Color is a difficult to understand topic anyway, and selecting appropriate colors for depicting experimental results not any less so. Now I found this excellent paper “Escaping RGBland…” (Zeileis et al. 2009) where you can make interactive educated choices – highly recommended. Zeileis et al. write “It is not necessary for the colors in a statistical plot to reflect fashion trends, but basic principles … should be adhered to.” and I will indeed change color choices in my analysis programs.

http://colorbrewer2.org