The brain is massively complex, so how have people tried to crack its mysteries? And what does this always have to do with the latest technologies that are available, And what does the history of neuroscience have to do with feeling bumps on someone's skull? Electricity, Frankenstein, animatronics, telegraphs, telephone exchanges, computers, LMS, and the next metaphor that we're going to use to try to capture the brain's magic. Welcome to Inner Cosmos
with me David Eagelman. I'm a neuroscientist and an author at Stanford, and in these episodes we seek to understand why and how our lives look the way they do. If you reach out and you wrap your hand around your coffee cup and then bring it up to your lips,
that all seems pretty effortless. But what's driving that motor action is a massive, invisible puppeteer living inside your skull, an organ made of tens of billions of very active cells called neurons, each one of which is firing off tiny electrical pulses tens or hundreds of times every second, and each of which is exchanging messages with thousands of
other neurons. And somehow from this vast electrical storm, you get your hand moving, and you get your perception of the tattooed barista in a rolled up flannel pulling espresso shots and their of students hunched over laptops, and the scent of dark roast and cinnamon. And you get your memory of your name and your first kiss and what you had for breakfast. And generally, from this lightning storm of activity in the brain emerges the shimmering.
Something that we call you.
Now, we humans have been trying to understand this wrinkled three pounds of jelly for a long time, because it's been obvious that this is where all the action is. Why is it obvious that is all happening in the brain. Well, if you lose your leg, or your kidney or an eyeball, those are all terrible events, but you.
Are still the same. But if you damage even a very small chunk of brain tissue, that can change you entirely.
It changes your personality, or your decision making or your thoughts. It changes who you are. So that led people to realize slowly that somehow all the action of your behaviors and your conscious experience is all tied to these three pounds of tissue. So that sounds good, but it's really hard to crack the mystery of how this thing works. And that's for two main reasons. First, all the action going on is microscopic. When you look at it, you can't see anything. It just looks like a big, wrinkly
thing with the consistency of mashed potatoes. You'll often hear me and others say that this is the most complex thing we've ever discovered on our planet, but that is a modern revelation. There's nothing obvious about that statement when you look at the brain, and in fact, ancient cultures used to throw out the brain at autopsy and be much more interested in other organs. So the brain's doing
all kinds of things, but you can't see that. The second challenge is that the brain is locked up tightly inside the skull. It makes sense that it's well protected as very delicate stuff, but that has made it especially hard to study. So what's happened is that people have been trying to decipher this very complicated puzzle for a long time, and the history of neuroscience is, in a
way the history of human self regard. Each generation peers into the darkness and sees its own technology reflected back. So in the seventeenth century, when hydraulics were the cutting edge of technology, the brain was imagined as a network of pipes and valves.
Thoughts and sensations flowed like.
Water, propelled by animal spirits that coursed through hollow nerves. Then in the eighteenth century, as electricity captured the imagination, the brain became a battery. It became a generator of sparks. By the nineteenth century it was a telegraph system, with messages darting along wires and transmitting coded signals from one station to another.
And then as we moved into.
The twentieth century with advancing technology, the metaphor shifted again.
The brain became a telephone.
Exchange, and then an electronic circuit and eventually a digital computer. Now the cool thing is that each metaphor brought new insights. So hydraulics inspired experiments on fluid pressure in the brain, and telegraphs shaped the search for nerve conduction, and the computer age fueled the rise.
Of neural network models.
But each metaphor also carries its blind spots. We become experts at seeing the brain through the lens of the day's technology, and sometimes we mistake the map for the territory. So nowadays we've got great tools. We have fMRI scanners that can track activity millimeter by millimeter. We have optogenetics that can tickle neurons on or off with light. We have genetic tools to clip out sections or even single letters. And yet the deepest mysteries of the brain still remain,
and they're still stubborn. How does a three pound lump of biological tissue generate the first person experience of being alive? How does matter become mind? And how do all the technology metaphors that we use today blind us to some of the details. So that's why today I called up Matthew Cobb. He's an evolutionary neurobiologist who studies smell and memory at the University of Manchester, and he's also a historian who has spent decades tracing the shifting stories that
we tell about the brain. He wrote a great book called The Idea of the Brain, and this is a fascinating history of neuroscience because it's not just a list of events, but instead it explores how science and culture and technology all braided together to shape our ideas and how time and again we've been certain that we were on the brink of a full explanation, only to discover
that the brain is more complex than we imagined. You can't read this book without getting a deep appreciation for the centuries of labor that.
It took to get us to the modern day picture.
The history of neuroscience is about hundreds or thousands of people making small contributions, and presumably each person could never know how seminole the contribution was. They were just providing one puzzle piece that would end up clicking together with other pieces in the future, but it's always impossible to see. So the key is that in the end, the story
of the brain is also the story of us. It's our technology, is what we understand at any point in history, and of course our relentless drive to understand the organ that is doing the understanding. So here's my conversation with Matthew Cobb. Okay, so, Matthew, I'm a giant fan of
your book, The Idea of the Brain. I thought it was extraordinary because it's not just a history of neuroscience that is, it includes that, but much more importantly, it's a history of how to think about the brain, how people have thought about it.
And one of your main points in the book.
Is that we always draw on our technology as a metaphor. So science is not just an accumulation of facts, but instead we frame things in particular ways depending on what's going on around us.
So let's start there this.
Issue that you've just described a metaphor, of the metaphors
that we use, and how that changes over time. That actually gives a frame to the book and makes the history, which otherwise might be a bit dull, actually much more interesting and enables you to think about how people could think the way they didn't, why they couldn't see anything different, Because that's always the problem with history, right, I mean, it looks so obvious to us, you know how why on earth didn't people realize that the brain is the center of thought?
Why do they think it was the heart? And one of the rules.
I think about history, in particular the history of science, is that you're not allowed to think that people in the past were stupid because they were amazingly smart.
They just didn't know as much as we know. And we only know.
That most of us because we've read it or we've been taught it in class. Relatively few people have actually made great breakthroughs that have changed how we think about the world. So when you've got these people who are very, very clever but can't see something, that the task is to try and put yourselves in their heads and say, well, what is it they don't understand And it can't just be the answer, you know, they didn't understand that nothing
goes faster than light or whatever. You've got to think about, well, what is it the framework that they're living in the world they're living in. Trying and put yourselves in their shoes,
and then it all starts to become clearer. And it also, I think for the reader, both the lay reader and for scientists becomes much more interesting because then you can start to say not only about now, but also about the future, about what might be coming and why we can't understand certain things at the moment.
You know.
One of the things that was quite stunning for me is I, like, you, have spent my whole career in neuroscience, but it was difficult for me to envision what it would have been like to live in neuroscience in an era where we didn't know something. So for example, the discovery of the action potential. This is where a neuron has a spike there, zips down the axon and carries signals rapidly in this way. I grew up just taking for granted that that's what neurons do.
That.
Of course, the brain uses electricity in this sense. But you know, you had a whole chapter about electricity and how people discovered that that was going on with the brain.
Give us a sense of that.
Well, it all began, I guess it began in America really with.
The ability to Franklin to actually bring electricity down very dangerous experiments as everybody knows, with kites and keys and that kind of thing. But in the eighteenth century people began to be able to store electricity. They could produce it in the form of static electricity. So they'd get some amber or any kind of resin and you know, just like you, you know, maybe you do.
This for your kid.
So you know when you were a kid, your dad would do this. You get a balloon, you rub it on his jump and he stick it on the wall. Amazing, right, that static electricity you can actually generate by putting say some wool onto a resin wheel. That you spin round and they would do These people would do these kind
of party tricks. They were called electricians, right, and they would go into your fancy house and they would they would do like one of these shows was called the Feathered Boy, and they'd get some hapless child and winch him up to the ceiling and then they'd charge him up with electricity using this amber and cloth, and then they'd throw a load of feathers in the air and of course they'd all stick on it. So they could do stuff like this. And also people were thinking about
actually using a very primitive form of electroshock therapy. So these electricians, when you could generate it, they would travel around the countryside, you know, itinerant electricians, so too particularly significant figures. So John Wesley, the founder of Methodism, and Maha, the French revolutionary, were both itinerant electricians in the UK in the eighteenth century.
Wandering around. Somebody say I'm.
Feeling really miserable and terrible, and he said, well, just hold on to this, and then they'd wind it up and bang, you get an electric shark. So scientists knew that electricity was doing something, but that didn't mean to say that it was how bodies worked, and it became increasingly complex as people were able to actually store electricity through something called a lidon jar. You'd actually put it into a generate this electricity and it would then stay.
But what happened it was then suddenly discharge if you touched it, like touching a cow fence.
Right, you know, you know you've been touched.
And so when French scientists didn't experiment with this, he got four hundred monks, which is they're all holding hands, and at one end he got one of these big jars full of electricity and he made the monk touch it, and then he watched us that they all jumped up as if the chargers down this line of monks must have been about whatever eight hundred meters more thousand meters
kilometer along a monk. And then finally they were able to use They were showing that even they're doing experiments on animals, like in particular frogs, and they'd noticed that if you stimulated a frog nerve on attached to a leg, then their muscle would contract. Now this didn't show you what was going on, because you know if you put acid on a nerve then that it will also contract so maybe electricy was just an irritant, or maybe it was actually in bodies right, And this was.
A big argument.
One of the things that people stuid was electric fish, so the electric eel. And they found this structure that was produced as shock and at the beginning of the nineteenth century chap called vaulta hence vault. He decided he'd do a bit of biomimicry. He said, okay, well they've got this electric organ. Maybe I can use that electric organ. I think I can mimic the structure of this which
had kind of layers and produce an electric current. And so he did created what he called a pile, which was layers of zinc and cardboard with los of acid around them. And it was because it was a black pile of pennies. And this would then produce electricity, but at a constant rate.
Now in English we now call this a battery.
So you've now got this continuous release of electricity. And then you can see, well, actually we can stimulate if we put these electrodes onto and they did very horrible experiments on animals, dead animals. You know, you could then make it if you put electrodes on either side of a cow's head, then its mouth would start to move,
and its eyes would roll in its sockets. And in one particularly awful experiment in London, then a criminal who killed his wife and child and had been hanged was immediately taken down from the gallows as soon as he was dead, taken into a small room with about twelve learned gentleman and this experiment was done on his dead body, and of course he would then, you know, his arms started flailing about and all the rest of it.
That these experiments convinced.
People that there was electricity in nerves, that it wasn't just an irritant, that it was actually some kind of organizing principle.
And by the way, this is starting to be one of the origins of Mary Shelley's frankenstn.
It's not known, but she did regularly go to something called the Royal Institution in London, which is still there, amazing lecture theater where science is communicated to the public, and various British scientists spoke there, and at the beginning in about eighteen fourteen, there were a series of demonstrations of this power of electricity to animate dead bodies on sheep's heads and so on. You can go to the theater and see it. I mean, it was a huge,
huge thing. And we don't know that Mary Godwin went, but it's pretty likely she did. And then a couple of years later, when she was whatever sixteen, she ran off with Shelley and they went on honeymoon. So they sat around in Morote ghost stories and she wrote Frankistan.
One of the fascinations that I enjoyed with reading your book was all these things that we learned from biology that have created new technologies, like looking at the electric eel and the invention of the battery that fade away. Now we have a gajillion times more batteries on our planet than we have electric eels, but we have forgotten where that came from, that biomimicry. So what I really
loved is the way you tie these things together. So tell us about phrenology, the pseudoscience of phrenology that was popular, and how you think this actually contributed in a positive way to how we think about neuroscience and modern times, even though phrenology itself was incorrect.
If you want to have an argument with a neuroscientist, your studies the human brain in particular go groups them together. Say is function localized? Is there part of our brain that is devoted to doing a particular thing and that if you remove it you can't do that thing?
Right?
The answer is kind of yes, and no. Man is complicated, right. But in the nineteenth century it tormented people a great deal, partly because of the influence of French philosophy, and that the identification of the brain and the mind caused a huge problem because Descartes, who was for the French, the philosopher and no need for any other. He insisted that the mind was a united structure and therefore the brain
must be as well. And if you look at the brain, it appears identical on either half, got two halves, and they look pretty symmetrical. So the French were very convinced that there was no localization of function. And yet there was this popular idea of phrenology, which is, you know, kind of grew up in the late eighteenth cent and was basically a version of oh, I don't like the look of his face, his eyes are too close together, or whatever, you know, version you he's got big ears,
or whatever. Way of looking at somebody's face and not liking them and thinking that they're not shifty or whatever. This was then theorized in all sorts of ways by various thinkers who argued that by feeling the outside of your head skull, you could tell the shape of your brain and the lumps on your head, because we've all got lumps. They reveal the existence of organs and sections of the brain that are doing particular things.
But rather than say, as.
You know, you might say, well they're devoted to speech or you know, impulsiveness or vision, they were all about moral virtues and you know, being goodness and so basically it's like astrology, you know. And people would go and it was incredibly popular in the nineteenth century, and the scientists got really really cross about it because they said, look, you know, your.
Skull's really thick.
You may have a big lump on your head, but there's no lump corresponding lump underneath your brain. So this is just rubbish. So this went on until the late the early twentieth century. So this is pseudoscience, and despite it being widely not accepted, but you know, ordinary people, ordinary people from Queen Victoria to Carl Marx all thought it was very interesting.
You know, there was nothing in it. It was complete, not of rubbish.
But what it's got in there is this suggestion that maybe there is localization of function. And this is eventually resolved in about eighteen sixty when a French scientist called Broker was studying the brains of patients who who'd have strokes and have lost the power of speech, and he found that consistently when he did the dissections, the front left and area of the brain was damaged in these
patients who had lost the power of speech. Patients who died and hadn't lost the power of speech did not have those leeches legiance.
And he was really unhappy.
About this because he was French and this could not be and yeah, I mean having I don't know if this has happened to you, David, but you know, sometimes you collect data and it.
Just pushes you into a corner.
You don't want to go to that corner because you don't think that's the way it is. But the data just in the end you've got no option. You're just going to say, well, that's the way it is, this is what happens. And poor old Broker had to say and shock the French establishment by saying there is localization of function. Of course, you can't see anything with the eye, the naked eye for this front left hand side of
the brain. And then a couple and if you saw this was a couple of years ago, quite astonishing paper was produced between a group of neuroscientists were doing fMRI scans and a woman in her thirties was a secretary who had a college degree, who wrote to them. She saw an advert, you know, saying we're looking for people to have brain scanned. She said, I've been told there's something interesting about my brain. And they said, okay, yeah,
come on in. They stuck her in the scanner and the left hand brain side of her brain is completely empty. There's nothing, there's a hemisphere completely and she has done this ever since birth, and yet she can speak perfectly normally.
You would not know if you didn't use a scan out, you'd have no idea.
I wrote about this in my book Live Wired. And you know, you see the same thing. When a child gets a hemisphere ectomy, they get half of their brain removed, let's say, because of an intractable epilepsy.
Turns out they're fine.
And this is because the left and right sides of the brain are sort of carbon copies of each other. They're redundant for the most part, and so you can take one out and the remaining hemisphere just rewires the real estate to drive the boat as it needs.
To as long as you're very young. So it's a plasticity thing, right.
They don't do hemispherectomies after about the age of eight typically, Yeah, yeah.
So, And there are examples of people who have had unbelievable bad damage from strokes who then recovered their function completely her adults. But as I'm always very careful to say when I explain this to people, is this is very rare and in general strokes are pretty bad news.
What's fascinating about that is sometimes somebody will get a stroke on their left side that impacts Broker's area, let's say, and they lose the ability to speak, maybe they get a lesion of Wernikey's area, or they you know, they lose the ability.
To produce language correctly, they have flu into phasia.
Okay, what happens is then they recover and people think, well, great, somehow that has you know, they have repaired that part of the brain. But what's in fact happened is it's just moved over to the other hemisphere and now they're taking care of language in their right hemisphere. And the way this was discovered this was, you know, fifty years ago, is then this poor person would let's say, have a stroke on their right side and they would lose language again.
And so what this demonstrates is the massive plasticity the ability to move stuff around to other territories on the on the brain.
Which also shows us why the answer is it localized? Man is is weird? So yes and no.
This is the difficulty that neuroscience has always faced with the question about localization because sometimes, you know, if a bomb drops on the runway of an airport, you'll notice that all the planes stop, but you didn't hit the airport itself, just the runway. And this is of course what always happens with the brain. When we see brain
damage and some function stops. We don't know if that's because that area was the function or just part of this larger network that has to be there for things to work for the planes to take off.
I mean, it's a very interesting philosophical and methodological problem which people have been arguing about for decades, since since the beginning of the twentieth century. And now, of course
we've got genetics, which does exactly the same thing. Or I knocked out this gene it's the gene for X. Well, very very occasionally it's the gene for X. But generally you've just pulled apart a component, some bit of the whole complicated network, and the whole thing falls apart, you know, And it's very difficult to know when this is going to happen. So analogies of things like bicycle wheels, you take a spoke out, you take a spoke out, and then eventually the wheel is going to collapse, but you
can't it's very difficult to predict when. And it doesn't necessarily matter which order you take the spokes out or whatever. So this is very hard and it's part of the problem is it's easy for scientists to get excited about this thing. I've got this tool, and it's easy for the public to understand because it makes perfect sense. But to get back to your starting point, to get to the metaphor, if you ask the average person street you know, what is the brain like a computer?
They go, yeah, well sure, but here's the thing.
You know, You've just explained that you know, strokes or whatever, you can recover various elements from a function from them. But if I take out the microphone from this computer, you just won't hear me. There's no other part of it. Is going to rewire it and go okay, well ill use the camera instead. I mean, we're done.
There will be no more podcast.
And so that that fixity in a machine and the flexibility of anything organic, but in particular the brain, that just shows they are a different order of thing.
So, by the way, I want to return to this issue about the metaphor of the brain as a computer, because you pointed out something surprising in the book that people were thinking about this the other way for a while. When von Neuman was developing the computer tell us about that.
We all say the brain is like a computer. But when the first computers, like we all use now, the one with as you said, von Neumann architecture, when that was being developed, what von Neuman said he was going to do, and this is his pitch to the US government in nineteen forty five to get the vast sums of money to build this thing. Why did he she said, I'm going to build you a computer and it's going to be a computer like the brain.
And why do I say that? Because I know how the brain's wired up.
How do I know that because two years earlier to researchers McCullough, CA and Pits have published a paper on logic.
It was called the Imminent Logic of the Nervous System in which they said, look the way the brain is wider. All the nervous systems are wired up.
You've got things called synapses, which are where one neuron sends a message on to another, and you can have a straightforward relay.
That's not very interesting.
What gets interesting is if you've got some conditionality about You could have a two neurons coming in and the onward neuron will fire if both of them are inputs are firing, or if one but not the other, and these basic logical structures, if and then and not. Anybody who's done any logic or any programming says, wait a minute, I recognize that. And that's what McCulloch said, right, This is astonishing. This is how nervous systems are wired up.
I'm going to build a computer like that now. I mean, it's not true. Nervous systems are not wired up like that. I mean, but it was a brilliant idea, and the idea of that there's this logical structure that enabled our brains or any brain function should produce an outcome to process information.
And that's what von Neuman did.
That architecture is hardwired into the computers that I'm using that you're all over the world.
Every computer in the world uses that.
Because he thought that mccullor ca and Pits had hit on while they had hit on something, but it wasn't particularly about neuroscience.
So let me step back for a second, because this idea, this relationship between brains and computers, this is what we're all very used to. But give us a sense of the metaphors that people had used previously.
The ancient Greeks, I mean, they thought about either many thought about something called numa pneuma. When you read this ancient Greek stuff and you think, what is this?
Is it air?
Is it wind?
So there's this aerial phenomenon which they thought was what was going on, but they didn't, I mean, and that because you know, wind technology is the height of their technology.
That's what they had, you know, that's what powered ships and so on.
Then The real breakthrough, I think came with Descartes, who was in a Parisian park in the sixteen twenties wandering around and they had a load of animatronics that these statues which all work by hydraulics, which was pretty damn cool and better in some respects and clockwork at the time.
And you know, you'd have a Hercules who would biff a dragon on the head with a great being.
He had a club and he biffed this dragon. You know, you go to a theme park and it was basically that. But you know, I mean, just like you see old CGI today and it looks crap. Mean, this would have looked awful, but at the time it was amazing. And the key thing is that descar I mean, he didn't think they were alive, but he thought, wait a minute, maybe that's how our bodies and our nervous systems work.
And he said, look, he's got this very famous drawing of this big kind of man baby touching a fire and he says, look, there's some kind of pressure from when you touch flane.
The pressure goes up into the.
Brain and then it bounces back it reflects, hence eventually reflex it bounces back and causes you to pull away from the fire.
Now very quickly people tried to do experiments to see whether this was the case. I mean, he constructed a whole theory about it.
By within twenty thirty years and seventeenth century, people had done fairly simple things like, you know, chopped a frog's nerve and no liquid came spurting out, so it didn't look like it was under some kind of hydraulic pressure. People thought of other things. Maybe they thought it was
like a vibration. There's something going down the nerve. If you hit a plank at one end, you can feel the vibration at the other, which's pretty smart, but there's not much you can do about it, and thinking about, you know, a load of vibrating planks in your head doesn't really help you much.
So even descartes idea wasn't really usable, but that was kind of how it was thought.
And then with electricity, the big breakthrough wasn't just that electricity is what's in this weird chemical way in neurons, But as soon as that was applied in the shape of the telegraph system, then people immediately drew an analogy both ways. They said look, here's there are maps the telegraph system in the UK in the eighteen forties with all the telegraph wires which were all running along railway
linees course, going down to London. And what people said is London is like the brain of the country and it receives information.
This is a word they use.
It's information from the provinces and it can also send instructions out and tell bits of the body politic what to do. And hey, we look at the nervous system, because they had fabulous dissections of the human nervous system. It's all going up into the brain and motor uron's coming down, et cetera. So this parallel was now we've gone from water that don't really make much sense or something electric. Now we've got a real communication metaphor of
the telegraph system. And this chap called Alfred Smith, he was absolutely convinced that literally this was what was going on inside the nervous system. And he even invented the facts or a kind of telegraph, the photo.
He said, if you could get.
A photo electric cell and send a man image down a telegraph, why you could should be able to reproduce it at the other end. So you'd actually get and that's what's going on in your eyes, he said, going into your brain. And he drew these amazing diagrams which and I show them to computation neuroscientists.
They get so excited because it looks like a primitive version of you know, the large language models that.
Everybody's excited about these days, all these crisscross interactions. I mean, I've tried to make head and tail of what he was getting at, and I think he was basically bonkers. And he hads to have looked. It looks it looks familiar, but it was actually insane. He made it, made a model of the brain out of brass, these strange bits of brass that he shows in his drawings, and you look at it and think, I don't you know what could this do? How could this process information in any
meaningful way? And there so you know, we got electricity. And then in the eighteen eighties you've got the problem about telegraph wise is it's just the static. They don't change, whereas we know that the brain has to respond to all sorts of different signals. And then the big change was the advent of the telephone and the telephone exchange because younger listeners.
Will have no idea what one of these is.
There's there's something called a telephone exchange where what would happen is you pick up your phone. You can see this in old silent movies on YouTube. You pick up your phone and there's a telephone operator who was normally a woman in the exchange, and she would a light would come on saying you picked up your phone.
She would connect a wire to your light, and you would.
Then ask for a particular number, and she would then connect the other end of that wire to that number. So you've got this flexibility. It's a switchboard, and I mean it's not bad. I mean it is in a way because you're rerouting information. There's nothing permanent, there's no fixed connections that can all be flexible. And that carries on into the First World War, and that's where ideas about information start to appear. Before this is mathematized, people
started to use the word information. In particular chap called Edgar Adrian, who's beloved of electrophysiologists, but unknown to most people. He won the Nobel Prize. Two of his students won the Nobel Prize. He was the president of the Royal Society. He was the vice chancellor of Cambridge University. I mean you could not get greater accolades than that. And now nobody knows who he is apart from the students at
the University of Lesdoo got an Adrian building. Anyway, he writes these popular science books and he tries to show to explain to people what's happening. His first person to record from a single neuron, to record a single spike. You can see it going down there as you described it. And he also shows a spike is either on or off. You either have it or you don't, so it's all or none, as it said. What the nervous system does with the information of spikes is to actually group them together.
And it's an in the code as he called it. He talked about codes in nervous systems, so are codes carrying information. And he has this lovely diagram of a again another poor old frog. It's the end of a stretch a neuron which is attached to a stretch receptor. It's a stretch receptor attached to a muscle, and he puts increasing weights on it. And as he puts more and more weights, the cell stretches and you get more and more spikes, so it's still it's digital in the
sense it's all or nothing each of those spikes. But what the nervous system is seeing is not zeros and ones. It's seeing an intensity code and a frequency code. How many spikes have there been a second perhaps even how they organized within a unit time.
These are all stuff that people are still very interested in.
So then when you finally get the computer all the in the mid late forties, early fifties, it all kinds of starts to make sense.
Yes, and this is where we this is part of my You know, we haven't gone any further, right, I don't think, so let me ask you this.
When I read your book, I was curious your opinion on this.
So you published this in twenty twenty, and since that time, we've had this explosion with the transformer architecture and large language models, which have changed the game a little bit, by which I mean, I think this is probably true
for both of us. You know, our whole career in neuroscience, we always looked at AI and sort of snickered at the idea that it was going to be anything as good as the brain, And then suddenly just a few years ago, it's got quite extraordinary, and so I'm curious how you think that changes the metaphors that we're using now. And I think we would probably agree that there's never going to be a perfect metaphor.
But how do you think LM's change the way we're talking about things?
Well, honestly, how they can help? How do they work?
Well, they don't know.
Yeah, so if you have Jeffrey Hinton, how does it work? He goes, I don't know.
So this is a really interesting thing though.
So for example, you and I both are interested in, let's say Eve Martyr's work where she tries to describe the lobster.
The example I give.
You great, Okay, great, so well here, why don't you go ahead and give it about Eve Martyr. But the point I want to make is that the questions that remain about what's happening in neuroscience, we have the same explainability crisis with lllm's.
So they're not distinguished in that way.
No, indeed, absolutely right, But rather it's so the example in the Eve martt so, yes, it's the lobster's stomach. She has spent a whole less a very smart woman spent her whole academic career studying the thirteen years that make up the system that controls them the activity of the lobster's stomach, and the lobster's stomach has to grind up its food a bit like a gizzard in a chicken, and it's got two rhythms, two ways of doing this, and this is controlled by the neurons. And she knows
everything about those neurons. She knows the genes that are expressed, she knows all about the neurotransmitters, the neural hormones that they bathe in that kind of modulate the way that it all activates. But at the moment, she cannot explain, using the tools we've had up to now, why those thirty neurons produce those two rhythms and not others, because according to the models, they should be able to do
other things, but they don't. And she can't predict, either experimentally or with a model, or by a conjunction of the two, what will happen if she is to alter the activity of one of those components. So, yeah, I mean, I take your point. This is, you know, we've got two completely inexplicable things we don't understand what's going on in the LLLM. But that's partly because the way it's built.
It feels like the LM should be so straightforward to understand in the sense that we've put it in there. It's a cartoon model of the brain in the sense that it's just units and connections between the units. And what this illustrates, I think is once things reach some level of complexity, it's not clear that we have the correct metaphors or the correct tools and science to capture it in some way.
Yeah.
I mean, like everybody, I'm amazed by what the LLMS can do. But I mean today, may this will all be old news by the time this is released. But today the big thing is chut GPT fives come out and it still doesn't know how many bees there are in blueberry, right, And it gets very indignant and insists that there are three because the capital letter counts for two, or because the bee in the middle for berry counts for two.
I mean, you know, so people are arguing with it, and this thing sounds real.
It sounds that it knows what it's talking about, you know, So it's that's very confident.
Here's what seems clear though, is that they are intelligent in a different way, as in, they get some simple things wrong, but they also do quite extraordinary other things that we're no good at.
So it's like we've invented a species that it is just a bit different from us.
So that is that true? I mean, you know you can why is it intelligent?
I mean you've got any Any machine that you have enables you to do things that you can't do yourself. I mean, draw a straight line, right, I can draw straight line on my computer in a way that I can't with my hand and a pencil, and I can do it better with a pencil than I can with a rock.
That's right.
But I think the surprise has been for us the way that it can put ideas together in ways that had seemed just a few years ago like some magical thing that only humans could do.
Writing a pair of you can ask you look, Jimmy.
A shakespeareance on it, where you know, poke Kiemon meets Godzilla on the surface of Mars, and it does it instantly and extraordinarily, And these sorts of things we would not have.
Expected even just a few years ago.
Yeah, and we probably would have expected they'd know how many bays they were in blue booth.
Yeah, exactly, so it's doing something.
But here's my question for you, is the metaphor has changed a bit just the last few years from von Neumann computers to having these large language models. That change a little bit the way we think about what might be happening in the brain, whether or not it's correct that that's what's happening in the brain.
For young people who are growing up right now thinking.
About large language models and thinking about the brain, the metaphors are going to carry over in terms of Okay, what if you just have a giant network with lots of connections and you can pay attention in the network to certain words. That's going to be the next crop of scientists are all going to draw on that metaphor.
I mean, in a way we've been here for a long time.
Francis Crick, as everybody knows, co discover the double heading structure of the DNA in nineteen fifty three with Jim Watson on the basis of work that been done by Maurice Wilkins and Russell and Franklin, and then he worked in molecular genetics and then in the mid seventies he decides he's going to move into neuroscience, and then I'm working on the book, and always in every chapter except
the chapter on neurotransmitters, Crick popped up. And he wasn't just popping up as you know, to somebody who was on the fringes. He was actually driving ideas and thinking and so on. In particular I was particularly impressed by was his embracing of something that was called parallel distributed programming, which was developed by Jeffrey Hinton and others in San Diego and around San Diego in California in the mid eighties, which is the ancestor of today's LMS.
He knew John Hotfield and hot.
Field and Hint of just won the Nobel Prize in Physics last year for their work on these neural networks and all the rest of it. And people were amazed by what these very very primitive neural networks could do. Because there'd been something called the AI winter from the excitement of the nineteen fifties. In the nineteen sixties and seventies, people thought, actually, this isn't just isn't going to work, and you.
Know, no funding FREEI was a disaster.
And then these tiny little programs were developed which could do things like learn the past tense is of English verbs, and they.
Would make mistakes like a human child.
That's so even if you told that told the computer that the past participle of go is went, it would say God, just like child. That's you know, so it's generalizing the e the most frequently accounted version, even though it's been told differently. So that got people are very excited, and Crick throughout this period was very impressed by these things, but he was continually arguing people saying, look, your models are amazing, but as models of the brain, they've.
Got to be biologically relevant.
They've got to be structured've got to have an organization, and that is similar to what's going on in the brain, which we didn't have much idea about. And if it relies upon, in particular, something called back propagation, which is where the signal comes back and goes up through the same connections in these computer programs, these connections equipment of neurons, and there isn't any backprop you know, it doesn't work that way. And yes, you know work arounds, but that's
not the way it works. He got very irate with his erstwhile colleagues and friends because of that, And that's clearly right. You know, if you want to understand the human brain rather than perhaps you know, understand the eerie alien intelligence, if you want to stand human consciousness, then you've got to keep that in mind, I think, and crit was right about that. So until we need to know how they're doing what they're doing, I think for it to be a really useful model.
So okay, a couple of things.
First, I still assert that the lms are going to change our metaphors the way that people of the next generation talk about the brain, correctly or incorrectly.
It's going to change the way they talk about it.
And and of course there's a history back to you know, parallel distributed computation and well before that as well. But but there are step changes, and I think we've just experienced one. So that's that's my assertion.
I was just wondering, if you know, I mean, you may well be right. You know, I can't. I I it'd be very interesting to see, and I just want a bit more meat on it.
Yeah, you know, it may it may be that it just requires some years before we see what that what that meat looks like on the book. Yeah, let me jump to another topic that I was that I loved reading about in your book. One of the things that I talk about on this podcast, probably more than anything else, is this concept of the internal model. In other words, your brain is locked in silence and darkness and it's
making a model of the outside world. And it does this eventually so it can better predict what's going to happen next. And the thing that I hadn't quite clocked until I read your book is that, at least the way that you put it, it's sort of correct me if I'm wrong here. It started with the young Craike in England who wrote a paper, and then other people picked up on this idea. Give us, give us that story about the internal model.
It really goes back to Helmholtz, who's the guy who discovers the all or nothing action potential. What Helmholtz said is that because there's a big argument about ah, is the activity in say, your optic nerve, is it qualitatively different from the activity in your auditory nerve? And what Helmholtz said, it's all the same. It's just activity. And the difference is is what your brain makes of that signal.
And what he says is that the brain gets stimulated and then it makes an inference about what that means, about what that is, and you can you can see this really easily, but don't do it very hard. The pressure eyeballs. Shut your eyes, pressure eyeballs, and you see colors, but there is no light. What is happening is those neurons are being activated by pressure and they're sending a signal to the brain and the brain, hey, I optic nerve it must be light, so you see light now.
Craig who was a psychologist in Cambridge, he wrote he wrote this little paper where he said, well, basically, what the brain must be doing is constructing this model of the outside world of things that it can affect, and it's trying to work out what the best things to do, how can modulate this outside was so effect Therefore the input that it receives in order to achieve whatever ends it may be interested in me.
They're eating you know.
I'm hungry. I'm right, my body's hungry. Therefore I need to find some food, so I've got to looking for it. And then the model that the key idea is that he's got this suggestion as you say that there's a model, a representation of some kind of the outside world, and
the way you can affect it is both things. It's not just a static photo, but it's also a set of potential alterations, and you can you can predict, Okay, if I go to the shop, I will be able to get a cream bun and I will satisfy my hunger. If I go to the bookshop, I will not be able to buy a cream, but you know, I can get a book. But that's not that satisfy a different kind of hunger? Is it the same kind of hunger?
Noe and so on?
So that model, I mean, is really quite astonishingly powerful, the idea that there's this representation at such a complex level of the outside world and of our ways of altering it that's been developed.
And I mean, he was working, like many people. He was working in the war, and he was working on how.
People could cope of work in the dark on chips and so on, and how they'd be able to predict a movement of enemy Aircraft's all.
This was kind of classic thing that scientists were involved in it.
And in his meantime he was thinking about, well, what does it all mean in terms of how the brain might function. So this little paper, and there's another one as well that he wrote, is astonishing the influential.
You say in the book that it's going to take us a long time to even understand very basic things. For example, there are a lot of connectome projects which look at, hey, where are all the cells and the connections. And let's say a maggot brain, a very tiny little brain. You know, what if we understood every single bit of the wiring diagram here and yet, as we know, that hasn't really unpacked the answer for us.
And so what do you see coming.
Down the line?
How, given this whole enormous, beautiful history that you've written of the field of neuroscience, what do you see in the future.
But what we're lacking is is ideas, ways of and maybe this is where the llms are going to do it without even telling us what the damn ideas are. You know, we'll be able to put all the data in, put all the connectomic data. I'm sure if eve Marda is not doing this, she really should check it in see what the AIS makes of it all. I think we need more ideas about very concrete things, not about how conscious, what consciousness is, or how it emerges. I think we need more data on that so we can
actually agree what we're studying. I guess the other thing that maybe you know, it's a negative thing to come back to Crick. You know, what he was focused on with Christophe Copp for the last twenty years was finding the neural correlates of consciousness or of retention of visual awareness.
And we've done that in that we can actually record from you know, people's brains, patients who are very kindly given up there allowing people to poke around in their brain while they're having an operation, and you can identify people have now even been able to reconstruct things that they have they have seen using the activity that's recorded. So it actually got some of the things that Crick wanted. And yet it's still not it's not it. There's something
still lacking. And maybe I mean the argument of the book would be it's going to be the next step in technology, and your argument that just being well, yeah, we've got that here, Well.
Let's see maybe that's it, or maybe it's going to be something further on.
Oh, surely it'll be something further on. And by the way.
One of your arguments is that are metaphors. While they can be helpful, they also always constrain absolutely absolutely.
Can't you know, they're they're I had this cute phrase, they're they're they're their frame, their frame. But a frame is not just you know, helping you put things on. It also limits you can't see outside of the frame. And that's that's why things look obvious in the past, because we can see the things that they can't. And that's why people sometimes think people in the past were stupid. But they're not very clever. They're just limited and so
to we. And this is when scientists, I mean, you've been very kind, you've read the book, you've thought about a lot. But when you just chat to people and they had the unwritten God, that's amazing. And then they said, well, what's the next big thing going to be into which I say, well, you know, if I knew that, I wouldn't tell you.
I'd be very rich living on my island somewhere. I had no idea.
But it's going to come and it But I mean, maybe you're right, I mean I'm maybe I'm not sufficiently tuned into the world in particular of kind of human euroscience and human brain studies to see that the excitement of the parallels with the with the llms maybe changing how people are thinking things. But certainly up until twenty twenty, which is when the book was finished twenty nineteen, there was very much in the tens or whatever we call it,
the teens, there's very much a sense of stagnation. I mean vast amounts of data which were people are drowning in. But also and then, what what are we going to do now this kind of uncertainty? Maybe that's being resolved.
Well, I think in your language it would just be another frame given to us by the technology things that we can't say beyond that.
But do you think that's happening? Is that a shift in the human.
Community for two reasons though.
One is looking at llms and saying, wow, that's quite extraordinary what they can produce. The other thing is just being able to use these as a tool to apply them to the massive data sets that we have shored up already and be able to say, oh, here are patterns of the data that we would never have seen ourselves. That was my interview with Matthew Cobb, neuroscientist and author
of the idea of the brain. So the central lesson that surfaces in his book is that we draw our metaphors from the technology that exists around us at the time, and as we invent new technologies that gives us new insight into how the brain might be functioning. But also metaphors can limit They are frames that can block out what we're able to see. This shows us something about
how science works. Despite everything that we learn in school, science is not just the accumulation of facts, but instead a grasping with our language towards something trying to get more effective metaphors. When we look back at the history of brain science, it's easy to fall into the trap of feeling like the present is the pinnacle that we finally arrived at the right way of thinking about this.
We smile at the hydraulic brain of the sixteen hundreds, the clockwork brain of the Enlightenment, the telegraph brain of the eighteen hundreds, the telephone exchange of the late nineteenth century. They seem like charming relics from a simpler time, but each of those metaphors were stepping stones for our thinking. They framed the kind of experiments that people built. The data that they noticed, the questions that they thought it
was even possible to ask. Every metaphor illuminated one part of the landscape, even while it left the rest in shadow, and that shadow is still with us. For my whole career, the metaphor of choice was the digital computer, the brain as hardware and software as an information processing device with inputs and outputs and circuits, and that idea has given us extraordinary advances like artificial neural networks and brain machine
interfaces and entire fields of computational neuroscience. But it also narrows our vision. And in my book live Wired, I wrote that everything we think about in Silicon Valley is trim and efficient hardware with a layer of software on top. But the brain is clearly so much more than that. That's why I use the term livewear to make this distinction clear. It's a system that is constantly reconfiguring its own circuitry based on experiences. In other words, what flows
through the network changes the network. Now we're all watching artificial neural networks like the Transformer architecture, which has changed our world, and those certainly seem to capture something a bit better. But whatever we use and whatever we come up with to make these distinctions. Will our descendants see our metaphors as limiting, just another product of our technological moment.
Of course they will.
For all we know, there may never be a final metaphor, no single model that captures it all. The brain has its own idio syncratic, massive complexity, and strangeness, making it not quite like a telegraph network, not quite like your laptop, not quite the same as an LLM. We might never have a metaphor that feels complete, because, as they say, the map is never the same as the territory. But the pursuit continues to compel us because every era's metaphor,
although it's flawed, still moves us forward. The hydraulic brain led to experiments on fluid pressure. The telegraph brain inspired the search for nerve conduction speeds. The computer brain pushed us into the era of big data and AI. Our past models were all part of the process. So I'm left with two feelings, humility in the face of the
unknown and excitement about what's next. Somewhere out there in the midst of the future, there is a technology or an idea that's going to give us a fresh lens one that we can't yet imagine, and with it a new way of asking the oldest question in neuroscience, how does matter give rise to mind? So as you go about your day hearing and seeing and remembering and imagining, consider this, the thing doing all that work is also
the thing asking the questions about its own functioning. Despite all that we've learned, it is still, in many ways an undiscovered country, and every day we take a step deeper into that uncharted land. Go to eagleman dot com slash podcast for more information and to find further reading. Join the weekly discussions on my substack, and check out and subscribe to Inner.
Cosmos on YouTube for videos.
Of each episode and to leave comments until next time. I'm David Eagleman, and this is Inner Cosmos.