Your brain is spread across two hemispheres distributed through billions of neurons yet your experience is unified. You don't hear with one consciousness and see with another. Something binds it together. What? You taste coffee and the bitterness and the warmth and the steam and the weight of it in your hand arrive as one thing, inseparable, a single experience you have never thought to question. Now take the warmth away from the bitterness. Put the steam in one room and the weight in another. You haven't lost any of the pieces. You have lost the coffee.
The neuroscience is clear about what processes what. Color in V4. Motion in V5. Shape in the inferotemporal cortex. Sound in auditory cortex. Touch in somatosensory cortex. Each area fires when its specialty arrives. None of them fires for coffee. There is no coffee neuron, no coffee region, no place in the brain where bitterness and warmth and weight converge into a single representation. The pieces are processed in different places, at slightly different times, by neurons that have no direct knowledge of each other's outputs. And yet the coffee is one thing. That is the binding problem.
Criticality gets partway there. At the critical point a perturbation anywhere can cascade everywhere, and NFT leans on this hard. But binding requires more than propagation. If everything reaches everything with no structure, you get noise, not coffee. The signal has to arrive organized. Topology is what organizes it.
Think of three friends who all know each other. The three friendships form a triangle, a structure with an interior, a space enclosed by the connections. Four mutual friends form a tetrahedron. Five form a shape that exists in four dimensions. The Blue Brain Project, applying algebraic topology to cortical microcircuits, discovered that neurons form these mutual-connection structures, called simplicial complexes, reaching up to seven algebraic dimensions. Neural activity generates high-dimensional topological cavities that appear, grow, and dissolve in coordinated sequences. These are not metaphorical dimensions. They are algebraically real topological features of neural connectivity.
Santoro and colleagues (2024) showed that interactions among three or more brain regions predict what a person is doing better than pairwise connections can. The effect shows up in restricted subnetworks, where higher-order structure adds real predictive power, even though whole-brain metrics wash it out. Binding through topology is structured and multi-scale rather than uniform.
Start at the bottom. Inside a single neuron, radical pair spin coherence creates correlations between tubulin subunits. These are tiny, local, and fast. Microsecond events in a protein lattice. By themselves they bind nothing.
Now zoom out. That neuron belongs to a clique, a group of neurons that all connect to each other. The clique forms a simplex, a higher-dimensional shape with an interior. Two neurons ten centimeters apart in your skull may be neighbors in the seven-dimensional simplicial complex formed by their shared clique membership. The back-action from a radical pair event in one neuron does not need to cross ten centimeters of brain tissue. It reaches the other neuron through algebraic proximity, through the interior of a shared shape. What looks like long-range correlation in three dimensions is local interaction in seven.
Now zoom out further. At the critical point, correlation length diverges. A perturbation in one simplicial complex can cascade into the next, and the next, and the cascade spans the whole network. Criticality is what connects the quantum scale at the bottom to the system-spanning dynamics at the top. The radical pair event is local. The simplex makes it reach. Criticality makes it matter.
That is how the coffee becomes one experience. The bitterness in gustatory cortex and the warmth in somatosensory cortex and the weight in motor cortex are processed separately in three-dimensional space. In the higher-dimensional topology of their shared simplicial structure, they are neighbors. They share an interior. That is the binding.
NFT predicted that topological complexity would decline before classical neural activity metrics during propofol-induced anesthesia. If quantum substrate coupling underlies topological binding, then disrupting consciousness should reduce topological complexity first and classical metrics second, because the substrate is upstream of the classical dynamics.
We tested this using the OpenNeuro DS005620 dataset (propofol sedation in 21 subjects), computing persistent homology (tracking how topological shapes appear and disappear over time) alongside classical metrics (Lempel-Ziv complexity and weighted phase lag index in the alpha band) across the transition from wakefulness to sedation.
The prediction was not supported. Classical metrics declined first in the majority of subjects. The proportion showing topology-first decline was 0.12 for both metrics, with p-values of 0.96 and 0.998 respectively. The data show the opposite of what was predicted.
What does this mean?
There are several possibilities, and it would be dishonest to reach for the most convenient one.
First, the prediction may simply be wrong. The binding mechanism may not operate as proposed, or topological complexity and quantum substrate coupling may not have the temporal relationship NFT assumed.
Second, EEG-derived persistent homology may not capture the relevant topology. The simplicial complexes identified by the Blue Brain Project operate at the level of cortical microcircuits, spatial scales below what EEG can resolve. The topological features we measured may reflect a different level of organization than the ones NFT claims are relevant.
Third, propofol may disrupt classical and quantum processes simultaneously rather than sequentially, in which case the temporal ordering prediction was based on a false assumption about the mechanism of action.
The simple temporal ordering prediction failed. We don't know which of these explanations is correct. The failure weakens the topological disruption order as a discriminative prediction for NFT, but it falsifies the prediction, not the binding hypothesis. The hypothesis must be refined or the prediction replaced with one that accounts for the data.
We report this because the alternative is public relations. The full TDA pipeline, dataset details, persistent homology computation, and temporal ordering statistics are reported in [Malloy, "Topological Disruption Order Under Propofol Sedation: A Negative Result," forthcoming].
Where we are. Three chapters of consequences, each further out on the limb. The quantum Zeno effect may be the mechanism of perceptual stability. Qualia may be the physical back-action of measurement on the measuring apparatus, and the hard problem may evaporate rather than relocate if the back-action loop and experience are the same thing. The binding problem may find its answer in higher-dimensional topology, though one topological prediction has already failed and that failure is on the ledger. Now the hardest question. If consciousness navigates, what is it navigating through?