In the philosophical and psychological debate on self-deception there is still a problem, that is both conceptual and empirical, in need of being accommodated into the framework of the major theories on offer. This is the so called «selectivity problem». Self-deception shows to be «selective» in that not all the subjects who are in the grip of a strong motivation to believe a certain proposition end up self-deceiving. That is, motivation to believe a proposition cannot can't be the whole causal story about why self-deception takes place, given that selfdeception is still selective and affects some motivated-to-believe subjects but not others. In this article, I offer a transcendental argument aimed at offering a working hypothesis on the causes of selectivity, and to this end I make a case for the epistemic character of the subject and the epistemic values she endorses as decisive to determine whether one subject ends up in self-deception or else in self-verification.