Map-Guided Theory Development: A Versioned Protocol for Thought Experiments, Semantic Audits, and Human-AI Research

Developing a theory requires more than collecting plausible propositions. Definitions can change between research episodes, separate witnesses can be combined incorrectly, and a successful local proof can cease to address the original question. We present Map-Guided Theory Development (MGTD), a protocol in which a versioned, semantically annotated research map mediates between conjecture generation, thought experiments, conditional deduction, criticism, and preservation. Claims retain their domains, quantifiers, interpretation, source versions, and epistemic status. Conjunctive premises are bound to a common application contract, alternative proof routes remain distinct, and a completed cycle includes a census-wide compatibility review plus deeper rederivation of affected conclusions. Minimal unresolved obligation sets provide a transparent representation of possible next research steps without asserting their truth or feasibility. We state elementary guarantees for the restricted deductive component and give constructive examples of the limits of structural and local checking. A retrospective case contains 1,278 recorded review items; it illustrates artifact design, not independent validation of the underlying consciousness theory. A small executable conformance suite rejects 12 deliberately constructed contract violations, accepts 12 benign controls, and checks frontier calculations on 1,024 finite rule graphs. The contribution is an explicit, reusable synthesis for theory-oriented human–AI research, not the invention of argument graphs, a universal discovery algorithm, or evidence of improved scientific productivity. Methods preprint, not peer reviewed. Human author of record and responsible depositor: Hongju Liu. Substantial AI assistance was used in the research synthesis, finite-model verification, draft, editing and publication preparation. This is not a validation of a consciousness theory or a measured improvement in research productivity.