Looking just below the eyes is optimal across face recognition tasks

Area of Science:

• Cognitive Neuroscience
• Visual Perception
• Computational Neuroscience

Background:

• Humans frequently fixate on the eyes when viewing faces, a behavior linked to social interaction and gaze direction.
• The precise role of eye movements toward the eyes in fundamental face perception tasks like identity, emotion, or gender recognition remains debated.

Purpose of the Study:

• To investigate whether eye movements during face perception serve an adaptive strategy to optimize performance in crucial perceptual tasks.
• To determine if gaze behavior in face recognition is a byproduct of social eye contact or has functional importance for basic perception.

Main Methods:

• Developed a Bayesian ideal observer model incorporating foveal visual constraints (foveated ideal observer).
• Compared model predictions of optimal fixation points with human eye-tracking data across identity, gender, and emotion recognition tasks.
• Validated the model using a large database of real-world faces.

Main Results:

• Human eye movements during face perception were found to maximize perceptual performance for determining identity, gender, and emotional state.
• Optimal fixation points, which varied slightly across tasks, were accurately predicted by the foveated ideal observer model.
• Models disregarding visual foveation or using center-of-gravity fixations failed to predict human gaze behavior.

Conclusions:

• Human visual system employs an adaptive strategy guiding eye movements to optimize face recognition and related perceptual tasks.
• The optimal gaze strategy involves fixating slightly below the eyes, leveraging the visual system's foveated nature.
• These findings suggest a functional role for eye movements in basic face perception beyond social signaling.