Something You Should Know - How to Get Better Results with AI & The Science of Healing Trauma

Starting point is 00:00:00 When you're with Amex Platinum, you get access to exclusive dining experiences and an annual travel credit. So the best tapas in town might be in a new town altogether. That's the powerful backing of Amex. Terms and conditions apply. Learn more at Amex.ca. today on something you should know why the expiration dates on some foods can be completely ignored then how to really get the most out of AI tools like chat GPT and the tricks that make all the

Starting point is 00:00:46 difference well let me tell you a secret you do actually get a better answer out of chat if you're polite to it. So in society, if you are rude to someone, you're less likely to get a helpful answer. So the model simulates that, and if you're rude to it, it doesn't give you such a helpful answer. Also, simple things you can do to be instantly more attractive, plus a fresh look at trauma, and does trauma truly shape your future forever? That is a myth that these experiences that we have will somehow negatively impact us for the rest of our life. The body and the mind, have such an innate ability to heal all of these experiences that we go through.

Starting point is 00:01:26 All this today on Something You Should Know. Did you lock the front door? Check. Close the garage door? Yep. Installed window sensors, smoke sensors, and HD cameras with night vision? No. And you set up credit card transaction alerts, a secure VPN for a private connection,

Starting point is 00:01:44 and continuous monitoring for our personal info on the dark web? Uh, I'm looking into it. Stress less about security. Choose security solutions from TELUS for peace of mind at home and online. Visit tellus.com slash total security to learn more. Conditions apply. Something you should know. Fascinating intel, the world's top experts,

Starting point is 00:02:07 and practical advice you can use in your life. Today, something you should know with Mike Carruthers. Almost every food you buy in the store today has an expiration date on it. And some of those expiration dates are very important, some not so much. Hi and welcome to something you should know. So you know it when milk goes bad, but other foods may leave you wondering, well, exactly how long are they good for? So here's a list of foods that you really don't have to worry about for a very long time. And salt is pretty high on the list.

Starting point is 00:02:45 Salt has been used to preserve other foods for centuries, so it should come. is no surprise that salt itself will last a long time. Apparently, adding iodine to salt does shorten its life a little bit, so if you have iodized salt, it should be good for five years, and other salts will last even longer. Now, honey is supposedly the only food that lasts forever. It's highly acidic and low in moisture, so bacteria can't really grow. The oldest jar of honey ever found is thought to be about 5,500 years old. So any honey you have is probably okay. Sugar. Now sugar manufacturers are required to give sugar an expiration date, but you probably don't need to pay much attention to it. Keep it in an airtight container and it will last almost forever.

Starting point is 00:03:39 And even if it gets hard, it's still edible when you soften it. Maple syrup, like honey, will last pretty much forever. Now, it can develop surface mold, and when it does, you skim it off, you boil the syrup, skim off the mold, and the syrup should be fine. Hard liquor. Now, cream liqueurs do have a shelf life, but hard liquor will last a long time. Now, the flavor can change. It seldom changes for the better, but it doesn't actually go bad. And dried beans. In a study at Brigham Young University, 30-year-old dried beans were still acceptable according to tasters. So the beans you have are probably just fine. And that is something you should know. Over the last several months, I've gotten more and more into AI and chat GPT, in part from doing some of the interviews here that I've done on something you should know with people who are really into it and have studied.

Starting point is 00:04:45 it and written about it. And the more I learn from them, the more I want to learn, because I'm pretty sure I have a long way to go to really understand what it can do and how to interact with it to get the results I really want to get. One person who has some great insight into this is Christopher Summerfield. In fact, he's one of the smartest people I think you will ever hear talk about AI in a way that we can all benefit and understand. Christopher is Professor of Cognitive Neuroscience

Starting point is 00:05:17 at the University of Oxford, staff research scientist at the pioneering Google Deep Mind, working in the field of AI research. And he is author of a book called These Strange New Minds, How AI Learned to Talk and What It Means. Hi, Christopher, welcome to something you should know. Thank you for having me. So I would guess that most people who listen

Starting point is 00:05:42 to this podcast have, by now, have dabbled with open AI or chat GPT to get a sense of what it's about. And I use it a lot. I mean, I just, and the more I use it, the more I really like it and like to experiment with it. But I have to admit, I don't really understand it. I mean, I know the principles I've heard. I've talked to people here about it. But can you explain it in a way that that maybe dives a little deeper, but not too deep, but really gives a good sense of what goes on when you type in a prompt to chat GPT or Open AI. How does it work?

Starting point is 00:06:23 Yes, so AI technologies have been around for a very long time. And the dream of building a system that can interact with people in the way that we interact with each other has been the kind of north star of computer science and AI research since like the 1950s, immediate post-war period. And what happened in around about 2010 or 2012 was that the coincidence of two things. So number one, we developed some new algorithms for processing data that allowed us to develop systems that rather than kind of just being out of the box

Starting point is 00:07:04 able to think and reason like we can, what they could do is they could learn from large quantities of data and gradually through that training process get better and better at doing tasks that we had set them. And so that coupled with the emergence of like lots and lots of data on the internet created a revolution, which people call the deep learning revolution. But everyone said, well, how far is this going to go? What can we really do with this? And then around about 2019 or 2020, people started to build very big models and those very big models started to learn to produce language in very similar ways to the world. The outputs look very similar to the outputs that humans produce when generating language.

Starting point is 00:07:50 And so chat GBT, which you mentioned, is just one example of a language model. That is a model that can respond to your utterances with outputs that resemble what another human would say. And so when I type in a question to chat GBT and get a response, what happened in those three seconds? Where did my question go, and where did the answer come from? What was used to create that answer? So the keystrokes that you made were converted to something called tokens. So tokens are just like digital representations of words or bits of words sometimes. And those digital representations were fed into a, they're sent to a data center and fed into a computational

Starting point is 00:08:35 stack, which is based on an algorithmic tool that's called the Transform. And the transformer, which was developed around about 2017, and it's actually the kind of, it's really the workhorse of modern deep learning. It's the reason that it's so successful, at least in the last few years. What that transformer does is it essentially tries to work out. If you've got a long list of tokens, like a sentence or even a paragraph of text, it has been trained to work out which bits of that input are related to which other bits. to give you a very simple example, if you say Peter and Jane walked down the street, he, and then you want to understand what the subsequent words in that sentence are, you know that he is most likely to refer back to Peter and less likely to refer back to Jane because it's a male sounding name and a male pronoun in English. So you can make that relational connection because you've learned that that's how language works. So language is full of cross references. And so the transformer is very good

Starting point is 00:09:44 at learning those relations. And in fact, it's very good at learning those relations not just for elements of a sentence like he and Peter, but actually for elements of all data that you feed to it, including, for example, if you fed it a whole book, maybe something that happens at the beginning of the book is related to something that happens at the end of the book. And the model has been trained to understand those relations. So it can process that all of those that relational information, and it uses that to make a prediction about what's going to happen next. And that prediction is then turned back into a token and fed to you as the output. So even though it feels like the model is responding to you as if it was a person, it's not actually

Starting point is 00:10:26 responding to you as a person. What it's doing is it's saying something that it is predicting that another person would be very likely to say in response to your query. But often I will ask a question of a person and the answer is, I don't know, but I ask GPT and it always knows. It always has an answer. It finds the information. Where does it find it? Yeah, so humans, of course, are overconfident. I'm a psychologist, so I can tell you this. So humans are overconfident on average in their replies, although as you pointed out, we're usually quite well calibrated. So we usually know when we don't know something and we know when we do know something, although not always. The models are much less well calibrated, and they tend to be even more overconfident than we are.

Starting point is 00:11:15 So even if they kind of don't know an answer, they'll very often give you an answer. So the question is, why do they do that? And the answer, there are many answers to this question, but probably the easiest to understand is the models are one of the ways in which the models are trained by the developers, so ChatDBT is trained by a developer called OpenA.I. One of the way in which they're trained is they're trained from human preference data. And so what that means is that the model will generate, during training, the model will generate two candidate answers, like answer A and answer B. And then the developer will ask a group of people, who are usually crowd workers, will ask them, do you prefer answer A

Starting point is 00:11:57 to answer B? And what they'll do is they'll use that preference data to train the model to make more answers that people prefer. So in theory, that should be a good thing. Models respond in ways that people like. That's, you know, in theory, means that they're going to be a better product. But of course, what people like is not always necessarily what is true. So people like it when you give a confident and eloquent answer to a question. And so lo and behold, they will choose the confident and eloquent answer. And they might choose it even if it's wrong. And so the model is trained to produce confident and eloquent answers to our questions. And it does so very, very effectively, but sometimes at the expense of truth. But when you ask it a question, I have a certain

Starting point is 00:12:45 kind of stove and it has this kind of light and the light's doing this thing. What could be wrong? Where does chat GPT find that information? It's fine to predict, but it's got to be based on some knowledge, where's the knowledge coming from? Yeah, so the knowledge comes from different sources. So the most important source of knowledge, the kind of foundational knowledge in the model, comes from the training data. And so I mentioned that the models are very, very big, meaning they've got lots of kind of what computer scientists would call trainable parameters. That means that there's basically lots and lots of things that you can change in the model to change how, adjust in the model to change how it works. But the other way in which they're big is that they have been trained

Starting point is 00:13:34 on lots and lots of data. So you sometimes hear people saying, these models have been trained on the entire internet. And that is not exactly true, but it's sort of an approximation to the truth. So what the developers have done is they've taken enormous swathes of text that you and I have written on the internet, whether that's like, you know, blog posts or social media posts or, you know, kind of books or news articles or just everything that's on the internet. And they have used that data in this predictive way to train the model to say the next, to produce the next token. Now, humans know a lot of stuff.

Starting point is 00:14:15 There's 8 billion of us. And so collectively, the information that we have put on the internet encompasses the vast majority of the things that we could, that we need to know about. And what's more, the way the transformer works, it allows it sometimes to stitch together bits of knowledge that even people haven't stitched together yet. So it will take knowledge about one bit of one field or discipline and it'll kind of combine it with something new in a very much in a way that a human might and come up with something that is unique and different and maybe even have. hasn't been said or thought before. And so the combination of this vast training data with the

Starting point is 00:15:00 computational power of this transformer architecture allows the model to answer questions like the one you articulated if you ask it how to fix your washing machine, even if it hasn't seen that specific washing machine or nobody has phrased that specific query in that way before, it'll usually be able to tell you something reasonably sensible. We're diving deep into how to use AI more effectively, and my guest is Christopher Summerfield. He's author of the book, The Strange New Minds, How AI Learned to Talk and What It Means. Hi there, Fred Greenhald here, director of audio dramas like DC high-volume Batman and Star Trek Con. However, my one true love remains all things spooky, and I'm excited to say there's a new season of my horror

Starting point is 00:15:45 podcast, Undertoe. This season is called Familiar Haunts, standalone horror tales that reveal how the past is never truly gone, and humanity may be the the most ruthless monster of them all. Familiar haunts is available now. Find it by subscribing to Undertoe wherever you get your podcasts, such as the app where you're listening to me right now. Search for Undertoe. Hi, I'm Adam Gidwitz, host of Grimm, Grimmer, grimist.

Starting point is 00:16:10 On every episode, we tell a grim fairy tale. Not the cute, sweet versions of the fairy tales that your children have heard so many times. No, we tell the real grim fairy tales. They're funny, they're weird, sometimes they're a little bit scary. But don't worry, we rate every episode, Grim, Grimmer, or Grimmist, so you, your child, your family can choose the episode that's the right level of scary for you. Tune in to Grim, Grimmer, Grimmist, and our new season, available now.

Starting point is 00:16:41 So, Christopher, do the different ones, there's, you know, Chachy BT, but there's a, there's a million of them. Well, there's probably not a million of them, but there's a lot of different interfaces that you can go to. If I ask the same question, you know, to several, do you generally get the same answer or everyone has its kind of own personality? Yeah, so the models do differ in their training data and that means both the bits of the internet that they've been trained on. So we call that the pre-training data. And then also different developers do different things at post-training. So they might, when they're measuring those human preferences that we talked about, they might instruct the raters, the human

Starting point is 00:17:23 to behave in different ways, and that will imbue the models with slightly different values. So there's been a lot of work to show that the different models have slightly different preferences over kind of social or cultural or political values, and that's because of the way they were trained. So famously, some models like Open AIs models have often been found to be slightly more libertarian and progressive on the political spectrum, whereas XAI, XAI's model, GROC, is often like a little bit more on the other end of the political spectrum. And that's because of the way that they've been, the models have been trained.

Starting point is 00:18:06 Very importantly, even the same model will not give you the same answer every single time you ask the same query. And that's because there's a degree of randomness in their outputs. So just like, if you asked me the same question twice in a row, I probably wouldn't respond verbatim exactly the same thing. Neither does the model. I've had the experience, and I know other people have talked about this, too, that when I type in something into chat GPT or OpenAI, I feel like I'm talking to a person. And it's very different than, say, when you type in something to Google.

Starting point is 00:18:40 That never felt like a person to me. But there is something about ChatGPT that feels very much, even though I know it's not a person, it feels like. It feels like a person. Yeah, that's absolutely right. And, you know, we have known about this phenomena actually since, for a few decades. So some of the earliest AI systems, which generated language, they did so in a very, very kind of rudimentary and stylized way by basically, you know, you can simulate a conversation by just always asking a question back at the other person. And so early chatbots did this trick. It's basically, they're basically just a bag of tricks.

Starting point is 00:19:21 And yet, people felt drawn in by these tools, even many decades before we had actually good conversational AI. We also, I think, just generally have a tendency to impute intentional states or mental states to complex systems. So, you know, when your computer shuts down or goes wrong, you get annoyed at it, right? There's no reason. It's just a lump of metal. No reason to get annoyed at it, but you do. And philosophers have called this, they call it something called the intentional stance. So we have an intentional stance towards machines. We believe that they have mental states. We're prone to believe that they have mental states. So chat GPT, of course, is just waiting for us to, you know, treat it like a friend. Right. Almost to the point where you think if you're, maybe if you're rude to it or you're not polite, you won't get as good. good an answer. Like you feel like you need to nurture the relationship and there is no relationship. Well, let me tell you a secret, which is that we know that you do actually get a better answer out of chat, GBT, if you're polite to it. And the answer to why that

Starting point is 00:20:36 happens is actually if you recall that these models are essentially predictive machines. So they're trying to predict the answer that another person would be most likely to give. So So in society, if you are rude to someone, you're less likely to get a helpful answer. So the model simulates that, and if you're rude to it, it doesn't give you such a helpful answer. But if you're nice to it, does it give you a better answer than if you're just neutral to it? It does, yeah. So it'd be nice to chat, TBT.

Starting point is 00:21:07 Isn't that? I don't think people know that. I think people would almost like to believe that to be true and act that way, but I bet most people, if you ask them, would think, well, no, I know it doesn't matter what I say to it, but that's not what you're saying. No, it does matter what you say to it. And so another example is that if you simply tell the model to think more carefully about the answer it's about to give you, it will also give you a better answer.

Starting point is 00:21:38 No, really? If you ask it to solve a reasoning problem and you just present the reasoning problem and say, what's the answer, then, you know, maybe it'll get it right. it'll get it wrong. But if you present the reasoning problem and you say, please think about this problem step by step, then the model is much more likely to give you a better answer to the problem. What are some other ways? Like when you said be polite to chat GPT and you'll get a better answer, I mean, that's just, I never thought of that. But what other things like that, if there are any suggestions on how to interact, how to prompt AI to give you more of what you really

Starting point is 00:22:18 want based on all the knowledge you have? Yeah. Well, I'm sure there are far better answers than I can give to this question online. There are millions and millions of websites that give you tips and tricks for using language models more effectively. But let me just say one thing that I think is a sort of general principle. or other language models, I don't think at the moment we should think of them like we would another person. So your friends and family will have distinct personalities. They'll have

Starting point is 00:22:49 distinct character that help you predict how they would behave. And people tend, most part, people tend to stay in character. So their character doesn't swing wildly from one day to another. Chad GPT is not like that. So what it does is it assumes a character which is based on the context that you've given it. So depending on the way that you interact with it, it will instantiate a context, and it will start to behave like people would behave in that context. So if you start to talk to it using lots of slang and swear words,

Starting point is 00:23:26 then it will, well, actually, that's a bad example because it's been trained not to use swear words. But if you start to talk to it using lots of slang, then it will talk back in a way which reflects that, It reflects what it thinks you're trying to achieve and how you think it's, and that sort of mode of behavior. So you should be aware when you're interacting with the model that in a way you are guiding it. So the tone, the demeanor of your replies, the topic of your replies is setting the context for the model. And it doesn't have a unique and different character like other humans do.

Starting point is 00:24:02 And it doesn't remember your character from the last time so you would, have to re-establish it every time? Well, so that depends on which model you're using. Like I said, these modern models have these memory features. One thing that you can do is, that's kind of fun, is you can ask the model to generate, and you can do this with chat dbt, and probably other models, you can ask it to generate an image of what it thinks you look like. That's a fun trick.

Starting point is 00:24:31 And so you might find that it's surprisingly accurate because, basically, basically, because based based upon lots of information that you've given it, it might be able to make a guess about various aspects of your age, your gender, maybe your ethnicity. And so the model is always kind of like adapting to you and how you respond to it. So really, it's up to the user. I mean, how the user uses it determines the results you get. What a fascinating conversation. I've been talking with Christopher Summerfield.

Starting point is 00:25:07 He is a professor of cognitive neuroscience at the University of Oxford, and he's author of a book called These Strange New Minds, How AI Learned to Talk and What It Means. And there's a link to that book at Amazon in the show notes. And Christopher, I appreciate you taking the time to talk about this. Thanks so much for the opportunity. And yeah, do reach out if you want to know anything. Do you love Disney?

Starting point is 00:25:31 Then you are going to love our hit podcast, Disney. Countdown. I'm Megan the magical millennial. And I'm the dapper Danielle. On every episode of our fun and family-friendly show, we count down our top 10 lists of all things Disney. There is nothing we don't cover. We are famous for rabbit holes, Disney themed games, and fun facts you didn't know you needed, but you definitely need in your life. So if you're looking for a healthy dose of Disney Magic, check out Disney Countdown wherever you get your podcasts. Hey everyone. Join me, Megan Rinks. And me, Melissa D.Montz, for Don't Blame Me, But Am I Wrong. Each week, we deliver four fun-filled shows.

Starting point is 00:26:08 And Don't Blame Me, we tackle our listeners' dilemmas with hilariously honest advice. Then we have But Am I Wrong, which is for the listeners that didn't take our advice. Plus, we share our hot takes on current events. Then tune in to see you next Tuesday for our listener poll results from But Am I Wrong. And finally, wrap up your week with Fisting Friday, where we catch up and talk all things pop culture. Listen to Don't Blame Me, But Am I Wrong, on Apple Podcasts, Spotify, or wherever you get your podcasts. New episodes every Monday, Tuesday, Thursday, and Friday. Trauma.

Starting point is 00:26:45 It's a word we hear constantly. We talk about people with traumatic childhoods, friends who have endured traumatic experiences, or even our own traumas. That word carries weight, suggesting something. far more lasting than just a difficult time. But what really is trauma? Does it always shape our future, or does it depend on how we define it in the first place? To help clarify and explain what's really happening in the body

Starting point is 00:27:16 when we experience trauma, I'm joined by Dr. Amy Epigion. She's a double board-certified physician in preventive and addiction medicine with master's degrees in biochemistry and public health. She's the author of, of a book called The Biology of Trauma, How the Body Holds Fear, Pain, and Overwhelm, and How to Healing It. Amy, thank you for coming on. Welcome. Thank you, Mike. It's really good to be here.

Starting point is 00:27:42 So let's get a working definition of what trauma is, because it seems to be, I don't know, different things to different people. Some things are traumatic for some people and for other people maybe less so. And so what is trauma? Absolutely. And I had the same question. And in fact, it was until I developed health conditions that I knew were related to childhood trauma, that I really questioned that because I was not one that had seen trauma in my childhood.

Starting point is 00:28:13 And so my definition now of trauma is focused on the body's experience. And there are very distinct physiology changes that happen when the body goes into a trauma response that makes that be a trauma up until that point it's just a stress not that stress is always great sometimes it is good and sometimes it is exactly what we need and sometimes it just is chronic stress that is not helping us but there is a very distinct change in our physiology that makes that stress turn into a trauma and the inner experience then it can be simplified down to anything that for any reason at that time overwhelmed us and our ability to respond. And so an example, just a quick example of that, I don't want to go too deep down the rabbit

Starting point is 00:29:07 hole here, but just a quick example of attaching example to that. Absolutely. So I'll share the story of my patient Elena. And Elena came to me after she developed autoimmunity in her late 40s and she had had a car accident a year prior. She had not made that connection at all, but me knowing the nervous system, I started to ask some questions. And it was not surprising that she had had a childhood where, again, she also would not have said, ever use the word trauma for her childhood. But as I'm asking her about when she had experienced overwhelm and when she could look back on her life and think that that was a moment in which I didn't know if I was going to be okay. And I just felt like something inside of me

Starting point is 00:29:54 shut down and broke in order to keep going. And she described an experience, multiple experiences, actually growing up with a single mother who she was always afraid of, are we going to have enough? Am I too much for my mother? My mother's already got so much on her plate. I need to not have any needs and she became hyper responsible, even at a young age. And so even though this would not be a typical event per se, like the car accident that she had, that we can also look at that, it was the experience, it was the inner experience of that was too much for a young child to understand, to be able to process and know how to respond in a way that was productive.

Starting point is 00:30:40 She was a young child. how would she know how to navigate adult problems? But that's exactly what happens to a lot of people, is that they are faced with a situation that they just are not equipped to understand enough or have the time or have the space to be able to respond in a way other than to just kind of shut down. Okay, but here's the thing about the word trauma

Starting point is 00:31:06 that I think most people would agree with this, that one of the things about a trauma is it's going to affect you for the rest of your life. That that's the difference. You break your arm. You go to the doctor, he fixes it, and maybe that's a trauma. It's certainly a trauma to your body, but it's fixed and life goes on. But the kind of trauma that you're talking about, we often assume that because they went through this trauma, somehow that's going to mess them up for the rest of their life.

Starting point is 00:31:40 Well, let me challenge that for a moment, Mike, because as we look at how does the body heal a trauma to the bone? A break to the bone is a trauma. It breaks the bone. It doesn't grow the bone. It doesn't strengthen the bone. It broke the bone. I talk about a patient Kenneth who came in. He was a veteran. He was coming into the VA hospital where I was a general surgery resident. He had broken his bicep tendon. Rather than being attached to his shoulder, it was sitting in the middle of his arm. That is breaking something because we've exceeded our capacity. That is the same thing that happens to us in any other of these other maybe more emotional traumas or psychological traumas that we want to associate that word with. And yet for the body, it's the same principles. It's the same thing to ourselves of this is more than I can handle. And so rather than completely die, I'm going to just break the bone. I'm going to break the muscle tendon so that the rest of me is okay. And then I can heal. And that's what we see in the body. The body heals the bone. The body heals the muscle as long as it has what it needs to heal. And I think we can all look at people

Starting point is 00:32:58 who've had some broken bones. And they've never really healed well from that. They still have chronic pain. It lingers. They've got issues. Same thing with that. tendon with my patient Kenneth. If we had not done the surgery to reattach his tendon, yes, he would have survived. Yes, he would have been able to go on with his life, but he'd forever have one arm that was really weak because his bicep muscle wasn't working. So the body needs certain things in order to be able to engage its own healing strategies. And it is no different than any other traumas that we want to say. The body is designed to actually heal from those traumas. The fact that it doesn't and the fact that we haven't moved on is actually a sign

Starting point is 00:33:45 that we have not given our body what it needs to do its own innate healing. But many people break bones, have traumas, and they're not bothered by them anymore. I broke a bone, I broke a bone in my arm from rollerblading many years ago. It affects me in no way whatsoever. So give me an example of a trauma that you consider a trauma that goes through a process and heals, and it's fine. You can look at an example of a car accident. That can be a trauma for some people because in that moment, they can be so overwhelmed by what's happening that sometimes they can faint, sometimes they can lose consciousness. This is the body's way of showing. This is way too much for me to be able to respond to.

Starting point is 00:34:35 I'm just going to check out. And how do we then come through those experiences? How do we actually give the body what it needs afterwards? This is where a lot of people are unaware and so unintentionally keep their body in a sense of danger. What we actually need is, first of all, to return to a complete sense of safety. You'll see this in nature all the time. So one example is a turtle.

Starting point is 00:35:03 And when a turtle experiences overwhelming danger, not just a stress, an overwhelming danger, they just retreat inside their shell. They pull their head in, they pull their legs in. And there's no way that they're coming out of that shell until they've sensed that they are safe to do so. Well, in a trauma response, our body is essentially doing the same thing. Our body is essentially saying, this is not just stressful. If it were stressful, I'd keep trying to run away.

Starting point is 00:35:34 I'd keep trying to fight. I would keep trying to respond. But this is so overwhelming that I'm just going to shut down. I'm going to retreat. I'm going to pull into my shell and just go into full protection mode. And as long as we are living protected, we can still function. But there's just this background sense of the danger might still be there. we always have that sense of the other shoe might drop.

Starting point is 00:36:03 And even when times are good, we are waiting for the next bad thing to happen because we know that it's coming. We know that we don't deserve good things or deserve to be happy or we can't experience that joy as long as our body is like, I'm not sure that that danger ever truly went away. So that is why actually knowing, having the skills and the knowledge, for what the body needs to know that a danger is truly over is the first step and the essential step to actually moving on in a way that we have healed from our past and not just bringing it along with us. But I want to go back to this idea and have you address it very directly, that people have this belief. And I can give you an example.

Starting point is 00:36:52 I know someone who, when he was 10 years old, saw his father die. And there's this assumption that this will affect him for the rest of his life because it was so traumatic. Is that necessarily so? It is not necessarily so. That is a myth, that is a misconception that these experiences that we have will somehow negatively impact us for the rest of our life. the body and the mind have such an innate ability to heal and not just move on, not just push through, but actually heal all of these experiences that we go through. We just haven't known how.

Starting point is 00:37:38 So I've had many patients who come to me and they almost bring their scorecard with them, this is what happened to me, this is what happened to me, this is what's happened to me. And it's a lot. And I look at these and I've even worked with people who have children who saw their sibling or someone else get killed and they haven't spoken in six years. And this idea that how do you ever recover from something like that? It is actually possible. These are not life sentences. And that's the first belief that I get to challenge with my patients is your past is not your life sentence. And as a physician, what I'm focusing especially on is the resolution to our physical health because a lot of people will tell me, look, I feel that I understand why I am the way

Starting point is 00:38:29 I am. I can look back at my life and the experiences that I've been through and I can explain and I understand my triggers. I understand all of that. But my body is still living under the impact of the past. My body will still not only have reactions like when I'm around my family, for example, that's a common example, but it also has an autoimmune disease. It also has digestive issues. I also have sleep issues and ongoing anxiety issues. So my body is somehow still remembering the past, even though I feel that intellectually, I know all of this stuff and I've been able to move on. But given your definition, everybody could have a different road to get to where you want to go. Some people would have the same experience and not

Starting point is 00:39:21 be traumatized because they don't fit your definition of being overwhelmed. Absolutely. And this is where we can see two people go on the same military excursion. And one comes back and it can assimilate with society and enjoy their family and not have illogical reactions. And another one is stuck at home in fear with PTSD symptoms. Same excursions, same experiences, same environment. What was the difference? The difference was that for one, they went into that more resourced. They went into that so that when they experienced those, whatever it was that they saw, whatever it was that they did, whatever was that they experienced, they were able to keep it at the stress level or experience that trauma response and resolve it at that time, whereas someone else wasn't able to.

Starting point is 00:40:16 Isn't one way, though, to address it, to just leave it alone, not leave it alone, but that time will, time is a tool that, and again, I use the example of, you know, when you lose someone, if someone dies who's very close to you, it certainly seems over. It seems like you can't sleep, you can't eat, it's horrible, but eventually time fixes that, that human beings go through the stages of grief and come to acceptance and life goes on. And there was no direct intervention to deal with the trauma. You just figured it out, worked it out, accepted it and moved on. 100% Mike and and this is what I hope that everybody hears is that the body has the innate capacity to move through even the hardest of experiences and have resolution so for those people who have

Starting point is 00:41:19 not had resolution that's what we get to become curious about and why not why did the body get stuck in that grief why five years later are we still ruminating about that person. Why five years later are we still having sleep problems? Why? Why? Why has our body or our mind not seem to fully recover when time should heal? And that was one of the first lessons that I had to learn with trauma. I became a foster parent. This is directly working with trauma. Anytime that you take in a foster child, you are working with someone who's been through trauma. I was under the assumption, Mike, that time heals all wounds, because that's what I've been taught. Time heals all wounds. All I needed to do was provide the love, provide the stability, the security, and then just

Starting point is 00:42:13 give it time. Well, this is what I did, and time was not healing. In fact, he was progressively getting worse, and I'm trying to understand why, because I've created an amazing place for him compared to his past, but he was still so stuck in his past that he could not, could not be actually present and relax into the security, relax into my love. He wouldn't even let me let my love in. And by the age of five, he was talking about how he would want to kill me. And this is how he would want to kill me. And it was obvious that time was not healing. So I was missing something. I was not giving him something that he needed to allow time to be healing. So it's not just that time heals everything. Time only heals when the body has what it needs. And then you compound that

Starting point is 00:43:09 over time. And yes, the healing and the resolution will come. Yeah. So when people stay stuck, when the guy comes back from military service and can't get on with his life, when you lose somebody and you can't time doesn't heal that when your foster son does not get better what's this trauma work that you've mentioned what is it specifically what so you wake up some morning to start your trauma work what is that i don't think it's what people think it is when i talk about trauma work now what i'm really talking about is i need you i need a person to be able to open up and feel and then be able to move through those feelings, not stuffing them back down, but move through those feelings to a sense of, I am safe enough, I am secure enough, and now I can keep myself

Starting point is 00:44:06 open to the world. Anytime that we find ourselves still guarding our heart, still protecting ourselves, that's how we can know, whether I know that I had trauma or not, I'm actually, my body is showing that it has these patterns of trauma. but it starts by doing what you call somebody you talk to a friend i mean when you say you know move through it and and feel your feelings a lot of people don't know what what does that mean yeah and i wouldn't have known what that means either so what i start what i start people with is feeling sensations in their body because i need to get that connection of their mind and their body part of living with chronic trauma has been a disconnection of our mind from our body so we tend to be more in our head and

Starting point is 00:44:50 we like to think about everything, analyze everything, understand everything, we sign up for more courses, we read more books. But we actually don't know how to be with these feelings in a way that will be productive. So I start people with having them feel different areas of their body. And sometimes that can be, I want you to feel your pulse. I want you to actually feel your heartbeat and whether we go to the neck and the artery, we go to the wrist and feeling their artery. but this is starting to form that ability for them to connect. In science, we call this interoception, your ability to sense and perceive your inner physiology and state.

Starting point is 00:45:30 Most people can't. And yet what the research shows is that looking at mothers specifically, if they don't have that ability to know and read their own inner state and physiology, they're not able to do that for their babies. and that's how then we pass on what you might call generational trauma is because a mom doesn't know actually how to connect with her own body and sense it and thus doesn't know how to connect and be able to hear the real messages behind what her baby needs. So this idea of connecting with our body and that can be so vague. And so I've made it very practical by saying we are going to

Starting point is 00:46:10 see how we can create a felt sense of safety right now. And I have people put a pillow or a blanket or a book or a purse over in front of their stomach. And this seems really weird to them at first. And they're like, why would I put this over my stomach? And I'm like, just do it. And then we'll see if this is helpful or not. And as soon as people do that, you should see it, Mike, their shoulders drop,

Starting point is 00:46:35 their face softens, their abdominal muscles just also. kind of relax and they're like, oh, I feel, I feel like I can take a breath. And I'm like, this is why it's important to just do some of this connection exercises at first. Because if I had told you that you were tense before, you would have been like, no, I'm fine because this is your normal. It's not until we actually can create a change in our inner physiology that we're like, oh, now I can feel a difference. And if this is what safe actually feels like, I don't think I've ever felt safe before. And that is when the real change starts to happen, Mike.

Starting point is 00:47:17 So if you had something happen where you had that reaction, you felt overwhelmed in your life. And so that's by your definition trauma. But now you don't feel overwhelmed. Time fixed it. It seems like it's in the past where it belongs. And so are you no longer trauma? Yeah, my response to that would be, are you still going into trauma responses? And it sounds like from your question, you're saying that no, this person is no longer going into that sense of complete overwhelm. And so they are no longer under the impact of that past trauma. Well, given how much we talk about trauma and traumatic experiences and the effects of trauma on people, I really appreciate you weighing in and giving your take on

Starting point is 00:48:10 I've been talking to Amy Epigion. She is a double board certified physician in preventative and addiction medicine, and she's author of the book, The Biology of Trauma, How the Body Holds Fear, Pain, and Overwhelm, and How to Heal It. And there's a link to that book in the show notes. Amy, thank you so much. All right. Thank you so much, mate.

Starting point is 00:48:35 If you would like people to think of you as more attractive and like Here are some solid suggestions backed up by science. First of all, be funny. People with a good sense of humor are considered to be more social and more intelligent. Surround yourself with other people. Studies show that people look better in a group because unattractive features are less noticeable. Get past the small talk. People feel more connected to someone who asks deeper, more sincere questions.

Starting point is 00:49:08 Get a pet. It shows you're a nurturing person and able to make long-term commitments. And smile more. A happy face can compensate for attractiveness flaws. The stronger the smile, the more attractive the face. And that is something you should know. We have a great team of people working on this show. Our producers are Jeff Havison and Jennifer Brennan.

Starting point is 00:49:32 Our executive producer is Ken Williams. I'm Mike Carruthers. Thank you for listening today to something you should know. know. You know, a lot of the guests who appear on something you should know have their own TED Talk. And did you know that TED has a podcast that drops a new TED Talk every day? If you love hearing fascinating stories, insights about human behavior, and mind-blowing concepts that weren't even on your radar before, I recommend you check out TED Talks Daily. You'll hear hot takes and deep dives from the world's leading thinkers, and learn answers

Starting point is 00:50:09 to questions like, how do I stop caring about what other people think? Or is my partner the one? Do aliens actually exist? Don't miss out on your daily dose of inspiration. Listen to TED Talks Daily, wherever you get your podcasts. I'm often asked, as you might imagine, what podcast do I listen to? And I actually have an eclectic taste and I jump around, try different ones. But I will say that I have a couple I'm very consistent about, and one of them is the Jordan Harbinger Show. It's kind of a little like something you should know, but Jordan goes in interestingly different directions. I do know that we share a lot of listeners, a lot of listeners who like this podcast like the Jordan Harbinger Show. Jordan is really good at getting his guest to

Starting point is 00:50:59 open up and share great insights. Recently, he discussed modern romance scam tactics. I mean, that's the lowest of the low, but you've got to know about them so you can fight back against them. And another episode he did was about how society has engineered a generation of lonely men. The show covers a lot of great topics, which, well, like I said, if you like this show, you're going to like his show. There's so much here. Check out the Jordan Harbinger Show on Apple Podcasts, Spotify, or wherever you listen to podcasts.

Something You Should Know - How to Get Better Results with AI & The Science of Healing Trauma

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.