Pour surmonter les limites des modèles neurocognitifs de la schizophrénie, nous proposons d'utiliser les développements récents de la théorie de la représentation. Nous esquissons un cadre conceptuel permettant de construire la notion d' espace de représentation ( ie d'univers subjectif déployable à partir d'une mémoire) et nous redéfinissons dans ce cadre les notions clés des syndromes paranoïdes : automatisme mental, délire, dissociation, autisme. Les schizophrénies apparaissent alors comme les cas-limites de fragmentation dissociative de l'espace de représentation. Nous illustrons cette approche en utilisant le Test de Saisie Visuelle Parallèle qui évalue le traitement symbolique d'informations visuelles interférentes. Les résultats de 12 sujets schizophrènes, stabilisés cliniquement, suggèrent la présence de zones de mémoire parasites liées à la fragmentation. L'analyse des données individuelles met en évidence l'équilibre dynamique entre l'activité mnésique interférente et l'émoussement émotionnel défensif.

Existent neurocognitive models of schizophrenia converge towards a core of impairments involving working memory, context processing, action planning, controlled and intentional processing. However, the emergence of this core remains itself difficult to explain and more specific hypotheses do not explain the heterogeneity of schizophrenia. To overcome these limits, we propose a new paradigm based on representational theory from cognitive science. Some recent developments of this theory enable us to describe a subjective universe as a representational space which is displayed from memory. We outline a conceptual framework to construct such a representational space from analogical representations that can be activated in working memory and are connected to a network of symbolic structures. These connections are notably made through an analytic process of the analogical fragments, which involves the attentional focus. This framework allows us to define rigorously some defense processes in response to traumatic tensions that are expressed on the representational space. The fragmentation of representational space is a consequence of a defensive denial based on an impairment of the analytic process. The fragmentation forms some parasitic areas in memory which are excluded from the main part of the representational space and disturb information processing. The key clinical concepts of paranoid syndromes can be defined in this conceptual framework : mental automatism, delusional intuition, acute destructuration, psychotic dissociation, and autistic withdrawal. We show that these syndromes imply each other, which in return increases the fragmentation of the representational space. Some new concepts emerge naturally in this framework, such as the concept of « suture » which is defined as a link between a parasitic area and the main representational space. Schizophrenia appears as a borderline case of fragmentation of the representational space. This conceptual framework is compatible with numerous etiological factors. Multiple clinical forms can be differentiated in accordance with the persistence of parasitic areas, the degree of fragmentation, and the formation of sutures. We use this approach to account for an empirical study concerning the analysis of analogical representations in schizophrenia. We used the Parallel Visual Information Processing Test (PVIPT) which assesses the analysis of interfering visual information. Subjects were asked to connect several small geometric figures printed on a transparency. The transparency was displayed above four photographs which were the interfering material. Then, subjects completed three tasks concerning the photographs : a recognition task, a recall task, and an affective qualification task. Using a case-by-case study, this test allows us to access the defense processes of the subjects, which is not possible with the usual methods in cognitive psychopathology. Twelve clinically-stable schizophrenic subjects participated in the study which also included a self-assessment of alexithymia by the Toronto Alexithymia Scale. We obtained 2 main results : (a) creation of items in recall or false recognition by 8 subjects, and (b) lack of the usual negative correlations between the alexithymia score and the recall, recognition and affective qualification scores in the PVIPT. These 2 results contrast with what has been previously observed for alexithymia using the same methodology. The result (a) confirms an interfering activation in schizophrenic memory, which can be interpreted in our framework as indicative of parasitic areas. The creation of items suggests the formation of sutures between the semantic content of photographs and some delusional fragments. The result (b) suggests that the apparent alexithymia in schizophrenia is a defense against interfering activation in parasitic areas. We underline the interest of individual protocols to exhibit the dynamic interplay between an interfering activity in memory and a defensive flattening of affects.