A posterior preference articulation approach to dual-response-surface optimization