version of this idea, called interactive trial and error learning, which has the property that it implements Nash equilibrium … behavior in any game with generic payoffs and at least one pure Nash equilibrium. Unlike regret testing (Foster and Young, 2006 …