s rho) between STRAIN indices and risk-corrected ambiguous choice and risky choice (Study 1).
• Multiple linear regressions predicting ambiguous and risky choice proportions from STRAIN indices while controlling for covariates: Study 1 (age, gender); Study 2 (age, gender, RPM IQ, household income, K-6 distress).
• Exploratory regressions examining timing (early life vs. adulthood), stressor type (acute vs. chronic), primary life domain, and core social-psychological characteristics (e.g., Interpersonal Loss, Role Change/Disruption) as predictors of ambiguity tolerance, controlling for covariates.
• Two-tailed tests, p < 0.05 significance threshold.

Study 1 (n=58):

• Participants showed risk aversion and ambiguity aversion. Lottery choice proportions increased with higher risk probabilities and decreased with higher ambiguity levels.
• Lifetime STRAIN exposure and ambiguity choices: Higher total lifetime stressor count and severity were associated with fewer risk-corrected ambiguous lottery choices (Spearman’s rho: count r = −0.33, p = 0.01; severity r = −0.39, p = 0.002). No associations with risky lottery choice (count r = 0.002, p = 0.98; severity r = −0.08, p = 0.51).
• Multiple regression (controlling age, gender): STRAIN indices significantly negatively predicted risk-corrected ambiguous choices; not risky choices. • Ambiguous choice: Count B = −0.040 (SE 0.019), p = 0.035; Severity B = −0.046 (SE 0.019), p = 0.020. • Risky choice: Count B = −0.016 (SE 0.024), p = 0.504; Severity B = −0.027 (SE 0.025), p = 0.282.
• Timing and characteristics: Early life stressors (prior to age 18) significantly predicted lower ambiguity tolerance; adulthood stressors did not. Acute stressors and stressors characterized by Interpersonal Loss and Role Change/Disruption were especially related to higher ambiguity aversion (exploratory, uncorrected).
• STRAIN descriptives: Mean lifetime stressor count = 14.55 (SD 10.70; range 1–53); mean severity = 35.22 (SD 26.75; range 3–155).

Study 2 (n=188) replication with additional covariates (age, gender, RPM IQ, household income, K-6 distress):

• Replicated risk and ambiguity aversion patterns.
• STRAIN indices predicted ambiguity, not risk: • Ambiguous choice: Count B = −0.025 (SE 0.011), p = 0.024; Severity B = −0.029 (SE 0.011), p = 0.012. • Risky choice: Count B = 0.009 (SE 0.017), p = 0.589; Severity B = 0.003 (SE 0.018), p = 0.869.
• Covariates: Age, gender, IQ, income, and K-6 did not significantly predict ambiguous choice. Household income positively predicted risky choices (e.g., B ≈ 0.153, p ≈ 0.021).
• Timing: Early life stressor count and severity significantly predicted lower ambiguity tolerance (count B = −0.022, p = 0.030; severity B = −0.023, p = 0.024). Adulthood stressor count showed a trend (B = −0.021, p = 0.059); adulthood stressor severity was significant (B = −0.023, p = 0.049).
• Stressor characteristics: Interpersonal Loss again emerged as relevant (trend-level).
• STRAIN descriptives: Mean lifetime stressor count = 14.78 (SD 12.05; range 1–65); mean severity = 35.50 (SD 27.64; range 1–144).

Overall: Across two independent samples, greater cumulative lifetime stressor exposure—particularly early life stress—selectively predicted higher ambiguity aversion, with no consistent relation to risk tolerance.

The studies addressed whether cumulative lifetime stressor exposure predicts economic preferences under uncertainty, specifically ambiguity aversion. Across two independent samples, greater total lifetime stressor count and severity were robustly associated with lower willingness to accept ambiguous lotteries (after risk-correction) but showed no relation to risky lottery choices. These associations persisted after controlling for demographic and cognitive/clinical covariates (age, gender, IQ, income, psychological distress). Exploratory analyses indicated timing and type specificity: early life stressors and stressors marked by Interpersonal Loss and Role Change/Disruption were most strongly related to ambiguity aversion, with adulthood severity also contributing in the replication sample.

These findings support the hypothesis that lifetime—and especially early life—stress exposure shapes subjective beliefs and appraisals under ambiguity, increasing avoidance of options with unknown probabilities. Potential mechanisms include associative learning linking uncertainty to unpredictability and uncontrollability, heightened arousal and negativity bias toward ambiguous stimuli, and developmental programming of neural circuits (e.g., amygdala, ventromedial prefrontal cortex) that process uncertainty and subjective value. The results clarify that the psychosocial determinants of ambiguity preferences differ from those of risk preferences, advancing understanding of decision-making under uncertainty and highlighting life stress exposure as a key experiential correlate of ambiguity aversion.

This work provides the first direct evidence that cumulative lifetime stressor exposure, quantified by the STRAIN, predicts economic ambiguity aversion but not risk aversion. The association is strongest for early life stressors and stressors characterized by interpersonal loss and role disruption. These findings suggest that experiences of stress and adversity can shape how individuals appraise and choose under ambiguous conditions, with implications for everyday financial, health, and legal decisions that often involve unknown probabilities.

Future research should: (1) establish causal mechanisms using longitudinal and experimental designs; (2) delineate sensitive developmental periods within early life (e.g., pre- vs. post-puberty) that most strongly influence ambiguity preferences; (3) probe computational and neural pathways (e.g., belief formation, learning under uncertainty, amygdala–vmPFC circuitry); and (4) examine generalizability across populations and real-world decision domains.

• Correlational design precludes causal inference; reverse causation or unmeasured common factors cannot be ruled out.
• Early life defined as <18 years may be too coarse; more granular developmental windows could reveal nuanced effects.
• Retrospective assessment of lifetime stressors may be subject to recall or disclosure biases, despite STRAIN’s robustness against mood and social desirability influences.
• Exploratory analyses of stressor types/domains were not corrected for multiple comparisons and should be interpreted cautiously.
• Although several covariates (age, gender, IQ, income, psychological distress) were controlled, additional unmeasured confounds may remain.