An Algebraic Framework for Indeterminate Predicates: Classification, Reasoning and Dispositional Dynamics
Thomas Bittner
Abstract
Knowledge representation standardly assumes bivalence: every entity determinately satisfies a predicate or determinately does not. Many domains (vague geographic regions, borderline biological taxa, mereologically indeterminate wholes) resist this assumption. The stabilizer formalism of quantum error correction is presented as a vector-based ontology language for classifying and reasoning about indeterminate predicates, with a dispositional dynamics: systems are represented by state vectors, predicates by Pauli operators, indeterminacy as an object-level property, not a meta-level family of precisifications. It complements, rather than replaces, the model-theoretic representation of classical first-order logic. The real (rebit) fragment is efficiently classically simulable (Gottesman-Knill), so no quantum hardware is required. Three determinacy modes (determinacy (D), gappy indeterminacy (G) and correlated indeterminacy (C)) generate, for three predicates, a complete classification hierarchy, its counts established previously, here derived as orbits of nested symmetry-group actions: 120 information kinds, grouped into 49 semantic kinds, 15 determinacy patterns, 7 information forms and 3 correlation classes. Applied to three mereological predicates, a representative subset of kinds is read explicitly, including correlated-indeterminacy patterns beyond bivalent satisfaction. Two results follow. The first is a violation-cost landscape, induced by a chosen dispositional specification, in which the ideal configuration is itself vague; its same-kind transformations are the normalizer of the unsigned landscape group in the group of local Clifford operations and predicate permutations. The second is a centralizer partition of the interaction space, deciding for each query whether it is revelatory or disturbing. The representability boundary lies between the semantic-kind and information-kind levels.