BioCaen : A Causal Qualitative Network for Cerebral Information Propagation Modeling

Functional brain mapping studies in humans may show contradictory results, as no one to one correspondence can be found between activated cerebral zones and cognitive functions . An explanation could be the networked physical organization of brain zones and the information propagation mechanisms through the network. As we focus on language-related brain subsystems, Al models are the single alternative to animal models . The brain being considered here as a physical, rather ill-defined system, AI qualitative approaches, especially causal methods, fit perfectly our purposes . The major constraint in the approach, i .e . the fact that phenomena related to the system's functioning must be time-ordered, is compatible with our knowledge on brain behavior . This paper presents a tentative two-level model of brain information propagation mechanisms . At the structural level, the brain anatomical structure is represented as a component network whose nodes are cerebral zones connected by propagating or inhibiting anatomical links (axon bundles) . At the functional/behavioral level, each zone is modeled by a causal qualitative network instanciated from a generic model . A component/connection approach derives the global functional model corresponding to a structural network-from the above models . As models must constantly evolve with new hypotheses and findings in brain research, we propose a flexible hypothesis simulator », BioCaen, for implementing them. BioCaen is an offspring of Ca-En (Bousson & Trave-Massuyes, 1993, 1994) that extends its capabilities by : (1) dealing with components, (2) giving more flexibility to time-variation expressions, (3) coding causal network nodes by couples of variables .