The impact of stimulus size and orientation on individual face coding in monkey face-selective cortex

Area of Science:

• Neuroscience
• Primate Vision
• Computational Neuroscience

Background:

• Face-selective neurons in the temporal cortex respond differently to various faces.
• The face-selective area ML in the middle Superior Temporal Sulcus is crucial for face processing.
• Understanding how these neurons respond to transformations like inversion and size changes is key to understanding face recognition.

Purpose of the Study:

• To investigate if neural response patterns in area ML can distinguish individual faces despite affine transformations.
• To assess the tolerance of face-selective neurons to changes in stimulus size and picture-plane inversion.
• To determine the impact of size and orientation on the reliability of decoding individual faces from ML neural activity.

Main Methods:

• Recorded activity from 57 ML neurons in two awake, fixating monkeys.
• Presented face stimuli at two sizes (5 and 10 degrees) and two orientations (upright and inverted).
• Used a classifier to predict face identity from neural response patterns under different conditions.

Main Results:

• Distinct patterns of activity across ML neurons reliably distinguished individual faces within specific size and orientation conditions.
• Classifier performance remained above chance for both upright and inverted faces, despite reduced average response to inverted faces.
• Decoding was largely preserved across stimulus size changes, but not across orientation changes (upright vs. inverted).

Conclusions:

• Individual faces can be decoded from ML neural activity irrespective of size and orientation, indicating robustness in face representation.
• Qualitatively different response patterns exist for upright versus inverted faces, suggesting orientation-specific processing.
• The findings contribute to understanding the neural basis of invariant face recognition in primates.