Why is it hard to assess thought disorder? Clarifying the third domain of psychosis

Formal Thought Disorder (FTD) presents psychiatry's central paradox: it is one of the robust predictors of poor outcomes and polygenic risk in psychosis yet remains poorly defined and rarely measured clinically. We systematically reviewed 50 years of FTD assessment (16 rating scales, 32 factor analyses) to understand this paradox. Research to date has implicitly treated FTD as a natural kind, a latent entity that causes observable signs. Yet, empirical evidence contradicts this assumption: we find radical heterogeneity in construct definition, non-replicability of factor structures at the item level, and no universally essential properties across items. We propose measuring FTD as a Constituted Practical Entity: a probabilistic cluster of linguistic-cognitive features whose interaction produces communication failure. In this framework, no single feature is necessary or sufficient; dysfunction emerges from their relationships, not from a single latent process. This reconceptualization reconciles the multi-dimensional nature of FTD with our attempts to measure it and offers a clear research path: establish consensus constituents, measure their interactions, and develop computational tools. Without addressing the conceptual foundations, technical advances will only perpetuate existing confusions. Our framework clarifies what is being measured in the name of FTD and guides the development of computational tools and clinically meaningful targets for this third domain of psychosis.