klayman-ha-positive-test-strategy-bayesian-optimal

Summary

This finding reframes the "positive test" approach — checking whether examples fit a rule rather than trying to disprove it — as actually the smart move when the rule you're hunting for is narrow and unlikely. It pushes back on the common assumption that this strategy is a reasoning error, showing it only fails when the target rule is broad and common, which matters for how the system should weigh evidence from hypothesis-testing tasks.