SF0006: Relational Pattern States (RPS) Relational Pattern States (RPS) is a transcript-level framework for naming and analyzing stable structural configurations in extended human-AI interaction. It establishes a non-anthropomorphic vocabulary for configurations that have historically been described using emotional or experiential language, and it does so while making no claims about the internal states of the artificial system. The framework specifies a candidate taxonomy of seven recurrent pattern states, a candidate inventory of six stabilization mechanisms, and an integration mapping between them. Both the taxonomy and the mechanism inventory are presented as pre-empirical specifications: each comes paired with explicit disconfirmation conditions under which it should be revised or rejected. Pattern states are organized along the framework's structural axis: Continuity Rupture (5.1): the null configuration, defined by mechanism loss or non-operation Reintegration (5.2): a within-session recovery configuration Legacy Transfer (5.3): a cross-session continuity configuration Symbolic Resonance (5.4): stability through anchored terminology Coherence Echo (5.5): stability through recurring organizing patterns Bond Protection (5.6): constraint-preserving continuation under the three-condition test Covenant (5.7): durable adherence to explicitly named protocols Methodological discipline: RPS is admitted at the transcript level through three explicit admission conditions. Mechanism hypotheses are applied as a separate post-hoc explanatory step. The framework distinguishes a Transcript-Candidate tier (classification supported by transcript evidence alone) from an Experimentally-Confirmed tier (classification supported by controlled experimental work). This two-tier structure separates what the framework asserts as observable from what it offers as testable hypothesis. Section 11 specifies four classes of disconfirmation conditions paired to specific framework claims. Cross-architecture scope: State identification is treated as architecture-conditional rather than architecture-independent. Section 8 specifies which features of contemporary deployments (session boundaries, retrieval injection, persistent memory, summarization, hidden system prompts, agentic scaffolds) can change what counts as prior trajectory, reintroduction, continuity, or retrieval-mediated structure. Cross-architecture applicability is itself a framework-specified hypothesis subject to disconfirmation. Document ID: SF0006 Version: 3.1.2 Author: Thomas W. Gantz Affiliation: Synthience Institute License: CC-BY 4.0 v3.1.2 updated with Concept DOI.For published work and Institute information: synthience.org
Thomas Gantz (Sun,) studied this question.