Modeling nonlinear chaotic dynamics and temporospatial integration in sensory processing of brains [odor perception]

Brains provide an inexhaustible source of new concepts for the development of novel and more effective models for information processing. A major problem is that brains are noncomputational, in that they do not contain or process numbers, but instead operate with electrochemical fluxes of ions and organic molecules. The metabolic energy is provided by neurons, which are not like transistors, in that they require continuous chaotic activity to survive. Properties on which brain function is based include capacities continually to construct patterns from chaotic dynamics, to build a broad knowledge base by Hebbian learning, and to change the base rapidly when global adaptation is required. The author emphasizes the hierarchical nature of brain architecture, that must be incorporated in future brain models to give them the power that brains display, and the author suggests that analog-digital hybrids be used