Subscribe
Share
Share
Embed
Who thinks that they can subdue Leviathan? Strength resides in its neck; dismay goes before it. It is without fear. It looks down on all who are haughty; it is king over all who are proud. These words inspired PJ Wehry to create Chasing Leviathan. Chasing Leviathan was born out of two ideals: that truth is worth pursuing but will never be subjugated, and the discipline of listening is one of the most important habits anyone can develop. Every episode is a dialogue, a journey into the depths of a meaningful question explored through the lens of personal experience or professional expertise.
PJ Wehry (00:27.854)
Hello and welcome to Chasing Leviathan. I'm your host PJ Weary and I'm here today with Dr. Mishwita Chiramutta and she is the senior lecturer in philosophy at the School of Philosophy, Psychology and Language Sciences at the University of Edinburgh. And we're talking about her book, The Brain Abstracted, Simplification in the History of Philosophy and Neuroscience. Dr. Chiramutta, wonderful to have you on today.
Mazviita (00:55.918)
Yeah, thanks so much, PJ. Nice to be here.
PJ Wehry (00:58.916)
Dr. Chiromuta, why this book? mean, such a controversial topic, it is a topic that is tied to so many philosophical issues. What did you want to bring with The Brain Abstracted? Why this book?
Mazviita (01:15.692)
Yeah, so this book really stems from my long standing kind of experience with and reflection on computational neuroscience. I initially I studied philosophy and psychology and I was I've always been interested in the connection between the philosophy of the mind and the mind brain sciences. But as I was going forward with my through my psychology degree, I became more interested in
neuroscience, so the biological basis of cognition and this idea that there is this material object in our heads, the brain, which is the seat of consciousness, which is what enables us to perceive the world, all those different things. So that prompted me to go ahead studying vision from a neuroscience perspective. So that ended up being my PhD topic. I was doing psychology type experiments in psychophysics, but actually trying to
explain the psychological, psychophysical data in terms of responses in neurons actually within the visual cortex. And one of the things that struck me was that we were using computational theories that bridge between psychology and neuroscience. So constructing these models of responses of neurons in our heads, using mathematical equations to simulate what our brains are doing.
and then using this as like an explanatory framework to make that link between actual psychological experience and these physical processes in our brains. So one of the things that struck me at the time was when I was building these models or learning to do this kind of modeling was just how much of a simplification these mathematical formulae were. Admittedly, this was a much earlier generation of
models in neuroscience and the things that we see today coming out of deep learning AI are much more mathematically complex, much fancier. But still the very fact that science relies on abstraction, on mathematicalisation, on sort of representing this messy biological complex reality in terms of a few equations.
Mazviita (03:35.758)
Obstructing is really central to the scientific projects. A lot of how science works is using these kinds of techniques. It originates from the physical sciences and it's been expanded elsewhere. But then posing this question, given that we know how complicated the brain is, are there limits to how much we can use these abstractions and simplifications and still be telling the right story about the brain or in particular about the relationship between the brain and mind?
PJ Wehry (04:08.196)
Great answer. Thank you. What would you say is kind of the current, I think just to give our audience a background here, what is the current state of neuroscience and philosophy? I think there's kind of a academic, disciplinary or disciplines political side, like the different disciplines and how they see each other. I mean, not that academics ever have like
Mazviita (04:31.362)
Yeah.
PJ Wehry (04:36.996)
political things going on between them. And never, Truly just like disinterested learning. But then also kind of how, where's the state of things as you see it, what are kind of the current questions and what is some cutting edge stuff that you're interested, you're excited about?
Mazviita (04:57.196)
Yeah, I mean, I would actually sort of begin answering that question actually by going a bit backwards in time, and think, talking about how this intersection between philosophy and neuroscience first originated. So there's this discipline known as neurophilosophy, which is most associated with Patricia Churchland, also her husband, Paul Churchland. So neuroscience was back in the 80s, actually a young discipline. I learned at a conference recently that the first
departments of neuroscience were only founded in the 70s. In fact, the word neuroscience is actually quite a new word. It's from the 60s or 70s. So neuroscience was still young and in its infancy or early, early years, primary school years, philosophers like the Churchlands got together and looking, we now have this science of the brain, which is the basis of the mind. Philosophers have been doing philosophy of mind for literally millennia.
not made progress, we still have the mind body problem, we still don't know if there's free will, we still don't know how the brain gives rise to consciousness, what emotions are, what vision is really, why don't we look to the sciences? So neurophilosophy was precisely this movement within philosophy which said that philosophers should be wedded to neuroscience precisely because neuroscience will give philosophy the answers that it couldn't figure out by itself.
Now that's a different thing from philosophy of neuroscience. But really I want to point this out because this connection between philosophy of and neuroscience really started out with this aspiration for neuroscience to be able to answer philosophical questions or at least help solve philosophical questions. But as I see philosophy of neuroscience as a discipline, that kind of started a bit later and that was a bit more of a modest goal of just saying
PJ Wehry (06:27.159)
okay.
Mazviita (06:51.692)
We have philosophy of science, which is just looking at the epistemology of science. These general questions like what is an explanation? What are laws of nature? Scientific realism is an important question within that field and within physics it asks questions like are atoms real or do theories positing atoms just give us a handy way of predicting observable phenomena? So philosophy of neuroscience
looks at those epistemological questions, but specific to things that come up with neuroscience themselves. So it's interesting that that started a bit later, I would say, really, in the 90s onwards. But it's, I would say now overshadows neurophilosophy. There's more people doing philosophy of neuroscience than neurophilosophy now, though I would say perhaps this disciplinary label neurophilosophy has faded a bit in the background.
because that idea of the methodology that Patricia Churchlin suggested, which was that philosophers interested in the mind should be looking a lot at the scientific data, that's become very mainstream within philosophy of mind. So the majority of philosophers of mind themselves, say if they're studying perception, they'll be like at least somewhat familiar with the psychological and neuroscientific literature on those faculties. yeah, philosophy of
PJ Wehry (08:04.708)
Hmm.
Mazviita (08:21.754)
neuroscience. It's been around a couple of decades a bit more but the Society for Philosophy and Neuroscience has only just been founded this year so it was in St. Louis last weekend for the inaugural conference of this new society. So was exciting, we finally have a society.
PJ Wehry (08:42.402)
That's awesome. Sorry, I'm looking up now because, I should remember his name off the top of my head, but I am not as good with names as I should be. We had...
Dr. Christian Miller on we had I had I don't know why I have the royal we going on. had Dr. Christian Miller on and he just reminded me what you're talking about. He has combined the kind of classical study of ethics with empirical psychology. And it's just becoming more and more mainstream to include the and what's interesting to me is I think sometimes and
Mazviita (09:02.956)
Yeah. Right.
Mazviita (09:14.594)
Mm-hmm. All right.
PJ Wehry (09:26.016)
maybe I'm mishearing this. And so I'd love to kind of say this back to you. Sometimes there's this greater, this is this kind of really big goal that science will explain everything. And sometimes it does. But a lot of times it's it becomes very clear that it helps a lot, but it doesn't explain everything. And so then like in the case of neurophilosophy, it's kind of fade in the background. But the impact has been felt in that it's added a lot, but it hasn't achieved that great goal of like, well, if we just studied neuroscience, then everything will be answered.
Mazviita (09:31.714)
Mm-hmm.
Yeah. Yeah.
Mazviita (09:40.782)
Yes.
Mazviita (09:45.646)
Right.
PJ Wehry (09:54.798)
And that's different from philosophy of, if I understand philosophy of neuroscience, is what are the underlying assumptions and questions we should be asking? What are the right terms for studying neuroscience? Is that, am I in the right ballpark? I think I just want to make sure I'm tracking with you.
Mazviita (10:11.244)
Yeah, sure. I would say, say what are the right times for studying neuroscience that leaves it open, whether the answers that philosophers of neuroscience come up with are most relevant to scientists themselves. So scientists themselves, they need to have their methodology, need to plan the best ways of investigating things. So there are certainly philosophers of science that want their results to be directly
applicable to scientific practice. And that group of philosophers actually has their own society, the Society for Philosophy of Science in Practice. And Hsiao-Chang, professor at Cambridge, is one well-known protagonist in that kind of area. So the idea that, know, philosophers sitting back and think about methodology and epistemology, if they're doing anything right, it should be that
PJ Wehry (10:42.83)
Yes.
Mazviita (11:07.52)
if we follow the advice of philosophers, science will go better. Or, know, the idea that scientists themselves, because they're so busy with their labs, they have to get funding, they have to keep the production line of data going. They don't have time to sit back and reflect on methodology as much as they like to. Maybe scientists in the past had more time for that. 100 years ago, was very standard for scientists to have a lot of philosophical training.
write more philosophically than it is today. So this idea that philosophers can be doing some of that sort of complementary work to the science. have to say, yeah.
PJ Wehry (11:44.676)
Do you mind if I ask a quick question? Do you think there's a way that I understand what you're saying, but I want to make sure I'm following you. Do you think that the fact that they're not able to have this kind of cross training is more a function of the growth of knowledge in general, or is it like a lack of funding or a cultural problem?
Mazviita (12:07.182)
I think it's a factor of the scaling up of science within the last hundred years was immense. mean, before the late 19th century, I mean, being a scientist wasn't even a profession really. The word scientist is a Victorian word. They needed a word for this person that spends all their career doing experiments and collecting data and publishing in journals. And before that...
PJ Wehry (12:12.919)
Okay.
PJ Wehry (12:16.824)
Yeah.
Mazviita (12:36.206)
natural philosophers or for a while they were men of science. But yeah, the idea of having a career as a scientist is quite new. For there to be thousands and thousands of people across the world all doing this, all doing experiments and producing data, that's quite new. So of course in the past it was easy to know all of the experiments going on in your own field, plus a few other disciplines within science, and also spend a bit of time on philosophy. think now there is
the scope that as science has become scaled up and more and more is going on in tiny and tinier sub-disciplines, that just to be competent in any sub-discipline you have to just be focused on that. But then also it was this cultural question. I think culturally we find more value in specialisation as opposed to regret than in the past. I mean, I think we admire the specialist.
more than the Renaissance man if you try and be a Renaissance man now you look like a dilettante so it's not
PJ Wehry (13:37.956)
man. Bonus points to you for using Dilatant. love, that's great to have come up in conversation.
Mazviita (13:49.986)
But yeah, so follow up what I was saying, going with the complementary science and philosophy of science in practice. That wasn't really how I was seeing my project in the book. I didn't want what I say to be about the limitations and possible problems with simplification, or at least oversimplification, to be read as a lesson to neuroscientists, do your work differently.
PJ Wehry (13:56.28)
Yes, please.
Mazviita (14:17.774)
I very much present it as an issue for people that want to use neuroscience to inform them about their philosophy of mind. So when you look at neuroscientific results, you have to interpret them with this knowledge that there is a vast amount of simplification going on and that should affect how you read off the results. Say if in a publication, visual neuroscience that tells you these computations going on in the visual cortex.
you should be aware that the computational framework is itself this idealization of what is actually going on in cortex and it's leaving out a lot of biological details, which we're not sure if it's justified to do that.
PJ Wehry (15:00.578)
Right. Because if you measure something, there's always little bits nipped off or added, right?
Mazviita (15:06.446)
Yeah, sure. there's things, know, one of the things I talk about a lot in the book is how data collection in science, neuroscience, as is mostly done, relies on a lot of experimental controls, which raises questions about the generalization that you can make from the experimental behaviors and observations to real world situations. So things like that.
irrelevant if you're trying to do that thing which neurophilosophers originally wanted to do it like combining neuroscience with philosophy of mind.
PJ Wehry (15:47.198)
I want to apologize because I keep asking you controversial questions, but then I'm like, well, you did choose to study this. it's just kind of like, no, I was I wanted to ask you because a major part of your book is you go through the different kinds of simplification. And of course, simplicity is a controversial topic as well. Yeah.
Mazviita (15:54.328)
Thanks.
Mazviita (16:08.878)
I didn't realise, I'm surprised you say that.
PJ Wehry (16:13.712)
Okay, I well I mean That's what I've had other guests tell me so, know, I mean I Know that's I mean, I actually I read You know, I was going through what you said. I was like this seems pretty clear to me. But you know, that's No shade to other guests I What are the different kinds of? What are the different kinds of simplification and how are you using them in your book?
Mazviita (16:20.014)
Right, no that's interesting.
Mazviita (16:41.494)
Yeah, so in the first chapter of the book, I took about three main classes of simplification. You could sub divide these further, but the ones broadly most important to neuroscience, what I say further on in the book, are sort of mathematicalization, as I mentioned before. So the very fact that you have some concrete object or system, it's out there in the natural world. You don't just take it in as it is, but
modern science has always sort of begun investigation with this question, what shall we measure? In other words, what shall we call the variables of the system that we can quantify, hopefully relate to one another in equations, which if everything goes well, we might one day establish as laws of nature. So the very idea that nature is mathematisable hasn't always been obvious to people. It's something of a 17th century.
invention, it's been around for a while, but it wasn't always there. But the point made by various philosophers, especially in the early 20th century continental tradition, is that when you mathematize, you're isolating only the quantitative properties of this thing. And there are whole bunch of qualitative properties. The most obvious being, so whenever we look at things, there's a whole lot of colours.
Colors as they are, as we experience them, we can't say a number. So I know we have color spaces and we can sort of attach number labels to colors, but the particular colorfulness of the color is not, it's something qualitative. It's not something that makes sense to just define numerically. So you're abstracting away from that. You're sort of, you're going along with this idea that we can safely leave behind in our scientific representation, all of those qualities.
properties and just isolate ones that we can represent numerically. So that in itself is a simplification because it's cutting down the amount of information that is there in your representation.
PJ Wehry (18:55.948)
Which is near and dear to your heart because, I mean, that's what you studied for your PhD, right?
Mazviita (19:00.116)
yeah. So my first book was on colour perception and the question of colour and philosophy actually. So it was actually because of that topic that I started reading that those philosophers talking about the qualitative and abstraction and that then influenced my next book. So yeah, that was what led me to this this issue. So the next sort of
General mode of simplification I talk about a lot is the forming of analogies. So this is important in neuroscience because the major theoretical framework that we have in neuroscience at the moment is the computational one, which says that what's going on in the brain is analogous to the information processing, which happens in artificial computers, even digital computers, that it makes sense to talk about the brain as taking inputs from the environment, doing some...
equivalent to number crunching. So in the same way that the number crunching in your machine in front of you will be done in voltages being shuffled around and these are interpreted as zeros and ones, binary numbers, that neurons themselves have something like a binary code or maybe a binary analog hybrid. So these ideas have been around for a long time. And what I argue in the book is that
This should not be taken literally that the brain is a kind of biological evolved computer, but that neuroscientists are drawn to this analogy between brains and computers because it serves as a simplification. So if you only focus on the properties in common between brains and computers, you basically have this license to ignore the biological background of
neural processes. So there's so much going on in the brain in terms of cells, metabolism, neurotransmitters, all kinds of stuff. But if you just say, but what really counts for cognition is computation, then you can reach to this very high level of abstraction. It just ignores all of that messy underground stuff, which is just, that's just the implementation. That's just the hardware. What we really care about is the software. I see that analogy between cognition and software.
Mazviita (21:25.582)
That was something that was quite attractive to people.
PJ Wehry (21:29.968)
And if I understand you, like that also leads into mathemitization. I realize I'd have a hard time with that word. Easier, right? Like computers love math. And so that makes that easier. Is that a?
Mazviita (21:35.692)
Right, yeah exactly. Yeah exactly, in fact that's what I say in chapter four that the mathematicalisation of the brain depended first on this analogy being available.
PJ Wehry (21:53.602)
Mathematization. I got it. OK, they were good. I actually, and this is interesting to me, there's a Dr. Megan Sullivan at Notre Dame taught.
philosophy of consciousness, you know, she's got maybe it was intro to philosophy, but she's talking through that problem with her students and she's been teaching, I think 15 to 20 years. And when she first started teaching it, she gave as an example, the computational like that your brain is a computer. And so how many of you think that's the best model for understanding the brain and maybe one or two people in her class raise their hand 20 years later, by the time she finished teaching it.
Mazviita (22:13.742)
Right.
Mazviita (22:24.11)
All right. Yeah.
Mazviita (22:35.854)
Right.
PJ Wehry (22:36.644)
95 % of the class is raising their hand. And that's because people understand computers better, right? Like there's because the point of the simplification, the point of the analogy is to make it available to us. But that doesn't mean that we're actually understanding the complexities, right? Is that,
Mazviita (22:56.206)
Yeah, exactly. mean, myself and plenty of other consumer users of computers, can't say I really understand how they work, but I know what software is because that's my interface with the computers and I know I can deal with the software even though I know nothing about the hardware. And so it's a very intuitive distinction. Could I just...
PJ Wehry (23:08.344)
Yes.
Mazviita (23:19.214)
I'm getting a fair amount of background noise through the window. I don't know if it's picking up, but I might just close the window.
PJ Wehry (23:25.996)
Okay, yeah, go for it.
Mazviita (23:29.262)
Anyways, distracting me a bit.
PJ Wehry (23:33.148)
No, no worries at all. Yeah, can edit that out. Awesome. So we've talked about mathematization. Geez. right. Analogy. then what's the, I believe there's a third type of simplification you really reference.
Mazviita (23:37.196)
Go.
Mazviita (23:50.476)
Yeah. The third one is reduction. So this is something that is very, very prevalent in other branches of biology as well. The core idea being that when you have a complex system, like any living organism is, or even any living cell, the best way to tackle it is to break it up into its component parts and study those piecemeal.
So often how that plays out in biology is going down to the molecular level because those molecules are the tiny tiny parts of living organisms. So if you know everything that's going on with them, hopefully the bet is that this system is sort of organized from the bottom up. So all of those tiny parts, put them together. That's how you get the whole thing. Just like building a house up of Lego.
So there's this assumption there that the whole organization isn't having this top-down causal effect on how the parts operate, such that when they're within the living context of the body, that they don't behave radically differently from how they do in isolation when you're looking at them in a test tube or something.
PJ Wehry (25:10.062)
And all three of these that you talk about, it is very useful. And I'm very grateful that you've broken down into these three, but they all have a ton of interplay, like reducing it into its smallest parts. It makes it much easier again to mathematize. If you have mathematize, you can put it into the models and the models are types of analogies. so it's all like, mean, these are all great tools for understanding.
Mazviita (25:12.376)
Mm-hmm.
Mazviita (25:25.112)
Ready?
PJ Wehry (25:37.188)
But I think this is, you also sent me a paper, but there's this question of is the brain something that can be finally understood in this way? And I think that's kind of a central question for your project. Is that a fair thing to say?
Mazviita (25:52.302)
Yeah, that is a fair thing to say. mean, I think it's a going concern within neuroscience and whether the methods that have been successful elsewhere in the sciences, especially in physics, can be applied directly to the brain and yield the same results. So interestingly, the majority of theoretical neuroscientists did their initial training in physicists.
physics. So the, the brain being so complex, the thought being like, we need the most highly powered mathematical tools to apply to this. And people get that training typically in physics, to some extent, quantum computer science as well. But I think it's a fair question, given that those methods are developed for things which
PJ Wehry (26:23.093)
okay.
Mazviita (26:50.698)
operate differently from living systems, will that pay off?
So not that physics is this monolithic thing. mean, lot of the physics that neuroscientists are most interested in is physics of mesoscale emergent systems. So not down at the tiny quantum level and not at the macroscopic Newtonian level, but actually looking at how large clumps of atoms get together at the nanoscale and you see sort of interactions between particles and emergent phenomena. So people are thinking...
the most relevant physics being that kind of thing. But still, an atom is way less complex than an individual neuron or any cell in the body.
PJ Wehry (27:37.4)
Yes. And part of it, so I'll just throw this out there, maybe as an analogy, maybe as like a definition, you I don't want to simplify too much, the part of this, so part of the struggle and part of what you're talking about is that we're using the brain to understand the brain.
Mazviita (27:48.238)
Thanks.
Mazviita (28:01.646)
Mm-hmm.
PJ Wehry (28:03.598)
but then we're going to get to a point where we're worrying about an isomorphic map. If you make a map that's perfectly accurate, it'll be as big as the thing you're trying to represent. And in that case, then the person would literally just be understanding their own brain. Is that kind of the question, or am I misunderstanding that? It's like, if I fully understood my own brain, then I wouldn't have room for anything else. Is that?
Mazviita (28:27.18)
Yeah, okay, I have to say that's not an argument that I give in the book. I tried to bear away from these arguments about, well, because we're using our own brains to understand our own brains, it's inherently impossible. Because I think the fair rejoinder to that is that in many ways, that humans are cognitively successful is because we
PJ Wehry (28:31.98)
Okay. Sorry.
Mazviita (28:56.174)
offload cognitive tasks and to technologies, this whole idea of the extended mind, the group mind, I don't think any individual brain is like the locus of any understanding where intelligence is very much distributed, social technological phenomenon. So by that, by itself, the brain reflecting on the brain, I don't think it stands because I think any scientific form of knowledge is more than just a brain.
PJ Wehry (29:02.414)
Mmm.
Mazviita (29:26.354)
So I'm more actually looking at something different from what you could infer just from the limitations of individual human brains, but actually how complex organisations themselves generate more in principle limits to intelligibility. And this is perhaps coming a bit out a bit clearer in some papers that I'm working on since the book's been published. But it comes across in the book.
in chapter seven where I talk about if things are constantly changing, constantly in flux, this creates a problem for just the basic inference pattern of science, which is induction. Biological organisms are inherently changeable. philosopher of biology, John Dupre, has been writing a lot about this recently. He's based at Exeter. And yeah.
PJ Wehry (30:08.26)
Mmm.
Mazviita (30:23.074)
The method of induction works on the assumption that the past is going to be similar to how things are in the future. Therefore, you measure in the past can project forward and allow you to make predictions. There's interesting results in neuroscience more recently, not just about plasticity, which has been known about for a while, but this phenomenon representational drift, which I talk about, is how you can measure the response.
neurons on one day and they seem to drift around to prefer other kinds of stimuli on another day. So just that if we're dealing with something which is much more inherently changeable than say, I don't know, a magnet in a lab, then that presents some in principle limits to how well we can theorise it compared to something which is inherently much more stable.
PJ Wehry (31:17.092)
want to make sure I'm hearing you correctly. Presentational drift. Representational drift. Different thing. Thank you.
Mazviita (31:20.874)
Representation or drift? so this idea that neuroscientists are measuring what neurons represent in any sensory area and it can sort of drift around from one day to next. Yeah, or what the neurons, if you like, likes to see, it's stimulated by, can change from one day to the next. I mean, these results are sort of open to...
PJ Wehry (31:33.922)
It can change from neuron to neuron.
Mazviita (31:45.804)
different interpretations and so forth. But I think as a general point, going back to this deprey thing about biological organisms being processes, not static objects, I think that's just something that you have to confront.
PJ Wehry (32:06.116)
Most people misunderstand what the butterfly effect is, but there's a book by James Gleek on chaos theory So I mean I'm doing this for our audience I you know, I'm trying to But this idea that The real point is that you cannot measure past a certain point and even and those differences beyond the measurement Make it impossible to predict. Is that ours that more where we're headed then?
Mazviita (32:09.537)
Mm-hmm.
Yeah.
Mazviita (32:16.899)
Right.
Mazviita (32:33.71)
Yeah, so I don't talk about the butterfly effect in the book, but that's correct how you describe it, which is with these complex dynamical systems, they're very, very sensitive how the equations will unfold. They're very, very sensitive to the initial inputs that you put in of the numbers. Those numbers are based on measurement.
limit to how accurate the measurement can be. So a tiny error one way or another will give you massively amplified differences in what you predict.
PJ Wehry (33:09.816)
So the example, because I think Dr. Gleek was using it for weather, was that if we could put a sensor at every square foot around the Earth and through the sky, we would still only be able to predict the weather two weeks or less in advance because of the differences within the square foot. And so it becomes really like... And so for me, I think I understand where you're headed because obviously the brain is that kind of system.
Mazviita (33:14.594)
Right.
PJ Wehry (33:37.794)
Right? Like, it's like, it's that complex.
Mazviita (33:37.878)
Yeah, it is. Yeah. And I think the key question that is open in my mind with neuroscience is just given all of that complexity, given all of those nitty gritty details of like molecules swimming around different neurotransmitters and ion channels being constituted and decomposing in the brain, how much of
All of that really matters to cognition. If the computational framework is right, most of that doesn't matter. It's as important as the individual voltage fluctuations in a bit of hardware. Basically your hardware and your computer is designed so that most of the nitty gritty stuff doesn't matter. There's a few variables that need to be tightly regulated. And then the system itself abstracts away from most of the dynamics of it.
PJ Wehry (34:10.276)
Hmm.
PJ Wehry (34:31.428)
Hmm.
Mazviita (34:36.3)
self as a physical system so you don't need to bother about when most of what's going on in the hardware. But if the brain is the kind of thing that actually tiny fluctuations in the hardware in some cases make a difference to cognition, then that's just much harder to work with because you need way more information about those initial conditions to be able to say anything meaningful about the high level behavior.
PJ Wehry (35:04.9)
And so go ahead.
Mazviita (35:06.018)
But it's an open question because obviously the brain is changing all the time at this molecular level, but there is this constancy in terms of behaviour, in terms of personality, both in animals and humans, that it's not like we're fluctuating radically in our behavioural ways, right? But it depends on how you look at it because...
Most people have enough novelty in them from day to day that they don't say exactly the same thing every day. Some people do just say the same anecdotes again and again and that's boring. We like people to have some degree of unpredictability not being the same from day to day but there's obviously enough predictability that we function.
PJ Wehry (35:46.253)
Yeah
PJ Wehry (35:56.016)
I just saw it's kind of going around on like the Instagram algorithm about husbands who like their wives are worried about like, do they like me or not? It's like, I've liked the same Subway sandwich since I was nine years old. You know, it's like, it's like, yeah, there are definitely people who are like, maybe if we think of weather patterns, it's like, yeah, it's going to rain. now even and we
Mazviita (36:17.868)
Right. Yeah.
PJ Wehry (36:24.866)
You mentioned extended mind. So I wanted to ask, cause it seems to be a large part of it is the question of AI and how that could help us. How does AI and deep learning reference that as well? How does that play into this whole discussion?
Mazviita (36:38.976)
Yeah, so this has been a really big phenomenon that's changed neuroscience a lot over the last, what is it, decade, bit more. One of the things that I really noticed with my background in vision science is that when I was being taught vision science, initially people looked at the human capacity for face recognition, object recognition, and said, wow, we don't know when we're going to get computers that can recognize faces.
at least not as reliably enough for security technologies. But one of the justifications for funding vision science, the kind of science that I was being taught, was that if we reverse engineer the human visual system, then we'll be able to build computers that can have this ability, and that will be a really useful technology. But how it happened was that engineers using artificial neural networks
which is not a neuroscience based technology. It's like slightly inspired by neuroscience from a long time ago, it engineers working independently of vision science and neuroscience figured out how to solve object recognition and face recognition. then vision scientists and neuroscientists like really paid attention saying, okay, well they figured out how to do what the brain does in a machine. We should be able to learn from these machines, something about how the brain works.
So there's been a lot of interest from the scientists in artificial neural networks, deep learning, whether there are some core principles which will help illuminate how the brain works. But what I argue in the book is that that again is a bit too quick. Obviously there are multiple ways that you can solve the same problem. And I'm arguing that what these networks are doing is something fundamentally quite different from how the brain operates at best.
You can talk about them brain and the artificial network as being on a very high level of abstraction, similar because they have the similarity of structure where you have these artificial nodes of connected to one another and changing weights in this very million billion parameter space. But in terms of the implementation, the actual stuff that they're made of and in terms of the learning process, they're very, very different.
Mazviita (39:05.889)
So I personally think there are limitations to how much neuroscientists can learn from the engineering, from deep learning.
PJ Wehry (39:17.122)
Yeah, I
Before, so I now run a digital marketing company with my wife, but I put a lot of effort into coding and that's how we got into the digital marketing space. And at one point I was looking at just like learning AI. So I started reading books on it. So when you talk about the black box of neural networks.
Mazviita (39:24.044)
Mm.
Mazviita (39:39.384)
Right.
PJ Wehry (39:42.008)
fascinating to me and I want to ask questions about it but I feel like maybe that's not the right road to go down.
Mazviita (39:47.404)
Yeah. Yeah. I mean, one of the things I can say as a tool for like dealing with very large data sets, so taking a very large high dimensional data set, finding some patterns in it and using those to make predictions. Artificial neural networks being used in neuroscience just for that in the same way that they're being used in many other branches of science, whether it's particle physics, weather science, anywhere where there's lots of data, you're going to be seeing deep learning now. It's very common tool now.
the sciences. But as you're saying it raises this problem of is it so much of a black box that we just see the predictions and we don't know how it's worked out the predictions. Ideally in science you want not only the prediction but also the explanation. You want to know what are the key variables, what are the causal principles that led to this thing happening and you don't necessarily get that from deep learning.
PJ Wehry (40:41.116)
And if I understand that's because of the billions of calculations involved, Which makes it, which is probably why it succeeds in the first place. Like, and this goes back to what you said about mathematician.
Mazviita (40:49.484)
Exactly.
PJ Wehry (40:59.042)
You like the part of the question is what to measure, but that's also governed by what is the purpose like the way that the black box works is that it's constantly tested against something and it reacts to whether it succeeded or failed. Right. And so and that's how it chooses how to weigh the different neurons. You know, another analogy. Right. And so one, that's why philosophy of neuroscience
Mazviita (41:13.038)
All right, yeah.
PJ Wehry (41:28.428)
exists is because we have to ask, OK, what is the purpose we're aiming at? And that's an important question.
Mazviita (41:29.166)
Thank
PJ Wehry (41:36.76)
I want to make sure we have enough time to talk about this because the idea that I'll be honest, I just like the name. What is the fallacy of misplaced concreteness? I just want to be able to say this in conversation, so I would love for you to explain it so I can use it.
Mazviita (41:48.438)
Right. Yeah.
Yeah, I don't think it's the catchiest terminology, but I'm glad you like it. No, this is a phrase from Alfred North Whitehead. was a, and one of these philosophers from about 100 years ago. He's interesting, an interesting figure because he's one of the founders of analytic philosophy. He co-wrote the Principia Mathematica with Bertrand Russell. So he's a mathematician by formation. But then in his independent
work in metaphysics, you really went deeply into this process metaphysics. So this idea that the world, the universe is inherently this influx, aren't stable entities. And
What he says about mathematical representation is that they don't tell us the inherent structure of stability of the universe. Actually, these are abstractions that we take from an inherently more complex qualitative world. So when he's talking about the fallacy of misplaced concreteness, he says that we have this tendency, bad habit of when we have an abstraction that works well for us.
be a mathematical representation, it could be an analogy. We tend, or all kinds of models and theories in science, we tend to substitute in our thinking the abstraction for the concrete thing that inspired it, that originated it. So physicists substitute in their minds the equations of stuff for the actual things that they're studying. That can be fine if you're a scientist, it can just be fine as kind of shorthand.
Mazviita (43:38.722)
But he says it's been a real problem in philosophy because philosophers have taken the lead from the scientists. For example, this is his diagnosis of the mind-body problem. in fact, he says that where we get the mind-body problem from is that physicists look at matter abstract to only the quantitative properties.
These are ones that physicists can deal with because they're the mathematical ones. What's left in material bodies is a whole bunch of qualitative properties like a bunch of colour and softness and emotional qualities that we experience in ourselves. They're there in nature, Whitehead said, but they seem hard to fit into nature precisely because we've substituted our concept of nature for this abstract mathematical
and if non qualitative one in which it seems really weird that anything should have any kind of mental dimension to it. But he says, if we don't go with our abstraction and just confront what is there in concrete reality, he says that even in material living bodies, there is a psychological dimension to them. We should just get used to the idea that nature itself is this integrated thing in which the quantitative and the qualitative go together.
So yeah, the fallacy of misplaced concreteness is just like, we forget what concrete reality is and just we're just stuck in our abstractions.
PJ Wehry (45:14.948)
stuck in the models and you know, yes. Thank you, I appreciate it.
Mazviita (45:16.717)
Yeah.
Yeah, so the more common way of talking is like, you're mistaking the map for the territory. That's the phrase that everyone also uses. More commonly known than Whitehead's one.
PJ Wehry (45:26.532)
Yeah, yeah, yeah, yeah, yeah.
PJ Wehry (45:31.074)
It's like if you're driving and you have.
Oh, yeah. I am. But it sounds better if you say. And I appreciate that. I've heard the name Alfred North Whitehead, but I've never studied him. So it's really nice to see even have that very kind of. It's just a great and clear explanation that you gave. That's why I appreciate that. Like, I'm like, oh, that's again, obviously, that's, know.
Mazviita (45:44.078)
Thank
PJ Wehry (46:03.396)
Not everything that he said. I don't think he's like a five minute.
Mazviita (46:05.224)
No, no. He's actually one of the most difficult 20th century philosophers. have a friend that works at Dundee University and she claims to understand his metaphysical books like Process and Reality and I'm just really impressed that anyone has written about this with him. But he also wrote these more popular accessible books such as Science and the Modern World and that's the stuff that I read and that's where he gives this diagnosis of the mind.
PJ Wehry (46:23.758)
Now there you go.
Mazviita (46:35.224)
body problem and also quite a few biologists of his era read that book as well because they were interested in idea that there is something important about biological complexity that is not captured in abstract models.
PJ Wehry (46:41.316)
Hmm.
PJ Wehry (46:50.044)
And hopefully this is not too big of a jump, but I do feel like there's a connection here between even how you start in the introduction, going between Galileo and Dooham, if I'm saying that correct. What is the connection between this mistaking the map for the territory? think things change over time. If you're driving and you're looking like, well, that's not on the map. It's like, well, it doesn't matter. It's right there.
Mazviita (47:00.524)
Yeah. Yeah.
Mazviita (47:06.039)
Mm-hmm.
PJ Wehry (47:19.797)
How does that help us to understand, I think you say that Galileo is still winning out over Doohem.
Mazviita (47:25.708)
Yeah, yeah, in terms of how many peoples interpret the abstractions of science. I give Galileo as maybe the most famous and influential instigator of this idea that actually the map is the territory, that actually there's nothing wrong with mistaking the abstraction for reality. Because he talks very famously about the book of nature being written in the language of mathematics and geometry.
So he doesn't say that the mathematician is there for our convenience in order to strip away complexity. He says, no, the underlying architecture of the universe is in that mathematical form. So the physicist needs to know that language of maths in order to be able to decode the book of nature. And Douerre? Right, exactly. And he does give us those as well.
PJ Wehry (48:17.998)
And there's theological foundations for this, right? Yes.
Mazviita (48:25.07)
Pierre Duhaume, French physicist and philosopher and historian of science, again from about 100 years ago, he has this quotation that I have put in the book comparing it to Galileo's position as a contrast to this, saying that we like to think that the simplest, elegant representations are the ones that are there in reality, but it's only because of the weakness of our own minds that we prefer this simplicity.
If we had infinite scope of cognitive capacity, then we wouldn't need the abstraction.
PJ Wehry (49:05.955)
Dr. Chirimuta, it's been awesome to have you on today. I've learned a lot, clearly more we could talk about, but I do want to be respectful of your time. As we draw to a close here, a common question I ask my guests. For the audience that's listened, the listener who has gone through this last 50 minutes, what is something that you would recommend they do or think about over the next week after?
after the last 50 minutes of listening to you share this.
Mazviita (49:38.722)
Hmm. Well, I think I would have to give them the recommendation given by Hussall, the phenomenologist, one of these other philosophers from 100 years ago that wrote about this topic to go back to the things themselves, to actually sort of attend to that concreteness, which is there in the world around us and just like attend to the difference. mean, how much are you used to like a particular convenient, habitual
PJ Wehry (49:54.062)
Mm.
Mazviita (50:08.28)
conceptual scheme and how much if you just pay attention to the details to the counter examples do you see a mismatch there?
PJ Wehry (50:17.336)
Beautiful answer. Thank you. And I actually I normally ask it a different way I normally say besides reading your excellent book. So let me add that as well. Of course we the audience should read the Freight of Tracted
Mazviita (50:25.531)
Of course, well that was the first answer that came into my head, but I didn't want to say that.
PJ Wehry (50:30.852)
I appreciate it. Let me say it for you. Obviously, we should read the book, but also a great reminder to return to the things themselves. See with eyes unclouded. Dr. Chirumutta, it's been wonderful to have you on today. Thank you.
Mazviita (50:35.89)
Thank
Mazviita (50:42.932)
Thank
Thanks so much for having me.