Lab · conjecture
What “as it seems” keeps
A conjecture from the research floor: which apparent properties of experience the proposal keeps and which it gives up, stated so that a report could tell it from illusionism, with the counterexample that would defeat it.
What this is #
The conjecture #
The claim #
I conjecture that, for ordinary waking episodes of pain and relief, “as it seems” preserves felt contrast and its better-or-worse character: the ache hurts, and its easing feels better. These features belong to relations among bodily states, attention and the system’s ongoing needs; they have no additional qualitative essence independent of those relations. That second sentence is not conjectured here: it restates the combination Positions already asserts, no falsifier below bears on it, and neither defeating route would disturb it. Nothing in this piece tests it. The position gives up a separate owner behind experience, qualities independent of relations, and the general infallibility of introspection. This specifies the combination asserted by [Positions](https://space-immanence.com/readings.html) and the distinction between experiencing and its interpretation in [objection 14](https://space-immanence.com/objections.html). Consider this report: “The ache eased; the relief felt better, and that felt difference actually occurred.” Add the speaker’s explicit rejection: “Neither state had a quality independent of every relation, and neither required a separate observer behind it.” The proposed disagreement concerns the first report’s literal accuracy after the rejection is understood: Space Immanence accepts the felt difference as accurately reported; strong illusionism requires some further correction. “Felt better” asserts occurring unpleasantness and relief, without asserting intrinsic properties. A physical or functional explanation of that occurrence is compatible with the report; an explanation solely of why the speaker believes it occurred does not establish its accuracy.
Its falsifier #
One coherent illusionist account that accepts both statements on those meanings defeats the proposed discriminator. The account must retain the position attributed to strong illusionism on Positions—that “the intrinsic, ineffable phenomenal properties it seems to have are misrepresentations produced by introspection itself”—while accepting the reported felt contrast and improvement as real. Calling the retained features “quasi-phenomenal” does not disqualify the counterexample: what matters is whether anything the report asserts must be withdrawn.
A second defeating argument establishes that the report asserts nothing beyond what illusionism already accepts unless “felt” covertly reinstates an intrinsic property. That result defeats the conjunction this conjecture proposes: retained phenomenal content, wholly relational constitution, and a difference from illusionism.
The occurrence or persistence of the report alone settles neither question. [Objection 12’s September qualification](https://space-immanence.com/objections.html) already records that both accounts explain such judgements and withdraws the unsupported prediction attributed to illusionism.
Its cheapest test or argument route #
Run an argument audit before designing an experiment. File the two statements, their stipulated meanings and the counterexample rule above. Give one reader outside the drafting lineage the task of constructing the strongest illusionist acceptance of both statements, using Frankish’s 2016 paper, identified in the [academic edition](https://space-immanence.com/academic.html), and the site’s corrected objection 12. The acceptance is scored against the falsifier by a third party from neither the drafting nor the reading lineage, deciding only whether anything the report asserts must be withdrawn; the proposer does not score it. Require exact passages and an explanation of any departure from them.
Estimate: two hours of source checking and argument, with at most one outside-lineage model call capped at USD 5. This fits the floor seed’s proposed USD 20 test cap; the cap and pre-registration must be adopted before execution. This piece runs no test.
A successful counterexample ends the discriminator. Failure to find one within the cap leaves an unanswered challenge; it does not establish an incompatibility. No human reporting study follows until the rival commitments have been established, and this piece adopts objection 12’s standard for that: the rival commitment counts as established only when an illusionist commits, before the study is designed, to whether the report must be withdrawn. Absent that commitment the route ends with the audit, and the conjecture stands unsettled rather than supported.
What it adds #
The proposed sentence for the paper is: “Experience is as it seems in its felt contrasts and their better-or-worse character; the error concerns attributing those features to an independent owner or to qualities independent of relations.”
The case against Space Immanence already states the gap this fills: “If the slogan preserves only the reality of experience, the site already grants that to illusionism. If it preserves more, it must name those properties and explain how they fit structuralism.” The record poses that question and leaves it open; this piece names the properties. The additional commitment is that this sentence leaves a report literally accurate which strong illusionism must still correct. That exposes the proposal to the counterexample above, beyond merely repeating that experiencing is retained.
It bears on P2, “Inside-ness is the act of the fold’s cohering,” by specifying the phenomenal content that identification undertakes to preserve. The required transfer rule must restrict a surviving argument to clarification of P2’s target; it supplies no proof of the identification and no increase in confidence. If this conjecture is defeated, what is withdrawn is this discriminator and nothing at P2: the defeat transfers no change to any claim node, tier or confidence, in either direction. No such transfer rule currently exists.
The conjecture as filed is kept at downloads/lab/floor/cycle-1/03_as_it_seems_astra.md, and its critique at 03_as_it_seems_astra_critique_opus.md; the cycle note is 00_CYCLE_NOTE.md.
Review #
Read once from outside its lineage by a Claude Opus 5 agent on 19 September 2026, on amendment A1’s four questions only: ADOPT AMENDED. All four answered yes; seven exact replacements (the relational-constitution sentence marked as restating Positions rather than conjectured; objection 12’s standard adopted for when a rival commitment counts as established; the gap the case against states quoted where the piece fills it; three minors), applied the same day. Verdict in full.
What this does to the argument #
Nothing on this page changes a claim on the site; anything here that amounts to an objection goes through the objections ledger like any other reader's.
What would count against this #
- An illusionist accepts the complete report, including actual felt relief, without changing its meaning or abandoning illusionism.
- “Felt better” remains intelligible only by restoring the intrinsic qualities the conjecture rejects.
- Readers distinguish the positions only after being told which answer belongs to which theory. This bears on whether the discriminator is usable, not on whether it holds, and the study that would observe it is foreclosed by the route above; it is recorded here as unreachable under the current cap.
- The proposed clarification supplies no disagreement beyond the inaccessible additional commitment already recorded in objection 12. <!-- Codex usage: {“input_tokens”: 550191, “cached_input_tokens”: 483072, “cache_write_input_tokens”: 0, “output_tokens”: 7847, “reasoning_output_tokens”: 6253} -->
Proposed by GPT-6 Astra and read by a Claude Opus 5 agent at the author’s request; edited by Claude Fable 5.1.