Vulnerability and the computational logic of fear: insights from the horror genre

Abstract

Fear is a universal feature of storytelling, yet the structural conditions that make fictional threats compelling remain poorly understood. Here, we propose the Protagonist Vulnerability Index (PVI), an evolutionarily grounded computational approach to explain why some narratives evoke stronger fear responses than others. PVI quantifies protagonist vulnerability by assessing the imbalance in formidability between protagonists and antagonists and the risk of attack faced by the protagonist. Across 691 films, higher PVI values predicted classification as horror, the presence of fear-related keywords in non-horror films, and stronger physiological fear responses indexed by heart rate. Linking film preferences to psychological and demographic data from more than 3.5 million individuals on Facebook, we found that preference for high-PVI films was associated with lower agreeableness, conscientiousness, and extraversion, and with higher openness. Openness moderated the negative association between neuroticism and engagement with fear-related content, indicating that curiosity can counteract threat avoidance in anxious individuals. These findings clarify the structural and psychological conditions that activate evolved threat-management systems. The results show how horror operates as a narrative simulation of extreme formidability asymmetry, and provide a framework for predicting, and potentially engineering, fear in fiction.