Watch Episode Here
Listen to Episode Here
Show Notes
This is a 1-hour conversation with Ali Hanson (https://braininitiative.nih.gov/ali-hanson-md-phd) on her path in science, including a ~45 minute presentation of her data in Hydra. This is the first of a series of videos I will do to highlight some exceptional junior scientists' work.
More information will be provided on a blog post on Ali's work at https://thoughtforms.org/
CHAPTERS:
(00:00) Journey to neuroscience
(09:25) Molecules versus bioelectricity
(13:39) Complex systems and Hydra
(26:09) Spontaneous activity and self
(56:39) Closing thoughts and teaser
PRODUCED BY:
SOCIAL LINKS:
Podcast Website: https://thoughtforms-life.aipodcast.ing
YouTube: https://www.youtube.com/channel/UC3pVafx6EZqXVI2V_Efu2uw
Apple Podcasts: https://podcasts.apple.com/us/podcast/thoughtforms-life/id1805908099
Spotify: https://open.spotify.com/show/7JCmtoeH53neYyZeOZ6ym5
Twitter: https://x.com/drmichaellevin
Blog: https://thoughtforms.life
The Levin Lab: https://drmichaellevin.org
Transcript
This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.
[00:00] Michael Levin: Thanks so much for joining me. How are you doing?
[00:02] Alison Hanson: Awesome, and thank you very much for having me. I'm excited. This is going to be fun.
[00:07] Michael Levin: I'm really looking forward to it. I want to shine the light on some of the younger people that are doing some amazing stuff, and I'm really excited about your work.
[00:15] Alison Hanson: Thank you. Yeah, me too.
[00:18] Michael Levin: Introduce yourself where you are now, and we'll go from there.
[00:24] Alison Hanson: I am Ali Hanson, and I'm in Rafael Yuste's lab here at Columbia. I'll just suffice it to say that I've completed my psychiatry residency, but my background was in cell and developmental biology and jumped into the field of neuroscience and using Hydra as a model system to study, to be able to see the whole nervous system at once. From there, what really interested me is the role of spontaneous neural activity, neural activity that's going on when the animal's not behaving. I'm now a postdoc finishing that and in the process of figuring out where I'll go next to start my own group.
[01:14] Michael Levin: Why don't you talk a little bit about your background and your journey as far back as makes sense of how you got here.
[01:22] Alison Hanson: Caitlin Clark went to my high school and she's the newest phenom. I wrote on my thing I was going to play basketball in the WNBA and then I never grew. But ended up playing basketball in college. I studied molecular biology and ended up doing an MD-PhD program. My PhD was in cell developmental biology and the molecular mechanisms of Wnt signaling. It was very, very molecular. I got trained in that belief system that molecules are a king. That explains not only how cells work, but how development works. My thesis was on the molecular mechanisms of Wnt signaling and ubiquitylation. Someone clued me in to the physiology course at the MBL in Woods Hole, which was a whole other way to do this. You can really zoom out and interact with people in many different disciplines. I was with pure mathematicians, physicists, computational scientists and people who had no experience with biology, and I was a biologist coming in. You do two weeks with different scientists on different projects and get exposed to different ways of thinking. In that, I ended up with Gaudenz Danuser as one of my mentors. I did a two-week course on modeling three proteins in the Wnt pathway. You can start learning something counterintuitive, something that you wouldn't know by just saying you're going to draw the cell signaling pathway on a two-dimensional piece of paper and draw a bunch of arrows and that's understanding. That doesn't actually explain anything. That was my first foray. Mark Kirschner and Andrew Murray were on the chalkboard doing all kinds of these things. There's a different way to do this: to find the simplest system possible and try to model as much of it as you can to gain some understanding. I was right at the end of my PhD, so I didn't have time to implement that and went back to do medical training. I took another foray, which I think is important to how I ended up doing this kind of work I'm doing right now, which is art. I left in the middle of my MD-PhD training to go study painting in Santa Fe, New Mexico with an amazing person who is Tony Ryder. That totally changed how I see and the way I think. It opened up my mind to how you approach reality with a fixed idea. I went in there. It was classical realism. You're trying to do portraiture and nude models, drawing and painting. I didn't have any experience with this at all, except for childhood classes. For an example: a shadow is dark, just black. If you have this belief, you'll never paint a shadow as it actually is, because there's no such thing as pure black in nature or reality. There's always some amount of light, some amount of color. Ted South Jacobs was Tony's teacher; it's a similar lineage in art and science, who wrote a whole book on the anatomy of human forms. If you paint a human eyeball that looks like an almond with a circle in it, you'll never, ever paint an eyeball that is true to reality. All of these concepts that I'm laying on top of these things are not real. Can I get rid of those things? That was part of that exercise for two years: trying to unlearn all of the things I had been told about what a human is and the features of it in order to paint what it actually is. Upaya Zen Center was in the backyard of this place. I learned of Zen Brain, which is about conscious complex adaptive systems. Also, the Santa Fe Institute was there, and I took classes. Ultimately, I was debating: do I go paint full time or do I do a postdoc? If I was going to just do a straight postdoc, I was going to do one on the origin of multicellularity. How do you go from one cell to a group of cells? That was what I was interested in. But then I got interested in: I really wanted to help people, and I wanted to do that through psychiatry.
[05:23] Alison Hanson: Is there a way to put all of that stuff together? Well, **** now I'm going to have to switch fields and study neuroscience. At the time, I'm sorry to any neuroscientist out there, but that's not science. That's not molecular. What is that? You're putting a human in an fMRI scanner, you're looking at blood flowing all around, you're not even looking at neural activity. This isn't understanding, this isn't mechanism, because for me mechanism was: what's the amino acid on the protein? That's a mechanism, right? That was the kind of thinking. But going to that Zen Brain Conference actually was one of those things where you realize some of these questions you just can't reduce to an amino acid in a protein; that model, that way of thinking doesn't compute here. Deciding, okay, I am going to pursue a residency in psychiatry and try to figure out what is neuroscience. I had to do all these interviews. I was essentially using the residency interview as a postdoc interview to say, who's doing neuroscience, where, what are they up to, and at what level of scale are people studying all of this? I was trying to see the lay of the land. I remember at one point I somehow got a hold of Sydney Brenner when he was in the hospital in Singapore and asked, "What level of scale do you think we should be studying the nervous system?" I can't remember his answer, but it was single cells; you could go down and you can go up. So that raised the question: where and how do you tackle this problem? I ended up reading Rafa's article, and George Church in Scientific American around that time about using complex adaptive systems theory to study the brain. You needed to see the whole thing. Yes. On that interview I talked to Rafa and he had a hydra on the screen. It was one of those moments: "What? No way. This is amazing. This is exactly what I've been looking for." All of my dreams coming true on this screen. It's a simple system. You can see the whole thing. It has a rich history in cell and developmental biology and evolution. It was all together in one thing. All right, that's what I'm going to have to do. I early on saw that there was this spontaneous neural activity in that animal. What is this? What is that? That got me very curious about where else that was found, which led to the human brain and the self. So it all was coming together. Then I was seeing patients as a resident and had a little bit of research time. I was really reading, thinking, and looking. I'll disclose that I'm a Zen student, and so I was also a scientist of my own mind. I've been exploring it from that angle a lot: first-person point of view and third-person with patience, and the "third-person objective" view of the system too. There's all of that at play all the time. That's the longest version I can muster.
[09:25] Michael Levin: The art angle is also quite interesting. I can't draw or anything like that, but I do a bunch of photography, and I agree with you that watching the light and trying to capture what you actually see or what you think you see is an interesting component of it. Since you have a background in molecular genetics and now neuroscience, how do you think about the kinds of things those systems are good at? How are molecular dynamics and the kinds of things you get out of those different from what you can get out of, for example, electrophysiology, in terms of the computations, the uses? What's different about those systems?
[10:14] Alison Hanson: You mean at the cellular level versus the system level or the neural?
[10:20] Michael Levin: If you were going to build things and your options were gene regulatory networks and diffusion versus electrical networks and synaptic things, what are the different dynamics that you can expect to get out of those systems? How are they different functionally?
[10:38] Alison Hanson: Yes, great question. Ultimately, one of them is slower. So molecules diffusing limits you in space and time. Whereas electricity allows speed and distance. So it just speeds up processes. I don't think there's anything particularly special about neurons. If we're talking about memory or computation, any of those can be implemented in any material. There's nothing special about neurons. This can happen in those molecules if you put them together in certain ways and you maintain a pattern that's relevant to the organism. That can be implemented in the molecules, but it just limits the space and time of that. Molecules are on a shorter length scale and take longer to move and communicate, whereas you can send electricity much faster over a longer distance. That's how I think about it.
[12:01] Michael Levin: I agree. That's definitely an important feature. People sometimes ask me what's magical about bioelectricity? What's different about electrical networks that's harder or maybe even not doable in typical molecular kinds of networks?
[12:19] Alison Hanson: I'm just thinking of this, intuitively at some point what do we as human beings use to communicate? We use electricity because it's the fastest thing. We use the smallest thing, the electron, that can go the fastest. That's how we connect globally via the internet and all of this. So of course nature would be using the same thing. I remember looking this up in one of your papers: when you get a wound, the first thing that happens is electrical communication and then the molecules. I was at some point thinking, one of the things I did every day, all day was running gel electrophoresis in a lab.
[13:01] Michael Levin: Yeah.
[13:02] Alison Hanson: You're generating an electric field to put your proteins where you want them. Why wouldn't nature be doing that, organizing itself, using electricity? You're using electricity because it's the fastest way to communicate from head to tail; something's up on this end, and then you bring the molecules over there. For me, it's about speed and distance. That's what I would say.
[13:39] Michael Levin: What are you working on now?
[13:45] Alison Hanson: I have a few slides I could bring up here. I can't see that. We don't really have to do a formal thing. Go for it.
[14:08] Michael Levin: Do as much as you want.
[14:09] Alison Hanson: This is the gist of it. As I've alluded to, it's what is going on with this spontaneous neural activity. This would be an outline of why somebody like me, mostly interested in studying the human brain, is interested in studying something as simple as this Hydra. Then that brings us into the history of spontaneous neural activity, what it might be doing, the tools needed to study that in Hydra, and some of these future directions. I'll go as quickly as possible. This is one of my favorite things to start with: imagine that you're a scientist and you're the first person to ever see an image that looks like this. Which is exactly what happened to Golgi when he put the Golgi stain on a nervous system for the first time. You're trying to make sense of "What is this?" He thought that this was just one giant reticular meshwork. It's all one thing. Other people at the time looked at the same image and saw something else. They zoomed in on these shapes that look like pyramidal things. Sherrington and Ramon y Cajal said, "What about these pyramidal things? What if those are how this works? What if those are the functional units of this system?"
[15:21] Michael Levin: The Golgi stain, was that random? Was he just trying random chemicals or did he have some idea that it was going to stain something interesting?
[15:30] Alison Hanson: I have no idea. I would have to look, that's a great question, why, how that ended up happening. But before that, there was no ability to see what it looked like. So there were all kinds of speculations on behavior and all of that, but no real way to see or visualize the nervous system. And then once you do, then you've got to make sense of it. This is just a fundamental thing about science. First of all, the method that you use to view the thing will determine what you see. This is exactly what happens. And Neil Thees will talk about this. And so this was happening at the same time as cell theory with Schlein and Schwann, because at that time, they saw cell boundaries, they saw lipid membranes. The functional unit of that system of non-neural tissues is cells. But Neil will make the point: if you had just had a nuclear stain, what you would have seen is just a bunch of dots floating in liquid. You wouldn't have known that there were these things called membranes. If you had an Eastern conception of what's the functional unit of tissue, it would be fluid with a bunch of dots floating in it. So here you have this at the beginning: this whole picture of a nervous system, and then you're arbitrarily deciding which parts are separate from the rest. And so the decision at that point was we're going to go with cell theory and the neuron doctrine that these little pyramidal shaped things are the units. And at the same time, Sherrington was focused on reflex arcs. So that solidified this idea that not only the structural unit might be a single neuron, but the functional one, because you hit your knee, you get one sensory neuron that goes in, synapses on another motor neuron in your spinal cord, and that's your reflex. So that gave them two ideas. A, that single neurons are the functional units, and then B, the nervous system is doing nothing unless you stimulate it from the outside. Those are just these two major paradigms taken from mostly spinal cord and peripheral nervous system and applied to the cortex. And that must be how it works. But I would argue that those two major ideas have left us in a situation where we still don't really know what the true functional unit, at least of cortex or non-spinal cord peripheral nervous systems, really is. Is it multiple neurons? Is it all one thing? Is it some amount of an ensemble? How many neurons are in an ensemble? How do those things interact to generate complex cognitive states and how do those go awry in neuropsychiatric disorders?
[18:09] Michael Levin: Do you think there's a single objective answer to that question? Is there an optimal subunit, or do you think it's more observer-dependent?
[18:23] Alison Hanson: What I'm most interested in is general principles. There's not one answer of what a functional unit is in a particular system, but the general way that works will probably be general. I don't need to introduce a complex adaptive system, but maybe others would be interested. Basically, you have a bunch of individual simple systems or subunits at this level of scale. If the brain is actually this complex adaptive system, maybe we're studying at the wrong level of scale. We can't see the functional units because we're studying one thing at a time. These can be electrons in a magnet, water molecules, ants in an ant colony, birds in a flock, or neurons in the nervous system. The key is you put enough of them together and they follow simple rules. If we think of electrons: an electron spins up and its neighbors spin down, then it needs to become spin down. Put enough of them together, and that won't happen with one electron or ten electrons; it's a property of many electrons. You get an emergent phenomenon at a higher scale, such as magnetism. That higher level can feed back down and coordinate and constrain the lower level. You're never going to see magnetism looking at a single electron; you only see it at a higher level. That higher level follows totally different rules than the lower level. In neuroscience, think of this as a TV screen. For most of history we've been doing single-pixel recordings, single-unit electrophysiology, which is critical—we needed to understand what a neuron is by studying one at a time, but you can't see the whole picture. Alternatively, you can zoom out with MRI, EEG, MEG—you get a global view but very low spatial and temporal resolution, so you can tell something's happening but it's hard to see what. The ultimate goal is to look at that TV screen with single-pixel resolution to see what picture emerges at the higher level: what can we see by looking at all the individual neurons? I'm interested in the human brain because of the human mind. But it's huge: 86 billion neurons with trillions of connections. It won't fit under the microscope. You can simplify the problem by using a mouse brain, but how are you going to record all those neurons? Even optically, we're currently up to, generously, 20,000 neurons out of 71 million, so it's not much of the whole TV screen. You could simplify further with zebrafish, Drosophila, or C. elegans, but these systems are still complex. Depending on the developmental stage, a zebrafish can have up to 10 million neurons and many cell types, all shoved together in complex brains and ganglia that are hard to disentangle. Sydney Brenner sat down with my mentor Rafa. Sydney pioneered work on C. elegans, and he said that's still too complicated. What if we try something even simpler? Let's do Hydra, which has 200 to 2,000 neurons, depending on the size of the animal.
[22:16] Alison Hanson: Depending on how you cut off the single-cell sequencing data, you get 11 to 12, maybe 13 different neural cell types. It doesn't have a true brain or ganglia, as you'll see momentarily. Not only is it simple, but it's an interesting position on the evolutionary tree, at least as far as I'm concerned. It's right after, in quotes, "the non-neural lineage" and then right before the bilaterians, which is the branch of life that we're used to studying in neuroscience, where you have a central midline structure with a left and a right side. Cnidarians are totally radially symmetric; they don't have any of those things. It's another instance where you can ask, what are these general principles? This nervous system looks nothing like that one, and yet it's able to generate complex behavior. Does it matter what the architecture of this nervous system is? Everyone's very interested and obsessed with getting the circuit diagrams, but does that matter? Can you get essentially the same functionality out of completely different architectures? I think that's what we'll find if we look. This is just what an actual hydra looks like with its mouth, tentacles, foot, and this hydra bud. This is important because hydra can reproduce asexually. Hydra essentially clones itself. If you keep feeding these animals instead of them getting infinitely big or long, they start budding and making a genetically identical copy. I think this is critical for you. They can also reproduce sexually. The same animal's stem cells can generate testes or ovaries or an egg and reproduce sexually. It has essentially, not exactly the same, but about 20,000 genes, very similar to our genome. The genes here are essentially the same as you or I, and yet it's building something completely different. It's small; it can be anywhere from 500 micrometers to 1 1/2 centimeters. It's transparent, it's robust, you can do all kinds of things to it. It completely regenerates its nervous system every 20 days. It's constantly making this nervous system and maintaining itself somehow. A lot of molecular tools are being developed, and it has well-defined behaviors. At rest it's just laying there doing nothing. It can do a very simple longitudinal contraction where it goes from long to short. The opposite of that is elongation, so it goes from short to long. It has more complex behaviors; these are arbitrary words I'm defining, I'm just saying these are more complex. One is eating. In lab, you feed Hydra Artemianopolya, just brine shrimp. Reduced glutathione will also elicit this feeding response. They'll use their tentacles to capture shrimp, rip them into their mouth, and shove a bunch inside their body. The most complex, I would say, is somersaulting. It will reach out, stick its head and tentacles down, contract, release its foot, flip over, and put its foot back down. In this way, it can, in quotes, walk. It can do all of this with a very, very simple nervous system.
[26:09] Michael Levin: On the behavior side, has anybody checked various learning, training? Can you teach these things anything?
[26:20] Alison Hanson: This is the new frontier. But it's also the new old frontier. As in all science, there were some old studies; I think some of the first were 1900 and then 1950 of habituation in Hydra. If you pinch them or even stick an electrode in them, they'll contract. If you keep doing that with the proper interstimulus interval, they'll stop; they'll habituate. There hasn't been, in the last 50 years, any of that repeated. I don't think it's because people have tried and failed. I think most people haven't revisited that. That's on the docket. I think it was Nematostella showing associative learning there, so other Cnidarians now have been shown to be doing more complex things than meet the eye. You get this hypostome. This is just a cartoon of the same thing: this nerve net, the tentacles. This is where they would have the buds coming off. This is the basal disc. This is the tissue. It's just two layers. It's not a true three-layer thing. There's no true mesoderm. There's just the endoderm on the inside and the ectoderm on the outside. All of these are also myoepithelial cells. They're epithelial cells with muscle processes at the bottom. So they're sort of a dual cell type. Not all of those are connected by gap junctions. You have a nerve net in the endoderm with sensory neurons poking into the inside of the gut, and then ganglion neurons in the ectoderm with sensory neurons poking out to the outside environment. It's thought that these do not communicate, but that might not be accurate. That's the current idea. There are three different cell types. You have i-cells, which are the stem cells that make the germ cells, the gland cells, nematocytes — a super cool cell type specific to Cnidaria that shoot out darts with toxins — and neurons. These all self-renew. The i-cells make these four, and the ectoderm and the endoderm self-renew as well. That's the gist of that. This is the current view of the connectome of Hydra. This is from Charlie David at LMU using an anti-adherin antibody that stains all of the neurons in the animal. It looks like this could just be a diffuse nerve net. In the ectoderm it's a north-south highway with more bipolar neurons. In the endoderm it's a more multipolar thing going radially around the animal. This is from Rob Steele at Irvine, where GFP is expressed in i-cells. There appears to be a little condensation of neurons in the foot and in the head. Keep in mind that what we're seeing depends on the tool that we're using to look at it. Which is true? Yes — they're both true. When you look at that, it's unclear how that will function; there's no obvious circuitry that would tell you, just from that structure, how it's going to function.
[30:40] Alison Hanson: Is it functioning as one thing, as many things, who knows? Until Christoph put in GCaMP6. So this is a molecule that will fluoresce every time the neuron fires. He made a transgenic animal expressing this in the eye cells. So all of the neurons had GCaMP6 expressed. He then put the Hydra in between two coverslips. So this is just 100 micron spacer. It's swimming around in Hydra media so it can still behave in two dimensions. And then just put this on a normal wide field microscope and came up with a movie that looks like this. Can you see all the pixels in the movie at the same time? And when you do, what do you see? Are they all firing completely independently? Total chaos? No. And are they all firing as one thing? No, it's something else. In order to figure out what the something else was, Christoph had to painstakingly circle every one of those neurons and track it over time to analyze that data and figure out what is happening here and came up with this general model, saying that out of that seemingly simple nerve net, you have these three independent units, ensembles or circuits, groups of neurons that are co-firing together. And the one that's most consistently associated with muscle activity is CB, where every time it fires, you get a longitudinal contraction, so it gets shorter. That's very clear that that's motor. These other two are less clear, rhythmic potential 1 and rhythmic potential 2. Rhythmic potential 2 in that early paper was most associated with radial contraction, so that's in the endoderm of the animal. When that fires, it gets skinnier and can do some egestion, so it will spit its food out or water from its mouth. And then rhythmic potential 1 is in the ectoderm, and when it fires in that paper was correlated with elongation toward light. But what was most interesting to me was that this was also happening when the animals are just sitting there doing nothing. And that is not new. We're noticing a theme. Reading literature is helpful. So this was found in the 60s by Pisano and McCullough, where they were just sticking an electrode into Hydra, and they found the same thing, that there was this rhythmic firing of some kind of neural network in the animal they didn't have access to. They just knew that this was happening with their electrode when the animal was doing nothing. And they called this cryptic activity. They said, what the hell is this doing? Because at the time, the idea is that the nervous system is there to generate behavior. Otherwise, it's off. Why at the base of the tree of life, the simplest nervous system, are we seeing all kinds of activity going on if it's not generating behavior? And so he hypothesized in one of his early papers, is this thing coordinating the nervous system, the quarterback of the thing, the conductor? And I've been told that he went off and started selling bikes and so never finished that work. And so we don't know. But it was discovered there with an electrode, and it was rediscovered with calcium. So it's method independent that you find the same thing. And that is what got me interested and brings us to this very brief history of what we know about spontaneous neural activity, which is very little. Because as I was saying, Sherrington at that time, the big mantra was reflexive brain, it's doing nothing otherwise. But his own student was finding something very different than that. He cut the head off of a cat and de-afferented the cat, no sensory input whatsoever, zero input to a nervous system. He had all kinds of activity going on to the point where the cat could still walk. So this was clearly internally generated neural activity, had nothing to do with outside input.
[35:00] Alison Hanson: That was just ignored. That was whatever, that guy's nuts. But whatever he's up to, let him do that. Around the same time, Hans Berger was inventing the EEG, where you have humans sitting in a chair, all kinds of electrical activity going on there as well. It wasn't just Pisano McCullough discovering this spontaneous activity in Hydra. Anybody looking at any cnidaria was discovering this kind of thing going on. No one's talking to each other, and so no one knows. It wasn't until the 2000s when Raichle and others at Wash U put humans into fMRI scanners. At the time, what was interesting was the task-evoked activity. Anything else that you couldn't explain was seen as noise, white noise on the TV, get rid of it. What we're interested in is what happens to the brain when you go to do the task. At some point he noticed that all of those areas in the brain that are active at rest seem to be specific, and they're specifically inhibited when the person is going to do the task. So he called this the default mode network, which became a very hot topic and has reignited interest in the study of spontaneous neural activity, which we now know is one of the most highly conserved aspects of all brains. This is from Buzsáki's review, and it spans a wide range of frequencies, up to 600 hertz, very ultra-fast, all the way down to ultra-slow of .01 to .1 hertz. This is found in all mammalian brains, regardless of size or structure. The current idea is that that's just noise, reverb. If you have a complex enough electrical system, you're going to get oscillations. It's not functional. It's not relevant. It's an epiphenomenon of this system. It's not doing anything. One of the reasons that remains is that you can't really image it or manipulate it in a specific way to determine: if you manipulate this oscillation, does that do anything? But again, it's taking 20% of total body energy and is one of the most highly conserved aspects of all brains that we've looked at. So the question is why or what is it doing? One of the most highly studied is the default mode network in humans, and this is related to the self in theory. It's most active at rest. It's ongoing all the time, and its function continues to be debated, but it's thought to be involved when you're internally focused. You're either remembering your past, thinking about yourself right now, or contemplating your future. The same sort of midline structures that I'll show you in a second appear to be activated when you're doing self-related processing. If I'm shown a picture of a stranger versus myself, the default mode network lights up when I'm looking at something related to the self. That's in any modality, whether pictures or hearing my voice. It's also linked to a number of neuropsychiatric disorders. You can't throw a dart at PubMed right now and not see the default mode network linked to something. I'm simplifying that to say that the general idea seems to be that in states like depression and anxiety, when you're too internally focused, the default mode network is too active. This is associated with rumination and worry, and vice versa. When, in states like psychosis, psychedelics, or meditation, you disrupt the default mode network, this leads to "ego dissolution." Some of the best data in human studies are when you put humans in and they can self-report on psilocybin, LSD, or DMT that whatever they think they are is dissolved.
[39:20] Alison Hanson: And then you can see what happened in the MRI scanner and that correlates with disruption of the default mode network. So the question is how would that have anything to do with the cell? What would be the mechanism by which the development network could implement something like a self? And there's the structural argument and the functional argument. The structural argument being that it's a central midline structure, so medial prefrontal cortex, the posterior cingulate, the medial temporal lobes, that these are major rich hubs in the brain that are connecting regions that would otherwise be segregated. So you can imagine if you had a brain where the visual cortex was not talking to the auditory cortex, was not talking to the olfactory cortex, and it was not talking to the internal brain stem system, that would be a disintegrated system. So is this central midline superhighway with these thickly myelinated axons connecting all of those brain areas, getting input from all of them and then sending information back? So that's the structural thing. And the functional piece is that it's one of the lowest frequency oscillators in the brain, of 0.01 to 0.1 hertz. This is now finally going to answer your question about the observers in the brain. Buzaki was saying this is what we've been doing as neuroscientists: looking at things from a "third party objective" view — you're looking at a brain with a bunch of squiggly activity, and then you're looking at its output, this behavior, and you're correlating those two things. If I see this squiggly activity, then I can predict that the animal, whatever organism, is going to do this behavior. That's helpful for us. That's very useful clinically. But that's not understanding. That's not what the brain is doing. The brain has to observe itself. The neurons have to observe their own information and then decide what to do. If you think of neuron A, B, and C here, neuron A has to observe the upstream activity from its inputs, neurons 1, 5, and 9. In order to do that, it has to be integrating that information over a certain period of time. Let's just say that's 100 milliseconds. So it reads that information and decides whether it will fire or not based on that upstream input. Same thing with B and C. You can think of those as the highest frequency oscillators in the system, the letters in the neural code. Then the downstream observer of A, B, and C is Q, let's say. It has to read out the input from A, B, and C, and therefore has to be integrating that input over a longer period of time to give time for A, B, and C to give it their input. So that would integrate over 200 milliseconds. It then reads out A, B, and C — reading out, let's say, "cab," a neural word. Looking over here, that would be like this frequency. So you would go from letters to neural words, and then this is going to read out over a longer time scale to get inputs from those to form sentences, and then paragraphs, and all the way down. Is it possible that you have this thing that's the lowest frequency oscillator, getting all of the input on the longest time scale in the entire nervous system to read all of that information out? This is a picture of what that might look like, laced on top of the brain. I like to think of this in terms of the visual cortex as the easiest for me to conceptualize. If we think of this little red thing here as the highest frequency oscillator, the letter of the neural code, if that's coding in the visual cortex at least for the simplest thing like edges in the visual field, then the next layer, layer 2, is getting input from that layer. And that has to be reading that over a longer period of time. So it has enough time to get input from, let's say, two edges for it to read out a higher level abstract concept or shape that looks like this. Then the next lowest layer, a lower frequency, reads out the input from layer 2 over a longer time scale so that it has time to read out this shape and this shape to form a higher level abstract concept of this shape, all the way down to the lowest frequency oscillator, let's say in this case the development network that's reading out all of the visual information, the auditory information, the somatosensory information, the olfactory, and internal state of the organism. In that way, it gets the highest level, most abstract view of both the external environment, the world, and the internal environment of the organism or the self. So can it be the reader, the integrator of the information, but then can it also be the coordinator top down?
[43:40] Alison Hanson: And in this way, we went through that diagram, but in this case, if we're thinking, is it possible that these low frequency oscillators can essentially organize higher frequency oscillators through various mechanisms of cross-frequency coupling: phase-phase coupling, phase-amplitude coupling. So is it possible this lowest frequency oscillator is essentially critical for setting up this hierarchy of oscillators and organizing them in space and time in the brain, such that if you get rid of it, that organization, that hierarchy might disappear. There's some evidence that when you give humans psychedelics, the brain goes into a more disordered or chaotic state. It's unclear if you get rid of this default mode network, do you get something like this where you have all these other oscillators that are all this? Or do the oscillators themselves go completely out of order and you just get complete chaos? It seems to me that it's not the latter, complete chaos, because if people report psychedelic experiences, they're still having some kind of experience. It's just less order: you might be having those ensembles playing in a different order than you normally would. All this is complete speculation. Because it's very difficult to image or manipulate the human brain, luckily this thing appears to be highly conserved. You have humans, monkeys, rats. A hierarchy of brain-wide oscillations has been found in a bunch of insects also. This gets into some really out-there stuff where we're talking about finding these things even in systems that don't have a nervous system. People have been looking at this in plants for a long time: when you put a microelectrode array in their root you see electrical spikes in plants that Belluska and others have been saying for a long time might be involved in information integration and communication plant-wide. Adamatsky put these electrodes into oyster mushrooms and found the same thing. These global organism-wide electrical spikes. In this animal he's doing interesting work about whether this is a fungal language and that kind of thing. The same thing with amoeba: finding these low-frequency electrical oscillations or spikes; it's unclear what they're made of in that case. In bacteria, in single cells, when Adam Cohen first made the voltage indicators and put those into single cells, he showed that even bacteria have these low-frequency electrical oscillations. If you played that movie and I didn't tell you they are bacteria, you might think those are a dish of neurons flashing. Once you put them together into a biofilm, those individual bacteria generate global biofilm-wide potassium waves that help them coordinate themselves as one giant unit. Not only can they do that within themselves, but they can do this long range. Biofilm one wants to eat and biofilm 2 wants to eat. If there's enough nutrients, they eat at the same time. But if there's not, they can coordinate nutrient sharing and then they oscillate in anti-phase. The general idea being that these low-frequency electrical oscillations are ways to coordinate otherwise autonomous units into coherent wholes, no matter where you are, with or without a nervous system. This gets a little bit—you have to take a lot of drugs to watch this—but this is what I just showed you. We're all very used to this: neural tissue being electrical, ions flowing through, and that's how you generate action potentials. This is where people get confused or get hung up. When I learned from your stuff I was like no way. Of course this is how this works. It's just slower. People have a really hard time thinking about different scales. Even in the biological sphere, we can't deal with different spatial scales. But this is just slower. So you can have ions flowing through gap junctions in non-neural tissue. I'd be very curious to know what you think or what you're finding in terms of any of these non-neural bioelectric networks, if you're finding anything, if there are any higher-frequency oscillations or lower-frequency oscillations, and if any of that is organizing even the non-neural tissue. We can get there in a second, but the idea is very similar: you have individual cells whose voltage, if there's some fluctuation, would be oscillating at some higher frequency. And those can be put into groups of a lower frequency to form a subunit, an arm.
[48:00] Alison Hanson: And then you can have a bunch of those arms being read out. And on up to build chunks of bodies based on a similar principle of these voltage circuits being read out by lower and lower frequency oscillators. And then in single cells, is it possible that instead of proteins being just the structural units of cells, which is what I've been taught my whole life, these things are actually themselves conducting electricity? Stuart Lindsay at Arizona has shown that you get nanosiemens currents flowing through all proteins. He just picked six at random — BSA, collagen, many things that you wouldn't expect to be conducting electricity. So is it the case that cells are actually true little living computer chips, where the proteins are forming electrical circuits on and off? And that metabolism itself, we've been taught, is generating ATP and energy, but it's also generating a bunch of hydrogen ions. Is it also giving us a pool of hydrogen ions to be flowing through the computer chip? That's what I was thinking. You can build all these things in different ways at different scales. What might they be doing? Could they be maintaining systems at criticality? In the human brain, for example, with the default mode network, it's thought to be at criticality, where it helps the system form patterns at a useful time scale. But you can also have a system that's totally on or totally off: brain death, or a tonic-clonic seizure, or total chaos. This appears to be what happens — you go towards chaos if you get rid of this organizer in at least the human brain. I don't know what happens in other systems if you get rid of that. Again, integrating all the bottom-up electrical information in the system and then coordinating it top-down. Not only would this thing maybe maintain the overall electrical organization or structure of the system, but the spikes of that oscillator are not totally constant like a mechanical clock. They're variable. I think that variability, in terms of external and internal input, could allow it to coordinate the entire system top-down and provide information about the overall state by how frequently it's firing. That's speculative. Low frequency: within a range of frequencies, local units get some kind of global piece of information indicating that this frequency means everything is stable and to keep doing their local job. A higher frequency signal means something else to local function, and maybe you need to speed up their processes. So again, is this thing conserved in Hydra? That's what I want to start testing there: does the Hydra cell phone serve as the organizer, the organism's unifier, an ultimate integrator of both its internal and external input to generate coherent adaptive behavior, i.e. a unified self. I've had to do all this manual tracking by hand. What I've been spending my time doing is trying to get this automatic, working with a group in Rafa's lab to make dual-color animals so that you can use that nucleus that never disappears while the animal is moving. In all other neural systems, you're putting a microscope on the head and it's not moving much. But this entire system is moving. You have to figure out a way to track it and extract all of those calcium signals from each neuron and end up with individual spikes from each of those neurons. What's another interesting, potentially unique thing about Hydra is that from Christoph's initial work when he put an electrode in at least one of the neurons, it appears that one calcium spike equates to about one action potential. Using optical information from this animal, you can capture every spike from every neuron and get a clear picture, because in other systems a calcium signal can equate to many action potentials in the mouse, for example. So it's hard to have that as your only readout.
[52:19] Alison Hanson: I'll just show you really quickly this beautiful movie. Those are all of the tdTomato and the GCaMP7 neurons there. You can track all of that and do the signal analysis. You can also do automatic behavior tracking using DeepLabCut so that you can do automated behavior and automated neural signal analysis and start making sense of this stuff. With that, I'll just end with a few teasers. This is just to show you, prove that Hydra lies there and rests and it still has this neural activity, which will flash in a second. You can start asking: what is the relationship between that network and its other two networks? Is that network actually predictive of global neural activity in this animal? What's unique about Hydra is this. These two other things that you can do, where this is the parent animal, this is the foot and the head, and this is a bud. This is an early bud coming off. This is, again, a genetically identical animal. This is one slab of tissue. But at some point, this genetically identical thing becomes two individuals. How does that happen? At this stage, their global neural activity is essentially the same, and their behavior is essentially the same. But at some point, that tissue becomes three different individuals. This is a bud here, a bud here, and the parent here, and they're flashing asynchronously, like different Christmas trees, and they start to develop completely independent behavior. You can go to town on those questions. You can also do all kinds of other manipulations: cut to make animals with two heads, and you would expect it's sort of like a split-brain patient. You're sharing one body, you have two heads, one foot. I would have predicted they'd act independently, but it also looks like Hydra's brain, like the plant, might be in its foot. These actually start acting the same, but you can actually break a Hydra. If you make a split foot, then it starts acting independently. This was contracting independently of this foot. Its nervous system is broken; it is no longer one coherent thing. Even the tentacles—the body here is flashing independently of this foot, independently of that one. It's having a hard time coordinating itself as one coherent thing. To conclude, low-frequency electrical oscillations are found throughout the living world. We don't really know what they're doing. The hypothesis is that they are overall electrical organism organizers that can do three jobs: maintaining systems near criticality, so that if you get rid of them you go toward chaos and disorder; integrating lower-level information; and, from that top-level view, coordinating everything from top down. I'm going to start trying to test these things and how these tools have been built. So there's the whirlwind tour. I'd like to thank all those people who contributed to all of this, and thank you for sitting here.
[56:39] Michael Levin: That was awesome. Thank you so much. Yes, super interesting. We're going to have to schedule a part 2 because there's a bunch of things I want to ask you and also talk about the cell folk concept, both in cells and embryos and then outside of electrophysiology in general, because we're starting to see something similar in other systems that are not natively electrophysiological at all. What you've got here is a very profound, generic concept that could apply to a lot of things.
[57:14] Alison Hanson: Yeah. Awesome. Love it.
[57:16] Michael Levin: Thank you so much. We'll talk soon.
[57:22] Alison Hanson: All right. See you.
[57:23] Michael Levin: OK. Thanks. Bye. Bye.
