Identifying spatially overlapping local cortical networks with MEG

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Neuroscience

Background:

• Macroscopic measurements like electroencephalography (EEG) and magnetoencephalography (MEG) may differentiate neuronal populations.
• Previous modeling studies suggest potential for distinguishing neuronal activity based on mean field signals.

Purpose of the Study:

• To investigate if distinct orientation columns within the primary visual cortex can be identified using MEG signals.
• To determine if MEG data can predict the orientation of visual stimuli based on activity at a single cortical location.

Main Methods:

• Utilized static, obliquely oriented grating stimuli to evoke sustained gamma oscillations in the primary visual cortex.
• Employed multivariate classifier methods on time-series MEG data from a focal cortical region.
• Analyzed both single-trial evoked responses (0-300 ms) and induced post-transient power spectra (300-2,300 ms, 20-70 Hz).

Main Results:

• Classification of stimulus orientation (left vs. right oblique) was significantly above chance for both evoked responses (11/12 datasets) and induced oscillations (10/12 datasets).
• Stimulus-specific information was preserved in sustained gamma oscillations, persisting long after initial sensory processing.
• Classification remained effective even after rank-transforming power spectra, indicating distinct network temporal signatures.

Conclusions:

• MEG signals can differentiate between neuronal populations based on their orientation columns.
• Sustained gamma oscillations contain stimulus-specific information, offering a potential marker for neural population activity.
• The temporal signatures of induced oscillations are characteristic of underlying neural networks and can be decoded.