Excited to share a preprint of work done together with folks from UCSF (@littleneuro), UC Berkeley, and U of Washington, using neural decoding and adaptive DBS to selectively amplify movement when intended and reduce dyskinesia when it isn't

@FiveThirtyEight About 1.5%? I wrote a jupyter notebook on Google Colab showing my approach using ratios of spherical surface area (. Here's a gif showing how that probability changes as we get less and less picky about our viewing angle.

RT @biorxiv_neursci: Selective modulation of population dynamics during neuroprosthetic skill learning #biorxiv_ne…

@cogcelia Hahaha let me know how it goes! I was afraid that I’d missed the crucial window of youth with my dearest Podrick, but if your cat can learn than maybe so can mine!

@cogcelia @gottapatchemall +1 for the comment, but... What. Is. This. GIF.......

RT @cogcelia: More exciting news from the Wallis Lab: we're teaming up w/ Dr. Tirin Moore & Dr. Krishna Shenoy to study attention & decisio…

RT @tsay_jonathan: To maintain calibration, our motor system adapts to changes in the body (e.g., fatigue) and environment (e.g., a windy d…

RT @FiveThirtyEight: Our 2020 Election Forecast:

RT @TessyMThomas: Our manuscript posted on medRxiv demonstrates the potential for simultaneous classification of gestures on both hands usi…

@TessyMThomas Btw I saw your medrxiv preprint and it’s on my reading list! Excited to check it out. Looks like you’ve been finding some really cool things in this biman control space.

@TessyMThomas ... the corpus callosum, so interhem cortical networks def play a role. We are using some new behavioral manipulations to ask questions surrounding these ideas as well.

@TessyMThomas Great Q! I think it’s very likely that it would. I’m intrigued by the idea that it reflects a bilat network that can control both limbs as a unified plant. Much of our arm use requires coordinating both sides, and we know that some forms of biman coord/interference depends on...

And BIG S/O to my incredible co-authors @c_mmerick @ivryrich Joni and Jose. Such incredible people and scientists. (9/9)

Of course, this is a preprint and we would LOVE to hear any thoughts/feedback! And S/O to some of the cool people and their work that motivated some of the questions we asked here @SpecificAmes @MarkChurchland @K_P_Cross @ScottLIMBlab (and any co-authors I couldn’t find!) (8/9)

These results may suggest two functional layers in the MC network: an effector-specific output layer and a bilaterally distributed computational layer. They’ve also proven essential for making sense of some of our other results that we hope to release soon (stay tuned!) (7/9)

Signals that were shared within single-units carried information capable of classifying target reaches for either arm and existed within a shared subspace for both arms. Separation of arm-specific signals came only in the simple form of localized variance (6/9)

This “localized” component emerged within PMd during preparation, became most pronounced following movement onset when M1 became strongly engaged, and principally involved the contralateral hemisphere. Now what about those units that liked both arms!? (5/9)

We found many units that modulated their activity during trials of either arm, but those that were dedicated to a single arm had much stronger modulation. This led to the majority of variance for each arm localizing within mutually exclusive sub-populations (4/9)

These questions constrain the set of possibilities for each hemisphere’s role in motor control, and have conflicting (or absent) answers in the existing literature. We addressed them (+ a bit more) using an instructed-delay task in NHPs while recording from bilateral PMd/M1 (3/9)