Preprint / scholarly article

Classification-Induced Cognitive Drift

K. Takahashi

Published
DOI
10.5281/zenodo.19306514

Full text PDF (Zenodo)

Abstract

This preprint develops a first-principles calculus for classification-induced cognitive drift in reflexive human and AI settings. It formalizes how disclosed classifications can change targets, evaluators, and later evidence under replay, repeated-measures, rollout, and observational comparison regimes.

Keywords

  • cognitive drift
  • reflexive classification
  • interactive kinds
  • looping effects
  • label feedback
  • performative prediction
  • strategic classification
  • algorithmic classification
  • human-AI interaction
  • evaluator drift
  • classifier state logging
  • contradiction-triggered revision
  • partial identification
  • causal inference
  • observational comparison
  • repeated-measures design
  • staggered rollout
  • interference-aware evaluation
  • auditability
  • deployment governance
  • transportability
  • deployment safety
  • AI safety
  • decision support systems

Identifiers and source records