Spatial working memory for locations specified by vision and audition: Testing the amodality hypothesis