Bidirectional predictive modeling of narcissists’ social exclusion and their hostile interpretations: a deep learning-based exploration of cognitive mechanisms

TL;DR

This study uses deep learning to explore how narcissistic traits and social exclusion interact, and how self-affirmation can reduce hostile interpretations.

Contribution

A novel deep learning model (HIBPN) is proposed to predict and explain the bidirectional relationship between narcissism, social exclusion, and hostile interpretations.

Findings

The HIBPN model achieved 69.9% and 83.2% predictive accuracy in modeling interactions between narcissism, social exclusion, and hostile interpretation.

Narcissistic traits were identified as the core driving variable, accounting for up to 75.73% of model weights.

Preventive self-affirmation reduced hostile interpretation bias in implicit narcissists by 5.56% in social acceptance predictions.

Abstract

This study aimed to clarify the interaction mechanism among hostile interpretation, narcissistic personality traits, and social exclusion/acceptance situations, address the limitation of traditional linear analysis, construct an accurate “trait-situation-cognition” prediction model (Hostile Interpretation & Bidirectional Prediction Network, HIBPN), and verify the intervention effect of self-affirmation on social exclusion. Two sequential experiments were designed: Experiment 1 enrolled undergraduate and postgraduate students, divided them by narcissistic traits into dominant narcissism, implicit narcissism, and neutral control groups, manipulated situations via Cyberball 5.0, quantified hostile interpretation biases, and validated situational effectiveness to provide modeling data. Experiment 2 selected implicit narcissists from Experiment 1, randomly assigning them to the self-affirmation intervention group (completing a value- and trait-oriented self-affirmation task) and control group (completing a neutral task). Consistent manipulation and tools with Experiment 1 were used to verify the intervention effect. The model integrated situational variables and narcissistic traits, establishing a closed-loop “antecedent-social exclusion-cognition” system and a bidirectional inference structure (the forward chain predicts social exclusion probability, and the reverse chain deduces the path of hostile interpretation bias). Results: In Experiment 1, under the 1:9 training-test set ratio, the predictive efficacies of the model’s two paths reached 69.9 and 83.2%, confirming significant interactions among the three variables; narcissistic traits accounted for 33.55 and 75.73% of the weights, serving as the core driving variable. In Experiment 2, preventive self-affirmation prior to social exclusion significantly reduced implicit narcissists’ hostile interpretation bias, with their subjectively predicted social acceptance rate 5.56% higher than that of the control group. This study reveals the dynamic interaction mechanism via HIBPN, providing a new cognitive prediction framework, and confirms the intervention value of preventive self-affirmation, offering a feasible approach to maintaining mental health.

Figures