Psychedelics & the Filter

Of all the doorways people associate with "other realms," psychedelics are the only one you can study inside an fMRI scanner. And the science is genuinely strange.

What the Science Finds

Psilocybin (from mushrooms), LSD and DMT don't simply rev the brain up. They reduce the activity and connectivity of the Default Mode Network — the hub that maintains our ordinary, self-centred model of reality — while at the same time the brain as a whole becomes more globally connected, regions that don't normally talk doing so freely. Robin Carhart-Harris's entropic brain hypothesis captures it: psychedelics increase the entropy, the range and richness, of possible brain states. Clinically, single doses of psilocybin produce durable reductions in depression and end-of-life anxiety (Johns Hopkins, Imperial College), and DMT reliably yields "breakthrough" encounters with vivid spaces and apparent entities.

Reading the Kernel's Ledger

Long before the neuroscience, Aldous Huxley borrowed Bergson's idea that the brain is a reducing valve — a filter that throttles an overwhelming flood of reality down to the trickle useful for survival. The framework restates this precisely: ordinary consciousness is a tightly sandboxed, heavily filtered read of the pool, with the Default Mode Network acting as the gatekeeper.

Loosen that gate and you get a wider, less-filtered read — more of the ledger arriving at once. The entropic brain, in these terms, is the access permissions being relaxed: more states reachable, more of the store legible.

Dreaming vs. Tripping — the Address Anchor

This theory gives consciousness a precise kind of access: reads are content-addressed and observer-local — your sense of self is the cue that pins each read to your own records. That one idea sharply separates a dream from a trip.

Dreaming keeps the anchor intact: you replay your own stored material, self-indexed, with the internal rules relaxed — which is why a dream feels real yet runs on its own loose "physics" (see Conscious Access). The pointer never leaves your region.

Psychedelics loosen the anchor itself. The Default Mode Network — the brain's self-model — is exactly what they quieten, which is why the hallmark is ego dissolution. Once the self-anchor weakens, two very different things can follow:

1 · A confused link — mis-addressing. Cue precision breaks down and reads cross-wire: fragments that don't belong pulled together, senses bleeding into one another, novelty everywhere. Still a read of your own data, just scrambled.

2 · A relocated read — another region. No longer pinned to "you," the pointer can drift toward parts of the address space not indexed to your instance — the model's version of "other realms," the entities, arriving somewhere with its own rules.

Both differ from dreaming for the same reason: the anchor is compromised, not intact. Dreaming stays home; a trip either mis-reads home, or leaves it.

Why Relocation Deserves to Be Taken Seriously

It would be easy to wave the second option away as "just hallucination." The framework doesn't let us off that lightly — because relocation is what it actually predicts. If reads are pinned to the self, then dissolving the self doesn't merely add noise; it releases the pointer. And the substrate that pointer could drift into is already non-local — the plumbing for a read that reaches beyond your own region is built into the architecture. On this model, a fully dissolved anchor reaching another part of the ledger isn't an exotic add-on; it's the default expectation.

The phenomenology pushes the same way. This is the old filter theory of consciousness (Bergson, Huxley): the brain may restrict a far larger signal down to a survival-sized trickle — in which case loosening it reveals more rather than inventing more. And what's revealed doesn't behave like a dream. Three features keep the relocation reading alive where "internal noise" strains to explain them:

It feels "realer than real." Users overwhelmingly report the space as more solid, detailed and lucid than waking life — the opposite of a degraded or scrambled signal.

The entities act autonomously. Unlike dream figures, the beings encountered are experienced as independent — they surprise, refuse, instruct, and persist against the user's will. Autonomy is exactly what reading something you didn't author would feel like.

Information feels received, not generated. People describe being shown or taught — meeting intelligences that seem to know more than they do. If that ever yields something the person could not have known, the internal story breaks.

None of this is proof. But it means relocation is not the fringe option to be talked down — on this model it is the leading reading of what a dissolved anchor does, and the question is genuinely open.

"But Everyone Sees the Same Code" — the Weakest Argument

Ironically, the point raised most often — that trippers worldwide see the same lattices, grids and "code" — is the weakest support for relocation, and it's worth conceding cleanly. These recurring shapes are Klüver's form constants, and you get the same geometry from a migraine aura, a flickering or laser light, pressure on a closed eye, even falling asleep. Mathematical models (Ermentrout–Cowan; Bressloff–Cowan) derive them from the wiring of the visual cortex. So the shared geometry is hardware — everyone reading through the same eyes and cortex — not proof of a shared external place. (A 2025 pilot study reporting consistent "code-like" descriptions of a diffracted laser shows exactly this: recurring shapes, no decoded information.)

But notice what that retires: only the geometry. It says nothing about the autonomy of the entities, the realer-than-real clarity, or the sense of received information — and those, not the grids, are what actually point toward a relocated read. Conceding the wallpaper doesn't empty the room.

Entities as System Processes

If relocation is real, the next question is what you would actually meet there — and the model gives an unexpected answer. Not other souls, not aliens, but the system's own background processes. In computing these are daemons: programs that simply run — maintenance, cleanup, routing, access-control. They have a function, not a personality — which is exactly how the entities are described.

The busy workers — "machine elves" forever building and operating — read as processes at work. The gatekeepers — guardians who grant or block passage, who "were expecting you" — read as access-control at the kernel boundary: you escalated, and the guard checked you. And the being met at death (see Birth, Death & the Record) reads as the cleanup-and-handoff process that reaps a terminating instance — with the life review as the integrity pass it runs over your record on the way out.

And here is why they seem to know everything. A gatekeeper runs at the ledger's level, not in our sandbox, so it has read-access to the whole record — where we only ever see our own slice. To a user-space visitor, a process with full ledger access would appear, quite simply, omniscient: ancient, all-knowing, able to read you completely, holding every fact there is. That is the council, the librarian, the being that "knew my entire life in an instant" — not a god, but a process with a higher read-permission than ours. It is almost exactly the old idea of the Akashic records and their keepers, restated as access control.

The honest snag: these states reliably manufacture a feeling of boundless knowing — the noetic quality — and the profound "downloads" people bring back tend to read as vague or empty once written down. So an all-knowing gatekeeper may be the state generating the impression of omniscience (with the social brain dressing it as a being) rather than a real read of the ledger.