Google acknowledges that people "believe that the incremental traffic from high ranking keywords, typically branded terms, is low, or zero." But they don't address the issue of branded keywords in the research paper.
I still think the numbers would be a lot different if it were isolated to non-brand or brand.
edit: they do mention in the conclusion that essentially, your mileage may vary, especially with respect to brand/non-brand terms. Still, it'd be interesting to see a study of it.