Scene content is predominantly conveyed by high spatial frequencies in scene-selective visual cortex

Area of Science:

• Neuroscience
• Visual Perception
• Image Processing

Background:

• Complex real-world scenes contain intertwined visual features.
• The visual system extracts information by disentangling these features.
• Spatial frequencies are a key component in image representation.

Purpose of the Study:

• Investigate the role of spatial frequencies in scene representation within the human visual cortex.
• Determine whether low or high spatial frequencies preferentially carry scene content information.
• Compare neural decoding of scene categories using different spatial frequency bands.

Main Methods:

• Decoding scene categories from brain activity patterns in scene-selective visual cortex.
• Using stimuli with full, low, or high spatial frequencies, controlled for contrast and luminance.
• Analyzing activity in areas including the parahippocampal place area (PPA), retrosplenial cortex (RSC), occipital place area (OPA), and lateral occipital complex (LOC).

Main Results:

• Decoding scene categories was significantly more accurate for high spatial frequencies than low spatial frequencies in scene-selective regions.
• Decoding accuracy for high spatial frequencies matched that of full spatial frequency images in PPA, RSC, OPA, and LOC.
• A dissociation was observed in PPA: categories were decodable from both high and low frequencies in posterior PPA, but only high frequencies in anterior PPA.

Conclusions:

• High spatial frequencies are critical for conveying the content of complex real-world scenes.
• This finding contrasts with previous behavioral and computational models emphasizing low spatial frequencies.
• Results support the idea that high spatial frequencies drive neural representations of scene categories, similar to full-spectrum images.