Identifying task-relevant spectral signatures of perceptual categorization in the human cortex

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• The human brain efficiently categorizes sensory input, like faces and objects, using specialized neural networks.
• Neural activity integrates sensory data with cognitive processes, but the specific spectral features reflecting this categorization remain unclear.

Purpose of the Study:

• To investigate how different neural response components contribute to categorical information processing in the human brain.
• To identify spectral signatures associated with visual category perception using intracranial recordings.

Main Methods:

• Trained Random Forest classification models to decode eight perceptual categories from human intracranial signals (4-150 Hz) in 100 subjects.
• Analyzed spectral features identified as relevant by the algorithm for perceptual decoding.
• Examined power changes in low (4-50 Hz) and high (50-150 Hz) frequency bands across sensory and non-sensory cortices.

Main Results:

• Demonstrated that network selectivity for categories correlates with specific patterns of neural power increases and decreases.
• Identified distinct spectral signatures in both low and high frequency bands associated with visual categorization.
• Revealed these signatures in both sensory and non-sensory brain regions.

Conclusions:

• Specific spectrotemporal patterns in neural activity characterize the mechanisms of visual category perception.
• This study uncovers novel spectral signatures related to visual categorization not previously reported in the literature.
• Findings highlight the importance of analyzing both low and high frequency bands for understanding neural processing of visual categories.