Medical Epistemology

Medicine operates under conditions that would paralyze most fields: incomplete information, time pressure, high stakes, and patients who can’t always articulate what’s wrong. A diagnosis is a bet — informed by evidence, but never certain until confirmed by outcome.

This makes medicine fundamentally different from laboratory science. In physics, you can repeat experiments until uncertainty shrinks. In medicine, you must act now, with this patient, using whatever information you can gather before the situation changes. The structure of knowledge is probabilistic from start to finish.

Atul Gawande’s writing explores how good physicians manage this uncertainty. They use heuristics — mental shortcuts that are usually right. They build decision trees that front-load the most dangerous possibilities. They recognize patterns from thousands of prior cases, like diagnostic thinking in any craft.

But pattern recognition fails at the edges. Rare diseases hide among common symptoms. Textbook presentations don’t match textbook cases. The best clinicians maintain productive doubt — confident enough to act, humble enough to revise. Donald Schön called this “reflection-in-action”: thinking while doing, adjusting on the fly.

Evidence-based medicine tried to make clinical reasoning more rigorous: randomized trials, statistical significance, systematic reviews. This helped — but created new problems. Trials study populations; clinicians treat individuals. The average effect obscures variation. Guidelines that work for most patients may harm edge cases.

The lesson: medicine is craft, not applied science. It uses scientific knowledge the way a carpenter uses physics — as background constraint, not step-by-step instruction. The clinical art is knowing when to follow the evidence and when the evidence doesn’t apply. That judgment can’t be codified.