Center for Information and Neural Networks

National Institute of Information and Communications Technology (NICT) Osaka University Advanced Telecommunications Research Institute International (ATR)

http://cinet.jp

Planning Ofce 7th oor, Nano Biology Building, Graduate School of Frontier Biosciences, Osaka University, 1-3 Yamadaoka, Suita City, Osaka, 565-0871 E-mail : [email protected]

Ibaraki country club

Kitasenri Station

Handai-Byoin-Mae Planning Ofce Station

Osaka University Hospital

Hankyu Senri Line Osaka Monorail

National Museum

Koen-Higashi- of Ethnology guchi Station

Senri Expo Park Suita I.C. Chugoku Suita Unobe Station Banpaku-Kinen-Kouen Station Senri-Chuo Station Chugoku Highway Yamada Station Suita Kita-Osaka kyuko Osaka Monorail

Monorail Bus Train Approximately 5 minutes on foot Approximately 1 minute on foot from “Handai Igakubu-mae” stop Approximately 30 minutes on from “Handai Byoin-mae” station, Hankyu Bus: From Senri Chuo subway station for “Handai Hombu-mae” or foot to the east of Kita Senri Osaka Monorail Saito Line “Ibaraki Mihogaoka” station, Hankyu Senri Line Kintetsu Bus: From Hankyu Ibarakishi station for “Handai Hombu-mae” (via JR Ibaraki station)

2011.12.26 Center for Information and Neural Networks CiNet Mission From Data Communications to Brain Communications

The development of ICT (information and Brain-inspired Intelligence communications technology) has led to fast and Systems massive data communications networks, resulting in an unprecedented impact on our lives. However, each year the volume and speed in which data are processed increase by over 50%, which risks compromising information transmission due to network congestion, unstable connections, and excessive energy use. Furthermore, users can be overwhelmed by the HHS abundant information. Therefore, constant Heart to Heart Science research is needed to sustain these ever-growing demands.

One example of research focuses on organic information processing network systems like Intelligent Innovative those inside the brain and biological cells. These networks function with an adaptability, Robotic Autonomous autonomy, and low-energy consumption that is not seen in present ICT. As such, these systems Networks Networks may make a new paradigm for future ICT.

At CiNet, we are studying the information and neural network strategies used by living systems, Instrumental including high-level brain function, and implementing these ƒndings to artiƒcial Technologies information and communications networks. Brain Imaging Techniques BFI BMI Network Brain-Machine Interface Brain-Function Installed Information Network Applications to Nursing Care, Problem-Solving Rehabilitation, Healthcare and Welfare Networks

Information and Neural Networks: A Council’s Report

The “Working Group on Global Issues” panel for the “Task Force solutions require human- and earth-friendly technologies. The communications networks that are robust to breakdowns in a system. at the link below, CiNet was inaugurated in the ƒscal year of 2011. Concerning ICT Policies in the Global Age” conference hosted by the combination of brain research and ICT can be used for this purpose. For these and other purposes, the Internal Affairs and To read the conclusions, please go to: Ministry of Internal Affairs and Communications gave discussion on In response, the ministry held the “Council Concerning Brain and ICT” Communications Minister, Vice Minister, and Parliamentary Secretary “Final Summary: Council Concerning Brain and ICT” how to overcome the severe local and global concerns faced by conference, which identiƒed priority issues. The list was long and sat down with experts, listening on how to solve these and other http://www.soumu.go.jp/menu_news/s-news/01tsushin03/01000017.html various countries around the world, with special focus on wide, including technologies that aid the physically challenged and problems by designing a research program that focused on brain environmental and medical issues. In particular, the council stated elderly, have low energy demands, and information and research and ICT. In response to their conclusions, which can be seen

01 Center for Information and Neural Networks CiNet Mission 02 Center for Information and Neural Networks Message Information and Communication of the Future

Increasing the collaborative work between neuroscientists, engineers, psychologists, and cognitive scientists will help realize more environmentally and user-friendly technologies

Special advisor

Current times have seen an unprecedented Seiji Ogawa growth in productivity and innovation, in large part due to ICT, which has been instrumental in our ability to work with large volumes of data at Management Staff high speed. However, the volumes and speeds of data processing continue to increase, such that Director General without innovations they will become unsustain- Toshio Yanagida ably large and complex. Therefore, the networks responsible for these tasks need to be constantly upgraded. However, this comes at the cost of Vice Director General Kazuhiro Oiwa more energy. HHS Director General, Advanced ICT Research Institute, National Institute of To resolve this problem, we are considering alter- Information and Communications Technology (NICT) native strategies. The brain, for example, is always transmitting and processing new and Vice Director General increasing levels of information without Makoto Imase BFI Professor, Graduate School of Information Science and Technology, concurrent increases in its energy demands. Network Osaka University Research in this area, then, may offer unique and profound solutions to the continuous problem of processing more information. Our Vice Director General Mitsuo Kawato vision is to bring ICT scientists together with BMI Director, Brain Information Communication Research Laboratory Group, Advanced Telecommunications Research Institute International (ATR) people from the neurosciences, psychology, and the cognitive sciences to develop brain-based applications to ICT. Instrumental Vice Director General Such research is expected to realize more envi- Yoshichika Yoshioka Technologies Professor, Immunology Frontier Research Center, Osaka University ronmentally and user-friendly technologies. For example, new information processing and control technologies based on mechanisms used by the brain offer considerably lower energy demands than current models. They will also help us design user interfaces where one’s thoughts are directly transmitted to a machine this by developing non-invasive techniques that allow turned back to make even better ICT, which will or information terminal, which then carries out us to measure brain activity with unprecedented create a circle of discovery and innovation in both the corresponding action. accuracy in space and time. information and neural networks. At CiNet, we have already made great gains on The results from these brain studies will then be

Director General Toshio Yanagida

03 Center for Information and Neural Networks Message 04 Center for Information and Neural Networks HHS Heart to Heart Science Understanding and applying how the brain identi es the “heart” of a message

Introduction of research

Communicating only the true message: sending what we need to send Applying the mechanisms for how the brain processes Understanding social action in terms of brain information science and receiving only what we need to receive information to better communications When a human being communicates with The brain extracts meaning even if the sensory others, the manner of communication varies information obtained is incomplete. We study the between individuals. Our group researches brain mechanism that processes incomplete brain mechanisms related to interactions information by focusing on how it produces during social activities. We are using this information coding and voluntary action, and use information to identify how to create The ultimate goal of communication is to overwhelms an individual, making it this mechanism to design an intelligent information environments that optimize communication

convey an intended meaning that is increasingly difcult to recognize the system model that Brain ICT Laboratory, Advanced ICT for a given individual. correctly interpreted. Ironically, advances heart. Senior Researcher Brain ICT Laboratory, can autonomously Research Institute, NICT Advanced ICT Research Institute, NICT solve problems when Masahiko Haruno in communications technology like speed It is to conquer this challenge that HHS is Tsutomu Murata given ambiguous Cumulative and capacity make this goal more and conducting research and development to Differences among individuals inputs. frequency more of a challenge, because of the assist people in understanding the Brain activity, Social activity Gene during social activities Which do you choose? excessive amounts of data to be meaning of words and recognize the Theory of brain temperature brain structure model Self(yen) Other(yen) processed. content of information by scientically v=1/t =C exp(- ) M/S 1 100 100 Normally, a person extracts subjective analyzing the higher brain functions that M: Difculty of hidden picture (can be decided quantitatively) 2 110 60 meaning from an excess or ambiguous recognize, understand and react to a com- S: Cognitive temperature of brain = Power of inspiration V: Speed of inspiration t: Duration of inspiration 3 100 20 amount of information. For example, munication. In this way, HHS research log(Search time/s) consider a poem. Can the reader con- seeks to improve BMI technologies by Director General, Advanced ICT Research Institute, dently interpret the author’s intended clarifying how higher brain functions arise National Institute of Information and meaning? To varying degrees, this concern and applying this knowledge to Communications Technology (NICT) How the human vision system works Neural representation and information networks underlying perception always arises whenever our senses receive improving the quality of human commu- Kazuhiro Oiwa Vision plays a predominant role among the The visual system, one of the brain’s most information. In any form of communica- nications. ve senses in daily communications and sophisticated systems, is very important for tion, the individual receiving the data We envision heart-to-heart science behavior. The world of vision is lled with human communications, as it processes a must process the information and extract leading to a better, more creative society. colors, shapes, motions, depths and textures huge volume of information in a highly the intended meaning, or what we refer To do this, we are proactively working that are then reconstructed by the brain. We ˆexible manner. Using the latest engineering to as the “heart” of the message. with a wide range of researchers from the study how the miracle of “seeing things” is methods and measurement techniques, we realized only from the light information that study how complex neural networks process Combined with the act of transmitting neurosciences, biophysics, information Professor, Graduate School of Frontier is captured by the eyes. Professor, Graduate School of Frontier information on shape and depth acquired by this heart, we are aiming to optimize Biosciences, Osaka University Biosciences, Osaka University technology, psychology and philosophy. individual to generate an image. what we call “heart-to-heart communica- CiNet provides an ideal place for such Ichiro Fujita Izumi Ohzawa tion”. However, the amount of data being collaborations. processed by today’s networks often

Understanding the brain’s communication network Investigating how the brain conducts working memory and cognition

Interpersonal communication is supported by Working memory is the brain function for temporarily information and communication between storing information required to conduct a cognitive various areas in the brain. We are studying task. Using fMRI (functional magnetic resonance the mechanism used by the brain to process imaging), we are investigating the human working and communicate this information by applying memory system and its brain mechanism, particularly a wide range of methods like psychophysics, focusing on individual differences. In addition, we are brain activity measurement seeking factors that affect the brain network respon- Professor, Graduate School of Frontier Biosciences, tools and clinical studies. Professor, Graduate School of sible for working memory by investigating emotion, Osaka University Human Sciences, Osaka University self-monitoring and self-consciousness. I got it Shigeru Kitazawa Mariko Osaka Imperfect Information Working memory network

Posterior Parietal Cortex Understanding the heart of the information Anterior Cingulate Cortex PreFrontal Cortex (Osaka et al., 2003, 2004, 2007)

05 Center for Information and Neural Networks HHS 06 Center for Information and Neural Networks BFI Network Brain-Function Installed Information Network Building a Low Energy Information and Communications Network Based on Human Brain Function

Introduction of research

By understanding and applying the mechanisms used for brain function, we aim to Energy-ef­cient and robust information network Simple, energy-ef­cient network control using uctuations develop computer networks and information and communications networks that control based on neural and biological functions It has been found that in living organisms mimic the exible, robust and autonomous properties of the brain. We are researching the topology and the response to an external force self-organization of robust information corresponds to the scale of the ”uctuation in networks along with developing the system. This behaviour can be described What is the difference between the all it takes to turn on a human brain. energy-saving information network as a ”uctuating complex system by using an attractor perturbation model. We are human brain and a CPU, the brain of a We seek to understand how these brain techniques by analyzing the topological and hierarchical structures of the networks seen applying this model in the research and computer? The former consists of properties emerge and apply these Professor, Graduate School of Information development of simple and energy-ef–cient in neural and biological functions and Science and Technology, Osaka University approximately 14 billion brain cells, while revelations to the design of better Professor, Graduate School of Information applying these properties into information methods for effective network control. Science and Technology, Osaka University Naoki Wakamiya the latter is composed of approximately a computer networks. One particular networks. Masayuki Murata End-to-end rate control billion transistors. While the clock speed property of interest in the brain is the Large Neural and Fluctuation – a measurement of the speed of a property of ”uctuation, which is biological networks Information network External forces ？ computer - of a human brain is approxi- commonly seen in many biological Extracting similarities and differences mately 100 Hz, that of a CPU is over- systems. Unlike most arti–cial systems, ？ whelmingly faster at several GHz, more brain activity is actually quite noisy Energy-saving and robust Small External ？ Fluctuation Forces than 10 million times that of the brain. (”uctuate). These ”uctuations are information networks ？ Furthermore, computers can be thought important for the brain to be Multihoming Professor, Graduate School of Information Response size (variation) connected together to increase their adaptive to its ever changing environ- Science and Technology, Osaka University processing power by even greater mental. Preliminary work has indicated Makoto Imase magnitudes. Consider the more than 700 that incorporating such ”uctuations into Modern ICT system design based on the brain Robust information network control based on brain dynamics million computers now online and how a computer network has similar bene–ts. We study brain functional networks by using much more powerful they are than when In the BFI (brain-function installed infor- Power consumption has become a serious constraint in systems, both at the device and theoretical models with particular focus on working as isolated units. mation) network component of CiNet, circuit level as well as in communications internal ”uctuations. The topological features Yet at the same time computer networks we strive to push this ”uctuation networks. The brain deals with such of these networks are analyzed with complex are inferior to the human brain in many research and its application to computer problems innovatively, by proactively network theory and an attractor perturbation ways. For example, the brain is much networks by expanding into other utilizing noise and ”uctuations rather than model to describe the effects of any external more ”exible than a computer, can solve unique information processing supressing them. Investigating and applying in”uences. From this research, we expect to Senior Researcher Brain ICT Laboratory, these mechanisms is expected to lead to Senior Researcher Brain ICT Laboratory, gain better insight into the brain’s dynamics ambiguous problems and has an imagi- properties seen in the brain to develop Advanced ICT Research Institute, NICT Advanced ICT Research Institute, NICT more ”exible and robust ICT systems. and to utilize this for the design of robust nation to which no computer can robust ICT. We believe this can be Ferdinand Peper Kenji Leibnitz control methods in future information compete. Less abstract, the brain is much realized at CiNet by “networking” the networks. more environmentally friendly than information sciences with brain science, fMRI Functional NW computer networks. In fact, while faster biology and sociology. Fluctuations in brain activity and more powerful networks usually

require disproportionate increases in Attractor perturbation energy, one additional watt of power is

Brain functional network

Information networks Arti­cial systems will consume 50% of the world’s Biological systems computing speed, integration Power consumption, amount of information, Robot control based on biological uctuations Human interfaces utilizing brain functions total electric-generating capacity by 2035 The human brain Cells 14 billion neurons Approximately 20,000 genes We are conducting research and development The human brain performs many functions of Approximately 50 trillion synapses (combined) on how biological ”uctuations can be used in which people are conscious, and others which No. of combinations to be controlled The fastest supercomputer No. of combinations to be controlled 1.0 × 106000 robot control, with the greater goal of take place unconsciously. We are researching how requires 30 megawatts – equal to the 1.0 × 1015 trillion designing robots that can do complex tasks to create interfaces that are natural and useful Increasing 1.5 to 2.0 x annually power used by all households on Energy consumption while remaining robust to an ever-changing for human beings in order to realize functions Awaji Island, Hyogo Prefecture Energy consumption Brain: A trillionth of a watt environment. Applying such theories, we aim that are useful for human consciousness to them Power consumption of routers as of 2010 Brain: 20 watts Such combinations would to design robots that resemble human beings by ef–ciently utilizing functions and Such combinations would 6000 Amount of information in the world (700 billion GB) require 1.0 × 10 Watts Professor, Graduate School of and behave like living organisms when exposed Professor, Graduate School of Information characteristics we are not aware of. Computing speed 15 Engineering Science, Osaka University Science and Technology, Osaka University require 1.0 × 10 trillion Watts in an arti­cial machine to dynamic and different environments. Poor robustness makes in an arti–cial machine Integration of LSI (Moore’s Law) Hiroshi Ishiguro Taro Maeda Intention Operator Realized movement Robot them vulnerable to accidents Movement of robot : z 2000 2010 2020 or sudden change Robust and energy ef–cient A system that can act out Control interface Operational movement : y an “intention” Advanced technologies Intention (operator) Intention (robot) Biological principles Segmentation of action: x Control interface Segmentation of action: x’ Operational movement : y Fｐ Fm Gc Gr Robot that can Perception of Generation of Learning Control of segmented movement movement controller movement imitate arm (Segmentation) (Continuation) Realized movement (Segmentation) (Continuation) movement Movement of robot : z Design of robust, energy ef­cient information systems Tracking movement of a robot that searches using Android → → → → → → using biological principles biological ”uctuations Normal voluntary operation: x y x’ z Derive intention: x’ z x y

07 Center for Information and Neural Networks BFI 08 Center for Information and Neural Networks BMI Brain-Machine Interface Technology

Developing Brain-Machine Interfaces for Human Care and Advanced Communications

Introduction of research

Applying how the brain processes information for better human care devices and technologies and advanced communication networks. Investigating the mechanisms for cognition and motor control and applying them to BMI

Technology that can extract information from the brain and then be applied to computer process is fundamental for brain-machine interfaces. We are investigating where, how and when information for cognition and motor control exists in the With its low birthrate and greying decoding information from one brain brain, and developing technologies that can extract this population, Japan is a nation aging faster (decipher and extract) and encoding it to information in an ef‰cient manner. The ultimate goal is to apply than most. This will lead to a large another (code and approach). Associate Director, Brain ICT Laboratory, these technologies to the real-time operation of a computer or Advanced ICT Research Institute, NICT expansion in health care, as it has been Currently, our best understanding of robot and to improving the cognitive abilities of human beings Technology that can smoothly reconstruct the motion of a Hiroshi Imamizu ‰nger by analyzing brain activity. Left: Three-dimensional estimated that by 2025, Japan will have 3 how the brain processes information is by themselves. view. Right: Projection view onto a two-dimensional plane million stroke patients to care for, with at looking at temporal correlations (Red: Reproduced from brain activity. Blue: Actual motions) least one million having chronic mobility between the brain activity recorded and and communication problems. Addition- the information manipulated in the ally, Japan has an alarming number of experiment ranging from sensory stimuli, suicides, having reached 30,000 annually motor acts and cognitive states. By Using BMI to enhance brain function and environmental adaptability mostly due to mental disorders and developing entirely innovative method to stress. This group too will require a great directly induce spatio-temporal patterns The wish for a healthy and active life is common to all humankind, and is expected to be even more so as societies become older. Recent progress in brain deal of personal attention to prevent of brain activity, we could directly Director, Brain Information Communication science is making it possible to develop new technologies that support this by such tragic acts. Because of factors like examine the cause and effect relation- Research Laboratory Group, Advanced Telecommunications Research Institute International (ATR) facilitating a number of human activities that become more dif‰cult with age. these, a third of Japan’s workforce is ships between brain activity and mind. In this project, we are developing new technologies for understanding the basis estimated to become caregivers for the This requires a very organic approach Mitsuo Kawato of neural activity by researching techniques that can exploit the brain’s plasticity disabled and the elderly by 2030. This is that combines the neurosciences, Research Manager, Brain ICT Laboratory, to enhance brain function, use BMI to expand human capability, and produce far too high a proportion for any biophysics, and information technology Advanced ICT Research Institute, NICT robots that can respond to their environments in anthropomorphic ways. As country, therefore requiring alternative along with new measuring devices Eiichi Naito these methods develop, we will then apply these techniques to understanding care-giving options including technolo- enabling BMI technology to revolutionize how changes in neural behaviour coordinate with changes in thoughts. gies and services that can replace the healthcare and welfare, while at the human component. same time advancing our understanding As such, we are focusing on brain - of the neurosciences and information machine interface (BMI) that can achieve and communication sciences. CiNet Supporting handicapped people with advanced ICT – Motor BMI sophisticated human care by studying provides the best environment for such a and applying the neurosciences. For project. In our project, advanced ICT is applied to infer a person’s intent Wireless BMI system Intracranial example, we are designing robots that by decoding his or her brain signals and using this information electrodes LSI, casing Wireless can be controlled by the brain signals of to drive various external devices that are used to support the communication system person with a given task. For example, we anticipate being able a user to aid in simple tasks during body area network nursing care. to help even the most severely disabled people who have lost analog LSI ampli‰er most of their motor and speech function by implementing Implantable communication In addition, we are investigating ways to device and battery robotic arms or electrical equipment that can be activated by use BMI with robot technology in infor- Professor, Graduate School of Medicine, Osaka University the user’s thoughts. Such therapeutic devices will be invaluable External devices mation and communication technologies. ・ Computer Toshiki Yoshimine in giving a disabled person more independence, with the ・ Robot The purpose here is to develop science ・ Electrical equipments greater aim of applying them to a clinical setting. ‰ction like communications technology where a machine can intuit a user’s feelings and thoughts in way people do when interacting with one another. Current noninvasive BMI technologies in Developing next-generation arti€cial retina – Sensory BMI Decoder Stimulation Japan are the most advanced in the electrode world, and are already being applied. Walk signal We are developing original arti‰cial retina based on BMI technology, Receiver coil Furthermore, BMI technology has the which would be safer and more sophisticated than current arti‰cial retinas potential to signi‰cantly alter our way of Image information (Fig. 1). Already, we found a totally blind patient was able to identify the viewing communications systems. One position of a chopstick when using our prototypes (Fig. 2). By increasing Fig. 1 ambitious goal is to use BMI technology the number of stimulating electrodes that can be used on a patient, our Visual rehabilitation in inter-personal communications by current goal is to restore reading and walking vision in blind patients. Professor, Department of Applied Furthermore, we are investigating the central mechanism the brain uses to Visual Science, Graduate School of Medicine, Osaka University recognize an object and developing an ef‰cient electrode stimulation Takashi Fujikado system for visual rehabilitation. Fig. 2

09 Center for Information and Neural Networks BMI 10 Center for Information and Neural Networks Instrumental Technologies Brain Imaging Techniques

Cutting-Edge Brain Function Measuring Techniques

Introduction of research

By developing and integrating a number of non-invasive techniques to measure brain

function, we seek to design new techniques that can help us apply brain processing Developing a magnetic resonance method to probe the brain BOLD strategies to information and communications systems. Professor, Immunology Frontier Research Center, Osaka University Yoshichika Yoshioka Extracting signals that represent brain function and activity Magnetic resonance is a unique method with which we can obtain information non-invasively at many hierarchical levels – molecular, cellular and whole body. We are using sophisticated magnetic non-BOLD Our understanding of the life sciences our ability to observe brain properties resonance scanners of 3.0 T and 7.0 T for human beings and 11.7 T for has dramatically improved thanks to that are currently impossible to extract. animals to observe never before seen properties both at the molecular- breakthroughs in measuring and imaging At the same time, the development of and individual-level that can measure and analyze more detailed brain Challenges in acquiring more techniques. In recent years noninvasive small and highly plastic brain function- functions including BOLD effects, molecular diffusion and cellular detailed information. methods, especially for measuring measuring devices in conjunction with metabolism. human brain activity, have become large and low-plasticity ones have the signicantly more sophisticated and potential to create new interfaces that effective due to their size-reduction, connects brain activity and information higher performance and operational terminals. The development of such speed. As a result, it has become possible measuring devices and systems is funda- Intra-cerebral neurotransmission networks: to measure faint and ne brain activity mental for the development of brain- Analyses with PET and MRI that was once not conventionally imaging techniques and an important Professor, Immunology Frontier measurable in real-time without placing part of combining information and Research Center, Osaka University Information is transmitted between neurons through the a great burden on the subject. The result neural network technologies. Yoshichika Yoshioka synthesis, release and receptor binding of neurotransmitters in a has been gradual clarication of the For this reason, we aim to establish next- synapse. It is possible to simultaneously measure the dynamics mechanisms and functions that drive generation imaging techniques including and receptor occupancy of labeled neurotransmitters and the brain activity. fMRI (functional magnetic resonance corresponding change in blood ow by using electric excitation Professor, Graduate School of Much of this has come from improve- imaging), phase-contrast cerebral blood eld 7-tesla MRI device, we will combine Medicine, Osaka University and an integrated PET/MRI system. With this technique, we are ments in both hardware and software, ow imaging, temperature function various kinds of imaging at the analyzing the intra-cerebral neurotransmissions by observing Jun Hatazawa Integrated PET/MRI system helping us realize unprecedented spatio- imaging, nerve ber function imaging, molecular, cellular and tissue levels to neurotransmitter, blood and energy metabolism behaviour. temporal resolution to detect more brainstem nucleus function imaging, and acquire high-denition information on details about the brain in less time. magnetic resonance spectroscopy biological function. This will require a This will only continue, as new non- imaging (MRSI) to acquire information highly interdisciplinary and comprehen- invasive methods with even better about brain function that can then be sive approach that investigates from the Studying brain network dynamics by Multiple types of spatio-temporal capabilities are used in ICT. In particular, by using a level of neural metabolism to high-level brain measurement Behavioral data data constantly being made. The result will be state-of-the-art super-high magnetic brain function. integrating multiple information sources

In order to better understand the function of the human brain, it is essential to study brain network dynamics that Deciphering brain signals fMRI fNIRS describe spatiotemporal changes in brain activity. To Method of integrating Functional magnetic resonance imaging Functional near-infrared spectroscopy multiple information sources ~From atomic and molecular levels to the entire brain~ achieve this, we are developing methods that can Statistical model Physical model integrate multiple information sources such as neural and Method of estimating Researcher, ATR Neural Dynamics Primary signal Information Analysis Laboratories behavioral data from multiple subjects as well as methods that can study brain dynamics including top-down physical Change in neural activity Okito Yamashita and bottom-up statistical models. Studying brain network dynamics ・Action potential ・Neurotransmitter (excitatory / inhibitory) Neurovascular Electroencephalography (EEG) Analysis method Brain activities during visual word processing Magnetoencephalography (MEG) Coupling Developing analysis tools for MEG Phonology MRS & PET MEG EEG MEG-measured brain activity and Magnetoencephalography Electroencephalography applying them to language function Secondary signal Tertiary signal We are measuring linguistic brain function to Vision evaluate how a human understands information. In Spatiotemporal Change in metabolism Change in hemodynamics Meaning order to identify spatiotemporal active sources in the brain activity ・Glucose metabolism ・Blood ow ・Blood volume brain, we are developing methods for analyzing MEG fMRI ・Oxygen metabolism Characters ・Blood oxygenation level Senior Researcher, Brain ICT Laboratory, and fMRI data, and using them to measure and model Advanced ICT Research Institute, NICT context (preceding information)-dependent word Start 0.2 second after Magnetic resonance spectroscopy (MRS) Functional magnetic resonance imaging (fMRI) processing and semantic cognition. word presentation Positron-emission tomography (PET) Near-infrared spectroscopy (NIRS) Norio Fujimaki Semantic processing: Yes No Optical coherence tomography (OCT)

11 Center for Information and Neural Networks Brain Imaging Techniques 12 Center for Information and Neural Networks CiNet: a new scientic partnership between

government, industry and academia National Institute of Information and Communications Technology (NICT) Promoting research and development and industry planning in the eld of information and communications technology, a vital sector for Japan’s economic growth. On January 7, 2009, the National Institute of Information and Communications Technology (NICT) and Osaka University came to an agreement on a large-scale collaboration for the study of brain function and applying these results to new ICT and networks. Since then, the Advanced Telecommunications Research Institute International (ATR), a leading institute in BMI technology, has joined to make this an even more potent partnership, with the future being expected to reign in even more institutes and companies that will expand CiNet’s impact in biology and medicine. Advanced Telecommunications Research (ATR) On March 10, 2007, CiNet held its kickoff symposium on information and neural networks to showcase its success and Institute International Quantitative brain research with focus on BMI vision. The minister of Internal Affairs and Communications was one of many to show support. technology With its laboratories based at the Suita Campus of Osaka University, where advanced research for non-invasive techniques on human brain function is well on its way, and by bringing together the learning and research environments at universities and industries, CiNet aims to take neuro-based ICT from science ction to science.

Osaka University Long and rich history of successful academic output in many elds and in producing many of Japan’s brightest and best scientists.

Distribution of research Cultivation of human resources

Commercial and industrial application Comprehensive Career Development

● Focus on need-based research and development ●Timely funding for research and development For members in industry and government

●Opportunities for university education and credit ●Opportunities to work with and mentor students Beneting from a University’s knowledge capital Nurturing and producing ● Cutting-edge research in various elds For members in academia ● Wealth of existing and potential intellectual properties Creating new research the human capital to ●Management opportunities in industry and government and management ●Planning of and participating in national-level policies create new ICT based Rich and diverse human resources networks on neural networks ●Teaching staff from all scientic elds ● Excellent and diverse student body research Training for the future Research devoted to national policies Nurturing and producing government, industry and academic

● National-level research activities leaders to support Japan’s strength in information and ● Strategic research and development funds communication technologies

13 Center for Information and Neural Networks Cooperation among Government 14