Interview: What’s wrong with the Human Brain Project?

By Neuroskeptic | July 11, 2014 5:42 am

The hot story in neuroscience this week has been the controversy over Europe’s Human Brain Project (HBP). On Monday, 156 neuroscientists signed an open letter calling for major reforms of the ambitious program, and pledging to boycott it if things don’t change.

hbp logo

Since then, the number of signatories on the letter’s NeuroFuture.eu site has grown to nearly 600. The HBP released an official response on Wednesday.

I emailed Zach Mainen, one of the drafters of the open letter, for comment.


1. In a nutshell, what is wrong with the Human Brain Project today?

There is a core of a perfectly reasonable, if unambitious idea: to build databases and simulations of parts of the brain. That’s being done in many labs already for many years. I did this kind of work for my PhD over 20 years ago. It is nothing revolutionary or paradigm-shifting at all.

Overlaid on that were and are:

(A) A huge amount of salesmanship, which the HBP now admits was ‘miscommunication’ about the true goals and nature of the project. This miscommunication is how the public and European government were sold on this project in the first place. Now that they are admitting this. That leaves us with a €1.2 billion for a project that has very modest and unclear deliverables and is not going to reach the lofty goals it promised. It leaves the neuroscientists who are in the project without a clear goal other than ‘providing data’.

(B) This is combined with a very problematic system for funding and managing science. The HBP is large enough to qualify as a medium to large sized grant program. Yet it is being run in the manner of a small individual lab sized project. The executive committee make the decisions on who receives funding, including how much money goes to themselves. Unlike a grant program, there is no external accountability.

This is likely to breed corruption and is not the way we should be running science. We have better models like the NIH competitive awards or Europe’s ERC program. We need to rethink the mega-project model exemplified by the FET (Future and Emerging Technology) and find better means to help scientists to collaborate. To do this we also need to fix the broken publishing and promotion systems.

2. In your opinion was the HBP flawed from the start, or was it a good idea that went wrong at some point after launch?

Since both (A) and (B) above were built in, it has always been this way. Many of us who joined the project agreed with some aspect of it but were quite naive in retrospect. We took the hype to the somehow acceptable salesmanship that would subside and were not aware of how the process would be run. This could have been known.

Unfortunately, the basic idea of taking grant money almost however you can get it is a rampant problem, especially when involving European consortia, which are large, incredibly bureaucratic.

3. Why has this come to a head now? Haven’t these problems been discussed for some time?

They have been discussed widely, at least the scientific problems. That’s partly why I’m taking the time to go into the policy issues. The reason it comes to a head is that it is up for review later this year so its a chance to do something.

4. What lessons would you say the U.S. BRAIN initiative ought to learn from HBP?

Mostly what to avoid. I think the U.S. made almost all the right moves to be honest. The NIH chose a committee, chaired by Bill Newsome and Cori Bargmann. They got advice from those involved in the Human Genome Project. The committee included a broad range of very respected neuroscientists who consulted a lot more people and they worked hard to come up with a large funding program.

Like the HBP, they chose technology, but they did so in a much more open and flexible process. The resulting program will not be administered by the people who made it up and will be run entirely on competitive calls. Europe could learn a lot from their example, in hindsight of course.

5. You have nearly 400 signatures now [nb: this was sent on Tuesday. It’s now up to 574]. Did you expect such quick success?

Yes. It has been clear that there is a lot of disapproval. The difficult part was coming up with a statement that scientists with diverse views on the problems and solutions could all agree on. We had the first 150, which was the hard part, in three days or so. This was critical. After that it becomes a different thing.

We are happy to have had enormous support from some very influential people who signed up early on and who put some kind of political commitment into it. We also respect those who chose not to sign, but encourage them to speak their opinion on this matter if they feel its important. We will introduce a section on the website for non-signers or non-PIs who want to comment, although there is already a lot of discussion on newspapers and blogs (hopefully like this one!)

6. I was interested when you said: “We need to find better means to help scientists to collaborate. To do this we also need to fix the broken publishing and promotion systems.” Do you see the problems with the HBP as being related to broader problems in science, then?

Yes, which are particularly acute in Europe.

7. Did you (or others) try to achieve change internally within the HBP and find it wasn’t possible?

Yes, we tried via internal channels, including the advisory boards and this resulted in very minor effects.

  • Joanne Williams

    The problem isn’t the management. There are plenty of Big Science funders that don’t follow the NIH model (e.g. NASA, the military). The problem is that it was a stupid idea to begin with. However, it was one of the least stupid of the “Big Ideas” Europe was vetting, which included developing robot companions, and developing AI systems to “guard the planet”. Here are the six finalists.

    http://cordis.europa.eu/fp7/ict/programme/fet/flagship/6pilots_en.html

    Which two would you have picked?

    • Deneb Needs Memes

      Criminy. Most of those proposals sound like they were written by someone who took several ccs of LSD after bingeing on anime and Stanislaw Lem novels.

      • Ibn al-Haytham

        I have to strongly object. Lem is great.

        • Deneb Needs Memes

          Indeed.

          PROGNORRHOEA or prognostic diarrhoea, a children’s disease of twentieth-century futurology (v. PRAPROGNOSTICS), which led to essential prognoses being drowned in inessential ones as a result of decategorization (q.v.) and created the so-called pure prognostic hum. (See also: HUMS, also PROGNOSIS DISTURBANCES.)

    • Ibn al-Haytham

      When I heard the six pilots’ pitches at a FET event, I started to wonder whether there was something funny in the food or those are really the six best ideas Europeans could come up with (probably neither, as the rest of the conf had an astonishingly high bullshit level too, so the whole thing reflects about EU policy makers).

      I would’ve gone with personalised medicine and graphene, the former had the clearest idea about what to do and why would it be useful and the latter seemed to be pure technology, so there would have been at least some industrial benefit.

    • Dennis

      These are all good goals to aim for. They all promise interesting and valuable results coming from simply trying it. These goals, including HBP, are not meant to be the sole purpose of each project. If I recall correctly, they were meant to be somewhat science fiction-ish.

  • Pingback: “The members of the HBP are saddened by the open letter posted on neurofuture.eu” (updated x2) | neuroecology()

  • Bill C

    It seems problematic to emphasize #6 in this context. If there are issues with the specific project on this topic that are in addition to the “broader problems” then address. It’s not that the “broader problems” about publishing and promotion (e.g. open access? tenure?) aren’t real. It’s just that when problems on a specific project get lumped in with these bigger, presumably longer term issues, it starts to sound like just “bitching”. (e.g. a colleague of mine expressed this frustration recently at a feedback session for awardees of a recent large NSF grant, namely that early career awardees were using it as a platform to complain about typical assistant professor issues.)

    • http://blogs.discovermagazine.com/neuroskeptic/ Neuroskeptic

      True. Any emphasis on #6 came from my direction – as these ‘broader problems in science’ are a longstanding interest of mine.

      • melanieaolson

        My Uncle
        Riley got an almost new red GMC Canyon just by some parttime working online
        with a laptop. visit their website C­a­s­h­f­i­g­.­C­O­M­

  • Amade Hossen Khoyratty

    Am 56 years old man , I had an AVR Metronomic Hall Kester 25mm at the age of 28. Four years later i.e in 1991 I got an embolism following a stroke on my right side. I am still suffered my right hand and right leg.
    There is still weaknesses in both hand and leg. I have been taking Eptoiene. Since it’s not available in Mauritius. Drs. from S.Africa advised me to tale LEXOTANIL 3mg in the morning. This tablet is helping my a bit. Over 18 years am taking this medication.
    If ever there is a better medication do let me know.
    Thanking you in anticipation.

  • Ibn al-Haytham

    The greatest problem with HBP in my opinion is (aside from the atrocious concentration of funds) that they are aiming to model the brain without any kind of theory or hypothesis about how or why it should work. They simply doesn’t seem to have any idea, neither well formulated nor simplistic. See Körding saying it better than me here: http://www.quora.com/What-are-the-main-objections-to-the-human-brain-project

    • Deneb Needs Memes

      Körding:

      I believe in the vision of whole brain emulation. I believe that this has got to be one of the most important objectives of mankind. I believe in the demand to make more than incremental progress (and strongly support the US BRAIN initiative).

      That’s a non sequitor. BRAIN has nothing to do with emulation. How is it that no-one seems to be able to keep the two initiatives straight?

      • Ibn al-Haytham

        Well, I don’t really know what’s going on in the BRAIN (he he), so he might be wrong there.

    • Dennis

      Computer modelling of neuronal networks is pretty old and works if sufficient data are available. There is no reason to believe it can’t be scaled up given sufficient data and computing power. IMO

      • Ibn al-Haytham

        Scale up to what? If you are talking about ANNs, modern ones do well on object classification tasks. I could also cite strategy learning results that look nice. How do these kind of functions constitute a complete human brain? And we only talking about the computarional level here, how all this is implemented in physiology is almost completely unknown.

        • Dennis

          Uhm, scale up the mini networks that we have to a whole brain? And talking about physiology, that’s the second dimension of scaling: the model can be based on single neurons functions or even down to receptor function. We aren’t talking about simply reproducing some psychokinetics.

          • Ibn al-Haytham

            You would still have to define the structure of the network, all the learning rules, stimulus presentation paradigms, etc. But I’m not speaking against this approach at all, this is meaningful and is done currently by Google et al. I was criticising approaches that do not define such a computarional framework for the modelling, which reduces the effort to the reproduction of some reaction kinetics.

          • Dennis

            What you need are the connections and the properties of those connections. With sufficient iformation on that level, the functions like learning should emerge in the model. We know from experiments which stimuli should produce which results. If a model that recreates cell properties ends up emulating behavior correctly, then the model is good. But if you have to put in additional assumptions on the function and the mechanisms, then you are engineering something, not modelling reality.

          • Ibn al-Haytham

            Well, those are some high expectations, I can only hope they will be met some day. However, learning and other behavioural stuff is also something you measure, just as you do with channel density. And you have made tons of assumptions when applying a set of differential equations to model channels, how are assumptions about macroscopic dynamics any worse? And of course the goal is not only reproduction but understanding too, I doubt that the blueprint of a modern CPU would convey much information about the logic of its operation to basically anyone, while the abstract specification of computational options would do so (and also about what properties are essential, e.g. you can use either NAND or NOR gates, the microscopic structure will look different, but equivalent computational functions can be constructed).

          • Dennis

            It’s not an expectation, it’s a goal. If you don’t reach that, HBP will not have the use that was proposed.

  • practiCalfMRI
    • Ibn al-Haytham

      Biologists — neuroscientists included — can’t hope for that kind of theory. Biology isn’t elegant the way physics appears to be. The living world is bursting with variety and unpredictable complexity, because biology is the product of historical accidents, with species solving problems based on happenstance that leads them down one evolutionary road rather than another. No overarching theory of neuroscience could predict, for example, that the cerebellum (which is involved in timing and motor control) would have vastly more neurons than the prefrontal cortex (the part of the brain most associated with our advanced intelligence).

      This paragraph, if I understand it correctly, is absolutely wrong.

      • WhitneyRFernandez

        Google>>CLICK NEXT TAB FOR MORE INFO AND HELP

      • Deneb Needs Memes

        No, that paragraph is not even wrong. The paragraph that preceded it, on the other hand—

        Different kinds of sciences call for different kinds of theories. Physicists, for example, are searching for a “grand unified theory” that integrates gravity, electromagnetism and the strong and weak nuclear forces into a neat package of equations. Whether or not they will get there, they have made considerable progress, in part because they know what they are looking for.

        —is absolutely wrong. Grand unified theories don’t include gravity, the term he was looking for is “theory of everything”. No-one is really sure what either kind of theory ought to look like. And progress towards either one has of late been absolutely dismal.

    • Dennis

      Useless piece obviously written by somebody clueless.

  • RKHS

    The US project is a failure and the European Blue Brain project has made great progress. We need more projects like Blue Brain and less like the US individual contributor model.

    The US human genome project was completed, not by the government but by a private corporation using the Blue Brain model. The government backed bureaucratic committee model failed.

    Blue Brain is open source. They are replicating studies that were done under the individual contributor model twenty years ago, but which were not converted into a cohesive, usable form.

    • Dennis

      To my knowledge, BRAINI is still in the phase where people apply for grants. It didn’t fail, nor did it succeed, yet.

      • RKHS

        The Blue Brain project is constructing an executable, open source, computational, software model of the human brain.

        The BRAIN initiative is just umbrella program for funding neuroscience research that is already happening.

        The Blue Brain project has very specific objectives and goals and measurable outcomes. The BRAIN initiative is effectively just a funding agency for vague neuroscience research.

        We already have immense amounts of data, immense numbers of research papers. More basic research and papers will not produce a better understanding of the brain. We have to finish the computational model and brain emulation.

        The problem with this article is that it is praising the BRAIN project, which will achieve nothing and attacking the Blue Brain project, which is the most successful computational neuroscience program to date.

        • Ibn al-Haytham

          More basic research and papers will not produce a better understanding of the brain.

          Then we are pretty much fucked.

        • Dennis

          We won’t be able to simulate a whole brain of any animal based on the data we already have. That is for two reasons: first, up to reccently physiological data wasn’t treated and stored in a way useful for large scale modelling attempts and there was a huge data loss because of that. Secondly, eletrophysiology has reached limits that imaging only surpassed partly and it comes with its own limits. Finally: the majority connections between the neurons are still not known.

          The BRAIN Initiative aims at these problems. And it indeed funds many approaches rather than only one, but that’s not a minus. If you knew about the general funding situation in USA you also wouldn’t be surprised that they use the additional money in such a broad manner.

          Finally, BRAIN Initiative is not a computational project, so I am somewhat surprised by your final words. You compare the empirical BRAIN Initiative approach with the computational HBP and then say, HBP is the better computational project… BRAINI doesn’t aim to simulate brain activity, they want to measure it.

  • http://autap.se Richard

    “There is a core of a perfectly reasonable, if unambitious idea: to build databases and simulations of parts of the brain. That’s being done in many labs already for many years. I did this kind of work for my PhD over 20 years ago. It is nothing revolutionary or paradigm-shifting at all.”

    Obviously Zach has a great deal more experience both with the HBP and with neural simulations than me, but from my understanding of the project proposal the whole point is that it isn’t just a continuation of how models have tended to be built for the past 20 or more years. Modelling in neuroscience is fragmented: everyone does their own thing tailored to very specific ends. You can learn something about a narrow topic from these models, but different models are often difficult to integrate or don’t generalise well at all. As Zach says, some of the kinds of models the HBP would implement aren’t really any different from the kinds he was looking at 20 years ago (even much of the software is the same – though maintained & updated). However, the real value would lie in starting to integrate different models and to have readily available data for model validation. This is non-trivial. The best we currently have for model sharing is ModelDB ( http://senselab.med.yale.edu/modeldb/default.asp ) which, while a very useful resource, is showing its age and is definitely not suitable for these sorts of activities (though Open Source Brain is a step forward: http://www.opensourcebrain.org/ ). Building the infrastructure and tools for data and model sharing *well* – to facilitate new discoveries and integrated models – will be very difficult. It won’t necessarily cause a “paradigm shift” in neuroscience theory, but could well improve the way neuroscience research is done.

    I am sure there are problems with the HBP and its organisation that I am not aware of, but a lot of what I have read in the press/on blogs is in contrast to my understanding of the project’s aims. Perhaps I’m missing something?

    • Ibn al-Haytham

      Let’s imagine a martian trying to understand human society through the written output of it. But in the beginning, he doesn’t know which patterns are scripture, and which are not intended to convey information by any sentient being (e.g. wallpaper patterns or tree leaf patterns). It would not make much sense for hime to create a huge database of patterns together with algorithmic means to generate them, at least not without trying to figure out which ones he will need in the end at least parallelly.

      • http://autap.se Richard

        I think you are suggesting that the HBP’s databasing goals are simply an exercise in data mining. I don’t think that’s the case. The data collection would not be unguided. We already have some useful models (as Zach says) to guide the data collection.

        If you speak to people in theoretical neuroscience, one of their complaints is very often that they don’t have access to the relevant data to help develop/verify/falsify their models. My understanding is that one of the goals of the HBP is to help rectify that situation.

        • Ibn al-Haytham

          This is certainly true. However, we have many models that sometimes contradict each other, and HBP does not aim to create common grounds on the theoretical side. And I think while data accessibility is a problem, for most models measurability is more of an issue. So databasing is all good, for two orders of magnitude less money.

          • angelabeisenberg

            my classmate’s aunt makes $68 every hour on the
            computer . She has been fired for 7 months but last month her paycheck was
            $15495 just working on the computer for a few hours. visit the site C­a­s­h­f­i­g­.­C­O­M­

  • Pingback: Spike activity 11-07-2014 « Mind Hacks()

  • rexreed

    The US program is flawed in a very different way. The NIH funds on the basis of a friends and family plan. Very little innovation driven by funding based on the friends and family plan intended to bank roll well connected highly political researchers who seldom generate powerful results.

    Adopting the NIH method is not a solution. It has been demonstrated that handing out money by lottery would generate superior results. One only needs to examine dementia research to see the negative effects of the NIH method.

  • Pingback: Opinion | Pearltrees()

  • Pingback: Post Of The Week – Saturday, July 19th 2014 | DHSG Psychology Research Digest()

  • Pingback: Public Health in the News – July 20, 2014 | NPHR Blog()

  • Pingback: Exploring the Brain | techstead()

  • Pingback: Do Rats Have Free Will? - Neuroskeptic | DiscoverMagazine.com()

  • Pingback: Do Rats Have Free Will? | AllDigitalNews.com()

  • Pingback: Do Rats Have Free Will?Fresh News Today | Fresh News Today()

NEW ON DISCOVER
OPEN
CITIZEN SCIENCE
ADVERTISEMENT

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Neuroskeptic

No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.

ADVERTISEMENT

See More

@Neuro_Skeptic on Twitter

ADVERTISEMENT
Collapse bottom bar
+

Login to your Account

X
E-mail address:
Password:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »