Matched Presents, Different Futures: History-Dependent Relational Response Geometry on a Superconducting Quantum Processor

Can systems matched on a declared present nevertheless show different future responses because they retain different relational organisation? We test this in a six-qubit hybrid quantum-classical architecture on IBM Heron hardware. Written and sham preparations use exactly the same five coupling values and the same number of native two-qubit operations; only the edge assignment differs. Under a frozen preregistered decision rule with randomised block orders, matched fast preparations displayed different later directional responses when the prepared slow classical relational configuration differed. Because the full-sample cosine alignment and norm-based retention estimates sat outside their original percentile intervals, a post hoc audit was performed from the sealed raw-count archive. That audit is reported alongside, and does not replace, the frozen rule. Claim boundary. The result is real under its frozen analysis. It is not autonomous cross-job quantum memory, and it is not evidence of a hidden quantum degree of freedom persisting between jobs. The classical coupling configuration is part of the preparation and is named as such throughout. No claim is made about quantum coherence surviving between executions. This deposit contains the manuscript in PDF and DOCX form together with the complete evidence archive: protocol freezes, compiled QPY circuits, inspection tables, randomised block orders, backend snapshots, raw-count files for both runs, IBM job ledgers, point estimates, blockwise endpoints, the frozen decision JSON, an artefact inventory and a checksum ledger.