KAWATO Dynamic Brain Project

Project Director: Dr. Mitsuo Kawato, ATR Human Information Processing Research Laboratory

Creative researchers are invited to apply from the fields of

The word "cerebellum" is Latin and means "little brain"; it is a diminutive of "cerebrum" which means "brain." This well describes the status of the cerebellum in the field of neuroscience. Compton*s Interactive Encyclopedia completes its description of the cerebellum by stating simply that it is below the cerebrum, behind the brain stem and controls balance and coordination. The intelligence that makes us human is considered to be the almost exclusive domain of the much larger cerebrum; the cerebellum is the center for more "primitive" brain functions.

This anthropocentric view of the cerebellum may change significantly by the time ERATO's Kawato Dynamic Brain Project is completed five years from now. This project will add research on computational theory, algorithms and expressions to the traditional hardware-focused approaches of neurophysiology, neuroanatomy and molecular neurophysiology to research sensory motor learning in the brain with particular attention to the cerebellum. Its ultimate goal is a better understanding of the higher functions of the brain - consciousness, thinking and language.

This integrated software-hardware approach to the brain has been most successful in studies of sensory input - motor output functions. The Kawato Project will use sen-sory motor functions as the basis for its entree into the research of higher brain functions. Research will focus on learning in sensory motor systems - the next higher brain function. Typical sensory motor tasks are the movement of the hand to a target position and the tracking of a target spot by the eye. When these tasks are repeated, in general, the respective hand or eye movements become faster, smoother and more efficient, in other words, learning takes place. Sensory motor learning can also result from visual observation and mimicry by the subject. Researchers will not only experimentally observe the neurophysiological changes involved in this learning, they will also construction computational models and carry out computational simulations of this learning.

A key principle of the Kawato Project can be stated, "If we cannot reproduce our understanding of how the brain works in the form of a robot, we cannot claim to understand how the brain works." An important part of the project will be the construction and programming of humanoid robots that can incorporate the computational models and algorithms developed to explain the sensory motor learning of experimental subjects.

In the past, the "primitive" motor control functions of the cerebellum were thought to be unrelated to the higher intelligence functions of the cerebrum. Recent research has found that motor control is intimately coupled to speech and handwriting recognition. The Kawato Dynamic Brain Project expects to significantly extend our understanding of learning and the cerebellum. Who knows? We may come to view the cerebrum as an appendage of the cerebellum.

ERATO's Kawato Dynamic Brain Project will begin October 1, 1996 and operate for five years with a total budget of approximately US$ 15 million.

Dr. Mitsuo Kawato graduated from the University of Tokyo Faculty of Science Laboratory of Physics in 1976 and received his Doctor of Engineering from Osaka University Graduate School Faculty of Basic Engineering in 1981. He was a member of the faculty of Osaka University Faculty of Basic Engineering until joining ATR*s Visual Perception Research Laboratory in 1988. Dr Kawato became Laboratory Head in ATR*s Human Information Processing Research Laboratories in 1992. He holds joint appointments as Visiting Professor at Hokkaido University Research Institute for Electronic Science (since 1992), University of Genova Laboratory of Robotics (since 1993) and Kanazawa Institute of Technology (since 1994). Dr. Kawato received the Toyama Science & Technology Award in 1986, the Yonezawa Founders' Medal in 1991, the International Neural Network Society Outstanding Research Award in 1992, the Minister of Science and Technology Research Success Award in 1993, the 11th Osaka Science Prize in 1993 and the 10th Tsukahara Memorial Prize in 1996.

"Neuroscience: Tilting Against a Major Theory of Movement Control", Science, Vol. 272, No.5258, pp.32-33 (1996) (Research News)

H. Miyamoto et al., "A Kendama Learning Robot Based on Dynamic Optimization Theory," Neural Networks, in press (1996)

Hiroaki Gomi and Mitsuo Kawato, "Equilibrium-Point Control Hypothesis Examined by Measured Arm-Stiffness During Multi-Joint Movement," Science, Vol.272, pp.117-120 (1996)

Hiroshi Imamizu, Yoji Uno and Mitsuo Kawato, "Internal Representations of the Motor Apparatus: Implications From Generalization in Visuomotor Learning," Journal of Experimental Psychology: Human Perception and Performance, Vol.21, No.5, pp.1174-1198 (1995)

Yasuhiro Wada and Mitsuo Kawato, "A Theory for Cursive Handwriting Based on the Minimization Principle," Biological Cybernetics, Vol.73, pp.3-13 (1995)

Yasuhiro Wada, et al., "A Computational Theory for Movement Pattern Recognition Based on Optimal Movement Pattern Generation," Biological Cybernetics, Vol.73, pp.15-25 (1995)

Frank E. Pollock, et al., "Perceived Motion in Structure from Motion: Pointing Responses to the Axis of Rotation," Perception & Psychophysics, Vol.56, No.1, pp.91-109 (1994)

Kawato, H. Hayakawa and T. Inui, "A Forward-Inverse Optics Model of Reciprocal Connections Between Visual Areas," Netword: Computation in Neural Systems, Vol.4, pp.415-422 (1993)

Shidara et al., "Inverse-Dynamics Model Eye Movement Control by Purkinje Cells in the Cerebellum," Nature, Vol.365, pp.50-52 (1993)

Mitsuo Kawato and Hiroaki Gomi, "The Cerebellum and VOR/OKR Learning Models," Trends in Neurosciences, Vol.15, pp.445-453 (1992)

Mitsuo Kawato and Hiroaki Gomi, "A Computational Model of Four Regions of the Cerebellum Based on Feedback-Error-Learning," Biological Cybernetics, Vol.68, pp.95-103 (1992)

M. Kawato, K. Furukawa and R. Suzuki, "A Hierar-chical Neural-Network Model for Control and Learning of Voluntary Movement," Biological Cybernetics, Vol.57, pp.169-185 (1987)