Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
We encourage researchers to utilize the full-text search on this page to navigate our oral histories or to use our catalog to locate oral history interviews by keyword.
Please contact [email protected] with any feedback.
Credit: M. Scott Brauer
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Michael Levin by David Zierler on June 29, 2020,Niels Bohr Library & Archives, American Institute of Physics,College Park, MD USA,www.aip.org/history-programs/niels-bohr-library/oral-histories/XXXX
For multiple citations, "AIP" is the preferred abbreviation for the location.
In this interview, David Zierler, Oral Historian for AIP, interviews Michael Levin, Vannevar Bush professor of Biology and director of the Allen Discovery Center at Tufts. Levin recounts his early childhood in Moscow and the circumstances leading to his family's arrival in Massachusetts. He explains his early interests in technology and physics, and he describes his multidisciplinary curriculum at Tufts that that explored computational and biological questions which fascinated him. Levin discusses his graduate work at the Harvard Medical School where he studied gene expression in embryos from a quantum level perspective. Levin explains his postdoctoral research with Mark Mercola and his work at the Forsyth Institute. He discusses his developing interests in bioelectricity and the challenges associated with presenting his work in mainstream academic science. Levin discusses in interest in cancer research and he describes his work as director at the Allen Discovery Center. Levin explains his decision to take on a joint appointment at Harvard, and he describes the utility of using the metaphors of computer hardware to explain biological systems. He describes his current interests in AI and the future of evolutionary biology.
Okay. This is David Zierler, oral historian for the American Institute of Physics. It is June 29th, 2020. It is my great pleasure to be here with Professor Michael Levin. Mike, thank you so much for being with me today.
Well, thank you very much for having me.
All right. So to start, please tell me your title and institutional affiliation.
I am a Professor of Biology at Tufts University. I hold the Vannevar Bush endowed chair, and I'm the director of the Allen Discovery Center at Tufts. I'm also an associate faculty at the Wyss Institute at Harvard.
Okay. Now, I'm quite interested as a historian of science in Vannevar Bush. I'm curious what connections there might be to his legacy and the chair that you hold in his name?
Well, he was an incredibly interesting fellow. Very prescient, thought of a lot of things that are, of course, with us now in terms of technology and so on. And I'm extremely pleased to hold that chair. I was given that honor at Tufts, and I'm really happy to be part of that legacy. He was an amazing man.
And when did your partnership with Harvard start?
I've been collaborating with people there for years. I think my official associate membership started about a year ago, so I believe at the end of 2018 or possibly 2019 was when I became an actual associate faculty. But Don Ingber's group and I have been collaborating for quite some time. And of course, I did my postdoc at Harvard Medical School, and got my PhD from there as well, so I have lots of contacts at that campus.
And what does Harvard offer you in the way of resources that you can tap into?
Well, the Wyss Institute is an amazing place both in terms of the human elements - the people there are just incredible, both the faculty and all the junior scientists that work with them, and the facilities are top-notch, so they have some really incredible aspects of bioengineering, of material science, of computational modeling. The cell biology, the genetics, all of these things are at the highest level there and it's great to be a part of it.
All right, so Michael, let's take this back all the way to the beginning. Let's start with your parents. Tell me a little bit about them. Where are they from?
We were all born in Moscow, Russia. My dad is a computer programmer, was for years. He's retired now. My mother was a concert pianist and had several careers after that. We left in 1978, when we emigrated from Moscow to the North Shore of Boston.
Was your family part of the Refusenik group?
Only in the sense that it was an incredibly difficult process to get out. We were not as politically active to the extent that some people were. I don't think my parents ever got arrested, as some people did, for demonstrating or anything like that. So I think on the spectrum of how many difficulties people had, we were probably on the luckier side. But we were part of that whole wave that was trying to get out because of the terrible conditions there.
Levin is a Jewish name, I assume that your family was Jewish, and that was one of the reasons that they wanted to leave.
Correct, yes. I mean it was bad for everybody. It was especially bad for the Jews, but it was a very problematic society for many reasons. And I think everybody, or almost everybody, would have gone if they thought, as my father did, that there was an alternative.
And how long had your family been trying to leave Russia prior to 1978?
I think we managed it in under a year, from the time that they sort of pulled the cord in terms of making that decision and filing paperwork, which was kind of a-- once you do that, there's no going back kind of a thing. Once people find out you're leaving, all kinds of hell breaks loose. From the moment that they did that, to the time that we actually got out was, I think it was under a year. It could have been a lot worse.
Was the United States always the goal, or was your family exploring other options like Australia or even Israel?
We would have gone anywhere, but the United States was the number one choice.
Did you have family here?
No, we were sponsored by a set of people at the Temple Sinai in Marblehead, Massachusetts, which is a Jewish synagogue where the people got together and sponsored a number of families to come to the North Shore of Boston. And they were an amazing group of people, and it certainly wouldn't have happened without them.
How old were you when you came to this country?
I turned 10 right as we got here.
So you must have had vivid memories of your first days in America?
Oh sure, yeah. I remember quite a bit. I remember my life before that, and I remember arriving, yeah. It was quite an experience.
How was your English prior to coming here?
Oh zero. None of us had any English.
Uh-huh. (laughs) How did you get up to speed? Watching cartoons? Did your parents enroll you in special language programs?
Nope, no special language programs. Cartoons, absolutely. Watching TV, trying to understand what was going on in the movies. That was number one. Went to a public school within probably a week or two weeks of getting here. Actually was well on my way to learning Spanish for a while, because I didn't realize that many kids at the school spoke Spanish. I had no idea what English was, and so I would just learn whatever the kids were speaking. And so it took probably about a year, as I recall, for me to get to the point where I was able to say everything that I wanted to say and make myself understood. I was still young. I think it was harder for my parents, but overall a real interesting experience, landing down in a place where everything is completely foreign and you don't speak the language.
Now, I assume in the Soviet Union, it was very difficult to be Jewish-ly connected? Was your family able to be more Jewish-ly connected, particularly by dint of being hosted by a Jewish community?
Yeah, the community was incredibly welcoming. Actually, everybody was. We found it amazing. You know, this was the middle of the Cold War and everything, and we thought there would be all sorts of difficulties in terms of assimilation, but actually people across the board were extremely nice and welcoming, and we were tightly integrated into the local Jewish community. We're still members of Temple Sinai to this day.
Oh, so in a sense, you've really never left.
That's correct, yeah. I live, I think, I've always lived within about a mile radius of where we landed in '78.
Are there other families with a similar background in your community?
Yes, absolutely. We were on the early side, but there were a few families who were already here and then a bunch more that came shortly after us.
Is there family in Russia, the former Soviet Union, that you have been able to stay in contact with? Or did everybody that was close enough, did they also leave?
Many of them left, but there are kind of more distant relatives that are still there, and my parents are in touch with them.
Have you ever had a desire to return?
Oh no. Absolutely not. I mean, it was a horrible place, and frankly, I don't even have time to go to nice places to visit on vacation. I'm definitely not going back there. (both laugh)
It's low on the list.
Yeah. Absolutely, very low on the list.
Mike, tell me a little bit about your education when you got here. Did your parents just throw you into public school, and you just sort of had to catch up on your own?
Yes. I was in public school and just trying to catch up with the language. Shortly thereafter, they were able to get me into a private Jewish school. This was Cohen Hillel Academy, also in Marblehead. I went there until high school; I started there in third grade and went to this Jewish school until Swampscott public high school.
Was your father able to keep up his career in computers when you got to America?
Remarkably, yes. What he did when we got here, within a couple weeks, was buy an ancient beater of a car that, as I recall, didn't even have seatbelts. He got out onto Route 128, which is this like big technology beltway that goes around Boston, and he just got out there literally knocking on doors of computer companies. And shortly thereafter, he managed to get a job, which is just absolutely amazing, because-- I mean, first of all, he didn't know how to drive really. That's step one. Because of course in Russia, hardly anybody had cars, so it's not like he had any experience with that. But just imagine during the height of the Cold War, you're a technology company, and this guy shows up with a very heavy obviously Russian accent, doing his best to make himself understood. And it's a testament to his incredible stamina and talent that he was able to persevere through all of that and managed to convince these people that they needed to hire him. He got a job very quickly in software.
What company hired him? What was his first job?
I believe his very first job was a place called Compugraphic, but the main job that he had for decades after that was at Digital Equipment Corporation. And I remember this very vividly, because that was one of the best parts of my childhood. I would go to work with him occasionally and just hang out among all the computers and look at the sort of piles of computer hardware that they were throwing away or trying to get rid of, and being able to go through all that and play with it. It was amazing. And he worked there pretty much right up until his retirement.
Did your father have a mathematical background in his education?
Yes he did. He majored in math and computer science. He actually, much like my mother, they both were very close to getting their PhD degrees right before we left. In fact they had both defended and it was just a matter of paperwork, and then as soon as people found out we were leaving, that got canceled. He was in mathematical computer science, and worked for the weather service; he wrote computer code to predict the weather and to do data analysis and visualization.
Growing up, did he involve you in a substantive way with the technical aspects of his work?
Yes. Not so much with the details of what he particularly was working on, but he was constantly bringing home things for me to look at. And these were a lot of books, these were a lot of parts, things to take apart, electronic kits similar things to play with. And it started, I remember this vividly too, at the very beginning when I was really young, I'm going to say probably five or so? I had asthma, and the thing with asthma is you can't breathe, and when you can't breathe you get worried, and when that happens, your throat closes up more, and that's a vicious cycle. So for a kid especially, it's a real challenge to keep a kid calm during those scenarios. And of course, there wasn't any decent medicine, and so what he would do is take the back off of our TV set.
We had one of the ancient, you know, giant black and white TV sets with the tiny little screens. And he would take the back off and I just remember staring in there and looking at all the little parts, forgetting for a while about the asthma, and I had this sort of thought that it was amazing that somebody knew how to put all those parts together in just the right arrangement. I mean, it clearly wasn't an accident. Somebody knew how to put that in, and it was also quite clear that you can't just mix it up and still have cartoons come out the other end. Somebody actually knew how to do this. And at that time, I thought this was incredible. This must be the coolest job in the world. Whoever figured that out, that's got to be what I do. But he always was extremely supportive of any kind of intellectual or practical interests of that type. So he would talk to me about these things and we would go to museums; we did a lot of that stuff together.
At what point did you start to exhibit strength in school, in math and science, yourself?
I'm still waiting for that to happen. I was not a great student. I did okay, but I was definitely not a great student. I was not on anybody's radar in school as one of the top students in terms of grades. There was a set of students who were universally known as the really excellent students. I was never in that group. Certainly did better in science and math than a lot of other subjects, but I was not a standout in terms of any standard metrics of testing.
When you were thinking about undergraduate programs, were you thinking specifically about science and the fact that you wanted to pursue a career in science?
A couple of things happened. I got pretty good with computers, and this was back in, I mean we're talking 1987, '88, '86, those years. And at that time, it wasn't the way it was later, where everybody can code and it's a very popular thing. At the time, there weren't a lot of people programming yet, and I found that I was able to make money writing software for people for various things. For local businesses, for research labs, whatever. I was able to code. And I was really interested in artificial intelligence, computer science, and it wasn't terribly obvious that I needed to go to college. I was doing pretty well, I was having a lot of fun making money writing code. And I sort of, on the one hand, kind of figured that maybe that's what I could just keep on doing. On the other hand, I really wanted to get into the science of artificial intelligence - go beyond coding and really get into computer science, per se. And so for that, obviously you'd have to go to college. And so that was the goal.
Right around that period, I rekindled my interest in biology. This started a long time ago as well, because when I was a kid, we had a neighbor who was really into insects. Collecting bugs and insects and things like this. I was always interested in technology, in physics. But I would look at these little creatures and think, well, you know, this is amazing, because it's pretty clear that these things are doing things and have functionalities that none of our technology has. And that maybe there's a lot to be gained by understanding how the biological world works, both in terms of just capabilities and robotics, but also in terms of the fundamental questions, like what is a thought, and what is memory and what does it mean for a physical object, whether it be a robot or a biological creature, that has goals and has memories and has an inner perspective and a first person outlook on the world and so on. I was always interested in that. And so going in, I went into college ostensibly to get a computer science degree, but by the time I got out, I ended up getting dual bachelor's degrees - one in biology and one in computer science.
Did your father have any useful career and academic advice for you, as you were coming up in undergraduate?
His advice was always one level above the details. So it's not that he had experience in any of the particular things that I was studying necessarily, but he always had excellent advice about how to relate to science, to difficult subjects, to people, and the way you should think about things and how to integrate that with your life and things like this. He always had a lot of really excellent advice about navigating these things in general.
Who were some of the professors as an undergraduate that you became close with?
Well, the closest early relationship I had was with my first mentor, Susan Ernst - she just retired a couple years ago. It was actually one of the great pleasures of my career, to go back to Tufts as a faculty member and to sit in on faculty meetings next to people I had as professors during my time there. Susan Ernst was a developmental biologist. She studied sea urchin embryo genesis. I remember approaching her and saying, "Okay, I'm a computer science student, know very little about biology. I think I would like to get into biology, to understand living systems from a physics and computational perspective. Can I do research in your lab?" And then I had written up some thoughts that I had had about computational approaches to understanding development, which were extremely naive biologically. And I approached her with all of this, and amazingly enough, she said yes and welcomed me into her lab. And I spent two and a half years, doing independent research with her on embryonic development. She is a very special person, and I think certainly was formative for everything I've done in biology since then.
Mike, did you essentially put together your own curriculum, or was there a pathway at Tufts for allowing students to think about science in very out-of-the-box ways like you were? So that, you know, computational questions and biological questions had sort of a course load infrastructure to them already. Or did you essentially have to create all of that yourself?
No, I created that for myself. There was no such path, at least as far as I knew. And in fact, even when I got to grad school, so we're talking '92-93, I had a well-known professor who was teaching cell biology, when he realized what my background was, he told me in front of the whole class that basically the admissions committee must have made a mistake and that I was wasting everybody's time, and that there's no way that this sort of background was going to be sufficient for getting at PhD at Harvard Medical School. I remember hearing that and thinking that, wow, this is definitely a kind of behind the times attitude - there's no way this is right. And now I think it's much more obvious that computational and kind of physics perspectives are a good complement for biology, but I guess back then it must not have been.
And I learned a few things during that whole period. One was that you absolutely have to make your own curriculum and get the tools that you need to do the work that you're going to do, and whether or not other people agree with that is something to kind of put aside and do what you need to do and then figure out later if it's working or not. And the other thing is to be really focused on whatever you're doing now with the intention of getting back to what you really wanted to do when you are ready. So I got a PhD in a very traditional genetics lab, in a very mainstream molecular biology project, because I wanted to learn those tools. I knew I was going to do very different things later on, assuming I survived the process, but it was clear that it was not the time to try to convert people to my way of looking at the world, it was time to keep your head down, learn what you need to learn, and make sure you're competent in the mainstream kinds of approaches and then see if you can do other things later on.
Now, the early 1990s is certainly, you know, to start thinking about how you can combine computational questions with biological questions, this is in many ways a very advanced way of thinking, particularly for an undergraduate, right? I'm curious, what were some of your intellectual influences, maybe even beyond the classroom, that got you even to think in this realm?
Yeah. I mean, realistically, I was thinking in a broad, sort of unsophisticated way. I was thinking about these things very early on, because I remember as a kid out in the yard with my friend, doing very primitive behavioral experiments with beetles and watching caterpillars emerge out of eggs and things like this. And I remember thinking that, in terms of engineering, we really don't understand any of this until we can make one ourselves, right? This is a common idea. And I thought, okay, so how would you make something like this yourself? Not necessarily in the same physical embodiment, but what does it take to-- I mean, these creatures were in many ways very goal-directed. They had specific things they were trying to achieve, and if you intervened or you moved them, or you put something in their path, they would figure out a way to get where they needed to go.
And I was thinking that it would be very important to try to understand what computational processes - what was the primitive thinking that was going on in there? That was first. And then of course the question of, how did they get here? Where did they come from and what did the first one look like, and what are they going to look like later on? My dad used to bring home all sorts of books when I was little, and then once we got to the U.S., we had a weekly ritual where we would, on Saturday mornings, go to Cambridge, Harvard Square in Boston, and we would go and we would hit the bookstores and we would come home with a pile of books. And a lot of the things that we would look at together and that we would buy were works in philosophy of mind, in cognitive science, these kinds of things. In fact, one of my favorite sets of books was all this stuff by Dan Dennett, which is another big pleasure of mine, because he and I are working on a paper together now. So it's kind of great to be able to do that, because I was reading his books from the very beginning, and thinking about these issues of what does it mean to have first person perspective and memory and goal-directed activity and so on? And it was just very natural that this had to be more deeply understanding the laws of physics and computation that evolution was exploiting to make these amazing creatures.
What made you realize that having a physics grounding was important for combining these two disciplines, of biology and computational science?
Well, I should say right off the bat, I don't have any real physics credentials. I don't consider myself strong in physics per se, even though I work in what I think could be fairly described as a really interesting branch of biophysics. It became clear that what living things were doing were exploiting the laws of physics to achieve specific behavioral ends. And by behavior, I don't just mean they're moving around, but behavior in terms of embryonic development, the cells were harnessed towards particular outcomes, like putting together various anatomical structures and so on. And clearly, in order to do that, evolution had to exploit various laws of physics. And it became really interesting to me to think about how these different laws of physics not only have constraints, but also as facilitators of specific biological phenomena that we were seeing.
I thought at one point I was going to be an astrophysicist. I was really interested in large scale physics. Planets and stars and things like this. And it was also very interesting to think about biological creatures and how complex they are. And how the laws of physics that govern these massive cosmological scale events, could also be harnessed in a very different way to make these really tiny things with way more complexity than anything we see in the astronomical world. And so it was clear that kind of physics, computation, and biology form this really critical knot that had to be all studied together. You couldn't take these things apart, I don't think, and do them separately. All the interesting stuff is right at the intersection of those three fields.
Did you have a senior thesis?
I did. As an undergrad, I had a senior thesis on the effects of magnetic fields on sea urchin development, and so that was kind of an early start to trying to integrate some of this stuff, looking at how a physical force of such a magnetism could be coupled to processes of embryonic development.
What kind of graduate programs were you thinking about as you were nearing the end of your undergraduate education? I mean, just from listening to you and all of the things that you were interested in, I can probably come up with ten very different kinds of programs that might have satisfied both your intellectual ambitions and your career goals. So I'm curious how that process played out for you.
Yeah, a couple of things were going on at the time. I really wanted to do this kind of work, and I knew that by then, I had done enough reading to know what kinds of things were considered mainstream, and that would be easy to find in a graduate program, and which things would be considered crazy or were very rare and hard to get hold of. And I had sort of decided that if I was going to try this, I would go to a fairly mainstream program to learn the state of the art in molecular biology, genetics, cell biology, and then apply it in novel ways if I could. The other big thing at the time was that it actually didn't seem very realistic that this would ever work out. At the time I briefly had a software company. My dad and I worked together to write some software that we were going to sell as a company. And it was actually taking off. So there was some question about whether I would just go off and do that versus grad school, but in the end I thought what I would do is I would try graduate school. I would take this as far as I could, and then eventually I would surely get kicked out and then at that point, I would go back to coding and then everything would be fine.
So I went to Harvard Medical School. I applied there to the PhD program, not the MD program, feeling that there was absolutely no way that I was going to be able to get in, because of the limited biology experience that I had had relative to all the other students that were applying. I think-- I don't know for a fact, I think what probably got somebody to accept my application was that at the time, I had a number of publications, all in the computer science field. These were papers that I had been writing basically on my own, and just submitting them to computer science journals. So by some miracle, I got accepted there. I was in grad school for four years, and most days I thought, "Okay, this is it. This is the day I get kicked out." That was kind of the general feeling I had most of the time through grad school, right up until the time somebody told me that they had "checked my box"-- I didn't know what that meant either, but it turned out that that was, when the PhD committee checks this box, it means you can defend and you can get your degree. But I was largely clueless about how things works, and my strategy was just, I was just going to keep going forward, I'm going to do what I need to do, and whenever it is that I get kicked out, that's when I'll go back to coding.
Now administratively, how does it work? There's a department of genetics within the Harvard Medical School?
And what's the connection there? Why would the department of genetics be in a medical school and not, you know, part of a biology program?
There certainly is one in the regular biology program, but it's certainly clear that the tools of genetics have a major role in understanding both health and disease. A lot of people in that program where I was, it was called the Cell and Developmental Biology program, were addressing very basic, fundamental questions. It wasn't all medicine, but at the same time, they were in an environment where all that work could very easily be applied to biomedical problems, because all the clinicians and the clinician scientists were right there. So it was a very integrated program where you could study very basic genetics in fruit flies and yeast and things like this. At the same time, right there was somebody who studied the genetics of various birth defects and so on. And all of this was pretty integrated. Yeah, I went there because I thought this was the number one place to learn the state of the art approaches. This is the best mainstream biology has currently for understanding living systems, and I wanted to become an expert in those tools and that way of thinking.
Did you realize in real time what an exciting moment that this was in the 1990s for genetics in terms of all that was going on at DOE and at NIH?
No. I had absolutely no clue what was going on at the DOE and NIH. But it was clear to me that this was a very exciting time, because it was clear that people were cracking through to a much better understanding of the genome, in terms of the factors that control the hardware of the body. And I was, you know, that kind of distinction was always kind of clear in my mind, that what we learn when we're going genetics is fundamentally, we learn where the hardware of the cell comes from, where the different building blocks, the proteins, that the cell has to deploy, where those things come from. And that that was essential, but beyond that, there would be really interesting questions of the software - how do cell collectives use that hardware to communicate, to solve problems, to store memories, to make decisions, and so on. So it was clear that that wasn't the end of the puzzle, but it was really a critical beginning to it.
Being in the Harvard Medical School, did you ever consider, if only for a minute, to pursue a medical degree?
No, never really thought about that. I mean, I'm really interested in taking some of the things that we've learned in my lab, and hopefully making an impact on medicine. I would love to be able to say that we've helped real, living patients. That would be fantastic, and a lot of our regenerative medicine work is aimed in that direction. But that's definitely, I never thought that seeing patients was where my personal strengths lie, so I figured I would help on the discovery of basic science, and then somebody else, the clinician scientists and then the MDs would take that into the clinic and help people.
What were the kind of courses that were required to get a PhD in genetics at Harvard?
Well, let's see. I mean there was of course genetics, there was cell biology, there would have been biochemistry, there were lots of different electives, you know. People were taking courses in virology and protein signaling, biochemistry, and immunology.
Did you have an opportunity to continue pursuing your interests in computers? Or did you sort of put that on hold?
No, I kept it up actually. It paralleled with a number of papers from that time period, where just sort of to keep sane, I would do these computer science things in my so-called "spare time" because I found the biology classes very hard, very challenging, and the computer science compared to that was easier, and so I was able to feel like, okay, I'm still competent at something, at least? It was nice to do some coding on the side. And so I kept that up.
How did you go about developing your dissertation topic?
I was in the lab of Cliff Tabin, and his lab had discovered this particular gene called Sonic Hedgehog, that was really important in limb patterning and it was important in neural tube patterning and so on. And one of the post docs there was showing me some expression analyses of this gene during early development. The later roles in the neural tube and in the limb, had gotten picked up by the various other post docs. But there was this really curious aspect to it early on, that if you look at the early embryo, there's this little curly q. It was very small, like a comma almost, and it always bent to the left. And we were looking at this, and nobody had taken that up to study it, because at the time, there had never been a gene that had been asymmetrically expressed, that was only on one side of the body. Every gene up until then was bilaterally symmetrical, and it wasn't clear at the time whether this was some sort of aberration or whether this was real or what the significance was. And I looked at it and I thought that, you know, this question of left-right asymmetry is incredibly interesting from a physics perspective. Years before that, I thought about questions like, if you were communicating with a space alien somewhere, and you could talk to them but you couldn't send physical objects back and forth, how would you explain to them what the words left and right mean? So assuming they look a little bit like you, you would say, "Okay, so you're standing there, your feet are pointed towards the center of your planet, and your major sense organs are pointed forward. Fine. Now your left hand is the one that, what?" It's really hard, right? There are some physics tricks to like move the right hand rule of the magnetic field and so on. There are some physics tricks you can use. But overall, it's a really hard problem to solve, and yet all embryos reliably do this. All normal embryos know their left from their right with a fairly low error rate. So how could this happen?
And so I thought about that for years, and it was more or less a total accident that here was for the first time a way to actually make some progress on this problem. This was '93. Back at this time, there was very little known about how left-right asymmetry worked in embryonic development, and no molecular information whatsoever. There were some mutants, there were some snails that coil the wrong way and things like that, but there were no molecular details. So I jumped on it immediately, because this, I thought this was just the coolest problem ever, not only because it has fundamental kind of physics interests, in the sense that the universe does not macroscopically distinguish left from right, although at the quantum level, it does - there's this parity violation at the quantum level. So I thought was just incredibly interesting and in the back of my head, I thought, "Wow, maybe embryos are picking up on that somehow, which would be totally cool. So that was it. Once I saw that, that had to be the PhD project.
Who was on your committee?
Boy oh boy, that is a great question. I don't remember exactly.
I mean a more general question would be, would the people on your committee, would they have reflected the multidisciplinary nature of the dissertation?
The dissertation itself proceeded in a very standard, molecular genetics way. I wrote the first draft of our paper and I had all this stuff in there about shaking hands with an alien and physics of parity violation. And all of that got crossed out. (laughs) correctly. Cliff said, "Look, you can't talk about any of that stuff in the first paper - your paper's about the genetics. You've got to get rid of all that stuff." So--
Save this for the New York Times bestseller, right?
Yeah, it wasn't for the Cell paper that came out. But I ended up being able to put it all back into the actual thesis at the end, the dissertation. The committee was extremely straightforward, developmental molecular geneticists. And by the way, the first few committee meetings, everybody though this was a total artifact. So the first thing they said is, "Why is it only on one side? That means you don't know how to do in situ hybridization properly. It's clearly an artifact." So what they said was, go find some other genes and make sure that they're expressed on both sides, and show that you can actually do a proper expression analysis. I said okay, fair enough. So I went off, got a bunch of other genes, and the first three that we looked at also turned out to be asymmetric.
Now, at this point, I started thinking, "Wow, maybe there is an artifact here of some sort." But as it turned out, then I kept going and I ended up looking at, oh, probably two dozen or so. Then we basically just realized that what was going on is that, actually a lot of these genes were asymmetric, and nobody noticed it because they were expecting them to be on both sides. In fact, I came across a number of people who had the same photos that I took of these expression patterns in their office, and people assumed it was an artifact, and nobody said anything or wrote about it. So our 1995 Cell paper is the first description of asymmetric genes, and it's got a bunch of them in there, because it turned out it wasn't just Sonic Hedgehog, it was a lot of the early developmental processes are fundamentally asymmetric.
Mike, when you graduated, or I should say, when you defended, what kind of opportunities were you looking for? What did you want to pursue after graduate school?
I was still firmly in that same mode - keep going with this as far as I can, and see what happens. And so at that point, I wanted to learn an amphibian model system, where I would be able to do regeneration and in particular I wanted to learn a system where you can see the very earliest steps of development right in front of you. I worked in chicken embryos as a PhD student. And chicken embryos are amazingly useful for a lot of different things, but by the time you see that egg come out of a chicken, it's probably got, oh, maybe 50,000 cells already. You're not seeing the earliest events of development in front of you because they take place inside the chicken. So I wanted to learn a model system where everything is there for you from day one.
And the reason that's important is because in my PhD thesis and in those papers, what we showed is an asymmetric cascade of gene expression, meaning there's this gene that's only expressed on the left because there's some other gene that turned it on that was also on the left, and there was another gene that turned it off just on the right, and that one was only expressed on the right. So you can sort of chase this back and back and back, and we did. But once you think about it, you realize that, okay, this is all great, but no matter how far you chase it back, you're always going to have the same question. Which is, whatever the first gene is that only turns on one side - is only transcribed, let's say, on the left or the right - why is that? And the answer can't be because its inducer was expressed there, because we're talking about the most upstream transcript.
And so it became pretty clear that we would have to ask questions about where developmental pattern comes from. You had to look at how you go from a single cell to this multicellular, amazing kind of a body that puts itself together during development. So I started looking for postdoc positions and the place that I particularly wanted to go, was the lab of Mark Mercola. He was in the cell biology department, basically across the quad from Cliff's lab, and he worked on frog development, in heart in particular. But the thing about frogs is that you do fertilization in a dish. So you collect the eggs, you collect some sperm, you put them together in a petri dish, and everything is right there for you from the very first moment of when that egg gets fertilized all the way through having a tadpole swimming around. And that's the sort of thing I wanted to learn.
And this is unique to frogs?
No, there are lots of model systems that do this. I wanted either frog or salamander, because I wanted something that was highly regenerative that you could do these kind of mix and match experiments, where there, we could paste different tissues, you can cut and paste different tissues and recombine them in various ways. All of that can be hard with mammals or birds or fruit flies, but amphibians are really good at it. But no, there certainly are other model systems where you can study the very early stages.
I'm curious if you saw the postdoc as an opportunity to branch into yet new areas of research? Or you saw this as a place to continue on with the research you had done as a graduate student?
Well, it was a little bit of both. So on the one hand, I was going to continue my work on left-right asymmetry, but on the other hand, I wanted to get beyond the transcriptional cascades because having found these transcriptional kind of chains of activation and repression, I wanted to know what was up stream of that. How does the very first gene know to be expressed just on the right side? And it seemed to me this process has to bump into physics at some point, because the genome can't tell left from right. So if you've got a set of DNA, that DNA is the same on the left and right sides. At some point somehow, there has to be different activation on one or the other. And there has to be some aspect of physics that dictates that. That's not something you can encode in the genome. And so it was clear that I needed to chase this upstream, and that's what I wanted to do in Mark's lab, and in particular there, we found this amazing bioelectric gradient that actually is formed in a very early stage that points rightward, basically, and that starts this whole asymmetric gene cascade going.
And so I wanted to pursue this work deeper into its origin. I wanted to know the origin of left-right asymmetry. I thought that what I would need to do there is to set myself up for having the tools needed and the knowledge needed to run my own lab, and do even weirder things on my own. And so that was the opportunity to learn all kinds of important skills as far as project management, interacting with editors and these kinds of things.
How much were you publishing during your time as a postdoc? Was this an opportunity for you to really get your name out there and look for collaborators?
Yeah. I published a number of papers at that time. In terms of collaborators, I think I did more of that when I was independent. The rest of the lab was not interested in all this bioelectric stuff. It's not like I had a lot of resources to commit to any collaborations at that time. The one thing I did start, towards the end of the postdoc, is I started to make connections with people who worked in neuroscience, inner ear, gut, and so on, because these folks had a bunch of constructs, meaning different pieces of DNA, for different ion channels. And I knew that my work in the future would really make use of these techniques. I wanted to develop ways to control the way that cells communicate with each other bioelectrically. And so I set up a lot of contacts with people who had expertise in these different kinds of ion channels that I then turned into tools to perturb native electrical signaling. And in fact, in the paper that came out of that, also in Cell with Mark, was the first use of ion channels, or in fact of any molecular reagents, to control the electrical signaling that underlies development. We had shown that you can target these ion channels in a way that would perturb the cell's ability to know which way was left and which way was right, for example.
When did you know it was time to move on from your postdoc? Was it a set period of time, or did the next opportunity come up for you before that?
It was not a set period of time. I spent four years in Mark's lab and I came across an ad from Forsyth Institute, which was an affiliate of Harvard Dental School, they were right next door, and they were looking for a basic developmental biologist to start up a lab and feed some of their applied efforts. And I was ready to give it a try - I was ready to try the things that I wanted to do on my own.
Mmhmm. And so what was your next opportunity? What did you want to do next?
Well, fundamentally, I wanted to make progress in developmental bioelectricity. So I wanted to establish the understanding. I wanted to do two things. First of all, in the short term, to establish bioelectricity as a molecular field. People have obviously talked about the importance of electricity in development regeneration in cancer for many years, probably over 100 years, but it had not really been addressed molecularly until then. And so what I wanted to do was to integrate that cool field with the field of molecular and cell biology, to show that we can now study it at high resolution and to really understand how these electrical signals are enabling cells to do the amazing things that they do at large scale. I wanted to establish those tools and to really show how bioelectrics matter during embryonic development, during regeneration, and during cancer. That was the short term.
The longer-term vision is, I always had this feeling that bioelectricity, while it's a fascinating mechanism, it's more than just a mechanism. And I always thought that if we treat bioelectricity as just another step in a long pathway alongside biochemistry and biomechanics and so on, that that was missing a large part of the puzzle. The special thing was not that it's electric. The special thing was that it's an incredibly convenient mechanism for processing information. It was a computational medium. And what was really special about it was not the details of the mechanics of it, but actually the fact that it was processing information. That it allowed all kinds of cells to do what we think of as neurons doing, which is to form a network that processes information and makes decisions and stores memories. And I thought, if I'm able to establish bioelectrics as a solid field and show that it matters for developmental and regenerative outcomes, the next step will be to understand it at a computational level, at an information processing level, not just as a mechanism. And that is what occupied my lab for the next 10 years after I started in my lab at Forsyth in 2000.
Who were some of the players, both individuals and the fields that they represented, that you felt like you really needed their buy-in, you really needed them to see where you were coming from, in order for you to get this effort of essentially creating this new field off the ground? Who were the biggest stake holders that would allow you to sort of gain mainstream acceptance in the broader scientific community?
Yeah, it's an interesting question. I thought about this a lot, because when I was younger, one of the things I was really interested in reading about was people on the fringes of mainstream science, both because they were studying something that nobody else was interested in, and the people that were way ahead of the curve. They had seen things that nobody else was willing to take on board at the time. And I read a lot about the history of it, both from their perspective and what the field was doing at the time, in terms of trying to understand how does one make an impact with a novel and kind of unusual approach to things? And I think I learned a couple of things from that and also from my own interaction with people talking about my ideas at an early stage of the game. And what I think I learned is there's a lot of inertia. And there's a lot of resistance to new ideas. But it's not because anybody is trying to hold it back or because there's some establishment that hates new things. A lot of people on the margins of the mainstream have this kind of adversarial view - that people are actively trying to hold them back. I don't think any of that is the case, there's no big conspiracy. Nobody's trying to hold back new findings.
However, what there is an incredible inertia, especially of the most successful people, because they're incredibly busy, their time is very limited, and they are already being rewarded for doing successful things that they care about. So getting those people to look at something new and to appreciate something new at an early stage, before it's obvious that this is something that they're going to make use of and it's going to be useful for them and so on, is really difficult. And this is what we found all along. That in trying to promote these new ideas, the biggest barrier was people who had not taken the time to dive deep enough into what you're doing to appreciate it. And you can't blame them, in a sense, because time is limited. Everybody's got a limited amount of time, and you know, there is not time in the day to delve into everybody's wacky idea that you haven't heard of before. So this, I always thought that this was a barrier and it's too bad, but it's not anybody's fault. It's just how it is. And so what I always thought is, our job is to make enough discoveries that make it clear that this stuff is important for other peoples' outcomes.
So for example, in all of my papers on bioelectricity, I always try to tie it into an existing pathway that people care about. In my postdoc work, you may not care anything about bioelectricity, but if you're interested to know why Sonic Hedgehog turns on just on the left side, bioelectric signaling is what you need to know. Or in regeneration. If you're interested in, what controls whether a tadpole tail is going to regenerate or not? If that's what you're interested in, there's no getting around bioelectricity, because we have shown that these bioelectric phenomena are really critical to having that happen. So I never really focused on convincing any one individual person. Or set of people. I didn't think that there were any kind of individual-specific gatekeepers. it was more of a general consensus in the community as far as what kinds of things were important and what were not important, and that what we needed to do was to show some really well-worked out cases where it's just super clear that, okay, if you want to understand this, meaning cancer or regeneration or development, bioelectrics is an important part of it. There's a general sense in the community that biochemistry is the key, so I needed to produce enough data so that people who care about some of these things say, wow, I need to read up on this because this is an important input into what I'm doing. And that's always been my message to my students.
Now, certainly, you know, your approach to sort of more mainstream science has been a continuing theme throughout your career. In other words, you didn't just have this unique relationship as a postdoc, and so I wonder if you could speak generally about, you know, when you say that in the late 1990s, if people were very set in their ways or they had a certain way of thinking about things, have you seen a broader acceptance in the scientific community for approaching interdisciplinary science in a more accepting way? Has it been less of an uphill climb in terms of the way you promote your ideas and how difficult it might be to get people to see where you're coming from? Has that burden lessened over the course of your career?
I should preface all this by saying that I am not complaining. I don't feel that I've been treated unfairly in bulk. I mean, there have been individual incidents, but overall, I don't feel that I've been treated unfairly. We've been well supported by mainstream funders. NIH, NSF, DOD, etc. We publish a lot. So, there's nothing to complain about there. However, the reality is that it has been absolutely an uphill climb in many ways. And while the details of it are getting better, just because there's now a larger body of work, there's other labs getting into this, it's now clear that this has clinical implications. So that part has gotten a little bit easier. The overall phenomenon of the inertia of science and just how difficult it is to get people to look at things from a different perspective, I don't think that's changed at all. Certainly not in the short time that I've been active. I think that takes a lot longer than that. And I tell this to all my students when they leave my lab: "Somebody who's very successful in one particular area of science, they can smell what's a good idea, what's a bad idea, what's going to work, what's not going to work. They're successful and they're well-calibrated on it. Those same people, no matter how intelligent or successful they are, are generally not well-calibrated on your ideas. They do not have a good sense or a crystal ball about what's going not going to be successful with what you want to do."
And so one of the things I tell my students is, you've got to be really careful at what advice you take on board. You have to take criticisms very seriously if they are specific. But I've heard two things my whole life, even from very successful people that I respect greatly. Don't do it, it's never going to work. It'll kill your career, it's not a good thing to be doing. That's A. And B, focus, focus, focus. Drill down. Don't do too many things, drill down on one area. And I think that realistically, almost all of what I think are interesting and impactful things that we've done in the last 25 years, consensus was - don't do this. And I don't think it's because anybody doesn't want you to be successful or anything like that. I just think a lot of people are not well-calibrated outside of what they love. They don't have a good nose for what is or is not a correct approach. And what that means is, we have to develop our own. And that's what I tell the postdocs and students leaving my lab. Nobody has the right intuition about what you ought to be doing except for you. And you need to develop it, and then, maybe, assuming that you've got the rigorous tools to take that forward, then maybe it'll work. But I have not seen any change in this big, very fundamental aspect that people are just not, they're not willing to give you the benefit of the doubt on things that they are not already familiar with.
Sure. Mike, can you talk a little bit about the Forsyth Institute and why you realized that it would be a good home base for your research?
Yeah, the Forsyth Institute was a very unique place. They were originally, I think, the first dental infirmary for children. They opened in 1910. And they were mostly about research on the dental kind of arena. So everything from the development of teeth to the bacteriology of the mouth, all kinds of things related to the oral health. And they needed some basic developmental biology expertise to bolster their applied efforts. And so I went in and had my interview, and the main question that I asked them was: is anybody ever going to tell me what to study or what not to study? And the answer was no. Because I had heard by that time, just talking to people about what junior faculty experiences were, that a lot of people were being told by the department chair, "Do this, don't do that. If you do this, you're not getting tenure." I was really interested in doing exactly what I wanted to do, because life is short and this job is hard, and if you're going to be putting in blood, sweat, and tears on it, you want to follow your very best ideas. So I asked "Can I do whatever I can afford to do? No one's going to tell me what to do or what not to do?" And they said yes, and sure enough, that proved to be correct. It was a very small institute. It had a great leadership and I had a supportive environment where I could write grants and grow my lab in weird new directions without pushback.
Was it the equivalent of an academic appointment? Was there a tenure track component to it? Could you bring in postdocs? Were you able to teach? How much was it like a more traditional academic appointment?
It was almost exactly like any traditional academic appointment. There was a tenure track. I had postdocs, I had grad students. I had a very light teaching load because I was doing it over at the med school. There were no undergrads. And so I would show up as a guest speaker to some classes, some advanced seminars and things like that. But yeah, it was a very standard appointment.
And in what way were you able to do your work better than had you been in a more traditional academic department kind of position?
Well, it's hard to compare it to a counterfactual possibility. I don't know what would have happened at another department.
Well, I mean, were you less burdened by not having to teach as much, for example?
That's for sure. I think what was absent, which was good for me at the time, was first of all the teaching. And second of all, there weren't a lot of people around me whose work was threatened in any way by the things that I was doing. So there wasn't a lot of pushback, because if I was finding the kind of novel things that upset some prior belief structures, there wasn't anybody in the immediate vicinity that was really worried about it. And so on the one hand, a junior faculty needs to have other people around to talk to, and that's certainly the case and I made plenty of use of that at the medical school. I had lots of friends and colleagues nearby. But it just meant that there was a lot less friction and a lot less local inertia that had to be overcome to get things out. And that was important for both the students who were working on kind of unconventional things, but also for myself to gain confidence to be able to send out both papers and grant proposals on very unconventional ideas that I think would have received a lot of pushback in a traditional department.
Did your interests in human health and particular medicinal and clinical type areas, did that expand during your time at Forsyth?
Not really, in the sense that we were in the practical sense, we were focused on very basic mechanisms. I mean, I always thought that if we crack this problem, ultimately, it would give rise to serious improvements in regenerative medicine healthcare. But I knew that that was way down the line. The things we were working on were very basic, fundamental problems, that nobody knew the answers to.
What do you see as some of your most important research that you did while you were at Forsyth?
Let's see, well, at Forsyth, there were a couple of things that I think were really critical, and some of it, it was done at Forsyth but didn't come out until later on, until after I had moved to Tufts. The first few years of establishing these bioelectric tools that demonstrate that we can effectively see, directly see, in real time in a living embryo or regenerating tail, or anything like that, all the electrical conversations that cells are having with each other. That had never been done before. That was an incredibly useful tool for us. Understanding that both regenerative events and complex development was implemented in modules that could be triggered by specific bioelectric states was very important. What I mean by that is for example, we found that if we observed, let's say, eye development and we found that the eyes were triggered by a particular bioelectric pattern in the early head, we could induce that same pattern elsewhere in the body and it would trigger the formation of a complete eye.
Now this was amazing for a number of reasons. Number one, it meant that it's not only the cells in the head were competent to become eye-- which was what the textbooks said at the time-- but that actually almost any cell can initiate eye formation. And the fact that you could do it bioelectrically was really novel. But the biggest thing that came out of this was how modular this was. Because we were not having to micromanage the creation of an eye. Eyes have many different cell types. We weren't specifying all of that. We were triggering it - like a subroutine call, for people who do coding. It's like a very simple trigger that kickstarts a program that you already know how to do. In effect, we’ve simply told the cells, " build an eye here." You don't micromanage or force the production of an eye by, let's say, 3D printing or specifically patterning, well, here I want the lens, and here I want the cornea and here I want some retina cells. Instead, we were telling the cell groups at a much higher level that here's where an eye goes, and then they do the whole thing, and then it stops when it's done.
So this idea that the bioelectric signal is a kind of large-scale plan of where the different organs go started coming out when we were at Forsyth. We also discovered this thing called the electric face. We found out that the early gene expression and the anatomy of the face is basically laid out by an electric prepattern that's there before those genes turned on. This was a natural pattern that was required for the frog to know how to put the face together, and if you went in and rewrote that pattern, if you edited the electrical states by targeting these ion channels, then the gene expression would change and the cranial facial anatomy would change. And so this idea of this bioelectric pattern memory as an informational scaffold that tells the cells what to do to form these various structures, that was I think one of the most important things that came out of our work. There was also other work on tail regeneration and showing that you could kickstart a whole complex appendage, including spinal cord and muscle and skin and blood vessels and so on, with very simple bioelectrical states. And also that's where we started our work on cancer, showing that the defection of cells from the normal electrical body plan would induce tumorigenesis in the absence of any genetic damage. And then better yet, that you could actually fix it by controlling their bioelectrics, you could normalize these tumor cells. So these are some of the things that I think were the biggest discoveries there.
Mike, were you looking for an access point to cancer research, or were you conducting basic research and it sort of dawned on you that it had relevance to cancer studies?
I've been thinking about cancer from day one, because the question isn't why do we get cancer? The question is, why is there anything but cancer? In the sense that individual cells are highly competent. Look at any unicellular creature, they know how to handle their needs (physiological, behavioral, morphological) on the scale of a single cell. So the real question is, how do they cooperate together to work on very large goals? Embryonic or regenerative cells have this large goal of building this massive organism. It's a much bigger goal than any individual cell can see. How do you harness individual single cell organisms to work together towards this amazing group project of building and maintaining a body? And once you think about it that way, it's inevitable that occasionally, that cooperative process breaks down. Cells that ignore or can't hear the signals that bind them to this large information processing network - their computational boundary shrinks. The scope of the “self” shrinks, whereas before, all of these cells are linked into this one massive whole. Once you're isolated from that, well, then your boundary is just the surface of a single cell, and everything else is “outside environment”. If you're a single cell that can't hear the signals of what you're supposed to be doing, the rest of the body is just external environment as far as you're concerned. So at that point, you would go where you want, you would proliferate as much as you could – that’s metastasis.
So to me, cancer was always a fundamental twin process to development and regeneration. Everything is about cellular decision-making and binding individual, smaller agents. And this is where my computer science/cognitive science interests are clear. Everything is about how the goals of individual cells scale up to be large, anatomical goals of building a body, building organs and so on. And how that process breaks down. And so to the extent that the electrical communication is really critical to cell networks having memory and making decisions, being able to coordinate with each other and so on, then it becomes a prediction of all of this world view that by manipulating the biometrics, you should be able to detect, cause, or normalize cancer. And this of course is another departure from the mainstream view, which sees irrevocable genome damage as causes of cancer. This is a completely different view, focused on an informational breakdown. And so for us, cancer became a really important context in which to test these ideas.
What have been some of the things that you found in cancer research that might be helpful in a clinical sense to treating cancer?
I preface this by saying that all of our prior work is in the frog model. So we are only now moving this towards clinical applications in human cells. I think we found three fundamentally profound things that are going to be important for cancer going forward. The first thing we found is that when cells disconnect from the electrical network of the body, when they lose that connection to this much larger, integrated unit, this is detectable by a fluorescent voltage reporting dye. So we showed that you can actually see precancer by simply tracking the voltage of cells, via dye microscopy. So this is potentially really important diagnostic technology, because you can actually see where the boundaries and the margins of this weird cell or group of cells are that are in the process of disconnecting from the rest of the body and going and transforming into cancer. So as a detection technology, that was the first thing. The second thing we showed is that contrary to the modern idea of cancer is always starting with genetic damage, we show that you can take a perfectly normal tadpole, no oncogenes, no carcinogens, no radiation or DNA damage or anything like that, and induce full-on metastatic melanoma by disrupting the electrical communication between two different kinds of cells. And so this is really important, because it gets at the root of what cancer actually is. And is it an irrevocable problem with the genome, or is it an informational disorder? And the reason that's important is not only for the etiology of what causes cancer, but for thinking about what kinds of things might be strategies for treating it? And the third thing we found, which I think is the most direct application to medicine, is that we found that once you do have a tumor, let's say induced by a nasty human oncogene, so we would take like KRAS mutations and things like this and throw them into tadpoles, and they would make tumors. You could then normalize those tumors, cause them to disperse. Now, not kill them, but cause them to disperse, and join other cell types and be normal, by forcing them to electrically communicate with their neighbors.
So what we showed is that the electrical state is dominant over the genetic state. The strategy is not only to track down every single cell and kill it, that it's possible that the electrical properties of the cell and its environment can normalize those cells and get them to function with the anatomical body plan and not towards single cell rogue behavior. And I think that's really critical, because it provides a different path forward than chemotherapy or immunotherapy. It suggests that what you have is an informational problem, and that this problem is resolvable at the level of physiology, not necessarily genetics.
When you became director, I wonder if you saw that as an opportunity. We spent a lot of time talking about for better or worse, how scientists collaborate, right? In your capacity as director, did you see an opportunity in your own small way to sort of improve some of those processes for how scientists are able to communicate their ideas and how collaborations can be done as productively and as effectively as possible?
Yeah, on a small scale. So our Allen Discovery Center has about ten PIs in it, so it's a microcosm of the bigger picture, and I was very fortunate, to have been able to recruit PIs in different areas that then all had to work together and address these problems from different perspectives. And you could see firsthand what are some of the barriers in communication, especially across disciplines. For example, computational people think very differently about what their mission is, and what it is that they're trying to achieve or understand, and what's important and what are the things that have to be captured and what are the things to ignore. I mean, these are critical decisions that you make in any area of science, and it was my job to try to facilitate those interactions. And so we, I think we did a pretty over all, we were pretty successful with some things, but the challenges are clear.
There are different cultures and different subfields, and I noticed this a long time ago. When I go to give a talk, I can always tell what kind of department I'm in based on which part of my talk makes people mad. And it's always a different part. Because there are things you can say in a neuroscience department that are completely obvious to them, and if you say the exact same thing in a molecular genetics department, people throw tomatoes. And they say that can't be right. And the same thing in bioengineering departments versus regenerative medicine versus computer science. Everybody has a slightly different worldview which means that things that are impossible/crazy/obvious/incremental advances, all of those boundaries are completely shifted depending on what community you're in.
So what were some of the ways that you realized you were being successful in your own small way?
Well, there are metrics like the number and the extent of the publications, you know, grants coming in, postdocs going out and getting their own jobs in the community. I think all of those things have been good. The latest edition of the most popular developmental biology textbook, the Gilbert and Barresi developmental biology textbook, now has two of our stories in it. So now students are encountering this stuff very early on. I think for me what's been the most gratifying to see is the increase in our understanding of how all of this works. And the further we go, the more tightly-integrated are becoming deep ideas from computer science and physics. So it's, the fact that this work has become more inter-disciplinary and not less so, and the fact that it wasn't just that we used some kind of calculation method from statistical mechanics, and then that was done and now we're on to other things-- the fact that some of these really, our latest papers are really using some very profound, deep ideas from neuroscience, from artificial intelligence, from computer science. And these things are all coming together in a more and more clear way, and some of the things that we've seen about what the biological software really is and how reprogrammable it is, and how to understand biological systems from a cybernetic perspective as goal-seeking information processing systems, not just mechanisms. This really tells me that we are on the right track and that we found something deep, because with every new experiment, we're able to pose questions that are more and more different from how people have been thinking about them. And how we know we're on the right track is when those questions suggest novel experiments that produce capabilities that couldn't have been-- that hadn't been done before.
So I always thought it was really important that if you have a different way of thinking about things, it's not enough to just reinterpret old data in a new way. If it's important, you've got to be able to derive something new that you're going to do that you otherwise wouldn't have done. And people ask me this all the time. They say, okay, that's a really weird way to be thinking about all this stuff. Why can't we just stick with the traditional way, and I say because this new way has enabled us to do experiments that nobody else had done. It suggests very specific approaches at the bench, it's led us to new advances and new capabilities. We've shown biological things happening that we would have never found if we weren't thinking about it in this particular way. And that to me is the mark of being on the right track for the theory portion.
Mike, in 2008, you significantly expand your appointments, right? You're at the Forsyth Institute, you have an appointment at Harvard, you also have an appointment at Tufts. So before we get into your unique work at each of these institutes, what was your thinking in terms of having, you know, literally and figuratively your foot in so many different places at the same time?
Yeah so around 2007, 2008, I started to realize that I had been naive in thinking that being at a medical school was going to be sufficient. I started work at the medical school, and I knew that while the cell biology and genetics expertise was first-rate, because it's a med school campus, there were no computer scientists, there were no cognitive scientists, there was no engineers, no physicists. And I thought, that's alright. Harvard and MIT are across the river, it's a 20-minute drive, we can bring all that expertise here. And after about six years, I realized that that really wasn't happening. And that what we needed to do, because our work had gotten more and more interdisciplinary, was to be on a basic science campus where all of that stuff was next door. I wanted to be able to have conversations with physicists and with computer scientists and so on. And so at that time, I started arranging my move to Tufts and that was complete in January of 2009. That's been a great decision in retrospect. It's worked out very well because we have a lot of really close ties with all kinds of different disciplines on campus and it's fantastic to be able to talk to people like Matthias Scheutz in artificial intelligence and Dan Dennett in philosophy of mind and cognitive science. And David Kaplan, who's a bioengineer, and mathematicians. Everybody's next door and you can walk over and talk to them. That's been great.
So this begs the question, why maintain the affiliation with Forsyth? What does that continue to do for you?
To be honest, I think the affiliation with Forsyth expired a year ago maybe or two years ago. We haven't really been doing anything together. The appointment had been maintained by them, but it's been a few years since we've done anything together, really.
So since 2008, your full "home institution," so to speak, has been Tufts?
Correct, well, 2009. So January 2009 is when we officially moved.
And how well-developed was the Center for Regenerative and Developmental Biology by the time you came on board?
It came with me.
Oh I see.
The TCRDB was my center that I established at Forsyth, and it came with me to Tufts. I brought it over.
And the other center that was formed was in 2016, which was our relationship with Paul Allen and the Paul Allen Frontiers Group.
Oh yes, so can you talk a little bit about that? How did you get to start working with Paul Allen?
Unbeknownst to me, Paul had started an effort to identify what he called the "dark matter of biology” for investment. He tasked his people to go around the world, talk to lots and lots of different scientists. It took them about, what, a year, I think, to find out what are the areas in which they think something very important is happening, but is at a very early stage. They wanted areas where the mainstream is not really buying into exploring yet, and significant investment at the right scale could make a massive difference to basic science and eventually to medicine. And so they called me up. This would have been, I guess, in the fall of 2015, maybe? And they asked me to come up to Seattle and give a talk. And I didn't know that anything else was going on. I came up and I gave them a talk, and went home, forgot all about it. And then eventually they called me and they said that, do I have—asked, would I be able to put together a proposal for a center on the scale of $10 million? A center to investigate this thing at-scale. And, absolutely - I've been thinking about this since I was young. So I said yes, of course, I have a plan. I wrote a plan, they had it reviewed, then came for a site visit. We had lots of different discussions, and then they approved it. In 2016, they started two Allen Discovery Centers. One was mine at Tufts, and one was Markus Covert’s at Stanford. And it was four years of funding, and it was incredible – a dream come true. And it allowed me to pull together these other ten or so PIs to work on this problem that I've been thinking about for years.
What have been some of the ways that your research has contributed to the larger vision that Paul Allen was looking to achieve in terms of, you know, his desire and ability to really pour money into basic science in the hope that it would really push the ball forward? What would you say is your contribution to that broader, philanthropic scientific push on his part?
Yeah, I think if we boil down the whole thing to a simple statement, I think that what we've advanced is the understanding of the software of life. I think that what we've found is that while genetics and genomics is really important, what they're doing is telling you where the hardware of the cell comes from. The next step is to ask, what is that hardware capable of? Can it do things other than whatever the default is? And how do we program it? You see, the majority of the field is working on rewiring. So people mutate genes, they change the way that genes turn other genes on and off, so they're literally rewiring the genetic circuits in single cells. This is what all of synthetic biology is about. This is what molecular genetics is about. They're rewiring the hardware. We, I think, have pushed forward a new way of doing things, which is to focus on the software and this idea that if your hardware is good enough, and my claim is that the biological hardware is absolutely good enough, then it's reprogrammable.
And what does "good enough" mean in this context?
Good enough means that it is reprogrammable. It means that it is such, it is constructed in a way that you can manipulate the outcome with inputs or experiences or stimuli, not by rewiring. When I give this talk, the way I illustrate is I have this photo that I show of somebody programming a computer in the late 1940s, and she's of course sitting there, and what she's doing is she's pulling wires out and plugging them somewhere else. So she's physically interacting with the hardware. And that is what programming used to be. But the cool thing about computers and what computer science has realized, is that if your computer is good enough, you don't need to ever touch the hardware. It's reprogrammable. You're going to give it experiences via the keyboard, you're going to give it stimuli, and those experiences are going to cause the algorithm to do something that it otherwise wouldn't have done. And so my argument has been that most of biology today is still stuck largely where computer science was in the 40s and 50s. We are still all about the hardware. Everybody's really interested in single cell molecular approaches, you know, editing the genome and molecular pathways. We're very much focused on the hardware. But what we would like is, is A) to be able to see whether this hardware is reprogrammable, and B) know how you can interact with it and get it to do whatever you want it to do with stimuli, with experiences, not through genomic editing, not through changing the cells, but by altering the information structure that those cells are using to decide what to do as a collective. And we've shown, I think, in the last four or five years, I think we've shown some amazing examples of that capability, and I think the fact that some of these things are possible suggests that biology has a huge future ahead aside from all of the advances of genomic editing and things like this. That actually interacting with the memory, the reprogrammability of the biological software is going to be a huge part of biology going forward.
I want to come back to this idea, this fascinating idea of the "dark matter" of biology. And I assume there that this is a metaphor that takes its cues from cosmology where dark matter and dark energy, you know, are things that are basically totally not understood, and yet we know that they comprise 95, 96% of the universe. So I want to know how useful you find that metaphor proportionately in terms of the percentage that we understand of molecular function? I mean, is that a fair way from your vantage point, knowing all that we don't know, is that a fair way of looking at biology circa 2020?
I think it depends how you ask the question. If you specifically focus on the molecular level and the cellular level, then I think the situation is not bad. I think our grasp of what's going on in terms of the molecules is actually pretty good. And I think the field is doing a very nice job of busting that open. However, I think what's really critical about this, and I think this is where the dark matter analogy is very apt, is that until you've learned to look for it, you don't even know it's there.
And so if you're really focused on the cell and molecular level, things are looking pretty good. It looks like you kind of have things under control. But I think the big elephant in the room here is not the cell and molecular level, but decision making on a large scale, anatomical level. There are many examples, where the fundamental question has nothing to do directly with information on the cell and molecular level but is much better-addressed at higher level of organization.
As biologists, what we're trying to do is reverse-engineer life. We're handed these amazing living forms and we need to figure out how they work. Imagine you were handed a device, let's say it's a chess-playing computer. And you wanted to analyze this thing and you had no idea what it was. You could, of course, spend all your time focused on the molecules. You could characterize the plastic, the silicon, the aluminum, and the copper. You could study all the materials that are there. And maybe you could even use Maxwell's equations and things like this to learn a little bit about what the physics of this thing is. But if you spend all your time at that low level of description, you're never going to get to the point where, of understanding what it actually does, and you're never going to extract the rules of chess, you're never going to learn what algorithm this thing is running, and in particular, what is the most effective way of predicting and controlling its behavior. Think about it from a medical perspective: if your goal is to fix a birth defect or regenerate a complex organ, trying to do this at the very lowest level (of molecules) is incredibly hard.
This is why none of us program at the hardware level anymore, because it's so difficult. Trying to control the features of a face or the location of an eye or anything like that, from the lowest level of individual molecules and pathways is brutally hard. It's unlikely that any of that is going to happen in our lifetime for regenerative medicine of complex organs. But if you knew that this device was a chess-playing computer and you had some understanding of the algorithm and high-level rules of chess (which are independent of any molecular facts), you could in fact do a very nice job of interacting with it in a predictable fashion because you would have some idea of what it's doing, what its goals are, and what its motivations are, and what signals you could give it to make it act one way versus another way. This is incredibly important, not only for the fundamental science of knowing what's actually going on in living creatures that makes them special, as opposed to non-living artifacts, but also for biomedicine. Because trying to achieve these complex outcomes bottom-up is incredibly hard.
And there's another example that I often give my students. If you have a rat and your goal is to get the rat to do a circus trick, to put a little ball in a little hoop, you've got two options. You can go the neuroscience route, and you can say, okay, I'm going to control every neuron in that rat's brain. I'm going to play it like a piano and I'm going to give it exactly the right signals to make the leg and other muscles go and make it do its thing. And maybe that's possible, and maybe someday we would have the complexity to under control, to be able to do that. I think it is certainly not going to happen any time soon. On the other hand, you could use a completely different strategy, which doesn't require you to know very much at all about how the rat works. And that's training the rat. You can give it rewards and punishments - behavior shaping, and just interact with the rat in terms of stimuli inputs, behavior, experiences, and get it to do what you want. People have been doing that with farm animals for, what, 10,000 years? Long before they knew anything about neuroscience. This tells you two things. This tells you that the second approach to complex systems is way more efficient than the former approach. Because you can get remarkably complex phenotypes even before you understand how all the parts work. And it tells you that if your system is the appropriate kind of system, then this is going to work. It's not going to work with a cuckoo clock. You can't train a cuckoo clock because it's not a reprogrammable system. You have to have a system that is intelligent and reprogrammable in its construction, and then you can take advantage of communicating with it via informational signals, not by physical micromanagement. And I think that's the future of regenerative medicine, to understand how that aspect works.
I want to ask another language-oriented question. You know, throughout our talk, you're quite comfortable using technical or computational metaphors to describe biological systems. So talking about rewiring a molecule or the hardware or the software of an animal cell, right? So I wonder, if what that suggests in a very deep way is that long term, these notional distinctions that we create between human constructs like a computer, and natural constructs, you know, from God, the universe, whatever you want to call it, that are not created by humans, if long term you tend to see these distinctions as more artificial than real?
Yeah. That's a very important question. Let's be clear that there are lots of people saying, "Life is not a computer, the cell is not a computer." Okay. What I'm absolutely not claiming is that living things are a computer in the sense that they are similar to the kinds of computers that you and I interface with every day. They are certainly not programed by an outside intelligence, they're not linear von Neumann-style architectures that take one instruction at a time and synchronously execute them, etc. I'm not saying that the standard computer with which people are familiar with is a good model for living things. It's important to realize that there are notions in computer science that are extremely deep, that have nothing to do with the physical implementations that we're talking about here. And this distinction between hardware and software is one of them.
And I think that that concept is incredibly apropos to living things, and I think not having paid attention to it is one reason for the limitations that we've seen in biology and medicine so far. And I think there are many such concepts that need to be appropriated from other fields, physics being one of them. The other one that I'm often called on is that I use a variety of cognitive terms in my talks. So I talk about cells having goals and memories and things like this, and a lot of people actually find that pretty uncomfortable, because they feel that this is mixing terms that don't have good mapping from, let's say, cognitive science onto development or regeneration or whatever. And I actually think that's very important, because there are artificial boundaries between fields, and we need to start erasing these boundaries. Some people think that human brains have unique properties like true intention or real memory or real goal directedness, and that other systems don't, and that when we talk about cells or tissues having memories or trying to achieve goals, that that's a bad misuse of a metaphor. When you make that claim, you are putting up some sort of a magic gap between the human brain and everything else. Where do you suppose we got these amazing properties? If we take evolution seriously, which I think we have to, then we are not allowed to assume that brains have some sort of monopoly on these complex functions, and that other systems’ apparent cognition is just “as if”.
Instead, what you have to do is imagine a continuum of very simple cellular functions that slowly evolve towards the kind of sophisticated things we see in human brains. You're not allowed to posit this magical gap where everything on one side is just chemistry, but humans have real thought. That kind of thing is untenable, given a modern understanding of evolution and computer science. And so one of the fields that we work in is called basal cognition. It's the idea that very simple things, metabolic homeostasis and other things that individual cells do, eventually scale up to complex, large-scale memories, and goal directedness. Not in a magical way, but in the way that control theory, engineers, and cyberneticists have known since the 1940’s. People are very uncomfortable with various pieces of biology having goals, and I suppose that that's because they think back to pre-scientific ideas about animism, but engineers have had a mature quantitative theory of physical objects having goals for decades. Cybernetics and control theory is very comfortable with physical goal-directed systems. And it's time that we integrate all that information into biology. And so, I fight strongly with people who think that this is a misuse of terms. I think these are limiting, artificial conceptual barriers that shouldn’t exist anymore given modern understanding of our evolutionary origins and of cybernetics.
So if you say that there is nothing fundamentally unique about the human brain, right? Evolutionary or in its possible distinctions with computers, does that sort of answer the question about any concepts of there being any metaphysical components to the mind, such as like the location of a soul or anything like that? Do any of those kinds of questions have any interest to you?
Well, they do very much, and this is where I started. All of this came out of my initial interest in the philosophy of mind and in asking what a self really is, and what is does it mean to have first-person perspective? These questions are absolutely very interesting. I think that what we've learned is this: there is no defensible way to think that the human mind is in some way fundamentally distinct from other types of living and soon non-living things that exist in the world. However, that doesn't mean that there aren't special aspects of being a conscious, first person perspective, that are not captured by the science that we have today. This is basically the hard problem of philosophy of mind. If it's a real problem, and I think that it is, but the jury's still out. But if that's a hard problem that exists, you have to think about how this plays out in other organisms. I see this every day; my son is now taking a philosophy class and they talk about human free will and human intentionality and all that. And I always poke at him as I do to colleagues I discuss it with. I say, you're talking about a strict case of your average, normal human, but there are these amazing other cases. There are people with split brains, which seem to have at least two distinct personalities inhabiting the same body with different goals and different psychological makeups. There are conjoined human twins that have a single body with 2 brains and 2 distinct personalities. There are human brain organoids that we can build in a dish that have learning and memory and behavior. They can learn to control artificial bodies and be instrumentized to drive simple robots and things like this. You can move memories from one animal to another by transplantation, either of molecules or of pieces of the brain. You can construct whole animals out of pieces of other animals, living hybrids. All of these categories show that building philosophical conclusions based on a “standard” human brain is very narrow – minds can be implemented in a wide range of bodies, including corner cases of great diversity.
Biology is incredibly fluid, and does not obey the limitations of simple categories. To think about human moral responsibility and freedom, just start rolling the evolutionary tape back, and ask, well, did humans 10,000 years ago have that? Yeah, probably. How about whatever pre-hominid ancestors we had 100,000 years ago? At some point, you go back to a simple primate. You say, okay, well probably we're not going to morally blame that creature for anything it's doing. So there's a nice, smooth gradation between that creature and the modern human; somewhere in there, we developed something that today we would call "moral responsibility." So what exactly was that? There's not a magical moment where there's just a break in this evolutionary chain and a “true human with moral responsibility” appears. We have to get out of the comfortable, crisp categories that we think exist. I'm writing right now a paper on synthetic living machines that we've made in the lab; we've made recently, some amazing synthetic living bodies that are made from frog cells, but they're not anything like a frog. They're doing completely different things. And we now see how this field can develop toward the formation of new cognitive architectures - we can see brains, and simple minds emerging from scratch in the dish. And whatever difficult problems we have about cognition and human brains, you have to be already asking those questions about these kinds of constructs.
Now, one big difference, obviously, between 10,000 years ago and today is that some of the projects that you're working on do have immediate moral implications, because the distinction between humans and machines are becoming less by the day. What do you think are the most important near and long term moral implications as these distinctions increasingly get fuzzier? Even in the course of our lifetime, not even looking ahead 100, 200 years.
Just to get one thing out of the way, I'm occasionally asked about moral implications of our work, which shows that you can have really great control over anatomical structure. We can make six-legged frogs and worms with two heads and things like this. And so sometimes people get worried about that. So I would like to sort of dispel that right off the bat. I think the amount of suffering in children and adults around the world makes it completely obvious that we have a moral obligation to pursue this work, to really drive regenerative medicine to the point where everybody can be healthy. I don't think there's any defensible moral stance for not doing this kind of work. I think that's simple.
The more complex aspect is, of this work, I think that what it's doing is really zeroing in on what it means to be a creature to which we owe some sort of responsibility. In the past, it was kind of obvious that, here are some living things and they have preferences, meaning they can suffer and they can enjoy happiness, and therefore to be a moral agent, you have to, for lack of a better word, be nice to them. Of course, that line has been drawn all over the place in human history. There are lots of societies where even being a certain kind of human was not enough to put you over the line of having rights. What we are seeing now is really a focus on trying to understand, what does it mean to be a system to which we owe something? So if you've got a Roomba, presumably you're not terribly worried about what happens to it, but at some point we're going to have artifacts that are so close to the kind of performance and information processing that we see in various kinds of animals, and possibly even humans, that it's going to be really not obvious what the barrier is.
And in fact, if you look forward, we're going to be surrounded by both constructed, meaning silicon, and biologically-grown, organisms that span a massive range of intellectual capacities. Some of them will be smarter than us in various ways, some of them will be much more limited than us. They will have different types of construction, whether they be alive or not. All the boundaries that we're used to are going to be erased. We're going to be faced with a multitude of every possible embodiment of mind. Different kinds of minds form simple to complex and better and worse than ours in different ways. It's going to be this massive proliferation of autonomous agents. And I think here's where the job of the ethicists really is going to get tough in trying to define some particular version of the golden rule that works for a situation where everything is up for grabs. Both structurally and functionally, you're going to be surrounded by a great diversity of creatures. There are going to be hybrid ones that are part biological and part not. There are going to be humans with various kinds of implants. There are going to be humans sharing brain data with each other directly. There are going to be various biological robots that were grown and developed and they have certain features of mammals and so on. There's going to be a massive proliferation of both bodies and minds around us, and the big challenge is going to be to figure out what we owe to these various creatures and how do you know what you owe a certain creature.
I don't want to get into a political discussion about where we are currently, but I am curious if you saw Andrew Yang's campaign as sort of a breakthrough moment in political discourse that AI is having massive national implications and it's not far off? We need to start thinking about these things now.
The thing is, advances in automation are going to have massive implications for the economy and for all areas of society whether or not any of this is really intelligence. So I think those two things are separate. We can have philosophical discussions about what it would take to get to general intelligence. Whether intelligence alone is enough to confer moral responsibility. And so on. The outcome of those things don't impact on the bare fact that just how excellent the automation is getting is going to have massive implications for society. This happened before in the Industrial Revolution, where nobody thought that the machines were intelligent. And yet the automation turned everything upside down. And this is going to happen again. Whether or not we believe that trucks that drive themselves, whether we owe anything to trucks that are self-driving, the reality is that their very existence is going to change a lot of the economics. We are absolutely going to feel the impact of all of these technologies. Not the least of which is this. The era in which you could have tangible evidence of something happening and show it to someone else is at an end. Think about it. For all of humanity up until mid to late 1800s, let's say, we had no way to objectively document anything for others. If you saw something, you had no way of proving it except telling someone, and they could believe you or not. You had no way of proving it.
Then we had a brief century and a half of photographs, where, I could get a picture or a video, and I could show it to people, and then everybody would see for themselves that yes, that really happened. You know, whether it be for purposes of police work or social justice or whatever it's going to be, you could take a video or photo and you could show someone, and that would be incontrovertible proof. That's over now. With all the deep fake stuff, and the ability of machine learning algorithms to in-paint scenes, to out-paint scenes, and to come up with convincing fake pictures, videos, images of people, of environments that never existed etc. - these are very difficult to distinguish from the real thing, and the ability to prove what actually happened with some kind of a record is gone. That was a brief blip on the human experience. And now, as before, somebody shows you a picture, you say, "How do I know that’s real – that an AI didn't create that?” You can't really trust almost anything you see anymore because these enormously powerful content creation algorithms are becoming widely available, for example to fake people’s voices and their facial images saying things they didn't say. In ten years, anybody is going to be able to produce a photo of anything they want. So these kinds of software systems are going to have massive implications for society in every possible way. And it doesn't hinge on them being morally equivalent, or conscious systems or anything like that.
What has been some of the research endeavors you've been involved with in the more recent past? The past four or five years?
My lab now works in a number of different fields. For example, we've been trying to achieve limb regeneration. And we've more or less solved it in frog. We're now trying it in rodents, and the idea is use particular kinds of bioreactors and bioelectric signaling to induce the regeneration of limbs - that's on the biomedical side. On the basic biology side, we've really been working this reprogrammability angle, and we've shown in the last few years that living tissues encode an electrical map of what their correct pattern is supposed to be, and if you injure them, they regenerate-- and this is particularly our two-headed worm work-- if you injure them, the cells consult that map to know what to build. And what we've shown is that you can actually rewrite that map, rewrite that electrical pattern in the tissue, and then the cells will build to whatever the new pattern is. And if the new pattern says two heads, well then the cells will build the two heads.
So this is an example of that reprogrammability, because we haven't genetically edited those cells at all. They're still the very original cells. What we found is that A) these cells read a particular recorded pattern in order to know what to do, and B) we've decoded an aspect of that pattern so that we know how to change it to make the cells do what we want. And this is just the beginnings of the science of reprogramming growth and form, so that ultimately you will be able to rewrite these pattern memories to whatever you want and get the cells, I think, to probably build almost anything you like. And so that's an example of the synthesis of a kind of cognitive science, and computational science perspective with biology, in that we are treating the cell collective as a cognitive agent that has an electric memory of what to do, and what you're trying to do is convince it to do something else as opposed to trying to micromanage every individual cell signaling to force it to do something.
So that's another thing we've been doing. We've had some nice progress on the bioelectric control of cancer. We've done some work on basal cognition and really trying to understand how slime molds and cells and tissues process information in making decisions to do what they need to do. We showed a few years ago the first model of regeneration that was discovered by an artificial intelligence we created. So this is kind of important. This is our effort on creating new machine learning tools for a bioinformatics of shape - moving beyond bioinformatics of gene sequences and proteins. We want to now get to a bioinformatics of a three-dimensional structure. And understanding functional experiments that people do that are already overwhelming the ability of any single scientist to be able to synthesize them all into a model. We created the first machine learning tool that tries to understand what's going on from a set of, I think, between 800 to 1000 papers on planarian regeneration, and to synthesize them all into a testable model of what's going on underneath. So in a way, trying to automate the most creative thing that scientists do. It's not just number crunching, but actually looking at the functional data and trying to take a guess at what the system is doing underneath. To try and generate hypotheses.
And then the latest thing is our work on the Xenobots. This had extensive coverage, including the New York Times, and lots of other places a few months ago, where we showed the amazing plasticity of cells, normal, unedited cells, taken from frogs, that could re-envision their multicellularity to build a new synthetic living machine with some guidance from an evolutionary computation algorithm simulated in software. It's the only living thing that I'm aware of that evolved not on earth, but in the virtual world of a computer. Basically, its evolutionary history was completely virtual, in a computer, and then we took the cells and we let them recombine in a particular environment and they implemented that novel body type with new behaviors, with new functionality, and yet a perfectly normal frog genome. And so that, I think, is pretty important because it tells us not only the plasticity of cells and how little we understand the relationship between the genome and the anatomy, but also it's the future of computer-designed organisms and synthetic living machines. That we're going to be able to make all kinds of novel bodies and minds as we talked about a minute ago.
Do you ever worry about opening a Pandora's box, where at some point humans are no longer able to determine what evolutionary developments happen when we're talking about, you know, new organisms that are outside of nature.
Yeah, I think that's absolutely an issue, and it's a bigger issue than just computer-designed organisms. We are surrounded by, and we are increasingly going to be surrounded by systems that we are not any good at predicting or controlling. We're talking about internet of things, we're talking about swarm robotics, we're talking about complex social and regulatory structures that have large-scale behaviors that are very difficult to infer or predict from what you know about the local environment. So you create something, you kind of know what it does, then you deploy a billion of them all over the world, and now the collective does something surprising. If I told you what the behavior of an individual termite looks like, you'd be hard-pressed to guess that what a huge swarm of them is going to do is build this massive mound of a particular structure with temperature control and all this other stuff the colony does. So these unintended consequences affect all swarms, whether they be swarms of cells or robots or people or ecological webs.
The interesting thing is that we face the exact same problem everywhere. So we have unintended consequences of social policies, we have unintended consequences of cellular rules, where you have cells with particular behaviors, and all of a sudden, bam, they make a salamander, and if you cut off the tail, it regenerates the tail, and you did not know that from looking at the cell or the genome of that cell. So and the same thing with swarm robotics and internet of things. We face this massive unintended consequence of swarm behavior, and I think this is the challenge, and I think it's an existential challenge for humanity, is to learn to program swarms. I think it's absolutely essential and our work on cells is, I think, part of how we're going to get there. So I think, yeah, this is absolutely a danger area. But we're heading there no matter what. And the only way we're going to survive it is if we learn, in the next couple of decades, we use tractable model systems like cells, like swarm robotics, and so on, to learn to determine the goals of swarms. So that when you deploy a bunch of these agents or a bunch of rules or a bunch of organisms, you have some control of what the swarm is going to try to achieve. It all comes back to goal-directedness. We have these goal-directed systems, who sets the specific goals? We need to understand this better.
Mike, for the last part of our talk, since we've already gotten up to the present day in terms of the research, I want to ask few broadly retrospective questions about your career, and then of course some forward-looking ones, which obviously has already been a theme of the things we've been talking about. So first, I want to ask, are you asking the same kinds of research questions that you started asking as an undergraduate? In terms of all of the ways that your scholarly interests, your research interests, have expanded over the decades, do you still remain connected to the original things in science that you were curious about? And are those curiosities still at the heart of what you consider to be the most important research questions even today?
I have to say that yes, in a certain sense, the questions that I started out, that drove me to be a scientist in the first place, is wondering about how mind and body interfaces. How goals and preferences and things like this can be created out of physical matter. I think that's still the number one question. The way in which I formulate those questions, I hope, is better than they were 40 years ago. But the fundamental questions are still the same. I think it's all about how information processing occurs in a physical world, and how evolution makes use of the laws of physics and computation to give rise to minds and bodies that do these amazing things that we see everyday. And I think that is still, that's what drove me to science, and I think that's still really the big question.
Are there questions that invariably hit a dead end in terms of your capacity as a scientist? Are there questions you have that by definition, the kinds of things that you're involved in are best-answered in a philosophical or even a spiritual context, that you're interested in?
There are lots of things that are important that we need to know that we have no idea about how to address empirically. On the one hand, I think research, in the sense of asking nature questions and getting an answer, is really the best way to address almost every question. Because in the absence of that, it's completely unclear how you know when you've found the right answer. We have ways of generating lots of different answers. People disagree about stuff, but we really as a civilization have really found only one broad method for coming to grips with knowing when you've got the right answer or not, and that's empirical testing against the real world. So overall, I'm still very committed to science as a way of knowing when you're onto the right answer or not. Having said that, I think it's very important not to believe that our ability to engage with the real world in experiments is good enough to give us reliable answers to many of the questions that are really fundamentally important. I mean, many of the things that we care about, we don't have any good ways to really study them. And these are issues of moral responsibility, these are issues of consciousness, although I'm aware that lots of people try to study consciousness. It's actually very, very difficult to do.
There are massive questions like this that we simply don't know how to answer. Not only do we not know how to answer them, we don't even know what a correct, possible answer would look like. If you're trying for, just to pick an example, if you're trying for a theory of consciousness that would correctly tell you what it's like to be a roomba, a Xenobot, some animal or a computer, whatever: what would an answer even look like? We don't even know how to represent a possible answer to that question, nevermind determine when the answer is correct.. So I think there are massive gaps in our understanding. I think fundamentally, we know way less than we think we know. And I think that I personally have a much lower than average threshold for ideas that overturn our conceptions. When people say, "Oh, that's impossible", I think this is real hubris - I don't feel that we really understand enough in a lot of deep areas to be able to reliably say that something is possible or impossible. I think we're just scratching the surface in a lot of the important questions. But really there's only one path forward - rational inquiry and as much as possible some sort of testing so that we have some idea that we're on the right path.
Mike, if you were able to look at all of your research accomplishments, how in assessing them in aggregate, how do you define success with any one of them? What are the feedback mechanisms that you know, you know, absent some breakthrough in a clinical trial or a commercial viability or things like that. When you're talking about ,you know, just doing basic science to try to get at some of these fundamental questions that you've been asking for the past 20, 30 years, how do you define success on any given one of those projects, and what does that say about the way that you do science? The way that you conduct yourself as a researcher?
Yeah, this is a great question, and I try to pose this to my students and postdocs all the time - it's really critical to try to define what success looks like. You know, what are you hoping to gain out of any effort you undertake? For me, success in this kind of enterprise, aside from any tangible outcomes, does things. One, it makes it very clear what needs to happen next. So I think a successful breakthrough, either on the conceptual side or on the experimental side, is something that says, "Aha! Given that, this is what you have to try next." It makes it really clear what your next activity should be. It pushes you towards novel experiments and novel thinking, modeling, computational analysis, whatever it's going to be, that you would never have done if you hadn't completed this step. So something that pushes you forward to new capabilities and new questions.
Second, a successful outcome is something that makes the overall pattern of how you understand the world to be clearer and more coherent. So to the extent that we've asked a series of questions, we've gotten some answers. All of this was successful if at the end of this, it now makes it more clear what in fact is going on in the world. Maybe in a multidisciplinary sense, it helps you to generate a coherent worldview that is simpler than all the naked observations that you have. I forget whose metaphor it is, but somebody said, you know, the edifice of science is you're building this building and some people sort of throw bricks into the yard, and that's good, but what you really need is to be able to assemble the bricks into a coherent edifice that's more orderly than a bunch of bricks lying around. And to the extent that in your own mind, and hopefully in the mind of the community that cares about your work, to the extent that this contributes to a vision that's bigger than the sum of its parts, that makes it more rationally understandable how life works. That's, to me, a successful outcome.
I want to ask two sort of "present-ist' questions about coronavirus. The first one is, do you see a way for the kind of research that you've done to be useful in combating COVID-19? In other words, one of the things that's very exciting to witness right now as a member of the broader scientific community is the way that researchers and scientists from vastly diverse fields have involved themselves in this global effort to deal with this unprecedented pandemic. I'm curious if some of the things you've been involved with, you know, over the course of your research, you see as particularly useful in dealing not just with this, with the particularities of COVID-19, but with the way that international society deals with pandemics generally?
Unfortunately not directly, no, there's not really anything that we're doing that directly impacts COVID research. Indirectly, though, I wrote an op-ed piece where I argue for the de-centralization of science. I call it the "science at home" movement. The idea is simple. It's just basically to link together all of the advances that have been made in things like the democratization of science research in developing countries, frugal sciences, all these different movements that aim at the same thing, which is to keep research going when you don't have access to large lab. If not for this epidemic, then for the inevitable next one, we're going to get kicked out of labs again. It's going to happen. So the only thing I can say on this topic, is that I think we have to pandemic-proof our research enterprise, and as much as possible, we have to figure out ways to do as much research as we can without having to be in the lab with other people. So on the large scale, this means automation and things like that. On the small scale, it means setting up the infrastructure to allow experiments that people can do at home and there's plenty of them. I know it sounds crazy, and there's a lot of things you couldn't possibly do at home, but there are many model systems that you actually can. And so what we don't have is the infrastructure to allow people to do this when everybody's kicked out of lab. And I think now is the time to start setting this up, so that we're ready for the next pandemic and research doesn't grind to a halt. There are ways to keep things going when people cannot all be in the same lab. So that's about the only thing I have to say on that topic.
One of the many structural problems that coronavirus has revealed. I mean, there's so many crises that have come as a result of this. One of them, of course, is the disconnect between science and society. I mean, the fact that we are living now in a time when wearing masks is a political statement, and we now see literally, you know, a new wave of this pandemic which is entirely avoidable if we did this properly from the beginning. It clearly is suggestive of a massive disconnect between science and society. And I want to ask this to you, not because obviously of your work with coronavirus, but because so many of the things that you're doing now currently and in the future, and in even more direct way, will have these massive implications on society. So I wonder if you could talk a little bit about the ways in which scientists can improve at least their end of this communication breakdown, so that society at large, or as many Americans as possible, have a greater appreciation for what it is that scientists do and that that communication can ensure that this relationship, this broader relationship between science and society does not play out in these destructive and counter-productive ways that we're seeing currently.
It's a difficult question. I think that overall, what we've seen since recorded history and especially since the scientific revolution, is that people's behavior as a group has numerous motivations; rational thought is only one of those motivations. I don't think this problem is modern. I think this goes back all the way to the beginning. There's only so far that scientific arguments are going to work - some people are just not motivated by it. So I think the best thing that science can do, and I think that the really the only way to win over the most, the biggest segment of society we can towards rational thought and progress, is to continue to demonstrate value via applications. I think it's really hard to get everybody on board with the love of research and the importance of research for its own sake. It's critical but it's just going to be really hard to get full buy-in on that.
And so I think the best tool that we as scientists have for getting people to go along with science as a critical way to guide their life is to continue to show the success of science in improving their life. And this means biomedicine, this means environmental impact, it means economically. We need to continue to elevate the human condition in every possible way and nourish practical implications for well-being. And to be really explicit that this is brought to you by science. That needs to be part of education from day one. Someone said once that science is the reason that the world makes sense and that we can have nice things. It's only because of science. And this needs to be widely-advertised.
I want to ask, Mike, two last questions. They're both forward-looking, and of course, it's so difficult to predict the future and where things are going, but I wonder if you could use the prism of some of your most impressive graduate students and the kinds of things that excite them, to sort of give a broad overview of the kinds of issues that are engaging the best scientists of the next generation. What are those topics that are greatest import, both scientifically, technologically, and socially, that your graduate students are engaged in that you see as the vanguard of where all of these issues that you've been working on in your career, where they are headed for the next generation?
What I see in the next generation is really a very clear appreciation of short term impact that their work is going to make. So people are thinking about, can we use synthetic biology to impact the meat industry and food and the way that food production works? What can we do with robotics that's going to improve the human condition? You know, these kinds of things. People are very keenly thinking now about how these various, how their own efforts are going to impact society broadly. And I think it's a very hard problem because oftentimes we don't know. There are lots of examples in science about research that led to applications that nobody could have envisioned, either for good or for ill, so it's hard. But I think those kinds of things are, people are very much-- What I'm seeing is much more engagement with the impact on society as opposed to just the focus on pure learning.
So the last question, Mike, it's a very simple question but I'm sure it'll be complicated to answer it. It is simply, what for the remainder of your career, what are you most optimistic about, and what gives you greatest concern? Regarding the issues that you've been researching over the course of your career.
That's an interesting question. What I'm most optimistic about is that I think the ability to integrate the different tools and different disciplines. It's going to really revolutionize our understanding. The ability to finally make living things from scratch -- when I say "from scratch" I don't mean form molecules, but even from existing cells — is going to really revolutionize some things that have been stuck for a long time. This gets to some philosophical problems, some deep conceptual issues of evolution. And I'm very optimistic that this kind of multidisciplinary consilience of developmental biology, cognitive science, and computer science, is going to drive really transformative advances and our understanding. The greatest concerns I have is, I suppose, just on a practical level, I think we have a real difficulty with unintended consequences. And I think we as a whole have not grappled enough with the scale-up of goals from individual agents to swarms. And I think this has really the potential to be a fundamental problem on a civilization scale if we don't get hold of it before our technology outstrips our ability to program swarms. From robots to living things, genomic editing, whatever. It all boils down to the same thing. Our understanding, our ability to drill down into details is getting better and better. Our ability to go back up and synthesize that information into a system-level-- despite the emphasis on complexity science and emergence and all that-- I think a lot of things are still missing from that whole area. And I hope that we can get there before our technology outstrips our understanding of how to do this, because it's an existential risk.
And just as a follow-on to that, both in the way that you describe what you're most optimistic about and what gives you greatest concern, it sort of begs the question that, you know, somebody can look at what you're doing where you are almost creating new life, and they can criticize this and say, you know, well what you're doing is you're playing God. And you might reject that because you might just say, well I don't believe in a God to begin with, so there's no concern for me acting as if there is something else. And yet the concern you have is that as a human, we are not able to see ahead of the curve. And of course, in the Judeo-Christian concept, that's one of the things that God can do. God is not bound by time, God can see the future. And so I wonder if you see any legitimacy to this concern. Not by your ability to create life, but in your-- Not "yours", in human's inability to see fundamentally where that's going, and so therefore the question would be, you know, maybe not in a spiritual sense, but in a moral sense, would somebody who says, looking at your work, this is problematic because you're playing God. Do they have a point to some degree?
I've heard this question before, and I don't think they have a point. I'm with that person in the sense that it is absolutely critical to weigh the consequences of our actions. I'm also 100% on board with the fact that we need to be responsible to ourselves, to our children that will follow us, and so on. However, the problem is this. Those people are usually pretty reasonable at evaluating the risks of action. What they fail to do is evaluate the consequences of inaction. Because it's very easy to say, "Oh my God, if you do XYZ, here are the things that are going to happen." But you never hear that weighed in comparison with, "Okay, but if we don't do that, here are all the things that are going to happen," right? People have this sort of implicit belief that you could screw things up if you do the wrong thing, but if we just don't touch anything, we're going to be fine. And I think that the history of humanity on earth says that you're absolutely not going to be fine if you don't do anything. And we know what it looks like when you don't do anything - life is short and nasty if you don't understand the world around you. Life is not good. And I think it's very clear that if you make a rational comparison between the potential consequences of doing research and not doing research, it's super clear.
I get phone calls every week and they come in two flavors. Some people say, "How dare you do this work? It's scary and I want you to stop." And then there are the people that say, "How come you're taking so long? My kid has a problem with XYZ. We need medical treatment. Why are you taking so long? Hurry up." And so all of those people need to get on the same page as to the fact that there is only one place that drives advancement in the quality of life, and that's science. There is no option of staying in some sort of Garden of Eden where everything is great and we just need to not mess it up. That option is off the table. We have a lot of problems that are absolutely going to be real civilization-level issues, and research is the only way we're going to make it through. So I think if you're worried about what any scientist is doing, weigh that against what's going to happen if that scientist does not continue. And see what happens.
Right. Right. Mike, it's been so fun talking with you. These issues are so important, and I kind of feel like, you know, we should check in in ten years and see where everything is, because it's all so fast-moving. So--
I really am so glad that we connected, and I really just want to thank you for spending the time with me today.
Cool. Thank you so much. Yeah, it's been a lot of fun.