Replication anxiety is eating science alive — but the deeper problem isn't fraud or p-hacking. It's that scientists rarely ask whether their conceptual vocabulary has kept pace with their data. The pragmatist philosopher Charles Sanders Peirce spent the 1870s arguing that scientific concepts are not neutral containers for facts; they are habits of inference, and like all habits, they can become so automatic that they start filtering out the very phenomena that would force revision. Meanwhile, the Nyāya-Vaiśeṣika philosophers of classical India — working from Praśastapāda's 6th-century synthesis — developed a parallel doctrine about 'category error': the idea that confusion in inquiry often isn't a failure of observation but a failure to notice that your organizing categories have become coarser than the distinctions reality is offering you. Put them together and you get something unsettling: a researcher can be rigorous, careful, and honest, and still produce a body of work that systematically misses what's happening — because the concepts doing the sorting are older than the questions now being asked. The practical move is almost embarrassingly low-tech: once a year, pick one core term in your field and ask whether the word was defined before or after the phenomenon you're currently studying. If before, assume the concept owes you an audit.

Name one concept you use routinely in your scientific work that you inherited rather than chose — then ask: what class of findings would it structurally prevent you from noticing?

Drawing from Pragmatist Philosophy of Science / Nyāya-Vaiśeṣika Classical Indian Philosophy — Charles Sanders Peirce (Illustrations of the Logic of Science, 1877–1878) in dialogue with Praśastapāda (Padārthadharmasaṃgraha, c. 6th century CE)