887: Andy Clark | How Our Minds Predict and Shape Reality

25:36So we're just going to have to sort of wing it here. But if you listen to a song that you know on the radio and it doesn't come through quite clearly, I can still sing along, I still hear the music quite well, our mind fills in the blanks. What does that mean? Does that mean my brain is actually playing the song to itself as opposed to the radio? I think that's exactly right. I think that's what's going on all the time. Even in the song you don't know, your brain has to in some way be playing it to you on the basis of the sensory evidence that you're getting. But in the case of the song that you do know, your brain's got all this better prior information that it can bring to bear on that process.

26:14And that just cleans up the signal. So you really do hear it that much more clearly. You don't just kind of think that you hear it more clearly because you know it better. This effect really is reaching down to perception. I think, you know, if we're not thinking about radios with bad reception, just think about listening to a song you know in the shower versus one that you don't know in the shower. Exactly the same effect. The sensory signal has become noisy, but prior information can really clean it up. And that's going on all the time, even in the case of you and me talking now. I'm using everything that I know about people and what they say and you and what you're likely to say and how your intonation works, etc.

26:58that I've learned over the course of this little chat to clean up the signal so it gets easier and easier as you go along. You're listening to The Jordan Harbinger Show with our guest, Andy Clark. We'll be right back.

27:37contacts has been doing this for over 30 years. They've got millions of lenses in stock, award-winning customer service, and perks like free returns, free exchanges, even free torn lens replacement. Jen loves their Express exam, which lets you renew your prescription online in under 10 minutes. It's a vision exam, not a full eye exam, but it gives you a doctor-issued prescription if you're seeing clearly with your current lenses, no waiting rooms, no dilation drops, remember those things? And with their best price guarantee, if you find your contacts cheaper somewhere else, they'll beat it. Getting contacts doesn't have to be a hassle.

28:05Let 1-800 contacts get you the lenses you need right now. Order online at 1-800-CONTACTS.com or download the free 1-800-CONTACTS app today. This episode is sponsored in part by Dell. It's time for Cyber Monday, Dell Technologies' biggest sale of the year. Enjoy the lowest prices of the year on select PCs like the Dell 16 Plus featuring Intel Core Ultra processors, and with built-in advanced features, it's the PC that helps you do more faster. Plus, earn Dell rewards and enjoy many other benefits like free shipping, price match guarantee, and expert support. They also have huge deals on accessories that pair perfectly with your Dell PC and make perfect gifts for everyone on your list.

28:40Shop now at Dell.com slash deals.

28:46Don't forget about our six minute networking course. It is how I managed to connect with all these authors, thinkers, and creators every single week and bring them to the show. The course is free over at Jordan Harbinger.com slash course. It's about building relationship building skills. It's not gross. It's not cringy. It's not going to make you feel awkward or look bad. And it takes a few minutes a day. And many of the guests on the show subscribe and contribute to the course. So come on and join us. You'll be in smart company where you belong. You can find the course again for free at jordanharbinger.com slash course.

29:15Now back to Andy Clark. I think you're onto something here that is super fascinating to me because if you're listening to a song that you don't know on the radio and it's got bad reception, you can't sing along the same way or you can't understand, I should say, the same way as you could with a song you do know. But you're right. The shower thing, I just experienced this the other day, I was listening to music in the shower and I was like, wow, you can really hear Spotify being played really well in my shower. And I brought my phone in there, right? And now that we have waterproof phones, my friend called and I had no fricking clue what she was saying.

29:49And I thought, wow, why is the audio so different on the phone than it is on Spotify? And there's something to the audio quality, but I really, now I think I knew what the song was saying and singing and I knew the music, but I had no idea what my friend was calling to tell me. And so I didn't have anywhere near the amount of error correction that I could implement in real time for that conversation. Yep, that's exactly right. And, you know, brains are full of machinery to do this. Loads and loads of downward connectivity leading from the sort of deeper areas of the brain out towards the sensory peripheries.

30:22You know, that connectivity is at least equal to the inward facing connectivity. And so brains are very, very active participants all the time in constructing in sensory experience. So what part of this is memory, e.g. I remember the Duran Duran track from listening to it a million times growing up, and what part of this is a totally different process that my brain is engaging in? Yeah, that's a very interesting question, because in a way it's all memory. What happens with these pictures is that perception becomes very much a sort of memory in that sense of prior trace driven process. There's an early book in this sort of area by um jeff hawkins i think it was and he talks about the memory prediction framework with a hyphen memory prediction framework it's a kind of picture of what cortex is doing in in his computational work um but yeah i think it's what happens here is that perception turns out to involve a great deal of memory and then the interesting question there becomes so you know what's going on in memory memory you know episodic sort of memory when i seem to relive an experience.

31:34But again, these things do begin to fall into place if you think that the machinery that you're using to actually perceive the world involves the ability to create those signals, if you like, using what you already know about the world from the top down, as people say in this literature. And of course, if you've got that, if you're using this top down flow all the time, just for ordinary perception, you can imagine that if you cut off the incoming perception, you can just drive with the top-down flow and then you get imagination. And so imagination and perception and memory are all really, really closely linked together.

32:12Indeed, it's really the kind of all different variants of the use of the same resources, which are predictions, prediction errors, and their weightings. Some of it sounds like, not all of this, of course, can fit neatly into one box, but some of this sounds like compression. And I mean data compression with computers. Are you familiar with this at all? Yeah, I mean, I'm familiar with it from sort of kind of motion compressed video and JPEGs and things like that. Those are beautiful cases of data compression by prediction, like in compressed motion video stuff. You're basically assuming that the next frame will be the same as the one before it, unless you're told otherwise.

32:53And so then, instead of having to transmit all of that information about the next frame, you just transmit the residual differences, the errors, if you like, that would result if you just assumed that nothing had changed. So now you get sort of errors saying something like, actually, stuff has changed over here with respect to the foot. Foot's a little bit further along here than it was otherwise. That turns out to be a very frugal way of encoding information, because basically, if you're predicting it already, you don't need to encode it again. So all you need to do is deal with the differences.

33:29And that seems to be going on in many, many sort of technological formats. And data compression more generally, which might be what you're talking about there, is a really, really powerful way of bringing out regularities and interesting sort of patterns in data. So yeah, I think brains are data compression machines, just as much as they're prediction machines. I'm by no means any sort of expert on compression or data compression, but I'm fairly sure that compression works by essentially using some kind of shorthand abbreviations, maybe even to store information and then expanding that again later.

34:06The video example you gave is probably better. I was thinking more like a zip file where you're opening something and it's suddenly larger because your computer can say, well, I know that this is usually this and we'll shorthand these 100 ,000 characters into this one character for the sake of opening this file. and you see compression errors in video especially older video where two people are talking and it looks okay and then a bird flies behind them and there's this weird blur that's sort of chunky that's a prediction error by the computer right because it didn't expect that kind of and then it's oh crap i gotta catch up and draw the bird and it's doing it too late yeah and it doesn't look right yeah that's a beautiful example and yeah that's exactly what it's like to be a brain encounter in the world is you know basically predict predict predict oh i missed something What's going on there?

34:54And then you try and sort of fill in that bit. So in some respects, and I touched on this on the top of the show, the world we experience is the world that we predict. But I really want to add limitations here because so many self-help grifters and guru wannabes or whatever you want to call them, they start selling the idea that you can force yourself to predict the world you want. And suddenly it'll become true a la, I think it's called manifestation bullshit. it. It's just pushed by tricky Instagrammy type people, or in some cases, folks that make it all the way to Oprah. But, you know, it's just nonsense.

35:30It's like you can just put these vibrations out there and your brain predicts this and suddenly you've got a Lamborghini in the driveway. Yes. Yeah. I mean, there are there are overturns of the sort of self-help thing in the predictive process in account of what brains do, because if you can alter the things you predict, you can nudge your experience around in that sort of way. Just like if I'm about to deliver a talk, for example, I get this tingly feeling in my fingers just before. And if I reframe that as chemical readiness to give a good performance, as opposed to anxiety that's going to make me crash and burn, that actually seems to be a helpful thing to do.

36:11So in that sense, changing what I predict can change my experience for the better. But of course, just predicting that I'm going to be able to return that tennis serve in just the way that I want to, that's not enough to enable me to deliver a good tennis serve. On the other hand, prediction is involved in delivering a good tennis serve. But most of the predictions that you need to make are the ones that your brain only learns to make by long and tortuous experience of trying to return serves. So it's that sort of hidden empire of prediction that needs to be right for things to happen. And unfortunately, you can't change it just by changing the top level and saying, you know, I predict I'm just really, really going to do a wonderful talk on quantum physics today.

36:55I don't, you know, if I don't know anything about quantum physics, that's not going to help. Right. No, it's funny you mentioned quantum physics because those same people that talk about the manifestation and the prediction, they're like, it's all quantum, you know, because your brain is changing reality through quantum something, something. Anything quantum, since people don't really understand it, is fair game for a lot of these grifters to make some shit up and then sell it, right? Because it's very complex and hard to wrap our minds around. Do our brains try to get better at predicting? I assume that throughout our whole life, our brain is just honing this skill.

37:29Yes, I would say so. I mean, you know, that's what brains love to do is kind of eliminate prediction error. And what that means in effect is that they love situations in which they can eliminate more prediction error than they estimated they were going to eliminate. And so that's sort of a, you know, if you find yourself in a problem space where you're doing better than you expected, you're going to hang around there. And those are the places you're going to learn the most. And so brains that like those sorts of error dynamics, as they're called, they're going to learn a lot of stuff and get better and better as they go along.

38:05And in fact, I think a lot of affect and emotion is related to these error dynamics. You know, we like places where we're getting rid of more than estimated amounts of prediction error. Like, that's probably why we do crosswords and Sudoku and things like that and why we enjoy mastering sport and all of those things. Yeah, I think there is something to that. It kind of goes back to what we talked about earlier with you're taking a lot of the enjoyment out of something. If you can just issue a command that says return a killer serve and you push J or whatever on your physical or not physical keyboard, it's not quite the same thing.

38:42Although there's a sense in which at the top level of your brain, that might be kind of what you're doing. In some way, you're sort of saying, OK, yeah, yeah, yeah, do it right. But the thing is that the well-trained brain unpacks that into all of these little predictions about what lower levels of processing and then all the way down to particular sort of muscles and spindles and so on need to be doing to get the job done. So, you know, it's that cascade of prediction that constitutes expertise. And that's how we get things done. So, yeah, a lot of the enjoyment is in having good cascades of prediction.

39:17I'd love to talk about some common illusions. It's hard to explain these, like the gold and white or black and blue dress. I'll link to this in the show notes. This was all the rage, what, three, four, five years ago, something like that. And nobody could figure out what it was. it took science, I think, a minute to catch up, or at least scientists to address this publicly. Can you explain this a little bit about what this was and, of course, what's happening here? Yeah, so this will be the, you know, that dress, as people often refer to it, the one that to many people looked blue and blue and white, and to me and a lot of other people looked gold and white.

39:54Now, because it's such a striking difference that people are so sure that they were seeing it the right way, it caused a little bit of an internet sensation for a while. The vision scientists jumped in right away and said, okay, yeah, we've known about this for a long time. It's because the different people's brains are making different assumptions about the direction that the light is coming from. So if you think the light is sort of coming particularly from, I don't know, above or the side or something, then it's going to look gold. If you think it's a more ambient kind of all-around light, then it's going to look blue.

40:32Something like that. And that story is good as far as it goes. But what was interesting to me was a subsequent study that was done by Walish, Pascal Walish, and some people at New York University. And what they showed was that the differences between the white dress and the blue dress seers actually corresponded, not perfectly, but in a very significant kind of way, to people who were larks and people who were owls. And so people that tend to get up early in the morning, they tend to see a lot of things under natural light with the light coming from a certain way. People that tend to be up very late at night, they tend to see a lot of things under artificial light with the light coming in that sort of way.

41:17And it turns out that they tend to see things as the dress as blue and the others tend to see it as gold. The blues in this case had it right because the dress was being photographed in artificial light. Wow, the blues were right. That is shocking. It's crazy, isn't it? It is crazy because I looked at this today and we'll link to this in the show notes. We'll have the image there. I almost wish we didn't tell them which was which, but it doesn't matter because it will look gold and white or it will look black and blue. I saw it as gold and white. But then I thought, okay, other people see it as black and blue.

41:51Let me now try to see this as black and blue. And I couldn't get there. I couldn't make my brain see it as black and blue, no matter how hard I tried. However, in this next illusion where it's auditory, I can easily switch between the two options. So this is an auditory illusion. It's green needle. You either hear green needle or you hear brainstorm. I can easily say, all right, I'm going to hear green needle. And I hear green needle. And then I go, all right, I'm going to hear brainstorm. And then I hear brainstorm. And I've got the soundbite here for people to play along at home. So here we go.

42:24Don't prime yourself to hear anything. Just whatever you remember. What do you hear?

42:39So I heard brainstorm, probably because it's the last thing I said. But now tell yourself you're going to hear the other thing. So if you heard brainstorm, tell yourself now you're going to hear green needle.

42:59It's incredible. And it's so obviously the same sound. And yet I do hear green needle or brainstorm, depending on what I tell my brain that I'm going to hear. It's amazing. I know it's fake. I know it's an illusion. I know it's a low quality sound thing from a little toy. And yet here we are. But I can't do that with the dress at all. I can't make the switch. What's going on here? That falls into place because in the green needle brainstorm case, what you're doing is you're installing a prediction by thinking green needle or thinking brainstorm or looking at the word. Sometimes people do this with the words written on a placard and pointing at them.

43:36And that's enough to get the prediction machinery into the right format to pull out the bits of the signal that correspond to whatever you've primed in effect you can't do that with the dress because you don't know how to as it were get the cascade of predictions into place that would be there if you were seeing the dress under artificial light you know you can't as it were imagine what the difference would be because your brain is busy doing all these compensatory things all the time And that's why actually you can spot that the dress is the same color when you move it from artificial light into natural light.

44:15So you're not going to think that the dress suddenly changed color if you do that, because brains are so good at doing all that compensatory work. So I think that the cases are interestingly different. One of them is one where you expect a top-level prime, like seeing the written thing green needle or thinking green needle, to be able to control the cascade of prediction. The other one, we've got no real way to control that cascade of prediction. So it's just about top-level controllability. I agree that it's a really striking and lovely demonstration of the predictive brain in action. To me, one of the things striking when I listen to those sounds is that the second time, it always sounds so much clearer.

44:59Yeah. Whatever way I've heard it the first time, the second time that comes, it sounds super clear. And then maybe it's going to flip again, you know, when it goes to the next time or something like that. But again, that's just, you know, having heard it one way once, your brain is suddenly busy predicting that in a very good and strong way. And so it's cleaning up the signal even more. So you've sort of got a double prediction effect there. One makes you hear it that way. And then because you've just heard it that way, it sounds even clearer next time you hear it that way. I looked for a higher quality version of this soundbite and I couldn't find it.

45:32But then I started to realize, I think the reason our brain can predict both of them is because it's low quality. It's the radio thing all over again. Yeah, that's right. It has to be staticky and kind of ambiguous. Otherwise we would clearly hear, everyone would clearly hear the same thing. It's a little bit like those old pictures of the spotty Dalmatian dog or something like that where at first you can't really see what it is and someone says it's a dalmatian dog hidden in there and once you you've sort of seen heard it seen it i should say that way then it can never look the way it did the first time you saw it again so yeah noisy signals dramatize what the brain is doing anyway all the time which is using predictions to clean things up and structure them reveal the stuff that you care about but yeah take away the noise and you'll lose the demonstration i'll play that one more time so people don't have to rewind again because i know people are like i gotta do this now again because i i played this like 50 times to see if i could trick myself or whatever it's very interesting it's not trick it's not saying brainstorm and then green needle or green needle and then brainstorm i've i've tried this um zillion times to see if oh is it different the second time and that's the trick.

46:47Nope. It really is the same thing. Really incredible. This goes a little bit deeper as well, though, into what we actually feel. Tell me about the nail in foot example. Talk about dramatic examples. Yeah. This is a construction worker. I think it was in New York. It was written up in the British Medical Journal. And it was they fell off some scaffolding and they saw a long nail go through their foot. And, you know, they were in intense agony. They were taken to hospital. They were given fentanyl and some other drug as well. And then they started to remove the nail and they discovered that the nail had passed very safely between the toes.

47:29So what happened is, you know, the workers were in the big construction boot. You got a perfectly good visual impression of a nail through what looks to be your foot, but it wasn't actually through the foot. Nonetheless, the agony that the construction worker was feeling, I'm sure was 100 % genuine. It's just that when the brain strongly predicts a certain kind of signal, just like with that phantom phone vibration that we started off talking about, it can construct that from whole cloth, if you like. So it's very interesting. Placebo effects have some of the same sort of profile, medically unexplained symptoms, chronic pain.

48:07There are many cases where I think we're going to start thinking about them much more in terms of the predictions that the brain's making. And pain is a very, very dramatic sort of case. What a strange day that construction worker had. Yes, yes, I think. And they've become rather famous as well. I don't know if they're sort of ever going to say, you know, that's me. I'm the construction worker in that thing. Yeah. Yeah. I wonder how many people tried it afterwards to measure the effect. You know somebody shot a nail through their foot or tried to, or tried to hit the toes and see if it would work.

48:40But is this why placebo works? Our brain predicts something's going to make us not feel as much pain, or it's going to wake us up, or it's going to lower our anxiety. And so the prediction is self-fulfilling in this case? Yes, I think that's right. And, you know, this works very well for some things. It doesn't work at all for others. As you were saying earlier, you know, you've got to be aware of the limitations of these things too. So, you know, placebos are very good at analgesia, for example. So, you know, making us not experience pain under conditions where we otherwise would. They're very good at fatigue.

49:12They're good at chronic back pain. But of course, no placebo is going to kill a virus. It's no good at using a placebo hand gel if you're worried about COVID or something. And likewise, no placebo is really going to get to grips with cancer, as opposed to cancer-related fatigue. So I think we need to be aware of the sort of limitations. But at the same time, the potential in sort of systematically exploiting these effects, I think, is really very large. It's even possible, apparently, although I only know one example of this, so treat it with some caution. But there's an anti-Parkinson's drug called apomorphine.

49:52and people that are trained on the real drug and get the anti-Parkinson's effect from the real drug can then be given a placebo, told it's apomorphine, and they get the same relief from the Parkinson's symptoms temporarily. It doesn't last forever, that effect either, unfortunately. But that's interesting because that's the case, as I say, it's the only one I know of, but it's one where you seem to be able to train the brain to do something quite systematic in response to a particular drug that can then be given as a placebo for the same effect. So again, I think one day when we understand a lot more about the precise pathways that are involved and why it is that some people get these effects more than others, there are conjectures already about why that might be so, then we might be able to do some sort of precision medicine that makes maximal use of placebo interventions for the cases and the people that are going to benefit most from.

50:56This is the Jordan Harbinger Show with our guest, Andy Clark. We'll be right back. If you like this episode of the show, I invite you to do what other smart and considerate listeners do, which is support our sponsors. We really do appreciate that. All the deals, discount codes, and ways to do so are at jordanharbinger.com slash deals. You can also search for any sponsor using the now up to the minutes-ish AI chatbot on the website at jordanharbinger.com slash AI. Thank you for supporting those who support the show. Now for the rest of my conversation with Andy Clark. It makes sense that this placebo and nocebo where it actually does something.

51:32Nocebo was kind of what we had with the nail through the foot that didn't really go through the foot, right? Exactly. Like, oh my God, I feel this pain. It's like a negative placebo. We did a whole episode on placebo with Joe Marchant or Marchant episode 716. That was important because I agree with you. It's important to know what placebo can and can't do because there's just this whole industry of people out there selling things that they say, oh, well, you know, this does this. And I go, no, it doesn't. That's nonsense. Well, even if it's placebo, that's still something. No, it's not. You're not getting placebo effect.

52:03You're conning people and they're not getting real medical care because they think that your stupid powder is curing their pancreatitis or something. You're a charlatan, you should go to prison. Yeah, yeah. And actually, you mentioned there Joe Marchant, and I think I read a paper of theirs that was about the honest placebo effect. I think it was them. And that's very interesting, too, that you can be given a placebo, told it's a placebo, have it clearly marked placebo, and yet still get the placebo effect from it. And I think that's really an important thing in thinking about the predictive brain that as it were what's going on there well it's because it's not your top level conscious prediction that matters your top level prediction might be oh it's a placebo it's not going to do anything if you're still getting those effects it's because there's all those unconscious predictions of relief that are coming from it being you know presented by someone in a white coat in a good bit of packaging um with you know some authority saying, yeah, you know, this is a placebo, but many people experience relief from taking it.

53:10So I think that's the idea that the hierarchies of prediction are mostly non-conscious is also important. And again, that shows us where the limits are of what we can do explicitly. For sure. Yeah, I think it's fascinating. Some of the stuff we covered on her episode was like, yes, if it's delivered by somebody in a white coat, or if it's injected, even if it's saline, and they say this is a painkiller, it's far more effective than if it's a pill. I even asked some, you know, typical stupid Jordan questions like, what if it's a tablet filled with powder versus a solid pill? And there was slight differences in there depending on what other medications the person actually took in their life, I think it was.

53:49And I'm drawing off memory here, but it was kind of like the way you took the medication. And was it your book where somebody was taking pink pills and they were getting relief and then the manufacturer said, ah, we're gonna make them blue now and they just stopped working? Yeah, that was my mother with one of her cancer treatments. And it was sort of, yeah, it was so sad because it's exactly the same active ingredient. And yet I definitely wasn't giving her the same relief. And that put me in a very interesting position because on the one hand, I believe very strongly in the power of placebo effects, particularly with regard to cancer related fatigue, for example.

54:28on the other hand I you know I wanted to say to her come on the active ingredients just the same so in terms of you know that pathway you should be getting just the same effects from it I to this day don't know quite how I should have dealt with with that situation if someone could end up having as much faith in the blue pill as they had in the pink pill it would be just as good yeah and that's maybe what I was trying to instill there but yeah it's um it's quite a can of worms the the whole placebo thing, you know, how much, obviously you take some efficacy away if you tell people they're placebos, but at the same time, it seems dishonest not to.

55:05I guess as long as the doctors and team that are treating them know it's placebo, it matters a little less if the patient knows, because you wouldn't want a doctor to go, oh, they're getting treated for this, they're already on a drug, if it's a placebo. But yeah, if it limits the effect, it's like, well, as a painkiller for somebody with a terminal illness, do you really need to, what would they really want to know in this particular situation? And I don't want to be that guy. But I thought when I thought when I read this in your book, I thought, man, why didn't he go to a compounding pharmacy and say, I need a bunch of bright pink fake sugar pills for my mother to take with her real medicine every morning?

55:38Yes, that's an interesting thought. I didn't have that thought. I actually didn't know that compounding pharmacy existed. Existed, yeah. But yeah, that would have been quite a good idea. Yeah. We, of course, had to make sure that she did keep taking the pill that had the effective ingredient in as well. So there'd be one extra pill now, but I'm sure that we could have found a way of pulling a veil over that. That's a good risk that I had not thought about. Like, what if she just goes, oh, I don't need the blue ones anymore. I just take the pink ones. And you're like, oh, mom, here's the problem.

56:09I got those are Tic Tacs that I've filed down. Oh, man. Oh, anyway, this must also be why our dentist will say ridiculous things like, you're gonna feel a little pinch and then she drills into my jaw with a drill or jams a syringe into my gums i'm like that was not a little tickle or a little pinch that was you just jammed a metal object into a hole in my face was way worse than what you said but in the of course in the moment and when i was a kid that used to work really well for me i'd go oh that was a pinch yeah good thing it was only a pinch but if i didn't get that warning it was just like getting punched in the face.

56:45Yeah. And imagine if you got the worst one in the one that says, you know, this is really, really going to hurt. Right. Yeah. So yeah, it's almost like whatever information you're given there is going to, it's going to nudge your experience around a little bit. Does it matter how much we trust the person just like placebo? Is there a white coat version of this is only going to hurt a little bit? And it's like, I don't believe you because you're, I don't know, you're wearing an eye patch and look like a villain. Yeah, I mean, this is this precision weighting bit of the formal apparatus here, at least.

57:19So the idea is that if a prediction that your brain is making is going to have much effect, it needs to be highly weighted. And one thing that authoritative sort of speech and good packaging and white coats do is they enable us to, or they force us to highly weight the predictions of relief or whatever else you're expecting from the pill or procedure. So yeah, a lot of things fall into place, I think, if we start thinking that things that look like ritual and trappings and so on and ceremony are actually ways of enhancing the precision weighting of our own expectations, changing our experience in ways that then follow suit.

58:01What about things like PTSD and prediction? It seems like some things such as, and I don't know if this is exactly PTSD, some things like flashbacks from war or sudden fight or flight responses, those might be caused by prediction issues. Again, I'm maybe grasping a little bit here, but when I was a kid, my friend's dad drove us to lacrosse practice and there was a helicopter for, I don't know, traffic or something. And he was at a red light and people were honking and he froze and he's like, do you guys hear that? And I was like, yeah, the helicopter? And he's was like, dad, you got to go. And he's like, you hear that?

58:38And we asked his mom later because we were like, what the hell happened with your dad in the car? He wouldn't go. He like had a crazy look in his eye for a second, not like a wild, scary one, but it was just like he was somewhere else. And she said, oh, did he hear a helicopter? And we said, yeah. And he goes, well, she told us he was in Vietnam and something, I don't know, it was the kid version of the story. But when he hears helicopters, he sometimes will have like a flashback and it's almost like something out of a movie. I asked my friend if he remembered that. And he's like, yeah, I do remember that because it was pretty rare, but it happened a few times in his life once while I was in the car.

59:13Is that a prediction error? I think it certainly involves a prediction machinery. So one of my Edinburgh University colleagues, Peggy Sears, has done some work on this, trying to distinguish between people that suffer from PTSD and people that don't in the sense of you can be put into the very same war zone situations. And some people will get PTSD and some people won't. So there is a difference there that needs to be explained. And what she found using various cohorts of people, war veterans, some of whom had PTSD and some didn't, was that the ones that had got PTSD had a different kind of response to prediction errors that preceded very aversive events.

59:58They had a very, very strong response to those. So they kind of, they changed their prediction routines very dramatically if they made mistakes like that, even in little sort of video game scenarios and the kinds of situation that she could put them in. And that might be the underlying difference, is that, you know, if you're going to update your responses and your model very, very rapidly on the basis of aversive events that follow failed predictions, like I didn't predict that helicopter, or I didn't predict that bomb going off or that sudden sound or whatever it was, then you're going to be more likely to get into these sort of locked cycles that characterize conditions like PTSD.

1:00:38So I think it is something going on in the prediction machinery that is creating and then locking in this sort of aversive and very dramatic response. Okay, so if predictions are survival and somehow a goal of our brains, Why didn't we just evolve to sit in the corner of a dark room and not move and only eat the same thing every day? You know, why do I like surprises and novelty and roller coasters? That doesn't, that's like the opposite of a nice, routine, smooth, easily predicted reality. And my brain loves that stuff. And I'm not alone. You're right. And this is a so-called darkened room problem for the predictive process in accounts of brain function.

1:01:19And, you know, why is it if we love nothing better than quashing prediction error, we don't just sit in a dark corner until we die predicting darkness and lack of food? Well, there's a quick answer to that. And then there's a longer answer. The quick answer is that the prediction machinery is focused inwards as much as it's focused outwards. And brains like ours, because they're the brains of evolved creatures, make certain kinds of prediction in a kind of chronic way. They predict bodily states that will keep us alive and functional. They predict having enough food. They predict having enough water.

1:01:56And indeed, they go looking. They send us looking, if you like, for food and water long before we actually need them, because it's a prediction of future need that constitutes the experience of thirst or hunger. So an example from Lisa Feldman Barrett there is that if you feel thirsty, you might take a drink of water. You'll feel immediately that your thirst is quenched. But actually, the water won't have any useful effects on your body for about 20 minutes. So the feeling that your thirst is quenched is the result of a prediction, just like the thirst itself was. So, you know, your body's got time to wait because it was predicting into the future anyway, if you see what I mean.

1:02:37I do, right. Right. So if I'm drinking water, it's not like, oh, good, this water that's barely in my esophagus, you know, barely in my stomach has now entered the cells that need it. No, my brain's just saying, hey, we predict that in a few minutes you're going to have what you need so you can stop telling the system that we need more of this. And that's purely a hallucination that we are now rehydrated. Yes, that's right. Just like hunger in a way is a sort of hallucination that you need food now, but it's a genuine hidden prediction that you're going to need food pretty soon. Right. Or that I'm just bored, unfortunately.

1:03:11Right. That's, yeah, unfortunately, my prediction machine is now, I think a lot of us have conditioned our prediction machine to think that we should eat when we're thirsty. We should eat when we're bored. We should eat when we're, I don't know, nervous, whatever it is. We cross those wires far too often. But I don't mean to escape the darkened room quite that easily. Ah, good. So, you know, step one of escaping the darkened room is we have these chronic predictions that we're going to get, you know, fed and watered and so on. And that's enough to keep you out of the dark and deadly room. But it's not enough to keep you out of the dark and boring room where nonetheless, you know, as you say, you're getting the same food, you're getting the same water.

1:03:48It's all going on. It's just not very interesting. But the reason that we hate rooms like that is the one that we touched on earlier, which is that predictive brains like nothing better than getting rid of more than expected amounts of prediction error. So they like getting rid of prediction error anyway. So they're already naturally curious. So the darkened room is not going to be very interesting there because there's not actually that much prediction error to get rid of. But also there are no good slopes of prediction error minimization there either. So you might start to invent games for yourself in the darkened room if you were locked in one.

1:04:25And that would just be you trying to create artificially some kind of arena in which you can experience a minimization of prediction error. In a way, that's what games are. So we leave the darkened room for so many reasons. There's so many questions that open up when we realize we're just hallucinating and filtering everything we take in. What about people who, for example, think that they're overweight when they're actually too thin? I don't know if anorexia is the exact thing, but you see people who think, I'm overweight. What is it called? Body dysmorphia or something like that? Is that what that is?

1:05:01Yes. And again, I think prediction machinery is certainly playing a role there. There are some interventions that suggest it's playing a role. So, for example, getting people with anorexia to inhabit a kind of virtual body with a normal body weight and maneuver that virtual body around sort of virtual arenas, that experience can impact the way that they feel and view their own body afterwards. So they tend to have unrealistic perceptions of their own body that can be partially remedied by sort of giving you a kind of a very, very strong nudge in the direction of making a different set of predictions.

1:05:43There'd be a lot of work to do to really unpick something as complex as anorexia, for example, using the prediction machinery account. But I think it will prove to be part of that because I think it's very intuitive somehow that we are somehow making all kinds of predictions about our own bodily state and about how our bodies ought to be and that those predictions in part constitute our reality. And so, you know, often we can have a sort of phantom fatness almost. I suppose, you know, if you think about phantom phone vibrations, maybe anorexia involves a kind of phantom fatness that is no harder to understand in some deep sense than the phantom phone vibrations.

1:06:29How do we make our predictions more accurate? Or is that just something the brain is always trying to do and there's not much we can do to influence that process? The answer to that has to just be practice, really. Yeah. This is all about expertise. Predictions underlie expertise and practice is what leads to expertise. So that's the thing that we can do if we want to get better at something practice. So there's no, unfortunately, there's no, there's no better. Wouldn't it be wonderful if there was some nice little gold or whatever it is? Yeah, yeah. That's funny. It's like the most basic advice ever.

1:07:06But hey, at least Now we have a scientific background for it, huh? Maybe some visualization would help because then at least we're getting through that. That's probably a whole... And careful framing, you know, all that stuff we talked about earlier, a little bit of well-placed self-affirmation, nice bit of framing, but lots and lots and lots of practice. I know we're running out of time here, but have you thought or researched at all what psychedelics can do to the predictive brain or these mechanisms? Is there any data on this? There is, actually. It's very interesting. A lot of the good work there has been done by the psychedelics lab in London, led by Robin Carhart-Harris.

1:07:44And what they seem to have been able to show is that the classic psychedelics, sort of LSD and magic mushrooms and so on, the classic psychedelics act on the brain in a way that seems to release us from the grip of top-level self-predictions. so at low doses they seem to just affect low level predictions hence visual effects and so on but at the higher more interesting doses they can undo some of the sort of predictions that you're making about yourself and if you're a depressed person for example just briefly experiencing what it's like to not predict yourself and the world in the ways that you normally do can be enormously liberating so it's quite a detailed account there's actually i you know obviously i'm glossing it very very rapidly there but it it makes contact with a particular structure of the kind of chemical molecules that characterize the classic psychedelics shows that their locus of action in the brain is going to be the sort of areas most involved in high level self-prediction and then show how Well, the psychedelics are kind of, I think, a phrase that Carhart Harris uses is shaking the snow globe.

1:09:00So they're kind of getting rid of some of those entrenched predictions, shaking up the snow globe so that you can start to experience new ways of experiencing yourself and the world. And, you know, that can be very powerful. Well, Andy, thank you so much. I predict this conversation will go over well with my listeners. And I thank you very much for joining us today. Fascinating stuff. I mean, it borders philosophical and can quickly be over my head. But you did a great job of dumbing it down for me. And I appreciate that. Thank you so much. I had a great time. We've got a trailer of our interview with Malcolm Gladwell, which is pretty timely right now.

1:09:38We'll discuss why the information we gather from face-to-face human interaction isn't as uniquely valuable as we think it is. And why television can actually make us worse at reading other people. Young African-American woman is in Texas, just has a job interview in a rural Texas town, Sandra Bland, and she's pulled over by a white police officer. Hello, ma 'am. They have a conversation. Do you mind putting out your cigarette, please? It quickly escalates. I will remove you. I'm giving you a lawful order. Okay, you gonna get me out of my car? Get out. Drags her out of the car. I will light you up.

1:10:07Get out now! Knock my head in the ground. I got epilepsy, you mother... She's put in prison, and three days later, she commits suicide in her cell. If she's in an Audi, her chances of being pulled over are lower. and if she's in and out with Texas plates, she's fine. Most of all, if she's white, there's no way he's pulling her over. And as I describe in the book, all of those inferences are deeply problematic. We have enormous confidence in our ability to draw meaningful conclusions about people based on very superficial evidence. Even though the plots of Friends are absurdly complex, no one in history has ever watched an episode of Friends and said they lost me.

1:10:44Yeah, what is going on in the show? Yeah, never happened. They do that because they're trained actors. If you watch a lot of TV, you can come to the false impression that that's what's going on in your face. But in truth, that's not true at all. And a significant number of people are what are called mismatched. And that is that their facial expressions under certain circumstances do not match the way they feel on the inside. The Amanda Knox case. An American teenager goes to Zia Urbara in Italy and gets falsely accused of murdering her roommate. and that case is all about the fact that Mendenox is mismatched.

1:11:18They have another guy who clearly did it and they drag her in. Why? Because she doesn't behave the way the Italian police and the British tabloid press think someone whose roommate has been murdered ought to behave. We are sending people to jail for years and years and years for crimes they had nothing to do with. Kids, I mean, she was like a college student, right? College student, yeah. For more from Malcolm Gladwell, including how the misunderstandings between people and cultures invite conflict, I told you this was timely, check out episode 256 of the Jordan Harbinger show.