Thank you!

GitHub - DiedrichsenLab/MultiTaskBattery Contribute to DiedrichsenLab/MultiTaskBattery development by creating an account on GitHub.

The MultiTaskBattery toolbox is live: 20+ implemented tasks, object-oriented design and a battery optimizer built in!

Docs: multitaskbattery.readthedocs.io/en/latest/
Github: github.com/DiedrichsenL...

With @carobellum.bsky.social Jinkang Derrick Xiang @diedrichsenjorn.bsky.social 🙌

How you arrange your tasks also matters more than you’d think.

Interspersing all tasks in every run lets you compute all contrasts within-run and leverage resting baselines across runs. Grouped designs force noisier between-run comparisons.

Multi-task batteries can parcellate entire brain structures. But they're only as good as their tasks.

We derived a data-driven task selection criterion (Minimal Collinearity) that outperforms others for parcellation & connectivity. 56 task maps released for battery optimization.

We tested both on cerebellar language mapping. Same subjects, same scan time.

Multi-task: 2× spatial consistency across individuals, and it cleanly separates language from neighboring working memory regions that the single-contrast kept pulling in.

Single-contrast localizers have been the workhorse of functional mapping — simple & effective.

But to isolate a coherent region or study region size, multi-task is preferable — assigning voxels by what they respond to, not how strongly. Threshold-free and robust to data quality.

You have 10 minutes of task fMRI to map functional brain regions in an individual. Three decisions matter:

How do you localize regions?
Which tasks do you include?
How do you arrange tasks in your experiment?

Here, we tackle all three! Preprint👇

www.biorxiv.org/content/10.6...