1887

Of quarks and neurons

An experimental particle physicist's foray into neuromorphic computing becomes part of a vast effort to understand the brain.

As a boy, not only did Karlheinz Meier take apart and reassemble the family television set, but he built his own broadband radio from vacuum tubes. "From the beginning I wanted to be a physicist, or at least somebody working with big devices and instruments," he says. "This is a classical career of the 1960s, I think, when technology was the most exciting thing on Earth. I was certainly influenced by things like nuclear power, the space program, and fast cars."

The first in his family to go to university, Meier ended up becoming an experimental particle physicist before eventually turning to the brain, his main focus in recent years. Today, he is a professor of physics at Heidelberg University and one of the leaders of Europe's €1 billion (roughly $1.4 billion) Human Brain Project (see Physics Today, December 2013, page 20).

Karlheinz Meier. CREDIT: B. Kindler/ Heidelberg University

Karlheinz Meier. CREDIT: B. Kindler/ Heidelberg University

In a telephone interview, Meier told Physics Today's Toni Feder why he became a physicist and about switching research fields. He describes his scientific experiences as "a lot of fun" and is outspoken about his views on science. The tweetable version is, "Interdisciplinary education and democracy in science are overrated."

PT: Why did you go into experimental particle physics?

MEIER: I started to study physics in my hometown, Hamburg. Those were really exciting days—1974, there was the so-called November revolution of physics.

That was the time when the charm quark was discovered. At the same time, the tau lepton was seen. And the J/psi. It was clear that there was more to particles than only the standard matter we are made of. Strange objects like the charm quark and leptons even heavier than the muon. We learned about that at university, and we visited DESY. I still remember seeing the electronic setup in one of those detectors. It was kind of a game changer.

PT: Do you mean it made you want to work in experimental particle physics?

MEIER: Exactly. Today you go to CERN, and it looks very elegant, most of the electronics is hidden in integrated circuits behind boxes. But in those days, the detectors had these sort of free-hanging cables, a huge amount of them, and there were flashing lights. It looked like a total mess. I found it amazing that this kind of system could pick up interesting events out of a collision. I wanted to do that.

PT: You did your PhD research at DESY, and then joined the CERN staff. What did you do at CERN, and why did you leave?

MEIER: At the end of my PhD, the next big machine was coming up. It was the proton–antiproton machine at CERN, in the beginning of the 80s. So I went there. It was a dynamic time. Real discoveries were made. Nobel Prize discoveries, like the W and the Z bosons. And strange things turned up, like monojets, the top quark, and many things which in the end all turned out to be nonexistent at that machine. But at least we learned how to detect that they are wrong. We started to understand how to do physics at hadron colliders, which is difficult because it's a messy environment.

I worked at CERN for almost seven years. My children were born in Switzerland. My wife lived with me there, but she wanted to go back to Germany. And the proton–antiproton days in Europe were over. LEP [the Large Electron–Positron collider] was coming up, but I wasn't interested.

I got a staff position at DESY. They had a new machine, and it was totally different. It wasn't electrons. It wasn't hadrons. It was both. It was an electron–proton collider, HERA.

I think it ran for 17 years. There were no really big discoveries, but lots of important, very precise measurements, especially about the structure of the proton. I spent only 18 months there, because then I got an offer from Heidelberg to take over a chair in experimental physics. I couldn't believe it. It was clear I had to accept.

PT: Was it a big change, going from working at research institutes to academia?

MEIER: Being a university professor is a totally different life. Let me exaggerate: It's characterized by having infinite freedom and no money, whereas at the research institutes it's infinite money and no freedom.

I continued to work at HERA, but with a Heidelberg group. And I started to build an LHC [Large Hadron Collider] group. It was clear that if you want to get funding for your university group, you have to take over a major responsibility. So we decided to work on a data selection system, the trigger. Simply said, it selects one out of a million events at the LHC.

I founded a microelectronics lab in Heidelberg. We got many people involved, because everyone wanted microelectronics chips. Also, students got very excited because of the good job opportunities.

PT: Why did you switch research areas from particle physics to the brain?

MEIER: Particle physics is great. It's an extremely exciting field, and it has extremely exciting instruments. They are doing wonderful discoveries. But if you are in it, it's always the same. The basic ideas are the same. The software tools are the same. The methods are the same. The day-to-day work, for me personally, I didn't find it so exciting any more. After 30 years, you feel you have to do something else.

I got interested in information science in general. Not so much classical computing, but, What is information? How is it connected to matter? How is it stored? How is it processed?

Naturally I talked to people doing neuroscience and I got interested in the brain. People told me there are electronic models for brain circuits. I read about models of neurons, synapses, networks, and it was interesting, and almost obvious—because we had all this infrastructure—to try to build a system like that on microelectronics. It started as an unfunded project at the end of the 90s.

At the same time, I wanted to have something in-house, where I could just go in the afternoon and talk to people, chat, and do simple experiments. Just have some fun. That's how it started. And I never intended for it to become big. But it started to go very well. We wrote papers, and people asked, why don't you ask for funding for this, it's an interesting topic.

PT: So how did you get involved in what has become a future and emerging technologies (FET) flagship, the Human Brain Project?

MEIER: We started some projects at Heidelberg. Then in, I think, 2009, I met Henry Markram. We had met before, but only worked loosely together. He is a very dynamic person, and he said, Let's [apply for an FET flagship] together. The flagships were supposed to be Moon landing–type €1 billion projects.

PT: What are you building for the Human Brain Project?

MEIER: We build this big thing, which is a system of 20 silicon wafers. Each carries about 50 million artificial synapses and 200 000 neurons, so in total we have a billion artificial synapses and 4 million neurons.

The system will be able to do interesting experiments of all kinds of artificial circuits. These circuits have very interesting features. They are very, very energy efficient. They are scalable. They are fast. They are compact. And they can learn. So they have all these wonderful things that our brain has. Our brain is extremely energy efficient, it doesn't need software, it is fault tolerant. So the obvious idea is that if you could transform this into an artificial system, it would probably be quite a game changer.

We have a whole set of applications we are working on. One is to take circuits from neurobiology, like insect brains, or subcircuits from rat or mouse brains. We implement them, and see how they behave.

There are theory papers, where we can show that neurons can actually sample from and take decisions based on probability distributions. That probably plays also a very important role in the real brain. For the first time ever, we have sufficient funding to build something really large. It should be ready in a bit less than two years. I am back to my dreams as a boy!

PT: Was switching fields difficult?

MEIER: It was easy for me. The reason was, it was a soft switch. I started with a little bit of the new field, and a lot of particle physics. And then there was a soft transition, which took almost 10 years.

In experimental fields, you have to find money to switch. If nobody finances you, it's just a dream.

Particle physics is a very well organized field, but I always found the meetings not so inspiring, because everybody kind of agreed—at least in experimental physics. It's always nice to fight a little bit scientifically from time to time. That is what I enjoy in this new field. When I go to meetings, I basically don't understand a word and people contradict each other all the time.

PT: How does particle physics experience play into your new work?

MEIER: I could not have done [the transition] without using all that I learned in particle physics. It helped technically and sociologically. The microelectronics is the same. And these people [in neuroscience] have never, ever worked in large collaborations. In fact, for the Human Brain Project, that is one of the bigger challenges. We have to show that it works.

PT: You are now in a very interdisciplinary area of research; what advice do you have for people interested in bridging more than one traditional field?

MEIER: Many people talk about interdisciplinarity, which is very nice. But it can also be a big mistake, because what people do now in some universities is special study courses, interdisciplinary study courses. They learn a little bit of chemistry, a little bit of physics, and a little bit of mathematics.

Some of them come and want to join our group. I talk to them, and ask what they can do, and they claim they know everything. But when I talk to them in some detail, they really are not good at anything. For our project, we need excellent mathematicians, excellent physicists, excellent biologists. They have to be good in their own field, and then they need to be able to talk to each other.

PT: Anything else you would like to mention?

MEIER: Experimental science is very much about leadership. At universities, there's a high value put on being democratic. I think science should not be democratic. A friend of mine told me that what science needs are "enlightened dictators." Those are top scientists who take decisions on their own, without an infinite number of committees and selection groups and things like that. Those are people who sort of drive the field. They really push things forward. They do it for the good of science, not for the good of themselves.

Comments

Submit comment
Close
Comment moderation successfully completed
This is a required field
Please enter a valid email address
ec53185b92e297f26a707ada48e7a20c ptol.magazine_postzxybnytfddd
Scitation: Of quarks and neurons
http://aip.metastore.ingenta.com/content/aip/magazine/physicstoday/news/10.1063/PT.5.9013
10.1063/PT.5.9013