Transcript of Episode 75 – Nick Chater: “The Mind Is Flat”

The following is a rough transcript which has not been revised by The Jim Rutt Show or by Nick Chater. Please check with us before using any quotations from this transcript. Thank you.

Jim: Today’s guest is Nick Chater. Nick is Professor of Behavioral Science at the University of Warwick School of Business.

Nick: Hello, everybody.

Jim: Glad to have you here, Nick. This should be fun.

Nick: Yeah, no, it’s a real pleasure to be invited.

Jim: Yeah. In fact, you don’t know this, but I’ve actually been following your work for quite a while, particularly, the work you’ve done with Morten Christiansen on the structure and evolution of language. I went back and looked at my records, and I believe the first paper I read by you guys was towards the connectionist model of recursion in human linguistic performance, which I think was around 1999.

Nick: Yeah, that’s right. And in fact, this very day, I’ve been working with Morten on a new book that we’re putting together called provisionally at least A Language Game. So we’re still working together 20 years on and having as much fun as ever.

Jim: Yeah, he’s a fine fellow. I got to know him when he was a sabbatical visitor at the Santa Fe Institute. We were both there, and we had some excellent conversations. So give him my regards.

Jim: Today, we’re going to talk about one of Nick’s books, The Mind Is Flat, and as always, you can find the link to that book on Amazon, I think, on our episode page at Now, this has got a little interesting history, a little unusual for the podcast, in that I actually read the book a while back, back in late 2018 as I check through my records. And I found it shockingly radical, as maybe some of you will, as we dig into some of the ideas on it. And I found it so radical. I said to myself, “I really need to read this again, to come to a firm conclusion about it.” But the rush of life, many other books to read, work to be done, I didn’t do it, until now.

Jim: And I said, “We have quite a few people in the cognitive science and artificial intelligence area on the podcast. I really ought to reach out to Nick Chater and give me an excuse to reread the book and maybe sharpen up my conclusions about it a little bit.”

Jim: And I got to say, it’s still radical as shit, but it holds together. And what’s more, it’s congruent with a great body of cognitive science, cognitive neuroscience and lab psychology research. And in retrospect, it should have been obvious, right? Since 2014, when I really took seriously to try and learn a little bit about cognitive science and cognitive neuroscience. I think I read 75 books and maybe 300 papers. And I kind of feel like Thomas Huxley and what he said about Darwin’s theory of evolution. How stupid of me not to thought of that. Really, it’s a radical idea, but I can’t find any holes in it.

Jim: Also, for those who want to learn more about it, there’s a course available at called The Mind Is Flat. We’ll have a link to that on the website as well at

Jim: Essentially what Nick is talking about in this book is that the view so many of us have had from culture, from science, from literature, from clinical psychology, et cetera, that there’s great depth in the human mind. And his argument is that this is not only misconceived, but it’s just wrong. The very idea that our minds contain hidden depths is completely wrong, and he starts off with kind of an interesting interlude about Anna Karenina. Why don’t you tell us a little bit about that and then let’s get rolling?

Nick: Yes. Well, thank you so much, Jim. It’s a wonderful intro. Yeah. So if you read a book, any book, really, you will have a sense that the characters become very real to you. And one of the things that’s very interesting is that they often seem realer than the actual people we know, or at least as real.

Nick: So for example, with Anna Karenina, you have a character develops through the book, who you have a sense of a tormented and troubled woman. The very end of the book, no plot spoilers are going to be objected to, as she jumps under a train and commits suicide. And it’s not quite clear to you as the reader, why she does that, but you still have a sense of a rich, complex person. And you think, “Well, maybe it was despair at being never been accepted into St. Petersburg society because of her wild lifestyle and her affair with her lover. Perhaps the affair with the level was failing.

Nick: There are any numbers of reasons for feeling hopeless, but also it could just have been a sort of momentary sense of … almost theatrical, making a theatrical gesture without really having thought it through at all. You’re not sure.

Nick: And I think what’s interesting is that you have a sense of rich character, a deep, interesting character, and one of the central characters of one of the most important novels ever written. But you really don’t have a very clear sense of what this character is doing and why. And one thing you might think is, “Well, if this was a real life character, a real person, not a fictional one, then we could simply ask her.” You could say, “Okay. You seem to be a very complex, three-dimensional, rich human being. And you did this extraordinary acts. Then you tell me why you did it.”

Nick: But of course, I think when we think about that a bit more, sort of obvious to us that actually Anna herself in contemplating, perhaps not really very deeply contemplating, her last desperate act, probably doesn’t have any particularly deep insights we don’t have ourselves. And she’ll be able to tell you a story. She better think, “Well, yes, I guess, what is bothering me at the moment? I think it must be this. It’s the hopeless, romantic relationship, like almost having tangled it [inaudible 00:05:36]. It’s the fact that I didn’t probably see my child. It’s the fact that I’m going to be ostracized by the aristocracy of some St. Petersburg and Moscow.”

Nick: But actually, she’s speculating just as much as we are. And I think that even if you would … And suppose imagine that she were to survive the incident, and you could ask her in her recuperating as I imagine it in the book, in a Swiss sanatorium. And you say, “Well, Look back and think through why you engaged in that act?”

Nick: And again, my feeling is that I’m going to try to sort of push on your intuition, that, that process of looking back at something one did, and trying to work out, “Why did I do that?” That’s really pretty similar. In fact, very, very similar, I think, to the process of trying to understand why a fictional character or another person did something.

Nick: What you’re doing is you’re looking at your behavior, looking at your experiences and trying to piece together what must’ve been going on. But what you can’t do, as it were look inside some kind of deep inner self and think, “Well, hang on. Other people have to speculate about my motives, but I can just see them. They’re just written in some ways, some sort of vault inside my mind. Let me have a quick look, a ruffle through, and I’ll tell you what it says.”

Nick: But I think we all know that to be illusory. But in fact, if you ask people to explain their behavior several times, they’ll often give you different, or at least not perfectly aligned answers, because we don’t fully understand ourselves. So really, what I want to do as an intuition pump at the beginning of the book, and obviously here too, is to make one thing … Well, it’s is an odd thing, because we have a sense of a rich character of Anna Karenina, we have a sense of rich characters in people around us and in ourselves. And I don’t want to understate at all the complexity of human beings. [inaudible 00:07:25] understanding, I think in some ways, that the idea that our mind has mental depths actually obscures, we’ll talk about this later, I’m sure, obscures how unbelievably creative and clever we are and what extraordinary beings we are.

Nick: So I’m not wanting to in any way reduce remarkable nature of human cognition, but I do think we should sort of allow the possibility that when I’m trying to understand my own actions, or Anna’s trying to understand her actions, or I’m trying to understand the actions of another person or a fictional character, I’m really doing the same thing.

Nick: And the final point is that when you think about fictional characters, and I think this is a bit of a shocker, when you think about fictional characters, it’s obvious that there’s no right answer. So if I ask you, “Well, come on then, what is the reason that Anna jumped under the train, really and truly?”

Nick: And of course there’s no real answer to that, because it’s just a story. There’s no deep question which could be settled by investigating Anna’s brain because there is no Anna with no brain. However, I think that’s exactly the right approach we should be having when we’re thinking about other people and ourselves too. It’s not the case that actually, if you could scan and Anna’s brain at a critical moment you could read off the fact that her real motivation was one thing or another. I think that’s itself an illusion.

Nick: So I think the creation of stories about ourselves is something we do all the time. We can create more or less convincing stories, and we can argue about the stories, but the idea that there’s some kind of ground truth is a dangerous mistake, I think. So it doesn’t mean that you cannot have sensible debates about why you acted as you did, just as you can for literary characters.

Nick: So if I said Anna’s concern was primarily a despair at the state of the serfs, you could reasonably say, “Well, there’s actually no sign of that in the rest of the book. Doesn’t seem to be at all about them. Lots of other people there.” In which case, you’d have a perfectly good argument.

Nick: So it’s not that you can say anything, but the idea that there’s a hundred of ground truth that you can get at is no more true, I think, for fictional characters than is of ourselves. And there’s no real sense [inaudible 00:09:30] there’s a real answer deeply lurking, deep within me.

Jim: Yep. I found that very good, as you say, intuition pump to kind of compare and contrast. Maybe we’re not that different than a fictional character. And of course the other side of the argument, kind of what you’re arguing against I suppose, is what started to manifest in the late 19th century and in through kind of the late mid 20th century, depth psychology, Freud, Jung, and those characters.

Jim: In fact, I’m going to read more than usual number of brief quotes from the book, to kind of get a sense of your position, where you’re coming from. “No amount of therapy, dream analysis, word, association, experiment, or brain scanning can recover a person’s true motives. Not because they are difficult to find, but because there is nothing to find. It’s not hard to plumb our mental depths because they are so deep and so murky, because there are no mental depths to plumb.” Right? Quite the opposite of the depth psychology guys.

Jim: “No one at any point in human history has ever been guided by inner beliefs or desires any more than any human being has been possessed by evil spirits or watched over by a guardian angel.”

Jim: And another, “The very idea of looking into our own minds, embodies the mistake. We talk as if we have a faculty of introspection to scrutinize the contents of our inner world, just as we have faculties of perception to inform us about the external world. But introspection is a process, not of perception, but of invention, the real time generation interpretation and explanations to make sense of our own world and actions. The inner world is a mirage.”

Jim: And then finally, “But of all this depth, richness and endless scope for exploration, it’s utterly fake.”

Jim: So you don’t hold back here. At first, I thought this was repetitious, but I did find it actually helpful to kind of move me psychologically off the ball of what we’ve kind of culturally found through our religions and depth, psychology, et cetera, that there’s all this stuff there to be discovered.

Nick: Yes, I think it is a very hard shift to make. And I think it’s something that on the one hand seems, as you were hinting in your introduction, on the one hand seems incredibly plausible and intuitive, and on the other hand seems just completely crazy. And I think one tends to oscillate between the two perspectives, and I think it takes quite a long time to feel comfortable with the fact that this sense that we have of being driven by [inaudible 00:12:04] beliefs, and desires, and motives, many of them mysteriously hidden from us, is actually just the wrong way to see things.

Nick: So I think again, I mean, thinking about fictional characters, I think is a good way in, because there in some sense, you want to be able to say, “Well, of course, characters in books, they’re kind of in some [inaudible 00:12:22] of metaphorical sense, driven by emotions and most [inaudible 00:12:26] so on, because when I tell a story about why they did what they did, I’ll say, ‘Well, her motive there was this, I think. I think she probably thought this was true, or she thought Vronsky wasn’t in fact being faithful to her.'” Or maybe she didn’t. But I’m giving you a rationalization, a story.

Nick: But of course, I don’t really that story is somehow substantiated in Anna’s brain and actually has driven her behavior. And I think the critical trick is to realize that, that’s not really true for understanding other people and ourselves either.

Nick: So one reason to think this is that when we look at the stories we tell about ourselves, of all shapes and sizes, whether they’re complex stories or everyday stories, about why we do things, the sound and consistent truth is that the stories we tell don’t seem to fit together.

Nick: So you asked me why I did something, and I’ll tell you. And you ask me, “Well, tell me more about that. Why did you do it that particular way?” And amazingly, I will just keep on talking. We’re amazing spinners of yarns. So I can keep on going, as if there’s this enormous library full on self-understanding I can draw. So like, “Why did you take that particular route.” And I say, “Well, there’s pretty busy on the other route. And it’s also just a little more scenic.” “And why is it more scenic? What would you say about … What makes it more scenic?” And I can tell you a story about that, and I can just go on indefinitely.

Nick: So you can either conclude from this … Well, first of all, the richness of my explanation is suspicious, because every time you create it, more appears. And you might think, “Well, have I just got this whole book loads of information, just waiting there to answer your every question, already to go, or am I in fact making it up as I go along?” I’m just thinking, “Gosh, you’ve got me there. I’ll cook up an answer. Oh, here’s an answer. You got to give me a new question. Oh, I’ll give you a new answer to that.”

Nick: And of course, if I’m improvising, if I’m doing that creative, making it up as I go along thing, you should be able to catch me out, because you should find that the answers I’m giving aren’t referring back to some common source, some sort of fixed book of my mind, which is just guides in my every action of thought. Instead, I’m like a novelist spinning a story, and my characters and the different lines of the plot are inevitably, if I’m not very, very careful, going to get tangled up and not quite make sense, and there’s going to be various inconsistency creeping in. Ad the truth is you ask people about that, that’s exactly what you see, inconsistencies everywhere.

Jim: Yeah. I think I got halfway there. Interestingly, as I was sort of processing this morning, my reaction to reading the book was, while I should have gotten all the way there, I was sort of halfway there, in that I had looked quite carefully at Mike Gazzaniga’s split brain research. And his strong evidence are just very fluid confabulation, right? We just make shit up to make the story as consistent as possible. And you do go through a lot of that kind of evidence. So let’s start talking a little bit about the evidence that supports the argument, and let’s start with where you do with visual processing and the grand illusion.

Nick: Yes. Yeah. So I think this is a very interesting place to start, and this is something that’s been much talked about in philosophy and psychology, starting [inaudible 00:15:41] Dennison and many other important philosophers and psychologists. And so, this is the idea, this question of when we consider the world around us and sort of look about, we have a sense of enormous richness. This is an illusion very similar to the illusion we’re going to come to in a minute. It’s an illusion that we’re thinking just about perception, not thinking about our relative and beliefs, but you’ll see that this is very parallel.

Nick: So as I look around the world, I think, “Yes, it’s fully colorful everywhere and seeing in crisp and clear detail.” If I look at a picture of some texts, or a page of text, I think I see words everywhere. Maybe how many words do I see at once? I don’t know, 50. Probably [inaudible 00:16:21] whole page worth, but a lot. I see words everywhere. I look at a school photograph. I think, “Well, I recognize all these people.” Maybe not all the ones in the far distant periphery, but I’m having a sense that they’re also recognizable.

Nick: So I have this really rich sense of my visual environment. And yet, we know that we know that is an illusion for very basic physiological reasons, apart from lots clever experiments. So the very basic physiological reasons, which we all know very well from in school biology, but tend to conveniently forget … Well, first of all, let’s take color.

Nick: There are two types of cell in the eye that detect color. There are cone cells and rod cells. Cone cells detect color, and rod cells just really detect sitting motion. They’re black and white detectors, so they don’t see color.

Nick: Now, the cone cells are very concentrated around the fovea, the center of the visual field. So as I look at something that part of my retina, that the light will be falling on is the fovea, a little clump of cells, really dense, about between one or two degrees a visual angle. It tapers off outside that. And that is where the cone cells live.

Nick: Now, there’s a little … around 10 degrees out, till about 10 degrees out, there are some cone cells, but not very many. So it’s roughly true that all of your color vision is in a little channel. It’s a bit like a sort of headlight. So where you’re looking right at a patch of color, you can read off as color. As you go very far outside the patch, you haven’t got the cells to do it. You simply can’t possibly know what color a thing is.

Nick: And that’s really weird, because then I look at the world now, I think, “Well, that’s just wrong. Isn’t it?” I can see color everywhere, but actually that’s not the case. So let’s just hold, hold that thought for one.

Nick: But it’s not just color. It’s also detail. So cone cells in the fovea are also what process details, sort of structure. And for that matter, if you go too far outside into the periphery of vision, the image is simply blurry. And in fact, very blurry if you went to 20 and 30 degrees outside of the visual center. So if I look at something and think of, “How much can I see 20 or 30 degrees off the center from where I’m looking?” Well, the answer is, it can’t be very much because the image is horrendously blurry. But again, I have no perception of that.

Nick: So I have a sense that the world is in perfectly clear focus. I see all the detail, I see all the color, but we know that’s wrong. Now, where does that illusion come from? What’s driving it? And I think the answer to that … [inaudible 00:18:54] there’s sort of two perspective, actually. I should say, it’s not just one view in this area. One perspective would say, “Ah, filling in.”

Nick: What your brain is doing is it’s getting really detailed, radical information from the fovea, but it’s sort of guessing about the rest of things. So that’s one possibility. We’re implicitly guessing. And as we move our eye around, we can sort of adjust the guesses, and see where they’re right and where they’re wrong. That’s one possibility. But the other possibility is that we’re in fact not filling in.

Nick: In fact, what we’re doing instead is just sampling a tiny portion of the image. And the sense that we see everything comes from the fact that as soon as I wonder about any aspect of the visual image, but I think, “What color is that?” Or, “What word is that?” Or, “Who is that in the picture?” I can immediately answer the question.

Nick: So the idea is that I haven’t got the answer preloaded. I haven’t guessed it either. It’s just that as soon as you ask me … if I asked myself a question about color or shape or identity of a word or anything else, incredibly speedily, my eye can flick into action, look at the thing, read off the color or shape or the identity of the word. And before you know it, in fact within a couple to 300 milliseconds, there’s an answer. Not so quick, but essentially it’s instantaneous to me. It’s as if I’ve got this information loaded into my mind.

Nick: So I think of myself, and then in some sense I really am, is having the world as it were, at my visual fingertips. Ask me anything about the world around me, I can answer it. In fact, that’s what it is according to the second view, to feel that the world is, as it were, present to you in complete detail.

Nick: It’s not that you’ve actually loaded this into your brain at all, it’s that you have it ready to sample and extract so speedily, that since you improvise the answer so quickly, that you have the feeling the answer was already there. And I think that’s an interesting …

Nick: First of all, it’s very interesting in itself, that our intuitions of our perception are so wrong. So we have the feeling of this great richness of the literal world. And that’s absolutely not the case. In fact, I’ll give you an illustration of that in a moment. But it’s also an interesting clue about our creative abilities and improvisational abilities as [inaudible 00:21:12], where we’re phenomenal improvisers.

Nick: And because we’re such good improvisers, we don’t realize we’re improvising. So you asked me a question about the color of some distant object, and I’ll tell you. And I think, “Well, I saw that all along. I knew what that color was. You’ve asked me, and I told you, but I had that loaded in my head.”

Nick: But in fact, if I wasn’t looking at it, I really didn’t have it loaded in my head, but I could answer the question so fast, it feels as if I did. And I think that’s true when we’re answering questions about our motivations and our beliefs and so on. You give me a question, I can answer it. And so speedy, am I with my answers, that I think, “I didn’t just make that up. That was in there. I knew the answer to that.”

Nick: I’m going to give a couple of quick examples in a perceptual context. One, I think, is really astounding intuitively, but it’s been known for a long time in eye movement research. This goes back to work in [Keith Rayner’s 00:22:03] lab. Now dead, sadly. A very remarkable pioneer of eye movement research.

Nick: So in his early work, when he was at MIT, when to Amherst after that. I think [inaudible 00:22:14] towards the end of his career. In his very early days, at MIT in the ’70s, he started working on a very simple, but very interesting paradigm. So you read a sentence which is presented to you on a screen, on a computer. And as you do it, they track your eyes. So they have a simple setup for watching where your eyes go, so you can see that the eyes quite accurately move as you hop across the screen.

Nick: So when you’re reading a sentence, you will be essentially jumping word, by word, by word, roughly. Sometimes short words get jumped over, and long word should be a couple of jumps. But essentially, your eye’s hopping, bounce, bounce, bounce, bounce, bounce across the screen in discreet lumps.

Nick: Actually, another illusion we have is that it feels as if our eyes are moving smoothly when we read from left to right. But in fact, that’s not right. [inaudible 00:23:05] jumps.

Nick: Oh, and just another thing, actually, if we just [inaudible 00:23:09] about the illusory nature of perception, another thing that’s also interesting and true is when you move your eye, you’re essentially blind. You can’t see anything while the eyes in motion. And so, as you read, you have a strange interplay of picking up information, complete blindness, picking up information, complete blindness. Again, we’re completely oblivious to that too.

Nick: Anyway, the important thing is what Rayner was able to do with the information about where your eye was looking, and that was to feed that back to the computer and to change what’s on the screen in consequence. So the clever trick was thinking, “Well, what would it be like if we give you a line of text and only where you are looking fairly exactly?” In fact, the critical window turned out to be about 15 characters, long. 15 counters around where you’re looking. In fact, five going to the left, 10 to the right. We read left to right. We’re slightly more interested in what’s coming next that what we just read.

Nick: You’re going to have a 15 character window, which is unchanged, untouched, no funny business. But all the rest of the words in the line of text will be Xs or Latin or anything you like. So what you’re going to do is you’re going to read a text, where as your eyes go from left to right, as they bounce across the screen, the actual text itself will change. So wherever you’re looking, text will appear as if you were seeing the whole sentence. But where you are not looking, then you’ve just got Xs.

Nick: So if I were looking over your shoulder, as you were doing this experiment, I’d see a line of Xs with some coherent text. The coherent would start on the left. It would go bounce, bounce, bounce, bounce, bounce across the screen. So to me, it would seem like a really weird thing. I’d see this strange pattern of some orderly, meaningful words, moving across a sea of Xs, or Latin, or something else. But if I asked you in the experiment, “Well …

Nick: But if I asked you in the experiment, “Well, how does that feel?.” The answer would be, “Well how did what feel? I was just reading some texts.” If I said, “Well, did you notice anything funny about… Did you see any X’s? Did you see any Latin?” The answer is, “No, I was just reading text normally. There’s nothing funny going on. I just read like I normally read.” And the point is that the, the most shocking is that if you give someone this kind of what’s called gaze contingent display so that you’re driving what they see based on the location of their eyes, you can give them a little tiny window of information around where they’re looking. And the rest is going to be essentially junk, and they will never know the difference.

Nick: Now it feels like the full text is available to them. They can see the whole sentence. And it’s also in some sense that’s right because as soon as they want to go back, “Let me just check back. I missed the beginning of the sentence. I’ll check back quickly.” It appears. Just as you go to look at it, it appears. You want to go jump forwards a bit that will appear too. So you have the sense that there’s, the full sentence is there for you, it’s at your fingertips. And it’s sort of is, but it really isn’t present because of the fact that the stimulus has been, large chunks of it have been deleted. And the same principle applies not just with text but applies to anything. So if I showed you a school photo, and you look at a particular face, but all the other faces are slightly blurred, that’s… You’re never going to notice that either. As long as where whichever face you’re looking at, I make sure to unblur that one just as you look at it, then you’ll say, “No, no, they were all in perfect focus.”

Nick: And in fact that has to be right because actually where you’re not looking, the world is blurred. That’s just the way it is. It’s blurred, and it’s pretty colorless, and it’s not very detailed. But we don’t notice this because we’re not looking. So I think that’s both weird and interesting in itself. And I think also it gives you a sense that it would be slightly odd to think of the filling in story as being right. Because if I’m just reading a patch of text, it seems a bit weird to say, “I’m filling in all the text that’s coming in, all the text that’s happened.” And I haven’t read it yet. How can I be filling the next stuff? It seems a bit of a strange perspective. But I think the better perspective is to say, your sense is a thing, being real to you is the sense that… Ask yourself a question about it and answer it really fast. And so that’s I think the same sense we have when we think about our own motivations, and beliefs, and so on.

Jim: That’s really interesting. And, you actually answered a question that I had while I was reading. I said, well, “Why would evolution have created the sense of more completeness than is really there?” And the answer you gave is, “The real world is relentlessly consistent. For facts to hold good in the same world, they cannot be contradictory.” And then you also said, “However, inconsistency and sparseness are not just characteristics of fiction. They’re also the hallmarks of mental life.” So interestingly, I took away from your words, that evolution has been playing a very clever grand illusion, which is to approximate the consistency of the real world with very sparse and inconsistent data. A pretty cool trick.

Nick: Yeah, I think that’s exactly right. Yeah, I think that’s exactly what the brain is doing. But if we’ve had… Imagine the opposite subjective experience. If you imagine that every time you looked at something, it suddenly sprang into detail in color. Then when you looked away, it faded into murky colorlessness. You’d have the sense that the world was changing. But the world isn’t changing. You just change where you’re looking. And when you look back, there it is again, nice and colorful. So we have the sense that what perception is trying to tell me is what the world is actually like. And if the world is not changing, I’m just moving my eyes, and I don’t want the brain to be saying, “Goodness, me. Everything is getting very… There’s color up here. Ooh, it’s gone again. It’s come back. Details appearing. It’s going. It’s coming.” The fact is there’s no actual change in the world so I don’t want to confuse the system by seeing changes where there aren’t any.

Nick: And the other point about consistency I think is crucial because the real world has to be consistent inevitably. So that means that when I… If, of course, if you’re in a devious psychological experiment, that gets strained a little bit because then you start to mess with people so you can get them to look around the picture. And one moment they’re looking at one thing, then they look at it again, and you can change what they saw and see if they notice. So you could start to mess with that. But normally, the world doesn’t play horrible tricks on us too often. And so we get along fine with this perspective.

Nick: But I think where we go wrong is thinking, when we imagine ourselves introspective, is thinking that’s the same thing. The thing is with introspect, with perception, there’s a physical, solid, external world, which is generating the data that we’re analyzing. When we’re introspecting, we’re not looking into some inner world. In fact, there is no inner world. What we’re doing is we’re interpreting our behavior and also to try to make sense of our actions, but we’re not able to do that by viewing, as it were, viewing some of the concepts of some sort of inner reality. And the trouble is because we’re trying to make sense of ourselves, because that’s fundamentally a creative process just as if we’re trying to make sense of a fictional character or another person, that is not a… There’s not a solid thing we’re perceiving. We’re not perceiving anything. We’re inventing, and the inventions will turn out to be very sparse and also very inconsistent.

Jim: Yeah. You gave a wonderful example where you move from the issues with our external perception to our internal perception. [inaudible 00:30:35] take a mental image and see what you actually know of it. You tell an interesting story about the tiger. Maybe you could tell that story.

Nick: Yes. So I think most of us have the intuition that we can imagine a tiger pretty clearly. And certainly we see tigers and other things in dreams and have a sense that it was vivid. It was almost, it was like it was really there. And so my question to my myself was to think, “Well,” and this is actually, this is the original example. It probably goes back further than this. It goes back certainly to [inaudible 00:06:04], psychologists and computer scientist interested in imagery. So the question is to ask yourself the question about your image of a tiger. So the thing that [inaudible 00:31:14] used to focus on is to say, “Oh, how many stripes are there on its tail?” To which the answer is, “[inaudible 00:31:21] hmm. Well, I ought to know the answer to that because I mean if I could imagine it that clearly. But let’s not leave aside the stripes of its tail because that’s [inaudible 00:31:30] really. The question is to think how the stripes worked around the body at all.

Nick: So for example, you might think, “Well, what direction are the stripes as they go around the body?” Do they go along, long ways or around the tiger’s body. And then what about the legs? How do they go?” So you try to imagine maybe a tiger’s stripes go round the legs and maybe round the body. So, but what happens when they join together? How does the striping pattern join between the legs and the body? How does that work? And it just gets very baffling. Now, if you were to look at a picture of a tiger, which I recommend that people might do, you’ll find that actually quite large parts of the tiger, they are the crucial, tricky bits to some degree, don’t have stripes at all. They’re just not striped. So that actually some of the questions you’re asking yourself, there’s an answer, which is, “Oh, that’s a white bit. Well, there’s just no stripes there.” And until you look at the tiger, you just can’t imagine that at all. You’ve also got the question of how the stripes work around the face. And it’s just something that when you see a tiger, you think, “Of course. That’s the only way it could be.” Try to envisage one. Try to count the strips. Try to imagine how they go, how they interface with each other. It’s absolutely hopeless.

Nick: And I think this is a really general sort of trick that our mind is playing on us to give me the sense that we can reconstruct animals, people’s faces, all sorts of letters, actually. If you just try to… Some of you may know how to do this, but try to write a typical lower case g, as printed in something like Times New Roman. What does that look like? And I’m just trying to imagine it now. And I know it’s pretty weird, but there’s not… It’s not like a g of handwriting that’s for sure. And yet this is something I’ve seen you, we’ve all seen, literally hundreds, thousands, millions of times in our life. We can’t even remember very clearly what that looks like.

Jim: I couldn’t tell you. It’s interesting. Yeah. Let’s dig in a little bit. We are a little short of time because there’s so much interesting here. Let’s make sure we move along. One of the things that’s sort of key on your argument is that basically everything is perceptual processing and that includes memory, which at first seems a little contradictory, a little off putting, but again, there are arguments from cognitive science that memories are really nothing much more than snapshots of perception. Could you talk a little bit about… And this is an area I’m very interested in. A lot of my work goes in here in the area of the linkage between consciousness, memory, perception and attention. So maybe give your thoughts on what memory traces are actually and how our processing of memory is very similar or perhaps identical to our processing of perception.

Nick: Yeah, yeah, absolutely. So I think the intuition is that we often have is that perception is kind of front end to mind sort of filters the input, makes sense of it. And then we store some kind of abstract representation of the output. We pop this into a memory and that memory system might be really different from the way perception works. It’s just that the next stage in the process as it were. And I think that’s a very misleading way to think. In fact, I think it’s misleading really to think of memory even really existing as a separate thing. Clearly, may remember lots of stuff, but the idea that there’s as it were a memory bank, which is somehow separate from the process of picking up and perceiving information, I think is a misguided one.

Nick: So this is a sometimes known as a proceduralist view of memory. So according to this sort of idea, you’ve got, what you’ve got is a set of mechanisms for perceiving the world or even for that matter also acting in it. So if it’s seems about remembering how to swing a golf club say, in some sense that’s… You’ve swung a golf club lots of times, maybe, in which case, your memory for doing that swinging, it’s not really stored somewhere separately from the motor commands that actually allow you to swing the club on a particular occasions. It’s just simply an accumulation of all those actions. And similarly, if you’re thinking about your memory of particular people or events or tigers, those are things that you have perceived in real time as you’ve encountered them. And those perceptions leave a trace. It’s not that they produce an output. They sort of give their verdict and say, “Ah, what happened there was this. I pop that in the memory bank.” It’s more that they, that the process of understanding the world from, all the way from the details of the low level vision features all the way up to and what was happening there. Was that person angry or happy or whoever they’re going with, that implement, whatever it was. All of that is that process is simply laying a trace as it works in situ.

Nick: So the idea of it from this perspective, [inaudible 00:36:19] perspective. And the idea is that if you have damage to a part of the brain that is involved in some space of perception, you should also have trouble with imagining or imaging that aspect of the world. And also you should have trouble remembering it too, unless you recoded that memory in a different form. So the thought here would be if, for example, you have an area to, damage to an area of visual cortex, which is concerned with say color or emotion or whatever it may be, let’s say color, then you will struggle to see colors and process them. But you’ll also find you imagine them either. And also when you remember images of objects, you won’t know what color they are. Now you might have remembered in words as it were tele boxes are red in the UK. I know that. I’ve… That’s verbal information I’ve got. So in that case, I can answer the question, what color is a tele box. If I try and visualize a tele box, if I’ve had damage to my visual cortex, I can’t do it because the memories are stored in the same place as the perception, or in case of actions, the memory of how to play tennis is stored in the circuits that are actually controlling tennis playing.

Nick: And I think, in general, we tend to miss this because we don’t think of perception as abstract as it really is. So you might say to yourself, “Well, yeah but then perceiving is one thing. That’s just sort of knowing what sort of objects there are and what color they are. But when I’m watching a movie or talking to a friend, I’m thinking about so much more than that. I’m thinking about your intentions and purposes. I’m speculating about human nature. Or I might be thinking about a mathematical problem and trying to do some geometry in my mind. Or I might be trying to work out how to put together a piece of furniture. All of these things I think should, you should view those as continuous with perception. And they’re all tasks, which are basically taking data from the senses and trying to work out what to do with it. And once we think of this as a relatively abstract, not just rooted in this kind of like the colors and shapes in front of us, but as a process of abstract pattern construction. And that’s really a good chunk of cognition is just all about that.

Nick: Now, having said that, of course, what makes us very good at this kind of task is that we’re not just locked into the current sensory input. So I can shut my eyes as it were and think about lots of things, which where there’s no sensory input coming in to me because I have all those past impressions, past thoughts, which I can review. But they themselves are just in the same format. They’re the same kind of thing. They’re just sort of past perceptual traces, past memories, memories of past perceptual experiences, rather than being in a completely different form.

Jim: Interesting. Now, one thing that you didn’t talk about much, and what I do focus a lot on in my own work, is as you’d say that, you did say that perception is not just a simple screenshot of what our eyes see because as you go through carefully, our eyes don’t see a screenshot. They see a whole bunch of little things. But one of the high level process things, which seems ubiquitous pretty much quite a ways down in the phylogenetic stack, as far as something like consciousness goes, is the ability to find objects. We perceive our existence in the sensorium as full of objects, not chunks of color and shape and lines and edges. We have, it seemingly at least seems to me and a lot of the research I’ve seen, a very deep hardwired machine for finding objects. In fact, there appears to be an object store in the parietal regions of the brain, the famous Mariah Carey neuron, et cetera, that you stimulate a specific neuron, and you pull up a very specific object or a general object like dogs or tables, et cetera. How does the concept of objects fit in to your sense of perception and memory?

Nick: Yeah, well I think that a very central concept actually. So I want… [inaudible 00:40:22] do quite a lot in the book is that we’re incredibly serial processors. So we can only really do one thing at a time. And in perceptual terms, I think roughly a way to put that is to say, we can perceive essentially one object at a time. So I think of objects as in a way themselves psychologically very, very fundamental because they are sort of the unit that the perceptual system can grab hold of. Now that means, for example, if I’m looking at a word, I can’t see other words going back to the eye tracker. As I read a sentence, roughly speaking, my eyes jumped word by word by word.

Nick: As I mentioned before, you sometimes jump small words, which are predictable. You can just guess those. And for long complicated words, you sometimes need to look at them more than once. Essentially you will read the beginning and then the end. But the… Yeah, broadly speaking, you’re picking up one word at a time. And although you have this feeling that you perceive all the other words, they’re not really there. Now I think… So that’s one interesting thing that the units, there may be a sort of basic level of unit that we can grab hold off perceptually. But another interesting thing one might think is, “But aren’t words made of letters?” So if I can perceive say a word, Dog, well surely I’m perceiving not, it must be perceiving three things. D, O and G because dog is made of three letters. So I must be able to perceive three things as well as one. It’s kind of weird. So three and one trinity is the dog. But I just think that’s misleading too because I think if you ask people to focus on the letter, they can’t see the word. If you ask them to focus on the word, they can’t see the letterings, there’s huge interference between. So there, again, it appears [inaudible 00:17:05] can see lots of words, and I see all the letters within them. And actually, I think that’s not right.

Nick: So let me give you an example from speech perception, which I think I talk about some in the book, which is the phony restoration effect. So this is a lovely effect where you play a word, and you splice out a bit of it and replace it with white noise. So it sounds like a word with a cough. And the question that the subject is asked is, “Where was the white noise. What did you miss?”

Nick: So the famous example is you play someone the word legislature, and I think it’s the sir, you take out the sir and you just replace it with a cough as it were. It’s just essentially just pure white noise. And the question is where was the cough? And people are hopeless at this. They often [inaudible 00:42:53] the cough to the boundaries of the words. It was just before or just after. And just generally they have, generally have no idea what they missed. And so that’s really very nice because that’s telling you that you perceive the word. You know you heard the word, in this case the rather strange and unusual word legislature. And you’re pretty sure of that, but you don’t actually know which sounds you heard. If I asked you, “Did you hear the S?” You think, “Well, I don’t know. I think I probably did.” But you’re wrong. You didn’t hear the S. That was the bit that was chopped out, but you just simply don’t know. In fact, that sort of phenomenon I think is everywhere in perception. If you perceive a whole, you don’t really perceive the parts. As you perceive one hole, you don’t perceive other holes.

Jim: Yeah. A lot of that-

Nick: Another example, which is… Sorry, go ahead.

Jim: I was going to mention that the field of gestalt psychology has been drilling into this for a long time. We perceive a car as a whole, but if we focus, and there’s some even some interesting thought about at what angle do you focus, et cetera, a car has tires and handles and a roof and a boot as you’d say in the UK, et cetera. But when we’re just seeing things go by on the road, we just perceive them wholistically as cars. It takes an additional level of effort and attention, and we’ll talk about attention in a minute, to decompose an object into its components. But you probably, as you point out, you can do one or the other. You can perceive the car as a whole, or you can perceive the tires, but you probably can’t do both simultaneously. And again, this has to do with this granular, but probably hierarchical structure of objects, that we’re operating in.

Jim: Let me drill into another question that I had, which is you don’t, other than very much in passing, dig into the distinction, if there is one and maybe your work would say there isn’t one, between the different kinds of memory with respect to their contents. And then I’m thinking in particularly episodic memory, i.e. rethinking the dinner I had with my daughter last night versus declarative memory. Paris is the capital of France. Do you think there is any difference, or are they just projections we put on to what are essentially a unitary memory process?

Nick: Yeah, I mean, I think… I’m not a… I mean this isn’t a central thing in my thinking, a central point in my thinking though I think it’s very, very important issue. So yeah, my natural starting point, given the general perspective I’ve been outlining, is to think of everything as fundamentally episodic. And so all you really have is a rich collection of specific experiences, and you run through your life processing these experiences as they’re coming to you. And they leave a trace and then walk on. And then those traces will affect how you process the new traces. But the… So essentially everything’s episodes. But we normally think of episodic memory in psychology, of course, when I can as it were pick out when the episode was. So if I say, “How do you know that a particular restaurant has changed its decor?’ You might think, “Well, I know that the because I was only there last Friday. Not so true, [inaudible 00:45:51], of course, [inaudible 00:45:53] but you’d think in the normal course of events, I was only there last Friday. I remember the event. I remember going in and thinking, “Yeah, wow, they’ve changed their decor.” So you have, in that case, a clear sense of where the episode route is.

Nick: And I think semantic memory or declarative memory is broader knowledge of things like the Earth goes around the sun rather than the sun goes around the Earth. And Paris is the capitol of France. And so on. I think of those things as fundamentally the same kind as episodic memory. We’ve just lost the episode. So if you ask me, “Who told, who first told you that Paris is the capitol of France?” And that’s really, “Oh, I’ve just no idea. It’s just people are always saying as it were.”And when was it that you realized that the earth went around the sun? Some time long ago. I don’t know. The difference from my mind would be not the, what you can do with the information. It’s just whether it’s traceable or not.

Nick: Now, I know there are some arguments looking at, for example, effects of brain damage, which appear to show that you can get damage to one memory, not the other. In fact, the one you normally gets is damage to the episodic system. So if you believe that there’s episodes, and there’s this more general knowledge, then that seems to be strengthened by the observation that if you have some kinds of brain damage, and also for people who have very severe alcoholism, you can get a loss of semantic, sorry, episodic memory. So you can have real difficulty remembering much about your life and where you were yesterday and almost your life story starts to collapse, but you may still know a fair amount about Paris being the capitol of France, and the Earth going around the sun, and so on.

Nick: But I think I actually, at least this is not a unique view to me at all, is a very standard counter to that is to say, “Yeah, but actually what those people have is a general loss of episodes.” And that shows up much worse when you ask them about specific episodes because specific episodes have been damaged. If you ask them general things, then any other number of episodes will help them answer the question. So if you me about whether Paris is the capital of France, since there’s lots and lots of memories, which will tell me the answer to that, then I only have to find one of them. But a really hard question, say, what was the last movie you saw? Because I have to pick out a particular episode. It was just a tough task, a difficulty problem.

Nick: And you see this all over the place in psychology, that when you have a single set of tasks, but some are harder than others, and then you damage, people have damage with difficulty, then they can’t do the hard things, and they can do the easy ones. It’s easy to think, “Oh right. Well, probably this hard and easy tasks are done by different systems, maybe different brain areas.” That’s the mistake I think what we maybe falling into that. But it’s a complex and knotty area [inaudible 00:48:34] although I am really fond of and wanting to push the episodic perspective, I think that’s the… It’s not something I can honestly say that this is a resolved question in psychology.

Jim: Interesting. Yeah. I actually had an idea when I was reading your book in that range, which was you talked quite a bit, which we’ll get to later, that much of brain processing is whole brain or nearly whole brain processing of one snapshot at a time. And then it produces an output that’s written to memory. And I had said, “Hmm, I wonder if there are kind of stripped down mini episodes, essentially where all the unnecessary details are eliminated.”

Jim: And so for instance, there’s an entity that kind of looks like an episodic memory, but is less rich, but just says, Paris is the capital of France. And every time I think that thought it gets rewritten. So we kind of combine these two thoughts to the one you just said is that Paris is the capital of France may exist in this stripped down semi episodic memory, hundreds or thousands of times. So sure it would be preserved over damage more than others. Now, the other form of memory that you didn’t address much, although again a little bit in passing, and here are the evidence from damage is stronger that it’s different, is procedural memories. We learn how to play the piano or ride a bike. And even if we have complete elimination of both hemispheres of the hippocampus, that seems to be preserved. That seems like maybe that is a different animal.

Nick: Yeah. Well, again, I want to say-

Nick: Well again, I want to say, I don’t think it is necessarily different, but the thing that is different is that certain kinds of procedural memories are going to be engaging particular cognitive processes, so different brain regions.

Nick: So, if it’s the case, for example, of playing the piano, is broadly speaking subserved by a relatively small sub-set, a relatively small-sub set of brain areas, or such that you can lose a fairly large chunk of cortex which I think we can, and some people have, and still be able to play the piano successfully. Then, you’ve got essentially dedicated neural hardware for this problem, that, that hardware can keep running.

Nick: Now, most problems we engage in don’t really have, or most things we do in daily life, don’t really have a dedicated hardware, but if you do some repetitive task for hundreds of thousands of hours, you may well come up with some dedicated hardware to do that task, reading would be another thing, and to some extent, language too. Language does use an awful lot of brain, high level differential stuff to actually figure out what was worth saying.

Nick: But, if you had a special dedicated brain region as I was saying, then that rather breaks this idea that you can only do one thing at a time. The reason that the brain is such a serial processor, the reason it does one thing at a time, is that it works as a giant distributed network, so the calculations its doing, the processing its doing are spread across billions of billions of neurons, which are all working in an interdependent cooperative fashion.

Nick: So, that means that if you’re trying to solve one problem with a big, cooperative network of neurons, then you can’t get the same network of neurons to do another problem at the same time, because they’re going to get muddled up. So, that implies that if you’re doing a task where you don’t have a special, as it were, dedicated piece of machinery set up, you have to put something together on the fly with the range of mental mechanisms you have lying around, that’s going to take up a lot of brain, and that’s going to basically wipe out the ability to do anything else very complicated.

Nick: But, if you have a dedicated brain, neural machinery, then you may be able to do, run that thing and still do something else. So for example, touch typists are able to read text and type it, if they’ve practiced this a great deal, whilst for example, holding a conversation. In fact, they won’t have any memory of what they’ve been typing, but they can just look at the text, type it out, and they can think about other things, they can have a chat, they can do that because they’ve basically built this very, very specialized piece of machinery over hundreds and hundreds of hours of practice.

Jim: I actually saw that. I had an executive assistant back in the, I guess it was early 80s, and she was an old school secretary before she became an executive assistant. She could type 70 words a minute and carry on a conversion with me, what the hell? Right? Very impressive.

Nick: It is amazing isn’t it? It’s an absolutely amazing thing to see, yes. The fact that it’s just become a completely automated process, and it takes hundreds of hours to do it of course.

Nick: Similarly, with piano playing, so when we’re learning to play the piano, we’re learning to type initially, we don’t have any special, we haven’t constructed any machinery to do it yet, so it’s using our whole brain, it’s struggling away trying to keep the process on track, so any slight interruption could completely scramble your typing or your piano playing. But, once you’re a master, then you can type furiously, you can play a piano piece, not with great beauty and expression necessarily, but you’ll be able to play it while having a conversation, because all of that lower level stuff has been, to some degree, automated.

Nick: Having said that, this automation isn’t perfect, so we tend to imagine for example, that when we’re driving, we can drive, as it were, on autopilot. I can see that I can do that, because look, I can have a conversation, I can see the radio, and I can think my own thoughts even though I am driving along.

Nick: But, it turns out that, that’s not quite right. I mean, most of the reason that we have such a strong sense of automatic driving is that most of the time when we’re driving, nothing is happening, and you’re just going along and the world is just flowing by, there’s no action to be taken. But, as soon as you have to take action, it tends to scramble any other activity you’re engaged in. An everyday example of that is that almost invariably, if you suddenly have to brake suddenly while driving, you will certainly stop speaking. No one carries on a fluid conversation and keeps on going while slamming on the emergency stop, everybody stops and you have to.

Nick: And, the experimental work by Hal Pashler at his lab at UCSD have shown this in a simulated driving task, where they give people a task of following along a car in a very simple driving world and the car occasionally brakes and you’ve got to slam the brakes of your car. It’s not a real car, it’s just joy stick stuff in a virtual environment.

Nick: But, when you do that, you are simultaneously asked questions, very simple questions like, I think it’s something like, for example one version is you get a tone and you have to say was the tone high or low, and you have to say it too, so there’s no physical use of a keyboard or anything.

Nick: But it turns out, if you ask me whether a tone or a beep is high or low while I’m braking, I can’t do both. I just have to do the braking, get the braking out of the way and then I think, all right, what was that question? Then, I answer the question.

Nick: So, even something where the task in the world is very simple, and the two things you’ve got to do, same as that tone high or low, or just keep the car on the road and brake when you need to, they seem like totally independent things, but actually get tangled up.

Nick: So, I think it makes all these remarkable cases, like the fact that people can type and play the piano, and chat at the same time while they do those things, makes them even more amazing, because it really is the exception rather than the rule of cognition.

Jim: Cool. Let’s move on to our next topic, and those people who listen to the show regularly know that it’s one of my obsessions which is attention. In fact, I’ve been known to say that we are our attention, and it has implications in all kinds of things, like for instance social media is constantly working to hijack our attention, et cetera.

Jim: The idea of attention has been moved over into AI, for instance some of the newest work in linguistic processes like the GPT-3 thing that a lot of us have heard about recently, which is this amazing piece of software from OpenAI that seems to have learned a scary amount of language by processing a trillion words or something. You give it a route and it goes off to the races and creates text. It uses a technology called transformer networks which actually uses attention as part of its model.

Jim: So, one of the things that is non-intuitive about our brain states and I think has huge implications for the contents of our memory, is this idea of the unity of attention, that it’s single threaded, mostly with some exceptions that you talk about, and that it processes by single steps. In fact, at my own work where I attempt to build simulated consciousness, how about that? But, for a white tail deer, not a human. I actually make that attention thing discreet, basically just have it change every 250 milliseconds, or not, either it changes or it doesn’t based on a whole bunch of conditions, et cetera.

Jim: But, it just single steps through basically one object at a time, which then activates a different brain state, et cetera. So, you go in quite a bit about this, this idea of the unity of attention and how it emerges from the concept of the whole brain, or the near whole brain processing. So, tell us about attention and how this fits in with your model.

Nick: So, I think Jim, the way you frame the question actually encapsulates the way I think, one should see this, which is to think of ourselves, to say that I am my attention, my focus of attention, and I am essentially a sequence of attention snippets. I think that’s a very useful way to think.

Nick: So, it’s very easy for us to imagine that as well as the things I’m attending to, my brain is engaged in all kinds of other activities, and it certainly is doing some things, it certainly is driving my heart, I don’t have to worry about that, it’s making me breathe and it’s doing all sorts of practical housekeeping things. But, I would want to argue that really most of the activity of your brain is focusing on whatever you’re actively doing now, if you’re speaking, if you’re playing a golf stroke, if you’re focusing on some music, that’s pretty much your brain’s primary business, it’s pretty much entirely focused on that one thing, and there’ll be some peripheral activity going on here and there.

Nick: But, it’s not, for example, the case that while I think I’m chatting to a friend, as a chunk of my brain which is wondering about a tricky problem I was stuck on yesterday. Or, there’s another part of my brain which is worrying about my finances, or there’s another part of the brain which is processing some resentment about some past slight from a work colleague, or all of this is, I think, completely wrong. Your brain, because it’s such a serial processor, it can only really do one complex thing at a time, if it’s doing one thing, it’s not doing anything else.

Nick: So, it’s really right to think of your brain as essentially taking on one challenge at a time, step by step, every few hundred milliseconds it’s making a new stand to let me look from here to there, or move from word to word, or review a thought or follow on with another thought, I’ll look at another part of an image. But, it’s a pretty slow process which pretty much blocks out everything else.

Nick: We do have an intuitive sense of being very serial I think, of struggling to do more than one thing at once. And yet, as with the grand delusion, we also have a sense that oh yeah, but while I’m doing this thing, I can see from a detailed world, I can hear all the sounds around me and I can feel my body on the chair. Well, I can now that I think about it, but I couldn’t before. And, I think we also do have the sort of implicit intuition that I’m doing lots of covert pondering as well, particularly if you’re taking a depth of psychology perspective, you have the idea that there’s all kinds of pondering of unexpected issues in unexpected ways, which is going on in the background. But, I think there’s no real evidence that, that’s the case, in fact I think the brain is really, because of the way it’s built, the fact is it’s this big distributed system, it’s really hard for that system to work on two problems at once with the same machinery.

Jim: I’ve actually found that to be a very interesting and challenging assertion on your part, and as I thought about it, I said you’re probably right. I will confess to being a person who has long believed in the background processing, because I felt it in myself, right? I’ve often written software as part of the work that I do, and not being a professional software developer, I can easily paint myself into corners and things, right? At four o’clock in the afternoon I go, fuck it man, I just can’t make any progress here. Let me go sleep on it, and I’m sure I’ll have the answer in the morning, and amazingly often, I do.

Jim: But, you gave a very interesting alternative way of looking at that, and I suspect that you’re right. Maybe you could talk a little bit about your idea of resetting and getting rid of assumptions et cetera. And then, an alternative argument for the phenomena that we can mistake as background processing.

Nick: I think this is a very powerful intuition that we all share, that you try to solve something and you can’t, and you come back and suddenly it’s clear. It is very easy to imagine that what’s been going on, and this is a whole idea going back to the fact that it’s a strand of [inaudible 01:02:03] psychology that we mentioned earlier on, that there’s this idea of incubation and problem solving. So, the problem is being mulled over intuitively even though you’re not actively addressing it.

Nick: I think the alternative is not, this is an old idea in conative psychology, but I think a very convincing one actually, is that there is something else that’s going on. So, what’s happening when you’re trying to solve a tricky problem is you’re searching in a space of solutions, and you’re in a cul-de-sac. If you weren’t in a cul-de-sac, you’d keep going and you’d just think I’m nearly there, I’ve nearly got it, and you would keep going until you got it. In fact, you stop and you get baffled because you’re in a cul-de-sac, you’re can’t, you’re going to have to go forwards.

Nick: So, if you keep going, you’ve somehow got to get yourself out of the cul-de-sac, and that’s often quite hard. In fact, you’ve often thought about possible ways out which clearly haven’t worked, because you’re still stuck, and it’s very hard to generate a new perspective on a problem which would get you out.

Nick: What a break with often do is just reset, so you just come at a problem, you’ve forgotten where you were and that’s good, because where you were was bad, you were in a bad place, and you just start again. So, that sense of now it’s obvious to me, is because now you’ve started from a different place, and that place just happens to be better. Often, it won’t be, and certainly often you’ll just be in another bad place or another place which you have to explore for a long time to work out whether you’re better or not.

Jim: The way I articulate it to myself when I read your theory was that my brain activations are still resonating away from the places I’ve had attention say for the last half hour, right, its upgraded or to a lesser degree, or it’s somehow impacting the whole brain network, everything I’ve paid attention to for a while, right? The brain’s state is conditioned by what I’ve been thinking about for the half hour. And, because I’m at a cul-de-sac as you said, those were not the right paths, right, and just stripping all that prior history of attention out or at least having it way lower activated the next morning, allows me to, whatever my internal processes are, and we’ll get to that in a second, converge to something that’s right, and it’s in a different set of attractors essentially.

Jim: I will say, it seems like you quote some evidence that it’s essentially just random, but doesn’t feel like that to me, and again, maybe this is another illusion that I am way better the next morning on solving a stuck point in programming than I was at 4:30 in the afternoon the night before, so it doesn’t feel random to me, and it maybe that there’s something deeper going on in clearing out a large amount of residual brain residences from the steps of attention.

Nick: It’s maybe that sleep is somehow very important to you, because I think there was some evidence that rest and sleep actually do lead to a better rebound defense. So, from the perspective we’ve been exploring, you’re more fundamentally shifting your play, your attractor, and getting unstuck more effectively, so that does appear to be the case.

Nick: But, if you’re trying to get experimental evidence that people are really doing incubation properly, its really impossible, or at least has proved to be excruciatingly difficult to get such evidence. So, the kind of thing you do is you give people so many experiments, give people for example, anagrams to solve and some of them you can solve, and some of them you can’t, and the ones you can’t you just go round and round and go, no I just can’t do it, so you then give people a break and then you come back and do them again. It’s very rarely the case that they do much better after a break, they maybe do a little bit, but it’s really … And, the length of the break doesn’t seem to make much difference.

Nick: So, it’s not as if you give them two hours and they’re much better, they’ve got all the anagrams sorted out, because they’ve had all that time to do the processing. No, it’s not the case [inaudible 01:05:53] slightly less stuck [inaudible 01:05:55] slightly better. But, it’s really much more consistent with a sort of diffuse, nudging you out of your cul-de-sac rather than my brain is just knocking off the anagrams one after the other, solving them all, knocking them back, bang, bang, bang, it’s not like that at all.

Jim: I should make a note for the listener here, just in this example that Nick gave us, each one of his claims he has backed up with actual evidence, and it’s well footnoted and the footnotes actually have interesting discussions, so these aren’t just dry footnotes. Footnotes are worth reading, which is not always the case for those who want to have some sense of confidence and Nick just isn’t making this shit up, even though of course he is, based on his own theory, right?

Nick: Absolutely, rather ominously myself, yes.

Jim: I’m going to hop to another topic, it’s one that I wasn’t convinced of, but I’m not going to say that I’m right to not be convinced, but I’m on the bubble, which is you argue that human thinking basically has no principles in it, right? We’re not really doing our thinking based on principles we’ve somehow extracted from our thinking about the world, and you even make the, to my mind, fairly extreme claim that even chess masters have no principles, they basically are just clever at how they remember real boards.

Jim: I wasn’t convinced by that argument, and in fact, let me play back the thing that caused my ears to go up, which is that you mention the well known result, that chess masters are no better than anybody else at recalling random boards, but they are way better at recalling the boards for real games, that either they played, or even any real game. Just show a chess master 10 positions from 10 random real games, and they’ll remember them. But, they won’t remember random boards any better than you or I.

Jim: I said, wait a minute here, doesn’t that say that there’s got to be some principal that’s been a filter applied to perception that allows the chess master to know implicitly at least, that one was a real board and one was another? And, why that particularly resonated with me, my motivation for getting deeply into conative science was something surprisingly simple. Through all my life I’ve played war games, back in the days when I was 10 years old and they were little cardboard pieces with tanks on them that you moved around on a hexagonal paper board, through all the evolution of the online stuff, et cetera.

Jim: And, in 2012, I learned a new computerized game called Advance Tactics Gold, which is very complicated, very realistic, very good and very hard, at least so it seemed. I sat down and learned to play it, it kicked my ass bad the first couple of games, I got a little better, a little better, and this is getting way, way, way more complex than Go. In fact, the calculations I did said that the average number of move possibilities is 10 to the 60th per turn, 10 to the 60th, as opposed to maybe 200 in Go and 30 in chess and five or six in checkers, drafts.

Jim: Anyway, I also stopped and thought about it a little bit, how many of these war games have I played in my life? And, gross estimate at that point was I probably played a game to conclusion 2,500 times in my life, which is a very low number compared to machine learning, right?

Jim: Continue this story, Advanced Tactics Gold, good game people, if you like games, check it out, it’s on Steam. After seven games I had basically extracted enough, and I would argue, applied my principles enough, so the sucker never beat me again. I probably played it a few hundred times since then, 100 times at least, won every time. Sometimes difficult to win, but I’ve won.

Jim: What we know about machine learning, there’s no way you could get that good on only playing 100 times, impossible, right? There has to be something else going on, so that’s my push back on no principles.

Nick: Well, I think it depends on what you count as a principle, because I think what you’re pushing back on is something I’m extremely sympathetic to.

Nick: So, if you think about how machine learning works as a model for the mind, which I think is an interesting thing to do, then the shocking thing is that it needs so much data, exactly as you’re saying, you’ve got this enormous space of possible games and you play in a tiny sub- space and you learn to play pretty well and start to win. Even though, you can’t possibly be doing that in some simple minded way comparing the current game to previous games, because the space is just way, way too big. Now, I think that’s absolutely true.

Nick: So, one view of what’s going on is that the brain must have the ability that the machine learning systems doesn’t have to work out some underlying, as it were, perhaps quasi-mathematical set of rules which tell you how to play well. So, instead of just thinking, oh I’ve seen a position like this before and I did this and it didn’t work, I’ll do something else, or it did work, so I’ll do it again. Instead of doing something like that, you’re somehow breaking out and having some general understanding, or some sort of general theory of how to play the game.

Nick: Now, I think that is wrong. I think the brain doesn’t do that very well. We have intuitions about how we play chess, or Go, or any game, but the intuitions are pretty poor. So, when, as I’m sure you know, when people started to build chess playing computers, one of the first things they had a go at is just asking chess players, how do you play good chess, which, is a very sensible thing to do.

Nick: But, actually chess players can’t say very much. They can certainly say things like well the queen, you don’t lose your queen, don’t rush your king into an attack, that’s a terrible mistake and control the center of the board. And, other things which are not worthless, but they are fairly incomplete and minimum. So, famously you have books, I can’t really think of the author, some great Russian grand master has a wonderfully titled book which says something like the principles of chess strategy, or something. What does the book contain? All it contains is a stack of cool examples, famous games, tricky problems, with an intuitive explanation of how to deal with this kind of case.

Nick: But, what it doesn’t do is say, now the principles of chess, here they are. Of course, if we need the principles of chess, if there were any, it would be useful then, obviously, building chess playing computers would be much easier than it is.

Nick: So, I think the right lesson to draw from that kind of case is to think well, what’s interesting about human learning, and this is what the quote, supposed principles of this chess book was telling you is, it’s saying, think deeply and reflectively and thoughtfully about the cases you encounter. So, you’re not just storing them as it were as raw data, and thinking if I see raw data that looks a bit like this again, I know what to do. You’re not doing that, you’re thinking in a very rich, metaphorical and logical way of thinking, now, what’s going on here? What kind of thing is this? Have I seen anything a bit like this before?

Nick: I think this is the ability that’s really astonishing, and it makes us so remarkable, it’s not that we can play chess by working out, as it were the mathematical principles about what to do, how to play a good chess game. But, what we can do is think this position I’m in now reminds me, possibly implicitly, of a number of other positions, and in those positions, the kind of thing I did was this, so probably this is the kind of thing I’m going to do now, so it’s a higher, more abstract level.

Nick: Now, how that works, how the brain is able to operate at this more abstract level, so that it can learn not just from the very specific thought position, but the class thought positions, how that works is a kind of holy grail of cognitive science and AI in my view, it’s the kind of thing that Doug Hofstadler was doing so much work on 40 years ago, and ongoing, is try to understand the fluidity and creativity of our ability to deal with our particular experiences, extracting analogically. This is a case, a bit like that previous case and doing that in a way that is coherent and that’s remarkable that you can do it, but I think that’s what we’re doing.

Nick: Because, if we were producing principles, well first of all, people can never articulate them, but also most complicated problems, and your game would absolutely be one of these I’m sure, there’s no real algorithm. The algorithm, if there is an algorithm to play that game well, is almost certainly very memory based, in situations like this, do this, in situations like this, do this other thing. It won’t be something more abstract than that.

Nick: So, I think if you want to call those principles, that more abstract we’ve seen in the world, then I think that absolutely is what the brain is doing. What it’s not doing I think is producing a set of rules or an abstract theory, which is a consistent understanding of how to play the game, or how the world works or anything of that kind.

Jim: Yep, I agree 100%. In fact, the word I always use, I didn’t use principles, I just used your word, principles

Jim: In fact, the word I’ve always used, I didn’t use principles, I just used your word, principles. I call those heuristics. And in fact, my mission, and I don’t know if I’ll succeed because I’m too old, God damn it, I wish I’d started this when I was younger instead of when I was 60 years old, is heuristic induction. How is it that we induce heuristics? Because there’s no doubt about … at least based on my introspection, but as you point out our introspection is dubious when I’m playing my game or any games. And in fact, since I did all that work with Advanced Tactics Gold, I even got the source code from the developer, wrote some evolutionary algorithms that improved his AI. I played some other games and I just see again and again and again, I develop heuristics, which are definitely not algorithms. You can’t say that these five or six or seven rules of thumb for that particular game are quite enough to win.

Jim: But if you have a brain that’s been tuned to extract patterns from games, you can then apply the heuristics in a pretty straightforward fashion. For instance, there’s a whole class of games where it’s really important that you keep your line consistent across the board, so guys can’t get past you, right? Kind of true in chess, but really true in some of these military games. And then there’s also, one of the things you learn when you play one of these games is, the relative efficacy of defense versus attack. For instance, having played several hundred titles over my life, maybe a thousand … no probably several hundred, a meta-pattern is that often … it turns out to be true in real life in war as well, that it’s often better to take the tactical defensive and get the other guy to attack you first and then counter attack, right? That turned out to have been the tactical brilliance of the German army in World War II, particularly in the later phases of the war. So it’s heuristics and how we induce heuristics from these otherwise flat patterns, I think, is still a huge open question and is right at the very core of my own interest in this space.

Nick: Yeah. And I think it’s at the … as you were hinting, it’s at the very core of intelligence, really. What is astonishing is that we can produce these sort of rough patterns, heuristics and so on. And they are, as you say, they’re not sufficient on their own to solve the problem, if you coded those into a computer it wouldn’t be a winning computer. It wouldn’t be a winning system on its own. But if you put that together with a crazy flexible brain that can somehow know, which heuristic is the right one to deploy. And when that’s a clash think, “No, no this is clearly the case that this one is much more important that one, but there may be some other cases where it’s the reverse.” That ability to kind of knit together, a difference of these heuristics is very remarkable. And I think we do forget just how unbelievably good we are at finding complicated patterns in very small amounts of data.

Nick: And so, the fact that you can play a game, with an enormous space of possible moves and get pretty good at it quickly, that’s very interesting. Similarly, science manages to produce principles and really does produce principles, they’re not in the heads of the scientist directly, but the things we have to produce as a community, the same in sciences, which have been very successful, such as physics. Amazingly, it turns out you can produce a relatively small number of principles, which explain huge amounts of reality. But you’re not doing that by, as it were, surveying the universe and just taking in as much data as you possibly can. In fact, you’re narrowing down on tiny, tiny amounts of data and sort of carefully dropping balls down slopes of different inclines and carefully watching pendulums go backwards and forwards. Then remarkably, it turns out that you doing this leads you to generalizations which helps you understand the universe as a whole, it’s just an amazing thing. But it really doesn’t work by being a kind of data hoover the way machine learning is, which is take a trillion words, or a trillion images and just find some, find the structure to all of them. The human brains seems to work in a really different way.

Jim: Yeah, it is interesting. And you know, and we alluded to it a little bit in passing, that one of the other cutting edge things learning on small data is something people are quite aware of as the distinction between humans and data … In fact, I noticed you had in your acknowledgement section my friend, Josh Tenenbaum, has done a lot of work on small sample learning and ways that are rather different than brute force machine learning.

Jim: And you know, this is very hot area, but anyway, the other area where this is … there’s a lot of tension right now is about language, right? Particularly with the new GPT3 out, which we alluded to earlier is this massive trillion word database that’s been processed by several billion node deep learning network, to produce a language model. And it’s scary how good it is. However, it does not appear to actually understand language. It doesn’t take a genius to figure out how to trick it. And so probably human language learning is something different than just brute force pattern matching on a trillion words. Because, frankly, we know we don’t get exposed to a trillion words. We have pretty good mastery of language by the time we’re, what? You guys work on that language three or four years old, right?

Nick: Yeah, [inaudible 01:20:13] pretty good. Yeah. Yeah.

Jim: Again, something like principles, something like, probably not Chomsky and universal grammar, but something like that, maybe is there?

Nick: Well, I think again, it’s going back to the point that cognition for me is perception. I want to see all of these cases where you’re abstracting from specific concrete experiences, but you’re maybe abstracting quite a high level of generality. So in learning a language which is something that Morten Christiansen and I have been thinking about a lot over the last few years, you should think of learning a language as learning a skill, which might intuitively sound to most people as very natural. Of course, it’s learning a skill, it’s no more or less than playing the piano or learning to play sports or indeed a game. But traditionally it’s not viewed like that. It’s viewed as, at least in Chomsky’s perspective, it’s viewed as a problem of learning a body of knowledge which tells you how … what sentences are allowed in your language and how they’re composed.

Nick: And that abstract body of knowledge is the real target of learning. And that is supposed to take you far beyond mere skill learning. But I think that the approach that Morten and I, and others, many others, would take is to see language as an abstraction from lots and lots of concrete experiences and what you’re learning, you’re learning specific words, you’re learning specific patterns of words, you’re learning specific relationships between singulars and plurals and so on, verbs and nouns.

Nick: And all of these patterns and constructions are learnt, the term of art in linguistics at the moment that’s the movement, construction grammar perspective takes this kind of perspective. You’re seeing the task of learning a language as learning all these little micro patterns and they gradually build up and interact, to give you a complicated system, which is the system of language that we speak. But really you’re learning it sort of piece by piece by piece. But you’re learning it as a set of sort of chunks of information. So chunks of knowledge, you’re learning about particular nouns and which verbs that go with your learning, particularly about how certain kinds of endings work. And you’re not, as it were, just blitzing the entire corpus that you’ve ever heard. And you’ll find these patterns and building on those patterns, are building incrementally more and more.

Nick: So I suppose there’s some very interesting step by step process of learning that we’re engaging in, when we’re learning a language, which is very different from the kind of blitz that the machine learning system will do, which allows us to learn from very small samples, the typical child will hear a few million up to perhaps 30 million or so in a very high linguistic [inaudible 01:23:00] household, it’s like 30 million words a year. So a four year old might’ve heard, it might between 10 and a hundred million, depending on the household, words. Most of which it would have been paid no attention to and probably make no difference at all. But that’s way too little, as you say, for a machine learning system. And the child’s also got the problem, of course, of doing all the phonology, it’s working out, “What were those sounds? How do I …” Just understanding the speech signal as an acoustic object and translating that acoustical stimulus into a symbolic form, that itself is a huge learning task. So the child’s learning feat is pretty remarkable.

Jim: Indeed … We’re running a little bit short on time here, I’m going to skip over a whole section on folk physics, folk biology and folk psychology, and the failure of good old fashioned AI, which is more or less implied by the failure of those two, but I would encourage listeners to read the book and go into that. Something that I think is going to be of real interest to folks is, I thought a very clever set of arguments on the nature of emotion. We think of, “Oh, I’m feeling anger,” or whatever that it’s a pure emotion. Well, there’s this distinction between emotions and feelings in the cognitive science literature, emotions are in the body, feelings in the head, roughly speaking. And you give some very clever examples on how that’s probably right. And what I had forgotten all about the Kuleshov Effect. So maybe if you could talk a little bit about how your overall model predicts actually, I think accurately, body plus context equals emotion, something like that.

Nick: Yes, absolutely. So perhaps I’ll start with the Kuleshov Effect. So Lev Kuleshov was a Soviet director in the early part of the Twentieth Century, and he noticed the effect that bears his name, which is if you show, say even still, although it could also be moving, a still image of a person and you change the context they’re in, although the image of them is exactly the same. Then they will … the viewer will interpret the facial expression differently.

Nick: So the famous example is of one of his actors, Ivan Mosjoukine, who has a rather impassive expression in the example that is in the book, and you place him next to a picture of a dead child, it’s actually a mock up of a dead child. It’s a sad looking picture of a child in a coffin. And you look at his face and you look at the coffin, you think he is expressing misery, this desperate sense of sadness, which … he’s controlling it, but you can see it there very powerfully.

Nick: And yet if you place the same face besides a picture of a bowl of soup, which does look particularly appetizing this soup, but mysteriously, he starts to look hungry and he looks like he wants to eat the soup. And then the final one is he’s going to … an attractive woman lying on a chaise longue. And you think, “Ah, I can see his lust in his eyes, even though again he’s hiding it well, it’s quite subtle.” And so in a sense we’re able to project, and do project, a different emotion into a face is I think very interesting.

Nick: Now the thing is that I think that’s a very good indication of what’s going on when we’re interpreting our own body state. So instead of looking at Ivan Mosjoukine’s face, I’m also, of course, in my daily life, I’m trying to interpret my own physiology. So I’ve got fairly simple model at my own physiology. I know roughly how much adrenaline is floating around. I know whether I’m feeling positively attracted to a stimulus or wanting to recoil from it, but there’s not actually a lot more than that. That’s not it, but that’s like … let’s take adrenaline, for example, correlate with my breathing and heart rate and so on. They’re really a very simple underlying physiological system I’m aware of. So that means that if I have a particular physiological reaction, I have to think to myself, “Well, what’s going on here? Why is this reaction coming from?” Rather like looking at Ivan Mosjoukine’s rather impassive face. And to work it out, as with the face, I look at the context. So I find myself in a situation where … Oh, if you let me give the famous example of a bridge crossing experiment, which is a lovely experiment.

Jim: Yeah, I love that one. That was like a hammer hitting me in the head when I read that.

Nick: Yeah, yeah. So this is a wonderful experiment done in the University of British Columbia in about 1974, I think, by a pair of brilliant social psychologists. So they station attractive women who were experimenters, who are holding a clipboard at one end of two bridges. And one bridge is a wobbly bridge, which is a suspension bridge, which is actually quite a scary bridge to walk across, sort of moves around as you walk, I think. And another bridge is just some regular bridge down, as it were, ground level.

Nick: This is an area where lots of undergraduates walks through. They’re picking off male undergraduates and giving their questions. And they say at the end, “Oh, by the way, for ethical reasons, I have to give you a chance to follow up if there’s any questions, so here’s my phone number.” And there really aren’t any questions although there isn’t in this little interchange, but the phone number’s there.

Nick: And of course the bridge-crosser might think, “Oh, well that was a very attractive woman, shall I give her a call? Just to get to know her better?” And so, and indeed, sure enough there are two things that are interesting, that they measured. One is how many of the bridge-crossers asked for the number, and how many don’t, and then how many of those that did ask for the number, how many did actually call.

Nick: Now it turns out that you get something like 50 or so percent more asking for the number and calling for people who’ve walked across the high bridge. Now, what’s going on there you might think? How can that be relevant? But it is relevant of course, from the perspective of trying to understand your own feelings if you’re monitoring a physiology. Because what happens to those people in the high bridge? They walk across the high bridge, and it’s quite scary. So that leads to all kinds of adrenaline flooding around the system. Then they meet the experimenter, and they think, “Well, I just seem to have this incredible adrenaline rush. Wow. I mean, the only reason it must be because this person is just somehow connecting with me in some fundamental way. I’ll get her number and call.”

Nick: Of course, in practice, the adrenaline rush is the bridge, but we’re not seeing it that way because we’re locked into the conversation we’re having now, and that’s our interpretation of our experience is a tremendous feeling of interest and attraction in this person. And in fact, that phenomenon I think is very, very general.

Nick: So another example, it might be a prosaic example, but rather clever, is the experiments by Daniel Oppenheimer, a psychologist now at Carnegie Mellon. He’s done many, many clever experiments in this kind of thing. But one very simple one is that you give people pictures of luxury cars, which are either badly or well photocopied. So, the first thing is, is you ask people which car they prefer. They tend to prefer the well photocopied car. And you’re not asking them about a picture you’re asking you about the car. But if they see a crummy photographed luxury car they think, “Well, no, I just don’t like that car so much.” Now if, on the other hand, so the same thing … So I’m looking at the image and I’m thinking … I’m just generally feeling a slight icky, queasy, I don’t know, just unsettled feeling about that image. What’s wrong with that? What’s wrong with the image? Of course the answer is, well, it’s badly photocopied, that’s what’s wrong with it. But we don’t interpret it that way, we just think, “Oh, I’m looking at a car. I’m feeling visually a bit unhappy, probably it’s because it’s a car I don’t really like.”

Nick: The wonderful thing about Danny’s experiments is he has an extra condition where he … So he tells people, “Oh, I’m so sorry. We had real problems with the photocopier. So some of these images are terrible.” And then the effect doesn’t completely go away, I think, but it largely goes away. As soon as people know, “Oh, right, yeah. It’s a beautiful car, but a crummy photocopy.” Then they’re okay. So similarly, I think, in the bridge crossing experiment had the experimenter reminded people, “Oh, you’ve just walked across a bridge, how much adrenaline do you feel?” Maybe if that had been the question, “What’s been the effect of walking across that bridge on how you feel?” Then they’d have thought, “Oh yeah, right, of course, the adrenaline is coming from the bridge, not from the experimenter.”

Nick: And so that’s a clue that it’s interpretative, the sense of emotion, the sense of attraction to a person, attraction to a car, sense of fear of walking across a bridge, and whatever it may be, these are interpretative states. You’re trying to work out what your body is telling you, in just the same way that looking at an ambiguous face, you’re trying to work out what that face is meaning. And, to do that you’re taking account of the context as well as your body. But you’re not … it’s not the case that the emotion is in any sense welling up from within, fully interpreted. Is doesn’t come from within you, wearing its concept on its face, as it were, rather you’ve got physiological response, contextual situation, and the interpretative brain is churning away and thinking, “I know what this is, it must be this.” But often of course those interpretations are wrong.

Jim: That’s really a powerful insight, I must say. And you give a very interesting example of the potential dangers of thinking the opposite. You tell the story of Bertrand Russell and his internalized view that he no longer loved his wife. Talk about that a little bit. And some of the dangers of assuming that these emotional states are deeper than they actually are.

Nick: I think it’s something we’ve all got to be really careful of. Yeah, so Bertrand Russell was cycling, I think, near Grandchester and he has some thunderbolt sense that he doesn’t love his wife. And although in fact the marriage then continues on for many years, seven or eight I think, it’s at that point sort of all over and it’s just a matter of escape and gradual extrication. And I think that his mental model and obviously Bertrand Russell was a phenomenal genius, and so, not through lack of sophisticated thinking, his mental model of what’s going on in emotions is from time to time your true feelings just burst forth. And when they burst forth you’ve just got to know, “That’s my real feeling, there it is, I should act on it.”

Nick: Which he did, now that may or may not have been the right thing to do. But I think not for that reason, because I think this sense that we have of deep things welling up, deep thoughts welling up from within is just fundamentally, fundamentally extremely fraught and illusory. And a similar thought is that in heated discussions with our loved ones, we can quite often find people saying things that one has that feeling of, “Well, that’s what you really think then. The truth has come out at last.” And I think, again, that’s a very dangerous perspective to have. We should view ourselves much more as creative fiction spinners, who can produce a fiction of one type, and of the reverse type and neither the listener nor the person creating the fiction should be really sure quite what story is the right one. So that sort of sense that there’s a kind of Rubicon that you can accidentally pass or that some sort of true insight, which will come to you and there’s no way back, I think it’s a very dangerous one which we should be very wary about.

Jim: Yeah, I think that’s really good advice. The theme throughout the book is that what’s really in our brain is both sparse and inconsistent. So don’t take it as seriously as we sometimes do, right? Because it’s provisional, the brain is just making this shit up on the fly. The confabulator, I’ve used that term for a long time, since I became aware of Gazzaniga’s work, I wish we could go into Gazzaniga, but we’re short on time. Let’s move on to some applied stuff now. With all this idea that it’s provisional, it’s confabulation, it’s sparse, it’s inconsistent. Nonetheless, we have built up culture, right? And culture takes all, this inconsistency and sparseness and sort of gets co-created with some attempts to put stakes in the ground.

Jim: And I was thinking of some extreme examples, one of which is the Catholic catechism. As a kid, I was raised a Catholic till I rebelled against it at age 11, I’ve told the story on the podcast a few times. But it was a pretty powerful system of indoctrination, right? So you just repeated again and again and again, you’re right down memory traces and you can spit out the answers to the 300 questions in the Baltimore catechism by the time you’re nine years old.

Jim: And then you look at Marxist-Leninism and their worship of the works of Marx and Engels. A good Marxist-Leninist could cough up Marx or Engels’ justification for almost anything. And then of course the similar systems in other religions, whether it’s Hindu, Buddhism, Taoists, and cultures more generally, select certain representations for maximum repetition. And that’s what culture seems to be.

Nick: Yeah, I think that’s absolutely right. No, I think there’s something deeply true about that. And I think to an extent we should think of ourselves as traditions. So each of us is a tradition, basically consisting of all our past thoughts and actions. So there’s this incredible degree to which we … because of our flow of experience and our continual flow of attention hopping from one thing to another, we built up this enormous repertoire of experience and that will condition the way we think about new things. So, each of … so I want to stress that we are incredibly rich sophisticated beings, and we’re enriched through our ability to tap into our own history, our own tradition. But of course we’re not just our own tradition. Actually, many of the things we know come from other people. In fact, we collectively create cultural traditions which are far more elaborate than any individual could create.

Nick: But that does also mean that it’s possible to try to coerce or corral people in a particular direction by putting some sort of cultural stakes in the ground and saying “Whatever you think about anything, this is one thing we all take to be inducible.” And to have a list of basic truths. Even though those truths may be very hard to understand. So you may have people who are convinced of some religious truth. So for example, the Trinity beings three in one, which is very hard to know what that really … what it is exactly one’s believing, but one can think, “Well, I’ve been told this is right. It’s very important. And I believe it.” Or, possibly, “I’ve been taught it’s not right, and it’s very important. So I definitely don’t believe it.” But whatever it is, whichever way round it is, one can have extremely strong sense of this is an important conviction to hold on to and defend, even without a really clear sense of what it is that one is holding onto and defending. And so I think we should always be skeptical of stakes in the ground. All stakes should be moveable I think.

Jim: Yeah. I think that’s one of the things that comes out of that work very strongly. And I hope disseminates out into the world, is that all this stuff is provisional, it’s inconsistent, it’s sketchy. And so, don’t believe that shit, right? Unless you can find objective evidence that reinforces it, and it points to me, even more, the importance of the scientific frame of mind. And I’m sure I’m going to piss off some of my postmodernist friends here, the liberal universalist mind, right? That doesn’t try to go from theory to everything and believes in tolerance and empiricism, essentially as ways to engage with the world, that we should be suspicious of these stakes in the ground. This has been a wonderful conversation. I didn’t get through all my notes by any sense of the imagination.

Nick: Yep, so, well, thanks so much, Jim. I mean, it’s been a huge pleasure to be on the show and lots of great questions and lots also for me to think about more, and I very much hope some of your listeners have found this useful and interesting. And, I hope it helps all of us just think a little bit more about both how we do work, how little we know how we work, and also the fact that we shouldn’t be too sure about any of our beliefs, but certainly not any of our beliefs about our minds.

Jim: Indeed. Well, this has been wonderful. I’ve enjoyed the hell out of it. And I’m glad I had the opportunity to read the book a second time and to come closer, at least, to your point of view.

Production services and audio editing by Jared Janes Consulting, Music by Tom Muller at