Jack Gallant
Cognitive, Systems and Computational Neuroscientist, Professor at UC Berkeley, and lab head. Check out our lab web site http://gallantlab.org For the latest news, publications, brain viewers, code and tutorials, and data.
- Voxelwise Encoding Models do some things differently from other GLM-related methods because the roots of VEM are in neurophysiology, not psychology. This 2006 paper describes system identification for neurophysiology. Change "neurons" to "voxels" and it all still applies. tinyurl.com/wu-etal-2006
- Are you on neurotree.org? This site was started @ UCB by my former grad students Stephen David (now Prof at OHSU) and Ben Hayden (now Prof at Baylor), and they've run it ever since (see tinyurl.com/neurotree2026 for history). Please keep your neurotree info updated, it is a great resource!
- Stephen got one thing wrong in the Transmitter article. He forgot that the tree was motivated by a hand drawn tree of my academic lineage that was posted in our lab. That got him interested in the larger lineage of all neuroscientists. And he was right, that lineage information is valuable!
- A recent paper made big waves arguing that fMRI data are hopelessly confounded and uninterpretable. But I'm always suspicious of strong claims that go against a large body of well-supported science. This important post by @alexanderhuth.bsky.social suggests that the paper is fundamentally flawed.
- Voxelwise Encoding Models (VEMs) are a great framework for modeling fMRI data, but it can be difficult to implement. We've made VEM accessible by providing software, tutorials and reviews that guide its use an implementation. Get it here: gallantlab.org/blog/2025-12... #neuroscience, #neuroimaging
- Sorry folks if Bluesky is truncating the link to our VEM blog post, here is a shorter link to the top level of the blog. The VEM post is at the top. gallantlab.org/blog/
- Put this new preprint by T. Zhang on your reading list! People drove a car to navigate in VR. Voxelwise encoding models were fit using 38 feature spaces (28,134 features!). We show the cortical navigation network comprises 11 regions arranged in functional gradients. www.biorxiv.org/content/10.6...
- This cool new study is, as far as I know, the largest computational modeling effort ever undertaken in fMRI. The methods developed for this landmark project push the boundaries on the breadth and depth of information that can be recovered from a single fMRI study. Great work from Dr. Tianjiao Zhang!
