Decoding in the Fourth Dimension: Classification of Temporal Patterns and Their Generalization Across Locations.

Area of Science:

• Neuroimaging and Computational Neuroscience
• Analysis of electrophysiological data

Background:

• Neuroimaging decoding commonly uses spatial features, overlooking temporal information in electrophysiological recordings.
• Techniques like electroencephalography (EEG) and magnetoencephalography (MEG) offer rich temporal data crucial for understanding brain function.

Purpose of the Study:

• Introduce the time-GAL toolbox for decoding neural time series using temporal information.
• Quantify decodable information in neural time series and assess cross-decoding generalization across sensor locations.

Main Methods:

• Developed the time-GAL toolbox implementing a decoding method based on temporal features in electrophysiological recordings.
• Quantified decodable information within neural time series.
• Utilized generalization across location (GAL) to characterize relationships between sensor locations via cross-decoding.

Main Results:

• Successfully decoded experimental conditions using temporal features from neural time series in two datasets (event-related potentials and steady-state visual evoked potentials).
• Demonstrated spatial cross-decoding in brain regions associated with visual and affective processing.
• Validated the effectiveness of the time-GAL toolbox for analyzing neural time series.

Conclusions:

• The time-GAL toolbox provides a promising, assumption-free method for analyzing neural time series across diverse paradigms and domains.
• The approach effectively quantifies differences in temporal patterns of neural information processing.
• Identified shared temporal processing patterns across different sensor locations.