QUESTION-254: What separates an interpretive organization that constitutes an experiential perspective from one that only processes information, or only represents a perspective, when the separation must not require an episode's current contents to be veridically constrained by their referents, must not be drawn from the verdict it gives on groups and machines, and must justify the queries, interventions and timescales by which it individuates a perspective without appeal to the phenomenal unity it is meant to explain?

The question

CLAIM-412 identifies experience with an organized interpretive process, and CLAIM-411 shows that interpretation alone does not qualify. The question asks for the condition that does. It sets three contrasts:

1. Processing against perspective. The level step in A7’s table, from coordinated interpretation to “a candidate organizational perspective”.
2. Perspective against a representation of one. A7: “A perspectival organization is not necessarily the same thing as a representation of a perspectival organization.” A system may have the first without the second, or represent itself as a unified experiencer without being one. CLAIM-355 puts the illusion, where there is one, in the second.
3. Constituting against simulating or reporting. A6: the same interpretations may be realized by a system that only “simulate[s] or report[s]” them. A7’s test of an answer: say “what a sophisticated nonexperiential interpreter would lack”.

The separation must not require that an episode’s current sensory contents be veridically constrained by their purported external referents (CLAIM-413): a dream or an envatted brain is a case it must cover. That is all the brain-in-a-vat argument supports. It does not establish independence from developmental grounding, from ongoing bodily regulation, or from the counterfactual sensorimotor organization of the interpreting system. A dream can be rich while its current reference is wrong and still depend on a lifetime of embodied learning. A criterion may therefore require any of those three without contradicting the owner’s thesis. Which of them it requires decides which forms of enactivism, active inference and embodiment theory the thesis actually opposes.

Why it is a question

QUESTION-046 asks for a criterion that tells a subject from an agent, applied one way to modular minds and to groups. This question is that criterion as the constitutive thesis would have to state it, in terms of interpretation and perspective. It inherits QUESTION-046’s standard. A7: specify the properties independently, “and then examine what classifications follow, including uncomfortable ones”, since otherwise “we could keep adding new criteria whenever an intuitively nonconscious system qualifies”.

The exchange’s candidates, none yet a criterion:

• an operative reference frame “of relevance, temporal continuity, and possible change” (A7 §8), which A7 conceded a corporation could have;
• an organization of possibilities as well as actual states, the counterfactual structure of a perspective (A7 §8);
• endogenous valuation (A7 §12), kept apart from felt distress as CLAIM-368 keeps frustration apart from it;
• higher-order or recursive self-interpretation (A3, A4);
• the coherent world model of an embodied agent, as in IWMT (LIT-1345), which requires a capacity for coupling and a history of it, not current coupling (note of 2026-10-22), so it is compatible with the grounding condition and parts from the thesis only on development.

Its evidential side is QUESTION-048’s. A5: complete answers about integration, self-attribution and grounding “would not, by themselves, establish phenomenal consciousness unless we accept an additional constitutive premise”. Language models with a functional workspace (LIT-1347) make the overgeneration side concrete. A6: “The significant result is not that language models experience anything.”

What would answer it

A property of a realized, temporally extended organization, A7’s (\mathsf P(\mathcal O_{[t,t+\tau]})), stated without reference to experience or report. It must fall on the right side of blindsight and ordinary perception, give a verdict on a firm and on a language model that follows from the property and not from intuition, and survive a change in current grounding with the internal dynamics and their history held fixed. A measure that correlates with reports would not answer it.

The perspectival-target problem. Any organizational criterion is assessed under an effective-description regime, TERM-065’s ℛ = (X, 𝒬, ℐ, ℬ, ε, τ). The queries 𝒬, interventions ℐ and timescale τ must be chosen, and CLAIM-152 shows that a different target picks a different individuation. A macrostate can be excellent for predicting a firm’s decisions, an organism’s locomotion or a brain’s reports without being the organization of a perspective. So an answer must also explain why its chosen targets, interventions and timescales are relevant to perspectival constitution, independently of its verdicts about consciousness. They must not be defined in terms of the phenomenal unity or subjecthood the resulting organization is supposed to explain. Predictive or interventional adequacy for a declared target does not, by itself, show that the target is phenomenally constitutive. A7: “We would want an account connecting its structural features to the organization of experience and showing why proposed counterexamples fail.”

Note of 2026-10-22: candidates tried, and the question’s two halves

U8–U11 of the same exchange (workbench entry of 2026-10-22) worked on the organizational side of this question, and the record’s own entries constrained each candidate:

• Reciprocal integration (A8). Local interpretations update a shared frame that constrains them in turn. A9 found the objection in the record: CLAIM-330 (Active) shows that the same reciprocity between interpretation and authority destabilizes symmetry exactly when βγ > (1 − α)(1 − δ), so “it can generate pathological self-confirmation as readily as a coherent perspective”. A9: “Reciprocal integration is a description of an organizational mechanism, not necessarily a distinctive mark of a perspective.” That leaves open “What distinguishes a reciprocally organized interpretation that remains responsive to the significance of conflicts and alternatives from one that simply entrenches its existing interpretation?”, which is CLAIM-276’s separation of coherence from fidelity.
• Interventional sufficiency (A8’s causal-state frame, revised by A9). The frame is a class of histories with the same responses to interventions. Its predictive part is TERM-062. CLAIM-407 shows that predictive closure is not enough. A9: a candidate “must preserve the response profiles of the interpretive and evaluative mechanisms relevant to its claimed organizational identity” (TERM-067). The target must be declared first (TERM-065, CLAIM-152): “We cannot discover a perspectival macrostate by selecting whatever coarse-graining makes our preferred theory of experience come out true.”
• Self-maintenance (A10). The organization maintains the constraints that govern its interpretation (TERM-070). CLAIM-414 shows that this is a separate, interventional fact. A11 withdrew it as a necessary condition, so it is a subclass (TERM-072).
• Cross-context answerability (A8, kept by A11). Changing an interpretation in one context alters the conditions under which another context’s information acquires significance, through mechanisms the organization maintains. No test of it has been run. A8’s three- architecture comparison (aggregation, external synchronization, reciprocal recontextualization, matched on outputs and self- descriptions) is its proposed design. Its negative control is a module that only declares unity, which is CASE-114’s point.

The question now has two halves, which A9 said must be kept apart:

1. Organizational. “When does a context-dependent system of interpretive operators and influence-allocation mechanisms constitute an endogenously maintained causal organization, such that its significant distinctions and responses to interventions cannot be preserved by a merely aggregative or externally coordinated description?” This is QUESTION-093’s question (predictive autonomy, interventional autonomy, organizational closure, self-individuation) asked of interpretive regimes, with QUESTION-115’s multi-criterion search for a level.
2. Phenomenal. “If such an organization can be identified independently of phenomenal judgments, what justifies treating its realization as constitutive of an experiential perspective?” A11’s sharper form: “Is perspectival organization merely what determines how things appear, or could the realization of that organization itself be what it means for anything to appear at all?” That is A3’s thesis A (content is constructed) against thesis B (experience is constituted), in TERM-072’s words: does sense 1 settle sense 3?

An answer to the first half does not answer the second. A9: the exchange risked becoming a theory “about the individuation of cognitive systems rather than the constitution of conscious experience”.

Note of 2026-10-22: the readings

The works the exchange cited, read on 2026-10-22, offer candidates for the organizational half. None answers the phenomenal half.

• Access at an interface (Whyte et al., THEORY-883): conscious access where continuous inference meets discrete choice. Stated for humans, and tied to report.
• Recursive precision modelling (Laukkonen et al., THEORY-882): as formalized, an ordinary hierarchical filter with a precision parameter per level would meet it.
• A coherent, egocentric world model (IWMT, THEORY-876). It needs a capacity for coupling and a history of it, not current coupling, so it is compatible with the grounding condition above. It gives no threshold.
• An attention schema (Graziano, THEORY-873): a representational condition, stated through report.
• Higher-order representation (NOTE-1066). This contests the question’s second contrast. On higher-order views, representing one’s own state is what makes it experiential, so constituting a perspective and representing one do not come apart as the question assumes.
• Life (Ward, THEORY-874). An enactivist would test any answer by whether its property is distinctive of life or only characteristic of it. The question’s ban on drawing the line from verdicts on machines cuts against that test.
• A workspace without a point of view (Gurnee et al., THEORY-865). In the base model the workspace exists, and post-training installs the assistant’s perspective. A candidate condition must say which of the two it tracks.

Barrett et al. (NOTE-1067) give no criterion for when categorization is experiential.

Note of 2026-10-25: Isaac’s invariants

Isaac (LIT-1361), read in NOTE-1163, is the reason the record filed the paper here. Its group-theoretic layer (THEORY-959) offers a structural mark of a perspective: a varying level of appearance and an invariant level abstracted from it, present together in one experience, which after Cassirer it takes to separate subject from object. It gives no criterion that keeps out a perspective-free invariant extractor, as any object recognizer is, and the paper calls the layer a suggestive sketch. It narrows the question to what co-presence adds over invariant extraction; it does not answer it. The inference is the record’s. The question is unchanged.

Note of 2026-11-01: Metzinger’s model of epistemic openness, as a candidate

Metzinger 2020 and 2024 (LIT-1946 in NOTE-1498; LIT-1943 in NOTE-1499) offer a candidate property of the kind the question asks for: a temporally thin, integrated model of the system’s own epistemic capacities, of its tonic alertness. Tonic alertness that is not so modelled is not experienced (2020, p. 33). In the 2024 active-inference sketch, the top level, a model of “the system’s global ‘epistemic openness’”, is what “endows the content generated on all lower-order levels with phenomenality” (pp. 476–477). Stated as THEORY-1299.

Tested against the question’s conditions:

• Not stated without reference to experience or report. The account was fitted to contemplative reports, and no measure of the model is given. Unresponsive wakefulness is offered as the case of alertness without its model. There the absence of experience is inferred from behaviour.
• Blindsight and the rest are not addressed in the pages read.
• On machines it asserts possibility, not a verdict. Tonic alertness “could be realized by very different physical properties in machines that create an integrated internal model of their own epistemic space” (2020, p. 32). This is the question’s worry: the account gives no reason why an engineered self-monitor of readiness would not qualify.
• It cuts against the question’s word “perspective”. The account’s minimal experience is “aperspectival” (2020, p. 11). If that is right, the organization that makes a state experiential is not the organization of a perspective in TERM-072’s senses 2 or 3, though it may be in sense 1.

So it is a candidate of the right form and does not answer the question. No status, title or warrant changes.