ONE OR SEVERAL SEMANTIC SYSTEM(S)? MAYBE NONE: EVIDENCE FROM A CASE STUDY OF MODALITY AND CATEGORY-SPECIFIC “SEMANTIC” IMPAIRMENT S. Carbonnel 1 , A. Charnallet 2, 3 , D. David 2 and J. Pellat 2, 3 ( l Laboratoire de Psychologie Expérimentale, Chambéry, Université de Savoie, France; 2 Service de Neurologie et Neuropsychologie, CHRU de Grenoble, France; 3 Laboratoire de Psychologie Expérimentale, Université de Grenoble, France) ABSTRACT Following cerebral anoxia, EC, a 55-year-old patient, exhibited a severe and clear-cut pattern of semantic impairments without general intellectual deficit or perceptual difficulty. EC demonstrated a complex neuropsychological picture including a massive visual agnosia and a complete lack of imagery, both of which involved all categories of objects (living and non living) and a category-specific word comprehension deficit limited to animal names. Findings are discussed in the light of the theoretical frameworks currently available in the area of neuropsychology. It is argued that neither the single nor the multiple view of semantics fully succeed in providing a satisfactory account of the data and a tentative interpretation of the whole pattern of impairment is proposed in the general framework of non abstractive conceptions of meaning. INTRODUCTION The issue of how meaning develops and how knowledge is evoked at any moment is probably one of the most fascinating in the area of cognitive sciences. Two fields of research are likely to play a privileged role in improving knowledge in this area: developmental psychology and neuropsychology. Developmental psychology provides insights about the essential question of how meaning is gradually acquired (see for example Nelson, 1985). It may be noted that the extensive field of connectionist modelling is of particular relevance here, since most connectionist networks are trained to learn the function they have to simulate (see Plaut and Shallice, 1993, for a recent example). With respect to neuropsychology, studies of object or word comprehension disorders are of particular interest, since the nature of the observed deficits and their possible selectivity can strongly constrain any theory of meaning. It is therefore not surprising that the pioneering works of Warrigton and co-workers (Warrington, 1975; Warrington and Shallice, 1979) on the various patterns of dissociation between word and picture comprehension have given rise to a stimulating debate which is still very much alive. These authors (see also Beauvois, 1982; Shallice, 1987) suggested that their results could only be accounted for by assuming that pictures and words do not give access to the same knowledge base (the multiple semantics view). This claim has been strongly challenged by the proponents of the unitary view of semantics (Caramazza, Hillis, Rapp et al., 1990; Riddoch, Cortex, (1997) 33, 391-417