Happy to share that our paper on sensory robustness in recurrent vision models has been published in Patterns (@cp-patterns.bsky.social) 🎉 with @marco-d.bsky.social, Karl Friston, Giovanni Pezzulo & @siegellab.bsky.social TL;DR: Top-down feedback + neural variability = Robust Vision #NeuroAI 🧵👇

Show Post

A new opinion piece argues that alpha rhythms in decoding may just be a confound of background alpha, citing our recent paper. But our findings are the *exact opposite* of what their confound would predict.

Show Post

I don’t totally understand the concern… I thought folks did these analyses to test whether single-trial activity was getting closer/further from the average-data (+ phase-random) boundary? Does your concern depend on whether the decoder is aligned with the average signal? eg all positive weights

Show Post

there's a U-shaped relationship between the rotational distance of a thing from a U and whether you call it a U-shaped relationship

Show Post

How much representational diversity exists around a trained network? A lot, it turns out! Moving a ViT’s weights by 1.3% while preserving its predictions, its representations end up less similar to the original than a random-init ViT’s ‼️ New paper: Hessian Null Space Continuation (HNC) 💫🧵 (1/12)

Show Post