Selection strategies as normative and descriptive models: Comments on Johnson's "The validation of concept-learning strategies."