The Necessity and Justification for Cognitive Concepts in Animal Learning: A Reply to Galvani