Following is my email interview with Thomas Lippert, professor and director of the Institute for Advanced Simulation and the Jülich Supercomputing Centre at the Forschungszentrum Jülich in the west of Germany, located somewhere between the cities of Cologne and Aachen. Lippert’s FZ Jülich is a central partner at the Human Brain Project (HBP), the Future and Emerging Technologies (FET) Flagship mega-consortium which is funded with €1 Billion, half of which coming from the European Union. Lippert heads the Subproject (SP) 7 at the HBP, named High-Performance Analytics and Computing Platform, tasked with providing the high-performance computational power in order to help the HBP megaproject simulate the human brain and its diseases.
I previously wrote some critical articles about HBP, about the hubris of its founding father Henry Markram, the bizarre way EU positively evaluated the performance of HBP without actually evaluating it, and also about the purpose of HBP and Flagship projects in general. Incidentally, the last article’s cover photo showed Lippert at a HBP conference. We got into a discussion over Twitter during HBP’s recent Open Day show in Glasgow, where Lippert kindly agreed to answer my questions about the purpose and orientation of HBP.
This interview with Lippert was indeed very useful, because Wolfgang Marquardt, FZ Jülich director, HBP Stakeholder Board member and past mediator of the anti-Markram coup at HBP, refused to tell me the secret of what HBP’s goal actually is. Lippert now indicates that the simulation of the human brain is indeed still a goal, and compares it to modelling weather or climate in supercomputers. He however also seems to indicate the limitations of the computer simulations, that the computational results need to be held against the actual neuroscience from the lab. For me, and probably some other biologists, this would mean the end of the road for brain simulations way before the travel starts. We simply do not know how complex biological neural systems generate even the most primitive forms of cognition, most likely we will never simulate the brain function of a snail, regardless of how big our supercomputers get. Brains are simply not big computers, just as birds are not little air planes. At least the dethroned HBP founding father Markram is still tinkering in the Swiss Lausanne with his Blue Brain supercomputer, where he occasionally invites gullible journalists to talk about his fantasies of simulating human brain. He also published a paper in his own Frontiers where he spoke excitedly of 11 dimensions he discovered hiding inside a brain.
EU is set to push FET Flagships regardless of anything, and they want more of them. As I learned from Lippert, even if the HBP were to be dissolved, the money would simply go to other Flagship projects, and the competing candidates can get indeed very silly (“From guardian angels to cuddly robots“). During the Edinburgh conference, I found out that the only Achilles heel of FET Projects which the EU can get upset about, is their boy’s club attitude (more here).
Indeed, it seems my tweets during the HBP conference hit a nerve, also Lippert recognises the problem as serious in the interview (his own SP7 doesn’t look too balanced, out of 10 key people listed only one is a woman). Lars Muckli, Glasgow-based computational neuroscientist and head of SP 3 (Systems and Cognitive Neuroscience) at HBP, wrote me during the HBP conference in October the following:
“There is an action plan to tackle gender inequality in HBP. An external company has been hired to advice HBP. There will be a gender conference in March 2018 in Madrid. And a dedicated task group meeting later this afternoon etc”.
Email interview with Thomas Lippert (TL), interviewer: Leonid Schneider (LS)
LS: HBP was started by Prof Henry Markram with the expressed goal to simulate human brain in a computer, the consciousness was expressly declared as possible, another goal was simulating brain diseases such as Parkinson’s or Alzheimer’s in supercomputers in order to find treatments for them. After the “coup”, the goals were seemingly dropped or reformulated, but they are still regularly mentioned, such as in the tweets from HBP last meeting. And there are of course the two platforms, for human and mouse brain organisation, but neither has Prof Markram involved.
1. Decisions in the human brain project are strictly following the rules set out in the FPA (Framework Partnership Agreement). They realize the principle of checks and balances. They force the HBP to follow democratic procedures. These procedures guarantee to avoid any “coup”.
2. “Human and mouse brain organization” are both sub-projects, they are not HBP platforms.
3. Prof. Markram runs the Simulation Platform, which is part of the joint platform of the HBP. It is linked through co-design projects with all other sub-projects.
LS: What is the current scientific (not organisational) goal of HBP?
TL: The HBP is not static. Its goals, fortunately, evolve. The brain is definitively a complex, multi-scale structure. The scientific goal of the HBP is to understand this structure and the function of the brain on its different levels and to elucidate how these levels interact. The levels range from molecules to large cognitive systems. Bridging these scales should allow disentangling cause-and-effect relations, a particular criterion is consistency. There are other scientific goals of the HBP. One is the creation of brain function inspired computing systems. Again another one is a Europe-wide federation of clinical data to constrain brain models. Certainly, an overarching goal must be to help to better understand, diagnose and hopefully cure brain diseases.
LS: Is HBP still about simulating human brain and its diseases in supercomputers?
TL: Modelling and simulating these models of the human brain was and is a goal of the HBP. HBP and other results show that one can get a better understanding of brain organization, e.g., to analyze and predict previously unknown features of molecules, cells and networks using simulation of models of the human brain.
Such a model-based computational science approach has been successfully applied in many other fields of research: examples are simulations of models of materials, models of elementary particles or models of weather and climate. Computational brain research is not special in this respect.
LS: Can you explain why you think brain simulation is at all doable, in general and specifically under the HBP setup?
TL: The process to achieve a better understanding involves experiments and observations, but also the creation of refined computer models and simulation. The computer is used as instrument to calculate the dynamics of these models, and to make predictions. The results are being compared with experimental results and the models can be refined. What is described is doable (and done in so many other fields).
What is more, the HBP is the right framework for such interdisciplinary collaboration: supercomputer specialists, informatics experts, theoretical and neuroscientists work together in integrated projects, so-called co-design projects, under the umbrella of the HBP. Many scientists have already joined, want and will join the HBP in upcoming open calls because of the benefits they can expect.
LS: Otherwise, what are the goals of HBP, aside of creating a platform or network, which for many does not look at all as a scientific objective?
TL: I vehemently object calling the building of scientific infrastructure to be “not a scientific objective”. In fact, research infrastructures are the scalable solution to let scientists participate all over Europe and beyond. They are built by scientists for other scientists.
LS: In view of these scientific targets, how is the investment of €1 Billion justified? Please consider that there if HBP were dissolved, the money would still be available for neuroscience or robotics researchers to apply to under Horizon 2020 programme.
TL: First of all, the EU funding is about €500 Million over 10 years. The EU expects this sum to be complemented by national efforts. Secondly, the money is provided by DG Connect (The European Commission Directorate General for Communications Networks, Content & Technology). Thus, Neuroscience can enjoy receiving additional support from a different DG beyond previously existing funding schemes to realize its scientific targets. If the HBP were dissolved the money would go to other Flagship or FET initiatives, not to neuroscience. In addition, neuroscience, brain computing or robotics researchers can apply under other H2020 programs, independently of the HBP.
LS: Is it correct that outside applicants can only join HBP with approval of established members, who annually pre-screen the applications before they are jointly submitted to the EU Commission?
TL: This is not correct. What is correct: The Scientific and Infrastructure Board (SIB) proposes open calls, which are endorsed by the Stakeholder Board, and agreed with the EC. The applications to the calls are being evaluated and ranked by independent international reviewers. There is no pre-screening at all. Please check the “partnering’ section of our website for more information.
[LS after-note: HBP removed a website with a reading list of own papers for the applicants which I mentioned here, but here an excerpt from the website “partnering” section, about potential Partnering Projects (PPs):
“Members of a candidate PP are advised to contact the HBP Project Coordination Office (PCO) prior to the submission of the formal application. The HBP PCO is the main point of contact for information about the application and association process, and can be reached at email@example.com”.]
LS: HBP was set up by three men, Henry Markram, Richard Frackowiak and Karlheinz Meier, and even now it is extremely male dominated, as it is evident here. Can you explain this? Does HBP plan to act on this gender imbalance?
TL: HBP was set up by a larger group of scientists who have written the original proposals than three persons. In the ramp-up phase, the executive board was then three persons and the board of directors was 12. We share the problem of gender imbalance with about 15% of leadership positions at this time being held by female researchers. Thus, the gender imbalance is substantial. The fact that the chair of the SIB, who is Scientific Research Director for the entire project, is female, is no excuse. We are recognizing the imbalance as a big problem, it is part of our internal discussion on all levels, and we actively work against it. We made some progress, it takes time, however.
To improve on gender balance, different measures are being discussed or have been already implemented. The HBP is fully committed and acting to compensate for the gender imbalance on all levels of the HBP.
You can read more about the plans to act on this issue here:
LS: Prof Marquardt was presented by the EU commission as independent mediator, I myself was educated by a German science journalist once that he has no relation to HBP aside of that. Of course Prof Marquardt is representing Germany at HBP Stakeholder Board, while the new scientific director Prof Amunts works in your FZ Jülich which director he is. Would you agree that FZ Jülich took over HBP control from Markram
using that opportunity?
TL: Please let me state the facts correctly: Prof. Marquardt was proposed by Prof. Markram of the coordinating site of the HBP, EPFL, to lead a mediation process in order to align the HBP Flagship with the European neuroscience. As a result of this process (which also involved many other researchers from the field), profound changes both to the mission and to the governance structure of the project were recommended and
later implemented in the FPA. We all are thankful to both, the EPFL and Prof. Marquardt for their efforts and this achievement. It was later, that Prof. Marquardt was elected SB member by the members of the German section of the HBP, which comprises researchers from many Universities and research institutions.
Prof. Amunts was a work package leader in the subproject on brain organisation and was elected by a majority of her fellow WP leaders to be their subproject leader for SGA1. In the group of the Subproject Leaders, the science and infrastructure board, she then was asked to take the position of the scientific director of the HBP. Other positions like the two SIB vice-chairs were chosen via election as well. These posts are
not set in stone and up for election again for each new SGA phase (every 2 years).
The coordinator of the HBP is EPFL. The elected leader of the stakeholder board, Matt Lambon-Ralph is from UK. i.e., the three chair positions in the HBP are hold by representatives of three different countries.
I emphasize that the processes are independent and, in short, I do not agree with the conclusion you anticipate with your question.
LS: In a recent NZZ article it is indicated that Prof Markram receives little or maybe even no money from HBP, and his SP6 (Blue Brain) is funded by Swiss state with CHF 2Mn annually. Is this correct? Is SP6 funded indeed externally by Switzerland?
TL: SP6, the brain simulation sub-project, of the HBP is carried out by several institutions. The group of Prof. Markram is part of SP6 and thus gets proportional funding. Prof. Markram is in addition supported in the Blue Brain project, which is an independent Swiss federal university project, I do not know the exact amount. Both projects, BBP and HBP, collaborate fruitfully, of course. Switzerland is a fully associated country to the H2020 program. Part of BBP is a Swiss national in-kind contribution to the HBP.
LS: Prof Lippert, many thanks for your interview.
“Blue Brain project, which is an independent Swiss federal university project”
Here are some comments below:
I think that: the Swiss BBP (Blue Brain project) is not directly supported by SNSF (Swiss National Science Foundation. SNSF is the most important and serious organism for the support of fundamental research in Switzerland). At first sight (to be checked) nothing appear under the name of Markram or Bluebrain, even for the last 12 years in the SNSF project database P3 (http://p3.snf.ch/).
I think present fundings of Swiss BBP may be supported by:
SNSF of several “smaller” projects related to Markram’s EPFL laboratory LNMC (Neural Microcircuitry Laboratory): single-cell gene expression, channelome, neuroanatomy, electrophysiology & microcircuits, plasticity, neuromodulation, autism. The “smaller” projects take place at EPFL and/or at Campus Biotech Center in Geneva?
In 2005, IBM BluegeneP super computer was acquired by EPFL, and not by SNSF (both federal money, but not same budget and way to apply). At that time, I think that IBM financed also research. In 2013, Consortium CADMOS with Geneva and Lausanne Universities and EPFL acquired an IBM Bluegene Q super computer, but not only for Bluebrainproject but also for many different research domains (see article https://actu.epfl.ch/news/a-new-super-green-supercomputer/). I think (to be checked) that IBM sold the machine at that time without supporting research of the BBP any more (the moment of when IBM stopped financing BBP may have been earlier than 2013).
BBP appears on the web page of Campus Biotech Geneva Center. This latter was heavily financed by private donor, Hansjörg Wyss Foundation. Did it give money also to the BlueBrain and/or HBP?
-> it means that BBP was never an important federal project compared to other neuroscience projects in the Swiss neuroscience galaxy.
-> the “physical place” of the Swiss BBP is not clear for me: EPFL and/or Geneva Biotech Center? The supercomputer is in Lausanne at EPFL. But how many BBP researchers are on EPFL and Biotech Geneva Center?
-> Even BBP is an eroded project, and it could be be a poisoned heritage for the new EPFL presidency since 2017?
Call me wholly unimpressed!
Those answers to your pointed q’s about the scientific rational of HBP are nothing but buzzwords mixed with a whole lot of BS!!
Step by step Oliver! I now invited Katrin Amunts to an interview, hope she agrees. She is the scientific director of HBP, runs the human brain organization subproject and is a biomedical researcher. I’d love to understand the neurobiology of brain simulation!
Pingback: Il "successo" del Gran Guignol - Ocasapiens - Blog - Repubblica.it
If I would have to evaluate as reviewer a research proposal aiming to simulate a Drosophila brain on a €1bn budget over 10 years, I’d nix them for being overly optimistic. If the aim were to simulate a mouse brain, I’d inquire as to what they had been smoking when they wrote that application and recommend they start taking their meds again. Simulating a human brain? Comparing that with simulating materials and such? I’m afraid I’d actually be left speechless, although probably rolling on the floor howling with laughter…
Pingback: Failed interview with Human Brain Project director Katrin Amunts – For Better Science
Pingback: Interview with Human Brain Project director Katrin Amunts – For Better Science
Pingback: Graphene Flagship deploys Stripy Stellacci to fight the Coronavirus – For Better Science