Day 7

Dec 9, 2025

Back to How Do We Learn Dec 2025 Aliveline

Day 7

It’s been a week since I started this aliveline. And I feel like I’ve done nothing! Amazing. This timeline is so insanely compressed - I am starting to feel a bit of pressure which is kind of against the point of this. Have to keep reminding myself that failure is expected & the goal is to try to fail in interesting ways. How can I fail more interestingly? What are super weird out of the box creative ways to approach this insane goal I’ve set?

The way a particular brain represents something is in a unique coordinate system, though it is likely related to how other brains choose to represent the same thing via some kind of linear transformation. So if you’re given a particular brain’s neural population activations.. can you reconstruct what the person is seeing/thinking/whatever? If you have some alphabet of neural population manifolds, I suppose you could compare against those, up to a linear transformation (rotation, translation)? Is there a “hash” you could take that preserves what matters so you can quickly compare against some library of manifolds? Shazam-ing a brain.

There is explicitly work (older preprint) that shows the above more or less for C. Elegans. Different worms conserve the same low dimensional manifold for motor activity, despite individual neuron activations being distinct. They call it “macroscopic universality, microscopic degeneracy” and they make a lot of analogies to statistical physics. Universality classes are definitely relevant here.

Last updated 2026-02-10