281 | Samir Okasha on the Philosophy of Agency and Evolution

0:00:00.8 Sean Carroll: Hello, everyone and welcome to the Mindscape Podcast. I'm your host, Sean Carroll. Sometimes now, as a card carrying philosopher, I'm a member of a philosophy department as well as a physics department. People will say either in some natural conversation or just blurting it out because they've been meaning to, Philosophy is kind of useless, it's silly. It doesn't address problems that are real and important. It doesn't take into account data and experiment and things like that. Now, most of these objections are prima facie, not worth paying attention to. But what really makes me laugh is the idea that philosophy doesn't matter, that it's not important, right. That it doesn't affect how we go through our lives. The simple response to all of these objections is that you can't help but do philosophy. You can only do it well or do it badly.

0:00:53.6 SC: Take, for example, a very basic question, what is a human being? What things here on earth count as human beings? This is a very important question. It's important question for the modern world, not only because we now have artificial intelligence, and you might wonder whether something that was created by human beings in a computer or in a robot should count as a human being. But also, what about things that clearly aren't human beings, but might deserve some rights that human beings have, like other species of animals here on earth. Where is the boundary line between human beings and other species? What about an Octopus, Octopus is clearly not a human being, but as we learned by talking with Peter Godfrey-Smith, they think in quite advanced ways. You might very well think that something like that deserves the kind of moral status as a human being. And you have to answer these questions, you might just say, Ah, here's my answer, and move on. You might be unreflective or not very deep about it, but that's not a virtue, that's just you being lazy.

0:02:01.2 SC: So, today's podcast with Samir Okasha, who is a Philosopher of Biology, is about these kinds of questions. In some places, it's exactly about these questions, what is a species? What is a purpose? What is an agent, things like that. But I think that more importantly, than any of the specific answers that are offered here, because Samir is very good at sort of saying, Here's what all the plausible answers are, without being too picky about which one is the correct one. The idea of thinking about these questions carefully and reflectively. Philosophy of biology is clearly related or similar to in some ways philosophy of physics, but also different because the things being studied, animals, plants, organisms, species, genes, evolution are all things that are emergent, right?

0:02:49.4 SC: They're higher level things. So where to draw the boundaries around them, becomes an important question, and it might get frustrating to you, but these are really, really important questions to policy, to morality, to how we get through our lives. So it's gonna be an important conversation I think, for getting straight on some of those issues. And we also address some big picture, very loud and noisy controversies in biology, such as the levels of selection, the kin selection versus group selections controversies, and so forth. So with that, let's go.

[vocalization]

0:04:32.2 SC: Samir Okasha, welcome to the Mindscape Podcast.

0:04:35.2 Samir Okasha: Thanks very much, Sean.

0:04:37.7 SC: I hang out with both physicists and philosophers myself, arguably I'm one myself. You're a philosopher of biology. And I guess, my first question is the relationship of philosophy to biology. And we have philosophy of science, and that often talks about theories and theory choices and things like that. In philosophy of physics, we have this subset called the foundations of physics, where we're really kind of doing physics, but in a way that would only let you get hired in a philosophy department, not in a physics department. Is there foundations of biology, is that a recognizable thing?

0:05:11.3 SO: There is, and that's to some extent what I think of myself as doing. However, I don't think everybody in philosophy of biology would recognize that label. So perhaps I should say a few words about how this sub-discipline of the philosophy of science came into being. I mean, I think it crystallized into a subject in its own rights in the 19, the early 1970s. And in part, this was because of a feeling in the Anglophone Philosophy of science community, that what passed for general philosophy of science, which dealt with sort of general methodological questions about the relation between hypothesis and evidence, about what explanation looked like, about how theory choice is made and so on. While purporting to be fully general in reality, was physics-centric. And so a lot of people started to ask the question, Well, wait a minute, where do the life sciences fit into this picture?

0:06:10.8 SO: And to start thinking of biology as an interesting sort of test case if you like, for general ideas from philosophy of science. And very often the question was, How well do these ideas about explanation confirmation or reduction, for example, transpose over from the implicitly assumed domain of the physical sciences to the life sciences? That was, I think, the initial impetus for the subject, but it quickly gained a following. And I think the motivation for linking philosophy and biology changed somewhat in the ensuing decades, in that very many philosophers and I think have hit someone like Daniel Dennett being a key person to illustrate this, came to think that certain ideas in biology, particularly in evolutionary biology, particularly to do with natural selection had a significance for philosophical questions that was quite unique, that one wouldn't find necessarily in other bits of science, including other bits of the life sciences. And so for that reason, evolution came to be at the centerpiece of this attempt to link up modern biological science with traditional philosophical ideas that constitutes the core of the philosophy of biology today.

0:07:33.7 SC: Yeah. It makes perfect sense to me that evolution plays the same role or that stands in the same relationship to philosophy of biology, that quantum mechanics stands in relationship to philosophy of physics, like centrally important and obviously raising some hugely important philosophical questions.

0:07:49.9 SO: Yeah. No, that, that's absolutely right. I mean, Darwin himself noted this. So in a famous comment he made in a notebook, never intended for publication. Then Darwin wrote, "Origin of man now proved. He who understands baboon would do more for metaphysics than Locke in an allusion to the... "

0:08:11.8 SC: Wow. Never heard that one.

0:08:12.5 SO: The 17th century English philosopher John Locke. Now he didn't expand on what he meant by that. And as you know, he was a very modest man, never intended that to see light of day. But I think it nicely captures the sentiment you express, Shean, that there's something about the theory of evolution by natural selection, that is of unique significance for philosophy. And this was a point picked up again by many thinkers in ensuing decades. It's interesting that the Philosopher Ludwig Wittgenstein in the Tractatus Logico-Philosophicus, from I think it was 1916 or 19, or something, who was not a fan of the, what we now call the naturalistic way of doing philosophy. So in that book, Wittgenstein says at one point, Darwin's theory has no greater relevance for philosophy than any other hypothesis in natural science. And you can almost hear the plaintiff tone in Wittgenstein's voice, as he says that, because the...

0:09:18.6 SC: Oh, yes.

0:09:19.4 SO: And I think the fact that he needed to say it, illustrates how the contrary opinion must have been in the air at the time.

0:09:26.5 SC: Well, okay, but then, yeah. So I completely agree that evolution clearly is going to be relevant, but there's fuzzy things going on in our understanding of evolution, right? Like what are the units of selection? We famously have had debates relatively recently. Is it the gene that is the unit of selection, is it the organism, is it the species, is it the, is it groups, is it kin, is it... I presume this is an active area of what you guys talk about.

0:09:56.6 SO: Very much so. And this was indeed the subject of my first book on evolution called Evolution And The Levels Of Selection from 2006, which was a philosophical attempt to come to grips with this, the very question that you discuss. Prompted by the fact that within the evolution of biology community, there were apparent disagreements about how exactly we should think of natural selection as operating. Should we describe it as a process that ultimately selects some genetic variance over others, perhaps the orthodox neodymium viewpoint and enshrined in popular works such as Richard Dawkins, The Selfish Gene. Or should we think instead of it as a process that fundamentally operates at the level of individual organisms, much as the way Darwin thought most of the time, or as some people would have it, should we adopt an expanded perspective according to which natural selection can in principle operate, at multiple levels of the biological hierarchy at once, including in some cases selecting whole groups over others, and in the more extreme versions of this, possibly even selecting some species over others?

0:11:10.4 SO: Those are ideas that for a long time, and I think to some extent still today, regarded as controversial, if not heterodox by the mainstream in evolutionary biology. But they have always had an important following, particularly this idea of group level selection.

0:11:28.4 SC: So what's the right answer?

0:11:31.2 SO: Oh, it's a long story, and the answer is, ever in philosophy and in most things is, it depends on what you mean. [chuckle] But I think one of the striking things about this debate and what attracted me to it initially was that, it was not always obvious whether the disagreements reflected empirical disagreements about the selective processes actually occurring in nature or were rather disagreements of a different character, where it was really a matter of convention, if you like. How one described how, which formulation of the same theory one adopted. So to put, to make that more concrete, so being this long clash between proponents of the Darwinian idea of individual level selection, which basically hypothesizes that natural selection, is choosing individual variants in a population, typically a population of conspecifics over other ones, according to the criterion of greater enhanced survival or reproductive success.

0:12:39.4 SO: So the better or fitter variants get selected as the bearers of those variants survive better than other individual organisms. But beginning with Darwin himself, the idea has been around that in certain cases, the process of natural selection could operate on whole groups. So, it could be group on group competition, if you like. And this is particularly relevant, given that many organisms live in groups of various different sorts from family groups up to larger tribes and larger aggregations. All within a single species in the case I'm thinking of. And it's inconceivable that in principle, some groups might do better than others. So the idea of natural selection, if you like, could be frameshifted from the individual level up to the group level.

0:13:37.2 SO: And the significance of this idea is that it's a putative mechanism by which, so-called altruistic behaviors could evolve. So something that's individually harmful or some sort of self-sacrifice on the part of an individual to help others, would seem impossible to evolve by the mechanism Darwin mostly had in mind, individual level selection. But if you think of selection as favoring some groups over others, then it's conceivable that things that are beneficial for the collective, even if they're individually costly, could prosper. And that's the historical reason for being interested in this issue of group level selection. But what attracted me to this debate, and I think other philosophers too, was that it wasn't really always clear whether the proponents of these two views were talking past each other or not. On the face of it, it started out as a resolutely empirical issue. I mean, here's a hypothetical process that we can model and describe, does it go on in nature or not? And if it does, how common is it and how important is it?

0:14:37.8 SO: That's to say, selection that the group level. But alongside that, you had a group of other people arguing that, no selection at the group level is really just a different way of talking about a process that they referred to as kin selection. Which is a process of selection on individuals that takes into account the fact that many behaviors an individual engages in, may have consequences for the reproductive fitness of it's kin. So in short, you had a situation where it wasn't clear whether we had a disagreement of fact or two different ways of stating the same facts or for the same hypothesis.

0:15:25.6 SC: I always felt a little naive about this, because it seems to me obviously true that altruistic behavior can be naturally selected for, and maybe this is, this group selection seems kind of natural to me. If I have a group of people who are fighting for each other and cooperating, I can easily see them surviving over a bunch of rugged individualists who are just in it for themselves.

0:15:48.4 SO: Yeah. No, that, I mean, that, that's exactly right. I mean, although there is a problem in that you are right, that there is, there would be a selective advantage to a highly cooperative, cohesive group, where everyone looks out for the group, rather than a group composed of rugged individualists who only look out for themselves and don't care about group welfare. However, one has to ask the question is that going to be evolutionarily stable? Because, you are certainly right that in group on group competition, the group containing the altruists will likely do better. However, if you think of the selective dynamics taking place within a group, then it would seem that the selfish free rider who looks out for themselves is going to outdo the altruist who cares about group welfare. And so the mechanism, it's not obvious that the mechanism can work, particularly over longer periods of time. Interestingly, Darwin himself put his finger on exactly this tension in the idea of natural selection in his book, The Descent of Man, in a famous discussion of the evolution of self-sacrificial tendencies among early hominids. So Darwin claimed that self-sacrifice was something one found in hominids, willingness to put one's own life on the line for the good of one's, of the tribe or the larger unit.

0:17:25.2 SO: And he then said, "Well, how could that have got there?" And the first thing he said was precisely to make the point that individually, that would seem hard to evolve by individual selection. So Darwin said he who was ready to sacrifice his life as many a savage has been, would often leave no offspring to inherit his noble nature. So what Darwin is saying there is that the altruist comes last.

0:17:50.5 SC: Well, yeah. Okay. So this...

0:17:52.4 SO: To put put your... But then he continued and immediately in the next sentence, Darwin continued, sorry to interrupt, by saying, "However, a tribe including many members who are always willing to sacrifice themselves for the common good might be victorious over another tribe. And this would be natural selection too." So Darwin himself in that very insightful passage, firstly points out the problem, points out the puzzle. How is it that a process of natural selection could lead individuals to behave in ways that are costly, but benefit others with a putative solution, namely the hypothesis of natural selection up at the group level.

0:18:35.0 SC: So, it tempts me to amend my previous suggestion and say that now the robust group will be one in which there are many altruistic cooperators, but also several selfish free riders. And perhaps the mixture of them will vary with time, depending on how much pressure they're getting from the outside world.

0:18:58.4 SO: Yeah, no, that's exactly right. Sean, you put your finger on it. It depends on the balance between the different selective forces. So you can immediately see why evolutionists realized that they had to mathematically model this. It was no good just to sort of bounce around scenarios the way Darwin was doing. And that effort really only came to fruition. It was quite significant time later. JBS Haldane, the English evolutionist in a 1930 book, The Causes of Evolution, gestured at the problem in formally and quantitative terms. But it was really only with work in the 1960s that people came to think in a systematic way about what the different selective forces are, and whether the mechanism of group level selection favoring altruism could evolve. And for a long time, the received view on that question was no. That it's an implausible evolutionary mechanism, essentially because the models, the early gen models designed to probe that question made this assumption that the turnover rate of individuals was much faster than that of groups. So if you like, the group would stick around for many individual generations, and that meant that selection or at the individual level would be faster and would lead the groups quickly to become selfish, dominated by selfish individuals.

0:20:28.1 SO: So in short, although it was still true that a cooperative or altruistic group would do better than a selfish group, the problem was that group level variation wouldn't be likely to arise in the first place, because the individual level process would tend to undermine it.

0:20:46.2 SC: It swamps it.

0:20:47.6 SO: If you like. However, more recently, opinion has changed on this matter somewhat, I should say. And the idea of what's now called multi-level selection has enthusiastic advocates.

0:21:00.6 SC: Well, that's kind of where I was gonna go, because at the philosophical level, this seems like a wonderful test ground for thinking about emergence, right? In some level, all of these goings on could be described as a bunch of atoms and molecules bumping into each other and obeying the laws of physics. And we're saying, but there's a more efficient surveyable, understandable higher level description. And in this case, unlike atoms going to fluids, when you go from molecules to organisms, or species or groups, it's messier and maybe all the emergent levels do interact with each other in some way.

0:21:39.8 SO: Yeah. No, I think I agree with that. Yeah. So there's the general issue of emergence and the idea of course of emergent property, the idea that sort of aggregates the things that are not fundamental particles, if you like, will often come to exhibit emergent properties, be they chemical bonds or larger things such as biomolecules or cells or whole organisms. And I think that certainly is right. And most of us think in philosophy, and I'm sure in most branches of science too, that although in principle, there's presumably a physical level story going on, that nonetheless, there are deep and important laws governing these larger entities in virtue precisely, because they exhibit these emergent properties in a systematic way.

0:22:41.2 SO: How the emergent property idea bears on this question of levels of organization and nature and levels of selection is, I think an interesting question. And one idea that's always been out there is the idea that natural selection acting above the individual level, where individual roughly means multi-cell organism in this context. So natural selection, acting at a higher level at the level of a family group or a tribe or a community or a species or something like that, requires that higher level entity to have properties that are emergent with respect to the lower level. So if you think, for example, of the contrast between an insect colony and a buffalo herd for example. A buffalo herd is basically just a bunch of buffalo all living together. Now, they do interact in interesting ways and so on, but it's stretched to think of the herd itself as having any emergent properties. Obviously, you could say some true things about the herd, and it has a certain population density, for example, occupies a certain geographical area, there's a certain total body mass and so on. And those in a sense all properties of the herd, but they clearly depend in a very direct and obvious way on the properties of the individual buffalo that make up the herd.

0:24:07.4 SO: Contrast that with something like a honeybee colony, which is an incredibly sophisticated grouping where those bees are divided into different castes. You've got the queen, you've got the workers who don't reproduce themselves, but who spend their time defending the nest, tending the queen's brood, foraging for food, working for the common good in short. And so you have a complex division of labor and what people often loosely refer to as a sort of functional integration of the level of the whole colony. And that's exactly the sort of thing that emergent property might use for the describe, that the term emergent property might usefully describe in a biological or ecological context. And it's no accident that things like honeybee colonies are often taken to be higher units of units of selection.

0:25:06.4 SC: Interesting. And so this is, you're bumping right up against the famous question of whether or not the higher level emergent properties are truly new or whether they could be derived from the underlying thing. I'm literally trying to finish a paper myself about this right now, but I'm the person you described at the beginning. I just have... All my examples are physics examples and I need to have more biology examples in the back of my mind.

0:25:31.0 SO: Right. And I think there certainly are some, yeah. I would distinguish two things. One is the general question of whether biological entities, including biomolecules for example, exhibit emergent properties, to which I think the answer is definitely yes in one sense of the term emergent property. From the question of whether emergent properties sort of feature in certain types of natural selection or must feature in certain types of natural selection, which is perhaps a slightly different question. And to see the distinction is just that one can ask the formal question even in the hypothetical circumstance that creationism were true.

0:26:13.5 SC: Yeah. Okay.

0:26:14.4 SO: It doesn't depend on evolution per se. It's really a question about levels of organization.

0:26:23.7 SC: My own experience in the emergence literature is that there are people who mostly bring up either physics examples or maybe computer science examples like the game of life. And there are other people who leap right to consciousness, and they wanna say the consciousness is strongly emergent. Would you advocate for more people living, spending some time in-between and thinking about biology?

0:26:46.3 SO: Maybe. Well, I think it's certainly true that there is an established use of the term emergent property in ecology, particularly where people talk about ecosystems having emergent properties such as the property of stability, for example, or a certain degree of complexity or a certain amount of nutrient recycling in the ecosystem at large or something. So people know what they mean when they talk about emergent properties in... Well, they pick out something that they all agree on at least. When they talk about emergent properties in a biological context, whether I would advocate that people interested in emergence in general expand into their repertoire of examples from the physical case and the consciousness case, to the more general biological case. I think it depends on what the goal of the inquiry is. It's long seemed to me that much of the talk about emergence is really that people are talking past each other.

0:27:52.8 SC: Oh, yeah.

0:27:52.9 SO: I think I'm not the first person to have made this obviously by any means. But I think it's interesting that I think philosophers and scientists often are talking past each other when they use the term emergence in that, for many philosophers that came to connote something really extremely puzzling. And there's some debate about whether those things really exist at all. Whereas I think for many scientists then, the notion that emergent properties exist is obvious. It's not up for grabs. So it's just another way of saying that there are things that aren't quarks if you like.

0:28:25.9 SC: It is hard and I've often tried to figure out a better word to use for what the physicists call emergence, 'cause I don't want to get confused when talking to people who come at it from a different angle. But it's a good word. We're stuck with it. We just gotta try to use it carefully, I think.

0:28:40.5 SO: Yeah. No, that's certainly true. There's probably no escaping it. You are right.

0:28:44.6 SC: For me, the other big philosophical issue, well, scientific/philosophical issue, foundational issue in evolutionary biology is the role of how much change, how much evolution is adaptive, versus just random drift or accidental spandrels or things like that. Is that something that you have tackled or thought about?

0:29:10.3 SO: Yeah. No, absolutely. I think of that as being... Well, in principle, an empirical issue at least when formulated carefully. So with your reference to spandrel then that brings to mind the famous paper by Richard Lewontin and Stephen Jay Gould from 1979, where they castigated traditional neo-Darwinians, for being too ready to assume that everything they find has an adaptive explanation and for telling what they call, Just So Stories. And I think it was a partially valid, if overstated critique in my opinion. I think they certainly engaged in one too many rhetorical flourishes in that paper. But they also made some sound points which is, one has to really consider quite carefully what the test is for whether some particular organismic trait really is an evolved response to some environmental situation or not.

0:30:15.9 SO: However, in the molecular biological age then, a whole lot far more sophisticated methods are available than when Gould and Lewontin were writing in the age of DNA sequencing, by which it is far easier for theorists to find signatures of adaptive evolution. And to get to the heart of this question of whether chance or adaptation is really the dominant feature, leading to change at the genomic level at least. I think at the phenotypic level then, although the... I mean the role of chance, the importance of chance processes can't be underestimated in evolution. They play an absolutely critical role, particularly in small populations where gene frequency dynamics will typically depend very heavily on population size. And if there's a small population, then chance is likely to be the dominant factor explaining changes in gene frequencies over time.

0:31:25.1 SO: However, the level of the whole organism's phenotype, I think the dominant view in a lot of biology has been that if we're talking about something that's extremely complex, if you're talking about, I don't know, some signaling pathway within a cell or some organ, if you think of a kidney or something like that, or some aspect of say a crab's exoskeleton, some aspect of anatomical structure. That's not the result of one gene, but is the result of many, many, many genes and a complex process of development each generation giving rise to it. Then it's inconceivable that that could be the results of chance alone. Certainly, mutations originally arise by chance. That's agreed on by all parties. But the traits in question are so complex and so clearly adaptive.

0:32:24.7 SO: Well, what's so apparently adaptive I should say, that the hypothesis that they could have evolved by anything other than cumulative rounds of natural selection seems implausible. So there's long been a tension within biology when people are thinking about chance versus adaptation, between those who are discussing the question at the genomic level, where they're discussing genome level evolution, changes in gene sequence, basically over time. And those who are discussing phenotypic evolution, where they're focusing on observed phenotypic characteristics of organisms. And there's a bit of a disconnect between those, and that a lot of genomic evolution doesn't have phenotypic consequences, in that many mutations, indeed, most mutations that are not deleterious are just silent. They make no difference. Very often will make no difference to the protein that the gene produces and so make no difference to the phenotype. And so you can have a lot of genetic change that doesn't manifest up at the phenotypic level. And that's why chance can be a dominant factor down at the genomic level, but playing less of a role, although still a role when one is considering the evolution of organisms phenotypes.

0:33:52.0 SC: Have you paid attention at all or followed the recent discussion about assembly theory and its role in the origin of life?

0:34:02.1 SO: Yeah, I have. There was a paper, wasn't there? About five years ago or so in nature I think that did the rounds or was it in PNAS that did the rounds about that. Yeah, no I don't have anything good to say about it, I'm afraid. I have read that. I have read that paper and I do remember thinking, interesting but overstated.

0:34:23.5 SC: Well, for the people who are listening, I give a tiny gloss on it. Sara Walker was a previous Mindscape guest, and she's one of the authors of the paper. But they seem to be saying I think overstated is very much my own reaction to it. They say something that is interesting and completely plausible, namely that the specific ways that complex structures can arise over evolutionary history are very path dependent. The specific steps are gonna matter. You're not just throwing things randomly together. But then they say, or at least they very, very strongly imply, therefore, we need to change the laws of physics to account for what is going on at a fundamental level. And that I just don't see at all.

0:35:07.0 SO: Yeah, no, that sounds like a big leap, that second conclusion.

0:35:12.2 SC: I'm probably being unfair.

0:35:12.3 SO: I remember thinking of when and... When I read that paper that it belonged in a lineage of sort of somewhat anti-Darwinian thinking. In that, the more extreme versions of Darwinism have always basically painted organisms as kind of a blank canvas, and just assume that look, whatever, anything could evolve, broadly speaking. That the constraints imposed by the laws of chemistry and physics are not particularly strong, relative to the range of amazing things, the ways that a lineage could go, the trajectory through evolutionary space that it could take. And so really, the actual trajectories taken are going to depend heavily on what was advantageous in the ancestral environments, and so the Darwinian story can proceed. And that contrasts with a different tradition that has always emphasized the importance of structure in constraining and physical constraints and physical chemical constraints, in constraining the type of things that can evolve, and sort of accord it a far lesser role to Darwin. Darwinism, thinking of Darwinism as basically a filtering mechanism for choosing among the few things that are physically feasible at all.

0:36:36.6 SO: And that way of thinking obviously downplays the significance of the Darwinian mechanism for accounting for what we have. And that's an old and fascinating tension, I think within evolutionary thinking or biological thinking, perhaps I should say.

0:36:54.2 SC: Yeah. I think one of the ways in which the assembly theory folks wanna contemplate, modifying the laws of nature is to make them more explicitly teleological. And that's always sort of a fraught word in discussions of biology and philosophy. The idea that there's motion toward a goal of some sort. It seems to me to be exactly the opposite. Maybe this is what you're saying of the Darwinian spirit.

0:37:20.0 SO: Yeah. Well, it's interesting. I think there are two ways of describing how Darwinian ideas bear on the issue of teleology or goal-directedness, which is, as you rightly point out Sean, is as a topic of age old philosophical concern. I mean, this goes all the way back to Aristotle and probably before. I mean one line of argument, and this was perhaps the dominant, probably still is the dominant view in evolutionary circles, is that what Darwin did was to basically banish teleology from nature. So it looked as if there was a Telos, that life had a kind of Telos. And what Darwin showed was that a brute blind, causal process that is not foresighted in any way, that simply is choosing between variants according to how currently good they are in the current environment could give rise to the appearance of purpose and of goal-directedness in the world.

0:38:25.4 SO: So, according to this view, what Darwin did is show that there isn't any teleology in nature and to show what looked to be teleological explanations, such as things like trees grow tall in order to get more sunlight, could be recast as valid causal explanations by invoking the feedback process of natural selection. So in a sense, it's perfectly true that trees grow tall in order to get more sunlight, but when we paraphrase that correctly, we realize that it's not a teleological explanation at all, but a brute causal explanation about the selective advantage to a tree being a little bit taller than its neighbors, and then the outcome of the cumulative process of selection that gave rise to. So that's the first view. It says Darwin banished Teleology. Second view, however, says, no, he didn't. But what Darwin did is naturalize teleology, or if you is to make the world safe for teleology [laughter] and to show how in fact it could be rescued.

0:39:30.2 SO: And an interesting concept arose here in the work originally due to an author called Colin Pittendrigh in the 1960s called Teleonomy. And then this was taken up by the later evolutionist Ernst Mayr and is still widely used today. So Teleonomy is meant to denote the sort of respectable form of goal-directedness that could survive the refutation of teleology by Darwin, if you like. So the idea is that Darwin throughout... The bad teleology, an example of the bad teleology would be the idea that evolutionary change itself is directed towards a goal, such as humans or something like that.

0:40:13.6 SC: Us. We're always the goal. Yeah.

0:40:15.6 SO: [laughter] organic change, right. So that's the bad sort. Or more generally the idea that natural selection has an eye on the future would be an example of the bad sort. The illegitimate sort. But what many theorists who talk about Teleonomy want to say is that organisms really do engage in goal-directed activities and behaviors. And that that's not just in the eye of the beholder and isn't something you can try and explain away in that aspects of say organismic reproduction and development of an organism and organismic behavior and of organismic physiology, all involve the movement of a system towards a goal. And that's implicit in how we speak about organisms. So imagine if you are observing animals engaging in some unusual behavior in the wild. The first thing that you say is, what are they doing? What are they trying to achieve?

0:41:17.5 SC: Why.

0:41:20.3 SO: Why are they doing that? Right. And the suggestion is that that's not illegitimate, that way to speak. That much animal behavior really is goal-directed. So take for example, a salmon that engages in that incredibly complex behavior where it navigates all the way back to its natal stream, thousands of kilometers from where it is in order to spawn. Imagine if you didn't know about the phenomenon of homing in salmon. Biologists are studying it first time and they track these salmon and find they're all doing this journey, first thing they're gonna say is what are they doing? What are they trying to achieve? What's their goal? [laughter] And so the underlying idea then is that a notion of goal-directedness is just a true descriptive fact about many organisms, about all organisms really, and that there's absolutely no reason to disavow that in the light of Darwin.

0:42:15.3 SC: This is actually very...

0:42:16.5 SO: So those are the opposing polls.

0:42:18.3 SC: Yeah. No, that's very, very helpful because I recently on the podcast talked to David Krakauer from the Santa Fe Institute. And he defined complex systems research as the study of Teleonomic matter. And I didn't bother to, in my mind or in the discussion, draw a distinction between teleological and Teleonomic, but you've explained it to me very nicely, and I kind of want to say Teleonomy is just emergent teleology. I don't know if that's okay.

0:42:49.5 SO: Right. No. And that may well be right. I mean, it's partly complicated by the fact that as ever these terms have conflicting definitions. [laughter] So in particular I referred earlier to the German evolutionist Ernst Mayr, who wrote important works over many decades in sort of '50s to the '80s and also touched on themes in philosophy of biology. And he borrowed this word, Teleonomy, but invested it with a completely different meaning from the one that Pittendrigh and others had originally coined it for. So I think one has to be careful with that, but roughly speaking I guess the definition is teleology real goal-directedness, Teleonomy apparent goal-directedness, if you like.

0:43:38.9 SC: Which very nicely segues into the other thing I wanted to talk about, which is there are, as you already have said, individual organisms that act in very goal-directed ways. Sometimes we call them agents, and you've written a whole book about this. But it is fascinating to me, forget about the overall goal-directedness of evolution or anything like that. Individuals have goals, right? I have goals. I know that's true and I think that I'm part of the natural world. In some sense over the course of evolutionary history, some organisms started caring about the future in some way. I'm sure that's an overly naive way to put it, but what do we think about how that happened?

0:44:22.8 SO: Yeah. No, I think these are interesting questions. One, way to think about it perhaps is this, I mean, if you think of humans like you or I, we have goals, in having this discussion we've got a goal, we've got a joint goal of making a podcast. And then in an hour's time, I'll have the goal of cycling home safely, cooking some dinner for my family or whatever. So think of that sense in which humans have goals. Then one standard way to think about that is to say that we have a mental representation of the goal. So there's a goal state. That we make a podcast. And you and I both mentally represent that state, meaning that we intend, we actually consciously intend to achieve that.

0:45:09.9 SO: Now, whether or not other animals mentally represent their goals is a complex discussion, within the field of comparative cognition. And opinions divide on that. I mean, I think most people would allow that. Certainly many primates do, possibly other vertebrates such as birds. Some authors want to extend belief and mental representation of a goal all the way down the animal kingdom. And some indeed would extend it to plants and microbes. But that's a more heterodox view, I think. But in any case, I think everyone would agree that not all life on earth has... Not all organisms that exist on earth have mental representations of their goals. But nonetheless, most of them, and perhaps all of them seem to have goals in a weaker sense, seem to engage in goal-directed behavior.

0:46:15.9 SO: Even if it's just something as simple as a bacterium swimming up an oxygen gradient. I mean, obviously it would be anthropomorphic to say that the bacterium is trying to reach an area of higher oxygen concentration. But nonetheless, in a fairly obvious sense, its goal in moving is to achieve a region of higher oxygen concentration, just in the sense that it consistently moves towards up the gradient and it stops once it's got enough oxygen. So one idea you might have is that conscious goal-directedness that involves a mental representation of a goal that humans and perhaps other animals have, evolved from the simpler sort of goal-directedness that's manifest in the teleonomic behavior of all living organisms, if you like. And then the task, if one accepts that conceptualization of the problem, the task then is to tell a plausible evolutionary story about how that might have occurred, what the intermediate steps might have been.

0:47:25.7 SO: And that's no easy task in part because it's not just a scientific one in that as I'm sure you know a lot of philosophers have long problematized this idea of mental representation of a goal or intentionality and wondered what it really amounts to. And indeed some have wondered whether it really exists. So it's not as simple as saying, now we just need to discover what the evolutionary pathway was from sort of implicit goal-directedness of the sort that even bacteria have to the conscious goal-directedness of the sort humans have. Because the endpoint there, the conscious goal-directedness of the sort humans have, not all theorists even agree on what that is. [laughter] if you see what I mean. And it is hard to tell the good evolutionary story about how some attribute came into being, unless one can agree on what that attribute is.

0:48:22.0 SC: Well, that's why we pay philosophers the big bucks here, right? To help [laughter] clarify these things. I mean, I think it is a trickier set of questions than they initially appear to be. I mean, a ball rolls down a hill and it comes to rest at the bottom of the hill. Essentially nobody would say that it had the goal of getting to the bottom of the hill, and maybe Aristotle would have, but I think we wouldn't. And when my cat breaks into the bread box to get the cat treats, I think it pretty clearly does have a goal of getting those cat treats. The bacteria is somewhere in the middle. Is it that we need a richer conception of goal-directedness, are there subdivisions or is it just that we've been too loose and we should pick the right one?

0:49:10.9 SO: I think these are very difficult questions. Yeah, I mean, one thing I try to do in my book about agency is just prize apart a number of different notions of agent, which I claim are actually work in different scientific fields. So I start with what we have traditionally meant by agent in philosophy sometimes called intentional agent. And according to this notion then to be an agent is to act rather than just to behave. And that means one has to explicitly, consciously have the goal of trying to achieve something. So intentional agency requires a certain psychology, you've gotta have a certain psychological makeup in order to be an intentional agent. Exactly who has that and who doesn't is perhaps a... Is a tricky question. But in any case many things don't.

0:50:00.8 SO: Bacteria don't, balls rolling down inclined planes certainly don't. The weakest notion of agent that I see at work is what I call a minimal agent. And I should note that I use this term differently from some other authors. And by minimal agent, I just mean, any entity that does something where the distinction here is between doing something and merely having something happen to one, if you like. And here I draw in a very interesting discussion by the philosopher Fred Dretske in the 1988 book, Explaining Behavior. Where Dretske just says, Look, contrast these two things, right? A rat is in a cage and it moves its paw to press a lever to get a pellet's learnt. It gets a pellet of food when it does that, an experimental rat. So there we say, the rat moved its paw, scenario one. Scenario two, an experimental biologist puts their hand into the cage and manipulates the rat's paw.

0:51:06.4 SO: And so in that case, the rat's paw was moved. But that's not a movement, that's not something the rat did. It's something that was done to it. And so the suggestion then is the most minimal notion of agency is just something that does things as opposed to merely having things happen to them. And how do we make that distinction precise? [laughter] Well, some philosophers have thought that the key just lies on whether the cause of the motion or the motor output is internal to the organism, or is some external force acting on it. And again, I'm not saying that that's completely clear, like everything in this domain, it's all a bit murky, but I think there's a very intuitive distinction there, and that I think serves to define the most minimal notion of agency. So we got the most demanding notion, intentional agent, the most minimal, no, sorry, the most undemanding notion, minimal agent. And between those, I think we get notions of agency in different fields, such as an AI, where people talk about an intelligent agent, and that just means anything that changes its behavior depending on the circumstance basically. So it doesn't always do the same thing, what it does depends on the circumstances, will implement stimulus response conditionals, if you like. If this, then do that.

0:52:25.6 SC: Does a volcano that is about to erupt count as a minimal agent?

0:52:26.9 SO: Well, it might do. Yeah. I mean, and I think it's not accidental that we use the active voice there. We say the volcano erupted, as well obviously, it's true to say there was an eruption of the volcano, but we think it's not wrong to say the volcano erupted. Now how much to hang on linguistic features like that is never clear. I mean, I think it can only be indicative. But I do think that there's no reason to think that minimal notion of agency picks out the biological domain particularly.

0:53:06.4 SC: Sure. Okay. And for the maximal notion for the intentional agent, clearly this idea of a mental representation of the world is crucially important. So let me kind of, and this is gonna be a hard question again, but that's why we're here. There must be levels there, the sort of richness and capacity of our mental representations of the world. I mean, does the bacterium have a one bits worth of mental representation of the world? There's more nutrients in this direction, less nutrients in that direction? Or is that going a bit too far, being too colorful?

0:53:44.5 SO: Yeah. I don't know. I think it depends. I mean certainly it has some internal state which one could take to be a representation of... Well, I don't know actually quite how bacterial chemotaxis works. I mean okay, what I'm saying isn't true, but could have been, let's say.

0:54:06.5 SC: Might be, yeah.

0:54:07.2 SO: You could well imagine a simple microbe, perhaps a bit more complicated than the bacterium, that had an internal state, which represented the amount of oxygen in the environment and the level of that internal variable dictated, along with other things, what the microbe does. Now, I think in fact, that's not how bacterial chemotaxis works.

0:54:30.9 SC: I think you're right. Yeah, that's right.

0:54:33.0 SO: But it could have been. So in that case you might well say that an internal variable or state of the organism represents some worldly circumstance, but whether it would be a mental representation is perhaps a bit less clear. 'cause you might say, well, to have a mental representation, you need to have a mind, to start with. And so maybe there is sort of independent criteria for what that means. But in short, I think I agree with you, Sean, that it's not a simple question to say exactly what mental representation is and what the bounds of it are. And indeed, I should say that in the philosophy of mind psychology, not all authors are happy with this notion of mental representation. So in some of his early work, Daniel Dennett made the argument that there wasn't really any fact of the matter about whether some creature has intentional states, such as believing that the world is a certain way, wanting to achieve a certain thing, classic sort of examples of mental representations.

0:55:37.9 SO: For Dennett then, the only question is, is it heuristically useful to treat it that way? And he insisted that it was a fool's errand to think that there was a distinction out there in nature between real intentionality that we humans have and the as if variety. In a way he wanted to say, Look, it's all as if, right? This is the idea of the intentional stance, that it's really just a question of the heuristic utility of thinking of something this way rather than describing it in terms of its physical makeup directly.

0:56:14.1 SC: Do we know...

0:56:14.2 SO: And I'm not saying Dennett was right about that, and I think he backtracked to some extent in his later work. And seemed happier with the notion of mental representation taken literally. But I think it was a very profound idea, in fact, that idea of the intentional stance and that idea that the question is really, is heuristically useful to describe things this way, not do they really have this attribute.

0:56:38.0 SC: Do we know much about the actual evolutionary history of mental representations or other aspects of agency? Were there big thresholds that we can point to where organisms developed a new capacity to do this kind of thing?

0:56:55.6 SO: Well, I suppose the big, probably the big threshold will be the evolution of a nervous system, of a rudimentary nervous system of some sort that enables the organism to process environmental information or environmental signals in a radically new way. And to respond adaptively as a result. A person who thought, at great length about this is the philosopher geneticist, Eva Jablonka. In her book, the Evolution of the Sensitive Soul, who adopts quite an interesting methodology of trying to find instead of taking the notion of consciousness as primitive, she tries to find a substantive correlates, if you like. For what we ordinarily call consciousness, such as unlimited learning potential, for example, the ability and principle to make associations between any two things. And to act on them, to take one thing, to be a cue for another thing, and then to tell an evolutionary story about how these came to be. But for me then it's, yeah, I mean, until you've got a nervous system, then you are a fundamentally different sort of a creature, I guess.

0:58:15.1 SC: Well, computers don't have nervous systems and you've already brought up AI a little bit. Is there some difference in principle between the level of agency in a large language model versus just a really big lookup table or something like that? Is it just a matter of our perception as outsiders or is there something that you can point to inside that really makes a difference?

0:58:40.0 SO: Yeah. No, this is a classic, classic question in the philosophy of psychology and artificial intelligence [laughter] to which I don't know the answer. My intuitive, naive reaction is to say, no, of course. They're not really intelligent. They're not really thinking. It's all of the as if variety, right? All it's doing is machine learning. [laughter] All it's doing is pattern recognition, and that's fundamentally different from the real thing. But then of course, the question is, well, what is the real thing? And is this just an anthropomorphic illusion that we have? I mean, surely what is thinking if not computing a function? And surely computers are doing that too.

0:59:26.4 SC: Have you, I mean, being the world's experts on what an agent is, have you been thinking a lot about AI?

0:59:35.4 SO: Well, to some extent in that. I think that what they're doing, the notion of agency that they're operating within in AI, as they call it, what they call an intelligent agent, it seems to me to occupy an intermediate position in my sort of taxonomy of agent notions between the real deal and the minimal thing. But I am conscious that, AI is a huge topic now, and it will be a [laughter], a natural next avenue for me to look into. But as yet, no.

1:00:09.2 SC: Is there something that you would personally take as necessary before you called something an agent, that people who do AI could propose as a benchmark?

1:00:22.8 SO: Well it's a difficult game to play that, isn't it?

1:00:25.4 SC: Yeah.

1:00:25.8 SO: Because I mean, you only famously Turing played that game. [laughter] And one could argue [laughter] that he lost. I mean. So he propounded this Turing test and everybody was happy with it, but then they made things the past it, and now we're sort of, now we want to say, well, therefore the test must have been wrong.

1:00:41.2 SC: Right. Yeah.

1:00:43.6 SO: Rather than saying that they're, that they've passed it with flying colors. So yeah, no, I don't know. I mean, I've always been partially sympathetic to the idea that, you've really got to move. And you've maybe got to have a body of some sort. So there's the famous HEIDEGGERIAN CRITIQUE of AI that says, Look, until you engage with the world through your body, then you could never really have thought, if you like. So, according to that line of argument, it's as almost a category mistake to think that you could have a thinking machine in less, the machine was, you had some sort of sensory capacities of some sort.

1:01:33.7 SC: But that seems very easy.

1:01:35.1 SO: Analog of a body.

1:01:36.0 SC: Yeah. How hard is it to put it in a drone or a robot or whatever?

1:01:39.7 SO: Maybe. Yeah. And if, yeah, as you write, as you say, if it's just about moving around, as you say...

1:01:43.5 SC: Or sensing.

1:01:44.3 SO: What could be easier, could be Right.

1:01:45.5 SC: Affecting the world.

1:01:46.2 SO: Or even se sensing, yeah. In a sense. Yeah. No, and always one confronts this fundamental question of whether any criterion that anyone puts forward is anything other than just some anthropomorphic...

1:02:00.9 SC: Right. That's the worry.

1:02:02.8 SO: Ideal that what that they've unwittingly taken on board and are now claiming to be the true hallmark of whatever attribute may in fact be shared [laughter] Yeah, that's right. I think it, it is one of those questions that's almost impossible to really even think about or to conceive, isn't it? Because I mean, at some level, you know, intuitively, well, you do know that humans, obviously were an evolved species. So, all of our cognitive capacities and our neural capacities have antecedents and have evolutionary antecedents, and homologs of many of them are found in other animals. And non-human animals. But nonetheless, we do have this deep sense that there's something uniquely human about us too.

1:02:57.9 SC: Well, good. We can put away the computers and AI for the wrapping up the end of the podcast here. But because we still left on the table, big human sized questions about free will and moral responsibility and things like that. Let me just ask what might be the easier question? Does the discussion of what counts as an agent from a philosophical evolutionarily informed perspective, help us understand how we should think about responsibility and free will?

1:03:32.6 SO: Possibly. I think the question of free will versus determinism is a really difficult philosophical question. And I think, interestingly, it's one of those questions that doesn't, I mean, where the basic formulation of the issue doesn't seem to change much despite all the great advances in neuroscience. And in comparative cognition and in evolutionary biology that we've made, I mean, the basic issue is really, seems to be a more, a metaphysical one, doesn't it? It's just that on the one hand, we seem to think that quantum stuff aside, the world's pretty deterministic. And on the other hand, we can't understand how if the world's deterministic there could possibly be free will. Now, I mean, the puzzle is deepened by the line of argument that says, Look, even if the world were indeterministic, that wouldn't help anyway.

1:04:34.0 SC: Yeah.

1:04:34.8 SO: Because what I mean, that's not our conception of human freedom either. Just the actions happening randomly with no causal antecedents. I mean, is that what we really meant by freedom? Surely not. So I mean, if one finds that persuasive, then it must clearly indicate that there's something amiss with our concept of free will. I mean, if we think that it's incompatible both with determinism and with indeterminism, the only way that could be just as a matter of logic, is if it's self-contradictory. I mean, the only you know, if any, if a proposition P is incompatible with X and incompatible with not X, then P cannot be true in any possible world, right? So, P is therefore self-contradiction. Now, this is a familiar old line of thought in the philosophical discussion of free will. But I do think it's an important one that is sometimes lost sight of in the enthusiasm for the idea that a biological or neuroscientific insights will resolve the issue.

1:05:41.5 SO: Because for me, the fundamental issue is really to do with what the concept of free will actually amounts to.

1:05:50.3 SC: Right. Well, and we can say the same thing about these questions of moral responsibility. Right? Does thinking hard about how humans came to evolve over evolutionary timescales nudge you in the direction of not being a moral realist. Or is it that it's completely independent considerations?

1:06:09.7 SO: This is a really good question. Yeah. So, I mean, for me, when one becomes appraised of evolutionary ideas and takes on board the idea of evolution by natural selection, then yeah. Moral realism is one of the first things that goes [laughter]

1:06:26.7 SO: Now, so moral realism, being the idea that there's an objective truth about what the right thing, the right course of action is or something like that. I mean, I've always been persuaded of that idea that sometimes called evolutionary debunking in the philosophy literature, which says that evolution debunks our belief, if we had one in moral realism or in the objectivity of the moral realm. Now of course, plenty of non evolutionists also didn't, had no truck with moral realism. Famously, David Hume. Well, maybe it's no coincidence that David Hume came pretty close to the idea of evolution by natural selection, but didn't quite get there [laughter] So Hume of course, was was a subjectivist about ethics thought that, the ethical values were not part of the natural order, but were projected onto the natural order by the human mind.

1:07:25.0 SO: And that's broadly what a view I would endorse too. And I do think evolution supports it or at least make not proves it, if you like, but makes it the most reasonable thing to think that the contrary opinion that there are objective values out there in the world would sit, would seem to demand that humans are somehow above the natural order. And so the Darwinian, the demonstration that humans are fundamentally part of the natural order, and we basically understand how we came to be here, seems to me to sit uneasily with any belief in moral objectivity or moral realism. However, again, as ever in philosophy, not everyone agrees [laughter] And there has recently been an interesting body of work, is suggesting that no form of moral realism might in fact be compatible with evolutionary biology. That may be, that could be objective moral truths which are going to be truths about cooperation or something like that. That there's an evolutionary explanation of why they are true, if you like. So we can, according to this line of argument, naturalize moral values rather than eliminate them.

1:08:51.0 SC: Okay. Last question. We let our hair down at the end of the podcast. So given all that we've learned about natural selection and agency and responsibility and things like that, what would you expect to find if we discovered life elsewhere? Do you think it would be more or less similar to humans? Or would it be entirely, utterly different?

1:09:14.3 SO: Yeah. It's a really difficult question. I mean my temptation is to again, give the philosopher's answer. It depends on what we mean by life [laughter] Which in a way is true. I mean, I take it that there's no particular reason to think that the biochemistry of life on another planet would be relevantly similar to our biochemistry. So I think that, yeah, we would not necessarily recognize the living forms as anything akin to what we're familiar with. Even the weirdest sorts of microbes that are as far from sort of direct perception direct human perception as any, as any part of the organic world. So not much like us in terms of intrinsic makeup. Functionally, I think we could expect them, you know, any life form to be engaged in some form of sort of metabolism or energy exchange with the environment.

1:10:24.9 SO: I expect one would have to presumably have some form of reproduction unless they're only, finite, many of them, and they just persisted for a bit, but then presumably they would die, through, I mean, chance. Misfortune is presumably gonna be found on any planet, is not just a phenomenon that we're going to have here. So presumably in order to escape that, you have to be able to reproduce, otherwise the lineage will just die out. So maybe it depends on whether you mean sort of conditionally on us discovering them. I mean, I think if you mean unconditionally then there's very little we can say about what life on other planets would look like. Other than that it would have to satisfy whatever condition we think defines life. But if you mean that we might discover, then it would have to be conditional on our having discovered them. And that may well make some things much more likely or, and other things less likely than they were unconditionally, if you see what I mean.

1:11:30.2 SC: I do and...

1:11:30.9 SO: They at least they'd have to be perceptible by or discoverable by human perception.

1:11:38.1 SC: I think it would be, at the very least, very nice to have more than one data point, when it comes to an ecosystem and a biosphere. That would help us both philosophically and scientifically.

1:11:48.8 SO: It certainly would, although those statistical arguments are powerful ones, aren't they? Just about, you know, how many possibilities there, there are for life and just how unlikely it is that we're really alone here in the universe.

1:12:02.8 SC: We'll see. It's a difficult question, but an empirical one ultimately. So Samir Okasha, thanks so much for being on the Mindscape Podcast.

1:12:09.6 SO: Not at all. My pleasure. Thank you, Sean.