Rachel Feltman: For Scientific American’s Science Quickly, I’m Rachel Feltman.
There’s a lot of uncertainty in the air right now, whether you’re trying to predict the outcome of an upcoming election or just trying to guess how much it will cost to gas up your car come next week. You don’t need a degree in neuroscience to know that all those unknowns are tough for the human brain to handle. But some experts think there’s actually a lot of nuance to explore about the cognitive burden of uncertainty—and that you might actually have the power to change how well you tackle it.
Our guest today is Maia Szalavitz, a contributing opinion writer for the New York Times and the author or co-author of eight books. She’s also the author of a feature in the latest issue of Scientific American that explores the emerging science of the human brain as a prediction engine. She’s here to tell us more about what that means in a world that feels increasingly unpredictable—and what we can do to make our minds more adaptable to chaos.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Thank you so much for coming on to chat with us.
Maia Szalavitz: Thanks for having me.
Feltman: So how did you first get interested in this topic?.
Szalavitz: Sure. So our world seems to be becoming increasingly uncertain, and it’s not the kind of uncertainty that we’re used to dealing with in modern America because we’re used to the law being predictable. We’re used to the government’s actions being reasonably predictable. We’re used to all kinds of norms that are being violated. We’re used to being able to think of “What do I wanna be when I grow up?” and that be an actual thing that you can choose rather than perhaps losing all jobs to AI.
So I was in the midst of my own uncertainty, and I’ve been thinking a lot about this idea of how the brain works as a prediction engine, and I was really interested in hearing from people with that perspective about, like, how to manage in this really unusual situation—I mean, I didn’t even get to mention climate change. You’re used to, like, summer being summer and winter being winter—so it’s really unsettling and really, you know, because our brains do work, as far as we can tell, as prediction engines, and we’re constantly having to update prediction errors, when there’s uncertainty, it gives your brain a lot more work.
Feltman: What exactly do we mean when we say we’re modeling the brain as a prediction engine?
Szalavitz: So essentially, all of our experience of reality is a hallucination under this perspective, which is really weird to think about just in itself. But essentially, in order to solve the problem of avoiding entropy and avoiding death and ideally successfully reproducing, biological creatures need a way of managing risk and figuring out what to do next.
And so our brains sort of optimize on, okay, we get this general sense—and it’s based on experience, and there’s obviously some genetics—but you start out with this general, you know, way of perceiving the world, and then you update it when there’s an error. And so, you know, my example that I use in the piece is that, like, if you are in your room, and you have a cat, and you look, and you see something out of your corner of your eye, and it’s curled up, and it’s gray, you have a pretty full picture of that cat in your head when you, you know, do that. And then you look, and, oh, it’s actually the towel that you forgot to put in the laundry, right? But there’s no, like, glitch in it going from cat to towel, right? It’s, like, completely seamless, and this is the way our entire world works. And there’s all kinds of really interesting implications of it for things like pain and how you act psychologically.
But that is kind of the fundamental idea, that we are always making predictions, revising predictions, and, you know, with major prediction errors, it becomes seriously problematic because we may not be able to, you know, eat and survive and do the things we need to do.
Feltman: So what are some of the implications for, you know, what this means in terms of, like, how we think things through and how we process the world? I guess what I’m getting at is, like, “Why is it helpful to use this prediction engine idea?”
Szalavitz: If you think about it, if your entire life is based on predictions, there are certain things that you are going to be good at and certain things you are going to be bad at. Staying in the moment is going to be a very difficult thing if your entire world is predictions, right? And if you’re constantly predicting things, then what you’re predicting has a very big influence on what you will perceive.
And so if you are constantly seeing the world in a really negative way, that will skew it even further. Or if you are constantly just trying to avoid things, it’s really, if you think about it, it has implications for absolutely every aspect, which is why I’m sort of struggling of, like, narrowing it down here.
Feltman: Yeah, absolutely. What about the implications for how cognitive stress impacts our bodies?
Szalavitz: Sure. So Bruce McEwen was the great researcher on the impact of stress and health, right? And he talks about, like, how stress gets under the skin. And at this very moment, I am having a personal experience of this because I had to take a course of prednisone to deal with a hearing issue, and that’s basically massive amounts of stress hormones, like, coursing through me and making me feel like bad coke or something.
But those stress hormones are really good in an emergency because what they do is keep you awake and alert and focused on dealing with what you need to deal with in front of you to escape the emergency. And that has a physiological cost because you spend more energy doing it, and therefore you produce more waste, and therefore there’s all kinds of different things that your body needs to manage that can get out of whack.
And when you look at the data, severe chronic stress, starting in childhood with things like abuse or neglect or other forms of trauma, raise risk for everything from, you know, obesity, heart disease, stroke, all of the mental illnesses, virtually. And it’s because of the way these hormones impact your brain, because if you have too much of these glucocorticoids, it can eat your hippocampus a little bit, and so it makes memory weird.
And a lot of these things are all about trade-offs, because in the short term, if you don’t survive, then the rest doesn’t matter, right? But in the long term, you are generating a debt in some sense.
Feltman: Well, speaking of mental illness, in your piece, you outlined how a few mental illnesses could be reframed as sort of disruptions in this predictive machinery. Could you tell us a little bit more about that?
Szalavitz: Sure. So the most obvious one is anxiety, because anxiety is worrying about the future.
Feltman: Sure, yeah.
Szalavitz: And so uncertainty is not going to be good for anxiety. And one of the things that was really interesting about this is that if we’re in this predictive world, then one of the key ways of dealing with anxiety is to get your predictions in line with reality as much as possible. But you can get stuck if you try to do that by just escaping because if you go in a dark room and curl up, yeah, you’re gonna reduce prediction error ’cause nothing’s going on, but you are not going to get the information you need to be able to manage the world. And so people with anxiety disorders tend to have a very high intolerance of uncertainty. We just don’t like it.
I am on the autism spectrum, and I like things to be predictable. I am not into extreme variability most of the time, except for in intellectual pursuits. But if you’re constantly having to worry about what’s gonna happen—am I gonna go out the door and get hit by a car? Or, does that person hate me?—then you’re spending a lot of mental energy, and if you start avoiding, then you will think that, like, you know, “Oh, I’m scared of dogs. I will avoid dogs. Didn’t get bitten. It worked. My strategy’s good. I will continue avoiding.” And so that creates a negative cycle that you can really get stuck in. And so while everyone with anxiety does not want to stop avoiding the things that they are avoiding, that is inevitably part of overcoming it.
Feltman: The researchers that you spoke to, published something about four phenotypes for managing uncertainty. Could you tell us what those are?
Szalavitz: The first one is probably where I would constitutionally fall, which is fragile. And what that means is: does not like uncertainty, tries to hide in a dark room, is congenitally oversensitive to stimulation. Or there’s another way to get into the fragile pathway, which is by, like, abuse or neglect or growing up in a really unpredictable situation. And so you kind of turn toward dysfunctional strategies like just hiding rather than actually exploring the world and resolving the prediction errors in a productive way.
And so the second type is called durable, and these are people who are growing up in a very, very predictable world, which would seem to be good, but if it’s so predictable that you never experience prediction errors, and you rarely experience the stress of having to deal with unexpected things, then you’re obviously, by nature, not going to be good at it, and you may underlyingly be fragile.
The third type is resilient, and this is what people tend to think is, “oh, this is, like, the best possible type,” because you don’t hide, you manage stress, you go and are curious about the world and learn, and by learning you build up your muscle of dealing with stress, right? But the authors of the paper argue that there’s a fourth type, which they very annoyingly call proentropic—which I would think it would be antientropic because you don’t want to die in entropy—but anyway, the point is that such people actively seek challenge and actively seek to, you know, deal with uncertainties and manage them successfully. And when you do that, you know, like anything else, the more you do something, the better you get at it generally.
Feltman: Yeah. Could you give some examples of, you know, what that would look like in the real world?
Szalavitz: You could think of it as anything from, you know, a firefighter who’s going into these very intense, dangerous situations and managing extreme stress very regularly. Or you could say, well, a scientist, you know, somebody who is, daily, challenging the way they see the world and, like, putting ideas to various tests and that sort of thing. So it could just look like anybody who really is trying to not avoid the world too much. Now the thing is: sometimes avoidance is productive, and you asked earlier about other mental illnesses that can be affected by this predictive thing, and depression is really one of them. You can kind of think of depression as being “let’s curl up in a dark room,” like a strategy of that. And, you know, people argue that, evolutionarily, that may be why we have this thing that we call depression in the sense that if you’re failing, sometimes withdrawing and, like, getting out of the situation is a good idea. And if you have a mechanism to make that happen, if you’re, like, failing, like, socially or in whatever ways, then that could be productive. However, if you get stuck there, then it becomes unproductive, and obviously the feelings of it can just escalate such that it’s very difficult to escape from without, you know, chemical or therapeutic help.
Feltman: So can understanding the brain as a prediction engine actually help us cope better with uncertainty? Your piece certainly had a very compelling narrative lead that got into this a little bit. So maybe you could start by sharing that with our listeners.
Szalavitz: When I was interviewing the authors of the paper that described those various phenotypes, they had just survived the Eaton Fire in Los Angeles, and they had been working on the paper during that, and the fire actually threatened their homes very significantly, and they had to, like, evacuate in a smoky, messy, scary hell.
And to top it all off, one of the co-authors was injured in a childhood diving accident, and so he is quadriplegic, and he uses a wheelchair, and the mobility van that transports his wheelchair was undergoing service. So he couldn’t get into his wife’s car, and they had to leave, and he had to, like, drive his electric wheelchair through, you know, the falling debris and the howling winds and the smoke and all of that to get to a place where a vehicle could get to that would be capable of him getting picked up.
Feltman: Yeah. That’s a tough one for the old prediction engine, I got to say. [Laughs.]
Szalavitz: No, exactly. And so that was why, like, when I was interviewing them, Laura Harrison brought it up because she was like, “This got put to the test for us very seriously, you know, in these circumstances.”
And so, you know, I wasn’t originally gonna write about any of that. I was just writing about the paper, and so then I had the anecdote that would, like, allow us to get into, you know, the very real-world implications of how we manage uncertainty. And, you know, they were talking about, while they were in that moment, like, thinking about these ideas and thinking about, like, “Okay, well, how do I respond in a proentropic or at least resilient way?” and, you know, “How do I manage the odds of, like, is it safer to stay or go?”
And so it was really fascinating to think about that. And when you do that, the other thing that becomes very clear is that managing your emotions is very important because your emotions are kind of there, they’re kind of decision-making algorithms to, like, guide you when under uncertainty, right? Like, this has worked in the past, so this will feel good. This doesn’t, so it won’t, right? So you have to be aware of those influences and whether they are guiding you in the right direction or not. And you also have to be aware that, like, when your brain is flooded with stress hormones, you are not thinking about the future the way you ordinarily do because you’re very narrowed in on where you are now, which is important when you’re getting out of a fire.
But it means that you’re not going to be at your most creative or at your most able to deal with abstraction. And so, you know, this is why, like, in national emergencies, it’s so difficult to manage because we just freak out. [Laughs.] I mean, I sort of feel like what happened, you know, with the pandemic is that that massive social stress moved everybody a notch higher away from planning and abstraction and deeper thought and into a more reactive and often aggressive, unfortunately, mode. So kind of understanding all of this in this context is really useful when you want to try to, like, cope with either on a personal or a, you know, political level the challenges that we are facing now.
Feltman: Well, and speaking of that shift in the wrong direction that I think we all feel very aware of, you made the point that these phenotypes aren’t really fixed, and that we can change the way our prediction engines work. Could you talk a little bit about how that’s possible?
Szalavitz: So obviously, we are limited, to some extent, by our temperament. Like, I’m never gonna be the chillest person in the world. That probably will not happen. But we can move away from our default responses toward, if they have become dysfunctional, toward more appropriate and, and helpful ones. And you do that by explicitly facing it—not in a way like tough love or force or any of that. And this is a lot of what cognitive-behavioral therapy is. It’s like, you know, “Okay, I walk into the room; I assume they hate me.” That’s my brain doing the assumption. It’s not them doing anything. So I have to be aware of that, and I have to say, “Well, what’s the evidence for my belief that they’re going to hate me?” And there probably isn’t any. And so I need to, like, let go of that and just go into the room with a better expectation. When you face these kinds of things, you have to not assume your perception is automatically correct or real, and you have to assume that it’s provisional.
And the more you can, like, expect surprise and, you know, work toward having the surprises ideally be pleasant ones, then you can get better at managing the whole thing—because it’s kind of amazing how much stuff is, like, building muscle, where it’s just, like, you start with, like, the lower weights, and you suck, and you’re bad at it, and then you repeat it, and then you actually see that you can really get a lot stronger, right? So this is true whether it’s managing stress or, you know, learning anything. But what we tend to do is, sort of, push people into the deep end, and then they’re like, “Well, I can’t do this. I’m bad at this.” It’s like, if you, for your first [time] trying to lift weights, [you] tried to lift 300 pounds, you would not succeed, and you would probably injure yourself, right? But you can, I can attest, build up to this by a lot of practice over many, many years. When it is stress or when it is, like, some of our psychological tendencies, a groove gets worn into your brain because it’s easier and because otherwise we couldn’t possibly handle the complexity of the world.
So one of the interesting things and implications of the predictive brain stuff is that if you can increase brain plasticity, you can interrupt some of these patterns and help people predict in a more functional way. And it’s really interesting to me that everything, with a few minor exceptions, that has antidepressant effects, whether it’s, like, SSRIs [selective serotonin reuptake inhibitors], exercise, ECT [electroconvulsive therapy], psychedelics, all of these things increase plasticity. And psychedelics do it in an especially interesting way because they seem to open up windows of plasticity that are only open at certain points during development. And so I find that fascinating.
Feltman: Well, thank you so much for coming on to talk to us. Your article was fascinating and it was great talking to you.
Szalavitz: Thank you as well.
Feltman: That’s all for today’s episode. You can read Maia’s feature online or in the latest issue of Scientific American, which is on newsstands now. We’ll be back on Monday with our weekly science news roundup.
Science Quickly is produced by me, Rachel Feltman, along with Fonda Mwangi, Sushmita Pathak, Allison Rodgers and Jeff DelViscio. This episode was edited by Alex Sugiura. Marielle Issa and Aaron Shattuck fact-check our show. Our theme music was composed by Dominic Smith. Subscribe to Scientific American for more up-to-date and in-depth science news.
For Scientific American, this is Rachel Feltman. Have a great weekend!