Treatment of multiply controlled problem behavior with procedural variations of differential reinforcement