The Poincaré conjecture states it simply: every closed three-dimensional manifold homotopically equivalent to a sphere is a sphere. For a hundred years nobody could prove it. In 2003 Grigori Perelman posted three preprints to arXiv and closed the problem. Turned down the Fields Medal. Turned down the million-dollar Clay prize. Moved into a communal flat in Kupchino. Worth remembering next time someone explains who sounds esoteric in this story.
A few uncomfortable observations.
A coherent thought behaves like a closed manifold
When a person thinks about something in sequence — builds an argument, arrives at a conclusion, returns to the beginning — their neural activity leaves a trace through state space. That trace can be recorded. In fact it is being recorded: in fMRI, in EEG, in neural embeddings.
For the last decade a discipline has been gaining ground in computational neuroscience called TDA — topological data analysis. The central idea is simple: instead of looking at brain activity as a vector in Euclidean space, you look at its shape. Persistent homology. Connectivity, holes, loops.
A strange regularity surfaces. Coherent thought — activity that closes into a loop. Anxious, scattered thought — activity that doesn't close; ends don't meet ends. Meditative states — fewer holes than the default waking mode.
The Poincaré conjecture has no direct role here. What does have a role is Poincaré's language, which turns out to fit what we observe, and that is already something.
What it means
Experiments worth knowing:
— Giusti et al. (2015, PNAS): topological analysis of hippocampal activity shows the brain represents space not as a map but as a complex of simplicial cocycles. Geometry emerges from topology, not the other way round.
— Saggar et al. (2018, Nature Communications): topological dynamics snapshots (Mapper algorithm) show that states of consciousness form distinguishable attractors in neural activity space. Meditators have a different topology from novices.
— Northoff and Huang (2017): "temporo-spatial theory of consciousness". Consciousness as a temporal structure with characteristic dynamics, rather than a point.
These data are not obliged to prove "consciousness is a topological manifold". They prove something different: topology is a fitting language for describing how consciousness works.
The distinction matters. Ontology answers "what is". Methodology answers "how to describe". We are on the second one so far.
Why look at this at all
Computational neuroscience today is roughly where physics was in the late 19th century: lots of data, no general frame. Everyone knows neurons fire. Everyone knows patterns are coherent. But what is coherent and how is described at a level Einstein would have called "butterfly collecting".
Topology is an attempt to give the general frame a mathematical skeleton. If coherent thoughts are closed loops in state space, if awareness has a specific persistent homology, if déjà vu is the intersection of current activity with a previously recorded trajectory — thinking becomes an object you can formally define.
"Math explains consciousness" overstates what we have. The cautious version: math offers a tool with which consciousness may at last start to be measurable.
The honest edge
Perelman, proving Poincaré, used the Ricci flow technique Richard Hamilton had been developing since the 1980s. It describes how the metric on a manifold changes under curvature.
There are papers trying to apply Ricci flow to neural connectivity graphs. Fashionable. Beautiful. Premature. A mathematical analogy becomes a theory only when it makes predictions you can test.
Topology of mind has so far produced language, metrics, hypotheses. Not a single prohibition — not a single "if this holds, that cannot be observed". Until there are prohibitions, the hypothesis is not yet science.
We are in the phase where language runs ahead of theory. A normal state of a field before a breakthrough — or before finding out there won't be one. The second is also a useful result.