Retinal input that is suppressed from visual awareness can nevertheless produce measurable aftereffects, revealing neural processes that do not directly result in a conscious percept. We here report that the face identity-specific aftereffect requires a visible face; it is effectively cancelled by binocular suppression or by inattentional blindness of the inducing face. Conversely, the same suppression does not interfere with the orientation-specific aftereffect. Thus, the competition between incompatible or interfering visual inputs to reach awareness is resolved before those aspects of information that are exploited in face identification are processed. We also found that the face aftereffect remained intact when the visual distracters in the inattention experiment were replaced with auditory distracters. Thus, cross-modal or cognitive interference that does not affect the visibility of the face does not interfere with the face aftereffect. We conclude that adaptation to face identity depends on seeing the face.