Common Sequential Organization of Face Processing in the Human Brain and Convolutional Neural Networks

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Human face processing involves distinct stages: detection and recognition.
• The precise temporal order of these stages in the brain remains debated.
• Previous research presents conflicting evidence regarding sequential face processing.

Purpose of the Study:

• To elucidate the sequential organization of face detection and recognition in the human brain.
• To investigate the computational underpinnings of this sequential processing.
• To determine if neural network models replicate observed human brain patterns.

Main Methods:

• Utilized intracranial electroencephalography (EEG) from human subjects.
• Applied representation similarity analysis to EEG data.
• Employed convolutional neural networks (CNNs) for computational modeling.

Main Results:

• Demonstrated a consistent "detection-first, recognition-later" pattern in human brain activity.
• CNNs trained for face recognition also exhibited this sequential pattern.
• The observed sequential organization emerged gradually during network training and was specific to correct predictions.

Conclusions:

• The human brain sequentially organizes face detection before recognition.
• This organization likely reflects computational optimization principles.
• Neural network models provide support for a computational account of this brain mechanism.