1
00:00:05,559 --> 00:00:08,959
Speaker 1: What's up with those illusions on the Internet where you

2
00:00:09,000 --> 00:00:12,559
Speaker 1: can hear the same sound one of two different ways

3
00:00:12,600 --> 00:00:15,640
Speaker 1: depending on the word that you're looking at. And why

4
00:00:15,640 --> 00:00:19,159
Speaker 1: do electrical outlets sometimes look like a face to you?

5
00:00:19,960 --> 00:00:23,440
Speaker 1: How can you have full, rich visual experience with your

6
00:00:23,480 --> 00:00:27,080
Speaker 1: eyes closed. And when you want to cross a street

7
00:00:27,160 --> 00:00:30,120
Speaker 1: and you hit that crosswalk button, are some of those

8
00:00:30,120 --> 00:00:32,680
Speaker 1: buttons fake and they don't actually do anything?

9
00:00:33,320 --> 00:00:34,519
Speaker 2: And why are there some.

10
00:00:34,640 --> 00:00:38,040
Speaker 1: Pictures that you can only see once you're told what

11
00:00:38,080 --> 00:00:41,920
Speaker 1: you're looking at. And although brains are often celebrated for

12
00:00:41,960 --> 00:00:46,200
Speaker 1: their parallel processing, what did they really be celebrated for.

13
00:00:49,880 --> 00:00:53,000
Speaker 1: Welcome to Inner Cosmos with Me David Eagleman. I'm a

14
00:00:53,080 --> 00:00:57,560
Speaker 1: neuroscientist and author at Stanford and in these episodes we

15
00:00:57,680 --> 00:01:02,160
Speaker 1: sail deeply into our three pounds universe to understand why

16
00:01:02,200 --> 00:01:05,040
Speaker 1: we perceive the world in the ways that we do.

17
00:01:13,520 --> 00:01:18,080
Speaker 1: Today's episode is about expectations and what that has to

18
00:01:18,120 --> 00:01:24,480
Speaker 1: do with perception. Unless you were living in outer space

19
00:01:24,720 --> 00:01:28,000
Speaker 1: or off the grid in twenty fifteen, your life was

20
00:01:28,120 --> 00:01:33,240
Speaker 1: touched by a very tiny, specific event that happened on

21
00:01:33,280 --> 00:01:37,679
Speaker 1: a small island in Scotland. Two young people were going

22
00:01:37,720 --> 00:01:40,680
Speaker 1: to get married there, and a week before the wedding,

23
00:01:41,080 --> 00:01:43,720
Speaker 1: the mother of the bride was shopping around for what

24
00:01:43,840 --> 00:01:47,680
Speaker 1: she was going to wear. So she finds some outfits

25
00:01:47,760 --> 00:01:50,720
Speaker 1: at a store down in Chester, England that she thinks

26
00:01:50,760 --> 00:01:54,080
Speaker 1: will look nice, and while she's making the decision, she

27
00:01:54,360 --> 00:01:57,480
Speaker 1: snaps pictures of each of them and she buys one

28
00:01:57,520 --> 00:01:57,800
Speaker 1: of them.

29
00:01:58,520 --> 00:01:59,880
Speaker 2: So she's driving home.

30
00:01:59,720 --> 00:02:03,560
Speaker 1: After words and she texts the pictures of the three

31
00:02:03,600 --> 00:02:07,320
Speaker 1: outfits to her daughter and she tells her that she

32
00:02:07,480 --> 00:02:10,560
Speaker 1: had bought the third one, and no one could have

33
00:02:10,720 --> 00:02:14,280
Speaker 1: ever guessed that this particular piece of clothing that she

34
00:02:14,400 --> 00:02:18,680
Speaker 1: sent a picture of, this one piece garment, is about

35
00:02:18,760 --> 00:02:22,440
Speaker 1: to become the most famous outfit that ever existed in

36
00:02:22,480 --> 00:02:27,080
Speaker 1: the history of humankind, because the daughter writes back to

37
00:02:27,280 --> 00:02:32,680
Speaker 1: clarify which outfit the mother had bought, and she texts, oh,

38
00:02:32,919 --> 00:02:37,200
Speaker 1: the white and gold one, and the mother texts back, no,

39
00:02:37,760 --> 00:02:43,240
Speaker 1: it's blue and black, and the daughter replies, Mom, if

40
00:02:43,280 --> 00:02:45,600
Speaker 1: you think that's blue and black, you need to go

41
00:02:45,639 --> 00:02:49,400
Speaker 1: and see the doctor. So the mother shows the phone

42
00:02:49,440 --> 00:02:52,359
Speaker 1: to her partner in the car, who, despite having been

43
00:02:52,400 --> 00:02:54,519
Speaker 1: there and bought the dress with her, looks at the

44
00:02:54,560 --> 00:02:57,000
Speaker 1: photo and says, yeah, I think it's white and gold.

45
00:02:57,680 --> 00:02:59,400
Speaker 1: So when they get home, they show the picture to

46
00:02:59,440 --> 00:03:02,600
Speaker 1: their younger, who agrees with the mother that the photo

47
00:03:02,639 --> 00:03:08,600
Speaker 1: looks blue and black. So, given this funny disagreement, the

48
00:03:08,760 --> 00:03:11,880
Speaker 1: bride to be posts the photo to her friends on

49
00:03:11,960 --> 00:03:16,560
Speaker 1: Facebook to settle this, and to her surprise, she doesn't

50
00:03:16,600 --> 00:03:20,840
Speaker 1: find consensus. Some think it's black and blue, others think

51
00:03:20,919 --> 00:03:25,800
Speaker 1: it's white and gold, and each person feels totally certain

52
00:03:25,880 --> 00:03:29,160
Speaker 1: about what they see. So for about a week, this

53
00:03:29,280 --> 00:03:34,200
Speaker 1: debate bubbles around in this small island community. The day

54
00:03:34,200 --> 00:03:36,920
Speaker 1: of the wedding arrives and the mother wears the dress

55
00:03:36,960 --> 00:03:40,280
Speaker 1: to the event, and the issue about the photo becomes

56
00:03:40,480 --> 00:03:43,080
Speaker 1: such a point of discussion that the musicians in the

57
00:03:43,120 --> 00:03:46,280
Speaker 1: band allegedly almost didn't make it onto the stage to

58
00:03:46,360 --> 00:03:48,720
Speaker 1: play because they were wrapped up in the debate.

59
00:03:49,720 --> 00:03:51,240
Speaker 2: So a few days after.

60
00:03:50,960 --> 00:03:53,200
Speaker 1: The wedding, one of the band members, who was a

61
00:03:53,280 --> 00:03:57,080
Speaker 1: friend of the happy couple, she posts the photo to

62
00:03:57,200 --> 00:04:00,520
Speaker 1: her blog on Tumblr, and by the end of the

63
00:04:00,600 --> 00:04:05,400
Speaker 1: day it gets five thousand comments, and soon enough, the

64
00:04:05,640 --> 00:04:09,839
Speaker 1: data scientists at Tumblr are examining this post because it's

65
00:04:09,880 --> 00:04:15,080
Speaker 1: getting fourteen thousand views each second. That's close to a

66
00:04:15,160 --> 00:04:20,600
Speaker 1: million views each minute. So a woman on the BuzzFeed

67
00:04:20,720 --> 00:04:24,360
Speaker 1: social media team sets up a poll about the color

68
00:04:24,600 --> 00:04:27,680
Speaker 1: for Tumblr users, and then she packs up and goes

69
00:04:27,720 --> 00:04:29,960
Speaker 1: home on the subway. And by the time she gets

70
00:04:30,080 --> 00:04:34,440
Speaker 1: off the subway, her phone is overwhelmed, and soon enough

71
00:04:34,480 --> 00:04:38,840
Speaker 1: the BuzzFeed page hits new records for how many unique

72
00:04:38,920 --> 00:04:41,360
Speaker 1: visitors were on the page at the same time, hitting

73
00:04:41,400 --> 00:04:45,599
Speaker 1: almost seven hundred thousand. The number of comments on the

74
00:04:45,640 --> 00:04:50,200
Speaker 1: original post increases tenfold that night. By late that night,

75
00:04:50,279 --> 00:04:56,120
Speaker 1: there are five thousand tweets per minute using hashtag the dress,

76
00:04:56,320 --> 00:04:58,640
Speaker 1: and by the middle of that night it's grown to

77
00:04:58,760 --> 00:05:02,560
Speaker 1: eleven thousand tw wheets per minute. Within the week, more

78
00:05:02,600 --> 00:05:06,400
Speaker 1: than ten million tweets are talking about the dress. This

79
00:05:06,680 --> 00:05:11,840
Speaker 1: was the dress that, as they say, broke the Internet. Now,

80
00:05:11,920 --> 00:05:15,360
Speaker 1: if you were, say a space alien, you might look

81
00:05:15,400 --> 00:05:18,920
Speaker 1: at all this human activity and think, wait, what, why

82
00:05:19,080 --> 00:05:22,440
Speaker 1: is the world stopping over a simple picture of a

83
00:05:22,560 --> 00:05:26,479
Speaker 1: piece of clothing in the UK. Now, the answer, as

84
00:05:26,520 --> 00:05:28,640
Speaker 1: you know, is that none of us humans would have

85
00:05:28,680 --> 00:05:33,040
Speaker 1: found it interesting either, except that someone that you loved

86
00:05:33,040 --> 00:05:35,720
Speaker 1: and trusted said, what do you mean you're seeing it

87
00:05:35,800 --> 00:05:39,880
Speaker 1: that color? It's so clearly the other color, And you said, wait,

88
00:05:39,960 --> 00:05:42,599
Speaker 1: what are you being serious? And they asked you the

89
00:05:42,640 --> 00:05:48,080
Speaker 1: same and then the awe sets in. You both realize

90
00:05:48,120 --> 00:05:51,320
Speaker 1: that you're looking at the same thing in the outside world,

91
00:05:51,680 --> 00:05:58,960
Speaker 1: and you're having different perceptions a different experience on the inside. Now,

92
00:05:59,160 --> 00:06:02,880
Speaker 1: no one was more excited about the dress than neuroscientists,

93
00:06:02,920 --> 00:06:08,000
Speaker 1: because for neuroscientists this was a terrific demonstration of what

94
00:06:08,040 --> 00:06:11,400
Speaker 1: we're going to talk about today. So to start things off,

95
00:06:11,520 --> 00:06:16,920
Speaker 1: let's just point out how important these kinds of perceptual

96
00:06:17,000 --> 00:06:20,760
Speaker 1: oddities are to neuroscience. I've spent a big chunk of

97
00:06:20,800 --> 00:06:26,760
Speaker 1: my career studying illusions. I've published scientific papers about illusions

98
00:06:26,800 --> 00:06:30,280
Speaker 1: in journals like Science and Nature, And some years ago

99
00:06:30,360 --> 00:06:33,960
Speaker 1: I wrote a review article in the journal Nature Reviews Neuroscience,

100
00:06:34,240 --> 00:06:38,400
Speaker 1: and I titled it Visual Illusions and the Brain, And

101
00:06:38,440 --> 00:06:42,760
Speaker 1: in that article I laid out how powerful illusions are

102
00:06:43,000 --> 00:06:46,599
Speaker 1: for figuring out what is under the hood. Sometimes I

103
00:06:46,600 --> 00:06:49,520
Speaker 1: feel like illusions are interesting only to ten year olds

104
00:06:49,560 --> 00:06:53,880
Speaker 1: and for most people they become nothing but entertainment. But truthfully,

105
00:06:54,040 --> 00:06:59,040
Speaker 1: illusions are microscopes for understanding what is happening in the brain.

106
00:07:00,200 --> 00:07:04,640
Speaker 1: Them we can reveal the systematic differences between what is

107
00:07:04,839 --> 00:07:07,719
Speaker 1: actually out there in the world and what we believe

108
00:07:08,160 --> 00:07:12,600
Speaker 1: is out there, And by dialing the illusion around carefully,

109
00:07:12,960 --> 00:07:16,840
Speaker 1: we can usually put constraints on how the network of

110
00:07:16,960 --> 00:07:22,440
Speaker 1: neurons must be operating. Now, most illusions are the type

111
00:07:22,440 --> 00:07:26,080
Speaker 1: in which we measure what's being presented in the outside world,

112
00:07:26,240 --> 00:07:30,080
Speaker 1: like two lines of identical lengths and you see it

113
00:07:30,160 --> 00:07:33,440
Speaker 1: as two different lengths, and we say, ah, there's a

114
00:07:33,480 --> 00:07:37,400
Speaker 1: systematic difference between what's on the page and what you perceive.

115
00:07:38,040 --> 00:07:40,000
Speaker 2: Or maybe I show you two.

116
00:07:39,720 --> 00:07:43,520
Speaker 1: Parallel lines against some background and you don't see them

117
00:07:43,640 --> 00:07:47,160
Speaker 1: as parallel. Or you look at a totally static picture

118
00:07:47,200 --> 00:07:50,920
Speaker 1: on a page and you swear that it's moving. But

119
00:07:51,000 --> 00:07:54,320
Speaker 1: the dress was interesting because it wasn't that traditional kind

120
00:07:54,360 --> 00:07:58,800
Speaker 1: of illusion. Instead, one person sees one thing and the

121
00:07:58,880 --> 00:08:01,480
Speaker 1: person standing right next to them sees another.

122
00:08:02,640 --> 00:08:04,160
Speaker 2: Now, what all.

123
00:08:03,880 --> 00:08:06,800
Speaker 1: Illusions, including the dress, tell us right away is a

124
00:08:07,000 --> 00:08:11,160
Speaker 1: foundational point that's not always intuitive, which is that we

125
00:08:11,240 --> 00:08:15,640
Speaker 1: don't simply look at the world and passively receive what's

126
00:08:15,680 --> 00:08:22,240
Speaker 1: out there. Instead, our brains actively construct our perception, and

127
00:08:22,640 --> 00:08:26,480
Speaker 1: different brains can do so differently. So now let's move

128
00:08:26,600 --> 00:08:30,680
Speaker 1: deeper into this mystery by turning to a different illusion.

129
00:08:31,000 --> 00:08:33,480
Speaker 1: That took over the Internet a few years later, in

130
00:08:33,559 --> 00:08:35,240
Speaker 1: May of twenty eighteen.

131
00:08:36,120 --> 00:08:40,040
Speaker 3: Laurel Laurel, Laurel.

132
00:08:41,080 --> 00:08:44,520
Speaker 1: Now, this was an audio file that was originally recorded

133
00:08:44,559 --> 00:08:47,800
Speaker 1: by a reader in two thousand and seven for vocabulary

134
00:08:47,840 --> 00:08:51,360
Speaker 1: dot com, and some students apparently re recorded that file

135
00:08:51,440 --> 00:08:54,679
Speaker 1: while there was some background noise in a room. So

136
00:08:55,280 --> 00:08:59,360
Speaker 1: a fifteen year old freshman in Georgia named Katie was

137
00:08:59,440 --> 00:09:03,120
Speaker 1: listening to that recording and she realized that she was

138
00:09:03,200 --> 00:09:07,560
Speaker 1: hearing some funny ambiguity, and she posted this little audio

139
00:09:07,600 --> 00:09:10,440
Speaker 1: clip on Instagram, and the next day her friend posted

140
00:09:10,440 --> 00:09:12,400
Speaker 1: it on Reddit, and then it got picked up on

141
00:09:12,440 --> 00:09:13,280
Speaker 1: Twitter and.

142
00:09:13,320 --> 00:09:14,800
Speaker 2: Soon it went nuts.

143
00:09:15,440 --> 00:09:20,120
Speaker 1: Why Because just like the dress, people can have a

144
00:09:20,320 --> 00:09:25,040
Speaker 1: different perception of the same item presented to their senses.

145
00:09:25,559 --> 00:09:29,240
Speaker 1: About half the people hear the word yanny and the

146
00:09:29,400 --> 00:09:32,200
Speaker 1: other half hear the word Laurel.

147
00:09:32,800 --> 00:09:38,600
Speaker 3: Laurel, Laurel, Laurel, Laurel.

148
00:09:39,840 --> 00:09:44,640
Speaker 1: Now, how can people hear different things? So hang tight,

149
00:09:44,720 --> 00:09:46,360
Speaker 1: I'll tell you in a minute. But what I want

150
00:09:46,400 --> 00:09:48,720
Speaker 1: to point out for now is that, just like the dress,

151
00:09:48,760 --> 00:09:52,040
Speaker 1: some people have one experience, some people have another, same

152
00:09:52,160 --> 00:09:57,920
Speaker 1: sound recording, different experiences. Now, the Yanny Laurel clip made

153
00:09:57,960 --> 00:10:00,960
Speaker 1: its rounds on the internet, but it about the exact

154
00:10:01,040 --> 00:10:05,040
Speaker 1: same time. In May of twenty eighteen, something even better

155
00:10:05,120 --> 00:10:09,920
Speaker 1: surfaced on YouTube. A guy had posted a video where

156
00:10:09,960 --> 00:10:14,199
Speaker 1: he was reviewing a children's toy from the ben Ten franchise,

157
00:10:14,679 --> 00:10:18,600
Speaker 1: and the toy lights up and says something. And here's

158
00:10:18,640 --> 00:10:22,320
Speaker 1: what it sounds like. It says the word green needle.

159
00:10:22,600 --> 00:10:34,080
Speaker 1: So listen carefully for green needle. Okay, well, that's not

160
00:10:34,280 --> 00:10:37,079
Speaker 1: actually what the toy was saying. It was actually saying

161
00:10:37,120 --> 00:10:41,600
Speaker 1: the word brainstorm, which is the toy character's name. So

162
00:10:41,760 --> 00:10:52,320
Speaker 1: listen for the word brainstorm.

163
00:10:52,440 --> 00:10:53,079
Speaker 2: Now, I just.

164
00:10:53,040 --> 00:10:56,840
Speaker 1: Played the exact same audio file in both cases, but

165
00:10:56,960 --> 00:11:01,800
Speaker 1: depending on your expectation what you were listening for, you'll

166
00:11:01,920 --> 00:11:05,400
Speaker 1: hear different things. So I'm going to play this file again,

167
00:11:05,840 --> 00:11:08,160
Speaker 1: over and over for about twenty seconds, and I want

168
00:11:08,160 --> 00:11:13,240
Speaker 1: you to think about brainstorm or think about green needle.

169
00:11:13,840 --> 00:11:16,840
Speaker 1: Try to go back and forth about which one you're hearing.

170
00:11:17,000 --> 00:11:19,720
Speaker 1: Switch your thinking from one to the other at any point.

171
00:11:39,200 --> 00:11:40,840
Speaker 2: So, what the heck's going on here?

172
00:11:41,000 --> 00:11:45,480
Speaker 1: How can a single audio file be heard two completely

173
00:11:45,480 --> 00:11:51,160
Speaker 1: different ways? Seems like magic, but it's actually neuroscience. All

174
00:11:51,240 --> 00:11:56,679
Speaker 1: these internet memes actually give deep insight into a fundamental

175
00:11:57,080 --> 00:12:00,640
Speaker 1: and rarely appreciated property of the brain. So I'm going

176
00:12:00,679 --> 00:12:04,640
Speaker 1: to unpack these illusions in a few steps. The first

177
00:12:04,720 --> 00:12:07,920
Speaker 1: clue to the mystery is that the brain does not

178
00:12:08,240 --> 00:12:13,280
Speaker 1: tolerate ambiguity. It really wants to come to a conclusion

179
00:12:13,440 --> 00:12:17,800
Speaker 1: about exactly what's out there. Now, that's a major daily

180
00:12:17,920 --> 00:12:20,560
Speaker 1: challenge for the brain because so much of what you

181
00:12:20,640 --> 00:12:25,080
Speaker 1: see or hear is ambiguous. You have data points that

182
00:12:25,120 --> 00:12:28,400
Speaker 1: come streaming into the brain through the eyes, or the ears,

183
00:12:28,480 --> 00:12:32,920
Speaker 1: or the fingertips, but often they could be interpreted more

184
00:12:32,960 --> 00:12:36,120
Speaker 1: than one way. So what does the brain do in

185
00:12:36,160 --> 00:12:41,520
Speaker 1: this circumstance. It locks onto a single way of understanding it.

186
00:12:42,320 --> 00:12:46,320
Speaker 1: In other words, if there are multiple possibilities, it'll force

187
00:12:46,440 --> 00:12:49,840
Speaker 1: an answer. Now let's pause for just a moment to

188
00:12:49,880 --> 00:12:53,880
Speaker 1: appreciate something here. When you read about the brain, you

189
00:12:53,960 --> 00:12:57,920
Speaker 1: always see it celebrated for its parallel processing. It can

190
00:12:58,000 --> 00:13:01,280
Speaker 1: do lots of things at once. But what it should

191
00:13:01,280 --> 00:13:04,240
Speaker 1: be equally celebrated for, the thing that no one ever

192
00:13:04,320 --> 00:13:10,280
Speaker 1: bothers to highlight is serialization. It takes lots of the

193
00:13:10,360 --> 00:13:13,840
Speaker 1: activity and it squeezes it down to one thing.

194
00:13:14,080 --> 00:13:15,559
Speaker 2: It serializes it.

195
00:13:15,559 --> 00:13:19,240
Speaker 1: It takes an information that could be interpreted in lots

196
00:13:19,240 --> 00:13:22,160
Speaker 1: of different ways, and it crunches it down to a

197
00:13:22,320 --> 00:13:23,520
Speaker 1: single interpretation.

198
00:13:24,800 --> 00:13:28,760
Speaker 2: Now, why is it so good at serializing, at.

199
00:13:28,600 --> 00:13:33,800
Speaker 1: Getting possibilities down to a single answer, Because fundamentally, your

200
00:13:33,800 --> 00:13:38,240
Speaker 1: brain has the challenge of controlling a giant body made

201
00:13:38,280 --> 00:13:41,800
Speaker 1: of trillions of cells, and when you come to a

202
00:13:42,320 --> 00:13:45,880
Speaker 1: tree in the path, it has to go either left

203
00:13:45,960 --> 00:13:48,719
Speaker 1: or right around the tree. Because of the physics of

204
00:13:48,760 --> 00:13:51,040
Speaker 1: the world, it cannot do both, and.

205
00:13:50,960 --> 00:13:53,360
Speaker 2: So it has to make a single.

206
00:13:53,400 --> 00:13:57,520
Speaker 1: Decision, go right or go left, and drag all those

207
00:13:57,559 --> 00:14:00,560
Speaker 1: trillions of cells with it. Your brain it has to

208
00:14:00,640 --> 00:14:04,960
Speaker 1: be good at taking possibilities and crushing them down to

209
00:14:05,080 --> 00:14:10,880
Speaker 1: a single decision. And it's the same with your perceptual life.

210
00:14:11,360 --> 00:14:14,600
Speaker 1: Your brain is used to dealing with a world where

211
00:14:14,640 --> 00:14:18,120
Speaker 1: it has to come to conclusions, having to say, look,

212
00:14:18,160 --> 00:14:21,760
Speaker 1: there are lots of possibilities here, but for me to

213
00:14:21,840 --> 00:14:24,520
Speaker 1: function in the world, I have to make an assumption

214
00:14:25,000 --> 00:14:27,600
Speaker 1: that what I am looking at is a piece of

215
00:14:27,640 --> 00:14:30,960
Speaker 1: food or a boulder, or a bear at a distance

216
00:14:31,120 --> 00:14:36,000
Speaker 1: or whatever. So the brain doesn't tolerate ambiguity, but it

217
00:14:36,080 --> 00:14:40,480
Speaker 1: always says, all right, this is my answer okay, So

218
00:14:40,680 --> 00:14:45,120
Speaker 1: now let's introduce one more perceptual illusion of this flavor,

219
00:14:45,600 --> 00:14:48,000
Speaker 1: and then we're going to unpack what's going on.

220
00:14:49,480 --> 00:14:51,080
Speaker 2: So surely you've seen this one before.

221
00:14:51,200 --> 00:14:54,080
Speaker 1: You draw the outline of a cube on a piece

222
00:14:54,120 --> 00:14:57,240
Speaker 1: of paper. You just draw a square, and then an

223
00:14:57,320 --> 00:15:00,720
Speaker 1: offset square, and then lines connecting the corners of one

224
00:15:00,760 --> 00:15:03,320
Speaker 1: to the corners of the other, so it's twelve lines.

225
00:15:03,400 --> 00:15:07,400
Speaker 1: It's the outline of a cube. This little wireframe drawing

226
00:15:07,600 --> 00:15:09,920
Speaker 1: is known as the Necker cube.

227
00:15:10,320 --> 00:15:13,479
Speaker 2: Now you've seen this before, but as you know, if you've.

228
00:15:13,280 --> 00:15:18,080
Speaker 1: Stared at one, it's perceptually ambiguous because if you stare

229
00:15:18,120 --> 00:15:21,320
Speaker 1: at this little wireframe, it looks like it's coming out

230
00:15:21,400 --> 00:15:25,400
Speaker 1: one way from the page, even though you could perceive.

231
00:15:25,080 --> 00:15:27,400
Speaker 2: The same drawing in two different ways.

232
00:15:27,720 --> 00:15:30,600
Speaker 1: Either the lower square is the face of the cube

233
00:15:30,640 --> 00:15:33,720
Speaker 1: coming toward you, or the upper square is the one

234
00:15:33,760 --> 00:15:38,280
Speaker 1: coming out toward you, but your brain makes a choice. Now,

235
00:15:38,320 --> 00:15:42,360
Speaker 1: you could imagine a space alien who looks at this

236
00:15:42,400 --> 00:15:45,600
Speaker 1: little drawing of the wireframe cube and says, okay, well,

237
00:15:46,040 --> 00:15:50,240
Speaker 1: both configurations of the cube are equally probable, so I'll

238
00:15:50,240 --> 00:15:53,640
Speaker 1: see it both ways at once. But we can't do that.

239
00:15:54,200 --> 00:15:56,960
Speaker 1: We have to see it one way or the other.

240
00:15:57,120 --> 00:16:01,800
Speaker 1: Your brain forces a single interpret and this is the

241
00:16:01,840 --> 00:16:06,280
Speaker 1: same thing that's happening with the other illusions with the dress.

242
00:16:06,400 --> 00:16:09,640
Speaker 1: You don't see it as both blue and black and

243
00:16:09,880 --> 00:16:12,520
Speaker 1: white and gold. And in a minute we'll see why.

244
00:16:13,160 --> 00:16:14,680
Speaker 1: The part I just want to say now is that

245
00:16:14,720 --> 00:16:17,960
Speaker 1: your brain concludes that it is one or the other,

246
00:16:18,080 --> 00:16:22,440
Speaker 1: and then it sticks with that. And likewise with Yanny Laurel.

247
00:16:23,080 --> 00:16:26,720
Speaker 1: Both sounds are present in the audio file, but you

248
00:16:26,840 --> 00:16:30,640
Speaker 1: don't hear Yanny and Laurel at the same time, stacked

249
00:16:30,680 --> 00:16:33,800
Speaker 1: on one another. And it's exactly the same thing with

250
00:16:33,960 --> 00:16:39,440
Speaker 1: brainstorm and green needle. Both interpretations are possible, but your

251
00:16:39,480 --> 00:16:43,920
Speaker 1: brain won't do both at once. It collapses the possibilities

252
00:16:43,960 --> 00:16:48,600
Speaker 1: to a single answer. In all these cases, even though

253
00:16:48,640 --> 00:16:52,160
Speaker 1: the data is consistent with either interpretation, your brain makes

254
00:16:52,160 --> 00:16:55,280
Speaker 1: a call. It goes left or right around the tree.

255
00:16:55,560 --> 00:16:58,800
Speaker 1: You very clearly perceive one or the other. And this

256
00:16:58,920 --> 00:17:03,040
Speaker 1: is because the brain isn't passively receiving the world. It's

257
00:17:03,200 --> 00:17:25,000
Speaker 1: making choices. Okay, but how does your brain know how

258
00:17:25,040 --> 00:17:30,119
Speaker 1: to collapse ambiguous data to a single interpretation. It does

259
00:17:30,160 --> 00:17:35,760
Speaker 1: so by leveraging assumptions, so let's go a level deeper

260
00:17:35,880 --> 00:17:38,920
Speaker 1: with the dress. Why does it happen that some people

261
00:17:38,960 --> 00:17:41,359
Speaker 1: see it one way and some people the other. It

262
00:17:41,520 --> 00:17:44,560
Speaker 1: happens because your brain sees a picture of a dress

263
00:17:44,560 --> 00:17:50,120
Speaker 1: in the shop and it makes dozens of assumptions totally unconsciously. Now,

264
00:17:50,160 --> 00:17:54,600
Speaker 1: what's amazing is that the assumptions aren't directly about the dress,

265
00:17:55,240 --> 00:17:58,760
Speaker 1: but about things you didn't even know you were thinking about.

266
00:17:58,800 --> 00:18:03,280
Speaker 1: What is the light source in the photograph? Is the

267
00:18:03,440 --> 00:18:07,840
Speaker 1: dress mostly being lit by fluorescent lights or by sunlight?

268
00:18:08,760 --> 00:18:12,240
Speaker 1: Is the dress facing a window or is the window

269
00:18:12,280 --> 00:18:15,840
Speaker 1: behind it? What time of day is it, what season

270
00:18:16,000 --> 00:18:20,320
Speaker 1: is it? Your brain is considering all of these questions,

271
00:18:20,840 --> 00:18:23,720
Speaker 1: and fundamentally, this all has to do with a computation

272
00:18:23,840 --> 00:18:30,080
Speaker 1: that it does known as color constancy. Color constancy is

273
00:18:30,160 --> 00:18:34,960
Speaker 1: this sophisticated ability of our visual systems to perceive the

274
00:18:35,000 --> 00:18:39,000
Speaker 1: color of something as constant even when the light source

275
00:18:39,040 --> 00:18:43,119
Speaker 1: the illumination changes. So let's say I'm wearing a white

276
00:18:43,280 --> 00:18:46,160
Speaker 1: T shirt and we're standing outside talking in the sunlight.

277
00:18:46,560 --> 00:18:50,120
Speaker 1: You will see my shirt as white. Now we go

278
00:18:50,280 --> 00:18:54,479
Speaker 1: indoors into the coffee shop and the illuminant changes. In

279
00:18:54,480 --> 00:18:57,720
Speaker 1: other words, the light that's bouncing off my t shirt changes.

280
00:18:58,359 --> 00:19:03,280
Speaker 1: Now it's fluorescent light compared to sunlight. The fluorescent light

281
00:19:03,359 --> 00:19:06,679
Speaker 1: has a different spectrum of colors coming out, and so

282
00:19:06,720 --> 00:19:09,760
Speaker 1: when those bounce off my shirt, you have a different

283
00:19:09,960 --> 00:19:14,600
Speaker 1: spectrum of colors hitting your eyes, and yet you still

284
00:19:14,600 --> 00:19:17,800
Speaker 1: see it as white. And then that night we go

285
00:19:17,920 --> 00:19:21,680
Speaker 1: into a dance club and the lighting is blue, and

286
00:19:21,760 --> 00:19:25,280
Speaker 1: yet you have no problem seeing the shirt as white,

287
00:19:25,600 --> 00:19:28,960
Speaker 1: even though it's mostly blue light reflecting off the shirt

288
00:19:29,240 --> 00:19:32,520
Speaker 1: into your eyes. And then afterwards we go sit by

289
00:19:32,640 --> 00:19:37,399
Speaker 1: a campfire and my shirt still looks white. Your brain

290
00:19:37,520 --> 00:19:41,359
Speaker 1: retains a constant perception of the color of the shirt

291
00:19:41,800 --> 00:19:45,240
Speaker 1: even though the wavelengths bouncing off of it are very different.

292
00:19:46,320 --> 00:19:50,159
Speaker 1: So what does this tell us, Well, it means that

293
00:19:50,240 --> 00:19:53,560
Speaker 1: the way your brain determines the color is not just

294
00:19:53,680 --> 00:19:56,560
Speaker 1: about the colors hitting your eye from the shirt. It

295
00:19:56,600 --> 00:19:59,720
Speaker 1: has to do with something else. And that's something else

296
00:20:00,119 --> 00:20:04,560
Speaker 1: is everything else in the scene. So when you're looking

297
00:20:04,600 --> 00:20:08,480
Speaker 1: at my shirt, your eyes are drinking in everything else.

298
00:20:09,119 --> 00:20:12,840
Speaker 1: The background, the color of the skin on my arms,

299
00:20:12,880 --> 00:20:17,000
Speaker 1: the color of the floors and walls, the colors of

300
00:20:17,359 --> 00:20:18,120
Speaker 1: all the other.

301
00:20:18,359 --> 00:20:20,680
Speaker 2: Jeans and shirts and signs in the.

302
00:20:20,640 --> 00:20:24,720
Speaker 4: Whole scene, and it uses all of that to estimate

303
00:20:24,800 --> 00:20:30,160
Speaker 4: the background illumination and then make the right computation about

304
00:20:30,200 --> 00:20:34,199
Speaker 4: the color of the shirt in the sunlight and the

305
00:20:34,280 --> 00:20:36,959
Speaker 4: coffee shop, at the dance club, at the campfire.

306
00:20:37,359 --> 00:20:38,520
Speaker 2: It's doing all of.

307
00:20:38,480 --> 00:20:43,080
Speaker 1: These computations, and this is what allows it to subtract

308
00:20:43,320 --> 00:20:46,840
Speaker 1: off the background lighting so that it can see what

309
00:20:47,040 --> 00:20:52,040
Speaker 1: color things are most likely to actually be. That's the

310
00:20:52,080 --> 00:20:56,760
Speaker 1: phenomenon of color constancy. The color of the shirt remains

311
00:20:56,840 --> 00:21:01,920
Speaker 1: constant even under different illumination, and that's what allows us

312
00:21:01,920 --> 00:21:05,720
Speaker 1: to see the colors of objects in the world consistently,

313
00:21:05,760 --> 00:21:10,120
Speaker 1: whether we're looking under sunlight or moonlight or firelight or whatever.

314
00:21:11,000 --> 00:21:14,480
Speaker 1: So the first lesson is you're not just seeing what's

315
00:21:14,600 --> 00:21:19,359
Speaker 1: out there. Your brain is actively interpreting information and serving

316
00:21:19,480 --> 00:21:22,040
Speaker 1: up a story to you. And I'll go into this

317
00:21:22,080 --> 00:21:24,399
Speaker 1: more in a future episode. But this is why we

318
00:21:24,440 --> 00:21:29,600
Speaker 1: can see strawberries as red. For example, when we change

319
00:21:29,600 --> 00:21:32,920
Speaker 1: the background color such that the actual light bouncing off

320
00:21:32,920 --> 00:21:37,440
Speaker 1: the strawberries is gray light, your brain can nonetheless say, okay, well,

321
00:21:37,640 --> 00:21:40,880
Speaker 1: given that everything else in the scene is now greenish,

322
00:21:41,400 --> 00:21:44,160
Speaker 1: I can subtrack that off and know that I'm looking

323
00:21:44,200 --> 00:21:48,080
Speaker 1: at something red. Now, in order to do all of

324
00:21:48,119 --> 00:21:50,960
Speaker 1: this that I've been talking about, your brain has to

325
00:21:51,040 --> 00:21:55,439
Speaker 1: make lots of assumptions about what the color should be,

326
00:21:56,200 --> 00:22:01,239
Speaker 1: and different brains do it differently. With the dress, you

327
00:22:01,320 --> 00:22:05,679
Speaker 1: see it as either white and gold or blue and black,

328
00:22:06,240 --> 00:22:10,680
Speaker 1: depending on the assumptions your brain is making. When you

329
00:22:10,840 --> 00:22:14,000
Speaker 1: glance at the photo on your phone, you have no

330
00:22:14,119 --> 00:22:17,720
Speaker 1: idea that your brain is doing all those sophisticated computations

331
00:22:17,840 --> 00:22:21,359
Speaker 1: under the hood to tell you what is the actual

332
00:22:21,440 --> 00:22:25,760
Speaker 1: color of this garment, given my assumptions about all the

333
00:22:25,880 --> 00:22:27,040
Speaker 1: lighting details.

334
00:22:27,760 --> 00:22:29,240
Speaker 2: The issue is that your.

335
00:22:29,000 --> 00:22:32,520
Speaker 1: Brain grew up in a particular environment, maybe with a

336
00:22:32,560 --> 00:22:35,439
Speaker 1: lot of snow or a lot of sunlight or fog,

337
00:22:36,119 --> 00:22:39,359
Speaker 1: and your brain makes assumptions about the time of day

338
00:22:39,720 --> 00:22:43,199
Speaker 1: and the season and the balance of artificial lighting to

339
00:22:43,320 --> 00:22:47,720
Speaker 1: natural lighting. To make sense of this little photo, what

340
00:22:48,040 --> 00:22:52,199
Speaker 1: hues does the lighting contain. If your brain ignores a

341
00:22:52,200 --> 00:22:54,920
Speaker 1: bit of the blue side, you'll see the dress as

342
00:22:55,040 --> 00:22:58,159
Speaker 1: white and gold. If your brain pays less attention to

343
00:22:58,200 --> 00:23:00,840
Speaker 1: the yellow side of the spectrum, you'll see it as

344
00:23:01,040 --> 00:23:05,000
Speaker 1: blue and black. You have no insight into the fact

345
00:23:05,240 --> 00:23:08,560
Speaker 1: that your brain is making all these assumptions under the hood.

346
00:23:09,280 --> 00:23:11,919
Speaker 1: Was the photo of the dress taken with the window

347
00:23:11,960 --> 00:23:13,120
Speaker 1: facing it or behind it.

348
00:23:13,200 --> 00:23:15,080
Speaker 2: Was it morning light or afternoon light?

349
00:23:15,760 --> 00:23:18,119
Speaker 1: And is your experience of the world based on the

350
00:23:18,160 --> 00:23:21,919
Speaker 1: fact that you are a mourning lark or you are.

351
00:23:21,800 --> 00:23:22,800
Speaker 2: A night owl.

352
00:23:23,080 --> 00:23:26,600
Speaker 1: One of my colleagues, Pascal Wallash, showed that people who

353
00:23:26,640 --> 00:23:30,119
Speaker 1: were early risers were more likely to think that the

354
00:23:30,200 --> 00:23:33,600
Speaker 1: dress was lit by natural light, and so they saw

355
00:23:33,640 --> 00:23:37,840
Speaker 1: it as white and gold, but night owls presumably had

356
00:23:37,880 --> 00:23:41,720
Speaker 1: more assumptions about artificial lighting, and they were more likely

357
00:23:41,760 --> 00:23:45,399
Speaker 1: to see the dress as blue and black. Your brain

358
00:23:45,560 --> 00:23:48,159
Speaker 1: is determining the color of the dress by comparing it

359
00:23:48,280 --> 00:23:51,240
Speaker 1: against the other objects of the background of the photo

360
00:23:51,640 --> 00:23:56,520
Speaker 1: and making its best guess at all these parameters. So

361
00:23:56,920 --> 00:24:00,320
Speaker 1: your brain relies on the answers to question is that

362
00:24:00,359 --> 00:24:03,439
Speaker 1: you didn't even think it was asking and the idea

363
00:24:03,600 --> 00:24:07,600
Speaker 1: of imposing assumptions. This is the same with Yanny and

364
00:24:07,680 --> 00:24:11,840
Speaker 1: Laurel in the auditory domain, or with green needle and brainstorm.

365
00:24:12,040 --> 00:24:16,840
Speaker 1: Your brain is imposing an interpretation. But what's interesting in

366
00:24:16,880 --> 00:24:20,640
Speaker 1: this case is that the assumption can be changed more easily,

367
00:24:21,000 --> 00:24:26,280
Speaker 1: typically by just staring at the word visually. Because your

368
00:24:26,400 --> 00:24:31,280
Speaker 1: brain is trying to disambiguate what it's hearing, and suddenly

369
00:24:31,520 --> 00:24:34,400
Speaker 1: it has lots of help from the visual system because

370
00:24:34,600 --> 00:24:39,800
Speaker 1: it sees a word. So the frequencies of both words

371
00:24:39,880 --> 00:24:43,800
Speaker 1: yany and laurel or green needle, brainstorm, they're contained in

372
00:24:43,920 --> 00:24:48,240
Speaker 1: the audio file, so just depending on how you listen

373
00:24:48,359 --> 00:24:52,560
Speaker 1: for it, you can hear one or the other. So

374
00:24:52,600 --> 00:24:57,120
Speaker 1: the brain constantly nails down its world by making assumptions,

375
00:24:57,640 --> 00:25:00,640
Speaker 1: and we see this with everything. And even though these

376
00:25:01,200 --> 00:25:04,199
Speaker 1: internet memes get all of our attention, the fact is

377
00:25:04,600 --> 00:25:07,560
Speaker 1: that our brains have to make assumptions all the time.

378
00:25:07,880 --> 00:25:10,840
Speaker 1: And this is because most of the inputs from the

379
00:25:10,880 --> 00:25:16,160
Speaker 1: world are quite noisy. For example, you can still understand

380
00:25:16,280 --> 00:25:20,120
Speaker 1: me even if my speech is choppy, or if I'm

381
00:25:20,160 --> 00:25:23,399
Speaker 1: speaking and there's lots of background noise like at a restaurant.

382
00:25:24,160 --> 00:25:27,200
Speaker 1: What's actually hitting your ears in these scenarios is a

383
00:25:27,320 --> 00:25:31,960
Speaker 1: very messy signal, But the brain imposes an interpretation about

384
00:25:32,200 --> 00:25:34,920
Speaker 1: what must have been said, and that's what you perceive

385
00:25:35,440 --> 00:25:38,800
Speaker 1: what you believe you heard. A lot of your cell

386
00:25:38,880 --> 00:25:43,320
Speaker 1: phone conversations are super noisy, but you typically don't realize

387
00:25:43,320 --> 00:25:48,919
Speaker 1: it because you keep making your reasonable interpretations. Now, this

388
00:25:49,040 --> 00:25:52,880
Speaker 1: is true of most of what is hitting your eyes

389
00:25:52,920 --> 00:25:56,399
Speaker 1: and ears. You don't catch a fraction of the data,

390
00:25:56,800 --> 00:25:59,439
Speaker 1: but your brain fills in the details to put together

391
00:25:59,480 --> 00:26:01,959
Speaker 1: a story. And this, by the way, is what's at

392
00:26:02,000 --> 00:26:04,680
Speaker 1: the heart of a lot of art and graphic design.

393
00:26:05,040 --> 00:26:08,000
Speaker 1: You just see a few curves and you interpret it

394
00:26:08,080 --> 00:26:11,720
Speaker 1: as a face, or a series of segmented lines and

395
00:26:11,800 --> 00:26:16,440
Speaker 1: you interpret that as a body. We are always operating

396
00:26:16,480 --> 00:26:19,919
Speaker 1: off thin data, but that doesn't stop us from coming

397
00:26:19,960 --> 00:26:24,440
Speaker 1: to clear conclusions. And before I explain how our neural

398
00:26:24,480 --> 00:26:28,040
Speaker 1: networks go about making these assumptions, let's just take a

399
00:26:28,080 --> 00:26:31,919
Speaker 1: second to look at how your brain is so imperfect

400
00:26:32,040 --> 00:26:36,919
Speaker 1: at this. Take paridolia, which is when you perceive a

401
00:26:37,119 --> 00:26:41,160
Speaker 1: meaningful pattern where none exists, Like when you look at

402
00:26:41,200 --> 00:26:44,760
Speaker 1: an electrical outlet and you see a face made up

403
00:26:44,760 --> 00:26:48,080
Speaker 1: of little eyes in a sort of surprised mouth. You

404
00:26:48,119 --> 00:26:51,760
Speaker 1: can't help but see that. Your brain imposes that interpretation

405
00:26:51,880 --> 00:26:54,920
Speaker 1: on it. Or you see a face in the clouds,

406
00:26:55,359 --> 00:26:58,840
Speaker 1: or someone sees the face of their local deity in

407
00:26:58,920 --> 00:27:02,960
Speaker 1: a piece of toast. Why does this happen, Well, your

408
00:27:03,000 --> 00:27:06,199
Speaker 1: brain is really wired up to see faces, and so

409
00:27:06,359 --> 00:27:10,960
Speaker 1: it triggers that interpretation whenever it sees three blobs in

410
00:27:11,040 --> 00:27:15,560
Speaker 1: the approximately right configuration, and the same thing can happen

411
00:27:15,640 --> 00:27:18,600
Speaker 1: with sounds, like when there's some weird sound and your

412
00:27:18,600 --> 00:27:22,680
Speaker 1: brain thinks it's a person shouting or someone calling your

413
00:27:22,760 --> 00:27:26,600
Speaker 1: name or whatever. This is the brain working to make

414
00:27:26,760 --> 00:27:29,560
Speaker 1: sense of the world around it. All it ever does

415
00:27:30,160 --> 00:27:34,800
Speaker 1: is look for meaning from data in the world. In fact,

416
00:27:35,640 --> 00:27:39,440
Speaker 1: typically the brain will try to impose an interpretation even

417
00:27:39,480 --> 00:27:44,080
Speaker 1: if you have random noise. That's the idea with rorshak

418
00:27:44,200 --> 00:27:47,560
Speaker 1: ink blots. You have these blobs on a page, and

419
00:27:47,640 --> 00:27:51,760
Speaker 1: your brain reaches for some way of explaining them. Oh,

420
00:27:51,800 --> 00:27:54,800
Speaker 1: that looks like a rabbit or an airplane, or an

421
00:27:54,800 --> 00:27:58,640
Speaker 1: emperor on a throne or whatever. And generally a lot

422
00:27:58,680 --> 00:28:03,800
Speaker 1: of life involves forcing patterns on random noise. Here's an

423
00:28:03,880 --> 00:28:07,520
Speaker 1: auditory example from my colleague Diana Deutsch, who has spent

424
00:28:07,600 --> 00:28:12,480
Speaker 1: her career pioneering auditory illusions. So here's an experiment where

425
00:28:12,520 --> 00:28:16,639
Speaker 1: she plays mixed up audio that doesn't say anything, but

426
00:28:17,080 --> 00:28:21,320
Speaker 1: it sounds like speech, and people will generally impose the

427
00:28:21,400 --> 00:28:23,880
Speaker 1: interpretation of words on these.

428
00:28:57,280 --> 00:29:01,600
Speaker 4: Come come, come, come, come, come, come, come, come come.

429
00:29:33,000 --> 00:29:38,160
Speaker 1: This is essentially the sound version of Rorschach blots. Different

430
00:29:38,280 --> 00:29:41,520
Speaker 1: people will generally hear different things, and it seems to

431
00:29:41,560 --> 00:29:44,520
Speaker 1: be related to what they are thinking about or what's

432
00:29:44,560 --> 00:29:47,640
Speaker 1: on their mind. So this all reminds us of the

433
00:29:47,880 --> 00:29:52,720
Speaker 1: power of the brain to impose meaning. Just think about

434
00:29:52,760 --> 00:29:56,360
Speaker 1: the situation when you're expecting a friend and you're looking

435
00:29:56,440 --> 00:30:00,240
Speaker 1: around for him at a crowded mall. Everyone looks like him.

436
00:30:00,240 --> 00:30:02,560
Speaker 1: For just a fraction of a second. You look at

437
00:30:02,600 --> 00:30:05,160
Speaker 1: someone's face and you think, oh, that's him, and then

438
00:30:05,560 --> 00:30:09,560
Speaker 1: five hundred milliseconds later, your visual system takes in more

439
00:30:09,640 --> 00:30:13,480
Speaker 1: information and decides out never mind false alarm, And then

440
00:30:13,520 --> 00:30:16,880
Speaker 1: we typically forget that we even thought that. But we

441
00:30:17,040 --> 00:30:20,240
Speaker 1: are expecting to see our friend, and so our brains

442
00:30:20,720 --> 00:30:26,240
Speaker 1: impose that expectation on lots of faces. Okay, so we've

443
00:30:26,400 --> 00:30:31,480
Speaker 1: established that brains take ambiguous signals and squish them down

444
00:30:31,520 --> 00:30:35,800
Speaker 1: to a single interpretation by use of assumptions. And that's

445
00:30:35,840 --> 00:30:38,520
Speaker 1: why we see the dress as one color or the other,

446
00:30:38,960 --> 00:30:42,520
Speaker 1: or we hear brainstorm or a green needle, but not both.

447
00:30:43,760 --> 00:30:46,520
Speaker 1: But how do our brains actually make their choice? How

448
00:30:46,560 --> 00:30:50,800
Speaker 1: do they pull this off? Neurally speaking, they do it

449
00:30:51,000 --> 00:30:57,360
Speaker 1: by combining bottom up information with top down information. Now,

450
00:30:57,760 --> 00:31:01,720
Speaker 1: bottom up means information and from the outside from the world.

451
00:31:02,080 --> 00:31:04,360
Speaker 1: What are the air compression wave of sitting my ear

452
00:31:04,440 --> 00:31:06,560
Speaker 1: drums or what are the photons sitting my retina?

453
00:31:07,080 --> 00:31:09,000
Speaker 2: Those are the signals that I am receiving.

454
00:31:10,000 --> 00:31:12,800
Speaker 1: But we don't interpret those bottom up signals that face

455
00:31:12,960 --> 00:31:18,120
Speaker 1: value because they're usually not sufficient. Instead, your brain melds

456
00:31:18,200 --> 00:31:23,880
Speaker 1: this with top down information, which means your expectations what

457
00:31:24,040 --> 00:31:26,600
Speaker 1: we think is likely to be true in the outside

458
00:31:26,680 --> 00:31:30,640
Speaker 1: world given our experience with it, and it's only in

459
00:31:30,920 --> 00:31:36,240
Speaker 1: combination the data plus our expectations that we see anything

460
00:31:36,400 --> 00:31:40,520
Speaker 1: in the world. And the surprise, I think the counterintuitive

461
00:31:40,600 --> 00:31:45,000
Speaker 1: part is that your prior assumptions, your expectations, the.

462
00:31:45,080 --> 00:31:45,840
Speaker 2: Top down part.

463
00:31:46,240 --> 00:31:50,680
Speaker 1: This is the overwhelming majority of what determines what you see.

464
00:31:51,880 --> 00:31:54,880
Speaker 1: For example, it seems like you just open your eyes

465
00:31:54,920 --> 00:31:57,400
Speaker 1: and there's the world, but in fact, when you look

466
00:31:57,520 --> 00:32:00,720
Speaker 1: at the visual cortex at the back of the which

467
00:32:00,800 --> 00:32:03,200
Speaker 1: is the place that receives the information from the eyes,

468
00:32:04,040 --> 00:32:07,400
Speaker 1: you find that only five percent of the input there

469
00:32:07,880 --> 00:32:09,280
Speaker 1: is coming from the eyes.

470
00:32:09,360 --> 00:32:12,280
Speaker 2: And the rest is all feedback activity.

471
00:32:12,640 --> 00:32:15,400
Speaker 1: In other words, ninety five percent of the data is

472
00:32:15,480 --> 00:32:19,120
Speaker 1: coming from higher levels of the visual system and other

473
00:32:19,400 --> 00:32:23,480
Speaker 1: areas of the brain. In fact, What is so crazy

474
00:32:23,680 --> 00:32:27,320
Speaker 1: is that you don't even need your eyes to have full,

475
00:32:27,960 --> 00:32:32,200
Speaker 1: rich visual experience. You can have this with your eyes closed.

476
00:32:32,680 --> 00:32:37,040
Speaker 1: And this is what we call dreams. And what's happening

477
00:32:37,080 --> 00:32:40,640
Speaker 1: here is that this is all internally generated activity and

478
00:32:40,720 --> 00:32:43,000
Speaker 1: none of it's entering through the eyes when you're asleep,

479
00:32:43,480 --> 00:32:46,440
Speaker 1: and it's not that much different from your normal vision.

480
00:32:47,120 --> 00:32:49,640
Speaker 1: So your perception of the world when you're walking around

481
00:32:49,800 --> 00:32:52,520
Speaker 1: is something like an awake dream.

482
00:32:53,600 --> 00:32:53,719
Speaker 5: Now.

483
00:32:53,760 --> 00:32:56,160
Speaker 1: I'm going to return to this issue in future episodes,

484
00:32:56,600 --> 00:32:59,000
Speaker 1: but what we want to concentrate on right now is

485
00:32:59,080 --> 00:33:02,320
Speaker 1: that your eyes are not simply a camera and your

486
00:33:02,480 --> 00:33:05,640
Speaker 1: ears are not simply a microphone. For those of you

487
00:33:05,720 --> 00:33:08,000
Speaker 1: who have been listening for a while to this podcast,

488
00:33:08,440 --> 00:33:12,040
Speaker 1: you know this is a major theme. Your brain is

489
00:33:12,200 --> 00:33:16,040
Speaker 1: locked in silence and darkness and needs to make assumptions

490
00:33:16,360 --> 00:33:20,800
Speaker 1: based on very thin data. So when I ask you

491
00:33:20,920 --> 00:33:24,880
Speaker 1: to think about the words green needle, that is top

492
00:33:24,960 --> 00:33:29,840
Speaker 1: down information that shapes how you interpret the bottom up data.

493
00:33:37,120 --> 00:33:40,440
Speaker 1: In contrast, imagine that you stare at the word brainstorm

494
00:33:40,640 --> 00:33:44,280
Speaker 1: while listening. You lock that in as your top down expectation,

495
00:33:44,440 --> 00:33:46,760
Speaker 1: and then that shapes your bottom up data.

496
00:33:46,840 --> 00:33:47,760
Speaker 2: And that's what you hear.

497
00:33:55,360 --> 00:34:00,200
Speaker 1: Even though both interpretations are available, your brain surfaces is

498
00:34:00,280 --> 00:34:04,280
Speaker 1: those features out of the landscape of data that match

499
00:34:04,440 --> 00:34:08,920
Speaker 1: what you're looking for. In other words, your expectations. What

500
00:34:09,120 --> 00:34:13,839
Speaker 1: you listen for is what you hear. And by the way,

501
00:34:13,920 --> 00:34:16,719
Speaker 1: all this is related to why lip reading works. When

502
00:34:16,760 --> 00:34:20,000
Speaker 1: you're in a noisy environment, you watch somebody's mouth while

503
00:34:20,000 --> 00:34:23,239
Speaker 1: they're talking, and in this way you combine a bit

504
00:34:23,400 --> 00:34:27,120
Speaker 1: of noisy auditory data with a bit of noisy visual

505
00:34:27,680 --> 00:34:30,560
Speaker 1: and that sharpens your guess for what was just said.

506
00:34:31,320 --> 00:34:35,440
Speaker 1: During the pandemic, a lot of conversations went misunderstood because

507
00:34:35,800 --> 00:34:39,120
Speaker 1: people were wearing masks and lip reading went out the window.

508
00:34:40,280 --> 00:34:44,840
Speaker 1: Now amazingly, this top down information is so important that

509
00:34:45,000 --> 00:34:48,279
Speaker 1: sometimes you can set up a picture where you don't

510
00:34:48,360 --> 00:34:52,480
Speaker 1: have any real prior assumptions and there's not enough information

511
00:34:52,640 --> 00:34:55,680
Speaker 1: in the picture to see anything. And only when I

512
00:34:55,800 --> 00:35:00,360
Speaker 1: tell you some interpretation does the bottom up information should

513
00:35:00,360 --> 00:35:01,319
Speaker 1: make any sense at all.

514
00:35:01,880 --> 00:35:03,120
Speaker 2: You can only see.

515
00:35:03,160 --> 00:35:06,480
Speaker 1: What's in front of you if you're given top down direction.

516
00:35:07,520 --> 00:35:10,880
Speaker 1: For example, I've put a cool picture on my website

517
00:35:10,920 --> 00:35:14,440
Speaker 1: at eagleman dot com slash podcast. Take a look at

518
00:35:14,480 --> 00:35:17,480
Speaker 1: this field of black and white blobs and see what

519
00:35:17,600 --> 00:35:20,520
Speaker 1: it looks like to you, And presumably it looks really

520
00:35:20,680 --> 00:35:23,200
Speaker 1: like nothing at all, just a bunch of blobs. But

521
00:35:23,320 --> 00:35:25,600
Speaker 1: if I tell you what it is while you stare

522
00:35:25,600 --> 00:35:29,600
Speaker 1: at it, then you suddenly see it. It seems immediately

523
00:35:29,760 --> 00:35:32,840
Speaker 1: obvious and you cannot see anything other than that, And

524
00:35:32,920 --> 00:35:35,440
Speaker 1: the only thing that's changed is that you now have

525
00:35:36,080 --> 00:35:40,600
Speaker 1: a top down expectation about what you're seeing, and suddenly

526
00:35:40,760 --> 00:35:43,680
Speaker 1: all these blobs make clear sense. I'm not going to

527
00:35:43,719 --> 00:35:45,360
Speaker 1: tell you what the blobs are here, but if you

528
00:35:45,400 --> 00:35:47,719
Speaker 1: go to the website and scroll all the way to

529
00:35:47,840 --> 00:35:49,799
Speaker 1: the bottom of the page, I'll give you a hint

530
00:35:49,880 --> 00:35:53,240
Speaker 1: there so you can enjoy the experience of not knowing

531
00:35:53,719 --> 00:35:57,560
Speaker 1: and then knowing. And because this is a podcast, I'll

532
00:35:57,600 --> 00:36:01,360
Speaker 1: give you an auditory example of this, again from Diana Deutsch.

533
00:36:01,840 --> 00:36:04,400
Speaker 1: So I'm going to take a piece of music that

534
00:36:04,560 --> 00:36:08,279
Speaker 1: you know, but I'm going to shift each note up

535
00:36:08,520 --> 00:36:11,960
Speaker 1: or down an octave, so one note might be played

536
00:36:12,000 --> 00:36:14,320
Speaker 1: an octave higher and the next note might be played

537
00:36:14,320 --> 00:36:17,520
Speaker 1: an octave lower. And I want you to identify the

538
00:36:17,600 --> 00:36:28,279
Speaker 1: piece of music. It's definitely one that you know. Now

539
00:36:28,360 --> 00:36:32,319
Speaker 1: I assume that you couldn't identify that piece. Now I'm

540
00:36:32,360 --> 00:36:34,760
Speaker 1: going to play it for you without the notes shifted

541
00:36:34,880 --> 00:36:45,880
Speaker 1: up or down in octaves. Now that you know the tune,

542
00:36:46,000 --> 00:36:47,920
Speaker 1: I'm just going to play that first one again and

543
00:36:48,000 --> 00:36:51,560
Speaker 1: you should have little or no trouble hearing the correct melody.

544
00:36:59,760 --> 00:37:01,920
Speaker 2: The only difference between the first.

545
00:37:01,680 --> 00:37:04,080
Speaker 1: Time I played it and the last time is that

546
00:37:04,320 --> 00:37:08,120
Speaker 1: now you have a top down expectation, and so it

547
00:37:08,239 --> 00:37:13,359
Speaker 1: switches from random noise to a tune. And so these

548
00:37:13,400 --> 00:37:18,080
Speaker 1: are all examples in which top down expectations are critical.

549
00:37:18,160 --> 00:37:21,880
Speaker 1: Without them, you don't have any interpretation at all. And

550
00:37:22,080 --> 00:37:26,640
Speaker 1: once you build an expectation, then the data have meaning.

551
00:37:27,640 --> 00:37:29,360
Speaker 1: You need to be told what to see in the

552
00:37:29,440 --> 00:37:31,600
Speaker 1: picture or to hear in the tune to get it.

553
00:37:31,880 --> 00:37:35,680
Speaker 1: And the only difference between before and after is whether

554
00:37:35,760 --> 00:37:39,680
Speaker 1: you have something to match it to some top down expectation,

555
00:37:40,200 --> 00:37:44,960
Speaker 1: and as soon as you do, then you perceive. Now,

556
00:37:45,160 --> 00:37:47,680
Speaker 1: just to be clear, this doesn't mean you can impose

557
00:37:47,920 --> 00:37:52,680
Speaker 1: any top down interpretation. It has to match sufficiently well.

558
00:37:53,280 --> 00:37:56,960
Speaker 1: The thing about brainstorm green needle is that the bottom

559
00:37:57,040 --> 00:38:01,000
Speaker 1: up data can match either one of the top expectations

560
00:38:01,080 --> 00:38:05,680
Speaker 1: for either word. You can hear green needle or brainstorm.

561
00:38:06,120 --> 00:38:10,600
Speaker 1: Because these are possible words that can roughly match the

562
00:38:10,680 --> 00:38:13,600
Speaker 1: bottom up stimulus with all of its noise, But you

563
00:38:13,760 --> 00:38:17,920
Speaker 1: can't hear something totally different like blue reader or my

564
00:38:18,239 --> 00:38:22,320
Speaker 1: penguin because you can't make a good enough match between

565
00:38:22,800 --> 00:38:26,440
Speaker 1: data and expectation. So there has to be a sufficient

566
00:38:26,560 --> 00:38:30,120
Speaker 1: match between the top down and the bottom up for

567
00:38:30,320 --> 00:38:34,040
Speaker 1: perception to happen. Okay, so let's come back.

568
00:38:33,880 --> 00:38:36,799
Speaker 2: To this issue about the assumptions that we make. How

569
00:38:36,920 --> 00:38:39,880
Speaker 2: do we know what to assume about the world.

570
00:38:40,480 --> 00:38:46,920
Speaker 1: Well, this relies almost entirely on our prior experience. For example,

571
00:38:46,960 --> 00:38:50,680
Speaker 1: when you're judging depth, like how far different things are

572
00:38:50,760 --> 00:38:54,000
Speaker 1: from you, which is again a totally unconscious process, you

573
00:38:54,080 --> 00:38:57,000
Speaker 1: can do this by comparing the images from your two eyes,

574
00:38:57,440 --> 00:38:59,680
Speaker 1: but this is only useful out to about thirty meters.

575
00:39:00,120 --> 00:39:03,000
Speaker 1: So it turns out the brain has other ways to

576
00:39:03,160 --> 00:39:06,840
Speaker 1: determine depth, and one of the main ones simply pivots

577
00:39:06,920 --> 00:39:11,400
Speaker 1: on its experience with the world. The visual system builds

578
00:39:11,520 --> 00:39:16,759
Speaker 1: up prior expectations about the relative sizes of objects. So

579
00:39:16,840 --> 00:39:19,200
Speaker 1: if you're standing outside and you see a dog in

580
00:39:19,280 --> 00:39:21,880
Speaker 1: the distance, then it takes up about as much space

581
00:39:21,960 --> 00:39:25,480
Speaker 1: on your retina as the truck over there. You can

582
00:39:25,560 --> 00:39:28,640
Speaker 1: assume that the dog is closer and the truck is

583
00:39:28,880 --> 00:39:32,719
Speaker 1: farther away. Why because a close dog will look a

584
00:39:32,880 --> 00:39:35,680
Speaker 1: certain size and a far away truck will end up

585
00:39:35,719 --> 00:39:38,360
Speaker 1: looking about that same size, and so your brain is

586
00:39:38,400 --> 00:39:42,759
Speaker 1: able to instantly make the proper assumption about how far

587
00:39:42,920 --> 00:39:45,839
Speaker 1: away things are. And you might be wrong, by the way,

588
00:39:45,920 --> 00:39:48,439
Speaker 1: maybe it's a miniature model of a truck that's really

589
00:39:48,560 --> 00:39:51,960
Speaker 1: close and a monstrously huge dog that's really far away.

590
00:39:52,560 --> 00:39:55,880
Speaker 1: But most of the time your assumptions are fine. So

591
00:39:57,160 --> 00:40:00,759
Speaker 1: data doesn't just come in from the world and get seen. Instead,

592
00:40:01,320 --> 00:40:06,960
Speaker 1: your visual system capitalizes on prior expectations. And although this

593
00:40:07,160 --> 00:40:11,080
Speaker 1: idea isn't always intuitive, it's not a new idea. In

594
00:40:11,160 --> 00:40:15,919
Speaker 1: the nineteenth century, the German physician and physicists Hermann von

595
00:40:16,000 --> 00:40:18,759
Speaker 1: Helmholtz was one of the first people to entertain this

596
00:40:19,000 --> 00:40:23,880
Speaker 1: model of perception. He suspected that the small amounts of

597
00:40:23,960 --> 00:40:27,600
Speaker 1: information dribbling in through the eyes were just too slight

598
00:40:27,800 --> 00:40:32,200
Speaker 1: to account for the rich experience of vision. So he

599
00:40:32,320 --> 00:40:36,439
Speaker 1: deduced that the brain makes assumptions about the incoming data

600
00:40:36,640 --> 00:40:40,800
Speaker 1: based on previous experiences, and he correctly surmised that this

601
00:40:40,960 --> 00:40:44,080
Speaker 1: is how the brain can use its best guesses to

602
00:40:44,320 --> 00:40:48,920
Speaker 1: rapidly turn a little trickle of information into a full picture.

603
00:40:49,960 --> 00:40:51,360
Speaker 1: By the way, if you want to look this up

604
00:40:51,440 --> 00:40:56,040
Speaker 1: in more depth, look up Helmholtz's notion of unconscious inference.

605
00:40:56,480 --> 00:40:59,400
Speaker 1: We infer what's out there, and it all happens unconsciously.

606
00:41:00,000 --> 00:41:02,839
Speaker 1: You can also look up Bayes' theorem as a way

607
00:41:02,920 --> 00:41:05,800
Speaker 1: of approaching this mathematically. One way to think about this

608
00:41:06,080 --> 00:41:09,160
Speaker 1: is that our judgments often rely not only on what's

609
00:41:09,200 --> 00:41:11,120
Speaker 1: in front of us, but also on all of our

610
00:41:11,239 --> 00:41:31,600
Speaker 1: prior experiences. So where we are so far is that

611
00:41:31,719 --> 00:41:35,719
Speaker 1: the process of perceiving the world, of interpreting what we

612
00:41:35,760 --> 00:41:40,160
Speaker 1: see or we hear, it's influenced by our past experiences,

613
00:41:40,239 --> 00:41:45,000
Speaker 1: which shape our current expectations, and that's what determines what

614
00:41:45,160 --> 00:41:49,000
Speaker 1: we think we see and hear. Now, it's sometimes the

615
00:41:49,120 --> 00:41:52,960
Speaker 1: case that your brain has more than one prior expectation.

616
00:41:53,120 --> 00:41:55,160
Speaker 1: It could be this, or it could be that, And

617
00:41:55,280 --> 00:41:58,680
Speaker 1: in this case it's easier to witness something very interesting,

618
00:41:58,680 --> 00:42:00,799
Speaker 1: which I want to tell you about now. So let's

619
00:42:00,880 --> 00:42:05,279
Speaker 1: return to the Necker cube, that little wireframe drawing. So

620
00:42:05,360 --> 00:42:09,160
Speaker 1: it's a very simple drawing, but it exposes something amazing,

621
00:42:09,400 --> 00:42:13,879
Speaker 1: which is a competition that is always raging under the hood.

622
00:42:14,880 --> 00:42:17,440
Speaker 1: Your brain is always trying to figure out what is

623
00:42:17,560 --> 00:42:20,320
Speaker 1: going on out there, and the way it does this

624
00:42:20,920 --> 00:42:24,759
Speaker 1: is by assessing probabilities. So this simply means you have

625
00:42:25,000 --> 00:42:28,320
Speaker 1: some networks that are saying, yes, it's definitely this, and

626
00:42:28,400 --> 00:42:31,240
Speaker 1: you have other networks that are saying, yes, it's definitely

627
00:42:31,320 --> 00:42:31,960
Speaker 1: this other thing.

628
00:42:32,800 --> 00:42:34,040
Speaker 2: In the case of the Necker.

629
00:42:33,880 --> 00:42:36,680
Speaker 1: Cube, you have one network saying the cube comes out

630
00:42:36,719 --> 00:42:40,000
Speaker 1: of the page this way, and the other network insisting

631
00:42:40,040 --> 00:42:42,080
Speaker 1: the cube comes out of the page the other way.

632
00:42:43,120 --> 00:42:45,920
Speaker 1: And in other illusions you sometimes have even more networks,

633
00:42:45,960 --> 00:42:49,000
Speaker 1: each voting for their thing. But the key to understand

634
00:42:49,560 --> 00:42:53,560
Speaker 1: is that it's a competition. All these networks are screaming

635
00:42:53,680 --> 00:42:57,160
Speaker 1: off and trying to dominate each other, and it's a

636
00:42:57,719 --> 00:42:59,960
Speaker 1: winner take all competition.

637
00:43:00,360 --> 00:43:01,960
Speaker 2: It's like king of the hill.

638
00:43:02,040 --> 00:43:04,520
Speaker 1: Whichever kid is able to get to the top of

639
00:43:04,600 --> 00:43:07,719
Speaker 1: the hill gets to push everyone else down. In the

640
00:43:07,840 --> 00:43:12,760
Speaker 1: case of local neural networks, when one is successfully firing

641
00:43:12,840 --> 00:43:16,600
Speaker 1: on all cylinders, it's able to inhibit the neighboring networks.

642
00:43:17,080 --> 00:43:21,440
Speaker 1: It releases neurotransmitters that keep itself propped up and at

643
00:43:21,480 --> 00:43:25,920
Speaker 1: the same time inhibiting the activity of the competitors, and

644
00:43:26,040 --> 00:43:27,800
Speaker 1: whichever network.

645
00:43:27,520 --> 00:43:29,800
Speaker 2: Is king is what you perceive.

646
00:43:30,840 --> 00:43:33,960
Speaker 1: And because it's a winner take all competition, there's only

647
00:43:34,080 --> 00:43:35,239
Speaker 1: one king at any time.

648
00:43:35,800 --> 00:43:37,120
Speaker 2: That's why you don't.

649
00:43:36,960 --> 00:43:41,600
Speaker 1: See all the possibilities at once. You only see the winner.

650
00:43:42,640 --> 00:43:45,360
Speaker 1: But here's the wacky thing with the Necker cube. It

651
00:43:45,520 --> 00:43:49,640
Speaker 1: really could be either way. It's equally probable that these

652
00:43:49,760 --> 00:43:52,399
Speaker 1: lions represent a cube this way or represents a cube

653
00:43:52,400 --> 00:43:55,239
Speaker 1: the other way. There's a fifty percent chance of either

654
00:43:55,280 --> 00:43:58,440
Speaker 1: of these. This is known as eque probable. So your

655
00:43:58,520 --> 00:44:03,000
Speaker 1: brain takes this equa probable stimulus and nails it down

656
00:44:03,080 --> 00:44:06,120
Speaker 1: to one choice or the other. But if you have

657
00:44:06,320 --> 00:44:09,000
Speaker 1: stared at one of these drawings for more than ten seconds,

658
00:44:09,120 --> 00:44:14,080
Speaker 1: you know that your brain changes its interpretation. If you

659
00:44:14,200 --> 00:44:16,680
Speaker 1: stare at this wireframe, it looks like it's coming out

660
00:44:16,719 --> 00:44:19,280
Speaker 1: one way from the page, but if you keep staring,

661
00:44:19,320 --> 00:44:21,920
Speaker 1: it'll switch so that it looks like it's coming out

662
00:44:21,960 --> 00:44:24,279
Speaker 1: the other way. And if you stare at this for

663
00:44:24,320 --> 00:44:27,040
Speaker 1: a little while, you'll see that it switches back and forth.

664
00:44:27,239 --> 00:44:29,440
Speaker 1: You see it one way then the other way. Your

665
00:44:29,480 --> 00:44:32,319
Speaker 1: brain will stick with one interpretation for a little while

666
00:44:32,440 --> 00:44:35,239
Speaker 1: and tell you that's what's in the world, and then

667
00:44:35,280 --> 00:44:40,680
Speaker 1: it will suddenly change its claim. Why because, as I said,

668
00:44:40,719 --> 00:44:43,200
Speaker 1: there's a fifty percent chance of interpreting the cube one

669
00:44:43,239 --> 00:44:46,040
Speaker 1: way or the other, and the brain cannot see both

670
00:44:46,080 --> 00:44:50,080
Speaker 1: interpretations at the same time, so it switches between them.

671
00:44:50,200 --> 00:44:53,600
Speaker 2: It's the king of the hill game, but the king

672
00:44:53,719 --> 00:44:54,600
Speaker 2: never lasts.

673
00:44:55,080 --> 00:44:57,919
Speaker 1: Someone always manages to knock that kid off the top,

674
00:44:58,040 --> 00:45:00,480
Speaker 1: and then the new kid has to defe and the

675
00:45:00,600 --> 00:45:04,719
Speaker 1: throne against other invaders. And that's precisely what happens with

676
00:45:04,880 --> 00:45:09,320
Speaker 1: these neural network competitions. One network wins, but it doesn't

677
00:45:09,440 --> 00:45:12,880
Speaker 1: last that long before it gets unseated. And then the

678
00:45:12,960 --> 00:45:16,680
Speaker 1: other network is active in a loop of self reinforcing

679
00:45:16,760 --> 00:45:20,560
Speaker 1: neurons firing. It gets to keep control be king of

680
00:45:20,640 --> 00:45:22,720
Speaker 1: the North for a little bit, but then the first

681
00:45:22,800 --> 00:45:26,239
Speaker 1: one unseats it again. So what you see with the

682
00:45:26,360 --> 00:45:31,520
Speaker 1: simple drawing is the ever present, active battle in your

683
00:45:31,600 --> 00:45:35,640
Speaker 1: skull to control perception. So, in other words, if you

684
00:45:35,800 --> 00:45:40,040
Speaker 1: have two possible top down models, either.

685
00:45:39,800 --> 00:45:44,080
Speaker 5: Of which could equally be right, they'll fight and you'll

686
00:45:44,200 --> 00:45:48,800
Speaker 5: believe whoever the temporary winner is, and then you'll believe

687
00:45:48,840 --> 00:45:51,000
Speaker 5: the next guy when he's back in power, and then

688
00:45:51,040 --> 00:45:51,760
Speaker 5: the first network.

689
00:45:51,800 --> 00:45:52,080
Speaker 2: Again.

690
00:45:53,520 --> 00:45:57,080
Speaker 1: Now the dress tends not to switch, and this is

691
00:45:57,160 --> 00:46:02,400
Speaker 1: because it's not equiprobable. Our brain has developed very clear

692
00:46:02,960 --> 00:46:07,120
Speaker 1: prior expectations about lighting and fabric and windows and so on.

693
00:46:07,800 --> 00:46:10,960
Speaker 1: So my brain makes an interpretation and your brain makes

694
00:46:11,000 --> 00:46:14,360
Speaker 1: an interpretation, and there's no reason for either one of

695
00:46:14,480 --> 00:46:17,560
Speaker 1: them to question it. It's like playing King of the

696
00:46:17,640 --> 00:46:20,359
Speaker 1: hill against some small puppies. No one's going to knock

697
00:46:20,400 --> 00:46:24,320
Speaker 1: you off the throne. And that's why it's so difficult

698
00:46:24,600 --> 00:46:27,600
Speaker 1: to change your interpretation of the dress, even when you're

699
00:46:27,719 --> 00:46:30,840
Speaker 1: told that some other interpretation is possible.

700
00:46:31,400 --> 00:46:32,920
Speaker 2: Your brain relies on.

701
00:46:33,200 --> 00:46:37,120
Speaker 1: Deep assumptions about the world, and it's generally just too

702
00:46:37,239 --> 00:46:41,040
Speaker 1: hard to unseat the monarch. But what the Necker cube

703
00:46:41,160 --> 00:46:45,160
Speaker 1: reveals is that our brain's interpretation of the world can

704
00:46:45,239 --> 00:46:49,160
Speaker 1: be quite active if there are other equally likely interpretations

705
00:46:49,239 --> 00:46:51,960
Speaker 1: to be had, So the way we see the world

706
00:46:52,440 --> 00:46:56,160
Speaker 1: can change from moment to moment. Now as just a

707
00:46:56,239 --> 00:46:59,640
Speaker 1: one minute tangent. The funny thing is that you think

708
00:46:59,800 --> 00:47:03,560
Speaker 1: you are making the cube switch interpretations by yourself. In

709
00:47:03,640 --> 00:47:06,279
Speaker 1: other words, you feel like you're doing it consciously when

710
00:47:06,320 --> 00:47:07,840
Speaker 1: the cube switches back and forth.

711
00:47:08,560 --> 00:47:11,520
Speaker 2: But let's say we measure this. You stare at the

712
00:47:11,560 --> 00:47:15,160
Speaker 2: little cube and you hold down one key when you.

713
00:47:15,239 --> 00:47:17,799
Speaker 1: See it in this configuration, and as soon as your

714
00:47:17,840 --> 00:47:20,520
Speaker 1: perception switches and now looks the other way, you hold

715
00:47:20,600 --> 00:47:23,120
Speaker 1: down the other key. And you do this for a while,

716
00:47:23,520 --> 00:47:26,120
Speaker 1: back and forth and back and forth. You hold down

717
00:47:26,160 --> 00:47:29,399
Speaker 1: a key to let me know which perception you are seeing.

718
00:47:30,200 --> 00:47:32,720
Speaker 1: And remember how amazing this is because nothing is changing

719
00:47:32,800 --> 00:47:34,239
Speaker 1: on the page. It's only in your head.

720
00:47:35,040 --> 00:47:35,440
Speaker 2: Anyway.

721
00:47:35,680 --> 00:47:38,319
Speaker 1: When we look at the data, it's clear that your

722
00:47:38,400 --> 00:47:42,240
Speaker 1: results follow a particular mathematical distribution called a gamma distribution,

723
00:47:42,840 --> 00:47:46,719
Speaker 1: which comes from a random process. For the efficionados, this

724
00:47:46,840 --> 00:47:50,719
Speaker 1: is consistent with a poissone process. All this means is

725
00:47:50,760 --> 00:47:54,320
Speaker 1: that this switching is random, and this is exactly the

726
00:47:54,400 --> 00:47:58,680
Speaker 1: distribution you would expect from randomness. Sometimes you have the

727
00:47:58,800 --> 00:48:01,640
Speaker 1: winning network holding on to the throne for a long time,

728
00:48:01,719 --> 00:48:04,000
Speaker 1: so as for a short time, and on average it

729
00:48:04,120 --> 00:48:07,600
Speaker 1: lasts this medium amount of time before it switches. The

730
00:48:07,800 --> 00:48:12,840
Speaker 1: point is you think you're switching consciously, but it's just random.

731
00:48:13,480 --> 00:48:16,920
Speaker 1: The reason you take credit is because you think, Okay,

732
00:48:16,920 --> 00:48:19,160
Speaker 1: I'm seeing it this way, and I really want to

733
00:48:19,160 --> 00:48:21,280
Speaker 1: make it switch the other way, and I'm going.

734
00:48:21,239 --> 00:48:23,279
Speaker 2: To consciously work to switch it.

735
00:48:23,719 --> 00:48:28,480
Speaker 1: Okay, almost there, not quite working, still trying, and then

736
00:48:28,520 --> 00:48:32,720
Speaker 1: it randomly switches and you take credit for it. Here's

737
00:48:32,719 --> 00:48:35,000
Speaker 1: an analogy to help us understand that. You know those

738
00:48:35,320 --> 00:48:38,200
Speaker 1: pedestrian crossing buttons that you push when you want to

739
00:48:38,320 --> 00:48:41,480
Speaker 1: cross the street, and the little walk signal eventually shows

740
00:48:41,560 --> 00:48:43,120
Speaker 1: up and lets you know that you're safe to walk.

741
00:48:44,000 --> 00:48:47,800
Speaker 2: Some fraction of those buttons are placebos. They're fake. You

742
00:48:48,000 --> 00:48:50,400
Speaker 2: hit them, but they don't do anything.

743
00:48:50,680 --> 00:48:52,920
Speaker 1: You wait for exactly the same amount of time that

744
00:48:53,080 --> 00:48:56,360
Speaker 1: you would have waited anyway, but you have a sense

745
00:48:56,400 --> 00:49:00,200
Speaker 1: of control, an illusion of power over the light, even

746
00:49:00,239 --> 00:49:02,560
Speaker 1: though the timing doesn't change one bit.

747
00:49:03,200 --> 00:49:06,400
Speaker 2: And this is exactly the situation with this switching of

748
00:49:06,640 --> 00:49:07,440
Speaker 2: the Necker cube.

749
00:49:07,960 --> 00:49:11,759
Speaker 1: You consciously try to change it, and when it eventually

750
00:49:11,960 --> 00:49:14,880
Speaker 1: changes on its own, you think, yeah, that was because

751
00:49:14,920 --> 00:49:18,400
Speaker 1: of me. But when we measure the switching times, it

752
00:49:18,520 --> 00:49:21,919
Speaker 1: doesn't change anything at all, whether you're trying or not trying,

753
00:49:22,000 --> 00:49:26,560
Speaker 1: whether you're banging on that button or ignoring it. Okay,

754
00:49:26,640 --> 00:49:28,560
Speaker 1: so now I want to zoom back up to the

755
00:49:28,640 --> 00:49:30,960
Speaker 1: big picture about what we've been talking about, which is

756
00:49:31,040 --> 00:49:33,800
Speaker 1: how your brain makes assumptions about things, and how in

757
00:49:33,920 --> 00:49:38,080
Speaker 1: some circumstances these assumptions can fight it out. So we

758
00:49:38,200 --> 00:49:41,880
Speaker 1: see this in language often take the example of puns.

759
00:49:42,440 --> 00:49:45,720
Speaker 1: Puns strike us as funny because we're able to switch

760
00:49:45,840 --> 00:49:48,839
Speaker 1: back and forth and see the same thing in two

761
00:49:49,000 --> 00:49:52,040
Speaker 1: different ways. What do you get when you drop a

762
00:49:52,239 --> 00:49:57,560
Speaker 1: piano down a mine shaft a flat minor. The point

763
00:49:57,560 --> 00:50:00,359
Speaker 1: about puns is that we know from the s mile

764
00:50:00,440 --> 00:50:02,719
Speaker 1: on the other person's face that there's some joke to

765
00:50:02,800 --> 00:50:06,239
Speaker 1: be had, and so we search for other interpretations, and

766
00:50:06,360 --> 00:50:09,200
Speaker 1: we can switch back and forth between them, just like

767
00:50:09,280 --> 00:50:12,719
Speaker 1: a Necker cube. But something I found interesting is that

768
00:50:13,160 --> 00:50:16,160
Speaker 1: brains can be lazy, and we don't always bother or

769
00:50:16,280 --> 00:50:19,759
Speaker 1: seeking other interpretations. If you don't have a strong enough

770
00:50:19,880 --> 00:50:24,399
Speaker 1: reason to have more than one interpretation, then you stick

771
00:50:24,480 --> 00:50:28,120
Speaker 1: with what you've got. And this is often true in language,

772
00:50:28,280 --> 00:50:32,600
Speaker 1: which is very low bandwidth and depends enormously on assumptions.

773
00:50:33,120 --> 00:50:35,279
Speaker 1: So the other night I was at a party and

774
00:50:35,440 --> 00:50:38,600
Speaker 1: somehow the conversation moved in a direction where I mentioned

775
00:50:39,080 --> 00:50:44,040
Speaker 1: the famous book by Rachel Carson called Silent Spring. It

776
00:50:44,200 --> 00:50:45,800
Speaker 1: just so happened that no one there had heard of

777
00:50:45,840 --> 00:50:48,880
Speaker 1: this book. So in a sentence. I explained that the

778
00:50:49,120 --> 00:50:53,320
Speaker 1: author had argued that if pesticide use continued, there wouldn't

779
00:50:53,320 --> 00:50:56,759
Speaker 1: be any more birds, and so the spring season would

780
00:50:56,800 --> 00:50:59,560
Speaker 1: come around and we would hear no more chirping. It

781
00:50:59,560 --> 00:51:03,879
Speaker 1: would be And I was sort of surprised when everyone said, oh,

782
00:51:04,880 --> 00:51:07,440
Speaker 1: like I had just cleared up some confusion for them,

783
00:51:08,080 --> 00:51:11,760
Speaker 1: because it turns out that when I had said silent spring,

784
00:51:12,440 --> 00:51:14,760
Speaker 1: the person to my left thought I was talking about

785
00:51:14,840 --> 00:51:17,720
Speaker 1: a spring like a creek, so she interpreted the title

786
00:51:17,840 --> 00:51:21,479
Speaker 1: as silent river, and the person to my right thought

787
00:51:21,520 --> 00:51:26,120
Speaker 1: of spring like boeing boeing spring. And the person across

788
00:51:26,160 --> 00:51:28,520
Speaker 1: from me thought I was talking about the word spring

789
00:51:29,120 --> 00:51:32,880
Speaker 1: like the verb to jump, so he pictured silent spring

790
00:51:33,440 --> 00:51:37,239
Speaker 1: as a lion springing on him silently. And this is

791
00:51:37,320 --> 00:51:39,800
Speaker 1: typical of the way that we take in little bits

792
00:51:39,840 --> 00:51:43,120
Speaker 1: of data and impose an interpretation on them, and then

793
00:51:43,160 --> 00:51:47,560
Speaker 1: we're done. Our brains aren't generally incentivized to keep looking

794
00:51:47,640 --> 00:51:51,400
Speaker 1: for interpretations. You pick one and that's it. And by

795
00:51:51,440 --> 00:51:54,600
Speaker 1: the way, that's typically what happens with Laurel and Yanny.

796
00:51:54,960 --> 00:51:57,919
Speaker 1: If you didn't know to listen hard for something else,

797
00:51:58,120 --> 00:52:02,920
Speaker 1: you probably wouldn't. And with green needle and brainstorm. Unless

798
00:52:02,960 --> 00:52:07,080
Speaker 1: you were told to switch your perception, you probably wouldn't

799
00:52:07,120 --> 00:52:10,440
Speaker 1: have even thought to try it. And so I often

800
00:52:10,640 --> 00:52:13,520
Speaker 1: wonder about the ways that we do this with many

801
00:52:13,680 --> 00:52:17,560
Speaker 1: things around us. We pick some top down model and

802
00:52:17,719 --> 00:52:20,080
Speaker 1: that seems to match the bottom up data, and it

803
00:52:20,200 --> 00:52:24,000
Speaker 1: doesn't strike us to examine further because we're pretty sure

804
00:52:24,080 --> 00:52:26,759
Speaker 1: we have a match. I'll leave this as an open

805
00:52:26,840 --> 00:52:29,320
Speaker 1: question for all of us to think about places in

806
00:52:29,400 --> 00:52:32,520
Speaker 1: our life that maybe we haven't even thought to re

807
00:52:32,840 --> 00:52:37,759
Speaker 1: examine more deeply. So to wrap this up, when this

808
00:52:37,960 --> 00:52:41,120
Speaker 1: woman in the UK sent a little cell phone photo

809
00:52:41,239 --> 00:52:44,600
Speaker 1: to her daughter about her dress, it not only broke

810
00:52:44,719 --> 00:52:48,800
Speaker 1: the Internet, the more importantly, it breaks for us a

811
00:52:48,960 --> 00:52:53,360
Speaker 1: critical assumption that almost everyone carries around, the assumption that

812
00:52:53,480 --> 00:52:55,719
Speaker 1: when I look at the world and you look at

813
00:52:55,760 --> 00:53:00,319
Speaker 1: the world, we see the same thing. The naive umption

814
00:53:00,600 --> 00:53:04,200
Speaker 1: is that there is simply truth out there and it's

815
00:53:04,320 --> 00:53:07,120
Speaker 1: just a matter of opening your eyes. But the dress

816
00:53:07,440 --> 00:53:11,479
Speaker 1: and hundreds of other illusions reveal that we don't see

817
00:53:11,560 --> 00:53:18,120
Speaker 1: the world out there directly. Everything is interpretation. We only

818
00:53:18,239 --> 00:53:22,520
Speaker 1: have a bit of data dribbling in through our peripheral devices,

819
00:53:22,600 --> 00:53:26,240
Speaker 1: our sensory organs, and that data enters into a brain

820
00:53:26,760 --> 00:53:30,440
Speaker 1: that's already churning and bubbling with its own activity, its

821
00:53:30,480 --> 00:53:35,400
Speaker 1: own expectations, and so all we ever perceive is the

822
00:53:35,680 --> 00:53:39,319
Speaker 1: best guess from our neural networks about what is going

823
00:53:39,480 --> 00:53:43,640
Speaker 1: on out there, given a little rough data and a

824
00:53:43,800 --> 00:53:48,759
Speaker 1: lot of assumptions shaped by our past experiences. So the

825
00:53:48,920 --> 00:53:52,440
Speaker 1: next time you see a face in an electrical outlet,

826
00:53:52,680 --> 00:53:55,719
Speaker 1: where you see a shape in aurorshack blot, or you

827
00:53:55,880 --> 00:53:59,480
Speaker 1: see the dress and feel certain about its color, just

828
00:53:59,600 --> 00:54:03,279
Speaker 1: remember you are not seeing the world as it is.

829
00:54:03,960 --> 00:54:06,319
Speaker 2: You are seeing it as you are.

830
00:54:10,560 --> 00:54:13,520
Speaker 1: Go to Eagleman dot com slash podcast for more information

831
00:54:13,840 --> 00:54:17,680
Speaker 1: and to find further reading. Send me an email at

832
00:54:17,880 --> 00:54:21,880
Speaker 1: podcast at eagleman dot com with questions or discussions, and

833
00:54:22,000 --> 00:54:24,120
Speaker 1: I'm going to be making episodes in which I address

834
00:54:24,200 --> 00:54:26,279
Speaker 1: those reaching.

835
00:54:25,960 --> 00:54:29,440
Speaker 2: Out on a narrow road from my internal world to yours.

836
00:54:29,800 --> 00:54:32,000
Speaker 2: This is David Eagleman, and thank you for joining me

837
00:54:32,239 --> 00:54:33,640
Speaker 2: in the inner cosmos.