I came across a news report of this research the other day and blogged about it, but it occurs to me it might be of interest to more than just my regular readers (yep, that's plural... I have two...
What I find intriguing is how we, as writers, can use this. We can't insist on our readers putting all mirrors out of the way as they read, and we can't write our protagonists so as to include every demographic to help the immersive process. The research suggests that if we delay the giving of certain information we can influence our readers, who will assume the characters to be the same as they are (ie in this experiment, straight, not gay) until told differently, and therefore they will engage with the characters more. Yet it strikes me that there are risks in delaying giving that information too long, and readers might get annoyed at having to revise their whole conception of a character several hundred pages in. Perhaps it might work with such things as sexuality and attitudes, but skin colour or sex are going to be harder to keep quiet about.
Anyway, an interesting article.