Individual-subject Functional Localization Increases Univariate Activation but Not Multivariate Pattern...

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Neuroimaging

Background:

• The frontoparietal multiple-demand (MD) network is crucial for goal-directed behavior.
• Multivoxel pattern analysis (MVPA) of fMRI data offers detailed insights into brain function, but low decoding levels have limited its application in the MD network.
• Standard MVPA approaches use group template ROIs, which may not account for individual differences in brain structure.

Purpose of the Study:

• To investigate whether using subject-specific regions of interest (ROIs) enhances univariate and multivariate fMRI analysis within the MD network compared to standard group template ROIs.
• To determine if subject-specific ROIs improve the decodability of neural representations and univariate activity.
• To assess the impact of different localizer tasks on ROI definition and subsequent analysis.

Main Methods:

• Employed three localizer tasks (spatial working memory, verbal working memory, Stroop) to define subject-specific ROIs within the MD network.
• Systematically evaluated univariate and multivariate fMRI results in a separate rule-based criterion task using both subject-specific ROIs and a group template ROI.
• Analyzed the reliability and similarity of univariate and multivariate results across different subject-specific ROIs.

Main Results:

• Subject-specific ROIs significantly improved univariate analysis results, showing increased activity compared to the group template ROI.
• No significant benefit or cost in multivariate analysis (discriminability) was observed when using subject-specific ROIs compared to the group template.
• Both univariate and multivariate results were consistent across subject-specific ROIs defined by the different localizer tasks.

Conclusions:

• Subject-specific ROIs offer advantages for univariate fMRI analysis within the frontoparietal MD network.
• The benefits of subject-specific ROIs do not extend to multivariate decoding in this network.
• These findings provide empirical support for the use of individual ROIs in fMRI studies of cognitive control, promoting standardization and comparability across research.