On the Persistence of Cognitive Explanation: Implications for Behavior Analysis