Recency, consistent learning, and Nash equilibrium