Psychological identity develops through the interaction of internal traits and external sociopolitical conditions. In Iran, repeated exposure to war, sanctions, and uncertainty has shaped identity across generations. This study uses machine learning to identify latent identity patterns and generational differences in a non-clinical Iranian sample (N = 620; ages 18–60).
Deep embedded clustering (DEC) was applied to decision-making and self-regulation traits. Four psychological profiles were identified, each defined by different combinations of impulsivity, coping flexibility, emotional regulation, and decisional insecurity. Profile entropy and drift indices were used to describe internal stability and ambiguity. SHAP analysis and counterfactual simulations were used to examine which traits most influenced potential profile change.
War-experienced adults were more likely to show stable but emotionally restricted profiles. Post-war adults more often showed profiles with higher entropy and less coherence. A simple transition model based on psychological proximity and entropy was used to explore possible movement between profiles.
The findings indicate that unsupervised learning approaches can identify non-clinical psychological risk and resilience patterns in culturally specific contexts. Generational differences suggest that sociopolitical exposure is associated with variation in identity organisation. These results contribute to the understanding of psychological adaptation in populations exposed to chronic structural stress.
Psychological identity machine learning deep embedded clustering decision-making profiles entropy behavioural drift counterfactual simulation Iranian adults generational differences resilience
| Primary Language | English |
|---|---|
| Subjects | Evolutionary Computation |
| Journal Section | Research Article |
| Authors | |
| Submission Date | January 2, 2026 |
| Acceptance Date | January 23, 2026 |
| Publication Date | January 30, 2026 |
| DOI | https://doi.org/10.26650/d3ai.1854272 |
| IZ | https://izlik.org/JA39WA57KW |
| Published in Issue | Year 2026 Volume: 2 Issue: 1 |