Skip to content

Psychedelics as Perturbation (Framing Note)

Status: Working Note

Scope: The epistemics of one perturbation class, read through this repository's instruments: what a chemically forced change of basis can and cannot teach a self-reconstructing system. This note is about cognition science, not practice — nothing in it is guidance, endorsement, or a claim about safety or therapeutic use; the clinical literature exists and is somebody else's subject.

Epistemic status: [HYPOTHESIZED] mapping over published mainstream neuroscience (REBUS, the entropic brain) and an older filter lineage (James, Huxley). Personal origin: the author's observation that the experience resembles the linear-algebra picture — changing one's standpoint; changing the basis.

Related files:

  • consciousness-as-global-availability.md
  • ../core/invariance-and-identity.md
  • ../computation/construction-vs-deduction.md
  • ../../lab/benchmarks/inverse-reconstruction/README.md

Failure conditions:

  • Resonance mistaken for evidence: felt profundity is a prior, not a verification. The standing rule applies especially here.
  • The basis metaphor flattering the experience: not every change of basis is a diagonalization; chemistry guarantees perturbation, not a useful representation.
  • Any drift from epistemics into practice. The moment this note reads as advice, delete the sentence.

The mechanism, in the repo's vocabulary

The current best mechanistic account — REBUS ("relaxed beliefs under psychedelics", Carhart-Harris & Friston 2019) with the entropic brain behind it (Carhart-Harris et al. 2014) — says: the brain ordinarily runs strong high-level priors that constrain perception and self-modeling; psychedelics pharmacologically reduce the precision of those priors; the system then samples states the priors normally cut away. Three translations into this repository's terms, each carrying real content:

  1. A change of basis. The everyday self-and-world model is a representation — a heavily privileged one (the brain's resting configuration is literally called the default mode network). The perturbation does not change the vector space; it changes the basis. Components that were off-diagonal in the default representation acquire visible coordinates. This is the author's linear-algebra intuition, chemically enacted — with the discipline the invariance note demands: a new basis is epistemically valuable only for what becomes legible in it, and only what survives the transformation back is a candidate for knowledge. Integration, on this reading, is invariant extraction — the sober basis is the referee.

  2. A self-intervention. The intervention experiment showed that systems at rest are generator-degenerate — opaque even to themselves; only perturbation re-excites the transients in which structure is identifiable. The default self-model is exactly such an attractor. A strong perturbation of it is the kick protocol, run on oneself: briefly, the system is off its attractor and parts of its own generator are visible that the resting state hides. (This is also why the experience resists articulation afterwards — what is glimpsed is a generator, and generators are possessed before they are exhibitable; cf. Construction vs. Deduction.)

  3. The self-reconstruction loop, transiently unbound. Ego dissolution is mechanistically legible through the consciousness node: the loop that reconstructs a self-model from the system's own traces and feeds it back temporarily stops binding. What remains is processing without the usual owner-tag.

The filter lineage — what the "antenna" intuition gets right

An old minority tradition — James's transmission lecture (1898), Bergson's filter, Huxley's reducing valve (1954) — holds that the brain receives consciousness rather than producing it. As a claim about a consciousness-field it fails the divergence test (see the equivalence-class analysis in the benchmark README: the field refuses to behave field-like under the queries that would distinguish reception from production, and survives only by retreating into unfalsifiability). But the tradition bundled two claims, and the second is good: that the brain's contribution is largely selection and reduction — the valve, not the spring. That half is REBUS avant la lettre, and it earns James and Huxley their place as anchors here.

The phenomenology also gets an honest explanation without a field: the self-model normally tags mental events as self-generated. When the binding relaxes, thoughts arrive untagged — and untagged traces feel received, not made. The antenna cosmology is the experience of broken authorship-tagging, reified. (In the vocabulary of the animism note: agent-attribution applied to one's own thoughts, displaced outward.)

What this note does and does not claim

It claims: a defensible mapping [HYPOTHESIZED] between a published mechanistic account and this repository's perturbation/invariance instruments — precise enough to generate questions (which trip-reported structures are basis-invariants? does anything survive return that controls cannot explain?) and honest enough to state its referee: verification happens in the sober basis, or not at all.

It does not claim: that the perturbation reliably yields knowledge (most of what a random rotation shows is noise wearing significance); that felt insight is insight (resonance is art's referee, not science's); anything whatsoever about practice.

Related work. Carhart-Harris & Friston (2019), REBUS and the Anarchic Brain; Carhart-Harris et al. (2014), The entropic brain; Huxley (1954), The Doors of Perception; James (1898), the Ingersoll Lecture (transmission theory). Mapping in the Related Work Map.