Evidence for Abstract Response Codes: Ear-Hand Correspondence Effects in a Three-Choice Reaction-Time Task*