Reflecting current neurophysiological findings, Nelson (this issue) proposes that some inferotemporal cortical areas have the potential for specialization for face-identification, and that these areas may use a critical period of sensitivity in infancy and childhood to become tuned to (the particularities of) faces that they will encounter. This 'revisionist' modularity stance takes account of developmental process in the manner suggested by other developmental theorists (Karmiloff-Smith, 1998). Face processing is modular because the skill we develop remains face-specific. Tuning (i.e. development), within this domain, is to particular dimensions of faces (inner parts, upright orientation), and to particular processing styles (configurational and holistic, rather than piecemeal), and results in the ability to discriminate faces, despite local variation in image quality and global variation in viewpoint and lighting. This approach respects and systematizes the data, and is productive of interesting research questions. One such question is how is the face-identification system set up and sustained in the developing infant? Is it 'just there' -a psychological tool that has evolved 'all on its own'? Evolutionary psychologists would wish to argue this, and the very rare cases of developmental and early-onset prosopagnosia could support this. However, much more common are delays or deficits in attaining face expertise in people with otherwise very varied cognitive styles and skills. This leads us to enquire whether there are developmental psychological drivers involved in tuning face-identification skills towards the adult style and proficiency: intrinsic needs and extrinsic skills which are the precursers of face processing skill, and which moderate its development.
Our evidence and speculations for the existence of two such drivers come from three developmentally anomalous groups who show qualitative and/or quantitative problems in processing faces. 'Failure of the face processing module' does not seem, to us, to be an adequate explanation for these youngsters' debilities: rather, one or both drivers of normal face processing may not be working efficiently. It is these distal causes that curtail the development of face processing in these groups and, by implication, enables the development of expertise in the face processing module in other children. The drivers we propose are