Mind & Matter
Mind & Matter
Evolution, Language, Domestication, Symbolic Cognition, AI & Large Language Models | Terrence Deacon

Evolution, Language, Domestication, Symbolic Cognition, AI & Large Language Models | Terrence Deacon

Download, watch or listen to M&M episode #141

No transcript...

About the Guest: Terrence Deacon, PhD is a Professor in the department of Anthropology at UC-Berkeley. He has written many papers and multiple books about the evolution of human language, origins of consciousness, and related topics.

Episode Summary: Nick and Dr. Deacon discuss various aspects of biological evolution, from natural vs. sexual selection to gene duplication and the consequences of domestication; the domestication of dogs and songbirds; human vs. non-human forms of vocal communication; ritual behavior & the origins symbolic cognition; artificial intelligence & large language models (LLMs); and more.

*This content is never meant to serve as medical advice.


Full AI-generated transcript below. Beware of typos & mistranslations!

Terrence Deacon 4:12

Oh, yes, I'm in the middle of a book called fouling up. How inverse Darwinism catalyzes evolution

Nick Jikomes 4:21

what is what is inverse Darwin Darwinism. So

Terrence Deacon 4:24

everybody asks, it's actually not anti Darwinism. It's a compliment to Darwinism. And I think it's a compliment that was there. In Darwin's own writing, though, he doesn't recognize it. And that's why I've used the phrase. So Darwinism in 1838. He came up with three ideas to send his chose up his notebook that basically drove the rest of his thinking. And it came up right after rereading or I don't know, they read the whole thing before this, but certainly reading Malthus on population this this whole problem Malthus realizes is that reproduction outpaces resource. And Malthus, of course, had this dire prediction that this would lead to terrible things in England in Europe. Because populations were growing so fast and resources were not growing fast enough. Darwin sees this and says, aha, but in the natural world, this is going to be the case. And that means there's going to be selection that favors a few over the others. Those that fit better with the contexts that can do better, they can outpace the others. And so this sort of drives his theory. And he comes up with three ideas, he says, the first thing is that I noticed that everybody realizes that the grandchildren are like grandparents, as basically they traits are inherited, they're passed on down. And he says, but but in fact, the second thing is that there is great variety and offspring. That is, although they carry some of those traits as a lot of variety. So that's a second story. That's a variation problem. Then the third one is, he says, But you know, then there's, there's great overproduction with respect to support of those offspring. And but those three things together, and you get natural selection. That's Darwin's basic idea. But notice that the Malthusian part isn't necessary. And it was what gave him the insight for natural selection. But if you think about things like gene duplication, that happens all the time in evolution, in fact, whole genome duplication does happen, particularly in plants. What happens is that the excess production doesn't necessarily mean that you're outstripping resources, doesn't take a lot of energy to produce extra genes, or to have a single gene that's duplicated, like in transpose on effect, there's not much energy use, but even in terms of multicellularity, think of the production of multiple cells, the reproduction by mitosis. Now, that builds a large multi cell body with maybe trillions of cells. There's some contexts in which some of those cells are in competition with each other, but not often. Usually, it's a context in which there's no competition, which is excess resources, in which overproduction is in fact, the way life has to work. In fact, to stay ahead of the second law of thermodynamics, you can't just repair things, you have to make extras, you have to have backups, you have to have duplicates, because something's gonna go wrong, eventually, we know that in terms of our own computer technology, that you got to have backups, you got to make copies of things. And that's how you keep things going. Well, evolution, of course works. But what happens when you don't have competition? When you don't have this subtractive effect, if you think about it, natural selection is a subtractive process. It requires variation. But it also requires overproduction. You have to make more than you need because you're gonna get rid of some.

Nick Jikomes 7:59

Yes, so you're taking away individuals who are unmatched to their environment, and leaving behind those that are better matched to their environment. But But I think I see where you're going here. If you think about something like gene duplication, that's not happening at all, you're just adding a new thing. And then that redundancy gives you space to evolve a new function without any actual competition there

Terrence Deacon 8:21

exactly are what it basically does is because there's redundancy, if one of them begins to degrade in some way, which will happen spontaneously. In fact, if you got two of them, then if one degrades, it won't be eliminated, unless it produces some troublesome project product. And one of the troublesome things that are going to happen if you've got to have something is that you can produce too much and too much can be a problem, I mean that we have a classic issue of that with respect to Trisomy 21, when the 21st chromosome is tripled, tripled, and said, Just dupe duplicated data produces Down syndrome. That is the overproduction of a bunch of genes is actually not such such a good thing. So oftentimes, when you over produce, it can also have negative consequences. So in fact, accumulating damage in the experts actually could be an advantage, you could decrease it. But the other thing is that you can also because damage on say one out of a few 100 nucleotides and the DNA molecule does a little bit a little bit of damage won't necessarily turn something completely off. It'll just kind of modify that function, oftentimes degrading it just a little bit. But there might be contexts in which a little bit of a change little bit of variation now actually complements what's already there. And so in effect, duplication, over duplication without elimination, allows enough redundancy that allows you to accumulate a little bit of variation and error without much cost.

Nick Jikomes 9:53

Yeah, and there's actually like, there's multiple ways this can go so for example, you know, gene duplication is relatively common, you get an extra copy of a gene, it's often not catastrophic. It can be, but it's often not catastrophic. And you can imagine, I mean, this, there's, there's examples of this that have been documented. You know, one way this could happen as you preserve the original function and Gene one, and Gene two just sort of goes in a new direction. But another way that you see sort of complexification evolve in the genome is what they call sub functionalisation. So that if you've got sort of two jobs that gene is doing, and then you duplicate it, it can be that gene one now just does job one, and Gene two just jumped to the now you've got the same functional outcome. But now you've got more complexity, more parts doing it, right. And

Terrence Deacon 10:40

that's the classic stories of those one is the hemoglobin gene, which is duplicated from myoglobin originally. And then the hemoglobin gene itself duplicates, and you've got an alpha and a beta. And they produce this, this tetramer, that now actually does a much more interesting job of carrying oxygen, because now you got four iron atoms, in this four part molecule, it's floating around your blood carries more oxygen, but also probably gets it off more easily to myoglobin in your tissues. But in addition, the beta hemoglobin duplicated a bunch of times, and now we have different hemoglobin through expressed at different times during gestation in the fetus. So that at times, because the fetal hemoglobin, of course needs to take oxygen from mother's hemoglobin, so it needs to have a slightly different oxygen affinity. But as a fetus grows, that difference in oxygen affinity needs to change as well. And when it's born, eventually, then you need to shift away from that because you know, the stronger your own hemoglobin holds on to oxygen, the more difficult it is to give it off to your tissues. So when you get when you're born, and you're getting oxygen easily just from the air, you now need to go back to a sort of a weaker hemoglobin, the standard beta. So this is a case in which it's duplicated many, many times. And in fact, the duplications all do different very different sub functions. One of them is within the tissue myoglobin, one within when the blood Oops, my my light went off, just turn my back on. And, and, and one has to do with a different timing in the lifespan. The other way that some functional sub functionalisation can work is of course, if rather than modifying the structural gene itself, modify the promoter in some way. So it's expressed a different time. Well, that's of course, the beta hemoglobins of that are duplicated into gamma and so on, in fetal life, different time in the in the lifespan, but you can also have it produced at a different part of the body, a different segment of the body. And, of course, that's what's happened with all the HOX genes that make up our body, that the same gene with a slight variation is now producing a different part of the body. And that different slight variation, now it can either produce, you know, very, very similar limb parts, like our fingers that are very similar to each other, or can produce very, very different structures, by virtue of now not being in competition with each other, but now actually complementing each other at a higher order level, that is a level of old body structures. So this, this process is one that occurs at many, many levels. And I've been focusing on all of those loans. Actually, the one that got me thinking about it originally had to do with the fact that we primates, need to eat vitamin C, we need to eat ascorbic acid and fruit, whereas almost no other mammal in fact, very few other animals in general need it. Because they have a gene that produces it, we actually have a pseudo gene, that somewhere back in our evolutionary past, we think somewhere greater than 50 million years ago, actually produce vitamin C and all primates, but the anthropoid primates and monkeys and apes lost this capacity. And, and they lost this capacity probably because they were eating fruit, and the fruit provided it from the outside world. So in effect, it masked natural selection on the gene and it allowed it to just pick up noise. So in fact, what we see now in most anthropoid primates is not only noise in the gene with lots of stop codons and the gene and so on, but in fact, is a major frameshift so that it's just complete nonsense now. And yet, that means that we're addicted to vitamin C, we have to eat it, we can't make it ourselves like like your dog or your cat can. So, as a result, a number of other things have happened, you know, we have changes in taste receptors, so that we actually have this pleasing interest in sweets and sour that is the acidic nature of vitamin C. But we also you know, subsequent to this the in, in monkeys and apes, duplicated the absence that is the the genes that are involved in producing light, receptive pay admits in a retina. And so the what amounted to the central pigment, a green sort of central frequency pigment duplicated got plugged right next to itself on the end of the X chromosome. And that one began to degrade. And as a result, it reduced its sensitivity to free to higher frequency became more and more red. In its reception. Now we have this sort of red green distinction, which of course, is critical to determine the ripeness of fruit, fruit is ripe, and you're giving a color signal to birds mostly. And initially, that says, Okay, now green, we were camouflaged, you can't see edible yet. But now I'm gonna make a clear change. And now you can eat me, well, that's a good thing to know, if you're a primate and you need fruit, you need to have this capacity. So one of things that happened here is that now a duplication outside the body. It's not just like a gene inside the body, but now it's outside the body created a whole bunch of other changes elsewhere in the in the body. So many genes are now contributing to the fact that we're getting enough vitamin C, getting enough ascorbic acid, we also have some, you know, blood borne molecules and so on that make it easier to take it in and pass along. A number of other adaptations responded now because we're, in a sense, codependent on something outside the body with duplication effect is both something inside and outside. And it's part of, it's going to play a role in, in all kinds of interactive effects, including social behaviors, the link with the outside or symbiotic, and commensal kind of behaviors that may be driving this or been driven by it. So a lot of these features, and what I'm interested in this new book, is to just go through the whole list fall from, you know, genes all the way up, I think, to even language, I mean, like, the classic example, where we depend on each other, you know, we have to be boring in the social group. So we're addicted to the social group, that's gonna do the same kind of thing that the vitamin C story would do.

Nick Jikomes 17:12

Yeah, I mean, there's multiple, there's multiple concepts that are that are important. So you know, to play with the vitamin C and the color vision example more. The basic idea here is, first, you have a lineage of primates that develops color vision, which is itself coming from an internal redundancy is happening spontaneously, you duplicate your opsin genes that encode our light sensitive proteins in our eyes. And now you can have this red green distinction, because you've got this syrup, you've got this extra gene that is now free to evolve. And what's interesting there is the redundancy is creating a context in which the novelty is actually coming from the relaxation of selection. So it's not just the subtractive process of taking away individuals who are not good at in a particular environment. But you've created this redundancy, the surplus, this extra gene in this case, and it's free to evolve, ie, I mean, what that really means is, there is no selection to maintain its existing function, because you've already got another backup copy. And then we're saying that ability, the color vision ability gives us the ability to detect ripeness of fruits and things. So now we've got external vitamin C, that we can readily find and adjust that then relax, it really does another relaxation of selection on our endogenous vitamin C production. And, you know, what's additionally interesting about this is once that relaxation of selection comes into play, and that endogenous vitamin C function gets mutated away. Right? There's, there's a sort of unit, you know, directional thing here, right? You can't, you can't undo that change. And so now you're really locked into, it's almost like creating more selective pressure for more novelty to evolve, because then you need to really, really make sure that you're never going to lose that ability to find the vitamin C in the external world.

Terrence Deacon 19:01

That's right. So you've got to have, you've got to like the flavor. You've got to have better abilities to move around in the trees. It's an open question whether the opsin duplication was before or after the degradation of the ascorbic acid Gene. Gene is called Gulo. For a complicated name, I won't go through it. But but that gene, we know we have a pretty good sense of when that duplicated and we're getting a pretty good sense of when we get I shouldn't say we get we had a pretty good sense of when that gene began to degrade. And by just comparing lots of different primates, and we have a reasonable sense of when the first duplication the opposite did in part because it turns out to be somewhat later. It turns out to be somewhat later because we see a different variant in older New World monkeys. And so it's clear that all the New World monkeys had already began to eat fruit and Become diurnal, their eyes get smaller, their teeth change in the fossil record. So it looks as though they're diurnal, which means they're probably already eating fruit at this point in time. But now we see that old New World monkeys separate. And it's not until the separation that we begin to see a slightly slight difference in how color vision is processed in all New World monkeys, Old World monkeys, like ourselves, were Old World monkeys in that sense, have this this three color vision all based upon one chromosome. And New World monkeys, it's across chromosomes. And so some New World monkeys, many New World monkeys are just die Chromat sanely, see in two colors are some by virtue of an interesting combination of their gene of their chromosomes become trichromatic. So it's an it's an interesting problem. And I think what we'll see, we begin to look at some of these other adaptations for ascorbic acid development, including the taste, taste changes, and so on, so forth, probably even changes in the teeth and locomotion and so on May may be involved as well.

Nick Jikomes 21:04

Yeah. But yeah, I mean, so it's just like, an essence, like a single, in this case, metabolic shift to losing the ability to produce vitamin C internally, it basically creates strong selective pressure that's going to be distributed across many different adaptations, right, you have to be able to detect fruits, by sight or other means, you have to be naturally interested in finding those things, you have to have the motor capabilities of obtaining them, and so on and so forth. And I think you know, where we're going with this, because there's probably something similar happening in language, the story of language evolution, but I want to talk a little bit more about some basic evolutionary concepts so that people have the proper toolkit. To that point, one of the things I want to get you talking about is, we can assume you know, this audience, you know, I've had many podcast episodes with with you and others on some of these subjects, you know, what we'll assume people know the basics of natural versus sexual selection. But I want to ask you about when these things come into conflict, so cases where an adaptation is favored by the opposite sex, and yet it makes it harder for an individual to survive in its physical environment? What are some examples of that? And why on earth would that conflict ever arise to begin with, right,

Terrence Deacon 22:22

that probably the best example of this is, with what are called widow birds in Africa, the widow bird is the male has an incredibly long tail. And during when it's displaying, and the mating season, basically flies low to the ground with a weird kind of slight slight that caused it to sort of go low and almost touch the ground and then come back up and fix the ground and come back up. But having a long tail means that that drags close to the ground means that you're probably pretty easily picked off by a predator under those circumstances. Now, it turns out that if, and this is sometimes called the handicap principle, that is basically these guys are actually risking their lives to display the females, because they have these incredibly long tails. But if a, if a number of males are out there, and many of them get picked off by predators, but one does not. And he's taking risks, then, if a female chooses that male to mate with, it's likely that that male is more healthy than the others more sophisticated in their flight, maybe has less susceptibility to disease, and so on, there are certain advantages that the female and her offspring will get from that process by picking a male who has survived despite these so called handicaps. The male on the other hand, that can actually survive under these circumstances, is also likely to have many more offspring. So it's an advantage for both males and females. Even though even though probably 80% of the males don't make it, most get picked off. One of the experiments that was done years ago, I can't right now remember the names of the experimenters but they basically clip the tails off of some individuals, and use super glue to attach that to other individuals. So some individuals had absolutely longer tails, and some individuals had shorter tail. So of course, that's going to correspond to the different predator effects. And they're going to be predict preyed on differently, but it turns out to the females prefer the individuals with extra super long tails. Now, what that would suggest is that although over time those guys will be probably be picked off and wouldn't make it at least for a brief period of time. Females don't have any limit on how long the tail ought to be. The longer the tail, the better looks like for them. Um, In part, because over time, there's going to be a balance at some point in time, as tails elongate in Burr in these male birds, it's going to be a trade off. So that a certain length of time, the amount, the risk you take of being preyed, and never reproducing. Compared to the advantage you have of maybe producing a lot of offspring, there's going to balance out so that natural selection, when effect set an upper limit on the length of tails. And, interestingly enough, does not change that for females. The Why wouldn't it change it for females. But ultimately, it's a pretty straightforward story. If the environment changes are fewer predators, better better for females to choose the longest tail, even if it's way longer than normal? So there's no real limit on female choice here, let's go to limit what females think is the optimal tail length. But there's going to be an optimal tail length set up by the environment. By virtue of you know, what's the probability of a male surviving with a long tail. So there's a case in which sexual selection which is driving this choice of males by females, and natural selection, which is setting an upper limit, so to speak on the length the tails, actually, you're setting up a balance, and what will happen is it will run to the point. So if you are predators in the environment, for example, we should expect that tail feather length should get longer. If there's more predators in the environment, we should expect that tail feather length in males will be shorter. So that to some extent, this balance between sexual selection and natural selection will actually be very useful in sort of fine tuning to the environment.

Nick Jikomes 26:41

And so I suppose that would imply that as selection pressure in the survival sense, gets lessened. So for example, if a species somehow finds itself in an environment that's less dangerous, there's less predators, you would expect male phenotypes to start to become more and more driven by meat, just raw meat preference rather than survival value. In other words, sexual selection will sort of beat natural selection in that sense,

Terrence Deacon 27:10

to the point that it hits this upper wall, it will be able to go forever.

Nick Jikomes 27:17

And I mean, what's interesting about that is, you know, when you start to think about our own species, right, we've we've created social structures that have eventually allowed us to create societies where the environment is not as consequential for us as it was when we were more primitive hominids walking around with with saber toothed tigers, you know, hunting us. In other words, we've removed some of that, select that natural selective survival pressure. And so the expectation would be that more and more of human male phenotypes over evolutionary time in our lineage should be more strongly. They're being pushed by natural sexual selection more than natural selection.

Terrence Deacon 27:57

And yet, the major predators on humans are humans. So this is going to change the balance in a different way. What it says is that cooperation may be the safest thing to be to be a cooperator. But to be fearful of others, some strange others, but via a very strong cooperator within your familiar group, these things might be an advantage. So there, there will now be a sort of different balance, different features will be unbalanced, and will be shoved one way or another. And, of course, to be a good cooperator. As a human being, you also need to be able to communicate well, you need to be able to tolerate the variation of your close relatives. And you need to some extent, cooperate, succumb to the the group's ideas, you maybe even need to sort of, you know, follow along, to stay in a large stable social group. So there are going to be a number of things simply because to acquire language, to avoid intergroup predation, so to speak, there gonna be a lot of very different features at work. Now, the troublesome thing is that as technology develops, of course, it's going to it changes the environment, it's going to amplify certain things, and damp others. Now, I don't know, I would guess that in the amount of time that significant technological developments, including, you know, weapons of war, beasts of burden, and so on, and excess resources and so on, there's going to be fairly recent in human evolution history, and probably not a lot of natural selection has been at work. I think that's probably a good thing, because it would have driven us I think, in even more extreme directions. Nevertheless, I think those biases that were already there can be easily amplified by the environment and the sake. And so I would guess that the last, you know, 510 1000 years of human prehistory has been a case in which those biases have been significantly driven, although I don't think it's made a big difference at the, you might say genetic, neurological level.

Nick Jikomes 30:08

You know, another. Another area that's very interesting is when we think about Darwinian processes that are happening within individual organisms over the course of development, and you know, this concept of redundancy comes back into play, but in a kind of a different interesting way. So, for example, you know, when we talk about things like gene duplication, these are stochastic molecular events that just sort of randomly come up with some frequency, but baked into the developmental process itself. In particular, the development of the brain is this sort of, you know, intentional, quote, unquote, over overproduction of synapses. And it's actually a feature of brain development, that you produce an excess, and then you, you take away that excess early in development. So can you talk about those kinds of competitive Darwinian interactions that our neurons display as we're developing? And why why things would be set up that way?

Terrence Deacon 31:02

Right. Now, that's a very interesting issue, because obviously, you're seeing both happen you're seeing is overproduction you're seeing remember that one thing is going to happen is that brains are going to be different in different species for different reasons. You know, a classic example is, as primates, both of our eyes face forward, but if you're, you know, a prey animal, you want to have more 360 degree vision. So you want your eyes to the side, there's certain things you have to give up to do that. But the question is, what with different positions of the eyes? Do you need to have mutations in the nervous system that say, okay, you know, your eyes are this far apart, in order to make it work, I've got to, you know, I gotta have this gene change. Well, in fact, even among primates, and among our own species, there's a lot of differences between the width of our eyes. One of the ways of, basically having a nervous system that doesn't have to sort of with each subtle changes in the periphery to change genes, you know, are going to revolve changes in connections in the end in sight, one way to do it, is to have the development of the nervous system, adapt to the body itself. So if you think about the Darwinian story, think, you know, think about the body as the environment, how is the brain can adjust to a variable body? Well, the answer is, it's going to work like natural selection does, you're going to generate connect extra connections, you're going to allow both options to take place, both all possibilities of eye position, for example, might work. But then you've got to use the relationship between the periphery, the input of the eyes, and the competition, now set up between synapses to organize vision in a way that matches the outside. So in many respects, embryonic development, and at many levels, including extra limbs, extra fingers, and so on people that are born with a sixth finger, for example, usually, it's a working finger. That's because the nervous system has said, okay, you know, it's developing in a way that has anticipated How many fingers you're going to have. It's not

Nick Jikomes 33:13

all pre specified, not pre specified, which is actually a huge feature, right? Cuz then you don't have to encode all of the detail in the genome. That's

Terrence Deacon 33:19

right. So, but this, what it does mean is that now epigenetic effects are going to be much more powerful. That's going to have to be important at many, many levels, once you allow epigenetic effects, that means also, the environment can begin to have more of an effect on the ultimate phenotype. And with the nervous system, we certainly know that if you if you damage peripheral systems in some way, you can actually change radically changed maps in the central nervous system. So and that's because it's responsive to this. But now, here's the other piece is that, you know, we have quite large brains for primates. How is this competition going to be modified? How's it going to be affected, we don't oftentimes think about the story of brain evolution in our own species, as an EVO Devo effect that is in which evolution has been affected by developmental effects. But clearly, the wiring of the brain since our brains are much larger in a smaller body, the wiring is going to be different because this the the Darwinian like effects are going to be modified. And I think about this sort of like I think about gerrymandering in elections, you know, we we shift the balance in some way or another, and you actually change the consequences that give some a better chance at competing and make others less likely to compete. Well, competition is driven in part by a phrase that's commonly used is that neurons that fire together wire together. Well, if you've got more of them in a certain place, they're going to have an advantage. They're going to have a gerrymandered advantage of taking over more space in the nervous system. Yeah,

Nick Jikomes 34:58

and Yeah, so you can essentially get get more votes by creating more neurons or altering the ratio of connections from one place to the other. But to illustrate this, I think I think a good example, and you can probably give many of this is this idea of, of natural selection and competition between neurons and from different parts of the brain. When you deprive someone of like one sensory modality, one part of their brain that we think of as being visual, or auditory, or whatever, can be taken over by another modality, are there some examples of that, that illustrate this, this competitive process

Terrence Deacon 35:31

in a brain as large as ours, it doesn't really have that kind of a massive effect. Even congenitally like people who are, don't really have a lot of auditory responsiveness in visual areas, for example. But a smaller brain, where distances are not so huge like in rats and mice, there's significant changes in the mapping structure that can take place. Probably the best studied in classic example, has to do with whiskers, the what we call fight Bresee, the faces of cats and mice and rats and so on these long, you might say extra thick, extra strong whiskers that stick out. It turns out that if you look in the part of the cerebral cortex, that is the map of what you might call tactile space of the body, there's actually a very precise map of whisker positions, you can actually stay in the brain of these animals after after birth and after death, and look at a very precise map that looks like a kind of a checkerboard of places that each response correspond to a whisker position. But in early life, you take a young rat or mouse and you clip out some of these whiskers, the map will be different. The map, the existing whiskers, will, in a sense have taken over the space of the missing whiskers on the map. So the map actually is now corresponding to the body, we also find this in the case of congenitally deaf mice, for example, you can see that, you know, because of the genetic effect, maybe disturbing the cornea, the cochlea of the auditory system, what can happen is that you see a shift in which visual and tactile areas have taken over some of the auditory areas, not completely, but basically disturbs that map in the human brain. And I would say, in most primate brains, because they're fairly large, that kind of competition for space is probably slightly different, it can't take over quite such a large area. So I suspect that it's minimized as brains get bigger. On the other hand, with the enlargement of brains, what we're seeing is a sort of functional version of what we were just describing in terms of redundancy by virtue, like gene duplication, for example. Now, if I have input going to the same area, but it's twice as large, I only, I may only need half of that, or a fraction of it, or, or I might only need to use some of the cells in it for the original function, and there can now be a relaxation effect that allows more differentiation and function in a larger brain. So I do think that there is an effect that's analogous to this in brain development. And it's partially because it has this two sides. It has, I like to use this old this phrase, you know, all work and no play makes Jack a dull boy, well, you know, all selection, and no variation and no overproduction. No relaxation, makes evolution dull, it can do as much, you really need to have this sort of freedom to produce variation, that's not a problem. And so it shouldn't surprise us that play in animals with complicated brains might also play a very significant role like this. That is, this is something we've ignored, because we think of complexity and, and greater capacity is always requiring work. And, and you know, you got to select on it, and you got to get rid of things and hone things in, we tend not to focus either on whether it's just a cultural thing, or just the history of the field, we tend not to focus on the play side of it much. And we mostly just talk about it in terms of variation. Well, it's not just variation, its tolerance to variation, that that that allows you to explore, just as play in, you know, in predators allows them to explore various means of playing with their prey or sneaking up on prey and so on. It's an exploration. So I think this is that's why I call it an evolutionary catalyst. Yeah. Makes it more likely that certain things will happen.

Nick Jikomes 39:53

Yeah, I mean, I think what you're saying is literate. So you've got built in redundancy and this over auction, that's just an intrinsic feature of the system of how brains develop, quite literally, play represents a different pattern of walking through a neural space. And a different pattern of exploring that space is going to give you a different selection pattern, which is going to, you know, it's ultimately going to give you a different set of features than you would if you were solely focused on survival. And you were quote, unquote, working in your environment all the time.

Terrence Deacon 40:26

Right? Well, and, and if you think about it, play in mammals, in general, is the context of relaxed selection, you make mistakes, you're allowed to make lots of mistakes, you don't even have to work with the real thing, you know, just with with effigies of prey, you know, I can be a cat chasing a ball of yarn, and practice, in effect, not because I'm practicing, being a killer. You know, it's just fun. And so part of it also has to do with, you might say, selection for having fun. And I think part of what we describe oftentimes isn't the knee in our sort of childlike like, you know, the way we like to play like we're playing with ideas now is in part just because we've got the capacity to do that. And that actually is an advantage, because it exposes lots of opportunities that we wouldn't have discovered before this, it allows us to combine things in unique ways. And of course, we human beings have, I've done this extraordinarily well. Once we've developed linguistic capacities, and our technologies and so on, it allows us to sort of offload all of this play also on to other things that we love to do. Including, you might say, the arts, the arts are an example of play.

Nick Jikomes 41:53

One of the, you know, one of the things that's, you know, the areas that that you've explored in some of your writing that I thought was very interesting, is domestication, and how domestication really ties into a lot of the things we've been talking about around relaxation or reorganization of selective pressures. You know, oftentimes, when we talk about domestication, we think of humans domesticating other animals like farm animals, and of course, right, if you're building a barn and the fence and making sure the wolves never get too close, you're relaxing the selective pressures in the environment that these prey animals would otherwise be exposed to. And therefore, they're going to evolve in different ways they would not have otherwise evolved. But we've also, as you've pointed out, and others have pointed out, we are also a self domesticated species. And that's also very interesting and important to think about. So just starting with the basics here, how exactly as sort of an evolutionary biologist, defining it in terms of selective pressures and constraints and relaxation of selection? What exactly is domestication?

Terrence Deacon 42:54

The problem is that domestication is many things. And we have one word for it. Clearly, we have domesticated species. That is, we've not imposed selection so much. In terms of the outside world, but we've selected we selected for animals that can work to get to can hang out together in large groups without fighting and competing. Therefore, we have to select against population stress effects, we have to select against aggressivity. And if we're going to have a bunch of them together that we're going to raise, we, as breeders select those, and as we select various traits in dogs, and so on, we're doing the selection, that's active. domestication. Clearly, that's that's going on with lots of human domesticated animals. And of course, plants, we select for certain features in plants by simply allowing some to reproduce more effectively, and others to not reproduce. So we're playing this role, the role that actually got Darwin thinking about this in the first place. That's why I call it selection. selective breeding is playing a role in this. And so the question is now for different species, what's being actively selected in terms of selective breeding, well, clearly reducing aggression. We even go farther further than this in particular, in a herd animals, where males might typically be competing and fighting with each other, and actually injuring each other in competition for mates and for controlling large groups of animals. We very often neuter the males, so they never get to that stage and so they can now actually stay in the large social group without being very aggressive and about being damaging. And yet we've also selected for more stress tolerance, because being in a large social group, and being confined may be stress So, but so we select for this and what we see, when we look at a lot of these domesticated species, when we say domesticated, typically, we mean by humans, domestic, of course refers to being in the home, you know, so basically hanging around us, we're gonna see that there's many of these features that are characteristic that we would want to make possible. So the human beings have played a significant role in this. And as a result, domestication in animals or going to eat actually produces changes in their bodies, in the same way that domestication of food crops has amplified, those parts that we will eat that will be editable, as opposed to those not, those are very distinctive, you might say, in enforced or human selected domestication. But then there's domestication in other ways. And we now think that dogs for example, self domesticated, those wolves, self domesticated to become domestic dogs, in part because they also needed to be able to find this wonderful resource, we provide garbage piles, you know, we things that we can't eat, that are still edible, if you got the right kind of digestive system and, and, and masticatory system that you can maybe crack into the bones, you can chew these things and, and get food that human beings would not want to keep. But that meant that that wolves that were more tolerant of each other in large groups around these, but actually do better that is they are better in social groups that were not so aggressive, that would be willing to hang out hang around human environments that might otherwise be stressful and dangerous, and might not be a threat to people so that people wouldn't go after them. They're going to self domesticate in order to gain access to this. But then human beings can then of course, take advantage of some of the things that they do well, which is finding prey, and so on, and so forth. So a first stage of self domestication leads to humans controlling domestication in dogs so that we get a really complicated interaction here. So the dogs in a sound self domesticated in order to be able to take advantage of a resource. We can say the same thing about humans. Now, what was the resource that we had to take advantage of? Well, certainly, in the savanna of our very early evolution, probably close to 3 million years ago, there was a lot of animals that have been preyed upon typically large bobbins, and so on, on the, on the fields and savannahs of Africa. The one thing that the predators couldn't do is they didn't have Jaws strong enough to break the largest long bones of big Barbets. In other words, the sort of the fevers and so on these these huge, very, very dense bones, and you know, dogs chew on them, but they never break through them unless you break them up and cut them up, so they can get the marrow. But I suspect that our ancestors for a long time, were using stones and things to crack open nuts. We already had, I think, probably maybe going back almost 4 million years, we probably had this capacity to crack things open. And so I don't think it would have been much of a stretch for our early ancestors to be into crack, figured that they can go in after the predators and scavengers had finished their jobs and still find food, their bones, but they had to do so now in a very different context. To do so in a context in which exposing themselves to predators out in the open savanna. That meant you had to have a cooperation, somebody had to keep the predators away while you're Did you hit your head down, chopping on this, this bone, maybe cracking a stone, so it became sharp, and realizing that the sharp edge could also scrape off some of the meat, you can now go in there a little earlier, but you know, really had to chase off the predators. But now you have a situation in which this is wonderful resource available. But you have to be able to work as a group, the competition between you has to be reduced, because you're trusting somebody to keep the predators away while you're getting the food. And the guys that are keeping the predators away, you have to trust that you're going to share that food. At some point, when you get back and climb up a tree where you're safe.

Nick Jikomes 49:33

You obviously need to be able to communicate with a certain level of sophistication.

Terrence Deacon 49:38

Right to Know who's doing what, when and to know that it's reliable. So all of this, I think sets us in a situation a little bit like the dog situation in which there is this wonderful source of nutrients out there. That's not available to anybody else. And then if you have the right tools and the right social organization, you can get to it, but it's going to select for all this that's going to select for many of the things that are selected for in normal domestication as a change in stress response and ability to handle being in large mixed groups possibly reduction in competition, possibly competition over females. And it might select for a form of communication that made it easier to do that. It's what I think language does, it allows us to communicate about the past the future promises, expectations, and so on, in a way that other kinds of communication does not. I think that had to be a very early stage in this process, and probably, I think, probably took a couple of million years to develop to the level we have. Now, that's a very, it's a very long term process. But in effect, I think it begins with a kind of self domestication. And of course, if you think about language itself, it already requires that you're in a social group, you can't get it unless you're in a social group, you can't get it unless you're in a social group that can survive over generations, because language has to be passed on and developed socially over a long period of time. It's not something that just develops in a few years or even a lifetime. So that's sort of a prerequisite. And I would guess that socialization and language sophistication probably co evolved over a long period of time, one pulling the other along, that being part of a very unique kind of domestication, that that we were involved in. But in this case, it's us. It's a self domestication story, all the way through.

Nick Jikomes 51:31

Yeah, I mean, one of the pieces that's fascinating there is, I mean, there's multiple threads, you could tie into this. But one of them that salient to me is, if you look at the metabolic and digestive adaptations human beings have compared to other animals, and you just sort of say, what does our metabolism digestion look like? Does it look the most doesn't look like a herbivore like a cow? Not really? Does it look like a carnivore? Does it look like this, that or the other? And the answer in terms of things like stomach pH and various other features of our GI tract is we have similar digestive systems to scavengers, particular meat scavengers. And that's basically ties into the story that you are telling them. I think others have told that a key piece of early human evolution, and I'm not sure how much explicit evidence there is for this, is that we were basically power scavengers, who could not only organize ourselves to ward off, you know, big scary carnivores who made kills. But we could take advantage of these food sources that gave us and once we could do that, they were not only food sources, but they're very, very nutrient dense, which maybe even helped unlock future brain evolution, basically,

Terrence Deacon 52:42

I don't have a lot of faith in that we needed food of a special sort to build big brains, there's a lot of theories like that, I actually don't think so much of a driver. It's a sort of, you know, if you've got it, then you can build it. I think it probably works the other way around. My suspicion is that that brain size may have evolved, in part because of a better food source. But we do know that that that human vegetarians and and people that don't get a lot of seafood and so on, don't have a lot of the particular kinds of lipids that are necessary for building the myelin of brains still have pretty normal brains and large sized brains. I don't think we have a lot of metabolic evidence that there is a the metabolism made brains possible. But this is an empirical question. And we'll find out as we begin to pursue this. And we'll know in part because of the genetics of it, as we learn more about the genetics, for for converting tissue, I mean, converting nutrients into neural, neurological tissue, how it's done, under circumstances of extremes. All of this will, I think, help us understand it, but we'll be able to also begin to look at the human genome. And, and to some extent, compare it to some of the Paleo genomes, that we now have to get some idea of how this has changed. But this unfortunately, we can't go back. Of course, we don't have paleo genetics that goes back more than, you know, then not even 100,000 years. So we're not going to get into our deep history this way.

Nick Jikomes 54:21

One have a really nice example of this sort of domestication and relaxation of selection phenomenon that will tie in a little bit more explicitly to getting us thinking about communication in humans are some fun examples in songbirds. I know that you've written about, you know, bengalese finches versus their their wild cousins, and how their songs are distinct and the distinctions arise out of, you know, the relaxation of selection that comes from domestication. Could you maybe rehash that story for us? And I can do that. Yeah.

Terrence Deacon 54:51

And first of all, we have not done I don't want to be clear, we have not definitively proved that is just relaxation that's involved. There. Probably Many things like any other thing, but we find in these birds we've now begun to do one of my former students began to do some good genomics in comparison comparison, the wild species, the domesticated species. And what we do find is that there's some significant genetic changes. Some of them are typical that we find in domestication, that is reduced stress response, and so on, and so forth. So that's, that's quite clear. So one of the first things is that we do know that almost all domestication requires a few changes that are very similar across groups. But then there's those aspects that are different. The story that that interested me initially was this, that the ancestors of bengalese Finch, they called the white back Munia, or white rump Munia, from China, Taiwan, and areas in Asia, basically has a fairly innately structured song. By that I mean, that it doesn't vary from individual to individual from group to group very much. And in fact, if you're isolated, you don't hear anybody singing. Nevertheless, if you're a young male who reaches maturity, you'll begin to sing, they go all songbirds go through what they call a sub song stage, as they reach puberty, which, which involves producing a song that is not like the adult song. But what they do is by listening to themselves, they correct it. And over the course of a few weeks, a young bird will begin to develop a song that sounds more like an adult of that species. And we assumed this is because they have what amounts to a kind of an auditory template that they've inherited, and maybe even a motor template that biases what kind of sounds they produce. But when they produce their first what they call some song, some people like to compare to babbling. I think it's similar but different in a lot of respects. But let me just describe the bird case, first of all, and that is they'll produce a vocalization that's got some features in the adult song, but not many. But if they hear themselves sing, they will correct it slowly. But surely, song after song after song, they'll sort of converge towards the normal adult song, we assume that's because they can hear themselves sing. And they hear that it's not quite right. We know this, because death and birds never do this. If they're definitely they always produce a weird song that stays weird, doesn't get corrected. They can't hear themselves sing. But now the question is what happens when you're domesticated? The the bengalese Finch that we now understand, was domesticated, at least in the the example of the Japanese that we've been looking at. This is work by a man named Cosmo Okinawa, who's in, in Tokyo, the university there, has been pursuing this his whole career. And what he finds is that the point, Rob Mooney, which has this very fixed sort of, you might say, innately pre specified song, that it nevertheless has to learn by listening to itself sing. The bengalese Fitch, which has been in captivity only for about 250 years, very short period of time, was raised, mostly bred for its coloration. And what happened is that for the most part, they bred away the brown color, the sort of camouflage the color that the wild species had. So it's mostly white with a couple of patches of color on it. And of course, you can have breed for coloration patterns, it turns out that one of the standard things that happens in domesticated species is a group of cells called neural crest cells actually play a role in both building the stress response, because they're, they actually become part of the adrenal gland, the adrenal medulla, but they also play a role in pigmentation. And so what happens during during domestication, as you begin to relax selection on this stress response, you also tend to produce kind of more modeled changes in coloration. And what I think that does is oftentimes produces lighter coloration. And so that clearly happened in these birds. And we've seen the evidence of it genetically. So they're clearly seeing this reduced stress response. And they were bred for this. It's just a coincidence that the stress response and the coloration started linked together. And this is just simply because of this, this epigenetic effect. But now, what's interesting is that the other thing that happened is that they began to produce more variable some not only that, they now produced a song that had to do with not their own singing alone. But if they heard another member of their species singing right after hatching, that was sort of fix in their mind. I think that it would become their new, you might say, auditory template of what Sound I put a song out to be like, and as they reach maturity, there is some song begins to converge towards what they heard. In fact, we now know that if they hear two or three, what we call tutors sing early in their lives, they'll oftentimes converge on a song that's a little bit like both of them, a mix of them. But they also have a song that's variable that changes from time to time from how they had they say may shift. Different you might say different motifs and different tunes get moved around with respect to each other sort of theme and very variation like effect, whereas the wild species is very stereotyped, Terry stereotypic, in the way it produces its song.

Nick Jikomes 1:00:40

So the species was being bred for coloration, not for song, but not song nonetheless changed that, but

Terrence Deacon 1:00:46

the song changed. So the the argument that I've made here is that if song was being used to do two things in the wild, but song does is it tells potential mates, I'm of your species, I'm not some other species. That's an important piece of information. The second thing it does is it exposes you to predation, kind of be a little careful about it. But in effect, again, it's this trade off between predation and showing yourself off. Again, typically males are producing this. And maybe if you're a better singer, you attract some females that you wouldn't do otherwise. Or if you sing more, if you're less stressed, and you're singing, maybe you'll do better. But of course, if you think too much, you might give yourself away to predators. So there's a bunch of trade offs here in the wild. In captivity, singing plays no role in mating, in captivity, if you're being bred because you have a certain coloration pattern, with another individual as a certain coloration pattern, that the breeders are uninterested in your song. And who thinks what plays no role? Basically, singing is now in a sense of relaxed, there's no, there's no sexual selection, on singing. In this case, if there's no sexual selection on singing, now, whatever it contributes to singing, whatever genetics contributes to singing, are relaxing their selection, and said over time, they can degrade. So my hypothesis was, and we it's this has now been tested out by another group in Edinburgh years ago, by just a simulation, not by actual looking at the genes yet, we as I said, we don't yet have the definitive genetic answer to this one. But what we can now say is that, that if this relax, that the template you have for what a soul ought to be like, is relaxed. Now, the only bias for what a sound ought to be like is what you heard your experience. But because it's not genetically pre programmed, there's lots of ways that experience and learning can now play a role that they couldn't have played before. Because now what you hear how social social you are, how you interact with other individuals, how interested you are and other songs, all of these things can now play a role in what song you ultimately sing. As you relax selection on just one part of the brain that's involved in this very, very, you might say, very strongly innate song structure as its relaxed. Now, many other aspects of the nervous system become potential biases and what song you produced. So why is this not clear is that in the bengalese, Finch, much more of the forebrain seems to be involved in song learning and final song production and playing with songs. And in this respect, but what's happened is that for the most part, we now see a transition in just 250 years, by virtue of relaxation, from a relatively innate form of communication that is passed on genetically to where the genetics degrades, and now song has passed on socially. But notice the analogy here to us. We need to hear others speak in order to pass this on. One of things has happened also is that in the bengalese, Fitch many more parts of the forebrain seem to be involved in solid communication, even though it's it's not playing a very critical role in terms of who reproduces with whom. vocalization in almost all primates that we understand is, for the most part, not cortical either. It's handled by deep brain structures in the midbrain and the brainstem, in part, but also in the basal forebrain. There's some areas that are involved, almost no cortical areas are involved. Language is almost exclusively controlled. We've got a lot of control in deep parts of the brain. But language recovery requires lots of the brain to be involved. We used to think it only involves a couple of areas and left hemisphere, Broca's area and Veronica's area. They were called. We now have a pretty good sense that a large fraction of both hemispheres is involved, particularly in the semantics aspect of language. So what's happened is that vocal communication in our species has now been shifted from a very few areas deep in the brain being distributed over a large part of the cerebral cortex. That's something similar to what we see in the bengalese Finch example. Obviously, there's major differences. One difference is that the song that's produced by bengalese Finch doesn't necessarily mean anything, right? Whereas language, of course, has this whole other symbolic aspect, it refers to things, and it can refer to things by its combinatorial. So language is very different in this respect. But in one respect, this might tell us that a relaxation of some innate communication might have played a role in our own capacity to communicate as we do. That is, it allowed this diversity allowed the play of cortical areas to now begin to play some role, and be now undergo also subsequent selection for the combinatorial interactions and so on. So I'm not making the claim that, that we just had relaxation of selection to get here. But as my sister factor population selection, probably played a role in this.

Nick Jikomes 1:06:26

Yeah, I mean, some of the analogy with the finch stuff is striking in many ways. It also starts to get you thinking about, you know, when you think about traditional or classic theories of language evolution, it's like why language is so you know, people like Chomsky, were basically saying, as I read him, while language is so complicated, how on earth are we going to explain how this evolved? In the fact that it's learned, like, Gen, each generation, each generation picks it up? De novo, essentially, you know, his, his ultimate answer to that was basically, well, the whole thing must be genetically pre specified. And this is offering a very different kind of answer.

Terrence Deacon 1:07:04

In fact, it's, it's almost the inverse, right? On the other hand, but we can so think about what what gets selected and what doesn't get selected. Selection requires that something is done the same way generation upon generation are done in a very similar way. He requires that a similar part of the body or the brain is involved generation after generation, that kind of redundancy is necessary for selection to act on a particular locus in the genome, because it has to have that kind of redundancy and repetition for the statistics of selection to work. But if you think about how words refer to things, we don't have any innate words. Yeah, yeah. If Chomsky theory was right, and we had innate grammar, why don't we have innate words, they say, Isn't that simpler than grammar, we don't even have innate words. And that's because the reference of the sound to a thing in the world is so disconnected, then it can be changed so rapidly, and we know even in existing languages, within a few 100 years, reference changes pretty, pretty radically. And within a few 1000 years, we've got languages that they can't be interpreted with. The same, they are so divergent. So first of all, there's probably not the kind of selection on language effects. But there will be selection on learnability and soci ability, there will be selection that favors the kind of attention we have the kinds of ways we produce sounds, because those are going to be done the same way generation upon generation. So there's a little bit of noise in the background. And if it's not disturbing this,

Nick Jikomes 1:08:44

no, I can barely I can barely hear it. All

Terrence Deacon 1:08:47

right. So the

Nick Jikomes 1:08:51

but I was just gonna, I can't really hear it hear anything. Hey,

Terrence Deacon 1:08:54

can you hear the humming in the background? Not really no, outside running a machine? I don't want to disturb what we're talking about. No, I can, I can't even hear. Okay, good. So, the issue is that there are many things that can be selected with respect to this in terms of language. But the fact that it has to be done socially, that in order to work language, we have to, we have to have acquired during our early development, the same pattern of interpreting sounds, we have to have this, this shared habit of interpreting sounds the same way. In a social group. That means we have to be good at social interaction, we have to pay attention to each other, we have to want to mimic each other to some extent. All of these are things that are going to be selectively favored. So we can see there's going to be a lot of selection for the biases to make this easy. But the innate part, the structure of language, and I think that even the structure of grammar and syntax, are very unlikely to be driven genetically. In fact, I have a whole other way of thinking about it, because I think that most of these, what we would call universals in language. I used to think that that language couldn't have universals because of this issue about selective capacities. You know why some things can be selected and not others? grammar and syntax, which is so complex and so distributed? Seems to me the last kind of thing, if you know, if that was innate, there ought to be a lot of innate words needed, or the case. And yet, there was a lot of syntactic and grammatical similarity across languages. Things like words, all languages have things like that play some role, like a noun or a verb. All languages have things like sentences. This is these are very common features. Why are they common? I think they're common, not because we have a sort of, we've inherited them genetically. Nor do I think they're the result of cultural processes. And I think that we've gotten trapped in this idea that there's only nature or nurture, or the interaction of nature or nurture. I think, in fact, that grammar and syntax have universals need for because it was neither nature nor nurture involved. I think this has to do with the structure of communication itself. And I think the best example of this that we know of, is mathematics. In mathematics, we assume that somewhere on some planet, circling Alpha Centauri, if it was ever possible, that they were intelligent beings, very different than us with a very different evolutionary background, we would assume that if they figured out how to work with quantities, that they would know that you can sort of now you can add recursive additions, you get multiplications, and recursive subtractions. And you get divisions. But there will be some numbers that won't be divided by anything but one in themselves. primes, we assume that every society, every culture, every intelligent population, and the universe will eventually if they figure out how to work with quantity, we'll figure out that they're things like prime numbers, prime numbers will be discovered. We don't say that people invent mathematics, they say they make discoveries in mathematics. And they make discoveries in mathematics, even though the mathematical tools they use are culturally created. Our variable decided, you know, we're going to use some, some marks for some things and other marks for other things. We're going to make those but but we discovered that zero is useful. We discovered their prime numbers, we discovered that we can come up with calculus, these are discoveries. It's neither nature nor nurture. It does require that we have these capacities and mental capacities to think in these terms. I think symbolic capacities are important here. But the discoveries because they're there in the world, and this has led many, I think mathematicians to think sort of platonically like there are these ideal concepts out there. My argument is that it has to do with with what's necessary in order to communicate quantity. And manipul manipulate quantity without confusing quantity. We can't have one equaling five spontaneously because we've done a bunch of manipulations, we've got to keep number separate. So we have something in mathematics, that is neither nature nor nurture, we discover it, because there's certain ways that you can manipulate quantity that maintain consistency. We have certain ways that you can manipulate symbolic communication and maintain agreement of reference. I know that I'm referring to things and I do so because there's got to be certain combinations have to happen. What are my favorite examples is why do we have senses? Why do we have things like verbs and nouns? Is it because they're there things in the world like verbs announced by this sort of their changes and actions and things? But it's not quite that simple. On the other hand, think of a word like loud, a word like loud, soft, noisy, cacophonous, you know, it brings up a lot of associations. But it doesn't by itself refer to anything in particular. But if I do this that was loud. I can say clapping my hands can be loud. But now doing this and saying clapping my hands. The noun part of that sentence could have been replaced by what we call an index. Something that is correlated with it, the sound is correlated with it. What we see is that nouns play a kind of fingerpointing role, they point to something that the verb will modify. In order for symbols, which are abstract and refer to other symbols, like loud, referring to soft to refer to something in the world, outside of this referential system of symbols pointing to symbols pointing to symbols to refer to things in the world, or to elsewhere in the communication that we've given, we've got to have pointers. Now serve as pointers. This is why we can use things like this or that that's loud, or this is loud, they just point to something. And not surprisingly, we human beings start our lives for the first years of our lives, with a sort of built in pointer. Yeah, we point to things. And it helps us acquire this. I think that in effect, what actually is happening over the first year of life is we're developing a lot of sophistication with this using this relationship of pointing of indicating that ground symbols in the world. So we hear words, and we point to things we hear words, we point to things, and it allows the first during the first year of life, us to begin to acquire this capacity. I think this is why it's so important that we have this curious bias that other species don't have this, we can use our hands to both manipulate things, but also to indicate things and children are just well divide device to do this. And this has been another I think selection, we've been selected to do this. Notice that also there are disturbances in human development, particularly in cases of autism. Although autism is a broad spectrum of things, a lot of autistic children do two things that are very atypical, and extreme autistic cases, they don't pay attention to, to where heads are looking for, or somebody is going to things how eye movement is working, they don't tend to point they don't tend to indicate and to follow others indication. This makes it very hard to acquire language. If you can't start out with his bias in which you, you learn how pointing works, you have to learn how indexicality works, it's going to be hard to get into language, you're going to be a late developer, you're going to use language differently.

So we already know that this is a feature, a clear feature of language learning, we we take it for granted with young children, that you know, this we point and we point they point we look at things they look at where we're looking, you know, this all a part of this has been brought out by a number of researchers a part of this process. But this ability to link symbols to indices to get them to ground. That reference in the world is neither nature nor nurture. This is necessary if you're going to communicate symbolically, it's built into what we call the semiotic infrastructure of language, you have to have this capacity. And there that's going to limit things. So to clap my hands now, and then five minutes later to say loud, they're not connected to each other. They don't, they don't indicate each other as well. They're not coupled. And as a result, the link is not there. So it also produce all kinds of other constraints on grammar and syntax, things have to be correlated with each other at the ends of, of nouns have to describe multiplicity, like plural, has to be attached to the noun, or have to be nearby the name. The past tense in verbs has to be attached to the verb or nearby, the verb has to be a modifier, and a modifier has to be next to it. Like the clapping was next to the sound loud. There are a lot of constraints that are now going to be built in. I think what we've ignored is as how many of these constraints are neither nature nor nurture. And that as we've developed language, as its elaborated, become more verbal and less pointing. We've had this sort of build more and more of those constraints. And no matter what language you speak, no matter where in the world you are, it's going to be different, done differently. But those constraints are going to be there anyway.

Nick Jikomes 1:19:32

Is that a fair way of summarizing that, that, you know, when you say that this is not a matter of nature or nurture, you know, the rules of grammar and the structure or language? It's not out there in the sense that you know, there are planets and rocks and mountains out there independent of us, but it is out there in the sense that as social creatures, there are just fundamental universal constraints on how we can intelligibly communicate with one another and that's, that's what we're seeing. Right

Terrence Deacon 1:19:58

and these are these are going to be as nice It's not about being a person. But it will be about being communicating with symbols. Once you communicate with symbols that have been ungrounded, that is the sounds don't are not like anything that they refer to, there's no likeness, there's no necessary correlation, even most of the words we've been using are not correlated with anything in our immediate surrounds. Those are all broken. Now you got to reintroduce those connections in different ways. That's just built into communication. The key here is that that's in the world, it's like gravity is in the world, we can take advantage of it because it's in the world. It's just there. This is just there, but not there unless you've developed your communication to a certain level. But so now what this does is it sort of inverts the whole Chomsky and like story, which says, Know, what we thought was the most innate part is, in fact, neither nature nor nurture. I mean, one of the arguments over the years has been, you know, maybe culture is what creates these universities, or no culture doesn't create these universals, culture is just sort of taking advantage of them there that are there, they didn't have to be there. Innately, they didn't have to be there in culture. So it's not a it's not an either or, or not a combination of the two. But you need both you need biases, and learning biases, and indicating any cultural tendencies to sort of allow us to work together to want to communicate to each other that wants to know what each other's thinking, and so one, all those things have to be there as well. So there's got to be all of those things that work, to then take advantage of these biases that are built into the problem with communication. So this has an effect, not just inverting the sort of innateness story, it basically says, oh, there's a lot of innate stuff. And there's a lot of cultural features. And they've they've got to be interacting, but they're taking advantage of structure that built into the very nature of communication itself. And that's the neither neither nor part of the story. That it actually is changing the debate totally. And, unfortunately, I think we have not come to the point where we can sort of do that parsing well yet.

Nick Jikomes 1:22:15

How do you so when we think about, so we can think about this, over the course of an individual's development, we can think about, say a one year old, who's really good at indexing and pointing into the world. And that sort of physical indexing capacity is then used to build up the symbolic capacity, and the more abstract parts of language and communication. When we think about how this evolved in our species. You said before that you think there was maybe a period, an extended period, you know, perhaps a couple million years, where we weren't using language in the sense that we, you and I are right now. But there was something like a proto language. How do you think about that arising? Is it is it something that doesn't have to do with say, gesture based communication, and simple rituals around? Interacting, you know, whether it's like in a meeting context or a feeding context? How do you think about the transition from language back to proto language and whatever came before that? Well,

Terrence Deacon 1:23:12

so let's start with before that was one of the things that I've done in my work is to begin to understand connectivity in the nervous system. One of the first questions I was asking, had to do with, you know, how is it that we're producing these sounds? And the answer is that most other species, particularly of mammals, don't have the ability to carefully time the production of the tension of laryngeal muscles and the production of breathing. Most innate vocalizations think about laughter and solving two of our innate vocalizations. They don't involve articulation. So when I laugh, my mouth is open. My tongue isn't moving around, my lips aren't opening and closing. It's a vocalization that's produced only by the autonomic nervous system that is, is produced by the fact that there's this automatic system that says I'm going to tie my breathing to be spasmodic and I'm going to open my mouth and tense my my larynx in timing with this. Sobbing another one. Again, I'm not articulating it. My mouth and tongue aren't moving in any kind of timing with this, this sort of open mouth grown scream, no articulation, this is typical of primate vocalizations. primate vocalizations are not articulated, the one close exception are given calls, which have a kind of that like is it what happens is it is a slight opening and closing in about very minimal again, no tongue articulation associated with it. So that's as close as we're gonna get to pretty complicated vocalizations. The other thing is that most primate vocalizations also involve sound production on the outbreath. And in breath. For us, probably the only case of this is in sobbing. Think about this in a sob, you know. But we make sound of the Enbridge. And notice that to make a sound on the in breath is painful for us. It's really difficult. All vocalizations, almost all human vocalizations with respect, with the exception of sobbing, are done in the outbreath. And in fact, we now know that there's some structures of the larynx that probably make this in breath sound, what we call engrossed in Grest. Sounds very difficult to make. There's a little bit in aggression in some some languages, it's very, very minimal. Now, so that's one thing. The other thing is that this separation of vocalization is very important. Because if we had a lot of conscious control, if most animals had conscious control of where their larynx was tensed or not, there would be a problem with eating and drinking. We have that problem because we could choke because we get with talking while we're eating and drinking. And we can and we say it goes down the wrong way. And it's very uncomfortable, very painful when a little bit of water or food gets stuck down there or it can be dangerous, can be deadly. Which which case the Heimlich maneuver is necessary to sort of dislodge food in the larynx. Now, that's because of course, these two pathways cross over each other. But normally, if you've got a system that says, Keep all vocalizations, automatic, and innate. And when you're eating and drinking, shut that system off, and have swallowing, and chewing and articulation of the tongue and so on. Being automatic also, but completely separate it, then we're safe is very unlikely that you're going to choke. There are other adaptations in terms of positional larynx, the way the epiglottis moves, and so on, that are also going to play a role in this. But the real challenge is, how are you going to be able to precisely time the movement of my mouth and tongue and lips, and jaw with breathing and with tension of the laryngeal muscles. The key is there has to be cortical control of the larynx. But to keep the larynx and breathing automatic, in other species, there is no cortical level projection down to the nucleus that controls these systems. It's called the called the nucleus ambiguous, ambiguous because it's hard to find in the brain, but it's there. And what we know in other species, and part of my work in the distant past, now was doing to try to find if there's any connections like that, from the cortex, projections that go right down to control this nucleus, which controls the muscles, the intrinsic muscles of the larynx, and some of the muscles of breathing, like, like the diaphragm. But what we know in us is that we can do this, the very fact that I am articulating, moving my mouth and tongue very rapidly in producing all of these sounds, and timing it precisely with what I produce a sound from my larynx or not. And the fact that I can sing that I can raise my tone of voice, or lower my tone of voice, by changing the tension on the larynx, independent of whether I'm aroused or not, tells me that there is a significant in humans a significant critical control of the lyrics. But this means that we can now serve we have this problem, you know, if I'm, if I'm speaking and suddenly, a very humorous thing happens, it oftentimes my laughter erupts my speech, I can't suppress it. Thesis systems are now potentially in competition with each other, because we have in humans superimposed a cortical connection onto this once automatic system. That's something that actually had to happen against all the potential costs of having this competition possible. So over time, this new connection pattern in which the cerebral cortex is now controlling laryngeal features and breathing so we can time our breathing so we can hold our breath, so I can tie my breathing so that I finish the sentence before I have to take a breath. All of these things have to be now controlled cortically and therefore learnable. And controllable in a complicated combinatorial way, those connections were there. And our ancestors, almost certainly were not there in the Australopithecines, were probably communicating like most other primates do, and probably had to develop over time. I think we talked earlier about this sort of gerrymandering notion that if, if certain areas are larger, they have a better competitive capacity. I think, in part, the large enlargement of the forebrain helped increase competition for synapses in the nucleus, ambiguous. cortical output now could outcompete other systems and stay connected to the nucleus ambiguous in our species, in part because this this cortical enlargement, I think that played a role in it. But what that meant was that there's a cost. It's changed the number of ways we can vocalize because, you know, when we're when ingressing sound, producing sound by breathing in articulation doesn't play much of a role. You know, my, what I'm what I'm trying to breathe in, I can't even do it. In fact, you just don't have the ability, but by producing sound out, and then moving your mouth and tongue and lips. Now you can produce a lot of diverse sense. But you've got to coordinate all these systems that I think took probably a million million and a half years to develop, I think it was probably well developed. Even in our ancestors. And our, you might say, our, our parallel lineage, the Neanderthals, and the Denisovans, that were around over the last 100 100,000 to 200,000 years. I think they also have these capacities, I think it developed early on. But this means that it couldn't have been the first thing that developed. And it probably developed along with enlargement of the brain, compared to the standard, because that's what will have determined how much of the cerebral cortex was controlling laryngeal activity. So I think, actually looking at how brains enlarge over the last 2 million years, it's probably giving us a sense of, of how this competition in which the cerebral cortex begins to control vocalization develops. And so I think we can see it as developing slowly over the last by 500,000 years ago, where we see brains almost as large as ours, I think it was probably well developed by then. But certainly, in the first, you know, 2 million years ago, as it begins to develop as for brains begin to develop with respect to this area deep down in the brainstem, I think we're beginning to see this asymmetric asymmetry developing. So what that suggests is that early on, symbolic communication was almost certainly not spoken.

Nick Jikomes 1:32:56

By gesture based or something to

Terrence Deacon 1:32:57

be more gesture based, it maybe was more ritualized gestures, interacting in some ways, one way to communicate symbolically, might be in effect, to make sounds, simple sounds, but associated with movements. If you think about the young kids who are just learning language in the first year of life, they do a lot of gesturing and kind of stuff. I suspect that that's not a bad model of what early language was like, it's, you're now coupling, this capacity. But the other thing that we see in young children is babbling. They produce, we're already big, they're already experimenting with this capacity to use the cerebral cortex to play with sent young children have this I don't think babbling was was an issue in our early evolutionary past. But maybe playing with ritual playing with, you know, acting things out collectively. I think that the auditory vocal part of symbolic communication, developed slowly over time. But now, why did it develop? I think there's a number of interesting reasons. Number one, it has to be learned if we're going to communicate symbolically, we have to learn how to do it all the same way, interpreted the same way. But in sign languages, for example, there's a symmetry with which the hand does which. If I'm to learn from you say, Here's my, here's my gesture. Like that. It's easy to learn it in mirror image. But we have to sort of invert it and do it with my right hand, not my left hand. But looking at me, you would be doing it in your left hand and I'd be doing it in my right hand. So there's asymmetry in terms of any kind of visual communication But if I do this, if I do this sound, PA, notice that you producing it doesn't require you to reverse positions, you don't have to take my position, you just have to hear it and produce it. Sound does not have this sort of spatial inversion is mirror image inversion that has to take place. So to learn to communicate visually, is more difficult. Learn symbolic communication, arbitrary movements of the hands or body requires, we have to automatically invert. Now, I think that that did happen early on, but it was not easy. I think it's one of the reasons why we do copying pretty well. And why we're interested in sort of taking another position. Another point of view, I think this this is makes that kind of thing easier. On the other hand, to shift to the auditory verbal mode of communication, bypasses that problem. That means it's going to be much easier to acquire copied sounds, and to produce many, many more variants easily. So I think that part of what was happening was also, the auditory vocal mode of communication is just easier to develop socially, to pass on, and to acquire many more variants very rapidly. And so it's clear that we've we still use pointing, we still are involved in gesture, gesture plays a major role in communication, face to face communication. And particularly, it's never missing from more linguistic communication. But linguistic communication can happen without it. And, and we can acquire it rapidly without having to look at each other, just to listen to each other. And, of course, we can produce much more rapid sound, as we can produce words in a very rapid rate. Because it's just articulation and sound. So it's a very complicated story. And I think one of the problems we have is, we've still been using simple stories to talk about the evolution of language, it involves development, it involves pointing, it involves neither nature nor nurture, involves social domestication. So we have to be self domesticated, to be in large enough groups that are stable for a long period of time. All of this stuff is a very complicated story. So that the human evolution story has to involve all of these features, epigenetic features, neither nature and nurture formal features, learning abilities, biases, in terms of attention, all of this stuff. I think we're not nearly as complicated in our thinking about this problem is we need to be and I think that's the future of the field, why this is such an exciting area, is there's so much to study. And this will also mean that there's so many other factors involved. And one of the other things that's important about this is once you have many, many constraints, affecting it, many, many biases distributed in many aspects of our nervous system. Now, some can fail, we can still do well, we have this meta stability, in a sense, because there's so many things contributing to it. Now, the redundancy story comes back. Because we've got many, many ways to acquire this. We can afford to have some of them fail, we can afford to have some of them not work. In the same way that that New World monkeys which have sometimes color vision sometimes don't, nevertheless, have other adaptations, that sort of take take into advantage take into the advantage of the fact that they can't see color change. But they can see other things, they can make other distinctions. Some redundancy in the system. And my guess is that, again, this brings us back to this sort of inverse Darwinian story that we need to keep track of that redundancy piece as well.

Nick Jikomes 1:39:21

When I think about when I think about the idea that that we have constructed the social niche for ourselves, that's allowed us to self domesticate, when I think about things like the, you know, I think you can describe language probably as an ornate phenotype, you know, and naturally that will get you thinking about sexual selection. So when I think about you know, natural versus sexual sexual selection, when I think about the constraints that are relaxed by having the social niche we've constructed for ourselves, what and why also think about sort of proto language and the earlier steps in language development or the the evolution language in our species, you know, I start to think about, okay, well, what are the other rituals that are universal, like languages across human cultures that also tend to be plugged right into this symbolic cognition capacity that we have. One of them that comes to mind is, you know, of the rituals that I think pretty much are universal across cultures that we have around mate selection. I think about a wedding, for example, there's explicit symbolism. It's obviously to do with reproduction. And I just wonder, what what do you think the other sort of ritual behaviors that were in that were contributing to the evolution of proto language? Actually were? Do you think they had to do with things like marriage and reproduction? Do you think they had to do with things like hunting and food acquisition? Some of the above all the above?

Terrence Deacon 1:40:51

Yeah. First of all, some of the above all the above, yes. My own work good dating back to the mid 90s. I was arguing that actually, the reproduction problem was a very significant one, it's in fact been taken up by there's been called deacons paradox, the paradox is, is this, that if you need to cooperate very closely and trust each other to gain food, so it's a bunch of males out there on the savanna, chopping up a carcass, and having some individuals fend off the predators and the scavengers, that could do harm to you, you then have a problem, you first we'll have to trust each other you say so intense competition can't be too strong, you've got to trust those that are protecting you. You get back to the tree, so to speak, where you can climb up the tree and, and eat safely? Well, one of the one of the problems is that now somebody's gathered the meat, but others who didn't need it. So if there's competition over resources, that's going to make it hard to get meeting tomorrow. Because now either you have to change roles all the time, in some systematic way, or there has to be an agreement about sharing. But this also goes along with offspring, and mates. So now, first of all, if the food you're getting is gathered in a dangerous environment, the worst thing to do is to bring a lot of newborns and young toddlers into that environment. Yes, yeah, the individuals that will wander off, will get picked off by predators and so on. So one of the things about going after a, a food source that potentially has dangerous associated with it, is it's going to sort of drive a little sexual division of labor. So mostly probably agreeing mates. But now here's the problem that males are also going to be in competition for, for reproduction, males are going to want to compete, as they have in most primate species. For mates in this respect, they compete by either chasing each other out aggressively or threatening in some ways, or more likely, being the only male or the alpha male in the group, in which no others are willing to compete with you, or you know, vie for access to females. But now you've got a problem, because now you've got both the sharing of resources problem, you've got the trust problem, you got the cooperation problem. But now, if you're in competition for mates, if I'm the I'm the I'm the guy out there protecting the one who's getting the meet, maybe I'll just sort of look the other way. And then I'll have access to more mates because that individual gets injured. Yep. So the Trust has to be there. But that means that in effect, there can't be a lot of competition over mates, the main competition has to be reduced. In order just to get that food resource, which is incredibly valuable, and inaccessible to any other species. You now need to have some way to sort of reduce competition for mating. In order to have this kind of trust, competent trust cooperation built. My guess is that that's going to require socially not only exchanging the resources we get from hunting, for hunting, but from scavenging and to share it with females and their offspring. But also, you're going to have to Innocence now make it so that there's less competition that one way to reduce the competition is to have exclusive meaning, or more or less exclusive meaning where some individuals are not going to be able to mate with other individuals and vice versa. If everybody is mating, then at this you're decreasing that sexual selection on mate competition, but how do you stabilize that? One of the challenges here is is this is of course, lots of money. literature is about literature has always been about this kind of competition among males and, and you know, infidelity and all this stuff. Yeah, this is the this drives social social systems today even the one thing that we find is that typically, these are these are crimes, these are things that we treat, as somebody has broken the rules. There's we've got to have social conventions, symbolic social conventions, that basically say, No, you can't do this, you can't do that. So think about how most I will call a marriage but most meeting, we set up in most human societies, marriage is a way of organizing meeting and organizing responsibilities around offspring care, and so on and so forth. And, and resource provisioning. What happens is typically, you get the marriage activities done, first of all, there's a bunch of symbolic activities that's about overtly promising. You know, it's, it's not just saying I do, and I will, and I won't, but in fact, demonstrating in some way, oftentimes, a marriage ceremonies involve some kind of marking, you know, carry a mark on the body, a tattoo or, or a scarf occation, or a ring, for example, something that will be permanently attached to the body that says, I've, I've done this, it's an index using our RS semiotic terminology, it's at the index like a signature is, it's attached to something. It's

Nick Jikomes 1:46:39

very public. And back in these times, it's not just your friends and family at the wedding, it's every human being you ever interact with?

Terrence Deacon 1:46:47

Yeah, that's right. Why, because you've got to be able to demonstrate, you know, who's gaining access and who doesn't have access, these are people who are likely to be both your supporters, we're going to find out if somebody's cheating on you, but also those individuals who could potentially cheat on you. You got to get everybody together. And you gotta get two families together to lineages together, because they're gonna hold the other lineage responsible for certain things. And then you do all this in the context of all this, that means you've got to have a way of communicating collectively, everybody's got to interpret things the same way. Or at least similar enough so that you can trust it. That I think is a great context. For developing symbolic communication. It's highly ritualized, it has to be distributed to a large group, you have to be able to trust that everybody interprets these actions the same way. The actions may have to enact something, and then show it to be blocked. So I think about a classic way of doing this might be what if we don't have length, we don't have any way to talk about this, maybe to have somebody act out, meeting or act out something and then have it interrupted or blocked. Now we actually have a way to say no, that's not going to happen. Everybody sees that this is not going to happen. Maybe everybody, everybody who is possible has to be involved in this sort of interaction play. We can begin to replace that maybe with a pantomime activity, a shortened version of it a compacted way of demonstrating that activity, or maybe a sound associated with it. I think that's the ideal context, I think, in which symbolic communication was probably first developed. It's developed there, because there's the most intense sexual selection threat associated with this, the most intense cooperation threat. That is, if somebody cheats, cooperation is off, nobody has resources, everybody loses. So now you got a collective interest in keeping it going. A collective interest in people, not just yourself, but others, your family members, your friends, who let you know that somebody is cheating on you, or otherwise stop the possibility of somebody cheating. This is a context in which I think most of those features are intense, plus the sexual selection features are the most intense. That's where I think this kind of communication probably evolved initially. And then, of course, expanded into many other domains. There's almost no domain in which symbolic communication with as well developed actually is advantageous. The question is under circumstances in which symbolic communication is difficult to acquire because you have very few adaptations for doing so. When would it most likely appear, it's going to appear in which the constraints are most intense, in which the cut the competition for keeping it going against intense tendencies to break it up, are our most intense, it's not going to be where it just helps things out. It's going to be where things are really intense. And I think this is where it's going to be most intense.

Nick Jikomes 1:50:28

What what I mean, I think you were kind of just getting to it, but like, what are the what are the features of language that we find attractive inmates? Does it have to do with the information being communicated? Does it have to do with you know, how things are being said it you know, when you think about if we just think about humans like zoo as well just think about other animals? When we look at how humans use language in mating rituals? What is it that males and females are each picking up on in the language use of potential mates? Is there a clear answer there? I

Terrence Deacon 1:51:00

think there's not directly although Darwin thought that that language evolved sort of like birdsong does, in which language was speech was a an advertisement. My suspicion is language by itself, probably not. But notice, think about song. The signs, don't argue that song preceded language, preceded speech. But in fact, when you listen to popular songs, most of them are about something having to do with mates or love or infatuation, or, or being cheated on or whatever. We still gravitate towards singing in which the content is about meeting in some respects. It's not not a surprise, that song also that music communicates emotion. What one of my favorite philosophers in the past as well named Suzanne Langer wrote a book called philosophy and a new key plus a lot of other books on on symbols and things like that. But But what she argued is that music is in a sort sense, a communication of feeling. In fact, she wrote a book called feeling and form in which she was basically saying that music is a sort of, it has a structure, that sort of like an emotional structure, elation, depression, excitement, resolution, musical tones, musical patterns have that feature now, it shouldn't surprise us, because, of course, most emotional communication

are produced by tonality, this is something we inherited from our primate primate ancestor. So changes in tonality in excitements in all kinds of attentional features and so on, are going to be communicated by change in tone. Notice it in language, also, what we call plus prosody in language, that is, the way I produce the words I'm producing, you know, that the tone of it, and so on, it can it can actually produce the emotion of what I'm saying, in this process. That's because we haven't lost those primary calls those those tendencies towards it, nor the intensity to to interpret them emotionally, they've noticed become, in a sense, subordinated to our symbolic communication. But I can amplify it in terms of prosody, but I can amplify it even further in terms of just music. And so it can now it's carrying a lot of that emotional communication. And it shouldn't surprise us that the major content of music has something to do with sexuality, particularly in popular song, particularly in music that's, that communicates to younger people. It's almost always about that. On the other hand, we do have some music that communicates over things like millet, Milton, music, music, about anger and so on, they get a little of that, but, but particularly around the time when, when people are more focused on meeting, you know, adolescence, and late adolescence, yep, music is all about sexuality. It's all about mates, all about falling in love all about who is a wonderful lover this, you know, she's so perfect. He's, he's so handsome. He's so wonderful, like all the nice things. And of course, in our species, it's not just being dominant and beautiful. It's also being a good caretaker. A reliable person, you know, I'll never cheat on you. That kind of communication we don't trust Just in speech, but even synth music, it sounds a little more influential. So I do think that there is a major feature of this, it's still with us, we've not stopped being primates, we have not given up the primate vocal call system, we have it there, it's just been reduced significantly, and made very distinctive by being incorporated into language.

Nick Jikomes 1:55:25

Another feature of language that's super interesting is that language itself evolves. It clearly evolves faster than, than biological creatures evolve, you know, orders of magnitude. Yeah, much faster. And, and, you know, you've written that there are basically that language is subject to selective influences, that are probably quite alien from those that affect our brains and bodies. So what exactly do you mean by a statement like that? Well,

Terrence Deacon 1:55:55

so number one, language learnability is going to be an issue. languages that are learned at a young age, and learned well, are going to be passed on more effectively, they're going to be more stable over time. So there's particular aspects of language, particular, phonetic structures that are difficult for young children to acquire, they're probably not going to be passed on Well, certain combinatorial effects, syntactic effects that are quite complex, that are not, you might say, intuitive, will not be picked up early on, we have to wait until sort of later child childhood to get these complexities. But as a result, they will also not be passed on as well, in part because early on the brain is very plastic. And what we acquire, acquire early on, tend to be more easily automated, and harder to get rid of. So in terms of if we think about language evolving, it has to have a reproductive component, and then it's got selection on it. Those things which are reproduced, more stably, and with with more fidelity over time, that means they're going to be more learnable. So I think what has happened is that languages have evolved and have to evolve to be more learnable at an earlier age. So one of things that we find is that children can learn multiple languages easily if they're exposed to multiple languages at a young age before their age three. But only a few of us do pretty good at learning multiple languages. Well, after that period of time, this is telling us that languages to some extent, language evolution and language change. And therefore, language acquisition is going to select for languages that are easily learnable and learnable. At an early age. So the structure of language evolution has got selection features, we're selecting our nervous systems, our development, our ways of communicating are all part of the, you might say, the environment in which to which languages have to adapt, one things they have to do is they have to adapt to brains, and how primate brains now, human bias brains still acquire this. But in particularly, if they can do it to younger brains that don't learn quite the same way that older brains do, that have certain disadvantages, and certain advantages. Those languages are going to be passed on better. And those features of those languages are going to be the languages that the features that most languages inherit. So I think language evolution has to be thought of, in terms of, it's evolving to us to be adapted to us. It's not evolving, because it's an organism, but it has many of the same characteristics.

Nick Jikomes 1:59:03

Yeah, in many ways, Bay is sort of like an organism, even though it's it's not,

Terrence Deacon 1:59:09

not clearly not an organism. And of course, the interesting thing is it's distributed. It's not just in me, you know, it can't be just in a single individual, it has to be distributed. So it has a lot of interesting statistical features that will have to do with social groups, and how groups form and break up. That's going to affect it. So languages have to have this capacity for conversation, for asking for telling, for explaining in all the features that languages have to do also with the fact that they're, they're a social phenomenon. Those that's part of the selection, the selection, the environment in which languages evolve.

Nick Jikomes 1:59:52

Given that, you know, Lang were given what we were talking about in terms of how language evolved the time Types of the types of things people are most concerned with that they use language for around mate selection around trusting the the individuals you have to interact with, to, you know, stay stabilized the social structure, you know, we all are things that we were discussing. What do you think the consequence, the consequences of the global, our sort of globalized population structure is today? Like we don't we no longer live in bands of dozens or just a couple 100 people, where everyone you know, and interact with is literally at your wedding? How do you think that and the technologies we used to communicate today, you know, our smartphones and all of this stuff? Can you anticipate any ways that might cause our language to evolve in a different kind of way than it has historically,

Terrence Deacon 2:00:49

not sure, it will cause our language to evolve in different ways, because we're mostly using it in person to person interactions, even like we are talking today, across long distances, and, and so on. But I think one of things that will happen is that we're going to be shifting more and more onto social phenomena. So, you know, in the past, when you wouldn't want to make an agreement, I couldn't trust you just to say, you know, your words, just like promised to do this, let's shake hands, let's touch. And that means, you know, so there's, we've got an index here of agreement, you know, we're not fighting, we're agreeing. When you can't do that, maybe what you can do is have a different kind of index, I give you a piece of paper that has some things that you read. And they say that this is what you can do, and you can't do, you have to indicate that you agree by writing your signature on it, your signature says that I was physically there, me, I was physically there, exposed to this, this is an index, it's a much more powerful link than just the Okey doke of language, language is disconnected in a way that a signature is more connected. We're now developing ways of doing that at a greater distance. One of the features is that as we begin to offload those direct interactions, issues of trust, issues of cooperation are much more difficult to maintain. And we have to develop extra tools to do this. One of the challenges is that also I think the internet has done something else that we've constructed these systems based upon. You might say, a logic that's irrelevant to the communication. So for example, in order to keep people looking at a particular website, I've got to make them like it. I've got to make them, you know, find all this next piece of information is more interesting, a follow that. So just for the financial structure of creating these systems, we've created certain biases and transmission, we've now noticed the consequences of these biases, that negative information, that surprising information, that that that threatening information, explicit sexual information, this stuff tends to be reproduced much more rapidly. So there's, there's a different selection pressures have become into play. And they come into play very rapidly, of what gets communicated. And that's going to structure things very, very differently. I think one of the things that, that I'm looking forward to, but not quite clear how it's going to affect affect us, is the way that that artificial intelligence is gonna allow us to communicate across languages effortlessly. That is simultaneous communication is going to be what everybody has available to them, as we have knowledge available to us by looking at our smartphones these days. All of that's there. But notice that this is going to get us back to where we started our conversation about redundancy. I start one of the chapters in my book by talking about smartphones, and how we've become addicted to our smartphones. I no longer know, the numbers, the phone numbers of people I used to know, I no longer know addresses, but I now have access to hundreds of more addresses and phone numbers than I ever had before. But if the power goes out, if the grid goes down, I'm really disadvantaged. I've become addicted at the same time. But this leads me back to a story that's as ancient as philosophy itself. In in Plato's Phaedrus. Socrates says that if you know men learn writing, they learn writing and reading is going to be and degrade their memories. They're gonna be stupid. Or maybe it's not a good idea. And yet, obviously, we've learned reading and writing. We now have offloaded a lot of memory onto this stuff. I don't necessarily know all the things that I've actually learned in classes, but I can go back to the books and find it. Button. Because of this, of course, I also know what Socrates thought or what Plato thought. Yeah, yeah, of course, the graded some capacities, but he gave us other capacities. This is that relaxation effect. There's been a relaxation of selection on memory. But it allows other things to develop. That's happening socially. I think it's going to radically in the next generation over the next couple of generations radically change the structures of society. I'm not sure it's going to do it in the in all useful directions. But it radically changed it.

Nick Jikomes 2:05:58

You know, a lot of people are now using and talking about and speculating about things like generative AI. So when you look at technologies like Chet GPT, and these large language models, we call them large language models, because we're feeding in human text made out of human language to them. Yeah. In what sense are the systems using or not using language and symbolic cognition?

Terrence Deacon 2:06:24

One of the things that I think is most interesting that's come out of this work is that we can also use large language models. Sticking in these letters, we have 20 letters for all the amino acids in a protein. We know the structure of some proteins, one of the problems is that we have not known for years and years how the sequence of amino acids cause proteins to fold. It turns out now beginning first of all, with a group called out of open AI Alpha folds, they fit in what we know about which amino acids follow which amino acids and how that causes a protein to fold. We know the folds of those proteins, some that we know, can we now just sort of train it on all those that we know and now feed in a random sequence of amino acid letters. And will it tell us how something folds? Well, it turns out that alpha fold, and now three or four different versions of this have used a large language model structure to predict how proteins forms. Now, here's the here's the important thing, proteins fold because of the chemistry, chemical environment, they fold because of hydrophobic and hydrophilic effects, because of stiffness effects, and so on. Because of the sequence of which they come out. Some are exposed to an environment before others are. It's a complicated physical chemical process. The large language models that predict protein folding, don't know anything about the chemistry or the physics. And yet, they're able using his huge statistical capacity to make good predictions about how proteins will fold. We can now use this statistical prediction to go back and figure out and I was we've got, it's like, you know, learn the math by going to the back of the book and looking up the answer, before I can figure out how to use the equation. That's what we've got, we've got the back of the book, there is no knowledge in the system. But the skeleton of the process, you might say the syntax, or you might even say the fossil of that information is there. It's structured, those physical processes produced structure. And now we can use statistics to look at the correlational structure, and then use that correlational information once we've got the structure to go back and try to analyze to learn. So this is helping the science. But now let's think about this the same analogy for language. We have terabytes of conversations on the web, that were scoured, to build into these systems, and then testing them on predicting how well it produced autocorrection kind of thing. The structure of that communication was on the web, was structured by people's interests, by meanings by references, by things in the world by science, by literature, the structure was produced there. That's the ANA analogy to the chemistry and the physics of protein folding. But notice that the large language model doesn't have any of that. It's not related to the world in any way. It's completely isolated, but it's got the structure that was there. I like to think about it as the fossil of the living speech, living communication. But if we've got enough fossil evidence We can reconstruct this. And now using our knowledge of language, and meaning and reference, we can now do what the biologists could do, looking at protein folding, and say, I can use that information. And I can reconstruct its meaning it's, in the case of the molecular biologists, I can learn something about the chemistry of it. In terms of URI, using this generative AI, I can learn something about the meaning of something that I might not have known. I can use it to create a Shakespeare sonnet, about planets and solar systems. Why? Because it's got that structure. But that structure, since it carries with it all the biases, all the structure and biases of communication, when I want new stuff into it, and have it come out, it'll come out in a way that's interpretable. It gives me information. So I think it's gonna be incredibly useful. But of course, as we know, it can be incredibly troublesome as well, for all the same reasons. But this is why these new technologies are going to be very useful in the sciences. Because they're giving us the answer before we know the details of the question yet. But we can use it we can retro dict, they can use it to go back and figure out okay, that gives me a clue as to how it was formed. Maybe I can use that. The problem is that we we tend to fail the Turing test here. We're so gullible. It produces something that looks like it understand something and we think, Oh, it's got understanding, it's got a mind. It's got sentience, it's got thoughts, none of that stuff. Any more than simulating the movements of planets. On a computer is gravity. There's no gravity, there's no mass. It's just the formal structure. But it's enough to do incredible things. We can use that formal simulation, to guess at new laws, we can use the simulation of protein folding in these systems to get it the chemistry and physics.

Nick Jikomes 2:12:14

Earlier we were discussing,

Terrence Deacon 2:12:15

by the way, I'm going to stop this here because if I don't break in five or 10 minutes, I'm in big trouble. Oh, I have a physician's appointment coming up.

Nick Jikomes 2:12:25

Okay. Yes, no problem. Well, Terrance, thank you for your time. I always love talking to you. This is the third episode you've done with me. I highly recommend for those listening, the first two episodes there about his books and complete nature and the symbolic species. We got into various subjects that tie into each of those. But if you're interested in this stuff, those are great places to dive in. And I'm definitely looking forward to your next book

Mind & Matter
Mind & Matter
Whether food, drugs or ideas, what you consume influences who you become. Learn directly from the best scientists & thinkers about how your body & mind react to what they're fed. New episodes weekly. Not medical advice.