Abstracts and References:
Richard Andersen
" Neural Prosthetics using Cognitive Control Signals "
Neural prosthetics research aims to develop brain-machine interfaces that can decode the intentions of paralyzed patients to enable them to operate assistive devices. Recent studies have found that high level cognitive variables can be decoded from the posterior parietal cortex and dorsal premotor cortex, areas upstream of the primary motor cortex. Healthy monkeys were trained to move a cursor on a computer screen using the decoded neural signals without their making any overt movements. The cognitive signals decoded include the goal and trajectory of the cursor movement in external (visual) coordinates. Since these areas are part of the neural circuitry for decision making, we were also able to decode decision variables related to expected value of the reward provided for the completion of each trial. The expected value signals indicated the probability, amount, and type of reward. The goal and trajectory signals can be used by paralyzed patients to operate computers, robots and vehicles and the expected value signals to assess the patients preferences, mood, and motivations. The demonstration of the decoding of these high-level attributes suggests that in the future other cognitive signals, such as those related to emotion, speech, and executive control, may also be used for prosthetic applications.
John Donoghue
"Turning thought into action: Using cortical ensembles in humans for direct neural control of computers and robots"
I will discuss our initial results from a human clinical pilot trial to develop a direct neural interface to control neural prosthetic devices. We have implanted two humans with tetraplegia from spinal cord injury with multielectrode arrays. Preliminary data from these participants show that neural signals persist in primary motor cortex years after spinal cord injury, that neural activity can be modulated by intention in the absence of movement, and that this activity can be decoded and used to control a range of devices that can potentially enhance independence of paralyzed humans. In addition, these results suggest that physical restoration of action via a physical connection from brain to the spinal cord is feasible for humans with spinal cord injury.
Serruya, M.D., Hatsopoulos, N.G., Paninski, L., Fellows, M.R., and Donoghue, J.P. (2002) Instant neural control of a movement signal. Nature 416:141-2.
Wu, W., Gao, Y., Bienenstock, E., Donoghue, J. P., Black, M.J. (2005) Bayesian population decoding of motor cortical activity using a Kalman filter Neural Computation: 2006; 18:80-118.
Suner, S., Fellows, M. R., Vargas-Irwin, C., Nakata, K., Donoghue, J. P. (to appear) Reliability of signals from chronically implanted, silicon-based electrode array in non-human primate primary motor cortex. Submitted to IEEE Transactions in Neural Systems and Rehabilitation Engineering.
Donoghue, J. P., Nurmikko, A., Friehs, G., and Black, M., J. (2004) Chapter 63. Development of a neuromotor prosthesis for humans, in Advances in linical Neurophysiology,Supplements to Clinical Neurophysiology, Vol. 57,
[Proceedings of the 27th International Congress of Clinical Neurophysiology, AAEM 50th Anniversary and the 57th Annual Meeting of the ACNS Joint Meeting, San Francisco, CA, USA, 15-20 September 2003] M. Hallett, L.H. Phillips II, D.L. Schomer, J.M. Massey, Eds., pp.588-602.
Serruya MD, Donoghue JP. (2003) Chapter III: Design Principles of a Neuromotor Prosthetic Device in Neuroprosthetics: Theory and Practice, ed. Kenneth W. Horch, Gurpreet S. Dhillon. Imperial College Press. pages 1158-1196.
Serruya, M., Hatsopoulos, N. Fellows, M. Paninski, L, and Donoghue, J. (2003) Robustness of neuroprosthetic decoding algorithms. Biological Cybernetics 88 (3): 219-228 March 2003.
The original publication is available at http://link.springer-ny.com.
Donoghue, J.P. (2002) Connecting cortex to machines: recent advances in brain interfaces. Nature Neuroscience Supplement, November 2002. 5:1085-8.
Dennis J. McFarland
"A Non-Invasive Brain-Computer Interface for Communication and Control"
Brain-computer interfaces (BCIs) can provide communication and control to people who are paralyzed. BCIs can use non-invasive or invasive methods for recording the brain signals that convey the user's commands. While non-invasive BCIs are already in use for simple applications, it has been widely assumed that only invasive BCIs, which use electrodes implanted in the brain, can provide multidimensional sequential movement control of a robotic arm or a neuroprosthesis.
In recent studies, we showed that a noninvasive BCI that uses scalp-recorded EEG activity (i.e., sensorimotor rhythms) and an adaptive algorithm can provide humans with a one and two-dimensional movement control. The adaptive algorithm used in this non-invasive BCI identifies and focuses on the EEG features that the person is best able to control and encourages further improvement in that control.
We now show that an extension of the same methods can provide sequential control consisting of two-dimensional movement to a goal followed by selection of the goal. The users first use two independent EEG control signals (i.e., a horizontal control signal and a vertical control signal) to move the cursor to hit one of several targets. Next, they either select or reject the target using a third EEG control signal.
These results suggest that people with severe motor disabilities could use brain signals for full mouse control, or even to reach and grasp objects with a robotic arm or a neuroprosthesis.
Support from: NIH Grants HD30146 (National Center for Medical Rehabilitation Research, NICHD) and EB00856 (NIBIB and NINDS)) and from the James S. McDonnell Foundation.
Wolpaw, J. R., Birbaumer, N., McFarland, FD. J., Pfurtscheller, G. & Vaughan, T. M. (2002) Brain-computer interfaces for communication and control. Clinical Neurophysiology, 113, 767-791.
Wolpaw, J.R and McFarland, D.J. (2004) Control of a two-dimensional movement signal by a non-invasive brain-computer interface in humans. Proceedings of the National Academy of Sciences, 101, 17849-17854.
Yukiyasu Kamitani 神谷之康
"非侵襲的脳情報復号化の可能性"
脳情報復号化とは，脳信号から未知の刺激や心理状態を推定することを指し，刺激や課題が与えられたときの脳活動をマッピングする従来の脳機能研究とは対照的なアプローチをとる．本講演では，ヒトの非侵襲的脳信号から知覚内容や運動状態を復号化することに成功したわれわれの最近の成果を紹介し，身体運動や言語を介さない情報伝達の可能性や復号化情報を利用した新たな実験パラダイムの展望を議論する．
Tomoki Fukai 深井朋樹
"Attractor dynamics of recurrent neuronal networks with spontaneous membrane potential fluctuations"
アトラクター神経回路のダイナミクスは学習や記憶などの高次脳機能 において重要な役割を演じているものと考えられる。大脳皮質の神経細胞は脱分 極性のUP状態と、静止状態にあたるDOWN状態という閾値下の二つの膜電位状態間 で、自発的遷移を繰り返すことが知られており、脳の内部状態を表現している可 能性や、その機能的役割について興味がもたれている。しかしながら、二状態間 遷移が覚醒状態の大脳皮質にも存在し、何らかの情報処理の役に立っているかは 未だ良くわかっておらず、議論が分かれるところである。本公演では近年、Ca2+ イメージング法を用いて明らかにされたような、大脳皮質神経回路の自発発火に おけるミリ秒精度でのUP状態遷移の時系列生成が、自発発火とスパイク時間依存 のシナプス可塑性を通じた神経回路の自己組織化により実現可能であることを、 計算論的モデルを用いて示す。また自己組織化された回路の神経活動において、 以下のような特徴が見出された。１）神経回路内に、規則的発火パターンを示す ペースメーカ的な神経細胞から、不規則なUP／DOWN状態遷移とバースト発火を示 すものまで、多様な活動パターンが生成される。２）自己組織化によりミリ秒精 度でのUP状態遷移の時系列生成を実現するためには、各神経細胞が明瞭に区別可 能な二状態をもつことが不可欠である。３）UP／DOWN状態を示す神経細胞の発火 パターンの決定において、抑制性細胞からの入力が大きな影響力をもつ。これは 主に、DOWN状態を調節する内向き整流K+電流とIPSPとの相互作用に起因する。こ れらの結果は、大脳皮質の神経回路活動はリカレントな興奮性、抑制性のシナプ ス入力と、細胞固有のイオンチャネルのダイナミクスとの複雑な相互作用で決ま る部分があること、そのため積分発火ニューロンなど単純なモデルだけで活動を 解析することには限界があることを示唆している。
Toshihiko Hosoya 細谷俊彦
"Dynamic Predictive Coding by the Retina"
Retinal ganglion cells convey the visual image from the eye to the brain. Their receptive fields generally have “Mexican hat” shapes in space and are biphasic in time. Thus the retina encode local differences in space and changes in time rather than the raw image intensity. This is regarded as an evolutionary adaptation to the natural environment. In the average visual scene, the light intensity at each point tends to be similar to, and thus predictable from, the intensity at neighboring points. The antagonistic receptive field effectively subtracts the predicted intensity from the actual intensity, reducing the signal dynamic range and increasing resolution. Yet animals encounter many environments with visual statistics different from the average. Here we show that when this happens, the retina adjusts its processing dynamically. The receptive fields change after a few seconds in a new environment. The changes are adaptive, in that the new receptive field improves predictive coding under the new image statistics. We show that a network model with plastic synapses can account for the large variety of observed adaptations.
Hosoya T, Baccus SA, Meister M. Dynamic predictive coding by the retina. Nature 436(7047): 71-7, 2005.
Katsuyuki Sakai 坂井克之
"Prefrontal set activity predicts subsequent cognitive performance"
In every day life the representation of rules is vital for goal-directed behavior, future planning and behavioral flexibility. The finding of neuronal activity that reflects behavioral rules has therefore made an impact on research in cognitive neuroscience. In psychological experiments the experimenter first outlines the task rules to be followed and we have previously identified activity in prefrontal cortex that reflects preparation for following those rules.
Here I will present new data that suggest the mechanisms via which this rule-related activity guides later performance based on that rule. We show that the sustained preparatory activity in the prefrontal cortex predicts the activity in the posterior areas during the task performance. We also show that the anterior prefrontal cortex interacts with the posterior areas in task-specific ways: the area with which it most strongly interacts depends on the specific task operation to be performed. These results suggest that the interactions prime selectively the task-related areas before the execution of the task and reduce the processing load in those areas.
Bunge S.A, Wallis J.D, Parker A, Brass M, Crone E.A, Hoshi E, Sakai K. Neural circuitry underlying rule use in humans and non-human primates. Journal of Neuroscience, 2005 in press.
Sakai K. Passingham RE: Prefrontal interactions reflect future task operations. Nature Neuroscience 6: 75-81, 2003.
Sakai K, Rowe JB. Passingham RE: Active maintenance in prefrontal area 46 creates distractor-resistant memory. Nature Neuroscience 5: 479-484, 2002.
Masami Kojima 小島正己
"Looking into brain function using genetic variants
-‘Jekyll and Hyde’ model of growth factor function in brain -"
With the sequencing of the human genome, it is possible to identify genetic variations, namely single nucleotide polymorphisms (SNPs), in a population. Approximately 6 million SNPs are believed to be the genetic variations in worldwide populations. However, a minority of those polymorphisms, which cause amino acid substitution of the protein, will be functional in nature. It would be interesting to examine how SNPs in neural gene affect brain function and to look for the new mechanisms in brain. We have now studied the modulation of brain function by SNPs in human brain-derived neurotrophic factor (BDNF) gene. The topics in my talk are: 1) high-frequency BDNF polymorphism (Val66Met) impairs an activity-dependent secretion of BDNF and human episodic memory, 2) low-frequency BDNF polymorphisms located near the processing site of BDNF (Arg125Met and Arg127Leu) inhibited the conversion of precursor BDNF to mature BDNF and consequently altered its biological action, a) from spine growth to spine retraction, b) from dendrite growth to dendrite retraction and c) from cell survival to cell death. These findings led us to propose a novel bidirectional regulation of brain function and development, namely ‘Jekyll and Hyde’ model of growth factor function in brain. Such integrative approach should give us a unique way of looking into brain function and contribute to the prevention of brain disorders associated with polymorphisms in neural gene.
Minor variation in growth factor gene impairs human memory. Science, vol. 299, 639-640, 2003.
Noriko Osumi 大隅典子
"脳構築の分子メカニズム"
ヒトの脳の中には1000億個のニューロンと、その10倍の数のグリア細胞が存在し、精密なネットワークを形成しているという。たった1個の受精卵から出発して、60兆個の細胞から成る体が、そして脳ができる仕組みは、長い進化の過程で備わってきたものだが、まさに驚異的というしかない。脳の細胞の元になる細胞（神経前駆細胞）がたくさん分裂して数を増やし、ニューロンやグリアの細胞に変化（分化）する。神経前駆細胞の増殖と分化は胎生期に爆発的に生じるが、生後も持続し、その程度は減りつつも生涯に渡って続くことが近年明らかになってきた。例えば、ラット・マウスでは側脳室周囲では一日あたり約80,000個、海馬では約9,000個のニューロンが生みだされる。ただし、これらの新生ニューロンがすべて生存するのではなく、大多数は1週間以内に細胞死を起こすが、それでも約1割のニューロンは神経ネットワークに組み込まれて機能すると考えられている。このことは、神経ネットワークが活動依存的にシナプスの組み換えや強化を起こすだけでなく、新しく産生されたニューロンを漸次組み込んで再編されていることを意味する。すなわち、神経ネットワークは従来考えられていたよりも、ずっとダイナミックに恒常性を維持しつつ活動していると考えられ、このような神経新生が脳の高次機能にどのように関わるかについての興味が大きくなってきている。実際に、鳴禽が季節毎に歌を学習する際に神経新生が盛んになることが知られており、ラット・マウスでも神経新生が記憶や学習に関わることが示唆されつつある。私たちの研究室では、このような神経新生における細胞の振る舞いや遺伝子ネットワークを明らかにする研究に取り組んでおり、とくに脳構築のマスターコントロール遺伝子とも呼ばれるPax6を中心とする遺伝的プログラムについて明らかにしつつある。興味深いことに、胎生期における神経新生で働く分子メカニズムは、生後脳でも使い回されて働いているらしい。また、神経新生の低下により感覚運動ゲート機構の異常が引き起こされる可能性を見いだしている。