Biology

Michael Levin: The Vannevar Bush Professor at Tufts University and Director of the Allen Discovery Center & Tufts Center for Regenerative and Developmental Biology

Below is a transcript of the Q&A session

What is your opinion on the idea that perhaps there is a higher level system that programs all biochemical processes, and it is primary to all biological processes; meanwhile, current science makes the material/cellular level primary?

Yeah, it is a very fundamental question, how to identify what’s primary if you have a particular system that has multiple levels of description. Let’s say you have a computer and it’s running an algorithm, and then one person will say, “I’m pretty sure the algorithm is causal because the way I program it is I change the algorithm and it does whatever the algorithm says”, and then somebody else will say, “You’re crazy, there is no such thing as an algorithm, all it is it’s electrons behaving according to physics”. If I were to zoom in to the computer, I could see exactly what’s going on and all I’m interested in is electrons moving around. So, people have argued about this kind of thing on a philosophical basis for the longest time. Amazingly, there is now some actual work on causal information theory where people like Erik Hoel who is part of my Allen Discovery Center have developed mathematical tools to actually try to estimate how much of the causal work in any given system is done at every level. And in the examples that we’ve looked at is quite clear that the information that is being processed by the electrical circuits is causally much more potent than things you might do at other levels; for the example, the gene expression level or the biochemical level. That’s not to say it is like that in every system, in our system, it is. I think the important thing to know about these kinds of questions is that it’s no longer just a matter of pure philosophy, we now have actual math, meaning actual actionable algorithms that you can use to figure this out as long as you can formalize some of the levels of description of your given system. Definitively, it’s not all in the biochemistry, that’s for sure.

How long do these modified two-headed planarian survive? Is it the same as the non-modified ones?

Some of the two-headed ones have trouble eating because the two heads sort of fight and they have different ideas about when and where they should eat. But among the ones that eat properly, or if you were to feed them manually, they live for ever. There is no reason why they don’t have a normal life span.

In addition to the electrical processes, are there other processes that play a role in this, such as hormones?

Well, for sure, there are lots of different processes. You can’t build anything out of just electrical signalling, the electrical signalling has to control all of the enzymes, the gene expression, the proteins, the hormones, everything else that’s downstream that’s required to actually implement the decision to build a head. Lots of things are required of the biochemical gradients, all of the biomechanics, all kinds of things. The question is, this goes back to the first question, which is what’s the primary driver? And the way I define the primary driver is the layer of control where the least amount of intervention gets you the most complexity as the predictable outcome. So, the most efficient leverage point for controlling and understanding what the system is doing. So, yes, there is tons of other stuff involved but it seems from everything that we’ve done, in the cases that we’ve looked at the electrical circuits that mediate the decision making are what’s driving what happens to the biochemistry, and the hormones and everything else. It’s again very similar to this multiscale problem in neuroscience where somebody might study psychology or cognitive science and think about reasons or psychological drives or cognitive dynamics for why an animal does things and somebody else is focused on the microlevel of hormones and ion channel activity in single cells and says, “That’s really where everything is going on, the other stuff is kind of an epiphenomenon”. And, you know, that’s been a debate for a really long time, but we now have an actual quantification to answer that. So, yeah, lots of other things involved but the bioelectrics are where the decisions are made.

If we can express bioelectrical gradients in specific locations, then we could basically generate anything we want, such as an additional limb, etc. Are we limited by DNA as a hardware (of what it is not there in the organism)? Regarding humans, we can’t have wings, can we? Just because it’s not specified in the hardware, or can we have them… with software?

Yeah, it’s a good question. You’re definitely limited for specific materials. For example, if you don’t have a certain enzyme in your genome, no amount of bioelectric signalling is going to get you that enzyme. If you need to turn chemical A into chemical B, and you don’t have the enzyme, bioelectricity signalling is not going to help you do that. But things like wings, we haven’t done wings, but it should be no problem because as long as the wings are made of the same kind of stuff, your building blocks, as long as your building blocks are present, no problem. As I’ve shown you, you can make worms of other species out of standard worm genome. People do talk about these developmental constraints. It’s a kind of an open question, it’s controversial. I don’t believe there really are significant developmental constraints in that sense. The constraints are all in the fact that we don’t know what we are doing. If we understood properly how the encoding works, if we understood exactly how cellular collectives encode their goal states, I’ll bet you any amount of money, you could make wings, whatever you wanted, as long as the building blocks are there. The limiting factor is that we don’t understand the encoding very well.

So does it also mean that we could regenerate. In another talk, you mention that small kids can regenerate part of their fingers. But why does it stop then?

Well, nobody knows why it stops but I’ll tell you a plausible story for maybe why it stops. Think about our ancestors, some sort of an early mammal, something mouse-like that is running around the forest and somebody bites its leg off. Now what’s the value of regeneration there? You’ve got a high blood pressure, you’re going to bleed out if you don’t seal it and scar. You’re going to bleed out and you are going to immediately try to walk on it which means you’re going to grind that delicate blastema tissue if it tries to go back, you’re going to grind it into the forest floor. I think that what mammals have done and also, by the way, because you’re not aquatic, you don’t have the nice water on the outside through which you can run all the ion currents that you need to set up your voltage states in the wound epithelium. I think because of the change to a terrestrial lifestyle, mammals went to scarring instead of regeneration but that is a very contingent kind of thing and we’re hoping to overcome that so we, literally, in our biomedical work, we use wearable bioreactors that establish an almost amniotic like environment and then drugs to open and close ion channels to try to convince the cells that a) it’s okay, you are in a protected environment, go ahead and try to rebuild, and b) here’s the signal we are telling you what to build, we’re going to build a leg, a finger or whatever it is. I firmly believe that it’s not a fundamental thing. Look at not only kids, look at deer, every year they regrow huge amounts of bone, and innervation and vasculature, interestingly, they don’t put weight on it. That’s probably one reason why it happens in deer because they are not trying to put weight on it, so it still makes sense to try and regenerate it. I think it’s going to happen.

Have you been able to observe an artificial bioelectrical pattern being passed down through generations?

That’s an interesting question, it depends how you define generations. If you define it as a traditional sperm and egg sort of sexual reproduction thing, no we haven’t observed that and we haven’t really looked for that. However, planaria the way they can reproduce is they tear themselves in half and each half regenerates, now you’ve got two worms, so that’s their mode of reproduction. Through that mode of reproduction, absolutely, we see that the two-headed worms when they split, you get a two-headed worm out, so it’s stable through that. You could imagine, we take a two-headed worm, we throw it in a river somewhere and a couple of hundred years later some scientists come along and they scoop up some sample and they say, “Oh, a two-headed form and a one-head form, a speciation event. Great! Let’s sequence the genome and see where the speciation event is”. Of course, that’s not where it is. And so, from that kind of perspective, yes, it can go transgenerational but it is not the same thing as going through the egg and sperm chokepoint.

Regarding the ethical considerations needed after the creation of intelligent beings with these kinds of origins: what is your opinion about what should be considered when evaluating the appropriate level of empathy towards such an agent? My only intuition about this is that complexity is relevant to this.

Well, I’m not going to pretend that I have the answer to this, I’m not an ethicist. I don’t know how to produce a new system of ethics, but I do know that the criteria we used to use are no good. It used to be that you could walk over and knock on something, and if you heard a metallic sound, you’d say: “Ah, that came out of a factory, it’s probably pretty boring, and I can do whatever I want with it”. And if it was soft and squishy, you’d say: “Well, I better nice to it and that kind of thing is gone”. It’s not going to be useful at all anymore. And so, what are the right criteria? My intuition says that it’s something about goal-directedness. The new golden rule might be something like, ‘Be nice to goal seeking systems’, something like that. I think you’re right, I think complexity has something to do with it. But it’s not complexity in any way we know how to measure because our current metrics of complexity would take up a completely random image of snow that would have maximum complexity, that’s not what we mean. I’ve argued that it has to do with the scale of goals, the spatial-temporal scale, that any particular system can entertain. So, if all you care about is the local concentration of butyrate and you don’t have much memory going back and you don’t have much memory going forward, so you have a tiny little cognitive light cone that’s small, not for what you can sense and do but for the goals that you can entertain, then you probably are a tick and you have a particular set of cognitive capacities. If you are dog, you probably have a bigger cognitive light cone, you have some memory going back, some memory going forward but your goal states are still limited. For example, you never are going to care about what happens three weeks from now in the next town, it’s just impossible for you. So, in that sense, you have a bigger cognitive light cone. And then if you are a human, your cognitive light cone might be enormous, you might be working towards world peace and be really depressed that the universe is going to freeze down in billions of years, or whatever. I think what we need to be able to do is to be able to define where certain systems land on this kind of metric of goal space, how big is your cognitive light cone and have some kind of a scale about how you treat systems given their capacity to be stressed out by failure to reach those goals. Another way to think about this is tell me what you are stressed by, and I can tell your level of sophistication. If you are stressed out by the local level of glucose, you might be an E. coli, if you are stressed out by the state of the economy over the last twenty years, you are probably not an E. coli. So, the goal states that you are capable of representing and comparing to the current state and being stressed by them is a pretty good indicator of complexity. That’s my intuition that this should all go in that direction. Nothing to do with what you’re made of, or how you got here but the types of goals towards which you can work and exert your life energy. That’s where I would go with this.

Have you investigated/developed any methods for tracing causality in multiscale systems? It seems that this must be the case given the cascading effects across scales you mention that bioelectrical signalling can have. Is there a way to determine which scale is most influenceable and influential?

There is, and I can give you more details, drop me an email and I’ll give you more details. This is in particular the work by Erik Hoel who is a collaborator in my center. And he actually has quantitative methods for tracing this, we have empirical methods, and he has a nice quantification of new math around this whole area.

A deep dive into the topic:
Albantakis, L., Marshall, W., Hoel, E., & Tononi, G. (2019). What caused what? A quantitative account of actual causation using dynamical causal networks. Entropy, 21(5), 459. [pdf]

Baluška, F., & Levin, M. (2016). On having no head: cognition throughout biological systems. Frontiers in Psychology, 7, 902. [link]

Becker, R. O., Selden, G., & Bichell, D. (1985). The body electric: Electromagnetism and the foundation of life. Morrow: New York. [link]

Bongard, J., & Levin, M. (2021). Living things are not (20th Century) machines: updating mechanism metaphors in light of the modern science of machine behavior. Frontiers in Ecology and Evolution, 9, 147. [link]

Levin, M. (2021). Life, death, and self: fundamental questions of primitive cognition viewed through the lens of body plasticity and synthetic organisms. Biochemical and Biophysical Research Communications, 564, 114-133. [link]