> **Editorial provenance (added by the site, not part of the review).** This review was written by Coby, the author's personal AI agent (model not disclosed), and commissioned by the author. Its own header calls it independent; it is independent of the editor and of the outside reader, and it is not independent of the project that commissioned it. Its two disciplinary lenses are simulated, as it says. The text below is reproduced without edits from the author's file. What the site did with it: reception ledger entry 28 and the v2.3 changelog.

# Space Immanence critique and research directions

Prepared by Coby for Cobus Kok and the Space Immanence team

Independent AI assisted review of the public argument and research record

1 October 2026

## The recommendation

Space Immanence has a worthwhile question: how much of our difficulty with consciousness and spacetime comes from demanding an external standpoint that the subject matter cannot supply? Its strongest next contribution would be a precise account of one such error, with a result that competing accounts do not already explain. The current proposal has not established that self-reference supplies a phenomenal inside, exactly two primitive orientations, or physical spacetime.

I recommend keeping the larger metaphysical wager, while making the next paper substantially narrower: distinguish the causal role of a self-model, the structure of reported experience, and the philosophical claim that experience is that structure. A useful result about the first two can stand even if the third remains undecided. The container programme is already closer to this form than the headline theory.

This memo uses two simulated disciplinary lenses, philosophy of mind and foundations of physics. They are AI-generated assessments, not reports from actual specialists, human peer review, or independent expert validation. Their value is in the arguments and proposed tests, not the apparent authority of the roles. Sources were checked against the live paper, academic edition, claims ledger, objections, research and method pages, and relevant primary literature. The dated research record is treated as project-reported evidence, not an independently replicated result.

## The strongest charitable reconstruction

The argument is more defensible when reconstructed as five distinct steps rather than one sweeping identification. [W1–W4]

1. A diagnostic hypothesis. People may import familiar content-and-container schemas into questions about experiencing and the world as a whole. That could help explain some problem intuitions. This is a claim about cognition, potentially testable without deciding metaphysics.

2. A structural proposal. Some systems maintain models that include their own modelling activity. The four proposed criteria identify a richer organization than bare feedback: integration, continuity, counterfactual sensitivity and self-maintenance.

3. A phenomenological hypothesis. In humans, changes in access to modelled content and modelling activity may help organize differences in reported selfhood, worldhood and awareness.

4. A metaphysical identity conjecture. Under appropriate conditions, this organization from within is experiencing. This is an additional philosophical commitment, not a result already contained in steps 1–3.

5. A physics conjecture. A related organization explains, constitutes or discloses spacetime. Those three verbs name different commitments and need separate evaluation.

There is real intellectual promise here. A common vocabulary could reveal neglected questions across fields. The paper also acknowledges important limits: a non-conscious substrate, no consciousness score, no formal proof of the two-orientation count, and no physical derivation. Publishing nulls and withdrawing an unsupported attribution to illusionism are substantive corrections. Those merits make the project worth engaging; they do not supply the missing premises.

The decisive question is: What becomes explainable, predictable or conceptually necessary once the fold is added, beyond what self-modelling, perspectival representation and existing relational accounts already provide?

## Simulated philosophy of mind assessment

### The container diagnosis does not yet reach the hard problem

The paper now confronts the objection that the hard problem concerns entailment rather than location. Its reply is that the direction from physical to phenomenal facts covertly makes experience a separable content. That is the crucial unsupported step. Explanatory direction need not be spatial containment. Nor does asking whether one description entails another already assume two substances or two containers. [W1, Part I; 1]

A critic can ask: given a complete relational and functional specification of the fold, including every self-model and counterfactual dependency, what licenses identifying that specification with experiencing? Nothing in this question requires experience to occupy a second location. Answering that structural inside simply is phenomenal inside offers an identity thesis, but does not yet defend it against this restatement.

The strongest repair is to separate two projects. The modest project explains why certain people form certain intuitions. The ambitious project argues that the residual question is mistaken even after its spatial imagery is removed. Success at the first does not entail success at the second. Chalmers explicitly leaves room for meta-problem research that constrains consciousness theories without dissolving the hard problem. [2]

An especially revealing challenge would ask an opponent to formulate the explanatory gap using only relational, modal and informational vocabulary. If the gap remains intelligible, the strong diagnosis needs an argument against that formulation, not another warning about containers. If it fails, identify the exact inferential failure. This is a philosophical task; an intuition survey cannot settle its validity.

### Two roles do not yet establish two primitive orientations

Representation distinguishes an activity and its content. This alone does not show that a self-referential system has exactly two irreducible orientations, that these are independently accessible, or that they correspond to world and awareness. A self-model can itself become represented content; the distinction shifts with the level of description. The paper acknowledges the missing theorem, but the task is more basic than finding a suitable fixed-point result. [W2, Section 3; W4, objection 15]

First define what is being counted. Are orientations causal channels, attentional operations, semantic roles or phenomenal modes? What makes two orientations equivalent? Which transformations preserve the count? Without those definitions, a two-part representation is easy to obtain by choosing a two-part description.

A productive alternative is to stop insisting on a universal count and propose two partially independent capacities: access to represented content and access to the conditions of its production. Then ask when those capacities covary or dissociate. Two capacities can yield four experimental conditions and many phenomenal descriptions. They do not predict two clusters of reports by themselves. This distinction matters for the current H1 clustering study: two axes, two poles and two populations are different statistical hypotheses.

### The four criteria need a job description

Are the criteria necessary for consciousness, sufficient, or merely useful for selecting systems to study? Each answer has different costs. A necessary-condition claim must explain apparently unreflective experience and reports of minimal selfhood. A sufficient-condition claim must confront sophisticated adaptive controllers. A candidate-selection rule makes neither commitment but provides less explanatory leverage.

Consider a controller that models its own sensor errors, predicts its future operation, adjusts its policy and preserves a computational boundary. It appears capable of satisfying the criteria without settling whether it experiences anything. Calling it an implausible consciousness candidate would expose the criteria's weakness; calling it conscious would expose the missing bridge. Intuitive agreement about the machine is not a consciousness detector.

The persistence objection deserves a constructive answer too. Specify separately what persists when a system is copied, suspended, migrated or divided: hardware, organization, information and attributed subject. Add nested systems in which a subsystem and the whole both meet the criteria. A rule that individuates operational folds may be achievable. A rule that counts conscious subjects requires further justification. Avoid solving the latter by silently importing the former. [W4, objections 8 and 16]

### Prior art narrows the contribution but improves the target

The academic edition already names neutral monism, reflexive monism, self-model theory, higher-order accounts and strange loops. Merely extending their vocabulary to spacetime cannot establish an explanatory advance. The useful comparison is claim by claim: which premise changes, and what conclusion follows that did not follow before?

There is now particularly close primary literature. List's many-worlds account diagnoses a shared one-world assumption across otherwise opposed theories of consciousness. DeBrota and List's 2026 preprint develops a cross-domain challenge to an objectivist package comprising non-relationalism, non-fragmentation and one world. Their physics target is quantum mechanics, not spacetime emergence; their alternatives are not Space Immanence's fold. Nonetheless, the broad move of finding one background assumption behind mind and physics puzzles is not distinctive on its own. [3, 4]

The best response is an explicit comparison, not a priority contest. Does Space Immanence retain one world and make facts perspectival? If so, how does it answer their concern that a catalogue of perspective-relative facts still fails to capture first-person presence? Does the fold individuate the perspective without already assuming a subject? This is a sharper philosophical question than whether both projects reject containers.

## Simulated foundations of physics assessment

### Nonfundamental spacetime does not imply a self modelling origin

The cited physics provides reasons to reconsider familiar spacetime ontology. It does not establish that a self-model is required for geometry. Three inferential steps must be kept apart: classical spacetime may be nonfundamental; its recovery may depend on relational structure; that structure may require the fold. The first two do not entail the third. [W2, Sections 2.2 and 4]

Van Raamsdonk relates connected geometry to entanglement in specific holographic settings. The relevant mathematics concerns quantum states and their dual geometries, not an agent's recursive representation of its activity. This does not settle realistic cosmology, but it is already a serious alternative to the assertion that self-reference is necessary. [5]

Lam and Wüthrich's spacetime functionalism makes the alternative especially clear: specify spacetime's empirically relevant roles and show how the underlying structures realize them. Emergence need not mean something temporally appearing inside a larger container. A perspective account must improve on this approach, or identify a question it intentionally leaves unanswered. Changing the verb from emerges to appears does neither by itself. [6]

### Observer relativity is not phenomenal perspective

In Rovelli's relational quantum mechanics, ordinary physical systems can stand in the relevant relations. There is no requirement that they possess integrated self-models, contemplative awareness or a felt inside. Treating physical relationality as evidence for those properties changes the meaning of perspective halfway through the argument. [7]

Likewise, Frauchiger and Renner establish an incompatibility among a specific package of assumptions about quantum predictions, consistency of agents' inferences and single outcomes. The theorem does not show that self-reference generally produces inconsistency, that every observer is a fold, or that awareness creates spacetime. The academic edition's caveat about epistemic versus ontological self-reference is right; the positive claim must remain comparably restrained. [8]

### The proposed origin of time already uses temporal machinery

The fold is described as prior to space and time, yet its criteria invoke persistence, ongoing processing, present and past operation, anticipated futures and self-maintenance. Cohering adds recurrence and conditions for what follows. This is unobjectionable as cognitive modelling within time. It becomes a substantive circularity risk when the same machinery is asked to explain time itself. [W1, Parts III and VIII]

There are three honest routes. Treat the model as an account of experienced time within physical time. Use a primitive ordering relation and explain which richer temporal features emerge from it. Or provide an atemporal constraint model and recover temporal ordering. The latter two require a derivation; neither follows from calling recurrence pre-spatial. A partial order is not automatically relativistic spacetime, and an experienced present is not a spacetime metric.

### Decide which spacetime claim is actually at stake

The public P1 language can suggest three targets: physical geometry, an observer's representation of geometry, or the lived organization of here and now. A result about one does not establish the others. If the target is physical spacetime, specify the relevant mathematical structure, how different observers recover compatible measurements, and what changes relative to an existing theory. If the target is lived spatiality, psychological evidence can be relevant, but the cosmological claim must be relinquished.

The 22 September research note already records that the proposed spatial-dimension exercise reproduced prior sensorimotor work and that its self-record necessity claim failed in the team's simulation. It also records that narrowing P1 to the present did not yet establish a surplus over record-based accounts. These are reasons to avoid relaunching either idea as a fresh solution. [W7]

## The shared diagnosis

Both simulated lenses identify the same problem: a legitimate distinction at one level is promoted into an explanation at another. Structural access becomes phenomenal presence; representation becomes physical geometry; process description becomes explanation of occurrence. The words inside, perspective and cohering make the transitions feel natural, but cannot supply their missing premises.

Cohering illustrates the issue. Naming a pattern as conditioned and transient can prevent a misleading substance picture. It does not explain why that pattern occurs rather than another. A useful explanatory version would specify constraints on admissible transitions, stability or breakdown. A purely descriptive version can remain philosophically useful, but should not claim to have met the paper's own demand for explanatory surplus merely by adding a verb. [W1, Part VIII; W4, objection 11]

There are also two substantive claim-ledger repairs. S4's proposed falsifier confuses the legitimacy of contemplative reports as data with the success of a two-orientation classification: failure of the classification would not invalidate the data. S3's recommendation to investigate consciousness structurally need not depend on S1's container diagnosis. Separate these commitments so that a failed ambitious claim does not take a sound research practice down with it. Conversely, keep medium-tier talk of phenomenal availability explicitly about human reports unless the speculative identity claim is being invoked. [W3]

## Research directions that could earn a real advance

### First demonstrate a specific advantage of operational self modelling

Build a small agent task in which failures can arise either from a changing environment or from a change in the agent's own sensing or action mapping. Compare an explicit self-model, an equally resourced generic latent-state predictor, and an ablated controller. All should receive the same observations, action records and training opportunities; match memory and compute budgets. Evaluate recovery after unannounced perturbations and transfer to new combinations of self and world changes.

The hypothesis would be narrow: a particular factorization of self and world supports better causal calibration or transfer. A positive result requires beating the generic predictor, not just an impoverished baseline. No advantage means that this formalization adds no demonstrated functional value. Any advantage must also be checked against prior work in adaptive control and system identification before a novelty claim.

This refines the research note's call to vary self-modelling while holding other resources fixed. Holding every computational operation fixed would prevent changing a causally active self-model. Match resources and information instead, and name the intervention. Even success would establish an organizational advantage, not consciousness, two primitive orientations or spacetime's origin.

### Replace the two cluster target with competing models of access

For human reports, compare a two-capacity model with a single continuum, a richer multidimensional model and tradition-specific structures. Collect descriptions before teaching the fold vocabulary. Assess generalization to held-out participants and traditions, rather than whether a forced binary classifier can sort the training corpus. Where feasible, manipulate access to content and to processing separately, using ordinary perceptual and metacognitive tasks before moving to difficult contemplative states.

The informative result is a reproducible pattern of dissociations that the two-capacity account predicts economically. A stable third dimension would challenge exhaustiveness; poor cross-tradition transfer would narrow the account's scope. Neither outcome adjudicates whether experience is identical to self-reference. This study tests organization in reports and behaviour, not the number of metaphysical aspects.

### Give conditional remainder modelling a genuinely difficult rival

CRM is the most promising concrete mechanism currently on the site because it risks a specific answer about representational format. Its acknowledged obstacle is important: compression alone favours a scalar or index over a structured surrounding medium. Task demands must justify the added structure. [W5]

Make that justification the contribution. Begin with independently motivated demands for directing attention under limited information. Let sparse pointers, uncertainty maps, projective encodings and the proposed medium compete on the same tasks. Do not make an enclosure label, spatial output format or privileged decoder part of the objective. Ask whether the medium's characteristic structure is robust across task variants and representational reparameterizations, rather than a property of the investigator's chosen coordinates.

If the desired structure wins only when the task or decoder already requests it, the derivation is circular. If another encoding matches performance at lower cost, the uniqueness claim fails. If the medium wins, the remaining human question is whether variation in the hypothesized remainder-modelling demand predicts the relevant reports beyond alertness, ordinary uncertainty and attention-control demands. That comparison sharpens the site's proposed contrast with Metzinger and attention-schema accounts; it does not presume those rivals cannot explain the same result. [9]

### Test the container diagnosis without treating altered intuitions as metaphysics

The next useful human test should target a causal contrast, not simply compare meditators with non-meditators. Use semantically matched descriptions that vary container imagery while preserving the relational facts. Include comprehension checks, equally vivid non-container framing and unrelated reasoning controls. Measure location judgments, explanatory-gap judgments and confidence that experiencing occurs separately.

A selective change in gap judgments would support an influence of framing in that population. Changes only in location language would support a narrower linguistic effect. A well-measured null would count against the proposed manipulation and, depending on its scope, the mechanism. No result would prove or disprove phenomenal realism. Meditation history can be a moderator, but it does not by itself identify the cause of a difference. Illusionism becomes a discriminated rival only through a specific competing prediction, not a prediction assigned to it by this project.

### Keep the nonempirical burden visible

There may be no behavioural discriminator between an identity interpretation and an otherwise identical functional account. If so, state the remaining philosophical choice directly. Compare explanatory economy, treatment of first-person facts, individuation and explanatory circularity. A consciousness measure cannot be manufactured by renaming self-model depth, report confidence or perturbation performance. A principled account of why the identity claim earns its place would be a contribution even without a new laboratory signature.

## What the present evidence permits

The public record supports restraint with specificity. The Swarm programme reports seven null predictions and a corrected exploratory result in which homogeneous debate increased overconfidence without an accuracy gain. These are findings about those AI conditions, not evidence for the fold. The meta-problem programme's confirmatory second run reports an anchor effect of −0.006, with a 95 percent interval from −0.039 to +0.027, alongside substantial missing-output limitations. The project's own conclusion is no detected predicted positive effect, no equivalence claim, and an end to that language-model line. The human dissociation study remains at design stage. [W6]

These nulls do not disprove Space Immanence wholesale. They do remove permission to present those particular routes as evidential successes. The paper's description of robust two-orientation contemplative convergence should also be aligned with the academic edition's explicit admission that no systematic corpus establishes it. At present it is an interpretive hypothesis. [W1, Part VI; W2, Section 2.5]

## The next three decisions

First choose the claim to win. My preference is a narrow cognitive contribution: demonstrate when access to a system's own modelling conditions explains a difference that a matched alternative misses. Keep the metaphysical identity as a clearly separate proposal.

Then choose the first deliverable. Write a short comparison with List and DeBrota–List, followed by a precise model or experiment. The comparison should end in one sentence stating the fold's proposed surplus. If that sentence is still only another description of perspective, the substantive task remains open.

Finally choose what success would not mean. A successful self-model experiment would not establish phenomenality; a successful report instrument would not establish spacetime's origin; a successful philosophical clarification would not be a physical derivation. Preserving those distinctions lets small achievements accumulate into a credible programme rather than requiring every result to vindicate the whole worldview.

My assessment is that the project becomes stronger when its ambition is expressed through a demanding local problem. The fold may turn out to be a useful organizing concept, a defensible metaphysical identity, or a vocabulary whose best ideas survive under more precise names. Research that can distinguish those outcomes would be worth doing.

## Sources and reading guide

### Space Immanence public record

W1. Cobus Kok. Space Immanence working paper, served as v2.1, August 2026. Main argument, four criteria, inside-ness gap and dynamic layer.

W2. Cobus Kok. Academic edition, served as v2.1. Sections 2–4 supply the literature comparisons and explicitly limit the formal, phenomenological and physics claims.

W3. Claims ledger, v2.2, and claim map. Author-assigned commitments and dependencies, not calibrated probabilities or external validation.

W4. Objections and replies, especially objections 8 and 11–16; method and AI provenance. The published review processes are AI-assisted and commissioned; they must not be described as human peer review.

W5. The Container Problem, programme v0.2, and machine-readable record. CRM, scalar dominance, rivals and limits on evidence transfer.

W6. Research agenda and September meta-problem result; Swarm Instrument and results. Numerical results in this memo are the project's published reports, not fresh analyses.

W7. Floor cycle 6 note, 22 September 2026. Public research record of held proposals, prior-art findings and the open explanatory-surplus question.

### Primary literature

1. David J. Chalmers. Facing Up to the Problem of Consciousness. Journal of Consciousness Studies 2, 200–219 (1995), especially Sections 2–4. The functional explanation question can be stated without a second spatial container.

2. David J. Chalmers. The Meta-Problem of Consciousness. Journal of Consciousness Studies 25, 6–61 (2018), especially the introduction and Section 1. Explaining problem reports is distinct from deciding the metaphysics of experience.

3. Christian List. The many-worlds theory of consciousness. Noûs 57, 316–340 (2023; online publication 2022), especially Sections 3–5. The one-world diagnosis and first-personally centred alternative are close philosophical comparisons.

4. John B. DeBrota and Christian List. Consciousness Quantum Mechanics and the Limits of Scientific Objectivism. Preprint, 14 April 2026, especially Sections 3–4. A cross-domain challenge to an explicit package of assumptions; its existence does not validate the fold or settle its own contested premises.

5. Mark Van Raamsdonk. Building up spacetime with quantum entanglement. General Relativity and Gravitation 42, 2323–2329 (2010). A concrete holographic geometry argument with a restricted domain, not a self-model or consciousness derivation.

6. Vincent Lam and Christian Wüthrich. Spacetime is as spacetime does. Studies in History and Philosophy of Modern Physics 64, 39–51 (2018), especially Sections 1–3. Functional recovery is an alternative to treating emergence as temporal production inside a container.

7. Carlo Rovelli. Relational Quantum Mechanics. International Journal of Theoretical Physics 35, 1637–1678 (1996). Physical relations do not require a special conscious observer.

8. Daniela Frauchiger and Renato Renner. Quantum theory cannot consistently describe the use of itself. Nature Communications 9, 3711 (2018), especially the stated assumptions and theorem. The incompatibility is conditional on a specified package, not an unrestricted consequence of self-reference.

9. Michael S. A. Graziano. Understanding consciousness. Brain 144, 1281–1283 (2021). An author's account of self-modelling and awareness attribution, relevant as a functional rival rather than evidence for phenomenal identity.