# AkashicOMNI Theory Card TC-B01 · Unsigned Candidate v0.1

**Question:** Can a named consciousness model predict above-chance matching to a target generated only after a person's choice is committed?

**Prepared:** 26 September 2026. **Status:** Unsigned proposal for adversarial review. No ethics approval, preregistration, participants, trial data or observed anomaly. **Companion:** *AkashicOMNI Study Plan and Protocol v0.1*, Track B. This card does not define or validate a physical field, and it does not change the published AkashicOMNI v0.4.3 specification, BQ001's UNRESOLVED status, or any evidence or canonical-edge gate.

## 1 · Exact proposition proposed for commitment

> Under the precisely specified one-choice, four-target, post-choice randomization procedure in the companion protocol, the probability of a match is **at least 0.30** across the eligible study population, rather than 0.25 under independently uniform targets and no information access.

This is a *candidate* statistical prediction, **not** a claim already entailed by panpsychism, animism, cosmopsychism or fundamental consciousness in general. The proposer must name a particular version of a model and explain why it predicts access to a future target, why 0.30 is the minimum expected rate, which conditions matter and which rival accounts give a different prediction. If that argument cannot be supplied, Track B may remain an independently interesting information-access test, but it is **not a test of that ontology**.

**Model name and version:** [proponent to enter]  
**Authoritative source and date:** [proponent to enter]  
**Model's mechanism or explicit rule linking it to the target:** [proponent to enter]  
**Why post-choice target access follows, rather than merely ordinary awareness:** [proponent to enter]  
**Why the effect is at least five percentage points above chance:** [proponent to enter]  
**Boundary conditions:** [proponent to enter before data]  
**Alternative predictions:** [critic and proponent to enter jointly]

## 2 · Observable and competing explanations

| Account | Registered expectation | What must be checked |
|---|---|---|
| Independence/null | For every locked choice, the later target is one of four uniform labels, independent of the choice; match probability 0.25. | Generator entropy and mapping, receipt order, duplicate handling, missing targets, access permissions, label distribution and serial dependence. |
| Candidate information-access rule | Match probability at least 0.30 in both independent cohorts under the exact procedure. | Pre-data theoretical justification, stage-specific effects, independent replication and clean technical audit. |
| Ordinary implementation or selection artifacts | Apparent excess can arise from leakage, target dependence, retries, selective exclusions, site anomalies or analytic flexibility. | Raw flow counts, signed first-choice and later-target receipts, append-only incident logs, masked exclusions, infrastructure separation, security review and new independent deployment. |
| Other unknown mechanisms | A reproducible effect might have more than one explanation. | Additional targeted experiments with distinct predictions; a match-rate anomaly alone cannot identify an ontology or a physical field. |

Generating the target *after* the choice probes a future-target assertion. It does not test access to an already existing hidden target. The deliberate choice of a post-choice design should be accepted explicitly by the signing proponent; no later substitution without a new card and registration.

## 3 · Locked experimental contrast

- One consented adult contributes one locked selection among four neutral symbols, then the separated service draws one uniform target label from two cryptographic bits. The target service accepts one signed first-choice receipt and issues one signed later-target receipt. No redraw is available to the front-end or investigator.
- Stage one: 1,200 valid first attempts, 600 per supervised recruitment channel. Stage two: a new independent cohort of 1,200, with separate team, deployment, keys and audit. Recruitment stops on *validity counts*, never match counts.
- A validity decision is made before outcomes are unmasked. All starts, invalid locked attempts and reasons are published as flow counts; technical errors are never hidden. Individual correctness cannot govern an exclusion.
- Primary supportive criterion from the companion protocol: **at least 340/1,200 matches in each stage**, with an exact one-sided null tail of approximately 0.00463 at 340. The probability of meeting this threshold in a single stage is approximately 90.2% if the true rate is 0.30. The probability of *both* independent stages meeting it at that boundary value is about 81.4%; a miss in one stage alone is therefore not a logical refutation of a 0.30 model.
- Both stage outcomes, site breakdowns, 95% exact binomial intervals, target-label distribution, audit incidents and extreme-case bounds for missing targets are reported, including null and compromised results.

The sample size and thresholds require independent statistical reproduction before registration. The hypothesis assumes uniform independent targets and independent first attempts; compromised audit conditions take precedence over a numerical result.

## 4 · What would weaken the precise 0.30 claim?

A failure to reach 340 in either stage means only that this protocol's **strict two-stage supportive criterion was not met**. To challenge the stronger *minimum-rate* assertion more directly, propose a separately prespecified model-challenge rule across the two complete, clean stages: if their combined **2,400 valid choices produce at most 662 matches**, the exact lower-tail probability under p = 0.30 is approximately **0.00496**. This is equivalent to an exact one-sided test against that boundary at alpha 0.005. The combined rule is used **only to challenge** the 0.30 minimum and never to rescue a stage that misses the supportive criterion. An independent statistician must verify and approve this added rule and its handling of model heterogeneity, missingness and site dependence before the first real choice; until then it is a proposal rather than a locked analysis.

At a true p = 0.25, the chance of observing at most 662/2,400 is about 99.8%; this rule has little risk of calling a clean chance-level result “compatible with at least 0.30” merely from low power. A clean result between these decision regions remains unresolved with an effect estimate and interval. Even an outcome in the challenge region weakens only this *specific rule under these conditions*, not all possible versions of a consciousness-fundamental ontology.

**Validity gate:** if target generation, event order, masking, independence or predefined exclusion rules fail materially, classify the result as compromised and register a clean repeat. A statistically unusual compromised result does not support or challenge the ontology.

## 5 · Claims this card does not license

No positive result alone establishes panpsychism, cosmopsychism, animism, universal experience, a consciousness field, information transfer through a particular mechanism or survival after death. No negative result alone disproves those ideas. Track A needs a separate theory card with a quantified sensor readout, calibration, rival contrast and an effect-size-based sample calculation. Culture-specific or animist traditions should be represented by authorised collaborators on their own terms, not conscripted as a universal laboratory mechanism.

The nested pyramid and sphere artwork and any “5D+” language are visual metaphors for perspectives and models. They are not measured spatial dimensions, instrument readings or evidence of a physical field.

## 6 · Preregistration and independent review record

Before recruitment, attach a timestamped protocol, all code hashes, mock-data audit, randomizer implementation, risk and consent materials, ethics determination, site list, funding and conflicts, data-access roles and independent review of the numerical rules. Register changes before the affected data are unblinded, retain previous versions and publish reasons. Make stage-one outcomes public before stage-two registration without tuning the latter to rescue the former. Publish all results, including nulls and security incidents.

| Sign-off role | Name and institution | Position, date and signature |
|---|---|---|
| Model proponent | [blank] | [accept exact rule / decline link to model] |
| Adversarial model critic | [blank] | [agreement, dissent and strongest rival] |
| Independent statistician | [blank] | [reproduce thresholds; approve or amend challenge rule] |
| Security / measurement engineer | [blank] | [verify generator, event order, isolation and audit] |
| Ethics lead / board reference | [blank] | [approval or determination number; not yet obtained] |
| Study principal investigator | [blank] | [responsibility, conflicts and resources] |

**Release gate:** This unsigned card is ready for critique, not recruitment. A signed card plus ethics and preregistration would authorize a specified study, not a v0.5.0 AkashicOMNI release. Claim Review Records should preserve source dependence, counterevidence, alternative mechanisms, dissent and a revision condition.

## Method references

- Melloni et al. (2023), [Cogitate adversarial protocol](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0268577), *PLOS ONE* 18:e0268577. Example of competing predictions and prespecification, not evidence for this hypothesis.
- [SPIRIT 2025 statement](https://www.bmj.com/content/389/bmj-2024-081477). Protocol transparency inspiration; this nonclinical task does not claim full trial-checklist compliance.
- [NIST SP 800-90A Rev. 1](https://nvlpubs.nist.gov/nistpubs/specialpublications/nist.sp.800-90ar1.pdf). Relevant random-bit-generator guidance, with independent implementation review still required.

**Authorship and transparency:** Human-directed AkashicNET research question; AI-assisted draft and numerical calculation. This document records no human signatures or external scientific approval.