subglobal1 link | subglobal1 link | subglobal1 link | subglobal1 link | subglobal1 link | subglobal1 link | subglobal1 link
subglobal2 link | subglobal2 link | subglobal2 link | subglobal2 link | subglobal2 link | subglobal2 link | subglobal2 link
subglobal3 link | subglobal3 link | subglobal3 link | subglobal3 link | subglobal3 link | subglobal3 link | subglobal3 link
subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link | subglobal4 link
subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link | subglobal5 link
subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link | subglobal6 link
subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link | subglobal7 link
subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link | subglobal8 link


第9回 冬のワークショップ

Large-scale Simulations and Database in Neuroscience


Special Session: Large-scale Simulations and Database in Neuroscience
18:00-18:50 Neural Tissue Simulation on Blue Gene
James Kozloski (IBM)
19:00-19:50 Imagining functional connectomics
Sebastian Seung ( MIT )
20:00-20:50 Reconstruction of an average cortical column “in silico”
Marcel Oberlaender (Max-Planck-Institute)
Topic Session
15:30-16:20 Automated parameter searches for large single neuron models: Purkinje cell excitability and plasticity
Erik De Schutter ( OIST )
16:30-17:20 Periodic microcolumnar modules in the neocortex.
17:30-18:20 Neuronal mechanisms to control odor-evoked specific emotions and behaviors
Topic Session
09:00-09:50 Understanding insect brain adaptability by analysis and synthesis
神崎 亮平(東大)
10:00-10:50 Feedforward gain regulation for quick sensorimotor control.
11:00-11:50 Spinal control of volitional movements: beyond the reflex and locomotion

Abstracts and References:

James Kozloski
"Neural Tissue Simulation on Blue Gene”

We developed a scalable, massively parallel application for identifying and databasing circuit connectivity in structural models of neural tissue during phase I of the Blue Brain Project (Kozloski et al., 2008; Markram, 2006). This approach catalogued "potential synapses," or touch sites, between neurons by exploiting a volumetric decomposition of the tissue, thus greatly simplifying the geometric calculation and aggregation of circuit statistics. Here we show two additional applications of this approach to neural tissue simulation and a method for estimating from first principles its computational requirements.

Our first application target area is computational developmental neurobiology. We present a method for simulating the growth of neuronal processes within a simulated tissue. Our method avoids physical overlap between neural processes by deforming their geometry to conform to the local tissue environment while preserving overall neuron topology extracted from a public database of real neuron morphology. We show that the effect of the signaling molecule reelin on developing tissue can be studied using force field simulation and parameterization techniques from molecular dynamics, and we test the hypotheses of Nishikawa et al. (2002) regarding the role of reelin in minicolumn formation.

Our second application target is distributed compartmental simulation of electrical properties of neural tissue. Our approach differs from previous simulation techniques that distribute neurons to the various processors of a parallel machine and solve the Hodgkin-Huxley equations over branched processes using the fully implicit numerical method of Hines (1983). Instead, we divide neuronal processes across volume boundaries employing the explicit/implicit method of Rempe and Chopp (2006), and we show scaling for our approach. We have integrated this approach with neuron growth and used circuit identification to drive the formation of local synapses and functioning emergent microcircuits, and we present preliminary work aimed at testing hypotheses regarding the origin of neuronal domains in developing neocortex (Yuste et al., 1995).

Evaluating the computational requirements for large-scale simulations of neural tissue is difficult for a number of reasons. Perhaps most importantly, whole brain connectivity maps (i.e., connectomes) are not yet available. With our novel volumetric decomposition of neural tissue onto the 3D torus of Blue Gene, grounded extrapolations of the computational requirements for large-scale neural tissue simulation become possible. We use our approach's computational cost per unit volume and strictly local communication pattern together with real anatomical constraints to extrapolate these requirements for a hypothetical exascale architecture running a whole-brain simulation.

Kozloski J, Sfyrakis K, Hill S, Schu¨rmann F, Markram H (Feb., 2008) Identifying, tabulating, and analyzing contacts between branched neuron morphologies, IBM Journal of Research and Development special issue on Massively Parallel Computing, 52:1/2(43-55).

Markram H, "The Blue Brain Project" (Feb., 2006), Nature Reviews Neuroscience, 7:153-160.

Sebastian Seung
"Imagining functional connectomics”

Technological advances could eventually make it possible to find connectomes, complete maps of all connections between neurons in a piece of brain. How could we use this structural information to understand brain function? Here I propose a computational approach in which the probability of connection between two neurons is modeled as a function of variables called cell labels. A label could include directly observable information, like cell type or receptive field properties. It could also contain hidden variables, which are inferred through computational analysis. This approach could be applied to test the long-standing theory that memories are stored in synaptic connections. During birdsong, neural activity sequences are generated in avian brain area HVC. The hypothesis that these sequences are stored in the excitatory connections of HVC could be tested by applying one-dimensional directed graph layout algorithms to the HVC connectome. Similarly, the hypothesis that spatial maps are stored in the excitatory connections of hippocampal CA3 could be tested by applying more complex graph layout algorithms to the CA3 connectome.

Marcel Oberlaender
Reconstruction of an average cortical column “in silico”

The group of Prof. Dr. Bert Sakmann at the Max Planck Institute of Neurobiology in Munich aims to understand principle mechanisms that underlie decision making in the mammalian brain.
The behavioral paradigm of our work is the so called “gap-crossing” experiment 1. It is a yes-or-no (binary) decision, in which a mouse chooses whether it crosses an approximately 6 cm wide gap to reach sweet milk or not. The decision depends on whisker deflection. If the whiskers reach and touch the other side of the gap, all mice reliably cross the gap. If the gap size is too big, mice never cross. The performance is independent of the number of whiskers. A single whisker (all others are trimmed) is sufficient to reliably trigger the crossing.
Each whisker has a functional correspondence in the primary somatosensory cortex (S1) that we refer to as the cortical column. Since the deflection of a single whisker is sufficient for triggering the crossing of the gap, the detailed study of physiology and anatomy of the corresponding cortical column will most likely reveal insights to basic mechanisms of the neuronal network that encodes sensory information from a whisker 2.
Each cortical column consists of more than 10000 thousand neurons 3, 4 and multiple neuron types. So far no experimental “in vivo” approach is capable to study the function and interaction of such a large scale ensemble of neurons on a cellular basis. We believe that the detailed anatomy of the network, the morphology of its constituent neurons and the wiring between them are key prerequisites for understanding cortical information processing. We therefore reconstruct an average cortical column as a model system for “in silico” network studies.
So far our group and collaborators physiologically characterized the cell-type specific input/output balance of a column 5, 6 and developed multiple tools to gather quantitative anatomical data such as the number and distribution of neurons in a column 7, mean characteristics of neuron types, axonal projection domains 8-10 and number of synaptic contacts.
Based upon this anatomical knowledge the generated average cortical column is realized as a network of full-compartmental neuron morphologies. A newly developed powerful simulation environment 11 allows for a quantitative “in silico” investigation of such large and complex neuronal ensembles. Finally, an interactive 3D visualization and analysis tool enables the correlation of simulation results to single cell, pair or network data from “in vivo” experiments.


  1. Celikel, T. & Sakmann, B. Sensory integration across space and in time for decision making in the somatosensory system of rodents. Proc Natl Acad Sci U S A 104, 1395-400 (2007).

  2. Helmstaedter, M., de Kock, C. P., Feldmeyer, D., Bruno, R. M. & Sakmann, B. Reconstruction of an average cortical column in silico. Brain Res Rev 55, 193-203 (2007).

  3. Meyer, H. S., Wimmer, V. C., Oberlaender, M., Sakmann, B. & Helmstaedter, M. The number and distribution of interneurons in a cortical column of rat barrel cortex. (in prep).

  4. Meyer, H. S., Wimmer, V. C., Oberlaender, M., Sakmann, B. & Helmstaedter, M. Neuron density profiles of a thalamocortical innervation column in rat somatosensory cortex. (in prep).

  5. Bruno, R. M. & Sakmann, B. Cortex is driven by weak but synchronously active thalamocortical synapses. Science 312, 1622-7 (2006)

  6. de Kock, C. P., Bruno, R. M., Spors, H. & Sakmann, B. Layer- and cell-type-specific suprathreshold stimulus representation in rat primary somatosensory cortex. J Physiol 581, 139-54 (2007).

  7. Oberlaender, M., Dercksen, V. J., Egger, R., Sakmann, B. & Hege, H. C. Automated three-dimensional detection and counting of neuron somata. (in prep).

  8. Oberlaender, M., Broser, P. J., Sakmann, B. & Hippler, S. Shack-Hartmann wavefront measurements in cortical tissue for deconvolution of large three-dimensional mosaic transmitted light brightfield micrographs. Journal of Microscopy-Oxford (in press).

  9. Oberlaender, M., Bruno, R. M., Sakmann, B. & Broser, P. J. Transmitted light brightfield mosaic microscopy for three-dimensional tracing of single neuron morphology. Journal of Biomedical Optics 12, - (2007).

  10. Dercksen, V. J. & Oberlaender, M. 3D Editing, Alignment and Splicing Environment for Neuron Reconstructions. EuroVis 2009 Berlin (in prep).

  11. Bastian, P. & Lang, S. (Stuttgart, 2007).

Erik De Schutter
"Automated parameter searches for large single neuron models: Purkinje cell excitability and plasticity”

Automated parameter searches have become an important tool in tuning neuron models and are an active research topic by themselves. The quality of the search is largely determined by the fitness measure used. Two approaches can be distinguished: feature based and trace based fitness measures. I will discuss the advantages and disadvantages of both approaches and then describe our phaseplane density method to fit voltage traces (Van Geit et al. 2007).
Based on this method we have searched the parameter space of a previously developed Purkinje cell model (Achard and De Schutter 2006). We found that a large number of models produced very similar complex firing patterns to current injection. These models were not completely isolated in the parameter space, but neither did they belong to a large continuum of good models that would exist if weak compensations between channels were sufficient. The parameter landscape of good models could best be described as a hyperdimensional foam.
This complex parameter space may give insight in the activity homeostasis controlling Purkinje cell excitability. Moreover it makes specific predictions on a coupling between intrinsic excitability and the potential to undergo synaptic plasticity at parallel fiber synapses (Steuber et al. 2007), suggesting a previously undiscovered link between the two processes.

Achard P, De Schutter E. Complex parameter landscape for a complex neuron model. PLoS Comput Biol. 2006 2:e94.

Steuber V, Mittmann W, Hoebeek FE, Silver RA, De Zeeuw CI, Ha¨usser M, De Schutter E. Cerebellar LTD and pattern recognition by Purkinje cells. Neuron. 2007 54:121-136.

Van Geit W, Achard P, De Schutter E. Neurofitter: a parameter tuning package for a wide range of electrophysiological neuron models. Front Neuroinformatics. 2007 1:1.

Toshihiko Hosoya

"Periodic microcolumnar modules in the neocortex.”


Reiko Kobayakawa

"Neuronal mechanisms to control odor-evoked specific emotions and behaviors”


Kobayakawa et al. Nature vol.450, p503-308 (2007)

神崎 亮平
Ryohei Kanzaki

"Understanding insect brain adaptability by analysis and synthesis”

昆虫はその微小な寸法という制限要因の中で、センサ・脳神経系を発達させ、さまざまな環境下で適応的な機能を進化させてきた。昆虫の小さな身体に潜む感覚・処理・運動能力は最近、「昆虫パワー」といわれる。このような昆虫機能は、われわれ哺乳類の複雑な脳神経系や、複雑化するロボットなどの機械の設計とは対照的である。単純・高速・経済的なセンサ・処理装置の技術開発には昆虫のセンサや脳神経系は魅力的な手本であり、また哺乳類の脳モデルとしてその設計には学ぶべきことは多い。わたしたちは、昆虫パワーを、鱗翅目昆虫のカイコガ(Bombyx mori)をモデル動物として、遺伝子・ニューロン・神経回路・行動にいたるマルチスケールの分析、分析結果の移動ロボットによる実環境下での検証・評価、さらには昆虫と機械を融合した「昆虫/機械ハイブリッド」の構築を通して、理解し活用する研究を展開している。

神崎亮平,倉林大輔 (2007) 生体-機械融合システムによる生物の環境適応能の理解と構築.計測と制御,第46 巻第12 号2007 年12 月号 934-939

Kanzaki R (2007) How does a microbrain generate adaptive behavior? International Congress Series 1301 Brain-Inspired IT III: 7-14

Emoto S, Ando N, Takahashi H and Kanzaki R (2007) Insect-Controlled Robot -Evaluation of Adaptation Ability - Journal of Robotics and Mechatronics19(4): 436-443

Seki Y and Kanzaki R (2008) Morphological classification of antennal lobe local interneurons in Bombyx mori by the intracellular staining under visual control method. J Comp Neurol 506: 93-107

Wang H, Ando N, and Kanzaki R (2008) Active controls of free flight maneuvers in a hawkmoth, Agrius convolvuli. J Exp Biol 211, 423-432

Yamagata T, Sakurai T, Uchino K, Sezutsu H, Tamura T and Kanzaki R (2008) GFP Labeling of Neurosecretory Cells with the GAL4/UAS System in the Silkmoth Brain Enables Selective Intracellular Staining of Neurons. Zool Sci 25: 509-516

Namiki S and Kanzaki R (2008) Reconstructing the population activity of olfactory output neurons that innervate identifiable processing units. Frontiers in Neural Circuits 2(1):1-11

Kazawa T, Ikeno H, and Kanzaki R (2008) Development and application of a neuroinformatics environment for neuroscience and neuroethology. Neural Networks 21( 8): 1047-1055

Hiroaki Gomi

"Feedforward gain regulation for quick sensorimotor control. ”


Gomi H, Abekawa N, Nishida S (2006) Spatiotemporal tuning of rapid interactions between visual-motion analysis and reaching movement. J Neurosci 26:5301-5308.

Kimura T, Haggard P, Gomi H (2006) Transcranial magnetic stimulation over sensorimotor cortex disrupts anticipatory reflex gain modulation for skilled action. J Neurosci 26:9272-9281.

関 和彦
Kazuhiko Seki

"Spinal control of volitional movements: beyond the reflex and locomotion”


Takei T, Seki K (2008) Spinomuscular coherence in monkeys performing a precision grip task. J. Neurophysiol. 99(4):2012-20.

Seki, K., Perlmutter,S.I. and Fetz,E.E. (2003) Sensory input to primate spinal cord is inhibited presynaptically during voluntary movement., Nature Neuroscience. 6(12): 1309-1316.

©2005 Mechanism of Brain and Mind