In the current study we describe J.M., a 15-year-old boy with a history of congenital brain abnormalities and concomitant visual-processing impairments. J.M.'s most prominent deficit is his impaired face recognition, but formal testing also revealed deficits in other domains of visual processing. One aspect that emerged from J.M.'s visual-processing assessment was a tendency to focus on local features and to rely on them for the encoding and identification of visual stimuli including geometric figures, objects, words, and inverted faces. In spite of this general tendency, he was impaired on tasks requiring the encoding of local features in upright faces. Moreover, his ability to distinguish between features in upright faces was significantly worse than that for inverted faces, the opposite pattern to that typically found in normal participants. What is it that keeps J.M. from applying his otherwise intact featurebased processing to upright faces? As proposed in previous reports of face inversion superiority in individuals with acquired face recognition impairments, we suggest that J.M.'s "inverted-face inversion effect" speaks for a specialized cognitive system that is mandatorily engaged by upright (but not inverted) faces, even when it is impaired and therefore maladaptive. In addition, since J.M. suffered from congenital brain abnormalities affecting the normal development of his face-processing skills, his performance suggests that specialized and mandatorily activated face-processing mechanisms are not entirely experience dependent, and that they can remain modular during development even if they don't function properly and are therefore maladaptive.