WEBVTT - Ep43 "How do we remember?" (Time Traveling Part 1) 0:00:04.680 --> 0:00:08.760 What is memory in the brain? How is that different 0:00:08.800 --> 0:00:12.640 from the way a computer stores information? How do you 0:00:12.680 --> 0:00:18.239 put together billions of specialized cells and you store your 0:00:18.400 --> 0:00:21.560 home address in their activity? And what does any of 0:00:21.600 --> 0:00:24.880 this have to do with the happy Birthday song? Or 0:00:24.920 --> 0:00:29.880 with squirrels hiding acorns, or with bards memorizing epic poems, 0:00:30.280 --> 0:00:34.200 or with people who cannot forget any of the events 0:00:34.360 --> 0:00:41.320 in their life. Welcome to Inner Cosmos with me David Eagelman. 0:00:41.720 --> 0:00:45.280 I'm a neuroscientist and an author at Stanford and in 0:00:45.320 --> 0:00:49.320 these episodes we sail into our three pound universe to 0:00:49.440 --> 0:00:53.240 understand why and how our lives look the way they do. 0:01:01.880 --> 0:01:06.039 Today's episode is about memory. What is memory? How does 0:01:06.080 --> 0:01:09.560 it work? How do details get stored in your brains 0:01:09.640 --> 0:01:12.760 such that if I say, what was the name of 0:01:12.800 --> 0:01:16.840 your fifth grade teacher? You can retrieve that name even 0:01:16.880 --> 0:01:20.000 if you haven't thought about it in decades? And how 0:01:20.080 --> 0:01:23.480 is this totally different from the way that computers store memory? 0:01:24.160 --> 0:01:26.959 So in this episode, I'll give you a foundation into 0:01:27.080 --> 0:01:32.479 understanding the landscape of memory. Now, this is the first 0:01:32.520 --> 0:01:35.160 of a three part series about the way that our 0:01:35.240 --> 0:01:40.840 brains constantly unhook from the here and now, and they 0:01:40.840 --> 0:01:46.240 go somewhere else. We are time travelers, and only because 0:01:46.280 --> 0:01:50.040 we do this so constantly, we don't even notice how 0:01:50.080 --> 0:01:53.440 amazing this is. I mean, everything you might study in 0:01:53.480 --> 0:01:56.480 a textbook about the brain has to do with Okay, 0:01:56.480 --> 0:02:01.000 here's how vision works. Here's how the visual quartet analyzes 0:02:01.440 --> 0:02:04.880 photons captured at the retina and makes an assessment of 0:02:04.920 --> 0:02:08.960 what's in front of you. And yet we're able to 0:02:09.160 --> 0:02:13.400 travel back to previous times. You can put yourself back 0:02:13.440 --> 0:02:17.440 in that fifth grade classroom, or the first house you 0:02:17.480 --> 0:02:21.720 grew up in, or the moment of your first kiss. 0:02:22.720 --> 0:02:27.320 You can simulate the sights and sounds and smells. You 0:02:27.400 --> 0:02:30.880 can remember how you felt in certain moments. You can 0:02:31.400 --> 0:02:36.640 recreate and reexperience that. You may be able to recollect 0:02:36.680 --> 0:02:40.600 lots of the details that were around you. To do that, 0:02:41.120 --> 0:02:46.840 you just time traveled across years or decades to place 0:02:46.880 --> 0:02:51.800 yourself in another era. Your visual system is not simply 0:02:51.840 --> 0:02:54.799 looking at the photons in front of you. Instead, it's 0:02:54.880 --> 0:02:59.600 now involved in running a simulation of what had transpired 0:02:59.680 --> 0:03:03.960 and an earlier time wherever you physically are right now 0:03:04.000 --> 0:03:07.919 listening to this podcast, You just time traveled. When you 0:03:08.000 --> 0:03:13.040 thought about your fifth grade experience. Now it turns out 0:03:13.040 --> 0:03:16.320 that brains don't just time travel backwards, but they can 0:03:16.360 --> 0:03:19.880 also move forward. So in the next episode, Part two, 0:03:20.320 --> 0:03:24.520 we're going to talk about prediction, and then in part 0:03:24.639 --> 0:03:27.680 three we're going to pull pieces of the puzzle together 0:03:28.320 --> 0:03:32.800 to unlock some wild surprises about how we experience the 0:03:32.840 --> 0:03:39.440 world emotionally given our time traveling talents. So today we'll 0:03:39.480 --> 0:03:42.240 start with memory. And I just want to say this 0:03:42.280 --> 0:03:44.920 is a huge topic that I could teach you over 0:03:44.960 --> 0:03:48.080 the course of years. So for today's podcast, we're going 0:03:48.160 --> 0:03:51.560 to take a high level ride to get a feel 0:03:51.800 --> 0:03:55.840 for the landscape. So one way to approach any big 0:03:55.880 --> 0:03:59.680 topic is to look at the extremes. So a lot 0:03:59.720 --> 0:04:03.440 of memories come easily to us the day we moved 0:04:03.440 --> 0:04:05.640 into our new home, or the day we got the 0:04:05.760 --> 0:04:08.640 job offer, or the time that the phone rang with 0:04:08.800 --> 0:04:13.520 news of somebody's death. But imagine having a clear and 0:04:13.680 --> 0:04:17.760 immediate memory for all of the events of your day 0:04:17.800 --> 0:04:21.040 to day life, such as the dinner that you had 0:04:21.160 --> 0:04:24.280 on November thirtieth, twenty twenty three, or the friend who 0:04:24.400 --> 0:04:28.600 visited you on May twentieth, twenty nineteen, or the swimming 0:04:28.640 --> 0:04:31.880 pool party you went to on July eighth, twenty seventeen. 0:04:32.000 --> 0:04:36.560 Imagine having a vivid, movie like recollection of what you 0:04:36.640 --> 0:04:41.520 did on today's day ten years ago. Now imagine that 0:04:41.560 --> 0:04:46.760 these recollections dominate your thoughts. They run in parallel with 0:04:46.880 --> 0:04:49.840 the events of your waking life, as you commute to work, 0:04:49.880 --> 0:04:53.840 as you talk with your friends, as you shop for groceries, 0:04:53.960 --> 0:04:56.880 as you spend time with your family. You would be 0:04:56.960 --> 0:05:01.920 right that other people would marvel and probably envy your 0:05:01.960 --> 0:05:06.400 exceptional powers of memory, but you might find the process 0:05:06.760 --> 0:05:12.000 uncontrollable and overwhelming. Some memories are painful, or traumatic, or 0:05:12.040 --> 0:05:14.640 just annoying. Then you might wish that you could be 0:05:15.000 --> 0:05:18.840 rid of them, and exhausted by their presence, you might 0:05:18.920 --> 0:05:22.520 finally contact a neuroscientist, hoping that they could help you 0:05:22.640 --> 0:05:27.880 understand what's happening. And that is the real life story 0:05:28.000 --> 0:05:31.840 for a woman named Jill Price, who is a mneminist, 0:05:32.160 --> 0:05:35.520 which means she has an extraordinary memory and she can't 0:05:35.760 --> 0:05:38.839 forget things. So some years ago she went to the 0:05:39.279 --> 0:05:43.640 University of California at Irmine and described her experiences to 0:05:43.680 --> 0:05:48.320 two memory researchers there my colleagues Magaw and Cale, and 0:05:48.480 --> 0:05:52.599 at first they were skeptical, but they investigated her memory 0:05:52.640 --> 0:05:55.640 with a series of careful tests and interviews over a 0:05:55.680 --> 0:06:00.000 period of five years, and the results showed that Price 0:06:00.320 --> 0:06:05.400 indeed had a rare and astonishingly powerful memory, at least 0:06:05.400 --> 0:06:09.919 for certain types of information. Given a particular date, she 0:06:10.000 --> 0:06:14.080 could recall within seconds the day of the week that was, 0:06:14.440 --> 0:06:17.200 and the details of what she did on that day, 0:06:17.520 --> 0:06:20.840 and the news were the events that occurred on that day, 0:06:21.440 --> 0:06:25.839 without any preparations. She easily recalled the dates of every 0:06:26.040 --> 0:06:29.040 Easter from nineteen eighty to two thousand and three and 0:06:29.200 --> 0:06:33.799 her activities on those days. She also gave correct dates 0:06:33.839 --> 0:06:38.600 and personal anecdotes for randomly selected news events like the 0:06:38.640 --> 0:06:41.599 start of the First Gulf War, or the bombing of 0:06:41.680 --> 0:06:47.159 the Atlanta Olympic Games, or the death of Princess Diana. Now, 0:06:47.200 --> 0:06:51.160 while her recollections were detailed and consistent, they were generally 0:06:51.600 --> 0:06:57.839 specific to events of high personal relevance, like illnesses or relationships, 0:06:58.200 --> 0:07:04.359 or big news stories. Interestingly, her general intelligence was average, 0:07:04.400 --> 0:07:08.160 and her memory for other types of information was really 0:07:08.200 --> 0:07:11.960 no better than anyone else's. She never excelled in school 0:07:11.960 --> 0:07:16.080 and wasn't particularly good at memorizing dates and history books 0:07:16.800 --> 0:07:19.880 and at one point her interviewers asked her to close 0:07:19.920 --> 0:07:24.280 her eyes and describe their clothing, and she drew a blank. 0:07:25.120 --> 0:07:28.400 So her memory abilities were so powerful, but they were 0:07:28.440 --> 0:07:33.800 also strangely limited. So when these researchers published their first 0:07:33.880 --> 0:07:37.840 report on Jill Price, she became a news item. Her 0:07:37.880 --> 0:07:40.920 story drew headlines around the world, She was a guest 0:07:40.920 --> 0:07:44.760 on Oprah, and she published her memoirs. But by the way, 0:07:44.960 --> 0:07:49.320 Jill is not the only mneminist. The first official report 0:07:49.400 --> 0:07:52.520 ofneminism that I know about was a book from the 0:07:52.600 --> 0:07:58.520 Russian neuropsychologist named Luria about a newspaper reporter named Sharyshevski 0:07:59.120 --> 0:08:03.400 who had the gift of an untaxable memory, essentially total 0:08:03.520 --> 0:08:08.120 recall for all the moments of his life. Shiashevski first 0:08:08.200 --> 0:08:11.240 came to him in the nineteen twenties because he got 0:08:11.280 --> 0:08:14.600 in trouble at a meeting for not taking any notes 0:08:14.640 --> 0:08:18.160 when there was a speaker, and Sharishevski responded to his 0:08:18.200 --> 0:08:22.520 boss by recalling the speech word for word, and this 0:08:22.640 --> 0:08:26.320 blew everyone away there, and it also blew Sharyshevski away 0:08:26.560 --> 0:08:29.280 that the other people couldn't do that, and so this 0:08:29.400 --> 0:08:33.360 started this thirty year relationship with Luria, who tested him 0:08:33.400 --> 0:08:37.160 in all kinds of ways and had him memorize math 0:08:37.240 --> 0:08:41.679 formulas or huge blocks of numbers, or even poems and 0:08:41.840 --> 0:08:46.120 languages he didn't speak, and in all these cases, Sharyshevski 0:08:46.200 --> 0:08:50.199 could memorize these in a matter of minutes with perfect accuracy. 0:08:50.760 --> 0:08:54.160 Luria tested and observed him for three decades and then 0:08:54.240 --> 0:08:58.720 wrote a book called The Mind of a Mneminist. So, 0:08:58.840 --> 0:09:03.439 although they are rare, neminists exist, and as scientists continue 0:09:03.480 --> 0:09:06.480 to search for the source of their remarkable abilities, it's 0:09:06.559 --> 0:09:11.679 worth asking why their stories capture our imaginations so strongly. 0:09:12.400 --> 0:09:16.000 Perhaps it's because we'd all love to have supercharged memories, 0:09:16.480 --> 0:09:20.800 Or maybe we see how failing memory is so debilitating 0:09:20.840 --> 0:09:24.120 in old age and we hope to escape the fate 0:09:24.200 --> 0:09:28.719 of our grandparents. Or maybe we recognize how little we 0:09:28.800 --> 0:09:34.080 actually retain from our vast, rich set of life experiences 0:09:34.720 --> 0:09:37.920 that shape us. So in today's episode, we're going to 0:09:37.960 --> 0:09:41.079 dive into where memory comes from and how it works, 0:09:41.520 --> 0:09:43.400 and this is going to give us insight into both 0:09:43.480 --> 0:09:47.319 remarkable memories and bad memories. We'll see what memory is, 0:09:47.400 --> 0:09:51.280 how it functions, how it fails, and what purpose it serves. 0:09:52.360 --> 0:09:54.959 So the first thing I want to tackle is the 0:09:55.000 --> 0:09:58.520 mistake we make by calling the storage of zeros and 0:09:58.559 --> 0:10:04.800 ones in computer memory, because computer memory is actually nothing 0:10:04.920 --> 0:10:10.640 like human memory. Computers store in their memory exactly what 0:10:10.679 --> 0:10:13.400 you give them. So if I write a document, I 0:10:13.480 --> 0:10:17.120 expect my computer memory to function perfectly such that it 0:10:17.160 --> 0:10:20.880 gives me exactly that document back. I don't want something 0:10:21.040 --> 0:10:25.880 similar or a degraded version of it. But human memory 0:10:26.000 --> 0:10:30.439 is a completely different ballgame because you're not storing things 0:10:30.640 --> 0:10:34.360 in zeros and ones. Let's imagine that you speak two 0:10:34.480 --> 0:10:37.960 languages and I tell you a joke in English and 0:10:38.000 --> 0:10:40.520 you turn and tell it to someone else in your 0:10:40.600 --> 0:10:44.520 other language. So the joke is obviously not stored as 0:10:44.640 --> 0:10:49.760 zeros and ones. You're not memorizing my exact words and inflections. Instead, 0:10:50.400 --> 0:10:55.040 the joke is stored in a conceptual space in your mind, 0:10:55.440 --> 0:10:58.920 the way that the characters and events fit together, the 0:10:59.160 --> 0:11:03.320 gist of the action. It doesn't matter the language with 0:11:03.480 --> 0:11:06.640 which you output it. And note that this is the 0:11:06.679 --> 0:11:15.920 same thing with a tune. When you remember a tune, 0:11:16.080 --> 0:11:21.120 it's not about the individual notes. Instead, it's about the 0:11:21.160 --> 0:11:26.600 relationship between the notes, and that's why you can transpose 0:11:26.720 --> 0:11:29.439 the tune to any other key and you still have 0:11:29.559 --> 0:11:40.080 no trouble recognizing the tune. It's not about the exact notes, 0:11:40.240 --> 0:11:44.400 but about the relationship between them. And this is the 0:11:44.480 --> 0:11:49.600 same thing when you're recognizing your friend's face. What has 0:11:49.640 --> 0:11:53.560 been stored in your brain is the relationship between their 0:11:53.600 --> 0:11:57.040 nose length and lip thickness, and distance between their eyes 0:11:57.080 --> 0:11:59.640 and shape of their ears, and where their hairline is 0:12:00.160 --> 0:12:04.840 and so on. It's not about the individual features, but 0:12:05.000 --> 0:12:09.320 the relationships between them. And by the way, this is 0:12:09.360 --> 0:12:13.520 the only way we could have good recognition memory because 0:12:13.960 --> 0:12:18.560 other details change. For example, imagine that you learn a 0:12:19.040 --> 0:12:22.000 picture of the face where the lighting is on the left, 0:12:22.280 --> 0:12:24.920 and later you see the same face, but it's lit 0:12:24.960 --> 0:12:28.920 from the other side. If you are memorizing pixel by pixel, 0:12:28.960 --> 0:12:31.480 then the pattern is totally different and you would never 0:12:31.520 --> 0:12:34.640 be able to recognize it or account for all the 0:12:34.720 --> 0:12:37.880 lighting differences that might happen in the world, much less 0:12:38.120 --> 0:12:40.920 the size differences if you see a big picture or 0:12:40.920 --> 0:12:45.760 a small picture. So understanding the relationship between things is 0:12:45.840 --> 0:12:51.440 the only way to have a successful recognition and literally, 0:12:51.480 --> 0:12:55.960 for decades, computer scientists were chasing the wrong ideas by 0:12:56.000 --> 0:13:00.760 trying to get pixel by pixel recognition until they realized 0:13:00.840 --> 0:13:04.959 that it's all about the relationships. And this is not 0:13:05.040 --> 0:13:07.920 just about vision, but this applies to everything we learn. 0:13:08.080 --> 0:13:14.040 Like facts, your brain encodes new ideas with respect to 0:13:14.280 --> 0:13:18.160 other things you have learned. Just imagine two people looking 0:13:18.200 --> 0:13:22.880 at a list of important dates in Mongolian history. If 0:13:22.960 --> 0:13:27.640 one of them already has a richly developed Mongolian history timeline, 0:13:28.200 --> 0:13:33.080 then new facts are readily incorporated into that person's network 0:13:33.120 --> 0:13:37.720 of knowledge. We don't simply memorize facts. We fit new 0:13:37.800 --> 0:13:43.440 data into the relationship lattice of our internal model. Everything 0:13:43.440 --> 0:13:47.280 we've learned before based on what was relevant to us. 0:13:47.840 --> 0:13:50.680 And I talked about this in episode twenty seven. Everything 0:13:50.760 --> 0:13:53.960 you learn is represented in terms of what you already know. 0:13:54.520 --> 0:13:57.880 So when you listen to this podcast, the points made 0:13:58.240 --> 0:14:01.400 would be meaningless unless you all already we're living a 0:14:01.520 --> 0:14:05.199 human life and knew what a brain was and understood 0:14:05.520 --> 0:14:10.640 the particular language I am speaking. We do not encode 0:14:10.679 --> 0:14:14.199 things as little files of zeros and ones. What's happening 0:14:14.240 --> 0:14:18.160 in the human brain is much different. So that's the 0:14:18.200 --> 0:14:22.080 first thing. And there's another reason why human memory is 0:14:22.120 --> 0:14:24.880 not at all like computer memory. And this is because 0:14:25.040 --> 0:14:27.880 over the past several decades, we've come to realize that 0:14:28.080 --> 0:14:33.040 human memory is not one thing, but instead it's made 0:14:33.120 --> 0:14:36.920 up of lots and lots of different subsystems. There are 0:14:37.400 --> 0:14:41.520 many things going on under this one umbrella term that 0:14:41.560 --> 0:14:45.400 we call human memory. So for the next few minutes, 0:14:45.880 --> 0:14:49.240 let's dive into the nuts and bolts of that landscape. 0:14:49.920 --> 0:14:52.840 The first thing to know is that memory is divided 0:14:52.960 --> 0:14:58.480 between short term and long term. So short term memory 0:14:58.960 --> 0:15:02.880 is about in information that decays in a really short 0:15:02.920 --> 0:15:05.240 time if you don't use it, like seconds to minutes. 0:15:05.320 --> 0:15:09.960 This is also called working memory by neuroscientists. The idea 0:15:10.680 --> 0:15:15.320 is that short term memory uses information to address immediate situations. 0:15:15.400 --> 0:15:19.120 So let's say I say, oh, hey, your verification code 0:15:19.200 --> 0:15:22.760 to log into LinkedIn is six one, nine, five, three 0:15:22.760 --> 0:15:26.680 to two. You can remember that for some seconds or minutes, 0:15:26.720 --> 0:15:30.720 but you're probably not going to remember that number next week, 0:15:30.960 --> 0:15:34.600 or probably even in an hour from now. There's no 0:15:34.760 --> 0:15:38.080 strict time limit here, but short term memory typically lasts 0:15:38.760 --> 0:15:42.480 just for the duration of the task at hand. Now, 0:15:42.520 --> 0:15:45.400 there are lots of ways that your brain helps itself 0:15:45.520 --> 0:15:48.480 long with short term memories. Sometimes you do this by 0:15:49.000 --> 0:15:52.400 talking to yourself like six one nine five three two 0:15:52.440 --> 0:15:55.240 six one nine five three two, or sometimes you might 0:15:55.440 --> 0:15:59.120 hold on to information by visualizing something in front of you. 0:15:59.480 --> 0:16:02.760 But how whoever you do it, working memory has a 0:16:03.120 --> 0:16:08.320 limited capacity for holding information. In nineteen fifty six, a 0:16:08.520 --> 0:16:14.040 cognitive psychologist wrote a paper called the Magical Number seven 0:16:14.120 --> 0:16:18.160 plus or minus two. Now, why that weird title is 0:16:18.200 --> 0:16:21.640 because he found that most people can hold about seven 0:16:21.720 --> 0:16:25.040 items of information in short term memory, sometimes a little 0:16:25.080 --> 0:16:27.840 more so as a less It doesn't matter whether that's 0:16:28.200 --> 0:16:32.000 shapes or locations or colors or numbers or whatever. You 0:16:32.040 --> 0:16:36.040 can hold about seven items. And by the way, this 0:16:36.080 --> 0:16:39.520 has been proposed to explain why telephone numbers in most 0:16:39.560 --> 0:16:43.040 countries use seven digits, because back in the day people 0:16:43.080 --> 0:16:46.520 often needed to memorize phone numbers and if they were 0:16:46.720 --> 0:16:49.720 like sixteen digits, no one would be able to do 0:16:49.800 --> 0:16:54.200 it anyway, So that's short term memory. I ask you 0:16:54.240 --> 0:16:56.680 to memorize a combination lock or a log in code 0:16:56.760 --> 0:16:59.600 or a phone number for some task and you say, cool, 0:16:59.680 --> 0:17:02.800 got it. That lives in your short term memory. And 0:17:02.840 --> 0:17:06.840 that memory is built of the activity of millions or 0:17:06.880 --> 0:17:10.840 billions of neurons holding onto a loop of spikes that 0:17:10.920 --> 0:17:15.000 run around and around across giant swaths of territory. The 0:17:15.040 --> 0:17:20.040 activity is kept alive, and that represents the thing you 0:17:20.240 --> 0:17:24.280 are remembering now. Often, something you hold in your short 0:17:24.359 --> 0:17:28.960 term memory tends not to transfer to your long term memory. 0:17:29.000 --> 0:17:31.639 But long term memory is the really interesting thing in 0:17:31.680 --> 0:17:33.520 the brain. So that's what we're going to concentrate on 0:17:33.840 --> 0:17:36.920 from here on out. Because with long term memory, it's 0:17:36.960 --> 0:17:42.440 not about active neurons. Instead, there are changes to the 0:17:42.480 --> 0:17:46.440 physical structure of your brain such that you can pull 0:17:46.520 --> 0:17:49.760 up information that happened to you a long, long time ago. 0:17:49.880 --> 0:17:54.720 You're actually changing the stuff of your brain. I mentioned 0:17:54.760 --> 0:17:58.240 in a previous episode that if a rock hits your 0:17:58.280 --> 0:18:03.520 windshield and leaves a you could say poetically that your 0:18:03.800 --> 0:18:08.400 car remembers the rock even long after the rock is gone. 0:18:08.400 --> 0:18:11.480 There is a physical mark on your car, and that's 0:18:11.520 --> 0:18:14.040 the way to think about it. With the brain, an 0:18:14.040 --> 0:18:18.479 event happens to you, and it leaves a physical mark, 0:18:18.960 --> 0:18:24.200 and that mark can be read out later. So as 0:18:24.200 --> 0:18:28.879 opposed to short term memory, long term memory involves brain 0:18:29.000 --> 0:18:34.000 systems that in code and store and retrieve information over 0:18:34.119 --> 0:18:37.760 long periods of time, anywhere from minutes to a lifetime, 0:18:38.119 --> 0:18:41.800 and the capacity of these systems is much greater than 0:18:41.840 --> 0:18:44.960 short term memory, as we see with someone like Jill 0:18:45.000 --> 0:18:50.040 Price or Shashevski. Now, amazingly, long term memory is not 0:18:50.320 --> 0:18:54.960 one thing. Instead, we divide it into two groups. On 0:18:55.000 --> 0:18:58.520 the one hand, we have implicit memory and on the 0:18:58.600 --> 0:19:03.960 other explicit memory. Now, implicit memory is the stuff that 0:19:04.080 --> 0:19:10.440 you can't express or articulate or consciously recall, like riding 0:19:10.520 --> 0:19:12.959 a bike. You have no idea how you do it, 0:19:13.000 --> 0:19:15.919 but you learned it and you remember it. You just 0:19:15.960 --> 0:19:20.520 can't articulate how you do it. It is implicit. And this 0:19:20.640 --> 0:19:25.600 category of implicit memory also includes unconscious emotional memories, like 0:19:26.080 --> 0:19:30.160 an aversion towards snowboarding years after an accident, even if 0:19:30.160 --> 0:19:33.879 the accident itself is long forgotten. Okay, so all of 0:19:33.920 --> 0:19:38.400 that is implicit memory. On the other hand, explicit memory 0:19:38.520 --> 0:19:42.399 is the type of information that can be consciously recalled 0:19:42.440 --> 0:19:46.920 and described, like facts and events. So think about something 0:19:46.920 --> 0:19:50.640 that happened to you that you can recall and describe, 0:19:50.720 --> 0:19:53.880 like this is where you went to college, or this 0:19:54.000 --> 0:19:56.600 is the name of that great movie you saw, or 0:19:57.119 --> 0:20:00.840 that's what your neighbor said to you last week. Explicit 0:20:00.880 --> 0:20:05.320 memory also includes memories of facts like the fact that 0:20:05.680 --> 0:20:09.760 Amazon dot Com sells books or that penguins can't fly. 0:20:10.880 --> 0:20:14.960 If you can say some recollection or fact out loud, 0:20:15.160 --> 0:20:19.760 then that is explicit memory. Now, I wanted to clearly 0:20:19.880 --> 0:20:23.359 lay out these two sides of implicit memory and explicit 0:20:23.400 --> 0:20:27.440 memory before taking a deeper dive to see why these 0:20:27.480 --> 0:20:47.040 are understood to be separate systems. So now let's return 0:20:47.119 --> 0:20:51.880 for a closer look at this first one, implicit memory. 0:20:52.480 --> 0:20:56.600 So think about memories involved in how to perform skills 0:20:56.680 --> 0:21:01.639 or habits like reading or typing, or whimming, or juggling 0:21:01.840 --> 0:21:06.520 or playing piano. You acquire these kinds of memory slowly 0:21:06.680 --> 0:21:09.879 through a lot of practice and repetition. The way you 0:21:10.000 --> 0:21:13.920 learn to perfect these actions is from your brain changing 0:21:14.040 --> 0:21:18.120 its structure. It is remembering. So even though you might 0:21:18.160 --> 0:21:21.560 not typically think of things like reading and typing and 0:21:21.600 --> 0:21:25.159 walking and so on as memory, they are indeed a 0:21:25.320 --> 0:21:29.000 form of remembrance, just one that you cannot articulate, And 0:21:29.040 --> 0:21:33.240 there are many forms of implicit memory. When you might 0:21:33.320 --> 0:21:36.680 have heard of is called priming. That's what happens when 0:21:36.760 --> 0:21:42.119 some past experience influences your future response. So imagine I 0:21:42.200 --> 0:21:47.840 show you the letters S, blank, P, and I ask 0:21:47.920 --> 0:21:50.159 you to fill in the missing letter. If you have 0:21:50.440 --> 0:21:54.080 just had a shower, you'll probably put an A in 0:21:54.119 --> 0:21:57.520 there to make the word soap. If you've just seen 0:21:57.600 --> 0:22:01.000 a steaming hot bowl at the table, you'll probably put 0:22:01.040 --> 0:22:04.320 the letter you in there to make the word soup. 0:22:05.400 --> 0:22:09.880 Your brain just recently experienced something, and that memory influences 0:22:09.920 --> 0:22:14.000 your behavior in the next moment. Now what's fascinating is 0:22:14.040 --> 0:22:17.600 that people with amnesia, which means they can't write down 0:22:17.720 --> 0:22:23.480 new explicit memories, can nonetheless show these kinds of priming effects. 0:22:24.080 --> 0:22:27.760 So they have no conscious recollection of having seen this 0:22:27.920 --> 0:22:30.919 steaming hot bowl at the table, and they'll deny that 0:22:30.960 --> 0:22:34.800 they ever saw that, but nonetheless they'll choose soup. I'll 0:22:34.800 --> 0:22:36.359 come back to this point in just a bit, but 0:22:36.400 --> 0:22:39.199 first I want to finish painting the different types of 0:22:39.320 --> 0:22:43.880 implicit memory. So another type of implicit memory is one 0:22:43.880 --> 0:22:47.240 you might have heard of called classical conditioning. This was 0:22:47.320 --> 0:22:52.320 first described by the Russian physiologist Ivon Pavlov at the 0:22:52.440 --> 0:22:55.800 end of the nineteenth century, so you generally remember this story. 0:22:56.000 --> 0:23:01.040 Pavlov saw that dogs respond to food by drooling or salivating, 0:23:01.280 --> 0:23:03.720 so he said Okay, I'm going to teach the dog 0:23:03.800 --> 0:23:08.480 through experience that the ringing of a bell predicts the food. 0:23:08.920 --> 0:23:11.560 The bell itself is totally arbitrary, but it comes to 0:23:11.600 --> 0:23:16.200 tell the dog that some food is coming. Pavlov could 0:23:16.240 --> 0:23:18.440 just as easily have used a pat on the head 0:23:18.440 --> 0:23:21.480 to predict the food, or a purple dot flashing three 0:23:21.520 --> 0:23:24.560 times or whatever. It's just some random thing that you 0:23:24.760 --> 0:23:28.400 link to food arriving a moment later. So he teaches 0:23:28.440 --> 0:23:31.879 the dog to associate the ringing of the bell with 0:23:31.960 --> 0:23:35.840 the impending delivery of food, and after setting up the relationship, 0:23:36.080 --> 0:23:39.400 the dogs salivate when he rings the bell, even though 0:23:39.400 --> 0:23:42.959 there's no food there yet, because the bell now is 0:23:43.080 --> 0:23:46.240 as good as seeing the food. And again, this is 0:23:46.560 --> 0:23:49.199 implicit memory because you don't have to be able to 0:23:49.760 --> 0:23:53.760 tell the story of a connection consciously in order to 0:23:53.880 --> 0:23:59.000 form this kind of learning, this memory. So that's classical conditioning, 0:23:59.280 --> 0:24:03.920 and there's another form of implicit memory, which is operant conditioning. 0:24:04.200 --> 0:24:08.280 Here you learn to associate your own behavior like pressing 0:24:08.280 --> 0:24:13.080 a lever, with something rewarding like food, or something aversive 0:24:13.320 --> 0:24:17.119 like an electric shock, and you gradually do more or 0:24:17.200 --> 0:24:22.880 less of that behavior as a result. Of remembering the consequences. Now, 0:24:22.920 --> 0:24:24.639 I'm not going to go into the details of the 0:24:24.680 --> 0:24:27.800 brain regions that are involved in these different forms of 0:24:27.840 --> 0:24:31.000 implicit learning. If you're curious, to go to my textbook 0:24:31.040 --> 0:24:34.520 called Brain and Behavior. But I'll just mention that learning 0:24:34.840 --> 0:24:38.119 any of these types of responses depends on some brain 0:24:38.160 --> 0:24:41.919 areas like the amygdala and cerebellum, and some structures in 0:24:41.960 --> 0:24:44.439 the brain stem. The key thing I want you to 0:24:44.480 --> 0:24:47.639 appreciate is that all of these forms of memory I 0:24:47.800 --> 0:24:52.640 just described are forms in which your brain changes itself. 0:24:52.680 --> 0:24:57.879 It changes its detailed internal structure in response to things 0:24:57.920 --> 0:25:01.480 that it has experienced so that it be behaves differently 0:25:01.600 --> 0:25:04.280 in the future. And all of these things I just 0:25:04.320 --> 0:25:08.960 told you about, that's just implicit memory. Now let's switch 0:25:09.040 --> 0:25:12.280 back again to explicit memory, the things that you can 0:25:12.400 --> 0:25:18.040 consciously articulate, like events and facts. Even explicit memory can 0:25:18.080 --> 0:25:22.240 be divided into two different forms. The first is called 0:25:22.680 --> 0:25:27.840 episodic memory. This is memory for past events or episodes 0:25:28.080 --> 0:25:31.440 that you've experienced, like a birthday party or a surfing 0:25:31.520 --> 0:25:35.720 trip you took. These are called autobiographical memories. And the 0:25:35.800 --> 0:25:40.800 key point is that, unlike implicit memories, these Episodic memories 0:25:40.840 --> 0:25:44.040 can be consciously recalled and described. They can have a 0:25:44.520 --> 0:25:47.879 sort of cinematic quality to them, and they usually have 0:25:47.920 --> 0:25:51.719 a particular context, like in the living room on New 0:25:51.800 --> 0:25:55.920 Year's Eve. You're also generally able to pick out specific 0:25:56.040 --> 0:25:58.840 objects and features and the surroundings, like oh, yeah, the 0:25:58.920 --> 0:26:01.439 sofa was over here my left, and there was a 0:26:01.440 --> 0:26:04.240 television set over there, and my uncle was sitting over 0:26:04.280 --> 0:26:08.080 here on my right, And you can recall very particular 0:26:08.560 --> 0:26:13.040 sequences of actions like oh, we were wearing party hats 0:26:13.080 --> 0:26:16.399 and the wine was poured and then it clinked glasses 0:26:16.440 --> 0:26:22.040 with everyone. So those are episodic memories. Now there's another 0:26:22.200 --> 0:26:27.399 type of explicit memory, not episodic, but semantic memories, and 0:26:27.440 --> 0:26:30.600 these are things that your brain remembers about the outside world, 0:26:30.720 --> 0:26:34.439 like the fact that a sheep has four legs and 0:26:34.480 --> 0:26:37.679 it makes a noise that sounds like bah, and it 0:26:37.760 --> 0:26:42.600 has a wooly coat. Semantic properties are more general than 0:26:42.800 --> 0:26:46.200 single events that you saw or experienced, and because they're 0:26:46.400 --> 0:26:50.400 independent of any one particular kind of sensory input, they're 0:26:50.440 --> 0:26:55.879 generally useful for organizing the world into categories of related 0:26:56.040 --> 0:26:58.879 things Like these are vegetables and these are animals, and 0:26:58.880 --> 0:27:02.880 these are tools, these vehicles and so on. So that's 0:27:02.880 --> 0:27:06.760 semantic memory. Now, I just told you about a whole 0:27:06.800 --> 0:27:12.160 bunch of different subtypes of memories. But why do neuroscientists 0:27:12.600 --> 0:27:16.879 divide memory into all these different flavors. Well, it's because 0:27:16.920 --> 0:27:21.399 clinicians have observed for many, many decades that sometimes a 0:27:21.520 --> 0:27:25.160 person can damage one part of their brain and lose 0:27:25.240 --> 0:27:29.840 a very specific sort of memory while not losing other sorts. So, 0:27:30.000 --> 0:27:34.280 for example, you might see a patient with severe amnesia 0:27:34.400 --> 0:27:39.040 for personal experiences. She can remember essentially nothing about her 0:27:39.080 --> 0:27:43.399 personal life, but she retains a good general knowledge for 0:27:43.640 --> 0:27:47.280 facts that she learned before her brain injury. So that's 0:27:47.320 --> 0:27:51.440 how we know that episodic memory is different from semantic 0:27:51.480 --> 0:27:55.239 It's actually underpinned by different structures in the brain, and 0:27:55.440 --> 0:27:57.800 by the way, you see this often in forms of 0:27:57.800 --> 0:28:03.280 dementia like early Alzheimer's disease, where a person's episodic memory 0:28:03.480 --> 0:28:07.440 suffers dramatically. They can't remember the details of their own life, 0:28:07.760 --> 0:28:12.159 but this is long before their semantic knowledge begins to fail. 0:28:13.000 --> 0:28:16.320 And on the flip side, in a rarer illness known 0:28:16.320 --> 0:28:22.200 as semantic dementia, episodic memory is preserved, while even basic 0:28:22.320 --> 0:28:26.119 forms of semantic knowledge are lost, like what a sheep 0:28:26.280 --> 0:28:29.280 is or what sound it makes. I won't go into 0:28:29.280 --> 0:28:31.240 details here, but if you want to look up more, 0:28:31.520 --> 0:28:36.600 episodic memory generally depends on the medial temporal lobes, while 0:28:36.680 --> 0:28:42.760 semantic memory involves the anterior temporal lobes. Okay, so now 0:28:43.000 --> 0:28:46.560 how did people start to figure out what brain regions 0:28:46.560 --> 0:28:49.640 were involved in what functions. Well, like many things, the 0:28:49.720 --> 0:28:52.360 answer is that this generally happens when a person gets 0:28:52.520 --> 0:28:56.360 damage to a very specific part of the brain. So 0:28:56.480 --> 0:29:00.200 let's zoom in on a particular example. In nineteen fifty, 0:29:00.640 --> 0:29:04.400 a particular area of the brain called the hippocampus came 0:29:04.440 --> 0:29:08.320 to the center stage in the neuroscience of memory because 0:29:08.360 --> 0:29:11.240 there was a man named Henry Malaison, who, by the way, 0:29:11.360 --> 0:29:15.440 was known for decades in the literature as HM because 0:29:15.440 --> 0:29:18.440 the patient's real name is never used while he's still alive. 0:29:19.000 --> 0:29:22.560 So anyway, Henry had suffered a head injury at the 0:29:22.600 --> 0:29:26.760 age of nine, and after that he had epileptic seizures 0:29:27.320 --> 0:29:31.000 and these got more frequent and severe, and the doctors 0:29:31.080 --> 0:29:34.720 tried to control the seizures, but eventually they couldn't control 0:29:34.760 --> 0:29:37.719 them anymore. So at the age of twenty seven, Henry 0:29:37.800 --> 0:29:43.240 went in for a neurosurgery because of where the seizures 0:29:43.280 --> 0:29:48.320 were originating. The surgeon removed the hippocampus on both sides 0:29:48.360 --> 0:29:50.640 of his brain, on the right and the left, as 0:29:50.680 --> 0:29:54.000 well as some of the regions that surrounded the hippocampus. Now, 0:29:54.720 --> 0:30:00.520 after the surgery, the seizures were all better, but Henry 0:30:00.560 --> 0:30:03.280 became one of the most famous cases in the medical 0:30:03.320 --> 0:30:07.600 literature because while he was otherwise fine, he had a 0:30:07.920 --> 0:30:12.680 severe amnesia that means he couldn't remember things, and specifically, 0:30:12.720 --> 0:30:16.880 he had an antaro grade amnesia, which means he couldn't 0:30:16.920 --> 0:30:21.320 form new episodic memories. And so that meant he would 0:30:21.360 --> 0:30:23.760 function just fine if you were talking with him, but 0:30:23.880 --> 0:30:27.800 after say fifteen minutes, he couldn't recall anything about the 0:30:27.840 --> 0:30:34.160 conversation because he wasn't converting that into new long term memory. 0:30:34.680 --> 0:30:38.000 So just imagine being one of the people working with him. 0:30:38.040 --> 0:30:41.520 You walk into the room and you introduce yourself, and 0:30:41.560 --> 0:30:45.000 you bring a giant Saint Bernard dog with you. Then 0:30:45.080 --> 0:30:47.320 the dog leaves and you chat with him for about 0:30:47.360 --> 0:30:50.280 five minutes and you ask him, hey, when I came 0:30:50.280 --> 0:30:54.120 in here, did I have anything with me? And he says, yeah, 0:30:54.160 --> 0:30:57.120 you had a giant Saint Bernard dog with you. So 0:30:57.200 --> 0:30:59.720 you continue to have a nice conversation about things. He's 0:30:59.720 --> 0:31:03.360 a small guy. Now you distract him for about another 0:31:03.400 --> 0:31:06.520 fifteen minutes and you ask him, hey, when I came 0:31:06.560 --> 0:31:09.120 in here, did I have anything with me? And he 0:31:09.240 --> 0:31:12.120 draws a total blank. He says, I don't think So 0:31:13.200 --> 0:31:15.920 now you leave the room and you come back twenty 0:31:15.960 --> 0:31:18.920 minutes later and you say, hey, when I walked in 0:31:18.960 --> 0:31:22.280 the room before, did I have anything with me? And 0:31:22.320 --> 0:31:26.600 what does he say? He says, who are you? As 0:31:26.640 --> 0:31:29.080 far as he knows, he has never seen you before 0:31:29.120 --> 0:31:32.640 in his life. So the researchers who studied Henry for 0:31:32.760 --> 0:31:37.239 years had to reintroduce themselves each time they walked in 0:31:37.240 --> 0:31:39.479 the room, even if they'd only left the room briefly. 0:31:40.120 --> 0:31:44.160 So the issue was that Henry could not form new 0:31:44.440 --> 0:31:50.120 explicit memories. But here's the interesting part. His implicit memory 0:31:50.440 --> 0:31:55.280 was fine. He could practice and learn new tasks like 0:31:55.800 --> 0:31:59.760 tracing a five pointed star viewed through a mirror, but 0:31:59.880 --> 0:32:03.200 he had no recollection of the practice in learning, and 0:32:03.560 --> 0:32:06.880 he expressed surprise at how well, he could perform this 0:32:07.040 --> 0:32:09.640 task that, as far as he knew, he had never 0:32:09.840 --> 0:32:13.840 seen before. Also, his short term memory was fine, well 0:32:13.880 --> 0:32:17.280 within the range of seven plus or minus two items. 0:32:18.320 --> 0:32:22.680 And so the tragic outcome from Henry's surgery helped to 0:32:22.800 --> 0:32:26.719 define the taxonomy of memory systems that we know today. 0:32:29.000 --> 0:32:31.440 And by the way, we've learned these same lessons from 0:32:31.600 --> 0:32:35.840 hundreds of other patients with amnesia since then. For example, 0:32:35.920 --> 0:32:40.240 somebody who can't form any new explicit memories can nonetheless 0:32:40.320 --> 0:32:44.360 learn an implicit task like the video game Tetris, and 0:32:44.400 --> 0:32:46.800 they can get better and better at the game. But 0:32:46.920 --> 0:32:49.480 each time you place them in front of the game, 0:32:50.200 --> 0:32:54.280 they claim genuinely that they've never seen this before. And 0:32:54.320 --> 0:32:57.560 what's fascinating is that if you wake them up when 0:32:57.560 --> 0:32:59.640 they're sleeping, if you catch them in the middle of 0:32:59.640 --> 0:33:02.760 a dream, they'll say that they were just dreaming about 0:33:03.120 --> 0:33:07.920 colorful falling blocks, but they have no idea why, because 0:33:07.960 --> 0:33:12.720 they have no conscious memory of ever having seen that before. 0:33:13.840 --> 0:33:16.600 By the way, if you saw the movie Memento, you 0:33:16.680 --> 0:33:19.440 know that was about a man who had lost his 0:33:19.680 --> 0:33:23.360 explicit memory. He had had a head injury that gave 0:33:23.440 --> 0:33:27.240 him amnesia. In his case, both in the forward and 0:33:27.440 --> 0:33:31.120 backward direction and tarot grade and retrograde, and so the 0:33:31.160 --> 0:33:33.600 only way that he could keep track of his goals 0:33:33.640 --> 0:33:39.160 through time was to tattoo information onto his skin. And 0:33:39.200 --> 0:33:42.160 I'll be talking more about amnesia in a future episode, 0:33:42.480 --> 0:33:44.320 but all I want to say for now is that 0:33:44.520 --> 0:33:47.640 for most of us, we are lucky enough to tattoo 0:33:47.680 --> 0:33:53.040 the information directly into our vast neural forests, and we 0:33:53.080 --> 0:33:56.040 can do this again and again throughout every day of 0:33:56.040 --> 0:34:00.640 our lives, and we never run out of room. Now, 0:34:00.680 --> 0:34:04.080 we know from Henry and many other patients that normal 0:34:04.240 --> 0:34:09.680 autobiographical memory depends on the integrity of the hippocampus. But 0:34:09.760 --> 0:34:13.120 I mentioned in episode number one of this podcast that 0:34:13.200 --> 0:34:17.880 there's another important player in the medial temporal lob memory system, 0:34:17.960 --> 0:34:22.280 and that's the amygdala, which is an almond sized structure 0:34:22.320 --> 0:34:24.880 that's just in front of the hippocampus, and this is 0:34:24.920 --> 0:34:31.360 involved in emotional memory. The amygdala assigns value positive or 0:34:31.440 --> 0:34:35.160 negative to things that it sees or hears or smells 0:34:35.200 --> 0:34:39.680 based on past experiences. And because it's so well connected, 0:34:40.200 --> 0:34:44.200 it can coordinate all the different prongs of an emotional 0:34:44.280 --> 0:34:49.200 response to something. It can coordinate the autonomic responses like 0:34:49.280 --> 0:34:52.920 an increased heart rate, and the endocrine responses like the 0:34:53.000 --> 0:34:59.120 secretion of stress hormones, and behavioral responses like fear and avoidance. 0:35:00.080 --> 0:35:02.799 So it turns out that while normal memories are just 0:35:02.880 --> 0:35:07.320 taken care of by the hippocampus, emotional situations like something 0:35:07.440 --> 0:35:11.600 very stressful or dangerous, those kick the amigdala into gear 0:35:12.160 --> 0:35:15.000 and memories get written down on what is essentially a 0:35:15.120 --> 0:35:20.319 secondary memory pathway. And that's why emotional events are more 0:35:20.520 --> 0:35:23.759 likely to be remembered, because in a sense, those are 0:35:23.800 --> 0:35:29.000 the most important memories. When something really emotionally important happens, 0:35:29.120 --> 0:35:32.200 that's what you want to write down. That's why memory 0:35:32.360 --> 0:35:36.600 exists to keep a record of the really important stuff. 0:35:37.560 --> 0:35:40.040 By the way, I spoke in the first episode about 0:35:40.480 --> 0:35:44.560 time seeming to run in slow motion when you're in danger, 0:35:44.719 --> 0:35:48.040 when you're in fear for your life, And if you're interested, 0:35:48.360 --> 0:35:50.600 you can go back and listen to the experiments that 0:35:50.680 --> 0:35:53.160 my lab did. But the punchline is that when you're 0:35:53.160 --> 0:35:58.320 in a very stressful situation, your brain writes down denser 0:35:58.520 --> 0:36:01.239 memories than normal, and so when you think back on 0:36:01.280 --> 0:36:05.640 what just happened, your brain pulls up much more detail 0:36:05.800 --> 0:36:09.680 than you would normally have, and you interpret that as 0:36:10.280 --> 0:36:13.520 that must have taken a long time because I hit 0:36:13.600 --> 0:36:16.320 the brakes, but because of all the ice, I couldn't 0:36:16.320 --> 0:36:19.320 get traction, and so my car slid into the intersection. 0:36:19.400 --> 0:36:21.839 And I saw the blue Toyota coming and I saw 0:36:21.880 --> 0:36:24.120 the expression on the other driver, and she hit the 0:36:24.200 --> 0:36:26.120 front of my car, and I saw the hood crumple, 0:36:26.160 --> 0:36:28.800 and the glass spider web, and the rear view mirror 0:36:28.880 --> 0:36:32.120 fall off, and so on and so on, and so 0:36:32.160 --> 0:36:34.440 when you think what just happened, what just happened, your 0:36:34.440 --> 0:36:38.479 brain assumes, Wow, that must have taken a long time. 0:36:38.520 --> 0:36:41.120 That must have taken many many seconds for all of 0:36:41.160 --> 0:36:43.880 that to happen in order for me to see that 0:36:43.960 --> 0:36:49.000 much detail. But in fact, time does not actually run 0:36:49.040 --> 0:36:53.440 in slow motion for you, as we demonstrated with experiments. Instead, 0:36:53.440 --> 0:36:57.320 it's a trick of memory. You just have more memory 0:36:57.400 --> 0:37:01.520 that you're drawing up. As I mentioned in the first episode. 0:37:01.800 --> 0:37:04.719 You can convince yourself of this because if you had 0:37:04.760 --> 0:37:07.920 a passenger on the car seat next to you, you 0:37:07.920 --> 0:37:13.640 don't actually remember his voice as saying, which would have 0:37:13.719 --> 0:37:16.160 to be the case if time were actually slowed down. 0:37:16.520 --> 0:37:19.920 So it's a trick of memory, and the retrospective illusion 0:37:20.080 --> 0:37:24.520 happens because the amygdala has come online because something very 0:37:24.600 --> 0:37:30.360 salient is happening, and it wrote down denser memories than normal. 0:37:31.280 --> 0:37:33.719 So that's what I wanted to say about the amygdala. 0:37:33.880 --> 0:37:36.600 But now we're going to return to the hippocampus, which 0:37:36.719 --> 0:37:40.759 underlies most normal day to day memory. And there was 0:37:40.800 --> 0:37:44.279 a very interesting discovery which won a Nobel Prize some 0:37:44.400 --> 0:37:47.600 years ago, and that is that the hippocampus is involved 0:37:47.719 --> 0:37:53.280 in encoding your location, your position in space. In other words, 0:37:53.719 --> 0:37:57.120 it's involved in saying, I'm in this spot, and the 0:37:57.200 --> 0:38:00.000 room I'm in connects to that hallway which leads to 0:38:00.280 --> 0:38:02.880 the door of that lobby, which has an elevator to 0:38:02.880 --> 0:38:05.960 go up to that other room. Your sense of where 0:38:06.000 --> 0:38:11.479 you are relies on your hippocampus. Your hipp campus, in fact, 0:38:11.520 --> 0:38:16.000 has these specialized neurons that we call place cells, and 0:38:16.080 --> 0:38:19.520 these neurons become active they fire off spikes only when 0:38:19.520 --> 0:38:23.160 you're in a particular spot in your environment. And there 0:38:23.200 --> 0:38:26.960 are other neurons known as grid cells, and these have 0:38:27.480 --> 0:38:31.640 multiple receptive fields that are arranged in a grid pattern 0:38:31.719 --> 0:38:35.360 that covers your local environment anyway. So all these cells 0:38:35.400 --> 0:38:38.839 work together to give you a very precise sense of 0:38:38.880 --> 0:38:42.880 your position in a room. And the hippocampus is not 0:38:42.920 --> 0:38:45.719 only about your current position, but it's also crucial for 0:38:46.360 --> 0:38:51.239 spatial memory. For example, imagine that you're in a labyrinth 0:38:51.239 --> 0:38:53.120 and you have to go down this hallway and come 0:38:53.120 --> 0:38:55.080 back to the center, and then you have to remember 0:38:55.120 --> 0:38:57.480 that you've been there, and then you go down a 0:38:57.480 --> 0:38:59.080 different one and you come back, and then you have 0:38:59.160 --> 0:39:01.000 to figure out a different one and go back and 0:39:01.040 --> 0:39:03.080 so on. It's really easy for you to do this 0:39:03.280 --> 0:39:06.640 because even in a brand new environment, you have a 0:39:06.840 --> 0:39:10.680 spatial memory, as in, I've already been down that hallway. 0:39:11.719 --> 0:39:15.000 But if you have damage to your hippocampus on both sides, 0:39:15.280 --> 0:39:19.080 then you can't do that task. You need your hippocampi 0:39:19.400 --> 0:39:25.480 for spatial memory. Now, interestingly, you find hippocampus like structures 0:39:25.600 --> 0:39:28.800 throughout the animal kingdom, not only in mammals, but also 0:39:28.880 --> 0:39:32.319 in birds and in goldfish, and in all cases this 0:39:32.400 --> 0:39:36.640 is involved in spatial memory. And across the kingdom you 0:39:36.719 --> 0:39:40.520 see that the size of the hippocampus is related to 0:39:40.640 --> 0:39:44.320 the demands of the territory mapping and the spatial memory. 0:39:44.760 --> 0:39:49.480 For example, squirrels spend the autumn months hiding literally thousands 0:39:49.560 --> 0:39:53.319 of seeds and nuts in different locations throughout their territory 0:39:53.719 --> 0:39:56.720 so they can have a steady food supply through the winter. 0:39:57.360 --> 0:40:00.520 And during this period the volume of their hippocampus increases 0:40:00.560 --> 0:40:04.560 by fifteen percent. Or as another example, some species of 0:40:04.760 --> 0:40:09.719 birds hide their food and these species have larger hippocampi 0:40:09.920 --> 0:40:12.799 than the birds that don't do this, and when they're 0:40:12.800 --> 0:40:17.760 doing their food hiding, that stimulates growth of their hippocampus. 0:40:18.160 --> 0:40:20.719 And this is what you find in humans too. Our 0:40:20.800 --> 0:40:25.759 hippocampi give us map like spatial codes, and these regions 0:40:25.920 --> 0:40:31.279 guide navigations. So consider London taxi drivers. I don't know 0:40:31.320 --> 0:40:35.839 if they do this anymore, but pre GPS, they had 0:40:35.880 --> 0:40:40.239 to memorize an unbelievably detailed map of London with thousands 0:40:40.280 --> 0:40:45.360 of destinations encompassing something like twenty five thousand streets, and 0:40:45.400 --> 0:40:48.359 they had to be able to verbally recite the most 0:40:48.400 --> 0:40:52.760 efficient roots between any two points, and just working from memory, 0:40:52.920 --> 0:40:55.560 say points of interest along the way, like the names 0:40:55.560 --> 0:40:58.439 of all the theaters that they'd pass in sequential order. 0:40:59.239 --> 0:41:02.000 And when neu oh imaging studies were done on these 0:41:02.080 --> 0:41:06.040 cab drivers, it could be seen that their hippocampi had 0:41:06.160 --> 0:41:11.640 grown compared with novice taxi drivers. And interestingly, this doesn't 0:41:11.680 --> 0:41:16.040 happen for physicians, who have to master a similarly large 0:41:16.080 --> 0:41:36.279 body of knowledge. But it's not spatial. Okay, So I 0:41:36.360 --> 0:41:38.960 just told you a lot about how the hippocampus is 0:41:39.000 --> 0:41:43.439 involved in understanding this spatial layout of things. But why 0:41:43.480 --> 0:41:46.719 do I mention the hippocampus encoding space in an episode 0:41:46.840 --> 0:41:52.200 on memory? While first, the world is full of spatial things, 0:41:52.280 --> 0:41:56.080 and traditionally our memories had to care about that, just 0:41:56.120 --> 0:41:59.360 like the squirrels remembering where they had buried their seeds 0:41:59.400 --> 0:42:02.560 and nuts. Also, you may have noticed that we are 0:42:02.600 --> 0:42:07.040 generally very good at remembering spatial information, like all the 0:42:07.080 --> 0:42:10.680 important rooms in a very large building. And many people, 0:42:10.680 --> 0:42:13.880 when they're trying to memorize something like a list of items, 0:42:14.200 --> 0:42:18.360 will use what is called a memory palace, which is 0:42:18.360 --> 0:42:23.600 where you associate these different items with different locations. So, 0:42:23.719 --> 0:42:26.040 for example, let's say I had to memorize a long 0:42:26.160 --> 0:42:35.000 list of words like apple, baby, clock, dennis, exhibition, flaggirl, horse, ice, jester, ladder, machine, noos, ocean, pigeon, radio, sheep, theater, 0:42:35.560 --> 0:42:37.640 and so on. What I might do is I might 0:42:37.960 --> 0:42:42.720 picture my house and I visualize the first item apple 0:42:42.920 --> 0:42:45.839 at the front door, and then I imagine walking in 0:42:45.880 --> 0:42:49.240 and I visualize the second item, let's say a baby 0:42:49.320 --> 0:42:51.919 in this case in a crib just inside the door. 0:42:52.200 --> 0:42:55.280 And then just past the baby, I picture a giant 0:42:55.320 --> 0:42:58.920 pendulum clock on the wall, and on the couch over there, 0:42:59.000 --> 0:43:02.440 I visualize a dentist doing his work on a patient, 0:43:02.760 --> 0:43:07.120 and so on and so on. I leverage location. I 0:43:07.160 --> 0:43:11.480 take advantage of these cells in my hippocampus that care 0:43:11.520 --> 0:43:16.040 about place to tie other information to them. And when 0:43:16.080 --> 0:43:18.800 I need to remember this list much later, I simply 0:43:18.840 --> 0:43:22.840 imagine myself walking through the house, and I can recall 0:43:22.920 --> 0:43:28.080 an enormously long list of arbitrary objects this way. This is, 0:43:28.080 --> 0:43:32.400 in fact, the oldest memory technique called a mnemonic device, 0:43:32.800 --> 0:43:36.560 that is on record. The ancient Greek and Roman bards 0:43:37.120 --> 0:43:40.400 used to tell their epic tales this way. This was 0:43:40.440 --> 0:43:43.359 before the invention of the printing press, so they had 0:43:43.400 --> 0:43:46.600 to memorize these things. But happily this was after the 0:43:46.640 --> 0:43:53.000 evolution of the hippocampus. Now, although some people purposely leverage 0:43:53.040 --> 0:43:57.000 these kind of techniques to memorize for a small fraction 0:43:57.120 --> 0:44:02.000 of the population. This happens naturally. In episode four, I 0:44:02.040 --> 0:44:05.640 talked about synesthesia, and one of the most common forms 0:44:05.680 --> 0:44:11.840 of synesthesia involves imagining things with spatial locations. For example, 0:44:12.200 --> 0:44:15.680 you see the days of the week in a circle 0:44:15.760 --> 0:44:18.359 around you physically, like Monday is over here, and then 0:44:18.360 --> 0:44:20.560 tuesdays there, and then Wednesday is up here, and then 0:44:20.600 --> 0:44:23.359 Thursday's over there, and Friday off to the side, and 0:44:23.440 --> 0:44:27.840 so on. And for each person. This is idiosyncratic, meaning 0:44:27.880 --> 0:44:31.000 that it's a different pattern for everyone, and a cinnasthete 0:44:31.080 --> 0:44:34.640 might see the months of the year in spatial locations, 0:44:34.920 --> 0:44:39.560 or the years historical in future, like nineteen seventy one 0:44:39.640 --> 0:44:41.200 is over here, and then it goes up and up 0:44:41.239 --> 0:44:43.319 through nineteen ninety, and then it's flat over to two 0:44:43.320 --> 0:44:45.360 thousand and five, and then it goes a little up 0:44:45.440 --> 0:44:47.880 and down through twenty thirteen, and then cuts to the 0:44:47.960 --> 0:44:50.799 left at twenty twenty, and then it suddenly dives down 0:44:50.840 --> 0:44:53.080 and then curves around, and then the future goes off 0:44:53.120 --> 0:44:55.920 behind you. And it's different for each cynisthete, and I 0:44:56.000 --> 0:44:59.200 gave many more examples in episode four, but the point 0:44:59.239 --> 0:45:01.840 I want to make here is that This form of 0:45:01.920 --> 0:45:06.719 synesthesia is closely tied to memory because all of the 0:45:06.760 --> 0:45:10.120 things that I just mentioned, like weekdays and months and years, 0:45:10.600 --> 0:45:13.560 these are all sequences that you have to learn, you 0:45:13.600 --> 0:45:18.040 have to memorize, and so in many people these sequences 0:45:18.480 --> 0:45:24.760 get tied irreversibly to spatial location. It helps the brain 0:45:24.800 --> 0:45:30.719 to remember something by tagging it with a place. So 0:45:30.760 --> 0:45:33.880 some people suggest that the way to understand the hippocampus 0:45:34.080 --> 0:45:37.719 is as a cognitive map. The idea is that the 0:45:37.800 --> 0:45:43.800 hippocampus originally created and stored territory maps for orientation and 0:45:43.920 --> 0:45:48.960 navigation and finding resources, but eventually the system adapted to 0:45:49.040 --> 0:45:55.520 create and store episodic memories. Also, because events usually have 0:45:55.560 --> 0:46:01.120 a particular setting in space, and because movement cross locations 0:46:01.160 --> 0:46:05.240 also involves movement in time, the hippocampus would be naturally 0:46:05.280 --> 0:46:09.839 well suited to capture sequences of events in time, in 0:46:09.840 --> 0:46:13.279 other words, episodes. And I want to mention that there's 0:46:13.360 --> 0:46:17.680 a cousin theory about the hippocampus that proposes what it's 0:46:17.719 --> 0:46:23.520 doing is it's storing the associations between elements of the events, 0:46:23.600 --> 0:46:26.520 just like what we talked about with the spatial relations 0:46:26.560 --> 0:46:33.200 between objects. So the hippocampus stores the temporal relationship between 0:46:33.239 --> 0:46:36.520 events like this happened after this, or it happened close 0:46:36.560 --> 0:46:39.520 in time or far in time, and it stores the 0:46:39.560 --> 0:46:46.560 relationships between pieces of information. It stores the relationships rather 0:46:46.640 --> 0:46:50.520 than the specific items. So this gives us a way 0:46:50.560 --> 0:46:55.640 to see how the original spatial function of the hippocampus 0:46:56.000 --> 0:47:02.920 could have evolved into episodic memory. Okay, so now I 0:47:02.960 --> 0:47:05.600 want to turn to what's going on at the very 0:47:05.800 --> 0:47:09.520 tiny level to ask how does activity in the brain 0:47:09.840 --> 0:47:14.520 cause lasting changes such that we have learning and memory. 0:47:15.640 --> 0:47:19.120 So almost all theories of brain plasticity, that's what we 0:47:19.200 --> 0:47:23.239 call it, when the brain reorganizes itself. These use the 0:47:23.360 --> 0:47:27.960 idea that the strength of connections between cells, the strength 0:47:28.000 --> 0:47:33.120 of the synapses, can be modified by previous activity. Okay, 0:47:33.120 --> 0:47:36.160 but how does that work? Well, First of all, if 0:47:36.160 --> 0:47:40.200 you imagine one hundred years ago, microscopes didn't have the 0:47:40.280 --> 0:47:44.640 magnification power to actually see neurons in their inner connections. 0:47:45.040 --> 0:47:50.160 So a century ago, scientists believed that brain tissue was 0:47:50.239 --> 0:47:54.759 a continuous network like the blood vessels. Blood vessels are 0:47:54.800 --> 0:47:58.080 a system of tubes where you have stuff running along it, 0:47:58.320 --> 0:48:01.640 and people assumed that was what was happening in the brain. 0:48:02.560 --> 0:48:07.160 But this idea was overturned by a Spanish neuroscientist named 0:48:07.200 --> 0:48:10.520 Santiago Ramonicjl. He spent a lot of time at his 0:48:10.719 --> 0:48:14.640 microscope trying to look at brain tissue, and because he 0:48:14.680 --> 0:48:17.960 was a photographer also, he had all these chemicals in 0:48:18.000 --> 0:48:22.239 his workshop, these stains that he put on to thin 0:48:22.440 --> 0:48:25.440 slices of brain tissue to see if that would allow 0:48:25.520 --> 0:48:29.120 him to see anything better. And some of his stains 0:48:29.600 --> 0:48:32.399 leaped into a few of the neurons and turned them 0:48:32.440 --> 0:48:36.280 all black, and he was able to visualize them under 0:48:36.320 --> 0:48:40.240 the microscope that way. So he put forward this idea, 0:48:40.280 --> 0:48:43.680 which turned out to be right, that the brain, instead 0:48:43.719 --> 0:48:47.360 of being a subway system of tubes, was a massive 0:48:47.520 --> 0:48:52.799 collection of billions of discrete cells. He called this the 0:48:53.320 --> 0:48:57.640 neuron doctrine, and this was a massively important step in neuroscience, 0:48:57.640 --> 0:49:00.759 and it eventually won him the Nobel Prize. Now, this 0:49:00.920 --> 0:49:03.960 idea that the brain is made up of lots of 0:49:04.160 --> 0:49:08.880 individual cells, this ushered in an important new concept because 0:49:08.880 --> 0:49:13.399 people began to realize that these separate cells they have 0:49:13.440 --> 0:49:17.080 to influence each other through these little connections between them, 0:49:17.280 --> 0:49:21.720 these synapses. And Ramonic Cahol was the first to suggest 0:49:21.800 --> 0:49:28.960 that learning and memory might occur by changing these synapses, 0:49:29.760 --> 0:49:34.000 so several decades later, in nineteen forty nine, the neuroscientist 0:49:34.080 --> 0:49:39.360 Donald Hebb made a specific proposal for how synapses should 0:49:39.440 --> 0:49:43.920 adjust to underlie memory. He said that if one cell 0:49:44.440 --> 0:49:49.480 consistently participates in exciting another cell, then the connection between 0:49:49.520 --> 0:49:53.800 them is strengthened, and if the first cell consistently fails 0:49:53.840 --> 0:49:56.879 to excite the second cell, then the connection between them 0:49:57.080 --> 0:50:02.080 is weakened. So this rule is often as cells that 0:50:02.280 --> 0:50:07.040 fire together wire together, and most models of memory formation 0:50:07.360 --> 0:50:12.520 employ this kind of rule. Now, when Hebb proposed this hypothesis, 0:50:12.560 --> 0:50:16.440 there was no experimental evidence to support it, and it 0:50:16.480 --> 0:50:21.279 took until nineteen seventy three for two researchers to discover that, 0:50:21.719 --> 0:50:25.040 in fact, that seems to be what happens between neurons. 0:50:25.160 --> 0:50:28.560 So you stimulate some neuron over here, and you measure 0:50:28.600 --> 0:50:31.759 the very tiny electrical response that it causes in this 0:50:31.880 --> 0:50:35.640 neuron over here. Then you blast the first neural with 0:50:35.680 --> 0:50:39.239 a bunch of electrical stimulation for thirty seconds, and then 0:50:39.280 --> 0:50:41.839 you try that first experiment again, where you give this 0:50:41.920 --> 0:50:44.120 guy a little zapp and you look at the electrical 0:50:44.160 --> 0:50:47.239 signal that it causes in the second neuron, and what 0:50:47.280 --> 0:50:50.520 you find is that you now have a larger signal. 0:50:50.600 --> 0:50:57.719 The synapse has strengthened, and that strengthening lasts. The connection 0:50:57.880 --> 0:51:01.120 strength between these two neurons has changed, and that change 0:51:01.200 --> 0:51:05.600 is held on to through time. So the synaptic connection 0:51:05.760 --> 0:51:10.600 can be modified as a result of the cell's history 0:51:10.640 --> 0:51:15.759 of activity. Now, a typical neuron has ten thousand connections, 0:51:15.920 --> 0:51:20.120 and what's fascinating is that each individual connection can strengthen 0:51:20.239 --> 0:51:24.160 or weaken according to its history. So the way that 0:51:24.239 --> 0:51:28.200 activity flows through a network of billions of neurons can 0:51:28.280 --> 0:51:31.640 be completely changed. By dialing the strength of this connection here, 0:51:31.640 --> 0:51:33.880 in that connection there, and so on for all the 0:51:33.920 --> 0:51:38.680 connections across the brain. By dialing the strengths up and 0:51:38.760 --> 0:51:42.520 down and holding on to those, you can store information 0:51:42.760 --> 0:51:45.200 in the system. And just to give you a sense 0:51:45.360 --> 0:51:49.840 of the size of the parameter space here, the large 0:51:49.920 --> 0:51:55.720 language model GPT four has one point seventy six trillion connections, 0:51:56.400 --> 0:52:00.759 but the human brain has about one hundred times that. 0:52:03.840 --> 0:52:06.000 So there are a lot of parameters that can be 0:52:06.040 --> 0:52:09.960 tweaked in the brain to store information. Now, by the 0:52:10.000 --> 0:52:13.320 way the birth of artificial neural networks like GPT, for 0:52:14.160 --> 0:52:17.440 is rooted in these discoveries from the brain from the 0:52:17.480 --> 0:52:20.520 past century. You have a bunch of units, and you 0:52:20.560 --> 0:52:23.840 have connections between these units, and you change the strength 0:52:23.840 --> 0:52:28.480 of those connections. That's how large language models work because 0:52:28.520 --> 0:52:32.400 making small, subtle changes in the way that units communicate 0:52:32.480 --> 0:52:36.919 a network can change the entire network's output behavior. By 0:52:37.120 --> 0:52:41.399 tuning the parameters of the network just right, or in fact, 0:52:41.520 --> 0:52:44.800 letting a network adjust its own connections according to some algorithm, 0:52:45.320 --> 0:52:50.279 a network learns to associate inputs and remember what it 0:52:50.360 --> 0:52:56.120 has learned. This is how we build artificial learning and memory, 0:52:56.200 --> 0:53:01.200 and modern artificial neural networks are extraordinarily impressive. But I 0:53:01.239 --> 0:53:05.400 think it's really important to note that although these connections 0:53:05.440 --> 0:53:09.600 between neurons have gotten all the attention, both theoretically and experimentally, 0:53:10.440 --> 0:53:14.960 there are lots of other possible ways to store information 0:53:15.239 --> 0:53:18.680 in the brain. For the last century, people have assumed 0:53:18.760 --> 0:53:22.960 that synapses are the key to memory, but the story 0:53:23.000 --> 0:53:28.000 has been complexified by recent decades of research. Where it 0:53:28.040 --> 0:53:32.240 stands now is that synaptic changes are necessary for learning 0:53:32.280 --> 0:53:35.560 and memory, but we really don't know if they are sufficient. 0:53:36.120 --> 0:53:39.080 It's still unclear whether these changes that the synapse is 0:53:39.680 --> 0:53:43.480 will be the only or even the most important mechanisms 0:53:43.520 --> 0:53:47.120 involved in memory, or perhaps whether they're involved at all, 0:53:47.320 --> 0:53:51.399 because the fact is that a dense net of intertwined 0:53:51.480 --> 0:53:57.920 cells has to orchestrate a careful balance between excitation and inhibition, 0:53:58.080 --> 0:54:01.279 otherwise the whole thing blows up into epilepsy or it 0:54:01.480 --> 0:54:04.640 sinks down into non activity. And it could be that 0:54:04.680 --> 0:54:08.120 all these synaptic changes we see are just to keep 0:54:08.160 --> 0:54:13.560 the system away from epileptic overload or synaptic shutdown, and 0:54:13.760 --> 0:54:18.520 memory maybe is stored in an entirely different manner. Our 0:54:18.640 --> 0:54:22.440 science is still young, and it's possible that we're still 0:54:22.560 --> 0:54:27.120 missing the core mechanisms of memory, because after all, there 0:54:27.200 --> 0:54:31.880 are adjustable parameters throughout the brain. You change something here 0:54:32.200 --> 0:54:35.240 and the network behaves differently there. So there are many 0:54:35.360 --> 0:54:40.080 other possible mechanisms involved in memory in the brain. For 0:54:40.160 --> 0:54:44.719 the cognitionanty, this could involve neurogenesis or changes in the 0:54:44.800 --> 0:54:48.800 excitability of the neuron, or the distribution of ion channel 0:54:49.239 --> 0:54:52.359 or the shape of dendritic trees and their spines, where 0:54:52.400 --> 0:54:57.000 the phosphorylation states of intracellular proteins, where the epigenetic codes 0:54:57.239 --> 0:54:59.719 and on and on. Okay, this is all in my 0:54:59.800 --> 0:55:01.959 tech book and we're not going to go into that here, 0:55:02.400 --> 0:55:05.080 but I do want to say that with so many 0:55:05.120 --> 0:55:10.200 degrees of freedom in biological systems, the number of possible 0:55:10.200 --> 0:55:15.080 ways to storm memory in a brain is vast. So 0:55:15.239 --> 0:55:18.600 why do we look at the connection strength of synapse 0:55:18.640 --> 0:55:23.040 as well? It's partially because that's what we can measure 0:55:23.080 --> 0:55:28.640 most easily. It's extremely difficult, essentially impossible currently to measure 0:55:29.000 --> 0:55:33.680 in a living animal changes in channel distribution or dendritic 0:55:33.719 --> 0:55:39.239 spines or epigenetic codes in single neurons. So as a result, 0:55:39.719 --> 0:55:44.239 almost the whole field of neuroscience just measures changes at 0:55:44.280 --> 0:55:47.520 the connections between cells. And that might turn out to 0:55:47.560 --> 0:55:51.120 be right, But if you could read the textbooks one 0:55:51.160 --> 0:55:53.680 hundred years from now, it might turn out to be 0:55:54.440 --> 0:55:57.720 that we were like the drunk, looking for our keys 0:55:57.800 --> 0:56:02.200 under the street light because that's where the lighting was better. 0:56:03.480 --> 0:56:07.640 So while artificial neural networks are awe inspiring and they're 0:56:07.719 --> 0:56:10.360 rapidly changing the world, we don't know for sure that 0:56:10.440 --> 0:56:14.680 they're doing the same things algorithmically that the brain is doing. 0:56:15.440 --> 0:56:19.120 They in a sense simplify everything. They are a clever 0:56:19.360 --> 0:56:22.960 step backwards from biology. In other words, you take the 0:56:23.440 --> 0:56:28.480 complexity of a single cell with the entire human genome 0:56:28.520 --> 0:56:31.520 in it and millions of proteins trafficking around, and you 0:56:31.640 --> 0:56:36.080 just imagine that it's a unit with simple connections to 0:56:36.160 --> 0:56:39.800 other units. And as I said, that has turned out 0:56:39.920 --> 0:56:43.880 in large language models to be shockingly effective. But we 0:56:44.080 --> 0:56:47.440 really have no way of knowing right now how much 0:56:47.560 --> 0:56:50.839 more we could get out of artificial neural networks if 0:56:50.840 --> 0:56:56.600 we included the rich complexity of actual biology. Maybe it 0:56:56.640 --> 0:57:00.759 wouldn't add anything, but maybe it would unlock in entirely 0:57:00.920 --> 0:57:06.160 new levels of function, making artificial neural networks more like 0:57:06.360 --> 0:57:11.160 a human with a sense of what information is relevant 0:57:11.160 --> 0:57:16.200 to learn and incorporating needs and goals and strategies and 0:57:16.920 --> 0:57:21.800 enjoying the experience of consciousness. The thing to keep in 0:57:21.840 --> 0:57:25.720 mind is that Mother Nature has had billions of years 0:57:25.760 --> 0:57:30.320 to try quadrillions of experiments, and we've only been making 0:57:30.440 --> 0:57:33.960 artificial neural networks for a few decades. So if I 0:57:34.000 --> 0:57:36.800 were a betting man, I would say there's probably a 0:57:36.920 --> 0:57:41.800 lot more to be discovered. It is certainly possible that 0:57:41.880 --> 0:57:48.120 we have not yet found memories Rosetta Stone and I 0:57:48.200 --> 0:57:51.560 want to point to one more thing about real brains 0:57:51.680 --> 0:57:55.040 that's definitely not captured in artificial neural networks, and that 0:57:55.200 --> 0:57:59.720 is the concept of forgetting. Remember I mentioned at the 0:57:59.720 --> 0:58:04.080 beginning about the mneminist named Sharyshevski, It turns out that 0:58:04.480 --> 0:58:10.400 Sherishevski's enviable memory went hand in hand with a surprisingly 0:58:10.600 --> 0:58:14.600 handicapped personality. He was crippled by the fact that he 0:58:14.640 --> 0:58:18.640 could not forget. Because when you have a memory like his, 0:58:19.680 --> 0:58:24.680 all the moments in life are retained. Past vandettas and 0:58:24.760 --> 0:58:28.880 things people did to slight you, and embarrassing moments and 0:58:29.080 --> 0:58:34.200 situations you've outgrown, and heartbreaks you would rather let slip 0:58:34.240 --> 0:58:38.480 into the past. All of these remain present and emotionally 0:58:38.560 --> 0:58:43.120 salient for you. And remember I mentioned the mnemenist Jill Price, 0:58:43.200 --> 0:58:46.120 who also had an untaxable memory. Just think about what 0:58:46.320 --> 0:58:49.240 drove her to seek help in the first place. Her 0:58:49.280 --> 0:58:53.479 memories were deeply emotional and sprang up throughout her day. 0:58:54.240 --> 0:59:00.240 Jill is constantly pouring over her past, an obsessive detailing 0:59:00.360 --> 0:59:05.560 mementos from childhood onward and becoming distressed by changes like 0:59:05.640 --> 0:59:08.760 moving to a new home. So for both Sharashevski and 0:59:08.840 --> 0:59:13.640 Jill Price. Having a perfect memory impaired them as much 0:59:13.640 --> 0:59:17.640 as equipped them, and this unmasks one of the great 0:59:17.800 --> 0:59:21.120 values of the way memory typically exists in most people. 0:59:21.560 --> 0:59:27.840 We don't retain everything, but instead items fade. One of 0:59:27.840 --> 0:59:31.360 my favorite quotations is from the French novelist and playwright 0:59:31.640 --> 0:59:37.160 Honore de Balzac, who wrote, memories beautify life, but the 0:59:37.240 --> 0:59:42.360 capacity to forget makes it bearable. So when it comes 0:59:42.440 --> 0:59:45.640 to thinking about human memory, remember that of all the 0:59:45.680 --> 0:59:49.160 things we talked about that differentiate us from digital computers, 0:59:49.440 --> 0:59:54.480 a very important one is our capacity to not remember everything. 0:59:55.320 --> 0:59:59.280 And even that which we do remember doesn't last too long. 1:00:00.080 --> 1:00:03.080 So let's wrap up. We've seen that human memory is 1:00:03.120 --> 1:00:05.680 not quite the same as a computer's memory. We don't 1:00:05.720 --> 1:00:09.720 store zeros and ones, but instead we hold memories about 1:00:09.800 --> 1:00:13.480 the gist of things, the relationships between things, and we 1:00:13.640 --> 1:00:18.240 forget through time and memory for us is stored in 1:00:18.320 --> 1:00:22.480 lots and lots of subsystems. We have different mechanisms for 1:00:22.920 --> 1:00:26.160 short term memory and for long term memory, and within 1:00:26.240 --> 1:00:31.920 each of those categories we have subcategories. For example, implicit 1:00:32.000 --> 1:00:34.360 memories that you can't articulate, like how to ride a 1:00:34.400 --> 1:00:38.440 bike and explicit memories like facts and events that you 1:00:38.440 --> 1:00:41.400 can talk about. And in all of these cases you 1:00:41.440 --> 1:00:44.919 can lose some of these types of memory without any 1:00:44.960 --> 1:00:48.919 effect on the other types. And now that we've seen 1:00:48.960 --> 1:00:52.280 an overview of memory in the brain, we will now 1:00:52.360 --> 1:00:57.520 be ready to turn to the main reason we have memory, 1:00:57.720 --> 1:01:03.600 and that is to predict the future. We retain information 1:01:04.120 --> 1:01:08.080 in the detailed configuration of our forest of billions of 1:01:08.160 --> 1:01:12.000 cells and trillions of connections, and the point is to 1:01:12.080 --> 1:01:15.840 develop a better understanding of the world so that we 1:01:16.040 --> 1:01:21.440 can know what will happen next. And for that, please 1:01:21.560 --> 1:01:24.800 join me in the next episode where we look at 1:01:25.080 --> 1:01:30.760 prediction how we simulate possible future worlds. This is one 1:01:30.800 --> 1:01:33.680 of the main jobs of brains, and this is the 1:01:33.720 --> 1:01:38.000 reason they retain memory. So right now, if you're imagining 1:01:38.080 --> 1:01:42.000 tuning into the next episode and feeling the emotional joy 1:01:42.360 --> 1:01:44.640 of what you will learn and what you will see there, 1:01:45.200 --> 1:01:48.280 this is your brain doing what it was meant to do, 1:01:48.880 --> 1:01:56.640 simulating the future. I will see you there. Go to 1:01:56.680 --> 1:01:59.800 Eagleman dot com slash podcast for more information and to 1:02:00.080 --> 1:02:04.960 find further reading. Send me any questions to podcast at 1:02:05.000 --> 1:02:08.240 eagleman dot com and I'll be making episodes in which 1:02:08.280 --> 1:02:13.840 I address those until next time. I'm David Eagleman, and 1:02:13.880 --> 1:02:15.640 this is Inner Cosmos.