Young children do not form representations of newly encountered faces as efficiently as do adults. A first step in explaining this difference, like any age-related change, is locating its source. A major source of the improvement is acquisition of knowledge of faces per se, as opposed to age-related changes in general pattern encoding or memorial skills. Two consequences of expertise at individualizing members of classes that share a basic configuration are known: a large inversion effect and a caricature advantage. It is possible that both of these effects reflect increased reliance, with expertise, on configuration distinguishing features. Several phenomena that indicate that inversion interferes with the encoding of configural aspects of faces are reviewed. Finally, developmental data are presented that confirm the suspicion that there are at least two distinct sources of the vulnerability of face encoding to inversion, perhaps reflecting two distinct senses of `configural encoding' of faces, only one of which is implicated in adult expertise at face encoding.