Richard Hahnloser leads the Birdsong Group at the Institute of Neuroinformatics, University of Zurich. His lab performs theory-driven systems neuroscience to investigate brain functions that can be characterized by a computational goal encompassing sensory inputs and motor outputs. This is achieved using behavioral experiments combined with electrophysiology, pharmacology, brain stimulation and correlative light and electron microscopy. Examples questions pursued by the group are: How do brains support the learning of complex tasks? How do neural circuits process sensory and motor signals? Is neural information processing optimal in some sense?
Richard Hahnloser gave a seminar at the Champalimaud Centre for the Unknown on the 19th of November and was kind enough to grant us an interview.
Do you remember the first time the idea of being a scientist crossed your mind?
Not exactly, but somehow as a little boy, maybe around six years old, I knew that I wanted to go to the ETH Zurich (ETH Zurich is an engineering, science, technology, mathematics and management university). This was where technology was and at that time I was mostly interested in skiing gondolas, these cabins that go up and down on mountains. I’d heard that the ETH was where you learned about such stuff, how to build such things. Eventually also when I was in an airplane for the first time: I became super excited and while my parents thought I’d be interested in visiting the cockpit to see how the plane was steered, I was actually just excited that you could lift something that big into the air, and interested in how this was possible.
Later on in life, when did you actually decide to become a scientist? Did you have a plan, were your career steps goal directed?
No, not really. I come from a family with an interest in art and art history, and so I was working for artists a lot while I was a high school student. At that time, moving in artist circles, made me consider becoming a musician but I was not a good enough pianist; I also thought of becoming a writer and I even handed my texts to a well-known local writer in the town where I grew up, but I soon also realized that fiction was not for me. I ended up studying electrical engineering which involved learning too many norms (the DIN norms of mechanical drawings) and these types of things. So I came to the conclusion that this wasn’t for me either and, very quickly, I switched to studying physics, and I was very happy there.
I guess I would select this topic again if I had to study once more nowadays even though it’s not what I’m currently doing!
While studying physics why did you decide to pursue a career in academia rather than striving for something more applied?
I was absolutely fascinated by theoretical physics, such as quantum mechanics and relativity. In high school I read about these topics and my mind was blown away by thinking about space-time issues, I really enjoyed thinking about these things. I wasn’t such a great student however until the very last exam when I realized “oh, now I have to have good grades otherwise I cannot become a PhD student”, so I really only performed well on the very last exam, allowing me to begin research.
And why neuroscience?
That was a very hard decision for me because I loved physics so much. However, I did practical work at a particle accelerator in Switzerland and this was very frustrating for me because I couldn’t get in touch with any of the real science. I felt I wasn’t engaging in actual research, I was just there working next to huge magnets doing nothing really enlightening. When I encountered neuroscience I had the impression that people used very simple tools, and that you could do neuroscience almost anywhere, even at home in your garage. I thought it was much more accessible than the particle physics I had been interested in. At this time I also read a text by a famous theoretical physicist, Steven Weinberg, where he described how a student had come to his office and asked him what research he should be doing; Steven replied “well what do you think you should be doing?”; the student was torn between two fields, one was not much explored and the other one very heavily explored; the student answered that he would choose the explored field with many well-formulated questions; Steven said “You just gave me the reason for choosing the other field, the one that is underexplored”. This can be hard for a student to realize, hard to imagine what could be there if you haven´t read and heard much about it, it’s just easier to pick things that are popular.
Is Neuroscience nowadays still very different, or is its approach moving a bit in the same direction as research in physics, with a big increase in the technology involved?
Neuroscience has definitely become more technological with all these new tools, that’s clear, but I have the impression that many people are still very clueless about what they should be doing, they are just collecting facts. I perceive neuroscience at the moment like some fields of botany, focusing on descriptions rather than on concepts. In contrast, in physics there are theories and a few equations, and people are still testing them today, they remain valid. There is no such thing yet in neuroscience. I wonder how neuroscience theory will look like in the future: it will probably not involve just a few equations that govern everything but it will maybe revolve around gathering data in such a way that principles can emerge or be easily tested. For me this is the biggest unknown, how theoretical neuroscience will look like in the future, because it’s clear to me that the need for theory will only grow with the increasing wealth of facts we collect. I look forward to seeing how the important theoretical concepts will emerge.
Is there still space for the one person – one project model of research, with an individual scientist designing the experiments, collecting and analyzing the data?
This type of encompassing individual work model will probably become more of an exception I guess, as people increasingly specialize. However, we are lucky that the tools we use in our research are still simple: we don’t need a Hubble telescope or a 30 Km accelerator. As long as the tools are very accessible and easy to operate, as long as things remain this way, I would say an individual can successfully combine many approaches. To a large extent, I think neuroscience can remain a one person show for some time.
WHAT DO YOU THINK THAT YOU, OR PHYSICISTS IN GENERAL, CAN BRING TO THE NEUROSCIENCE FIELD?
I think physicists bring mainly mathematics. We like to frame things in quantitative terms since we have a natural tendency to be reductionists and try to find governing principles. This contrasts with those biologists who have a tendency to just appreciate and emphasize diversity. I think both approaches are important in the end; but this reductionist theory approach is required, as I believe that there will be some kind of set of simple principles that will dictate a large body of experiments.
Which questions fascinate you most in neuroscience? Why did you choose to work with zebra finches and use birdsong as a model to formulate your questions?
I like questions about how biological systems work, about the algorithms, the information processing part; I like to see to what extent we can compare brains with man-made concepts. For example, I investigate how birds evaluate their singing performance and I try to understand whether we can compare their strategy to computational algorithms such as linear assignment algorithms, or whether birds have an inverse model, which are very powerful controllers in robotics, for instance. Another example comes from research I like to follow in the literature on deep networks, which addresses whether deep networks (extremely good at classifying images) can be used to describe brain functions. I really like these analogies with artificial intelligence and machine learning.
As for how I ended up in my line of research, there are two main reasons I would say. One of the reasons is that I met the people I ended up working with casually and just enjoyed discussing their science with them, and so the work projects just emerged naturally. This happened with my two postdoc advisors as well as my PhD advisor: I just liked interacting with them, found them inspiring and wanted to learn from them. In essence, it just happened: if some of them had studied something else I might have done something else, so I chose, to a logic extent, the people. The reason for studying birds however, is also related with the fact that when I was starting there had been a major discovery: the first recordings from populations of neurons in singing birds had been performed and the neurons exhibited a temporal firing precision in the song on a sub-millisecond time scale – this sub-millisecond precision in the brain of a behaving vertebrate was just unbelievable. I thought this was a fantastic place to look for brain functions. Typically what I had learned from cortex is that neurons are Poisson point processes and that there is noise all over the place and suddenly there is this place where everything is very stereotyped and precise. I found this precision to be very attractive.
Neuroscience nowadays is a very big field, ranging from molecular biology to experimental psychology. How do you find the interaction between the different fields? Does it exist? Is it optimal? Do you have an opinion about this?
I do have an opinion on this matter: I don’t really believe in bottom up simulation without principles, like to take molecules, put them into synapses and put electrical-chemical signal transmission into networks and just assume that things will fall into place – this is not the approach I would choose. I would call this reverse engineering. The approach I’m trying to take is forward engineering, which entails looking at behavior and describing it using the simplest possible reductionist model, and then looking at how brain manipulations affect the animal and thus the simple model. I try to always stick close to the behavior and focus a lot of attention on understanding it.
Do you think that we will be able to explain the human mind in neurobiological terms?
Probably not, but not because the technology is not there, rather because I imagine the experiments you would have to do are unethical. I might be wrong though, but in principle, I would say that if we understood all the principles about neurons, connections, memory, recalling of memories and all these things, the reduction approach should in principle work.
do you have an idea of how these principles would work, what they would be?
Well, I don’t know but of course I can just tell you the principles I like. I like principles, such as social principles: if neurons were little units that cooperate and compete and there is something equivalent to money, maybe some molecule, they try to pass it on, fight for it and store it; I believe that some kind of economic ideas would be very powerful, but that would need to be explored.
If you could change one thing in science, what would it be?
Probably the way we publish. I think the internet could allow completely different mechanisms of publishing. It is very inefficient having to fight with two reviewers for half a year or a year; it helps nobody, it’s a waste of money and time, and it should be changed. I think that whenever people have a finding they should just publish it, and not, like it is done nowadays, wait to have four figures and then piece together a story with an introduction and discussion. I think we should forget about this approach and just put our figures sequentially out there, as we uncover new results; if at some point we can link these into a discussion that is fine, but in general I think there should be a more honest, less storytelling way of publishing data.
But don’t you think that we humans are very addicted to stories and narratives that put facts in perspective?
Yes, and you could always publish the story at any time a bit later, but it doesn’t have to be published in the same data package as each individual observation. When you publish a paper, it develops a life of its own that you can’t control…suddenly your paper is influential for something that you didn’t think of. Nobody can predict the future and that should be respected; that’s why the data should be published as it is and then, later on, you can always tell stories about it, whatever you want.
Wouldn’t that make our lives as scientists more difficult in terms of keeping up with the literature and all the knowledge out there?
I don’t think so. I think that if science is published in little piecemeal quantities instead of just having one big paper full of narratives, you can have the motivation behind your findings and the references that agree and disagree. Then, search engines can just pull up everything that links with your findings, regardless of whether it supports them or not. I think search engines will just get better and you’ll have the information that you want at your fingertips, in no time.
What would you choose to be if you could live an alternative life being something else than a scientist?
I think I would have my own company, doing something, selling something, some kind of science technology related instrument…maybe there’s a life for me beyond research and teaching…but I don’t know….
Do you see teaching as burden side-product of a career in research?
No, teaching can be a lot of fun, but it’s really fun when you devote a lot of attention to it. You do have to do things fully to really appreciate them, and I like teaching because I’m always trying to learn something new. But there is also a downside of it: it takes time and you have to grade work, meet with students who need lots of help and sometimes I wish I had a bit more time to focus on some other things.
In your opinion should scientists have more time to think? Would you like to have more time to just think about your research?
So, time is really very important, you have to really be on top of it. Like one of my friends says, time is your worst enemy, the one you have to dominate. I have enough time to think, I think I take the time, but it is true that nowadays you have to do many things. Even for your CV, if you’re young you need to have done social work, you have to be community driven and all this stuff. In the old days, apparently, they had conferences with only one talk per day, or maybe two, so they took much more time for discussions. I guess it’s a bit sad that having time to think might be disappearing. Sometimes when reading old papers, it is mind blowing what they could do without computers, just by thinking. Today we have all these tools, but does it really mean more? I really would like to question that! Maybe people used to be better thinkers than today because we are so diluted in tools and technology and have to spend time on them.
João Afonso is a PhD student in the Circuit Dynamics and Computation lab at the Champalimaud Neuroscience Programme.
Edited by: Márcia Aranha (page editor), Clara Ferreira (editor-in-chief)