Dignāga, the 5th-century Buddhist logician, spent his career doing something that looks strange until you see why it matters: he insisted that every valid inference requires not just a positive correlation — 'where there's smoke, there's fire' — but a negative one too: 'where there's no fire, there's no smoke.' Both conditions together. One without the other and you don't have a rule; you have a habit you've mistaken for a rule. The technical term is vyāpti — pervasion — and the negative half is called the vipakṣa test: can you identify a clear case where the consequent is absent and confirm the antecedent is absent there too? Most mental models we carry around have only ever passed the positive test. We built them from cases that confirmed them. The vipakṣa test is the pressure Dignāga says a rule must survive before you've actually earned it as a rule — and most of our working models have never been subjected to it. The practical move is simple and uncomfortable: pick a model you've been applying confidently, and go looking not for more confirming cases, but for a domain where the outcome is clearly absent — then check whether your supposed cause is absent there too. If it isn't, you don't have a principle. You have a pattern with a jurisdiction you haven't mapped yet.

Name a mental model you've applied across multiple domains — then identify one domain where the outcome it predicts clearly doesn't occur. Is the supposed cause absent there too?

Drawing from Buddhist epistemology (Dignāga-Dharmakīrti school of Pramāṇavāda) — Dignāga (5th century CE, Pramāṇasamuccaya)