The Bandwidth of Diagnostic Horizontal Structure for Face Identification

Area of Science:

• Cognitive psychology
• Neuroscience
• Computer vision

Background:

• Horizontal spatial frequencies are crucial for face identity.
• Sensitivity to these frequencies impacts face identification and the inversion effect.
• The precise bandwidth of this information and human processing remains unclear.

Purpose of the Study:

• To characterize the bandwidth of diagnostically relevant spatial frequency information for face identification.
• To investigate how human visual system tuning aligns with the distribution of this information.
• To compare human performance with an ideal observer model.

Main Methods:

• A 10-alternative forced choice face identification task was employed.
• Faces were filtered to isolate horizontal or vertical structures at varying bandwidths.
• Manipulated stimuli contained diagnostic information in unknown orientation bands.

Main Results:

• The most diagnostic information for face identification is concentrated in a narrow horizontal orientation band.
• Human observers utilize a significantly wider range of orientations than is optimal.
• Performance varied systematically with filter bandwidth, indicating sensitivity to orientation bands.

Conclusions:

• Face identity information is primarily conveyed by specific horizontal spatial frequencies.
• Human face recognition is less spectrally tuned than theoretically optimal.
• Understanding orientation processing is key to understanding face perception.