Brain-machine interfaces are better when decoding exploits the intrinsic manifold geometry of brain activity, by @elbusch.bsky.social et al. Translates electrophys work (sadtler 2014 Nat Neuro) to humans, bridging fields (sys and cog neuro) that need far more crosstalk. doi.org/10.1038/s415...

Interesting work on low dimensionality of higher order neural representations in PFC.

Really delighted to get copies of our new book, Elements of Functional Magnetic Resonance Imaging. A labor of love over the past 10 years, along with @fmristats. Available for pre-order: mitpress.mit.edu/978026204504...

Good summary of the state of our understanding of representational drift. Personally interested in it as a solution to catastrophic forgetting in artificial neural nets. Hasn’t been fully established or profiled in humans yet either, raising an interesting possible direction for future work.

I’m very sad to share that Howard Fields has died. He made seminal discoveries in pain modulation, placebo, and addiction. He was also a wonderful mentor and friend who taught me how to think about the brain and the joy of scientific discovery. www.iasp-pain.org/publications...

This is the May cover of NN but it has already been debunked on biorxiv www.biorxiv.org/content/10.6... Such a clear statement about the ridiculous pace of publishing (not just NN).

Backprop in artificial neural networks (ANNs) is famously biologically implausible. This paper hints at an alternative solution to the credit assignment problem: spatial segregation along cortical dendrites instead of feed fwd v. err feedback type temporal segregation. www.nature.com/articles/s41...

Brain maps differ between people, but whether encoded information also differs depends on where you look. In some regions different layouts encode the same information, and representations converge like in artificial neural networks (ANNs). New preprint: www.biorxiv.org/content/10.6...

Some billionaire recently advocated against introspection (tinyurl.com/y2rkty43), claiming it was a modern invention. Reminds me of a paper claiming introspection may be culturally contingent, and rises and falls with social cataclysms like world wars (tinyurl.com/4yttw84s). Fun read.

Dimensionality reduction may be the wrong approach to understanding neural representations. Our new paper shows that across human visual cortex, dimensionality is unbounded and scales with dataset size—we show this across nearly four orders of magnitude. journals.plos.org/ploscompbiol...

www.economist.com/finance-and-... There is no substitute for American scientific investment. Since 1945, when Vannevar Bush convinced Truman of the merits of public research funding, only Russia and China have even tried to compete. If the USG stops funding science, the global economy will feel it.

Our new @NatureHumBehav paper shows why we need a holistic pain biomarker framework. ML on blood tests, brain/bone imaging & genetics predicts clinical diagnoses but falls short on subjective pain. Adding psychosocial (mood, sleep, stress) boosts both. shorturl.at/vxxeO

“A 25 percent reduction in public R&D spending would reduce GDP by … [an effect] comparable to the decline in GDP during the Great Recession.” tinyurl.com/vweadewh Meanwhile the government is proposing a 44% cut to the NIH www.science.org/content/arti...

Good article touching all key points succinctly. “U.S. government … research and development has accounted for at least 1/5 of total factor productivity growth in the U.S. business sector … a greater return than than private R & D has produced.” www.foreignaffairs.com/united-state...

Human ctrl of neural interfaces is more effective when the intrinsic neural manifold is accounted for. biorxiv.org/content/10.1... First demoed in macaque by @aaronbatista.bsky.social now translated to humans by @elbusch.bsky.social. Gotta pay attn to those animal population code manifold studies!