Research protocol · preregistered v0.1 · 2026-09-12

Integration–Continuity Experiment v0.1

A preregistered protocol for testing whether stronger cross-signal integration predicts greater task-state continuity and faster recovery under controlled perturbation.

Protocol and research architecture by John Brajer. This experiment tests an operational systems claim. It does not measure subjective experience and must not be interpreted as proof or disproof of consciousness.

Question under test

When task conditions are held constant, does stronger integration across relevant signals predict better continuity through perturbation, shorter recovery latency, and stronger preservation of task state?

Claim boundary

The preregistered claim concerns observable task-state continuity. Any analogy to consciousness remains interpretive and outside the experiment's evidentiary scope.

Preregistered variables

Φproxy

Cross-signal integration proxy assigned or measured for the matched condition.

C(t)

Continuity score across the perturbation window.

R(t)

Recovery latency after perturbation.

S(t)

Identity or task-state preservation after perturbation and recovery.

Locked comparison structure

The v0.1 protocol compares low-, mid-, and high-integration conditions under the same task template, deterministic seed policy, task budget, perturbation window, scoring function, and nuisance settings. The primary predictions were fixed before empirical comparison:

Higher integration should be positively associated with continuity.
Higher integration should be negatively associated with recovery latency.
The combination of higher integration and lower recovery latency should predict stronger task-state preservation.

Secondary analyses are exploratory. If matched preregistered conditions show no relationship, or a stable relationship in the opposite direction after checksum-verified reruns, the v0.1 integration claim fails.

Passive observational baseline

After preregistration, a live passive-baseline collector was added to observe existing production artifacts without writing to root sources, repositories, orders, or workspaces. Its data-source selection was frozen before comparison rather than adaptively choosing whichever signal produced a preferred direction.

Baseline v0.1 uses one matched low/mid/high observation set from a single UTC observation window. Its source roles are fixed: move for Path Expansion planning state, gate for execution truth, and deploy for deployment/release state. Raw observations remain distinct from normalized experimental measurements until the runner boundary.

Execution receipt

The first live passive-baseline workflow completed successfully on source commit 26185ca7…. The run validated and preserved one baseline artifact with SHA-256 digest 369316f1deed13125c829502b9de11fc0c582969ba10fe1f539d20e015191153.

A successful collection run establishes reproducible execution and provenance. It does not, by itself, establish support for any preregistered hypothesis.

Provenance

Protocol preregistration: Trillsverse Gate commit f75713cac0a9383752a2badea567d835b0d28bfe, September 11, 2026.
Passive-baseline execution architecture: commit aa332ad8be6bbab695a71a83eb7fa732c228d9fd.
First successful live baseline trigger: commit 26185ca77f6d475614bb958ab5c34aaa693cfc2f; workflow run 34649462991.
Authorship: John Brajer. Automation and AI tooling assisted implementation, validation, and publication packaging without changing the preregistered claim boundary.

Machine-readable record

The protocol's public machine representation is available at /integration-continuity.json. Future empirical results should be versioned separately so the preregistration remains historically immutable.