The BRAIN Initiative and its links to MSM Theme

Return to 2014 Multiscale Modeling Consortium Meeting

The BRAIN Initiative and its links to MSM (Bill Lytton, Raj Vadigepalli leads)


NIH Contacts for MSM Questions about the BRAIN Initiative:

Susan Vollman, Ph.D. Behavioral and Cognitive Science Research Branch (BCSRB) NIDA (301) 435-1315 

German Cavelier, Ph.D. Program Officer - Office of Technology Development & Coordination (OTDC), Office of the Director, NIMH (301) 204-0510 Cell (Preferred)  or (301) 443-9124  (Office)\


Keynote Speaker: Terry Sejnowski

IMAG Discussants: Sri Raghavachari (NSF), Justin Sanchez (DARPA), Ned Talley (NIH), Jacob Vogelstein (IARPA), Susan Volman(NIH)

BRAIN Inititative Report to NIH BRAIN 2025: A Scientific Vision (Brain Research through Advancing Innovative Neurotechnologies (BRAIN) Working Group Report to the Advisory Committee to the Director, NIH June 5, 2014)

External links

Aug 25, 2014 NIH Council Discussion (video), Roster, Agenda

White House BRAIN Initiative:

NIH BRAIN Initiative:, Funding

Ned Talley

From last year's brain intiatives, NIH will be awarding over 50 awards over $40M

For next year, reissue some of last year's RFA.  There will special focus on integration computation and experimentation, another on human neuroscience (suppporting today's technogies to do human research, unlock industry developed technologies), multiscale approaches to understanding neural dats (e.g. macro-signal for understanding human signals)

NSF BRAIN Initiative:

Understanding the brain website at the NSF - 36 awards just funded to develop innovative solutions using cyberinfrastructure

Collaborative Research in Computational Neuroscience (CRCNS)

Computational Data Enabled Neuroscience - algorithm development of data analysis

DARPA BRAIN Initiative:

1) DARPA SUBNETS program - 2 projects already running

2) DARPA Restoring Active Memory after brain injury, real-time formation of recall in memory in humans - 2 teams working on projects

3) DARPA Neural Function Structure and Activity - 2 projects funded for full 3D readout of cells in awake-behaving activity, demonstrating applicability in humans, all data will be shared


Not part of the BRAIN initiative formally, but have 6-7 programs in neuroscience on the streets


Adaptive Reasoning and Problem Solving (SHARP), including Transcranial Direct Current Stimulation

Large Scale Cognitive Models of Brain Function - understanding brain processes leading to suboptimal decisions

Coming soon:  Untapped - predicting from brain structure and function to cognitive capabilities

Coming soon:  MICRONS - Machine Intelligence from cortical networks, leveraging tools to understand computation in the brain

President Obama has indicated federal interest and support in a BRAIN (Brain Research through Advancing Innovative Neurotechnologies) to “unlock the mystery of the three pounds of matter that sits between our ears” (President's remarks on BRAIN initiative and American innovation). This initiative will involve a number of IMAG members including NIH, NSF, and DARPA. Much of the focus thus far has been on the novel imaging technologies that will be required to provide dynamic brain activity mapping. However, it is clear that detailed multiscale modeling will be essential in order to unlock this mystery, and that development of new MSM techniques will be 1 of the new technologies that will be needed. Modeling is, in fact, the essential key that will be needed as the data begins to fall into place.

The focus of this theme is to explain the critical role of MSM in enabling the success of the BRAIN initiative, and to explore answers to the question: “What could MSM do with data on every spike (or every PSP) in every neuron in a brain area?” The focus of the talks will be to understand how MSM can best contribute to this funding initiative. Therefore there will be no scientific presentations per se, but rather talks directed to the application of MSM as a set of new technologies that will be useful for BRAIN. Terry Sejnowski will be giving an address based on his experience as a member of the advisory board to NIH Director.

KEYNOTE SPEAKER: Terry J Sejnowski

Abstract: One of the major goals of the BRAIN Initiative is to integrate theory, modeling, statistics, and computation with experimentation. Rigorous theory, modeling and statistics are advancing our understanding of complex, nonlinear brain functions where human intuition fails. New kinds of data are accruing at increasing rates, crying for new methods of data analysis and interpretation. To enable progress in theory and data analysis, we must foster collaborations between experimentalists and scientists from statistics, physics, mathematics, engineering and computer science.

Other speakers will be from the 3 involved funding agencies NIH, NSF and DARPA, as well as a representative from IARPA. We will follow with a general discussion.

In addition to the particular points about brain modeling, we will extend the discussion to encompass shared issues and problems that will allow this theme to be of general interest to the MSM community. These topics include: 1. Development of new cell imaging technologies. 2. how to handle the yottabytes of "Big Data" generated. 3. Verification of simulations. 4. Matching analysis of simulations with direct analysis of data. etc.

An interesting perspective on the BRAIN project is in the Charlie Rose show from 7/14/13: included Story Landis (NINDS), Tom Insel (NIMH), Bill Newsome and Cory Bargmann (chairs of the advisory committee), Eric Kandel as moderator. Another recent document is. Another question that arises is the relation (if any) with the EU's Human Brain Project. One consideration is that the US project is focused on on development of new technologies, while HBP has proposed to utilize existing technologies for their analysis.

Discussion notes (from Bill Lytton):  

The word "analysis" is in the announcements to indicate that need work to interpret and understand these new data (credit to Susan Volman for this).
1. The methods being developed to analyze large datasets from physiology can also be used to analyze large datasets from simulation.
Do bioinformatic techniques for systems where all state variables are available (i.e. simulations) differ from analysis of experiments where only have a few indicators (e.g. neurophysiology)?
2. Need new measures to assess fundamental features -- patterns of molecular features that will demonstrate phenotypic patterns of activity. 
3. Simulation methods should include multiphysics (distinguish multiphysics from multiscales).  We need to add astrocytes and neurovascular coupling -- both different scale and different
  physics, e.g. biomechanics and electrodynamics and chemical reaction-diffusion simulation.
Please let's avoid using the word 'new' in describing these methods.  All of this is working with old methods but these are now applied to different (new) problems and combined in different (new) ways 
4. Fine-grained resolution (bottom-up detailed modeling) vs making use of lumping of the lower levels when modeling at higher scales.
HBP (European Union's Human Brain Project) is more explicitly about bottom-up.
HBP is more about simulation whereas BRAIN is more about neurotechnology (which might include neurosimulation) but not redundant with HBP.
5. Desideratum: simulator interoperability; but difficult to achieve
6. Desideratum: database sharing from experiment to simulation (parameterizations and trajectory targets), from simulation to experiment (ie for prediction assessment) and from experiment to prediction.
7. Desideratum: automated systems for developing, running and comparing simulations.
Circuit level is specifically an interest of BRAIN -- development of realistic neuronal network configurations as opposed to idealized configurations that are often analyzed (since these idealizations are amenable to dynamical reductions and other mathematical techniques).  Need to articulate approaches that can deal with circuit dynamics.  Note that Terry Sejnowski gave example of spindle formation in thalamus and corticothalamic circuit in his talk.
It is also important to characterize the cell types involved.  There is a suspicion/hope that one can find molecular and functional logic in the placement of these cell types in the circuit.
Need to define what is state-of-art and what specifically could be done in near future.
Would like to considering the brain as a whole unit from molecules up to behavior (multiscale modeling) -- how communication between different levels takes place what are the necessary things at the levels -- go from ion channel but then can do firing as hopf bifurcation
Susan Volman and German Cavalier are happy to provide further information and advice as RFAs come out.