By: Nils Muhlert
Professor Marcus Raichle has played a truly pivotal role in the discovery of the physiological basis of functional neuroimaging. During the 1980s he helped to discover the relative independence of blood flow and oxygen consumption during changes in brain activity; in the 1990s he identified the ‘default mode’ of brain activity; more recently, his team carried out critical work into the infraslow activity of the brain. Marcus is currently the Alan A. & Edith L. Wolff Professor of Medicine and a Professor of Neurology, Radiology, Neurobiology and Biomedical Engineering at Washington University School of Medicine in St Louis, Missouri. He took time out of his busy schedule to tell us about his seminal studies on measuring blood flow and exploring the rhythms of the brain.
Nils Muhlert (NM): First, can you tell us about how you became interested in neuroimaging?
Marcus Raichle (MR): I didn't start out my life aiming to become a doctor or a scientist. But the bug of doing research was planted dang early on.
It started with Fred Plum - he was my mentor in neurology. I was introduced to him in medical school in year one in our neurophysiology course. Plum said: "I'm not training you to be a clinician, I'm training you to be a researcher." Most of his pupils became clinicians anyway but you were surrounded by really neat research and I got hooked into it. I had a rotation with Plum in my third year and was soon enamored with this business of what the brain was doing. I became interested in how metabolic diseases affected the brain. The beauty of that idea was that, as opposed to much of Neurology, it was something that was reversible. I got into this with both feet at that point as a resident.
This introduced me to people like Kety, Ingvar & Lassen. There was no organization, no journal, it was very informal. I went to a meeting, and at that meeting Neils Lassen said: “let's go have a drink.” We went to a bar, he said, "it's your turn to host this meeting." That's how it worked.
In 1981 we hosted that meeting in St. Louis, and all the heavies of the world in brain circulation and metabolism were there. That meeting was where the International Society for Cerebral Blood Flow and Metabolism was formed. Louis Sokoloff was elected the president and a journal, the Journal of Cerebral Blood Flow and Metabolism (JCBFM), was created.
Ingvar and Lassen were particularly interested in function and blood flow. They started with a small number of detectors over the head and the number increased. There was a famous study on 12 Swedish firemen. In those days you had to put a needle into the carotid artery (it would never pass an internal review board now). They used this technique to demonstrate focal changes in brain activity with blood flow. Ingvar & Lassen then developed a Xenon technique with inhalation and they got up to 256 probes over the head.
But this also came on the heels of a story that goes all the way all the way back to 1878. William James, in his principles of psychology from 1890, has a whole chapter on the brain. Within that chapter he describes this marvelous experiment by Angelo Mosso - the first cognitive neuroscience experiment ever published. Angelo was studying people with bony lesions in their skull, some with syphilis. One Italian construction worker was hit on the head by a brick that fell on a building site. He had this defect, like the fontanelle of a child, where you could see the brain pulsating. Mosso had been part of a group that was interested in brain pulsations. He was in the lab one day and he had this guy Fertino with this pulse meter and blood pressure. At noon, the church bells rang and the clock in the room chimed. When that happened the pulsations over the brain in the frontal cortex went way up. Mosso turned to this fellow and said, “Well, should you have said a prayer,” as was customary, and by God the pulsations went up again. Amazingly, he asked him to multiply eight by 12. When he asked him this, the pulse went up and a moment later he answered the question and the pulse went up again. This led to this book, which I translated, on the circulation of blood in the human brain.
Mosso’s work was, however, brushed aside. A very famous British physiologist at the time said: “This is bunk. This can't be true” - a typical scientific rebuttal. When others started to record from single neurons with electrodes, the story of the pulsating brain was forgotten.
Many years later Harvey Cushing, the grandfather of American neurosurgery, admitted a fellow who was losing his sight. It turned out he had an arteriovenous malformation over his visual cortex. They operated on him and did the best they could to remove it. And of course, it left a defect. The patient reported that every time he looked at or read something, he heard a whooshing sound in his head. So they laid him on a chaise lounge, put a newspaper on the floor underneath him, and asked him to read it. They recorded the sounds from the back of his brain and heard a change in blood flow. This is the first demonstration of the effect of attention on brain blood flow, which is fundamental to an understanding of the BOLD signal. I love this because this stuff just kept getting lost in the dustbin of history.
NM: In this oral history series we always ask about the background to their neuroimaging work. You’re certainly the first to go as far back as the 19th century (laughs)!
MR: I really love it. The historical perception of all of this helps us understand how science evolves. It explains how people think about recording electrically from the brain, the neuron doctrine and understand how this dominates a story. Everything else gets pushed aside. Then every so often, something comes along and says, "Wait a minute."
NM: And turning to recent events - what do you see happening with neuroimaging in St. Louis and in the US more broadly now?
What's fascinating to me is the almost exponential increase in the sophistication of what we can observe in the brain. We started with using fluorodeoxyglucose (FDG) PET to measure metabolism. Then came blood flow and cognition. All of that was seminal. Our 1988 language paper was the culmination of how you did it. But then there was this idea that there's lots going on in the brain all the time - the so-called spontaneous activity, the “brain's dark energy” - that raises the issue as to what's going on.
I have to backtrack slightly. We were doing this with PET - measuring blood flow by using an isotope that decayed so fast, that in 10 minutes whatever we had injected was gone. We were interested in mapping the brain like Neils Lassen and David Ingvar, except with PET. One of the big surprises in all of that was two experiments we carried out in the late 1980s.
The presumption in Ingvar & Lassen’s work, and all the way back to Mosso, was that if the blood flow in the brain went up, it must be because the brain needed more oxygen. This makes complete sense: if I shut off all the oxygen available to me, I have about 15 seconds to say my last words. But the evidence for that relationship between blood flow and oxygen simply didn't exist.
Neils Lassen said to me “we need some evidence for this. If you work on the relationship between blood flow and oxygen, then we'll invite you to this prestigious meeting in Copenhagen.” So we carried out these internal carotid injections. Through this we found a correlation that everybody expected us to find: patients showed decreases in blood flow and decreases in cerebral oxygen utilization. We published it, and it was my most cited paper. As we got into the PET world further, it seemed to me that we ought to nail this one, we ought to do a full size PET experiment where we measure metabolism and blood flow, and we stimulate them. We needed real data, but then we couldn't reproduce it.
We would stimulate the visual system with a checkerboard or buzz the hand with a vibrator, and the blood flow would go up, just like we'd expect, but the oxygen consumption didn't. Then we did a second experiment where we looked at the glucose consumption, and it did go up. The reception to these papers was quite something, people thought, “This is terrible science, it can't be true.” There's a quote from Lou Sokoloff where he says, “This breaks the rules of biochemistry.” But it turned out that this was the foundation for fMRI, that set of experiments.
Later, I received a call from Seiji Ogawa. I'd never heard of Ogawa or David Tank or the people at the Bell Labs, but I received this invitation to give a talk there. I went and gave my talk. Seiji Ogawa took me over to his lab and showed me his rat experiments. They looked at what happened when you stimulated a rat and saw the very same thing. They reproduced our finding but in an MRI scanner. I thought, “Holy smokes.”
Ogawa published his work and called it the blood oxygen level dependent (BOLD) signal. The people at the Mass General - Bob Turner and that crew - they independently came to the same conclusions. Neither group knew the other were doing it. There were four papers published simultaneously in one year that basically said, this is how you do it. One from the NIH, one from Germany, one from the Mass General and one from the Bell Labs. They all came out together in one year.
At that point we had access to a far more sophisticated tool. PET still has a role in all this at a metabolic level, but as a brain mapping tool, MR is clearly the way people do it. It's almost dangerously too simple. You have access to the scanner and of course, through the major producers, Siemens, General Electric, Phillips, you have software that means you don't even have to think about it, you just acquire images. The issue that came to the forefront was: How do we think about this? What is the BOLD signal?
MR is still a metabolism and blood flow technique and therefore it's an indirect measurement of what the brain is doing. Various people suddenly got involved to understand the BOLD signal. Nikos Logothetis in Germany was probably the most visible. This provided a summary of what the brain was doing. There was nothing special about it, it was just a low pass filter of all that electrical activity. Since there were blood vessels involved, there was neurovascular coupling. But that didn't diminish the use of MR. People were continuing to make interesting discoveries with fMRI using all kinds of cleverness. And then along came the spontaneous activity story.
NM: And can you tell us about your role in defining the default mode network?
Well you have to go back again to another person: Bharat Biswal, a grad student at the Medical College in Wisconsin. His mentor, Jim Hyde, was a senior fellow in the MR world and they were curious about the noise in the BOLD signal. He carried out the world's simplest experiment: He had people move their hands so he could see where the motor cortex was. And then in the resting state said, I’ll look at the signal in that region to see if it was correlated with signal elsewhere. And, by God, it was correlated with activity in the motor cortex on the other side.
We had a big MR lab right next to our lab. Colleagues of mine asked what I thought of this discovery. It was interesting. Various people looked at this in St. Louis, and it was hard to reproduce. There was some skepticism about it. It just lay in the background for me (although we’ll return to this later).
In the meantime, this default mode story emerged serendipitously. We had been running cognitive studies, and the whole idea was that in order to study behavior, you have to be in control of that behavior. You want to be very precise about what you do. The experiment that epitomizes this was the processing of words. You want to know how you read a word, or how you say a word. So the first thing you want to know is what part of the brain is involved when you look at or hear the word. We had been in the habit of acquiring a resting state scan, where you did nothing, with each of our scans. It was a matter of habit. There was no master design.
I remember I looked at the difference between doing nothing and looking at a word. What I noticed was that not only did things go up, but if you flip the subtraction around, something went down. Specifically, something in the middle of the back of the brain. So I created a folder and started looking at this region. It was showing up in all these different experiments. I had a folder in my office called MMPA: medial mystery parietal area.
Gordon Schulman visited my office with Mike Posner, they were working with Maurizio Corbetta and had a great interest in attention in the brain. As an overarching part of what the brain does, when I'm attending to you, and we're talking, there are certain parts of my brain that are particularly active. They wanted to show that this was true across all sorts of different experiments. So they did a meta-analysis of nine of our PET experiments. They didn't initially find this attention system, which is now well documented, but they noticed a region at the middle of the back of the brain that was going down every time. Gordon brought that to my attention and I gave him my MMPA folder. We published a paper that shows that it decreases. We got interested in it and pursued it. This wasn't an artifact but people were very critical. So we wrote a paper that was summarily rejected.
We, the authors on that paper, had a meeting one day. We were trying to figure out how we would get beyond this problem. It seemed like what we were looking at had to be real. It wasn't an artifact. I was walking back down the hallway to my office and it suddenly hit me. I thought, “It has to do with the definition of an activation.” When you have an activation, you have a dissociation between oxygen consumption and blood flow. I turned to a friend of mine, Colin Derdeyn, who had been collecting a lot of resting state data for metabolic reasons. I asked if he had data like this that I could use, and he said, “Sure.” I said, "Did you ever look at it?" - “No, we never looked at it.” So I put it together. It was very clear that the areas that became the default mode network were active in the resting state. And that was the 2001 default mode of brain function paper that has now been cited well over 10,500 times.
But there's an addendum to the story, because you had Biswal in the background and the default mode paper, which was pretty unassailable at that point. And then along comes Mike Grecius from Stanford. I was at a meeting in Sendai, Japan. I was by myself wandering through the poster room and there was nobody else in there. I saw a poster by Mike Grecius. He had carried out the Biswal experiment, but instead of putting the region of interest over the motor cortex, he put it over the MMPA region. And there was the whole default mode network. I contacted him, and said you need to publish this. I was the editor at PNAS and I said I'll see to it that this gets published.
That introduced the default mode network in a way that was pretty defensible. All of a sudden, we could interrogate the organization of the human brain at this very large scale level without asking anybody to do anything.
Lurking in the background throughout all of this we had this question - what was this BOLD signal? Where does it come from? That fascinated me for a long time. If we were going to get taken as serious scientists in the neuroscience world, we had to know what that signal is. Lots of people, including Logothetis, were working on it. But the impression I kept getting was this is a summary measure.
One of my graduate students, Jin-moo Lee, became involved. We collaborated with neurosurgeons with electrodes on the cortex and ultimately got serious about this with mice. This was one of those immensely satisfying experiences where graduate students at Washington University were talking about this and brought together four major labs: the optical imaging people, single unit people, a cell biology lab and our lab. Together we rustled up money for the experiment that was later published in neuron in 2018. This is the story of infraslow activity: how it relates to the ongoing excitability of the brain and how it's organized.
It's amazing, but every time you think you've discovered something you find out that people alluded to it before. It turns out that there was a prescient article from a Russian physiologist, a woman named Aladjalova. In 1955, she published a letter in Nature on infraslow brain activity. Then she wrote a book—which we've translated and I want to publish—where she anticipated a good bit about this and yet it just got swept under the carpet. It too ended up in the dustbin of history.
We are moving closer to a sophisticated understanding of what signals from fMRI tell us about the human brain. We often talk about mice, rat and even monkey models of, say, depression, but ultimately you'd like to be able to relate it to the human. And, for the first time, we're talking about the whole brain. We're thinking about the whole brain being involved in everything we do. It doesn't matter if you're recording 10 neurons, 100 neurons, 1000 neurons or 10,000; there are 100 billion of them. Otherwise it's like going into Times Square, pulling out 10 people and saying, I can figure out what New York is all about. This isn’t to devalue the basic neuroscientist’s science, but we need to work together.
When we're talking about a mental illness, we have a plethora of questions about serious problems. But the answer to those problems is not a hole in your head somewhere. It's not your cerebellum, it's the relationships between brain regions. People look at the resting state in a static sense, but now we can talk about traveling waves and even turbulence. Our ways of looking at brain function are maturing. And we’re applying ever more sophisticated techniques across the levels of neuroscience. We can't discard the firing rate of a neuron. But at the same time, we have to consider the astrocyte and the oligodendrocyte, how all these cells work as a community.
NM: And now there’s increasing clinical applications of resting-state network imaging. I know your former student Michael Fox is very involved in some of that work - that must be nice seeing that coming out...
MR: The idea that you can think creatively about diseases is interesting. Mike is one of my great graduate students - I love Mike in many ways because I love ideas and I always remember going up to Mike’s cubicle and saying, “what do you think of this.” and he’d say, “I don’t buy it.” You learn!
I just sat through a meeting organised by Steve Hyman at MIT on how we think about mental illness. I ended up giving a keynote address which spooked me a bit. Here I was, surrounded by all these geneticists and basic scientists. The fact that we’re having conversations across these levels about really sophisticated problems is deeply interesting. I think that imaging will make a contribution.
If you take Alzheimer’s disease, there we have a pandemic of serious proportions. With imaging we can now see the plaques and tangles. But we’ve lost sight of what the basic problem is. Of course, plaques aren’t good to have and tangles are the predecessors of dementia. But we need to ask the deep question of why it behaves like it does. One of the attractive issues that we tend to focus on is the fact that Alzheimer’s is primarily a disease of the default mode network. So why is that anatomy important? You can then ask, “Why do we have amyloid plaques? It’s like putting salt into a glass of water. If you keep putting salt in there and get the concentration high enough then you get salt crystals. Same thing for amyloid. You ask, ”Why is this happening?” You get into the whole issue of activity within that system. You end up in the dialogue between the hippocampus and the cortex. This is a very beautiful but complicated relationship where activity at times can be explosive. And if it gets out of control then the salt crystals accrue and you have a problem.
The hippocampus gets smaller in people with Alzheimer’s. It gets smaller, but its metabolism is going up. Something is amiss. You can put together a very interesting story that you’ve basically lost control of this part of the brain - but to understand why, you need to combine basic neuroscience about the physiology of the hippocampus, cell biology that tells you the story of amyloid, what goes with it, and what it does when it’s out there besides making plaques. The range of topics here is enormous. The trick is thinking about how we pull it all together. One of the things that I love about this time is that we are beginning to have these conversations. And those conversations very much need to be encouraged.
NM: You played a part in the creation of OHBM - and you’d been working with Peter Fox on the earlier meetings. What did you think OHBM would achieve?
MR: It was an interesting experience. There was some tension initially. There was a concern about whether we split between the metabolism and blood flow people on the one side and the brain mapping people on the other. There was this meeting with Gazzaniga, Frackowiak and others. There was a certain sentiment from my side - I had grown up in the blood flow and metabolism community - and they weren’t against mapping the brain. But the feeling was that that world had become very interested in things like stroke and animal models and that it was going its own way. The brain mapping people had these tools and so were going in another direction.
I was really interested in the brain mapping picture. When OHBM started in 1995, we had published our paper on cortical processing of single words—the 1988 paper—so we were all really interested in brain mapping. At the same time it was awful to see this divorce happening. Some people went their separate ways. But I have friends in both worlds.
NM: So it wasn’t all acrimonious then?
MR: No, there was a comfort zone for both groups - where people could do their own things. I think we should keep track of both sides of the fence. Those that want to be critical of fMRI say that we have this problem of neurovascular coupling and it’s an indirect measure of what the brain is doing. My sense is to say, “Wait a minute, you need to keep track of the whole story.” I respect people and we still publish an occasional paper in JCBFM, usually around issues of what to believe.
NM: Are there any memories that stand out from attending OHBM?
MR: I have so many friends in these communities. Attending OHBM means engaging with people you’ve known for a long time. You often pick up new ideas. It gets you in sync with various people. For instance, I recently attended OHBM in Vancouver BC. There, I thoroughly enjoyed a symposium by Smallwood, Margulies and these people that are looking at hierarchical states. I’m a big fan of their work. I had an opportunity to sit there and listen to them and meet these guys.
A sad note is that OHBM in Vancouver was one of the last times I saw Karl ZIlles. He had carried out some of the seminal work on receptor pharmacology and receptor anatomy in the brain. He was there with new ideas about the default mode network: its genetics and receptors. I was shocked when I heard he’d passed away, as we were just recently working on a manuscript.
It’s those personal connections that go right back to Lassen and Kety. I’m deeply invested in the science but also in the people that make the science, whether that’s Frackowiak, Albert Geddes or Bob Shulman, all these characters that make this up. Seymour Kety was also a real gentleman scientist, a really grand person.
Neils Lassen is a good example - he was a very thoughtful fellow and had a big influence on a sidebar of my career. Neils once got a hold of me. He said, “I’m going on an expedition to Pakistan, with the Birmingham medical research expeditionary society. They’re organizing expeditions to the major mountains of the world.” I told Neils how jealous I was, having read books on Everest and other major mountains. Here they were, going to Pakistan and measuring brain blood flow. About a month later, Neils calls me up and says, “I’ve damaged my knee and can’t go - will you go instead?”
So I spent the summer of 1987 on the mountains in the borders of Pakistan and China. We hauled in this xenon and a detector system and measured blood flow at altitudes all the way up to 18,000 ft. The group still holds the world record for measuring human brain blood flow at altitude. We published a number of papers on this: acclimatization at altitude. It was a whale of an experience. When Neils died I attended a memorial service and gave a talk. I couldn’t resist telling this story.
Throughout attending scientific meetings I got to meet all these interesting people. People whose work I admired. It was as much an enriching experience as the particular facts that you learnt at the meeting. It was kind of who’s doing what and what do you think of these people.
NM: A final question: what do you see as the future of neuroimaging over the next 10 years?
MR: I think we’ll get an ever more sophisticated window into the human brain. That’s the overarching sense I have. We’ve opened a door to incredible possibilities. Every time we stumble through a new way of thinking about this BOLD signal, it demonstrates that we’ve not yet tapped all we can learn from it. Over the next 10 years we’re going to hit some bumps along the road, but the level of sophistication of how we think about it is really increasing.
At the same time, the increasing sophistication makes it ever more difficult for people to keep up with the argument. How do you actually calculate resting state? You might say: ‘just put them in the scanner and see the activity.’ To an extent that’s true, but the sophistication of all the maneuvers you do to get there, all the tricks you do is becoming increasingly difficult to understand for someone new to the field.
To exemplify this, my friend Gustavo Deco in Barcelona comes along and says the brain is turbulent. He appends that by saying that the great minds of physics believe it is the most challenging question remaining in physics today: what is turbulence and how do you think of it. So we’re now grappling with one of the most challenging questions in physics. I think, “Holy smokes.”
NM: It’s been a fascinating history into the origins of both fMRI and of OHBM - thanks so much for taking the time to share this with us.