Visual looming is a primitive for human emotion.

Area of Science:

• Neuroscience
• Computer Science
• Animal Behavior

Background:

• Defensive behaviors to looming threats are conserved across species.
• The mammalian superior colliculus processes looming information for threat detection.
• The link between looming detection systems and human emotional experience is unclear.

Approach:

• Developed a shallow convolutional neural network (CNN) inspired by the Drosophila visual system.
• Tested the CNN's ability to predict defensive blinking in infants and superior colliculus responses in adults.
• Assessed the CNN's predictive power for self-reported human emotions using naturalistic video clips.

Key Points:

• The CNN accurately predicts defensive blinking to looming stimuli in human infants.
• The CNN replicates superior colliculus responses to optical expansion in adult humans.
• CNN's processing of visual stimuli predicts subjective emotional arousal in humans.

Conclusions:

• A computational model of looming detection can predict both defensive behaviors and emotional experiences.
• Findings suggest a connection between evolutionarily conserved survival systems and human emotion.
• This work bridges computational neuroscience, developmental psychology, and affective science.