I think a lot of folks working on adaptive DBS don't realize they're working on a canonical control theoretic problem - without having any of the control theoretic background needed to solve it.

Contrasted with trying to reinvent your own approach to inference.

Also: engineering approaches to inference are strictly better for complex systems. Period.

If your analysis pipeline is not amenable to engineering approaches to inference, then you're better off spending time finding a way to represent your data so that it becomes amenable.

Why linearize?

I assume a requirement of consistency? "The sky is blue and not any other color. The sky is red and not any other color." can be written mathematically, but it's quite different than a sentence that doesn't contradict itself.

Funny enough, I think they meant "in a set theoretic way" informally...

$100 I can guess their demographics... best not to conflate conditional distributions with joint distributions...

as one of those "we" you keep bringing up - I found that abstract + literature quite interesting and important.

I believe this (friendly piece) aligns with your question about the absurdity of reducing strong emotions & compulsions to happenings in the brain. Bonus: it captures a diversity of researcher perspectives about the best ways to think about it.

www.thetransmitter.org/the-big-pict...

I love the model comparison approach (closer to the medical inference engine) - but if your models are hyperparametrized, there's an implicit "big data" in there, no?

Maybe not "foundation" since that's tighter linked to Transformer/LLM, but I still don't see an explicit "small data" vision.

I'd put these on the NeuroAI vision board:

@tyrellturing.bsky.social's Deep learning framework
www.nature.com/articles/s41...

@tonyzador.bsky.social's Next-gen AI through neuroAI
www.nature.com/articles/s41...

@adriendoerig.bsky.social's Neuroconnectionist framework
www.nature.com/articles/s41...

Yup (where foundation ≡ big data derived).

Has anyone outlined + whipped up votes + published an alternative vision in a similarly public way?

I want to be supportive of my NeuroAI colleagues, but who is working on alternative visions?

What about one that seeks to bring ML/AI to where clinical neuro(engineering) is already working magic?