Possibly poor wording, my intent was to say that you cannot dismiss or underestimate statistics in modeling the mind because the damn thing is a sophisticated statistical information retrieval system (with some higher-level structure on top for doing things like debating mind modeling on HN).