The Conversation
I. Revolutionary Rumblings
Beyond 1:1 maps
HB: Let me start with rats, because that seems to be where things started with your research, as you describe it in your book, Beyond Boundaries. So tell me about that, tell me about your first work that started with John Chapin, right?
MN: Yes. At Hahnemann University. Well actually, very few people know—and that’s one of the reasons I mentioned it in the book—that this whole field of brain-machine interfaces actually started with these experiments in rodents.
John was in Philadelphia looking for a postdoc to help him and I was graduating in Brazil at the University of Sao Paulo. And we both had the same idea independently of what we wanted to do in the future. He put an ad in Science that was not supposed to be answered, because it was one of these ads—
HB:—where he had someone in mind already?
MN: Yes, because he had someone in mind already. But he had to put up the ad to justify hiring this person. And I answered it.
HB: What a bummer for the other guy.
MN: Yes, exactly! He got hired later anyway, as it happens. So I answered it and John just called me straight away. It was the late 1980s, and back then it was very rare to get an international phone call in Brazil, particularly at the university. Anyway I got this phone call and I could barely speak any English, but John was all excited on the phone saying that I should come to Philadelphia for an interview, because what I wrote to him was exactly what he was thinking about.
It was this idea of recording from large numbers of neurons, many neurons simultaneously. Because until then the dogma was that we would just record one at a time, out of billions of neurons in the brain, and you would find the answers.
And each of us—for different reasons, that’s the very peculiar aspect of it—came to the same conclusion from very different directions. We wanted to look at what a population of neurons does when it fires together.
So John had started pursuing this, both as a graduate student and as a postdoc. He is an electronics genius, and he was building his own system. I was doing something different, but I really wanted to devote my entire career to that. I knew that’s what I’d wanted to do since I was a graduate student—a medical student, actually. So we got together and we started developing this technique to record from multiple brain cells simultaneously.
And the rat model was perfect. In particular, the whisker model, because there was a lot of physiological work done on anaesthetized animals, and there was this suggestion that you had these maps —these precise maps of the facial whiskers of the rat at different levels of the brain—and that in these maps each whisker was represented very clearly by a cluster of cells.
HB: There’s a one-to-one map from the regions to the actual neurons.
MN: Exactly, a one-to-one mapping. An isomorphic map, we say: the distribution of the rows and columns of whiskers in the face, into a three dimensional map at each level of the trigeminal pathway, the pathway that conveys tactile information from the face all the way to the cortex.
HB: Just to back up for a moment: I’ve only very recently become intimately familiar with sensory capabilities of rats. They use these whiskers to detect all sorts of things, right? They use them as a sensory device to get a sense of where to go and whether an opening is large enough or an opening is small enough—
MN: Like your fingertips.
HB: And you mentioned something before about being anaesthetized. And one of the interesting things I learned from your book is that previously when people were trying to study neural behaviour, neural pathways and signals, the actual subject was anaesthetized. There wasn’t this idea of having a live subject that was there in front of you that could be probed.
MN: Yes. Most of what we knew in the beginning, in the 50s and 60s, was done in anaesthetized preparations: that’s how we studied sensory systems like the visual system, the tactile system, the auditory system.
In the mid 60s a researcher at NIH, Edward Evarts, created a preparation to record from individual neurons in awake, behaving monkeys. And that was a big revolution, but it was still one neuron at a time. And people do it to this day, the preparation has survived for six decades: it’s a very useful preparation.
But we wanted to know this: since you have these neurons at different levels that represent the same whisker, how do they fire? How do they respond to a tactile, mechanical stimulation of the whiskers in an awake, behaving rat?
HB: So you’re able to put electrodes in the rat’s brain, and you mentioned how you were inspired by Joseph Silk’s book about the Big Bang to measure the neural electromagnetic signals in a distributed way by putting the electrodes all around its brain.
MN: Yes, I’ve long been fascinated by astronomy. And during medical school I read his book on how they linked the radio telescopes in England to measure the radio signals.
HB: So this is a form of interferometry.
MN: Yes. Of course we couldn’t mirror the entire system, but the logic was very similar. So I wanted to see how different sources —neurons—interact, how they work together to underlie some perceptual capability, in this case tactile. And rats love to use their whiskers to judge openings in a wall to see if they can run through it, which is why I like to compare it to fingertips.
HB: So this is what you suggested to John Chapin: this idea of this distributed measuring device, putting these electrodes all over the rat’s brain while the rat is awake to be able to measure the signals in real time, or close to real time. This was the idea, right?
MN: Yes it was this. And the key aspect there was to record from multiple structures simultaneously, which remains one of the key features of our laboratory to this day. Twenty-five years later, we introduced this approach of looking at multiple structures simultaneously and lots of neurons in each structure at the same time.
The first night we started we recorded 26 neurons simultaneously. For us, that was out of the universe, because people were recording 1. And 25 years later we’re recording 2000 neurons simultaneously. I have a plot of how this progressed over time: it’s really exploding now because of technological advances.
But the idea was to test a theory that was out there. It seemed very logical, very clear, that each neuron in each cluster of neurons that represented a part...