The Window

A theory of this scope gains nothing by predicting what a simpler theory already covers. The whole question is whether it says one specific thing the deflationary story cannot, and whether that one thing is true.

The first essay made a prediction, and the technical companion built the smallest model that makes it precise. This third piece has a narrower job: to say what would settle it. It does not defend the picture. It hands a skeptic the experiment that could take it apart, and admits the one result that would leave the theory intact but redundant.

The discipline here is a single distinction. The deflationary account (call it the entropic-brain view) already says that profound experiences are the brain becoming more globally connected and less predictable: boundaries down, signal diversity up, oneness reported. That account is well supported. Neural complexity, measured as the diversity of activity patterns in the signal, rises with the richness of the psychedelic state and falls toward the floor in anaesthesia and deep sleep, where no subject remains. If The Weave and the Window predicted only that, it would be a metaphysical decoration on settled neuroscience.

The question this companion exists to answer is therefore not whether the account fits the data (most things fit the data) but whether it says one specific thing the deflationary story cannot, and whether that thing is true.

Part One

The cheap prediction and the expensive one.

Recall the structure. Lucid unity requires two things in the brain at once. The boundary must dissolve: the self-maintaining networks (the default mode chief among them) must de-segregate, their walls thinning until the brain is globally connected, its partitions gone. And a subject must remain: the system has to stay integrated and complex enough that someone is still present to undergo the experience. Set on two axes, these give four cells: subject present and boundary intact, ordinary waking; subject present and boundary dissolved, the experience itself; subject absent and boundary intact, a rare borderline; subject absent and boundary gone, dreamless sleep and deep anaesthesia, no one to feel anything. Intense lucid unity lives in exactly one cell.

That is the cheap prediction, and it should be conceded as cheap. The conjunction (boundary down and subject preserved) is something the entropic-brain view delivers for free. Increased global connectivity with preserved complexity is just what “a rich state, not a blackout” means in that framework too. If the programme stopped here, it would have merely repackaged a known result in unfamiliar vocabulary.

The expensive prediction is different in kind. It is not about either axis alone. It is about the shape of the frontier between them: the boundary in the two-dimensional space that separates “lucid unity” from “oblivion.” The toy model in the companion makes that frontier bend. Because connection is finite and cannot be everywhere at once, opening a system’s edge draws integration out of its interior. The deeper the boundary dissolves, the more it costs the interior that keeps a subject present. Pushed far enough, the interior can no longer hold a subject, and consciousness fails. The claim is that how far the boundary can be opened before that happens should depend on how much interior integration was present to begin with. A more strongly integrated mind should tolerate deeper de-segregation before unity collapses.

In the geometry of the four cells, this means the line between the “unity” cell and the “oblivion” cell is not vertical. It tilts. With more interior integration the line moves outward, allowing more boundary-dissolution before collapse; with less, it moves inward, and collapse comes sooner.

State it as a number and the disagreement becomes testable. The claim is an interaction term. Regress the survival of lucid unity on boundary-dissolution depth and on baseline interior integration, and if this account is right, the coefficient on their product is positive and non-zero. A simple, one-axis entropic-brain account has no reason to predict a non-zero coefficient: on that minimal boundary-down/entropy-up view there is a fixed threshold of dissolution beyond which everyone, strong or weak, tips into oblivion at the same depth. (A richer neurobiological model could in principle reproduce an interaction; the claim is only that the minimal account does not predict one.) One coefficient, and its sign separates this theory from that minimal account. Everything else in this companion serves to measure it.

Rivals

What else could explain the window.

The entropic-brain view is not the only account on the field, and a theory looks less isolated, and more scientifically serious, once it has stated how it sits beside its neighbours. The fair test is not whether a rival predicts that boundaries dissolve — most do — but whether it predicts the same five things together. Put to each account, the questions are: does it predict boundary dissolution; does it predict a subject preserved (high complexity, not a blackout); does it predict the moderation interaction, that baseline integration shifts the collapse frontier outward; does it predict plenitude rather than cessation at the floor; and does it predict route-invariant structure, the same boundary–unity function however the state was induced? The third question is the discriminating one. Most rivals answer the first two yes and the third no.

The entropic brain

The closest rival and the main foil throughout. It predicts boundary dissolution and preserved complexity directly, and is route-agnostic, so it shares the first, second, and fifth answers. What it does not predict, on its minimal one-axis form, is the moderation interaction: a fixed dissolution threshold should claim strong and weak minds alike. That single gap is the whole experiment.

REBUS (relaxed beliefs under uncertainty)

The leading predictive-processing model of psychedelics: high-level priors are relaxed, loosening the self-model. It predicts boundary dissolution elegantly and is naturally route-flexible through any 5-HT2A-like relaxation of precision. But it is a theory of boundary collapse, not of the interior strength that must survive; it offers no principled reason that baseline whole-brain integration should buy more headroom before the subject fails, and nothing about plenitude versus cessation at the floor. It predicts the cheap conjunction, not the expensive interaction.

Predictive processing / precision-weighting (general)

The broader family behind REBUS: experience as precision-weighted inference, dissolution as a drop in the precision of self-related priors. Same verdict, and for the same reason — it speaks to how the boundary opens, not to a second, interior axis that moderates how far it can open. An interaction is not forbidden here, but it is not entailed either; it would have to be added by hand.

Global neuronal workspace

Consciousness as global broadcast across a frontoparietal workspace. This account is strong precisely on the subject axis — it has a natural story for when a subject is present versus absent — but weak on unity as such: it does not obviously predict that boundary dissolution with the workspace intact yields communion rather than ordinary rich consciousness. It could perhaps be bent toward the interaction, but it does not reach for it, and says nothing about the floor.

Thalamocortical arousal models

Unity and its loss as a function of arousal and thalamocortical gating. These models excel at the cessation end — explaining the blackout of deep anaesthesia and sleep — but treat lucid unity and oblivion as points on a single arousal axis, which is exactly the one-dimensional picture the interaction is designed to break. No second axis, no moderation, no plenitude claim.

The softer accounts

Absorption, suggestibility, and personality models predict who dissolves, not the brain-state structure of the dissolution, and so are orthogonal to the interaction rather than rivals to it. Meditation-specific attentional training and route-specific neurochemistry both predict differences between routes — the opposite of the route-invariance this account stakes out, which makes them sharp tests rather than competitors: if conditioning on de-segregation fails to screen off the route, this theory loses and they gain.

The pattern is the point. Every serious account predicts that boundaries fall and, in its better forms, that a subject can remain. Almost none predicts that a third quantity — baseline interior integration — should bend the frontier between them. That is the narrow ground this theory stands on, and the reason the interaction, not the dissolution, is the thing worth measuring.

Part Two

Measuring the two axes.

Measuring an interaction requires first measuring each axis cleanly, and here the difficulties begin.

The boundary axis is the more tractable. De-segregation of the default mode network and the collapse of ordinary network modularity are among the best-replicated correlates of ego-dissolution; reduced integrity within the default network, and the decoupling of its hubs, track self-reported dissolution across psilocybin and LSD studies. So “how far has the boundary opened” has a defensible operationalization: within-network integrity falls, between-network connectivity rises, and the brain’s modular structure flattens out.

The subject axis is subtler and more important, because the theory lives or dies on telling “lucid unity” apart from “nobody home,” and those can look similar from the outside: both are quiet, both are egoless, both resist report. Here the right instrument already exists and was built for exactly this discrimination: the perturbational complexity index, the measure validated against anaesthesia and disorders of consciousness precisely to answer “is a subject present” when behaviour cannot. It is the closest thing the field has to a needle that reads subject-presence directly, where every other measure has to infer it from what someone can say. If any measure can mark the moment the interior fails, it is this one.

The most important point in this piece is a warning. Not every measure of “complexity” is the integration the theory means. The theory’s “interior integration” is the whole-system capacity that keeps a subject bound into one point of view. It is not the rigidity of any single control network. This matters because the one existing finding that relates a baseline brain property to depth of dissolution points the awkward way: individuals with lower diversity of executive-network nodes were found to be more likely to dissolve under psilocybin. Read carelessly, that is the opposite of “more integration buys more headroom.” The theory’s only escape is that executive-network node diversity is a regional rigidity metric, not whole-system integration, and that a stiff executive network is plausibly part of the boundary machinery that has to relax, not the interior strength that has to survive. That escape may be right. But it is also exactly the kind of after-the-fact distinction that lets a theory avoid any test. So the program must close that door in advance: the interior-integration measure has to be named, justified, and registered before the data are seen, with the prediction stated in its terms and no other. The natural candidate is a whole-brain measure of signal complexity taken at rest (baseline perturbational complexity, or Lempel–Ziv diversity across the cortex), a whole-system quantity rather than a regional one. Naming it now is what turns this worry into a commitment rather than a loophole. Choosing the metric after looking would prove nothing but the analyst’s flexibility.

Part Three

The design.

The experiment is a within-subject, multi-route study, and it has to be within-subject because the prediction is about how an individual’s own frontier sits: comparing a person to themselves at different depths, not strong people to weak people across the noise of a population.

Each participant is characterized at baseline by the registered interior-integration measure, taken in ordinary rested waking. Then dissolution is induced and deepened along a graded path, with boundary-dissolution and subject-presence sampled continuously as depth increases. Crucially, the path must approach the individual’s own collapse (the depth at which lucid unity gives way to oblivion), because the prediction is about where that collapse falls, and a frontier that is never approached cannot be located. Several induction routes should be used. No route is privileged; the claim is about a pattern that should hold regardless of how it was reached: high-dose serotonergic psychedelics, where breakthrough past a threshold brings the dissolution of space and time; deep absorptive meditation, where the jhāna states show rising signal diversity and a drift toward critical dynamics; sensory deprivation; and the controlled descent of anaesthetic induction, which has the unique virtue of carrying a subject all the way to the floor under measurement. If the same frontier-shape appears across routes, it is a property of the architecture and not of any drug.

The analysis is the moderation test, stated in advance. Does baseline interior integration moderate the depth of de-segregation a subject can sustain before subject-presence falls below the conscious threshold while unity is still rising? If this is right, the interaction is positive: higher baseline integration, deeper tolerable dissolution before collapse. A simple one-axis entropic-brain account expects none: a fixed collapse depth, baseline-independent — though a richer model could potentially reproduce one. Both nulls and the alternative are registered before collection, along with the single integration metric and the single subject-presence metric. The result that matters is the sign of one coefficient.

This is a demanding design. It asks to bring people toward their personal threshold of oblivion and measure them right at the edge, where reports thin out precisely because the reporter is fading. Anaesthetic induction makes this ethically approachable because the descent is routine and reversible and already monitored. The psychedelic and meditative routes approach the edge far more gently and may never quite reach it. It is entirely possible that the cleanest version of this experiment (titration to individual blackout with simultaneous report) is one that can only be run in the anaesthetic corner, and that the other routes can only populate the approach, not the edge. That is a real limit, and it means the decisive coefficient may be estimable only at one corner of the space. The program should state that up front rather than promise a clean curve it cannot ethically draw.

A second axis the within-subject design does not address is other species. The interaction the theory predicts (more baseline integration tolerating deeper dissolution) needs a wide spread of baseline integration to reveal itself, and within humans that spread is narrow; across species it is vast. If the same frontier-shape holds when integration is varied phylogenetically rather than between individuals, that is converging evidence the deflationary account has no reason to expect. The limit is the one that haunts the floor: animals cannot report, so comparative work can populate the structural axes but not the felt discrimination between lucid unity and no subject at all. It broadens the external evidence without reaching the felt question.

Part Four

The floor.

There is a second claim here, and it is best held as a wager rather than as a co-equal prediction — because, unlike the window interaction, it may not be experimentally decisive at all. The wager is plenitude: that at the very floor of the deepest dissolution there remains a positive, contentful sense of being-one — fullness, presence, communion with something, not mere cessation. A no-ground view claims the opposite limit: that the floor is empty, the constructed self simply dropping away with nothing underneath to commune with. This is the live disagreement with Buddhism, and it is contentful, not cosmetic. The two pictures say different things about what is there when everything else is gone. But the reason to demote it from prediction to wager is structural, and the next paragraphs are about that: the floor may be unreportable in principle, so this claim cannot be made to carry the weight the interaction can.

So the sub-study asks: at the deepest reportable dissolution, is the character of experience contentful presence or contentless cessation? Here a structural bind appears that no instrumentation dissolves. The further the subject dissolves toward the floor, the less subject remains to register anything, including fullness. The very state that would confirm plenitude is the state least able to report it. As the subject goes to zero, the report goes to zero with it, and the floor may lie permanently just past the last point from which anyone can speak.

There are partial mitigations, and they are only partial. One can sample as near the threshold as report survives and ask whether the trajectory is toward fullness or toward emptiness: whether what is approaching as the subject thins reads as plenum or as void. One can use the fine-grained methods of micro-phenomenology, trained retrospection that recovers texture ordinary questionnaires miss, on the deepest states people can return from. One can compare the near-floor signatures of routes that participants describe as “full” against those they describe as “empty” and ask whether the brain marks a difference. But none of these reaches the floor itself, and this sub-study may simply be undecidable in principle: the plenitude-versus-cessation question could be one where the structure of the phenomenon prevents the measurement.

Yet the undecidability is itself informative. If careful work shows that the approach to the floor is systematically toward fullness, with the void reports clustering instead with the failure of the lucid state rather than its deepening, that is real evidence even if the floor stays out of reach. We will likely never stand at the floor itself. But if everything we can measure leans toward fullness, that still counts for something.

Part Five

What each result would mean.

A program should say in advance what each outcome forces. So, plainly:

If the interaction is positive and the approach to the floor runs toward plenitude, the theory has done something a metaphysical ornament cannot: it has predicted a specific relationship the deflationary account denies, and it has survived. That would not prove consciousness is fundamental. It would show that the architecture it describes is doing real predictive work.

If the interaction is zero (if strong and weak minds tip into oblivion at the very same depth of dissolution), the theory stays true but turns redundant. The phenomenology would still be real, the four cells still correct, but everything distinctive would collapse back into the entropic-brain account, and Occam’s razor would remove the extra structure. This is the outcome the view most fears and most needs to keep in view: failure by superfluity rather than by contradiction, and the more likely of the two failures.

If the interaction holds but the floor runs toward cessation rather than plenitude, then the discriminating physics survives while the metaphysical heart does not. The positive-ground commitment (the claim that there is one awareness underneath rather than dependent arising all the way down) would fail, and the view would have to concede the central point to the non-dual traditions it wanted to correct, keeping only its account of the parsing.

One caveat applies to all three. Even the cleanest win is a correlational, structural result. It would show that the right relationships hold among boundary, integration, subject-presence, and reported unity. It would not be a derivation of which something the state is like from its connection-structure: the rule that takes a physical state and outputs its felt character. That rule does not exist, for this theory or any other, and it sits exactly under the central assumption. A positive interaction would make the architecture credible. It would not close the hard problem, and no experiment in this companion pretends to. And there is a blunter limit alongside it: because the interaction is derived at the substrate scale and reaches the brain only by analogy, a null result could indict that bridge rather than the theory itself. The test is sharp, but not perfectly clean, and a program that means its falsifiability should say so.

Coda

What the companion delivers.

The first essay claimed only to locate the mysteries, not to remove them: to mark which claims are physics and which are philosophy, and still stake out a prediction that could be wrong. This companion is what that costs in practice. It converts a sweeping picture into a single coefficient and one hard phenomenological question, names the corner where each may be unmeasurable, and states in advance the result that would make the whole structure redundant.

This is less than a theory of everything, and more than most such theories offer: a defined point at which the account can be tested, together with a clear statement of where that test may not reach.

References