why is an old-fashioned behaviorist linear-association model not adequate to explain complex organized movements such as the difference between a horse's amble and its gallop?