Udyotakara, the 6th-century Nyāya philosopher, spent considerable effort distinguishing between a model that is merely undefeated and one that is actually confirmed — a gap his tradition called the difference between avyāhata (uncontradicted) and siddha (established). Most mental models live permanently in the first category while impersonating the second. They've never failed visibly, so we promote them to truth. The neuroscientist Roediger and Karpicke's 2006 work on the 'testing effect' showed something structurally parallel: retrieval, not re-exposure, is what actually tests whether knowledge is encoded — passive re-reading creates a feeling of familiarity that mimics mastery. Your mental models do the same thing. Every time a model gives you a plausible story about what just happened, you experience a low-grade sense of confirmation — but a story that fits the past is not evidence the model will hold forward. The discipline Udyotakara was pointing toward is not skepticism but specificity: for any model you rely on, ask what evidence would count as genuinely establishing it, not just failing to break it.

Pick one mental model you've used at least three times this week — what would it actually take for that model to be proven wrong, not just unused?

Drawing from Nyāya (Udyotakara's Nyāyavārttika) — Udyotakara (6th century CE, Nyāyavārttika), cross-referenced with Henry L. Roediger III and Jeffrey D. Karpicke (Washington University, 'The Power of Testing Memory', Psychological Science, 2006)