FOCUS In Sound #36: Leenoy Meshulam
Octopus during active sleep video:
Welcome to FOCUS In Sound, the podcast series from the FOCUS newsletter published by the Burroughs Wellcome Fund. I’m your host, science writer Ernie Hood.
In this edition of FOCUS In Sound, we get to know a young researcher who in 2022 was the recipient of a Burroughs Wellcome Fund CASI award, the Career Awards at the Scientific Interface. Those awards recognize outstanding scientists who have made significant contributions at the interface of biology and quantitative sciences, bridging the gap between disciplines, and fostering innovation. It’s a five-year grant totaling $500,000.
Leenoy Meshulam is a theoretical physicist who is also interested in biological phenomena, especially nervous systems and the brain. She explores the interface between physics and neuroscience. She received her PhD from Princeton University, after completing her master’s degree in physics and biology at Tel Aviv University. She is now a Swartz postdoctoral fellow at the University of Washington in Seattle.
Leenoy’s research endeavors have already taken her in some fascinating directions, which we will hear all about, including a remarkable recent publication about the sleeping habits of octopuses.
Leenoy Meshulam, welcome to FOCUS In Sound!
LEENOY: Hi Ernie. It’s very nice to be here.
ERNIE: Let’s start with your latest accomplishment as a co-author of a paper in the journal Nature called “Wake-like skin patterning and neural activity during octopus sleep.” Tell us about the overall findings…
LEENOY: So this paper concentrates on a main finding where we saw that the octopus, much like in other animals, actually has two stages of sleep. So we managed to prove that every about hour the octopus goes into a different kind of stage of sleep—an active sleep bout, where similarly to REM sleep for humans, for example, the type of brain waves and the neural activity changes, and this is accompanied by a lot of color and pattern changes on the skin of the octopus while the octopus is still asleep. And so the way the cycle looks is, we have about an hour of sleep that is not the active part, and then a few minutes bout kind of like REM, with lots of color changes on the skin, different kind of neural activity that accompanies it, and then back to the stage that is most of the sleep. Something to keep in mind is that this is the first time that neural activity was recorded in this way in the brain of an octopus. So this is a sleeping octopus we managed to put a Neuropixel in, which means that there is an electrode inside the brain of the octopus that’s, we’re able to filter brainwaves out of to see what the signals look like. Also to see spikes in the brain of the octopus. And we can really see the activity of the brain while this is happening. So it’s both the underlying activity of sleep and the behavioral aspect of the two stages of sleep, and very high-resolution filming of the pattern changes on the skin of the octopus. The skin of the octopus is normally a system that we’re looking at for things like the camouflage of the octopus, right. This is why it has the ability to change so much color to have these coordinated, beautiful patterns that look like the coral reef that it is trying to match behind it. It also has texture changes there. And we just didn’t know before that this happens during the sleep of the octopus. So there are just these bouts of color changes that happen during sleep.
ERNIE: So you were really able to correlate the visual and the brain phenomena?
LEENOY: Exactly. So this is the underlying neural activity to this active sleep bout that makes it so interesting, because we can actually look at what is happening in the brain and show that this is really sleep. You can really see when you’re looking, if you look at the plotting in the paper, you can see the immediate, sharp transition into active sleep that is change in the neural activity, and if you’re looking at the brightness of the color of the body of the octopus, you can immediately see a drop, because it starts having color instead of being transparent. So the transition is immediate and is completely synchronized.
ERNIE: Just so everyone is aware, we will post a link to video of the sleeping octopus along with this podcast. It’s well worth seeing! Leenoy, I understand that these findings point to the idea of convergent evolution. Would you elaborate?
LEENOY: There are multiple elements of the system of the octopus that are similar to what we see in other animals. And because the cephalopods, which is the family the octopus is part of, so that’s cuttlefish, octopus, and squid—these are the cephalopods—the convergence in the evolutionary tree from us, for example, we’re talking 600 million years ago. That is around the time that on earth we moved from single cellular organisms to multi-cellular ones. It’s a very, very long time ago. And yet, there are a lot of parallels. So for example, the visual system, the eye. The octopus has a camera-like eye, much like ours. It has some things that are different, and some things that are very much the same, the basic structure of camera-like. And with the sleep, we didn’t know if this is a system that is similar to ours or not. We didn’t know that they have two stages. Humans definitely have two-stage sleep that is very well researched. We know that there is offline learning that without which we can’t actually make progress in learning. It is incredibly important to have rest periods, like any person who tried to play an instrument or tried to juggle or tried to learn anything, motor or not, knows, right, without rest, without sleep, both of these things, there are no leaps in the learning. You can only get so far. And so that two-stage sleep is something that is incredibly important, and we know from human performance and from other animals. The fact that the octopus has two-stage sleep as well definitely points to thinking about this as, oh, we have two types of stages of neural activity, two types of the behavior during the sleep, and maybe this is also a system that is a parallel to our system, and the mechanism behind it might be the same. And because we are able to penetrate the system and measure the system in a very different way than what we do in humans, for example, the fact that we can actually see the output of the motor system on the skin, basically record the entire output, because we can film the entire octopus (definitely can’t record the entire brain of a sleeping human, right?) suggests that if there are parallels we might be able to get to them by researching this animal, where we have a different kind of access to a system that might be functioning in a similar way.
ERNIE: So what was your role in the project?
LEENOY: Right, so maybe the most important thing to say is that I’m only the theorist on this project. The main thing that was happening here is that the incredibly talented and incredibly hard-working group of Sam Reiter in Okinawa, the Okinawa Institute of Technology, that’s a university that is in Japan, the Okinawa island, has amazing facilities of marine biology and marine neuroscience to be able to research things like this. This group concentrates on cephalopods, and fortunately for me, they have agreed to collaborate with me. So there are multiple people who are driving this project, who did all of the filming, all the looking inside the brain, all the imaging, all the recordings of the surgeries, and have lifted this incredible, incredible project with Sam, the PI of this group, who designed this all and looked into everything, and that’s just an amazing, amazing feat that they achieved. The collaboration with a theorist like me comes in with looking at both the short-term and the long-term goals of this project, designing the ideas to begin with, such that we have experiments and computational work and theory work, because you need all of these elements in order to create understanding of the system. Recording is not enough, computational models is not enough, you need to think through all the elements of the scientific endeavor. It is a very interdisciplinary science. We have physics inside it, and coding inside it, computer science and biology inside, and molecular things and imaging things, and that’s a lot of types of skills that had to come together in order to create this paper. So I am the part that is more methodical, that is more looking at computational things and thinking how to design those things to begin with, because we have a lot of thinking through the long-term goals of this project, and what is the kind of resolution we would like to measure something in. And it’s the collaboration, the multiple viewpoints, I think, that is the important part that is underlying this research.
ERNIE: Leenoy, how did this collaboration with the Okinawa Institute of Science and Technology come to be?
LEENOY: Maybe something to mention is that I am coming from a background of physics. I am a physicist who’s interested in biological systems, and specifically in the nervous system and in the brain more than anything. Like we said, the system of the octopus is a very special one, where you are able to actually record all of these pigment cells on the body, we call them chromatophores. These are specific pigment cells that change for the camouflage and also during sleep. There are about a 100,000 of them on even those small octopuses, that cover them. And you can measure their color and their size for a long time very well in very high resolution. That means that this is a system that for a person like me who is in interested in systems where we have a lot of small elements that are pretty similar to each other and interact to create a global pattern, something that is emerging for a global pattern, needs to be coordinated from a lot of small interacting elements, it’s the ideal system. My Ph.D. was on recordings in the hippocampus of mice, where you could even 2000 neurons that are simultaneously recorded, and you could definitely look at the kind of physics I’m talking about, which is looking at all 2000 and looking at the collective behavior of them, looking at the global patterns, looking at what emerges from a lot of interacting things together that individual ones could not carry on their own. And after this I thought, well, really what we would like is to be able to look at an entire system like this, and to be able to measure to the highest resolution each one of the small elements. And the octopus system, the cuttlefish system, this camouflage is something where we know the function really well and we can measure the underlying elements really well, and look at the entire system. Sam Reiter, who is the senior author and experimentalist on the paper that we’re talking about, had a beautiful science paper on the camouflage of cuttlefish. And after they published it, I read it, and I looked at it and thought, this is just the perfect system for doing the kind of theoretical work and the kind of computational analysis that I’ve been most interested in. And I contacted Sam, who said that he is starting a lab in Okinawa and he would love to collaborate, and I am grateful to this day. A very important element of our ability to collaborate was the fact that OIST, the university, has a Theory Visiting Fellows program that was incredibly helpful in me being able to visit and spend months in Japan in order to design this project with them, to help in whatever way I can, to my work with them, and to think through our future goals together. And so this program basically allows for residency of theorists, having big white boards and big blackboards around, to be able to come and discuss and interact with the institute, which has been incredibly insightful and really helpful to be close to the lab and to the data and to the octopuses and to the researchers who are doing this incredible work.
ERNIE: Will the work on octopus sleep continue? And will you continue to be involved?
LEENOY: Absolutely. First of all, as long as Sam and his group allow me to be involved, I’d love to keep being involved. I’m feeling very fortunate that they do. In a very immediate way, we have the very next steps coming up, which is that there is more computationally heavy parts and theoretically heavy parts that are a different paper that we are going to eventually wrap up, that have to do with different kinds of calculations than the main finding that we’re seeing here. And then after this, we would love to be able to record from the octopus when it is also awake, not just when it is asleep. And I’m sure you can imagine that this comes with incredible technical challenges, but also this group is so good that I think they’re about to figure it out, so a lot of motion, right, being able to put an electrode, as a rigid thing, in a super-flexible animal that has no bones, that makes for a lot of noise artifacts. However, if you’re able to stabilize the system and to prevent some of the movements or have something that is more flexible in the recordings, then maybe you can have more neural recordings and you can do more and look at the awake octopus and the camouflage, and see how this affects what’s happening during sleep. Do they have any types of learning or effects from what has been their experience during their wake time to what they seem to be replaying on their skin during sleep time. I’m very much looking forward to that.
ERNIE: We will certainly keep our eyes open for that (so to speak). I know that much of your previous work, Leenoy, has been on the theoretical side, so I have to ask, was this an especially fun project for you?
LEENOY: It’s a great question. It is an especially fun project, but I feel sometimes like all the projects seem to be especially fun in their own way. Octopuses are an especially fun system, maybe that’s the best way to say it. In a biological way, there are a lot of things about them [that] are very special, and not that many people work on them. So the kind of neural activity that you look at and the kind of challenges that you have to face and think about are very different. It is also incredibly fun when one has a very nice mathematical problem to solve, or when you’re looking at mice who play video games and you’re looking at their memory, right, which is an element from things that I did before. Or trying to make decision making, which is part of the projects that I’m having now. So I think every model system has its own kinds of fun. Octopuses are particularly fun in the sense that nobody knows anything about them, so you get to have just the best conversations. And maybe a really fun thing that happens specifically with octopuses is that the first basic questions of an audience when you give a talk or somebody engages on this topic are always incredibly interesting, because people would like to know how smart are they? What does their nervous system look like? Do they spike? Do the neurons, are they active in the same way that ours are? Is their consciousness the same? What can we expect from them? How do we think about the nervous system? Is it concentrated? Is it distributed? And so the questions are incredibly fun. Finding out these things that in other systems we have already figured out and in this system we are yet to find out is incredibly fun. Finding out what are the right questions to ask mathematically is particularly fun in this system and the thing we are finding most illuminating. And interacting all over these kind of new landscapes is very wonderful. This new advantage of technology that has progressed, that Sam’s group now have, like this high-resolution, the neuropixels and recordings in the brain, that’s all just new technology that is now allowing us to start looking into the system in the kind of rigor that we can do in other systems, and ask critical questions. And this is incredibly exciting.
ERNIE: So Leenoy, let’s explore your research on the interface between physics and neuroscience. What is the relationship between the two, and how is it shedding new light on the principles of brain function?
LEENOY: Thank you for asking this. I think the basic principle of the subfield of physics that I’m coming from, from statistical mechanics, and of brain function, are in a way very similar. So let’s think together about how a magnet works, for example. This is something we have the physics of very well figured out. A magnet is made of a lot of spins. Every spin can have a direction, can be up or down. But magnetism as a property is not a property of one spin. Only if you put all of them together, then you get a collective property, something that emerges in the system. A magnet, all of it together has magnetism, not one tiny element. This idea of a collective property is language and math that you can see immediately how it’s very attractive for the brain, because one neuron doesn’t have a function, cannot carry out, throw the ball, cannot underlie decision-making. It does not have any consciousness. But many neurons together have something magical happening. They are together able to coordinate complex patterns of activity that underlie all of the behavior we have, everything. And so, the idea that we can take this kind of physics, this kind of math, and think how to create this for the nervous system, I find very attractive. And this is also why I am working in this kind of interface, why I think eventually the entire field of physics of biology, and physics of life, that taking these kinds of concepts not as an application but as we are trying to create physics and trying to create concepts and trying to understand the principles of a whole new kind of system, of a biological system that is active and is changing and is dynamic and is a lot more complex than a magnet and has many functions, is going to give us some avenue to understand and to gain insight into both the mathematical principles that are underlying it, but also the function, because we definitely need new kinds of insights, right, we need new physics and new math and biology, because the brain is incredibly complex. We are still trying to figure out even very basic questions. I remember sometimes when I give outreach talks in schools to children, I start talking about the brain, and the very first smart girl who raises her hand asks, “How does memory work?” And the first thing that happens is I need to say, “Well, we’re not quite sure yet!” And it is amazing, because when you ask a very basic question in physics, it seems like we have figured out some of the kind of childlike questions, or at least many of them. And in neuroscience, we’re still yet to figure out even the childlike questions, and there is something very exciting about this. And I’m hoping that the interdisciplinary approach is going to help us out in actually being able to figure this out.
ERNIE: It seems like the neural networking in the human brain is similar to what’s going on with artificial intelligence. Is that accurate?
LEENOY: The field of AI and artificial intelligence has made incredible strides in the past decade, the past two decades, I would say. Principles that were there since the eighties in the last ten years with a lot of computational power have just [made] incredible strides. There are a few things to keep in mind that are still different between the research of AI versus the research of the brain. So these things can definitely benefit each other, and have been benefiting each other a lot. In AI, the kind of basic units in the networks that we’re talking about, the very structure of the artificial neurons look very different than what is happening in the brain. There may be a few basic things that are very different. One is that the kind of connection in between the different layers, between the neurons, is one-directional a lot of the time, while in the brain we have a lot of what we call recurrent space, a lot of feedback that is happening. So it’s a different kind of processing. Another thing that is different is that we as humans, with our neural networks that are not artificial, are able to learn things from looking at them once and being able to extract patterns. The thing that remains a challenge for artificial networks, and that we know that for the brain is a very basic property, that we’ve yet to figure out how an artificial network can really do this well, to learn many things and to learn them fast. And maybe the last thing is that the artificial networks are something that you build and they come in with no evolutionary constraints that have designed them. You can build a lot of flexibility, a lot of degrees of freedom into them. The brain is under evolutionary constraints. There are rules of what you can and cannot do. And yet, it functions magnificently. And so, how do you create a system that can actually by design that is under evolutionary constraints in development get to be what the brain looks like? It is under compactness constraint, it is under the type of learning, the capacity constraint, it has its own kind of biological substrates that need to process, that need to be able to support all this. We are still a little far from being able to just take the research in AI and know that this is what is happening in the brain. That being said, a lot of things that we thought we would never be able to replicate that the brain can do, we can now do in AI, in a different way than what the brain does, but still in a way that is very impressive and is definitely very insightful as to how other systems can maybe implement these kinds of things with other kinds of constraints.
ERNIE: Leenoy, I’m going to hit you now with a question you asked as the title of your presentation at the Okinawa Institute: “Can We Hope for Simplicity When Describing the Brain?” So, can we?
LEENOY: That remains a really good question. I think when I titled my talk this way, I was hoping that the audience was also going to engage with this. And maybe try to be also a little optimistic. Because part of the challenge of biological systems that are that complex, like the brain, is that it really seems like there is a lot of devil in the details. A lot of exceptions, a lot of things that are very hard to find a rule for, a lot of details that remain incredibly important, and maybe we don’t have underlying principles. The worry is the person who tries to simplify then generalize, which are maybe the two big objectives of a theorist, definitely of a person who comes from a field like physics, where we have for very complex systems, the ability to write at least much of what’s happening in the system in simplified equations that actually give us a lot of the behavior of the system. The worry is that we won’t be able to do it, that it’s too complex or there is too much detail that is important that we’re not sure what to strip down and still retain the essence of the system. That is the biggest, biggest challenge of what researchers like me do. I am optimistic, because with the kind of modeling and the kind of research that we have been pursuing, you can definitely see that there are trends that you can write down simple equations and principles of and get a lot out of, and capture the essence of the system. And so specifically if we look at how to systematically simplify a system with inspiration that comes from statistical physics, concepts like normalization group, we can systematically simplify and still retain the essence of the system. So that makes me very optimistic. It doesn’t mean that we’ve managed to do this in every possible function of the brain, and it doesn’t mean that did it in every possible brain region, but we have hints, which mean that we might be able to get one day to underlying principles that when you ask a question, one can communicate within under two minutes of, how does that work? Instead of really, really complicated drawings.
ERNIE: Leenoy, what has the CASI award meant to your career at this point and your ability to pursue your scientific interests?
LEENOY: The CASI award has been incredibly helpful to me. In a very honest way, I just want to say I’m so grateful, because as a postdoc, you are really trying to find your own way and your own ideas in order to, if you want to keep on going with an academic career, which I do, in order to know what your group, what your research group one day is going to pursue. What kind of ideas, where is your strength, where your curiosity lies most, after your Ph.D., after your main stage of training. And as one does this, I found that I got interested, for example, in the octopus system, because I thought, wow, this idea of trying to look at lots of coordinated elements that give something global, and look at the emergence, and use these kind of concepts, is incredibly attractive to me and I would like to dive deeper into this. And for this,, the ability to just look around and say, what is the best system for it, and be able to just go for it, because you know you have the back of a fellowship that will fund you to pursue your ideas. You will even tell them your crazy ideas, and instead of telling you, “Wow, that sounds like it’s never going to work,” they will fund you. I remember my interview, where they were like, “Okay, let’s try to do this.” It’s so refreshing and so, unlike some other interactions for scientific endeavors which sometimes maybe encourage you for more incrementalism, the CASI and the Burroughs Wellcome as a foundation, the entire vibe seemed to be, “Take all the courage you want, we are behind you.” Pursue any system you want, pursue multiple model systems in parallel to get inside the comparison between model organisms, which is what I would like to do. The nervous system is incredibly complex. “We’re with you on this, try to take all the physics you can, all the neuro you can, go wherever you need, go to Japan, we will support you.” It still feels like being part of a family that is so incredibly supportive of brave endeavors and of curiosities that might completely fail, but might succeed, and it gives such a calm to understanding that this family is going to be with you, and is encouraging you all the way through, and it is just a wonderful experience.
ERNIE: Leenoy, you are doing tremendously important and fascinating work, and best of luck for continued success. Thanks for joining us on FOCUS In Sound.
LEENOY: Thank you so much, Ernie, it was really nice to be here.
ERNIE: We hope you’ve enjoyed the program, and will join us again next time. This is Ernie Hood. Thanks for listening!