FEATURES OF THE BRAIN SYSTEMS FUNCTION (EXPERT OPINIONS)
|
|
Brain-Style Information Systems Research Group
The brain is an astonishingly-complex system in which a large number of neurons process information via their emergent dynamic interactions. This research group aims to elucidate theoretically the underlying principles of information processing within the brain and to implement these principles using modern technologies. More specifically, our targets include mathematical studies of the principles and mechanisms of information creation in the brain, research on flexible heterogeneous systems that are open to the real world, and computational neuroscience focusing on modeling the mechanisms of the human brain.
1. Information Synthesis
----- Lab. head Shunichi Amari -----The mind has a memory system in which an image of the world is systematically stored. Rather than passively storing this information, the brain organizes the world image actively based on past experiences and synthesizes new information based on experience. The brain recognizes the outer world, obtains new ideas, plans motor commands to realize one's intentions, and so on. All of these govern information synthesis and creation in the inner world of the brain.
Our laboratory aims to elucidate the fundamental principles of information creation in the brain by mathematical and information-theoretic means. To this end, it is necessary to study how information is encoded in the brain and to determine the capabilities of dynamic interactions of neurons in this representation. Information is encoded not merely in the firing rates of neurons but also in the spatio-temporal patterns of excitation, including higher-order correlations, across groups of neurons. Information geometry plays a fundamental role in the study of such spatio-temporal stochastic patterns.
Primitive visual features such as contours of object images and brightness contrast, extracted in the primary visual cortex, are gradually integrated through feedforward connections to higher cortical areas in order to organize visual percepts. This process for perception is also modulated by activities transmitted through feedback connections. This laboratory is engaged in experimental and theoretical studies of columnar organization in the visual cortex as neural substrates for visual information representation and processing. Using electrophysiological and optical imaging methods, we investigate spatial patterns of columns, their mutual connections, and the influence of visual experience on the development of columns. Based on findings in these experimental studies, it is also attempted to reproduce columnar structure using the self-organization theory, and to build computational models of cortical circuits to obtain a better understanding of perception.