A note on holonomy groups and sigma models