You have to be careful about the studies on depressive realism, though; a lot of them are poorly designed. I remember reading about one study (and I'm not 100% on the details) where subjects basically interacted with a machine that flashed lights or something at random, and the depressed ones were more likely to come to the realization that their actions had no effect on the machine. The flaw is, if you design a scenario where people's actions really do have no effect on the outcome, of course you're going to discover that depressed people have a more accurate perception of reality--not because they actually do, but because you've put them in a situation that corresponds to their cognitive bias.
