How could one critique the assumption that conditioned responses will always occur immediately after the conditioned stimulus?