WEBVTT - Ep70 "Why do our memories drift? Part 1: The War of the Ghosts" 0:00:05.080 --> 0:00:08.800 When you look at medieval European art, why do the 0:00:08.880 --> 0:00:13.480 people look fine but the lions look so strange? And 0:00:13.520 --> 0:00:16.720 what does this have to do with Native American folklore 0:00:17.239 --> 0:00:21.040 or I witness memory of a car accident, or what 0:00:21.079 --> 0:00:25.759 a person remembers three years after watching the nine to 0:00:25.760 --> 0:00:28.720 eleven attack on the World Trade Center, And what does 0:00:28.760 --> 0:00:32.279 any of this have to do with flashbulb memories or 0:00:32.360 --> 0:00:37.360 misinformation or the telephone game that you played as a child. 0:00:40.760 --> 0:00:43.760 Welcome to Inner Cosmos with me, David Eagleman. I'm a 0:00:43.840 --> 0:00:47.360 neuroscientist and an author at Stanford, and in these episodes 0:00:47.440 --> 0:00:51.000 we sail deeply into our three pound universe to understand 0:00:51.120 --> 0:00:54.560 why and how our lives look the way they do. 0:00:55.160 --> 0:00:58.400 Today's episode is part one of a two parter about 0:00:58.640 --> 0:01:05.119 memory and why it drifts. So I want to start 0:01:05.120 --> 0:01:08.240 with this fascinating observation that you can notice if you 0:01:08.360 --> 0:01:12.800 look across visual painting in Europe all through the Middle Ages. 0:01:13.480 --> 0:01:17.880 Painters got better and better through time at painting architecture 0:01:18.240 --> 0:01:22.320 and the human form and mountain scapes, but they were 0:01:22.400 --> 0:01:27.200 absolutely terrible at painting lions. If you look at lions 0:01:27.360 --> 0:01:30.720 in these medieval paintings, you'll see that they generally look 0:01:31.000 --> 0:01:36.280 quite different from their real life counterparts. They have exaggerated 0:01:36.360 --> 0:01:40.679 features like overly large heads and bodies that are too long, 0:01:41.319 --> 0:01:45.759 and tufted tails that aren't really like actual lion tails, 0:01:46.200 --> 0:01:49.280 and they often look more like large, fierce dogs or 0:01:49.320 --> 0:01:53.200 mythical creatures. I'll put some pictures on eagleman dot com 0:01:53.240 --> 0:01:57.120 slash podcasts so you can see how strange these lions are. 0:01:57.520 --> 0:02:02.200 But why is this? The answer is the medieval European 0:02:02.240 --> 0:02:06.120 painters spent a lot of time with architecture and with 0:02:06.440 --> 0:02:09.440 people and with mountain scapes, but almost none of them 0:02:09.680 --> 0:02:14.080 had ever been to Africa and therefore seen a real lion, 0:02:14.520 --> 0:02:16.399 or for that matter, had been to India and seen 0:02:16.440 --> 0:02:21.480 an asiatic lion. So they had the extremely tough challenge 0:02:21.520 --> 0:02:26.080 of painting something they had never actually seen. Now, to 0:02:26.200 --> 0:02:29.160 be clear, it's not that they had never seen a lion, 0:02:29.200 --> 0:02:32.200 it's that they had never seen a real lion. All 0:02:32.240 --> 0:02:36.000 they had ever seen were versions of lions painted by 0:02:36.040 --> 0:02:40.360 other medieval painters, who presumably had seen pictures of lions 0:02:40.360 --> 0:02:43.440 painted by other medieval painters, and so on back to 0:02:43.960 --> 0:02:47.880 someone who had at some point seen a real lion, 0:02:48.320 --> 0:02:51.360 and so This notion of what a lion looks like 0:02:51.800 --> 0:02:56.200 gives us a visual example of the operator game. This 0:02:56.320 --> 0:02:59.880 is also known as the telephone game or Russian scandal 0:03:00.320 --> 0:03:02.840 or pass the secret. So you remember doing this as 0:03:02.840 --> 0:03:05.919 a kid. One person whispers, let's say a word or 0:03:05.960 --> 0:03:08.760 a little phrase in your ear, and then you whisper 0:03:08.800 --> 0:03:11.120 it to the next person, and they whisper it to 0:03:11.160 --> 0:03:13.440 the next person, and so on, and by the time 0:03:13.440 --> 0:03:16.160 it gets to the last person, she shouts it out 0:03:16.400 --> 0:03:18.600 and you all get to see if it's the same 0:03:18.760 --> 0:03:22.040 word or phrase that the first person said. And the 0:03:22.120 --> 0:03:24.760 joy of the game, of course, is that the message 0:03:24.800 --> 0:03:28.079 gets distorted in transmission, and by the time it ends 0:03:28.160 --> 0:03:31.320 up somewhere it can be very different from how it began. 0:03:32.360 --> 0:03:35.600 So today and next week we're going to dive into 0:03:35.720 --> 0:03:41.000 a strange question, how your own memories are like the 0:03:41.080 --> 0:03:45.720 Operator game. In other words, we tend to erroneously believe 0:03:45.920 --> 0:03:49.240 that when something happens, we record that in the brain. 0:03:49.320 --> 0:03:53.280 It's written down in the brain as a memory, the 0:03:53.280 --> 0:03:56.080 way that a computer might hold a little file. And 0:03:56.120 --> 0:03:59.280 then every time we retrieve that memory, every time we 0:03:59.360 --> 0:04:02.640 pull that back up, we are viewing that little movie again. 0:04:03.280 --> 0:04:06.080 So it comes as a surprise that real human memory 0:04:06.160 --> 0:04:08.680 in the brain is nothing like a movie, but much 0:04:08.720 --> 0:04:14.680 more like the Operator game, wherein the message becomes increasingly distorted. 0:04:14.840 --> 0:04:17.840 For example, let's say I show you a picture of 0:04:17.880 --> 0:04:21.120 something that you're not super familiar with, like an old 0:04:21.520 --> 0:04:24.919 Polynesian war mass with a sort of strange shape in 0:04:25.000 --> 0:04:27.240 some different lines and circles on it, and then I 0:04:27.360 --> 0:04:32.120 ask you to draw it. Later, your drawing will probably 0:04:32.279 --> 0:04:35.200 drift from what you actually saw, and the next time 0:04:35.279 --> 0:04:37.599 after that, when I ask you to draw it again, 0:04:38.240 --> 0:04:41.920 your memory will presumably be influenced by what you drew 0:04:42.000 --> 0:04:44.479 the last time. And if I ask you to do 0:04:44.520 --> 0:04:46.960 this over and over again, let's say once a month, 0:04:47.040 --> 0:04:51.120 there will be something of a steady progression from drawing 0:04:51.240 --> 0:04:55.560 one to drawing ten, because you are playing the operator 0:04:55.640 --> 0:05:00.160 game with yourself. Each time you retrieve the memory, it 0:05:00.200 --> 0:05:03.279 is influenced by what you thought it was last time. 0:05:03.560 --> 0:05:06.200 So to dig into this, we're going to start today's 0:05:06.240 --> 0:05:10.200 episode with a short story. This is a Native American 0:05:10.440 --> 0:05:14.440 folk story called the War of the Ghosts, and it's 0:05:14.440 --> 0:05:16.520 read here by actor Sean Judge. 0:05:17.080 --> 0:05:21.640 War of the Ghosts. One night, two young men from 0:05:21.680 --> 0:05:24.800 Eguilak went down to the river to hunt seals, and 0:05:24.839 --> 0:05:28.960 while they were there it became foggy and calm. Then 0:05:29.400 --> 0:05:32.760 they heard war cries and they thought maybe this is 0:05:32.800 --> 0:05:35.960 a war party. They escaped to the shore and hid 0:05:36.000 --> 0:05:39.640 behind a log. Now canoes came up, and they heard 0:05:39.680 --> 0:05:42.520 the noise of paddles and saw one canoe coming up 0:05:42.520 --> 0:05:45.600 to them. There were five men in the canoe, and 0:05:45.640 --> 0:05:48.719 they said, what do you think we wish to take 0:05:48.720 --> 0:05:50.800 you along. We are going up the river to make 0:05:50.839 --> 0:05:53.880 war on the people. One of the young men said, 0:05:54.480 --> 0:05:58.200 I have no arrows. Arrows are in the canoe. They said, 0:05:59.160 --> 0:06:01.760 I will not go along. I might be killed. My 0:06:01.839 --> 0:06:04.560 relatives do not know where I have gone. But you, 0:06:04.880 --> 0:06:07.200 he said, turning to the other, may go with them. 0:06:08.320 --> 0:06:10.760 So one of the young men went, but the other 0:06:10.880 --> 0:06:14.800 returned home, and the warriors went on up the river 0:06:14.960 --> 0:06:17.800 to a town on the other side of Kalama. The 0:06:17.839 --> 0:06:20.120 people came down to the water and they began to fight, 0:06:20.360 --> 0:06:24.000 and many were killed. But presently the young man heard 0:06:24.000 --> 0:06:26.840 one of the warriors say, quick, let us go home. 0:06:27.200 --> 0:06:32.000 That Indian has been hit. Now he thought, oh, they 0:06:32.040 --> 0:06:36.080 are ghosts. He did not feel sick, but they said 0:06:36.200 --> 0:06:40.159 he had been shot. So the canoes went back to Aguilak, 0:06:40.320 --> 0:06:42.440 and the young man went ashore to his house and 0:06:42.480 --> 0:06:46.440 made a fire, and he told everybody and said, behold, 0:06:46.839 --> 0:06:50.040 I accompanied the ghosts, and we went to fight. Many 0:06:50.080 --> 0:06:52.160 of our fellows were killed, and many of those who 0:06:52.200 --> 0:06:55.320 attacked us were killed. They said, I was hit, but 0:06:55.400 --> 0:06:58.760 I did not feel sick. He told it all, and 0:06:58.839 --> 0:07:03.359 then he became quiet. When the sun rose, he fell down. 0:07:04.320 --> 0:07:08.640 Something black came out of his mouth, his face became contorted. 0:07:09.400 --> 0:07:12.880 The people jumped up and cried he was dead. 0:07:13.680 --> 0:07:16.400 This probably seems to you like a bit of a 0:07:16.440 --> 0:07:20.760 weird story. It seems like it's not particularly well told 0:07:20.880 --> 0:07:25.120 or clear. So given the strangeness of the story, why 0:07:25.160 --> 0:07:29.080 did it become quite famous in the psychology community almost 0:07:29.080 --> 0:07:32.160 a century ago in nineteen thirty two. It's because of 0:07:32.200 --> 0:07:36.600 a researcher named Frederick Bartlett, who was a psychologist at 0:07:36.640 --> 0:07:40.680 Cambridge University. He chose this story, the War of the Ghosts, 0:07:41.120 --> 0:07:45.960 to examine the way that our memories change with time. Now, 0:07:46.000 --> 0:07:49.040 he chose this particular folk tale because he wanted something 0:07:49.080 --> 0:07:52.600 that wasn't crystal clear, so that it might be slightly 0:07:52.680 --> 0:07:57.480 more susceptible to changes in the retelling. That way, he 0:07:57.480 --> 0:08:01.560 could examine the way in which it changed in the retelling, 0:08:01.920 --> 0:08:04.800 and the retelling after that, and so on into the future. 0:08:04.920 --> 0:08:08.000 In other words, he used this story to see if 0:08:08.040 --> 0:08:12.920 he could really shine a light on the constructive character 0:08:13.040 --> 0:08:15.800 of memory. In other words, the way that recalling a 0:08:15.840 --> 0:08:22.360 memory is something like reconstructing what happened your brain rebuilding 0:08:22.400 --> 0:08:25.360 its best guess at a memory, rather than the way 0:08:25.440 --> 0:08:29.040 a computer simply retrieves a file of zeros and ones 0:08:29.040 --> 0:08:33.480 without loss. So in Bartlett's study, he had participants read 0:08:33.679 --> 0:08:36.600 the War of the Ghosts. Then they were instructed to 0:08:36.720 --> 0:08:40.319 recall and write down the story as perfectly as they could. 0:08:41.200 --> 0:08:43.600 Now it's only been about a minute since you heard 0:08:43.640 --> 0:08:47.440 the story, Think about how you would retell it now, 0:08:47.480 --> 0:08:51.200 what details you would remember from that story. Not surprisingly, 0:08:51.280 --> 0:08:55.480 Bartlett found that the story drifted from the original upon 0:08:55.640 --> 0:08:59.040 the retelling. Here, for example, is a person who is 0:08:59.120 --> 0:09:03.160 asked to remember and reproduce this story, and he did 0:09:03.160 --> 0:09:07.240 this multiple times. Here he is on his tenth reproduction. 0:09:07.120 --> 0:09:10.480 War of the Ghosts. Two Indians were out fishing for 0:09:10.559 --> 0:09:13.800 seals in the Bay of Manpapan when along came five 0:09:13.880 --> 0:09:17.400 other Indians in a war canoe. They were going fighting. 0:09:18.120 --> 0:09:20.520 Come with us, said five of the two, and fight. 0:09:21.440 --> 0:09:23.760 I cannot come, was the answer of the one, for 0:09:23.920 --> 0:09:25.800 I have an old mother at home who has dependent 0:09:25.840 --> 0:09:28.600 on me. The other also said he could not come 0:09:28.640 --> 0:09:31.920 because he had no arms. That is no difficulty. The 0:09:31.960 --> 0:09:34.800 others replied, for we have plenty in the canoe with us. 0:09:35.679 --> 0:09:38.880 So he got into the canoe and went with them 0:09:39.120 --> 0:09:42.800 in a fight. Soon afterwards, this Indian received a mortal wound. 0:09:43.679 --> 0:09:46.240 Finding that his hour was come, he cried out that 0:09:46.320 --> 0:09:49.480 he was about to die. Nonsense, said one of the others. 0:09:49.760 --> 0:09:53.120 You will not die, But he did now. 0:09:53.200 --> 0:09:56.439 Bartlett also had other participants read the story and then 0:09:56.480 --> 0:09:59.960 reproduce it at intervals, very far apart in some case 0:10:00.320 --> 0:10:04.599 over years. So here's another participant. This is subject p 0:10:05.360 --> 0:10:07.760 on his first reproduction. 0:10:07.720 --> 0:10:11.600 War of the Ghosts. Two youths were standing by a 0:10:11.640 --> 0:10:14.600 river about to start seal catching when a boat appeared 0:10:14.600 --> 0:10:18.000 with five men in it. They were all armed for war. 0:10:18.920 --> 0:10:21.800 The youths were at first frightened, but they were asked 0:10:21.800 --> 0:10:23.760 by the men to come and help them fight some 0:10:23.920 --> 0:10:26.719 enemies on the other bank. One youth said that he 0:10:26.760 --> 0:10:29.360 could not come, as his relations would be anxious about him. 0:10:29.880 --> 0:10:32.400 The other said he would go and entered the boat. 0:10:33.360 --> 0:10:35.920 In the evening, he returned to his hut and told 0:10:35.920 --> 0:10:38.840 his friends that he had been in a battle. A 0:10:38.880 --> 0:10:41.840 great many had been slain, and he had been wounded 0:10:41.840 --> 0:10:45.000 by an arrow. He had not felt any pain, he said. 0:10:45.679 --> 0:10:47.480 They told him that he must have been fighting a 0:10:47.480 --> 0:10:51.960 battle of ghosts. Then he remembered that it had been queer, 0:10:52.280 --> 0:10:56.440 and he was very excited. In the morning, however, he 0:10:56.520 --> 0:11:00.760 became ill, and his friends gathered round. He fell down 0:11:00.880 --> 0:11:05.920 and his face became very pale. Then he writhed and shrieked, 0:11:06.280 --> 0:11:09.960 and his friends were filled with terror. At last he 0:11:10.000 --> 0:11:14.720 became calm. Something hard and black came out of his mouth, 0:11:15.480 --> 0:11:17.880 and he lay contorted and dead. 0:11:18.400 --> 0:11:21.760 And here he is again when he's asked to reproduce 0:11:21.840 --> 0:11:24.240 the story. After thirty months. 0:11:24.559 --> 0:11:28.640 War of the ghosts, some warriors went to wage war 0:11:28.679 --> 0:11:31.800 against the ghosts. They fought all day, and one of 0:11:31.840 --> 0:11:35.320 their number was wounded. They returned home in the evening 0:11:35.640 --> 0:11:39.240 bearing their sick comrade. As the day drew to a close, 0:11:39.440 --> 0:11:42.560 he became rapidly worse, and the villagers came round him 0:11:43.520 --> 0:11:47.480 as sunset, he sighed something black came out of his mouth. 0:11:48.920 --> 0:12:00.280 He was dead. 0:12:05.360 --> 0:12:08.120 So what he generally found is that the story became 0:12:08.640 --> 0:12:12.240 shorter with each reproduction. But here was the key. He 0:12:12.360 --> 0:12:15.680 realized that this was more than just the telephone game, 0:12:15.960 --> 0:12:21.079 where a signal just becomes noisier each time it's repeated. Instead, 0:12:21.120 --> 0:12:23.760 with the War of the Ghosts, he found there was 0:12:23.840 --> 0:12:27.880 a pattern to the way the story changed. First of all, 0:12:28.280 --> 0:12:34.000 it typically became more coherent to the speaker, as Bartlett wrote, 0:12:34.240 --> 0:12:39.040 quote no trace of an odd or supernatural element is left. 0:12:39.240 --> 0:12:43.480 We have a perfectly straightforward story of a fight and 0:12:43.600 --> 0:12:47.920 a death end quote. So Bartlett studied the character of 0:12:47.960 --> 0:12:52.679 the changes, and he found that these happened by transforming 0:12:52.800 --> 0:12:57.160 details into things that are more familiar and conventional to 0:12:57.320 --> 0:13:01.440 the person doing the remembering. Sometimes the order of events 0:13:01.480 --> 0:13:05.679 would change, and things would commonly get omitted, like the 0:13:05.720 --> 0:13:08.520 ghosts getting sliced out of the story pretty early on, 0:13:09.160 --> 0:13:13.079 or the wound becoming a matter of flesh not spirit. 0:13:13.559 --> 0:13:16.280 All in all, he was interested in the way that people, 0:13:16.760 --> 0:13:21.160 through time made the parts grow more coherent for themselves. 0:13:21.240 --> 0:13:25.080 In other words, the distortions were driven by a person's 0:13:25.400 --> 0:13:28.040 schema now, what does that mean. It means that people 0:13:28.120 --> 0:13:31.840 morph the story to make it consistent with what's going 0:13:31.880 --> 0:13:34.960 on in their heads. In other words, the changes you 0:13:35.000 --> 0:13:38.160 make to the story are navigated by your internal model, 0:13:38.320 --> 0:13:42.800 or what psychologists call your schema, your mental framework for 0:13:43.080 --> 0:13:48.679 organizing knowledge and guiding perception. This is your brain's unique 0:13:48.679 --> 0:13:52.199 way of seeing the world. So the way the story 0:13:52.320 --> 0:13:55.520 was distorted in terms of the plot and characters and 0:13:55.520 --> 0:13:59.240 events is different in your brain versus someone else's brain. 0:13:59.840 --> 0:14:04.080 There's a lot of differences between individual schema, which is 0:14:04.120 --> 0:14:06.920 to say, if you read the story to someone who's 0:14:06.960 --> 0:14:10.480 really into boats, they might remember something differently than someone 0:14:10.480 --> 0:14:14.400 who's really into Native American history, or someone who's a 0:14:14.400 --> 0:14:17.360 physician and thinks about what causes a person to die. 0:14:18.160 --> 0:14:20.960 And by the way, it's not just individual differences, but 0:14:21.000 --> 0:14:24.960 also your culture plays a massive role in shaping the 0:14:25.000 --> 0:14:28.960 way that you interpret and remember information. So as you 0:14:29.040 --> 0:14:34.040 recall a story, you unconsciously mold it to align with 0:14:34.160 --> 0:14:38.400 your cultural background and your beliefs and your expectations. So 0:14:38.440 --> 0:14:41.960 this all has to do with the differences in internal models. 0:14:42.320 --> 0:14:46.360 Bartlett's study the way that people can't label something until 0:14:46.440 --> 0:14:51.360 it is recognizable so transformations moved in the direction of 0:14:51.480 --> 0:14:56.360 the conventions in their head. So I think about Bartlett's 0:14:56.400 --> 0:14:59.920 study all the time, for example, and how we consume news. 0:15:00.520 --> 0:15:04.280 Any news story we read this month is remembered differently 0:15:04.520 --> 0:15:07.640 by you than by someone else who reads it through 0:15:07.720 --> 0:15:11.720 a different political lens. Watch anything on the news, and 0:15:11.760 --> 0:15:14.040 then talk to anyone you know who sees the world 0:15:14.040 --> 0:15:17.239 a little differently, and you'll find that it's not impossible 0:15:17.720 --> 0:15:21.080 that they remember different parts of the story than you did, 0:15:21.440 --> 0:15:26.120 and or they remembered them differently. Their distortions in the 0:15:26.160 --> 0:15:29.640 story are not accidental, and nor are yours. They are 0:15:29.760 --> 0:15:35.440 ways of making things internally consistent. Now, psychologists have a 0:15:35.520 --> 0:15:39.640 lot of ways of talking about this, like confirmation bias, 0:15:39.800 --> 0:15:43.440 which means we tend to seek out and remember information 0:15:44.040 --> 0:15:47.440 that is consistent with our internal models, and there are 0:15:47.520 --> 0:15:50.280 other ways of talking about this. I've recently did an 0:15:50.280 --> 0:15:53.560 episode on conspiracy theories, and one of the things I 0:15:53.720 --> 0:15:56.960 highlighted is the way that we are attracted to theories 0:15:57.000 --> 0:16:01.640 that reduce our cognitive load or our dissonance. In other words, 0:16:02.080 --> 0:16:05.720 we prefer a clear and easy story. And all this 0:16:05.880 --> 0:16:09.760 is consistent with what Bartlett found in nineteen thirty two. 0:16:10.040 --> 0:16:13.440 We don't remember the story as it is presented to us, 0:16:13.840 --> 0:16:17.880 but rather as our filters allow us to store it 0:16:17.960 --> 0:16:21.960 and recall it. And when I think about the driftiness 0:16:22.080 --> 0:16:25.600 of memory, there are other angles that need to be surfaced. 0:16:25.640 --> 0:16:27.400 And here's the main thing that we're going to talk 0:16:27.440 --> 0:16:32.800 about today. Rather than memory being an accurate video recording 0:16:32.920 --> 0:16:35.880 of a moment in your life, it is a fragile 0:16:36.200 --> 0:16:39.760 brain state from a bygone time that has to be 0:16:40.280 --> 0:16:45.320 resurrected for you to remember. So here's an example. You're 0:16:45.360 --> 0:16:48.760 at a restaurant with a couple that you know who 0:16:48.840 --> 0:16:54.200 are visiting town, and everything you experience triggers particular patterns 0:16:54.200 --> 0:16:57.800 of activity in your brain. For example, there's a particular 0:16:58.040 --> 0:17:02.040 pattern of activity sparked into life by the conversation between 0:17:02.120 --> 0:17:05.720 you and your friends. Another pattern is activated by the 0:17:05.760 --> 0:17:09.720 smell of the coffee. Another pattern is activated by the 0:17:09.760 --> 0:17:13.480 taste of a madeleine cookie. The fact that the waiter 0:17:13.760 --> 0:17:16.600 is limping with a cast on his leg is another 0:17:17.080 --> 0:17:21.159 memorable detail, and that's represented by a different configuration of 0:17:21.200 --> 0:17:25.560 neurons firing. All of these constellations become linked with one 0:17:25.560 --> 0:17:30.360 another in a vast network of neurons that the hippocampus 0:17:30.680 --> 0:17:36.400 replays over and over until the associations between these scattered neurons, 0:17:36.480 --> 0:17:40.520 these distant points of light, this all becomes fixed into place. 0:17:41.320 --> 0:17:44.640 The neurons that are active at the same time will 0:17:44.760 --> 0:17:50.359 establish stronger connections between them, sells that fire together wire together. 0:17:50.880 --> 0:17:55.720 The resulting network is the unique signature of that event, 0:17:55.880 --> 0:18:00.959 and it represents your memory of the dinner with them. Now, 0:18:01.320 --> 0:18:04.639 let's imagine that six months later, you taste one of 0:18:04.680 --> 0:18:07.880 those Madeleine cookies just like the one you tasted at 0:18:07.880 --> 0:18:12.280 the dinner with the couple. This very specific key can 0:18:12.440 --> 0:18:18.320 unlock the whole web of associations. The original constellation lights 0:18:18.400 --> 0:18:21.760 up like the lights of a city coming online, and 0:18:21.880 --> 0:18:27.160 suddenly your back in that memory. But although we don't 0:18:27.160 --> 0:18:30.880 always realize it, the memory is not as rich as 0:18:30.920 --> 0:18:34.160 you might have expected. You know that your two friends 0:18:34.160 --> 0:18:36.600 were there, and he must have been wearing a button 0:18:36.680 --> 0:18:38.920 up shirt because he always wears a button up shirt, 0:18:39.400 --> 0:18:42.959 and she was wearing a blue dress or maybe it 0:18:43.000 --> 0:18:46.919 was purple, might have been green. If you really probe 0:18:46.920 --> 0:18:50.400 the memory, you'll realize that you can't remember the details 0:18:50.400 --> 0:18:52.600 of any of the other diners at the restaurant, even 0:18:52.600 --> 0:18:56.240 though the place was full. So your memory of their 0:18:56.280 --> 0:19:01.280 dinner visit has started to fade. Why well, you have 0:19:01.359 --> 0:19:05.000 a finite number of neurons and they are each required 0:19:05.080 --> 0:19:09.720 to multitask. Every neuron in your head participates in different 0:19:09.840 --> 0:19:15.440 constellations at different times. Your neurons operate in a dynamic 0:19:15.720 --> 0:19:20.320 matrix of shifting relationships, and heavy demand is always placed 0:19:20.359 --> 0:19:23.840 on them to wire with others. So your memory of 0:19:23.880 --> 0:19:28.000 the dinner has become muddied as the neurons involved have 0:19:28.080 --> 0:19:33.239 become co opted to participate in other memory constellations. The 0:19:33.560 --> 0:19:39.320 enemy of memory isn't time, it's other memories. Each new 0:19:39.359 --> 0:19:44.359 event needs to establish new relationships among a finite number 0:19:44.400 --> 0:19:48.880 of neurons. Now, the surprise is that a faded memory 0:19:49.320 --> 0:19:53.720 doesn't feel faded to you. You assume, or at least 0:19:53.720 --> 0:19:57.320 you into it, that the full picture is there. And 0:19:57.400 --> 0:20:00.680 the situation is even worse than this, because because it's 0:20:00.720 --> 0:20:04.920 not just that the memory is fading, it's actually drifting. 0:20:05.440 --> 0:20:09.760 So imagine that in the intervening year since the dinner, 0:20:10.080 --> 0:20:14.119 your two friends have broken up. Thinking back on the dinner, 0:20:14.160 --> 0:20:20.280 you might now misremember sensing red flags. Wasn't he more 0:20:20.400 --> 0:20:24.160 quiet than usual that night. Weren't there moments of awkward 0:20:24.280 --> 0:20:27.439 silence between the two? Well, it'll be difficult to know 0:20:27.600 --> 0:20:31.199 for certain, because the knowledge that's in your network now 0:20:31.600 --> 0:20:34.919 now that they've broken up, that changes the memory that 0:20:35.040 --> 0:20:40.160 corresponds to. Then you can't help but have your present 0:20:40.760 --> 0:20:46.080 color your past, so an event may be perceived somewhat 0:20:46.080 --> 0:20:50.120 differently by you at different moments in your life now. 0:20:50.240 --> 0:20:54.840 Psychologists have been studying the malleability of memory for many decades, 0:20:54.880 --> 0:20:58.080 and one of the pioneers in the field is Elizabeth 0:20:58.160 --> 0:21:02.199 Loftis at the University of California, Irvine. She transformed the 0:21:02.240 --> 0:21:07.040 field of memory research by showing how susceptible memories are. 0:21:07.280 --> 0:21:10.000 So one of her experiments goes like this. She has 0:21:10.040 --> 0:21:14.719 people watch films of car crashes and then she asks 0:21:14.760 --> 0:21:17.600 them a series of questions to test what they remembered. 0:21:17.920 --> 0:21:20.480 And what she found is that just the way she 0:21:20.800 --> 0:21:25.480 asks the question influences the answer she gets back. So 0:21:25.640 --> 0:21:28.960 if she asks how fast were the cars going when 0:21:28.960 --> 0:21:32.120 they hit each other, she gets an estimate of the speed. 0:21:32.440 --> 0:21:35.480 But if she asks how fast were the cars going 0:21:35.840 --> 0:21:40.480 when they smashed into each other, witnesses give a different estimate. 0:21:40.800 --> 0:21:45.040 Not surprisingly, they believe the cars were going faster when 0:21:45.040 --> 0:21:48.040 she uses the word smashed. Now, if you heard my 0:21:48.160 --> 0:21:51.800 episode on eyewitness testimony, you know that these sorts of 0:21:51.880 --> 0:21:54.479 problems are seen all the time in the legal system. 0:21:54.600 --> 0:22:00.359 For example, there's a related issue known as the misinformation effect. 0:22:00.760 --> 0:22:03.720 If you witness an event, but then later you're told 0:22:03.760 --> 0:22:06.320 something about that event in terms of what happened at 0:22:06.320 --> 0:22:09.000 the scene, or who is there or what they look like, 0:22:09.640 --> 0:22:12.280 that will become part of your memory too, and you 0:22:12.320 --> 0:22:14.879 may not be able to distinguish what you've been told 0:22:15.160 --> 0:22:19.159 from what actually happened. So Loftis and her team study this, 0:22:19.359 --> 0:22:22.399 for example, by showing people a picture of a car 0:22:22.440 --> 0:22:25.800 at a stop sign, and then afterwards, after the picture 0:22:25.880 --> 0:22:29.720 is removed, they give a text description of the same picture, 0:22:29.960 --> 0:22:33.480 but in the text they say it was a yield sign. 0:22:33.880 --> 0:22:36.920 And then they have people draw the picture as they 0:22:36.960 --> 0:22:40.639 remember it, and people tend to draw the original scene 0:22:40.680 --> 0:22:44.800 with a yield sign. So even after an memory is encoded, 0:22:45.440 --> 0:22:48.399 new information coming in like that it was a yield 0:22:48.440 --> 0:22:53.000 sign will change the retrieval. You'll believe that the whole 0:22:53.040 --> 0:22:56.720 time you saw a yield sign there instead of a 0:22:56.760 --> 0:23:14.760 stop sign. Now all the sounds surprising. At first, you 0:23:14.800 --> 0:23:18.560 think I can distinguish my own memory from something someone 0:23:18.640 --> 0:23:22.760 else said, but it turns out we can't. And we'll 0:23:22.800 --> 0:23:26.040 dive more into this in a quite shocking way in 0:23:26.119 --> 0:23:28.679 next week's episode. But I want to give you a 0:23:28.760 --> 0:23:32.160 deeper sense of this now, about the changeability of memories. 0:23:32.200 --> 0:23:36.840 So here's another problem. It's a relative of the misinformation 0:23:36.880 --> 0:23:41.080 effect that psychologists and legal theorists have studied, and this 0:23:41.119 --> 0:23:45.359 is the issue of co witness contamination. The idea is 0:23:45.400 --> 0:23:48.000 that if you see a crime and I'm standing there 0:23:48.000 --> 0:23:50.640 and I see it too, and then we start talking 0:23:50.760 --> 0:23:54.240 with one another about it, we can't help but influence 0:23:54.320 --> 0:23:59.480 each other's memories. If you remember that she had curly hair, 0:23:59.520 --> 0:24:01.760 but I say I'm pretty sure she had straight hair. 0:24:02.200 --> 0:24:05.280 Or if you think she was unathletic, but I say, no, 0:24:05.400 --> 0:24:09.200 she was really athletic, each of our statements become part 0:24:09.280 --> 0:24:12.840 of each other's memory, and we more and more come 0:24:12.880 --> 0:24:17.320 to believe things that we didn't originally. And we see 0:24:17.359 --> 0:24:21.159 these kinds of memory problems all the time in police lineups. 0:24:21.560 --> 0:24:25.280 For starters, it's really difficult to remember the face of 0:24:25.320 --> 0:24:29.200 a perpetrator. In a crime scene. So police and researchers 0:24:29.440 --> 0:24:32.760 have tried every which way for people to have an 0:24:32.800 --> 0:24:37.119 easier time reconstructing a face. You maybe describe it to 0:24:37.160 --> 0:24:40.600 an artist, or you reassemble the face from a selection 0:24:40.640 --> 0:24:44.280 of different possible eyes and nose shapes and mouths, and 0:24:44.440 --> 0:24:46.760 in the last decades you can do this with three 0:24:46.840 --> 0:24:49.880 D avatars on a computer. But whatever tech is introduced, 0:24:49.880 --> 0:24:55.360 it doesn't matter. People's performance is terrible. Why because memories 0:24:55.440 --> 0:25:00.439 are not like photographs. So one issue that researchers started 0:25:00.520 --> 0:25:05.600 noticing and studying in the nineteen sixties was police suggestibility. 0:25:06.119 --> 0:25:09.760 So imagine you're shown a lineup with several people and 0:25:09.800 --> 0:25:13.119 you have to decide which person you saw doing the crime. 0:25:13.880 --> 0:25:17.040 But imagine that the police already have their man in mind. 0:25:17.119 --> 0:25:20.560 They already think that it's Steve. Whether or not that's correct, 0:25:20.920 --> 0:25:23.000 they believe it's Steve, and they want you to say 0:25:23.040 --> 0:25:25.239 it's Steve. There are all kinds of ways that they 0:25:25.280 --> 0:25:28.119 can suggest that to you, even if it's unconscious on 0:25:28.200 --> 0:25:31.640 their part. So for example, positive feedback. If you say 0:25:32.320 --> 0:25:36.160 I think that's the guy, they say, yeah, good job, 0:25:36.200 --> 0:25:39.119 that's what we think. Also, it turns out that the 0:25:39.200 --> 0:25:44.040 positive feedback influences the confidence of the eyewitness, and then 0:25:44.119 --> 0:25:48.280 when the trial starts months later, the eyewitness says, I'm 0:25:48.400 --> 0:25:51.520 absolutely certain that it was Steve, even though he doesn't 0:25:51.560 --> 0:25:55.520 remember that. He was not certain at all, but because 0:25:55.560 --> 0:25:58.199 of the positive feedback, which can even be quite subtle, 0:25:58.600 --> 0:26:02.919 his confidence goes way up. And decades of psychology studies 0:26:02.920 --> 0:26:06.480 in the laboratory have verified the power of this kind 0:26:06.520 --> 0:26:10.080 of suggestibility. The main thing that everybody's worried about with 0:26:10.200 --> 0:26:14.880