Prosopagnosia is a neurological syndrome in which patients cannot recognize faces. Kecently it has been shown that some prosopagnosics give evidence of "covert" recognition: they show greater autonomic responses to familiar faces than to unfamiliar ones, and respond differently to familiar faces in learning and interference tasks. Although some patients do not show covert recognition, this has usually been attributed to an "apperceptive" deficit that impairs perceptual analysis of the input. The implication is that prosopagnosia is a deficit in access to, or awareness of, memories of faces: the inducing brain injury does not destroy the memories themselves. We present a case study that challenges this view. LH suffers from prosopagnosia as the result of a closed head injury. He cannot recognize familiar faces or report that they are familiar, nor answer questions about the faces from memory, though he can (1) recognize common objects and subtly varying shapes, (2) match faces while ignoring irrelevant information such as emotional expression or angle of view, (3) recognize sex, age, and like-ability from faces, and (4) recognize people by a number of nonfacial channels. The only other categories of shapes that he has marked trouble recognizing are animals and emotional expressions, though even these impairments were not as severe as the one for faces. Three measures (sympathetic skin response, pupil dilation, and learning correct and incorrect names of faces) failed to show any signs of covert face recognition in LH, though the measures were sensitive enough to reflect autonomic reactions in LH to stimuli other than faces, and face familiarity in normal controls. Thus prosopagnosia cannot always be attributed to a mere absence of awareness (i.e., preserved information about faces whose output is disconnected from conscious cognitive processing), to an apperceptive deficit (i.e., preserved information about faces that cannot be accessed due to improperly analyzed perceptual input), or to an inability to recognize complex or subtly varying shapes (i.e., loss or degradation of shape memory in general). We conclude that it is possible for brain injury to eliminate the storage of information about familiar faces and certain related shapes.

In the primate visual system, the identification of objects and the processing of spatial information are accomplished by different cortical pathways. The computational properties of this “two-systems” design were explored by constructing simplifying connectionist models. The models were designed to simultaneously classify and locate shapes that could appear in multiple positions in a matrix, and the ease of forming representations of the two kinds of information was measured. Some networks were designed so that all hidden nodes projected to all output nodes, whereas others had the hidden nodes split into two groups, with some projecting to the output nodes that registered shape identity and the remainder projecting to the output nodes that registered location. The simulations revealed that splitting processing into separate streams for identifying and locating a shape led to better performance only under some circumstances. Provided that enough computational resources were available in both streams, split networks were able to develop more efficient internal representations, as revealed by detailed analyses of the patterns of weights between connections.