Researcher at Google and CIFAR Fellow, working on the intersection of machine learning and neuroscience in Montréal (academic affiliations: @mcgill.ca and @mila-quebec.bsky.social).
Blake Richards
Loading...
How do brains plan actions towards goals?
To get at this question we studied mice navigating complex mazes as goals changed on every trial 🧵
Work with @thomasakam.bsky.social @behrenstimb.bsky.social @kristorpjensen.bsky.social now on BioRxiv: www.biorxiv.org/content/10.6...
Our new paper is out this week in Nature Neuroscience! www.nature.com/articles/s41... We built a BCI that works with the brain's natural geometry — and we found that people could learn to play a video game with their brains in <1 hr of training. This efficiency is groundbreaking & here's why:
1/7 Excited to share my last PhD article, just accepted to ICML 2026! In it, we (me, Alexandre Payeur, Guillaume Lajoie) used dynamical systems theory to study "local" learning in linear recurrent neural networks. See link for the paper, and thread for a brief summary. arxiv.org/abs/2606.00243
Video
Charlatans taking advantage of vulnerable people - just the lowest, most grotesque sort of behavior...
www.theguardian.com/society/2026...
hot damn. looks important.
www.science.org/doi/10.1126/...
See the paper card (for link and the blog) on the recent manuscript on inhibitory neurons across different layers of the cortex!
ps...the universality paper (quoted tweet) will be a follow up on this one.
neurovium.science/papers/Inhib...
NEW PAPER. Why do larger networks train better?
"Because they contain more candidate *sub*networks that can learn the task" → lottery tickets
This popular explanation uses an appealing but misleading metaphor🧵
We propose an intuitive alternative grounded in theory: escape dimensions
Busch et al. use nonlinear neural manifolds to help humans gain rapid control over a noninvasive brain–computer interface, allowing them to learn how to play a video game with real-time fMRI neurofeed...
www.nature.com
Biological and neuromorphic recurrent neural networks (RNNs) are subject to spatial and temporal locality constraints on the information that can plausibly be used during learning. A common strategy t...
Bring your cell typing questions to our "Describe Your Neurons like the Allen Institute" Q&A workshop.
Our experts will answer your questions live on cell type tools, datasets, and real-world applications.
🎟️ https://alleninstitute.org/events/neuroscience-cell-types-webinars
1/ We work hard to factor "noise" out of behavior. But the brain doesn't.
Strip away noise and you may miss what the brain evolved to do
I wanted to share this pre-Bluesky paper where we found premotor (M2) corticostriatal circuits encode a broad history of behavior, beyond just action + reward 👇
Kevin Mitchell
Desperate US parents pay up to $20,000 a session for a procedure scientists say could be bogus
Previous brain-wide association studies (BWAS) have linked specific environmental and behavioral variables to brain variability. In this work, we mapped 649 variables to children’s brains and compared...
Prior experience is used by the brain to guide adaptive behaviour during decision making. Here, the authors show that mice also selectively use information learned through recent and longer-term exper...
This work shows that cortical interneuron morphoelectric properties are organized by two separable axes: molecularly related, depth-independent cell-type structure and gradual, depth-dependent shifts ...
Chihuly gardens is my "pilgrimage" whenever I come to Seattle. The sacred beauty of colors and forms, this time in the "Chandelier" room resonated w the manuscript that Felipe Yanez & I are working on together: "The universality of form/function of cortical inhibitory neurons"...
Le chef du Bloc québécois, Yves-François Blanchet, affirme que le projet de train à grande vitesse (TGV) de la société Alto ne va pas dans la bonne direction et que son parti ne peut pas l’appuyer.
