Linking brain-wide multivoxel activation patterns to behaviour: Examples from language and math.

Area of Science:

• Cognitive Neuroscience
• Neuroimaging Analysis

Background:

• Establishing direct brain-behaviour connections is a primary goal in cognitive neuroscience.
• Standard functional magnetic resonance imaging (fMRI) analysis methods often involve numerous choices, potentially introducing bias into brain-behaviour correlations.
• Both traditional region-of-interest (ROI) and multivariate pattern analysis (MVPA) techniques can be susceptible to selection bias through voxel selection or data reduction.

Purpose of the Study:

• To demonstrate a novel fMRI analysis method that avoids voxel selection and data reduction.
• To reveal strong brain-behaviour correlations using a simple, direct approach.
• To investigate the efficacy of whole-brain linear regression in uncovering neural correlates of behaviour.

Main Methods:

• Applied plain linear regression to the entire brain volume as a single multi-voxel pattern.
• Treated each whole-brain volume as a single data point for analysis.
• Used only neural data and condition labels, without behavioural information, as input for the classifier.

Main Results:

• Strong brain-behaviour correlations were identified without any voxel selection or data reduction.
• A simple whole-brain linear regression model yielded correlations comparable to those from specific ROIs or more complex classifiers.
• More sophisticated classifiers like linear SVM and regularized logistic regression produced similar results to the basic linear regression.

Conclusions:

• A simple, whole-brain linear regression approach can effectively identify brain-behaviour relationships without the biases inherent in voxel selection or data reduction.
• This method provides a rigorous, direct, and unbiased strategy for investigating brain-behaviour connections in cognitive neuroscience.
• The findings suggest that complex analytical steps may not always be necessary for robustly uncovering neural correlates of behaviour.