I fear if I think too much about this conundrum I'll soon have blue smoke coming out my ears, like a robot in Fat Trek tricked into self-destruction by pondering a paradox in its programming.
I must wonder, though: is it true that depressives are more realistic than others? Is that in comparison with "healthy" people or only with the manic phase of manic depression? De Sousa doesn't make it clear and although he mentions empirical evidence, he doesn't cite it.