State or Signal? Observable-Relative Causal Aliasing under Receiver-State-Dependent Relabeling

Preprint of the study “State or Signal? Observable-Relative Causal Aliasing under Receiver-State-Dependent Relabeling”. This work investigates a limitation of interventional causal individuation: two underlying causal organizations can remain indistinguishable when the observable retained for comparison discards response information that differentiates them. In a controlled synthetic finite-state testbed, a preregistered and blinded confirmatory experiment evaluated whether a response-order criterion could distinguish direct interface use from receiver-state-dependent interface remapping. The preregistered confirmatory result was negative: the primary criterion matched 150/150 E1 targets and 49/50 E2 targets (199/200 overall), but failed the conjunctive E2 requirement. This negative result is preserved unchanged and is not interpreted as validation or near-success. Post-confirmatory analysis of the sole counterexample identified and characterized the mechanism underlying that failure. Receiver state can induce a contextual relabeling of the interface that changes the underlying causal response field while leaving the retained observable invariant. The paper defines this failure mode as observable-relative causal aliasing: causally distinct underlying responses become observationally equivalent relative to the representation used for causal individuation. For the frozen intervention design, the exact-profile blindness mechanism admits an exact symmetry characterization, while a distinct ordinal-blind route arises when cardinal response profiles differ but their weak order is preserved. Progressive refinement of the retained response representation strictly reduced the observed indistinguishability set. Under the frozen eligible System-II generator, complete two-edge collapse occurred in 46 of 10,000 instances (0.46%). The resulting contribution is a concrete, formally characterized failure mode of interventional causal identification and a methodological warning about treating invariance of an observable as sufficient evidence that the underlying causal organization is preserved. All mechanistic, algebraic, prevalence, and refinement analyses are explicitly post-confirmatory and do not retroactively alter the preregistered result. The claims are deliberately scoped. The study concerns a controlled synthetic testbed and causal identification relative to specified interventional observables. It does not establish a universal state/signal ontology, universal causal non-identifiability, a universal failure frequency, or claims about consciousness. Software, confirmatory records, post-confirmatory analyses, and reproducibility materials: https://doi.org/10.5281/zenodo.21995651 Public source repository: https://github.com/rubenalexandrebrito-source/observable-relative-causal-aliasing