Visual attributes of spiders associated with aversiveness in spider-fearful individuals: A machine learning analysis

Authors:
Preprint
Publisher website
Google Scholar
Abstract
Spiders are among the most diverse and evolutionarily successful taxa in the animal kingdom. Nevertheless, their popularity with humans is low, and arachnophobia, which is characterized by an intense and irrational fear of spiders, is one of the most common specific phobias. While it is clear that spiders generally evoke aversive responses in humans, it is a highly understudied question which visual attributes of spiders trigger these emotions. Here, we categorized a set of 313 images showing spiders and spider-related content according to a variety of visual attributes, and performed a machine learning analysis to investigate which attributes were associated with the mean fear, disgust, and approach-avoidance ratings of 152 spider-fearful adults. Predictive models were able to account for approximately 70%, 67%, and 60% of the variance in mean fear, disgust, and approach-avoidance ratings, respectively. Visual attributes indicating the subjective size, texture, prominence of the legs, subjective distance, and the overall presence of a spider were significant predictors of aversive responses. Our results shed light on the visual features that contribute to spider-related aversiveness. This knowledge may help in the selection and creation of stimuli that are particularly effective in improving classical exposure therapy, as well as novel computerized exposure-based treatments.