[luu]

I wonder how much of the effect is because black on white is actually superior, and how much is because people have so much more practice at reading black on white. It would be interesting to see the studies repeated, using sub-groups that are used to black on white and white on black, respectively.

[alaithea]

My thoughts exactly. It would be hard to find many people who almost exclusively read white on black, though. I think the groups people break down into are likely the average people from "book-land", where all is black on white, and techies who use dark background terminals, but are fluent in both. In which case, it would make the most sense to test people who have extensive exposure to both (techies) for comprehension, etc.

[G_Wen]

An interesting experiment would be to test with close captioning which, in North America at least, uses white on black. Another side experiment could be conducted to see if it has any impact on how much attention the show receives (if white on black is simply better at retaining attention than black on white).

While this only tests a certain setting where subjects are used to seeing black and white it would provide what you are looking for. A long term experiment where people recognize black on white as the norm would be difficult to carry out.