Seol mar théacs é seo: Robust versus optimal strategies for two-alternative forced choice tasks