Lab’s latest pub out in NeuroImage, "Task activations produce spurious but systematic inflation of task functional connectivity estimates" doi.org/10.1016/j.neur…
Evoked activations inflate task functional connectivity, but corrected by subtracting mean evoked response
Evoked activations inflate task functional connectivity, but corrected by subtracting mean evoked response
An alternate link to the paper is on our lab's website: colelab.org/#publications
Easy-to-read Python code available demonstrating the effect (task-state functional connectivity inflation by task activations), as well as how to correct it: github.com/ColeLab/TaskFC…
We estimated that task-state functional connectivity tends to be inflated about 3x in empirical fMRI data due to evoked co-activations (when not corrected for)
We also found that idiosyncratic task-state functional connectivity inflation patterns (even connectivity decreases) could be predicted based on evoked co-activation patterns in empirical fMRI data
Realization during this project: mean evoked neural responses are only a small percentage of overall variance during task performance
Most neural activity is either spontaneous/intrinsic or evoked that varies trial-to-trial. This activity drives task functional connectivity
Most neural activity is either spontaneous/intrinsic or evoked that varies trial-to-trial. This activity drives task functional connectivity
There is an extensive Frequently Asked Questions (FAQ) table in the supplementary materials:
colelab.org/pubs/Cole2019_…
Sample of questions:
Does removing the mean evoked response remove too little variance?
Does removing the mean evoked response remove too much variance?
colelab.org/pubs/Cole2019_…
Sample of questions:
Does removing the mean evoked response remove too little variance?
Does removing the mean evoked response remove too much variance?
