← New Books Network
New Books Network
Kevin J. Mitchell, "Free Agents: How Evolution Gave Us Free Will" (Princeton UP, 2023)
00:32:26
An interview with Kevin J. Mitchell
Loading summary
Transcript33 lines
- [00:01]
A
Welcome to the New Books Network.
- [00:05]
B
Hello everyone, and welcome to another episode of the Princeton University Press Ideas Podcast, a joint production of Princeton University Press and the New Books Network. I'm Mark Klobus and today I'm speaking with Kevin J. Mitchell, author of the book Free Agents How Evolution Gave Us Free Will. Kevin, welcome to the New Books Network.
- [00:24]
A
Thanks very much, Mark. Thanks for having me.
- [00:26]
B
Well, thanks for agreeing to be on our show.
- [00:28]
A
Sure.
- [00:28]
B
I was wondering if you could start us off by telling our listeners something about yourself.
- [00:33]
A
Sure. So I, I'm, I work in genetics and neuroscience at Trinity College Dublin in Ireland, and I'm broadly interested in how the brain gets put together. So what are the instructions in the genome that, that specify how to wire a human brain? And how does variation in those instructions lead to differences in behavioral traits and personality and intelligence and also risk of conditions like psychiatric disorders or neurological conditions? So that's been sort of my broad area of interest and it's led me to thinking about some sort of deep philosophical issues like the nature versus nurture question or this or this question of free will.
- [01:22]
B
That is one of the things I found most fascinating about your book, which is that you're showing how this question, which is long consumed us and has long been very prominent in considerations of, say, philosophy and questions of philosophy, is one that you're showing how we can approach it through the study of the biological sciences, the evolutionary sciences. What led you to undertake this book?
- [01:49]
A
Well, there were a few things. One of the ones actually led from a prior book that I'd written also with Princeton, which is called Innate how the Wiring of Our Brains Shapes who We Are. And one of the sort of messages of that was that really we're not blank slates. We're born with some psychological predispositions, our individual natures and those natures affect our personality. And that really, you know, personality traits really are descriptors of patterns of decision making through time. So the idea that, you know, is that our decision making is informed by our personality traits which is affected by our, our genetics. And an obvious kind of question then is, well, if that's true, then am I really in charge of things? I mean, I didn't decide how my brain would be wired. So does that mean that I don't really have free will, that I'm just kind of a pre programmed robot in a way. And that's a concern that was, you know, many readers raised, but it also was nagging at me, frankly. And then there was also, there's a kind of a Current trend these days for many people in neuroscience, especially to declare that free will is an illusion. To say that actually, you know, what, you think you're making decisions, and it's sure it feels like that, but really it's just all the neural circuits firing and we can see them. Look, we can put you in a scanner. We can see them working. We can go into an animal and we can activate different sets of neurons and make them do A or B, make them sit up or roll over, or we can change what they're thinking. We can implant memories in mice, you know. And so all of this neuroscience makes it really, really tempting, I think, to fall into this idea that really it's just neural circuits firing. It's not, you know, all of the rest of it, the mental states and so on that we are experiencing don't really have causal power. It's just a mechanism that's. That's whirring away and driving the behavior of the organisms in which those mechanisms happen to sit. And so those, those sort of worries, you know, they're not easy to dismiss, actually. And there's a reason why people think like that. And so what I wanted to do was try and figure out, is there a way to think about free will that is still naturalistic and scientific? It doesn't go the whole other extreme where you'd have to posit some kind of immaterial spirit or a ghost in the machine that somehow gives you an immaterial self control over. Over the mechanisms that. That doesn't seem to work. So what I wanted to do was find a way, a framework in which we could see how you can have free will and agency as biological capacities that aren't mystical or mysterious, whereby the self, the organism, can really be in charge of what it does.
- [04:49]
B
You begin that process by considering that question of free will. And you've already talked about it to some extent, but I was wondering if you could, perh. Perhaps elaborate a bit upon your approach to free will and how you conceptualize it for purposes of examining it in this book.
- [05:09]
A
Yeah, well, so one of the things I do is avoid locking in a definition right from the get go, because I think that has caused all kinds of problems. People define free will in a certain way and then have arguments about it, but it's just because of the way it was defined. So, for example, if you define free will as having absolute freedom to act free from any prior causes whatsoever, then that's just an incoherent notion. I mean, an organism or an agent that's Doing things with no prior causes in mind would be just a random behavior generator. So it's a kind of just a notion that makes no sense. It inflates free will to a level where only a mystical, magical kind of process could actually satisfy those criteria. And the same is true for what you can think of as a dualist way of thinking where you can say again, there must be a ghost in the machine, a different kind of thing from the physical stuff, so there must be some mental stuff or substance or entity in there that that's controlling things. But again, that. That basically rules itself out right from the get go because it's. It really relies on a kind of a supernatural framing. So what I wanted to do instead was just start with the phenomenon. Let's agree on what we're talking about to begin with. And the phenomenon really is that we feel like we make decisions. So we go around all day long, we're making choices. We're not doing it at random. We're doing it for reasons. And we know that because we can talk about those reasons and we can even tell each other about those reasons, and we can wonder about other people's reasons for doing things. So that phenomenology is really the bedrock of our everyday experience, that we're in control of ourselves. It's not that we're acting without any constraint, but that we have some degrees of freedom and that we ourselves can exercise our own agency to decide what we do. And so with that kind of framing, then the question is, well, how could that be? Because you kind of run into this problem that once you start thinking, well, maybe it's just the neural circuits, maybe it's not me, but why stop at the neural circuits? Maybe it's just physics. Neural circuits, after all, are made of atoms and molecules and electrons and quantum fields and particles and so on, and they're going to obey the laws of physics. So how could me thinking something or wanting something have any effect on what happens to all those molecules and be able to cause my physical being to do certain things? So there's a mystery there, and that's the challenge, is to try and figure out how to address that. And one way to do it is to just tackle the metaphysics head on and try and think in the abstract, well, how could it be in principle that that kind of thing could happen? But the approach that I take is very much grounded in what actually led to our having these capacities, or at least seeming to, which is the process of evolution. And so what I do is really start at the beginning and say, well, look, if you want to understand how it is that a human being can control what they do, you really have to ask how does any living thing control what to do? And one way to get a handle on that is to start with the simplest things like single celled organisms, and then see how evolution built the control systems that allow them to manage their own behavior over time.
- [08:45]
B
That's one of the things that I really enjoyed about your book, was you take it, you introduce the reader to the approach by going back to the beginning, by starting with the most basic elements and then showing as. As life gains in complexity, so too you're building to the explanation of how it works in humans. I was wondering if you could elaborate upon that, that in. In a bit more detail in terms of the emergence of life and how this process develops and how in that emergence over time beings develop agency, which, which I thought was. It was. Was very fascinating examination there.
- [09:26]
A
Yeah, thanks. I think it's a. I think for me it was a key way to ground some concepts that are otherwise a bit vague sounding. Things like purpose and meaning and value. They don't sound very scientific, but they are and we can naturalize them. And one way to do that is to think, well, what does it mean at all to be a living thing? What is a living thing really? And ultimately it's a set of processes that are organized in a certain way and that are capable of doing some work to keep themselves organized that way. So even a single celled organism is basically, it's got, you know, some physical structure, a bunch of molecules organized a certain way, and then it's got a lot of internal dynamical processes of metabolism and so on, and that pattern persists. And amazingly, even though the bits that make it up, you know, are in flux, they'll go in and out of the organism. It may not have any of the same atoms in it, you know, from one day to the next, but the pattern persists. So it's a bit like a tornado, right? The pattern is there even though the individual bits are flying in and out of it. The difference, of course, is that a tornado blows itself out pretty quickly and living things do some work to make sure that they don't blow themselves out. So really the living thing in that sense, first of all, it's very dynamic. It's this set of interlocking processes and it's an integrative whole entity. And that's really the crucial thing, is that the whole entity there is doing this integrative work where all of the parts are constraining all of the other parts to make up that whole thing and keep it persisting through time. Now one of the challenges, of course, is that, say a bacterium may be configured a certain way with its metabolic pathways. It's taking in some nutrients and it's performing its metabolism and it's getting rid of waste products and so on. But the environment may be changing and so it may be that the food it was eating disappears, maybe it uses up all the food. So it might, for example, have to reconfigure its metabolism to make use of some other food source as the environment changes. But another way that it can keep itself persisting is to be able to move. And that's a new thing in the, in the universe. Things that can move themselves about, you know, they're not just acted on by other forces. They as an entity can move in the world. And once that evolved, then it, it became really adaptive to be able to know where to move. And so organisms became aware of things in their surroundings. They have sensors on their, their outside, even single celled organisms do. And they can detect things like sugars or dangerous chemicals or whatever, and they can move towards the sugars and away from the dangerous chemicals. So even in that simplest sort of scenario, you've got purpose, which is to persist, you've got value, which is whether something helps you persist or not. So moving towards a sugar is a valuable thing to do. You've got meaning, which is the signal that comes in, into, you know, it's not a, the bacterium is not being driven around by energy or forces. It's responding to information about what's out in the world. And it's configured in a way such that its actions are appropriate response to that. And it's also, you know, really integrative and holistic because the bacterium is not just responding to one thing at a time. It's got loads and loads of signals out in the world. It's got all kinds of context that it's, that it, that it integrates in order to behave adaptively. And once I think we, once we see that, then, you know, the idea of an, of an organism being an integrative whole thing that controls its own behavior in the world is not so mysterious. And from there we can just sort of elaborate as we go through the evolutionary pathway that eventually led to human beings.
- [13:22]
B
You make it clear that what you're describing, those, you know, choices, shall we say, that that are being made, are not necessarily simple, reactive or, or incidental that, that they are that they are indeed responding to inputs and, and making choices. I, I found, for example, your, your, your examination of determinism to be, to be very interesting because you talk about how there are various factors involved. It's not simply a matter of like, you know, the old, you know, getting your, your knee knocked and just having it reacted. There are choices being made, but they're three key factors that you identify.
- [14:01]
A
Yeah, and I mean the key thing is to get away from a kind of a mechanistic view of causation where it's just the old sort of atomic billiard balls knocking into each other. Newtonian sort of view, or even an electrical kind of a view of what's happening in the nervous system. So when organisms evolve that are multicellular, then they have a problem in that they need to coordinate their bits. So if they're going to move in the world or they're going to act on the world, they need to be able to coordinate their parts. And the nervous system evolved as a means to do that, to connect up muscles and coordinate actions, but also to sense what's out in the world. So again, just like in the single celled organism, what's important is to know what's out in the world and then to know what's a good thing to do about it that will be an adaptive thing to help you persist through time. And so really what's driving these, or what, what the nerve nervous system runs on is it's not electrical energy, it's not like electrical circuits that are just driving each other. It's meaning, right? These neurons want to know what's the meaning of this signal that I'm getting. And the organism wants to know what's the meaning of these internal neural patterns that, that the organism can infer what's out in the world and link it to its sort of stored memory and drive or choose one out of many possible physical actions in its repertoire in response to that, based also on its own motivations and its own internal states. So we have all of this biology that's really designed as a big control system to keep organisms persisting for as long as they can and of course, you know, ultimately reproducing and so on. But the, but the important thing is to, is to realize that's an informational system, right? It's a system of meaning. It's not a system of just mechanism in that Newtonian kind of sense.
- [16:04]
B
It really is fascinating to think about how these, you know, the sophistication of these, of what, you know, for us are Incredibly simple organisms.
- [16:16]
A
Yeah.
- [16:17]
B
How do we get from there to the development of higher order features?
- [16:22]
A
Yeah, so. Well, one of the ways, as I said, is when multicellular organisms arose, and there's a whole sort of really fascinating history about how what are called eukaryotic organisms arose. So fungi and animals and plants and so on. And one of the key developments is the invention of the nervous system. And so if you have, in a single celled organism, you've got some sensors, you've got some motors that can drive the bacterium around, say, and you've got a few bits of biochemistry intervening that will relay signals and that will also integrate signals both over multiple kinds of signals and over time to drive an adaptive behavior. But really, sensation and action are pretty closely coupled in single organism simple organisms. In more complex organisms like us and many animals, we've got neurons that can sense things out in the world, we've got neurons that can control muscles so they can drive action. But we've also, as the nervous system got bigger and bigger, got all these intervening layers, and that's where the sophistication really emerges. And it does so in multiple ways. First of all, it allows us to make sense of things like visual signals. So, you know, a bacterium or simple things like worms and so on, they can respond to molecules that they directly contact. So they're smelling, smelling things effectively, but that means they're directly contacting them, or they can respond to touch. But their senses are all very immediate. It's things that are very close to them right now. So they inhabit the here and now, and they're very reactive. But when vision evolved then, and hearing as well, those are senses that don't directly detect objects. They just detect disturbances in the electromagnetic magnetic field or vibrations in the air. And then the organism has to do some work to infer, well, what's causing that pattern of photons that are hitting my retina or vibrations that are hitting my eardrum. And so we have all this sophisticated levels of circuitry in the nervous system that let organisms infer what are the objects that are out in the world, because that's what they care about, right? They don't care about photons hitting their retina. That's not meaningful. The meaningful information is here's that object, here's this object out in the world, and then they can link that to their stored knowledge about things. So the other huge, huge advantage of the nervous system is it's designed for learning. It's designed for making connections between different things. It's designed for knowing that you know, the connecting the properties of things, but also connecting events in time so that when A occurs, B always occurs after it, or if I do C, D is the result. And organisms can learn from that so that when they see what's out in the world, they can have a sense, based on their own past experience of what's a good thing to do, what are the causal properties that are at play in that scenario. And then they can evaluate the possible outcomes of their actions internally without having to test it out in the world and possibly die. Right? So they can sort of simulate it ahead of time and say, you know, what if I do this, that's not going to turn out well, I shouldn't do it. And so thereby, you know, they both evolutionarily, as the nervous system got more complex, the powers of agency got more sophisticated. But that's also true over an individual lifetime as individual organisms learn about the world and accumulate causal knowledge and that really gives them causal power in the world, the more that they know about how the world works.
- [20:03]
B
That's what that step that I thought was, was, was especially intriguing that how you go from the notion of simply, you know, reactive to where you start this, this, this process and the way you define it in terms of meaning. I was wondering if you could perhaps elaborate a bit upon how you use the, the, the notion of meaning in terms of describing this, this development towards, or how it connects to the concept of free will.
- [20:32]
A
Yeah, well, so one of the key sort of issues, key counter arguments, is this idea that it's all just neural circuits firing, right? I can say what I want about these neural circuits. Meaning, Meaning something like representing there's something out in the world or representing an internal state or a belief or a desire or a motivational state and so on. But the rejoinder can always be, look, it's just complicated electronics. Basically it's going to drive whatever is going to happen. What they mean doesn't have any causal power in the system. And actually I want to reverse that view. I think that's completely backwards because actually if that were true, then it should be the case that if you change the details of the neural firing that what happened should change. And that's not always true. In fact, it's often not true at all. So neurons are noisy in the first instance. They're not really, you know, it's not the fact, it's not the case that all the details of all the neural firing has causal power or even percolates through the system. A lot of those details are lost. What Matters is what the patterns mean. So for example, when one neuron is monitoring the activity of another one, it may be sensitive to the rate of firing over some time period. And within that time period, it doesn't matter if there's a, you know, 10 spikes that come all at once, or if they're spaced out, or if it's 5 and then a gap and then 5. What matters is it's 10 spikes over that time period. So the meaning of that pattern is in the rate of firing. It's not in the details. And that's true when you look at populations of neurons where they collectively encode something like, for example, in the visual system, where we abstract information about an object in the world or even get up to a level where we know there's a face and we recognize a face, and it doesn't even matter where it is in the visual world, visual scene, what angle it's at, we know who it is. So we have an abstract representation of that. That doesn't depend on the specifics of the neural firing. So what's important to what happens is the meaning of those states, the conceptual content that they have. The neural firings at any given moment are incidental. They may be arbitrary. They may change from instance to instance. That's not where the causal power lies. The causal power is in the meaning. So the neural patterns only have causal power in the system by meaning A or B or C. And I think that's a way of flipping the script a little bit. So rather than saying all this cognition can be reduced to neural firings, I would say no. The neural firing is only have causal power because they represent these cognitive elements. I think that's a really, really key perspective shift. I think that, that, that I hope can get us beyond a kind of a simplistic reductionism.
- [23:26]
B
Now, what you described here so far is the process by which biologically we've developed the, these cognitive abilities and how they, you know, can, can you can stand for free will. So how can you make the connection then that, that you do? Because to be honest, it's a little, the, the complexity kind of eludes me. But so then how does this simply not simply be, say, for example? Well, this is just simply more complex, reactive process. I mean, where does that element of choice come in?
- [24:06]
A
Yeah, well, so I would say rather than thinking about choice, I like to think about choosing as an activity. So any organism, when it encounters some scenario, will do this kind of calculation. It'll see what's out in the world. What do I know about it. What are my internal states, what are my motivational drives for various goals? And also crucially, what are my long term activities? Because any organism doesn't just react to things in the moment. We sustain patterns of activity through time. And that's especially true in humans because we have plans that can last for decades, right? We make commitments, we have policies, of course, we have a lot of habits that we've developed over our lifetime. But all of those things are there right in any given moment. We're not just reacting to things, we're carrying out some sustained behaviors and we're accommodating to new things as they happen. And so I like to move away from this notion of just momentary choice between binary decisions because it's just a super, super artificial framing. And of course, that's how we study things in the lab. Tightly controlled experiments, isolated stimuli, binary decisions. It's even how philosophers often design their thought experiments, funnily enough. But it's not how organisms encounter the world. And they don't just react to things, they sustain patterns of behavior through time. And so, you know, ultimately, if an organism is doing that, and it's doing this holistic, you know, integration across all these goals, it's optimizing its behavior over many, many sort of different stimuli, many of which, you know, it has ambiguous information about. But it's, it's trying its best to figure out what's the best thing to do in a really holistic way that's not being driven just by isolated parts. It's absolutely the whole thing working, different elements working, you know, over. They're concerned with different time frames, you know, some concerned with very long term goals, some concerned with just what are the motor actions I need to do to execute this behavior. That's all sort of nested together. And once you've got that, I don't see what else you would want to say. It is the organism as a whole, as a whole entity that's controlling its behavior and that's choosing between options in any, in any given moment. Now, the one sort of further element in humans that's really crucial is that we have an extra level to these control systems. It's not just that we have cognition that has, you know, beliefs about what's out in the world and some representations of desires and motivations. We have an extra level that can kind of look down on those and think about those thoughts and reason about those reasons. And importantly, we can articulate them, we can tell each other about them, which enables this incredible capacity for collective action. That I think really sets humans apart from other organisms. So we call that metacognition or introspection or executive function. And that for me is the level at which I'm happy to say, look, that's free will. That's doing things for reasons. It's being potentially at least aware of your reasons, being able to rethink them, to change them for other reasons. And so once you're there, I think we're at a point where I'm happy to say, yeah, that's a fairly, that's that that meets the fairly standard, you know, description of the phenomenon that, that we set out to explain.
- [27:37]
B
Yet your book does more than simply explain the phenomena. It also considers the implications of it, both for ourselves as beings and also for this, this new hot topic that we, we seem to be seeing everywhere nowadays, which is this question of artificial intelligence. I was wondering if you could perhaps explain how, how this helps us to better consider these broader questions about identity, the notion of free will and also the notion of this concept of an artificial intellect and whether or not it too by this criteria possesses free will.
- [28:16]
A
Yeah, well, first of all, I think it sounds like a very non sensationalist conclusion, but actually I think despite what you might hear from other neuroscientists or physicists or other philosophers, our notions of free will, I think are pretty good. And our systems, in the legal system, for example, of thinking about responsibility, they're really framed around the fact that free will is a capacity and that it varies, it varies between people. So young people may not have as much cognitive control as adults do and someone who's under the intoxicating effects of drugs may not have as much control and so on. So we already, I think, have a pretty sophisticated understanding of those things. I don't think it really has to change. I think it may be informed as we better understand the biology. And we can see cases where, you know, cases like dementia or schizophrenia or things like that, where we will have a more nuanced understanding of how those impinge on cognitive control and agency of persons more broadly. I think even scientifically or philosophically, I think we can defend biology as a science unto itself. It's not just complicated physics. Organisms are doing things for reasons that physics just doesn't encompass. I think that's important and I think agency is a really central idea in biology that hasn't really been realized as a hugely important factor that distinguishes living things from non living things in that living things do things. They're not just acted upon, they can act as causes in the world. And then for artificial intelligence, I think the systems that we have right now are very, very capable of doing a limited number of tasks incredibly well in some cases. But to get to artificial general intelligence, I think we may have to build an agent. I think we may have to build something that is able to intelligently behave in the world, to intervene in the world causally, to learn about causal relations in the world, to have some reason to care, to be what philosophers call a locus of concern, and to be me. And we may have to embody it in some way in order to do that. And then if we start thinking about that and we look to what we know about how evolution has built natural intelligence and we can see the architectures that enable it, in principle, we may be able to duplicate those in an artificial system. And then we have these huge, huge ethical questions about whether we should. And that's the whole other area that, you know, I, I, I cop out in the, in the book and I just leave it hanging for, for readers to consider.
- [31:00]
B
Well, with all due respect, I, I don't blame you because not only is it a huge question, but you tackle so much as it is in the book.
- [31:07]
A
Yes, it would have been a lot longer if I'd have gone into that. Yeah.
- [31:11]
B
Well, we appreciate the time you've taken to speak with us, but before we go, could you tell us what you're working on now?
- [31:16]
A
Sure. What I'm working on right now is thinking kind of back to my roots a little bit about development, Thinking about how the form of an organism is encoded in its genome. And people argue about that term encoded. It sounds very computer like and it isn't like that. It's like something, but it's not obvious what it's like or how we can understand basically why kittens have, why cats have kittens and dogs have puppies, and also why kittens and puppies have different nature between them. You know, where do those, how do those instincts get wired in to their brains and expressed in behavior? And so I think that's an interesting topic. It's got me musing on all kinds of things. So I think I'm going to spend some time thinking about that.
- [32:04]
B
It sounds like a fascinating topic. I look forward to seeing what you do with it.
- [32:07]
A
Yeah. Thanks very much, Kevin.
- [32:09]
B
Thank you very much for taking some time out of your schedule to speak with us. I hope you have a wonderful day.
- [32:13]
A
Thanks, it's been a pleasure. And thanks for having me on.