Predicting Marital Forgiveness after Infidelity Using Explainable Artificial Intelligence and Relational Trauma Indicators

Abstract

The present study aimed to predict marital forgiveness following infidelity using explainable artificial intelligence models and relational trauma indicators while identifying the relative importance of trust, attachment insecurity, psychological distress, and relationship quality in determining forgiveness outcomes among married individuals affected by extramarital relationships. This cross-sectional predictive study was conducted among 548 married individuals in Turkey who had experienced emotional, sexual, or combined forms of partner infidelity and remained in their relationships following disclosure. Participants were recruited from counseling centers, family therapy clinics, and online support networks. Data were collected using standardized measures of marital forgiveness, relational trauma, attachment insecurity, psychological distress, dyadic trust, and relationship quality. Several machine learning algorithms, including Elastic Net Regression, Support Vector Regression, Random Forest Regression, LightGBM, and Extreme Gradient Boosting (XGBoost), were developed and evaluated. Model performance was assessed using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Explainable Artificial Intelligence techniques based on Shapley Additive Explanations (SHAP) were employed to identify and interpret the relative contribution of each predictor variable to forgiveness outcomes. Correlation analyses indicated that marital forgiveness was significantly and negatively associated with relational trauma, attachment anxiety, attachment avoidance, and psychological distress, while showing significant positive relationships with dyadic trust and relationship quality (p < .01). Among the evaluated machine learning algorithms, the XGBoost model demonstrated the highest predictive performance, explaining 84% of the variance in marital forgiveness (R² = .84), followed by LightGBM (R² = .81) and Random Forest Regression (R² = .78). SHAP analyses revealed that dyadic trust was the strongest predictor of forgiveness, followed by relational trauma, relationship quality, attachment anxiety, psychological distress, attachment avoidance, and time since disclosure. Cross-validation analyses confirmed the stability and robustness of the final model, with R² values consistently ranging between .83 and .85 across validation folds. The findings demonstrate that marital forgiveness following infidelity can be accurately predicted through explainable artificial intelligence models that integrate relational trauma indicators and psychological relationship variables. Trust restoration and trauma recovery emerged as the most influential determinants of forgiveness, highlighting the importance of trauma-informed and attachment-focused interventions in couples recovering from betrayal. The application of explainable artificial intelligence not only improved predictive accuracy but also enhanced understanding of the mechanisms underlying forgiveness after infidelity, offering valuable implications for assessment, treatment planning, and the development of personalized therapeutic approaches for distressed couples.