1
00:00:00,000 --> 00:00:04,710
The following content is
provided by MIT OpenCourseWare

2
00:00:04,710 --> 00:00:07,199
under a Creative Commons
license.

3
00:00:07,199 --> 00:00:10,185
Additional information about
our license, and MIT

4
00:00:10,185 --> 00:00:15,162
OpenCourseWare in general, is
available at ocw.mit.edu.

5
00:00:15,162 --> 00:00:15,660
PROFESSOR:

6
00:00:15,660 --> 00:00:18,130
Good afternoon.

7
00:00:18,130 --> 00:00:23,590
Before I forget, today's
topic is cognition.

8
00:00:23,590 --> 00:00:28,210
And in part, a good chunk of
this lecture is going to be

9
00:00:28,210 --> 00:00:34,650
based on work done by Kahneman
and Tversky, who won the Nobel

10
00:00:34,650 --> 00:00:38,110
Prize for a bunch of this
work some years back.

11
00:00:38,110 --> 00:00:41,070
A couple of years back, now.

12
00:00:41,070 --> 00:00:44,450
The reason that is of particular
interest is that

13
00:00:44,450 --> 00:00:48,680
Danny Kahneman is here
tomorrow speaking.

14
00:00:48,680 --> 00:00:51,810
E-51, 4 o'clock, right?

15
00:00:51,810 --> 00:00:55,890
So if you happen to be a,
interested in the material and

16
00:00:55,890 --> 00:00:59,490
b, at leisure at 4 o'clock
tomorrow, I'd go hear him.

17
00:00:59,490 --> 00:01:04,880
He's a very good speaker and
there aren't that many folks

18
00:01:04,880 --> 00:01:06,830
who I get to talk about
in psychology

19
00:01:06,830 --> 00:01:08,380
who win Nobel Prizes.

20
00:01:08,380 --> 00:01:10,730
So, you should go hear him.

21
00:01:13,730 --> 00:01:16,200
That's one characteristic
of today's lecture.

22
00:01:16,200 --> 00:01:18,860
I realize that the other
characteristic of today's

23
00:01:18,860 --> 00:01:25,650
lecture is that I get to tell
you a bunch of things that are

24
00:01:25,650 --> 00:01:31,000
applicable to the forthcoming
election.

25
00:01:31,000 --> 00:01:34,120
And to politics in general.

26
00:01:34,120 --> 00:01:39,210
This particular swath I'm
cutting through, the large

27
00:01:39,210 --> 00:01:45,330
area of cognition, includes a
bunch of topics that if you've

28
00:01:45,330 --> 00:01:49,020
watched the debates thus far,
or if you go and watch the

29
00:01:49,020 --> 00:01:53,460
debate that's Friday, right, the
next one's Friday evening,

30
00:01:53,460 --> 00:01:56,510
you'll be able to see some of
the stuff I'm talking about

31
00:01:56,510 --> 00:02:00,960
today in action.

32
00:02:00,960 --> 00:02:04,510
There are two large parts to
what I want to talk about.

33
00:02:04,510 --> 00:02:09,480
And this is described, at least
in rough detail, in this

34
00:02:09,480 --> 00:02:12,720
abstract at the top
of the handout.

35
00:02:12,720 --> 00:02:18,350
We can make the distinction
between the way, using the

36
00:02:18,350 --> 00:02:22,410
computer as a model for how the
brain works is sort of a

37
00:02:22,410 --> 00:02:25,380
common game, rather like
using the camera as a

38
00:02:25,380 --> 00:02:26,730
model for the eye.

39
00:02:26,730 --> 00:02:29,020
There are important
differences.

40
00:02:29,020 --> 00:02:32,020
One of the important differences
that I'll talk

41
00:02:32,020 --> 00:02:35,330
about, that I started talking
about last time, is that if

42
00:02:35,330 --> 00:02:41,820
you call up a document off your
laptop, you are expecting

43
00:02:41,820 --> 00:02:45,140
that thing pulled out of memory
to be exactly what you

44
00:02:45,140 --> 00:02:47,220
put into memory.

45
00:02:47,220 --> 00:02:52,430
As we saw, as we saw towards the
end of the last lecture,

46
00:02:52,430 --> 00:02:55,020
that's not true about
human memory.

47
00:02:55,020 --> 00:02:57,430
I'll elaborate on that a bit.

48
00:02:57,430 --> 00:03:02,110
The other thing you expect is
that your computer will do

49
00:03:02,110 --> 00:03:04,290
mathematical and logical
operations

50
00:03:04,290 --> 00:03:07,180
correctly and quickly.

51
00:03:07,180 --> 00:03:10,600
And that also turns out
to be not particularly

52
00:03:10,600 --> 00:03:15,130
characteristic of human
cognitive behavior.

53
00:03:15,130 --> 00:03:17,530
For interesting reasons that I
will describe in the second

54
00:03:17,530 --> 00:03:18,560
part of the lecture.

55
00:03:18,560 --> 00:03:22,910
So let's start with this issue,
this point, that says

56
00:03:22,910 --> 00:03:28,350
that recall is not just taking
stuff out of storage.

57
00:03:28,350 --> 00:03:33,040
Last time, I talked about a
couple of examples of that.

58
00:03:33,040 --> 00:03:36,520
And those are the ones labeled
as old on the handout.

59
00:03:36,520 --> 00:03:38,790
There's the context
of remembering.

60
00:03:38,790 --> 00:03:44,940
If you are depressed, your
memories of childhood are more

61
00:03:44,940 --> 00:03:48,390
gloomy than if you are
in a good mood.

62
00:03:48,390 --> 00:03:52,270
There is retriever's bias, that
was the example I was

63
00:03:52,270 --> 00:03:59,930
giving last time, with these
experiments where the knife in

64
00:03:59,930 --> 00:04:03,410
a simulated altercation that you
see, the knife might move

65
00:04:03,410 --> 00:04:07,070
from one person's hand to
another person's hand in your

66
00:04:07,070 --> 00:04:10,320
memory, depending on the
particular biases and

67
00:04:10,320 --> 00:04:12,680
associations in your memory.

68
00:04:12,680 --> 00:04:18,520
Now, let me tell you about a few
more ways in which memory

69
00:04:18,520 --> 00:04:22,560
is not just like pulling
a document off

70
00:04:22,560 --> 00:04:24,850
the hard drive somewhere.

71
00:04:24,850 --> 00:04:28,700
One of these is what I've
labeled on the handout as

72
00:04:28,700 --> 00:04:31,830
failures of source monitoring.

73
00:04:31,830 --> 00:04:35,950
Source monitoring is what we're
busy making sure you do

74
00:04:35,950 --> 00:04:37,750
correctly on your papers.

75
00:04:37,750 --> 00:04:42,090
Telling us where the information
comes from.

76
00:04:42,090 --> 00:04:45,390
And evaluating the information
on the basis of

77
00:04:45,390 --> 00:04:46,740
where it comes from.

78
00:04:46,740 --> 00:04:53,840
It turns out that that's not
as easy to do as one might

79
00:04:53,840 --> 00:04:55,030
think it ought to be.

80
00:04:55,030 --> 00:04:58,160
Indeed, I got this quote from
Euripedes on the handout that

81
00:04:58,160 --> 00:05:02,180
says man's most valuable trait
is a judicious sense of what

82
00:05:02,180 --> 00:05:03,680
not to believe.

83
00:05:03,680 --> 00:05:09,480
If you hear something that's
clearly false, is it possible

84
00:05:09,480 --> 00:05:12,640
for you to not be influenced
by that.

85
00:05:12,640 --> 00:05:14,900
This turns out to have been a
question that philosophers

86
00:05:14,900 --> 00:05:15,970
have worried about.

87
00:05:15,970 --> 00:05:18,160
Descartes thought that
the answer to the

88
00:05:18,160 --> 00:05:19,440
question was yes.

89
00:05:19,440 --> 00:05:24,490
If you hear a statement that's
clearly false, you can label

90
00:05:24,490 --> 00:05:28,470
it as clearly false and, in
a sense, put it aside.

91
00:05:28,470 --> 00:05:31,530
Spinoza, in contrast, thought
that it would be very nice if

92
00:05:31,530 --> 00:05:32,140
that were true.

93
00:05:32,140 --> 00:05:34,410
But he doubted that
it really was.

94
00:05:34,410 --> 00:05:38,690
This has subsequently been
tested experimentally.

95
00:05:38,690 --> 00:05:41,280
And I don't have enough colored
chalk around, but I'll

96
00:05:41,280 --> 00:05:42,450
simulate it.

97
00:05:42,450 --> 00:05:45,170
Here's what Dan Gilbert
and his colleagues

98
00:05:45,170 --> 00:05:47,850
did a few years back.

99
00:05:47,850 --> 00:05:50,810
They did an experiment
where you read little

100
00:05:50,810 --> 00:05:54,260
biographies of people.

101
00:05:54,260 --> 00:06:00,600
And the material you're reading
is color-coded.

102
00:06:00,600 --> 00:06:03,100
And so you're reading along.

103
00:06:03,100 --> 00:06:05,920
And then some of it, here, we'll
use straight lines for a

104
00:06:05,920 --> 00:06:06,840
different color.

105
00:06:06,840 --> 00:06:09,410
Some of it is red.

106
00:06:09,410 --> 00:06:11,520
If it's red it's a lie.

107
00:06:11,520 --> 00:06:12,920
Don't believe it.

108
00:06:12,920 --> 00:06:16,330
But if it's green, it's true.

109
00:06:16,330 --> 00:06:20,670
And if it's red it's a lie,
and true, and so on.

110
00:06:20,670 --> 00:06:22,260
Read a paragraph
of this stuff.

111
00:06:22,260 --> 00:06:24,170
We're going to make sure that
you've read it because we're

112
00:06:24,170 --> 00:06:26,770
going to test you afterwards
and make sure that you

113
00:06:26,770 --> 00:06:28,430
actually know what it said.

114
00:06:28,430 --> 00:06:32,670
Because it's very boring to
discover that you can ignore

115
00:06:32,670 --> 00:06:35,700
lies if you don't ever
hear the lies.

116
00:06:35,700 --> 00:06:39,620
But we know you've heard
these various lies.

117
00:06:39,620 --> 00:06:43,530
And then what we'll do is ask
you about this person.

118
00:06:43,530 --> 00:06:48,090
The metric they were using was
a scale between college

119
00:06:48,090 --> 00:06:50,980
student and criminal,
I believe.

120
00:06:50,980 --> 00:06:54,150
And you were asked how criminal
you thought these

121
00:06:54,150 --> 00:06:55,660
particular people.

122
00:06:55,660 --> 00:06:56,570
So you're reading along.

123
00:06:56,570 --> 00:07:01,520
And so-and-so is taking 1803.

124
00:07:01,520 --> 00:07:05,410
And steals calculators.

125
00:07:05,410 --> 00:07:10,150
And does the problem sets
reasonably reliably.

126
00:07:10,150 --> 00:07:12,090
And then fences the
calculators.

127
00:07:12,090 --> 00:07:15,790
And so on, right?

128
00:07:15,790 --> 00:07:21,220
The result is that even though
you know that the statements

129
00:07:21,220 --> 00:07:28,100
are false, your opinion of this
hypothetical person is

130
00:07:28,100 --> 00:07:30,870
lowered by negative lies.

131
00:07:30,870 --> 00:07:33,770
And, reciprocally, raised
by positive lies.

132
00:07:33,770 --> 00:07:42,780
You can't completely discount
the information that you're

133
00:07:42,780 --> 00:07:45,630
hearing, even if you
know it's a lie.

134
00:07:45,630 --> 00:07:48,960
Now, if it is not immediately
obvious to you that this has

135
00:07:48,960 --> 00:07:52,860
direct relevance to the current
political campaign,

136
00:07:52,860 --> 00:07:55,520
you can ask yourself about the
behavior you can see in this

137
00:07:55,520 --> 00:07:58,890
campaign, or in any other
campaign that you ever care to

138
00:07:58,890 --> 00:08:00,430
subject yourself to.

139
00:08:00,430 --> 00:08:07,250
Which is, you get charges that
are made, typically by

140
00:08:07,250 --> 00:08:10,360
surrogates of the candidate.

141
00:08:10,360 --> 00:08:14,500
So a clear example, this here
would be the charges about

142
00:08:14,500 --> 00:08:19,850
Kerry's Vietnam behavior,
brought by the Swift Boat

143
00:08:19,850 --> 00:08:23,920
people, not directly by
the Bush campaign.

144
00:08:23,920 --> 00:08:25,640
And I should be able to
think of a flip side.

145
00:08:25,640 --> 00:08:29,270
Oh, the National Guard
stuff for Bush.

146
00:08:29,270 --> 00:08:31,450
Regardless of the truth, we're
not going to get into the

147
00:08:31,450 --> 00:08:32,500
truth assertion of this.

148
00:08:32,500 --> 00:08:36,550
But, in both cases the really
nasty attacks are raised by

149
00:08:36,550 --> 00:08:37,990
surrogates.

150
00:08:37,990 --> 00:08:45,720
If it is then proved to be a
lie, then the candidate can

151
00:08:45,720 --> 00:08:52,000
have the great and noble
opportunity to run around the

152
00:08:52,000 --> 00:08:58,780
country saying, look, people
should not be saying that John

153
00:08:58,780 --> 00:09:01,060
Kerry molests small animals.

154
00:09:01,060 --> 00:09:03,280
It's a lie that he molests
small animals.

155
00:09:03,280 --> 00:09:07,200
Anybody who says that should
given a dope slap.

156
00:09:07,200 --> 00:09:09,850
And I would never say anything
like, John Kerry

157
00:09:09,850 --> 00:09:11,210
molests small animals.

158
00:09:11,210 --> 00:09:15,470
And I will tell all of my people
not to say that John

159
00:09:15,470 --> 00:09:17,080
Kerry molests small animals.

160
00:09:17,080 --> 00:09:21,550
And you can go home with this
righteous feeling around you

161
00:09:21,550 --> 00:09:23,100
saying that I've done
the right thing of

162
00:09:23,100 --> 00:09:24,560
condemning this smear.

163
00:09:24,560 --> 00:09:27,930
Not only that, I've spread it
around a bit, and the dumb

164
00:09:27,930 --> 00:09:29,970
suckers are going
to believe it.

165
00:09:29,970 --> 00:09:32,550
It's not going to swing the
election one way the other,

166
00:09:32,550 --> 00:09:34,080
but it's going to push things.

167
00:09:34,080 --> 00:09:39,520
And political campaigns do this
absolutely routinely.

168
00:09:39,520 --> 00:09:42,420
Perhaps not out of a completely
Machiavellian

169
00:09:42,420 --> 00:09:45,100
understanding of the psychology,
but because they

170
00:09:45,100 --> 00:09:47,180
know it works in some fashion.

171
00:09:47,180 --> 00:09:48,820
I don't know that they're
sitting around reading Dan

172
00:09:48,820 --> 00:09:54,590
Gilbert's papers to figure
this sort of thing out.

173
00:09:54,590 --> 00:09:57,960
This has applications elsewhere
in your life, too.

174
00:09:57,960 --> 00:10:00,970
It's very important
to understand --

175
00:10:00,970 --> 00:10:05,230
I'm going to take zinc because
it's going to keep me from

176
00:10:05,230 --> 00:10:06,640
getting colds.

177
00:10:06,640 --> 00:10:08,190
Now, why do I think that?

178
00:10:08,190 --> 00:10:14,150
Is that because my aunt June,
who's the medical wacko of the

179
00:10:14,150 --> 00:10:15,870
family thinks that's true?

180
00:10:15,870 --> 00:10:18,290
Or is that because I read it in
the Journal of the American

181
00:10:18,290 --> 00:10:19,510
Medical Association?

182
00:10:19,510 --> 00:10:22,450
It makes a difference where the
information comes from.

183
00:10:22,450 --> 00:10:24,600
So, some sort of an ability
to evaluate

184
00:10:24,600 --> 00:10:25,810
the source is important.

185
00:10:25,810 --> 00:10:28,260
And the fact that we're
not very good at it is

186
00:10:28,260 --> 00:10:29,530
problematic.

187
00:10:29,530 --> 00:10:32,850
A particular subset of the
problems with source

188
00:10:32,850 --> 00:10:36,400
monitoring is what's known
as [? crypt ?] amnesia.

189
00:10:36,400 --> 00:10:40,140
And that's the problem where you
don't remember where your

190
00:10:40,140 --> 00:10:42,750
own ideas come from.

191
00:10:42,750 --> 00:10:50,090
The canonical form of this,
in a setting like MIT --

192
00:10:50,090 --> 00:10:54,520
actually, this is the part of
the lecture for the TAs.

193
00:10:54,520 --> 00:10:57,990
The canonical version of this
is you go to your doctoral

194
00:10:57,990 --> 00:11:01,320
adviser with a bright idea.

195
00:11:01,320 --> 00:11:03,740
Yeah, I got this cool idea
for new experiment.

196
00:11:03,740 --> 00:11:07,210
Your doctoral adviser looks at
you, says, that's stupid.

197
00:11:07,210 --> 00:11:08,020
And you go away.

198
00:11:08,020 --> 00:11:10,490
You know, you're kind
of devastated.

199
00:11:10,490 --> 00:11:12,710
Next week, in lab meeting.

200
00:11:12,710 --> 00:11:16,910
Your doctoral adviser has this
really bright new idea.

201
00:11:16,910 --> 00:11:18,440
It's your idea.

202
00:11:18,440 --> 00:11:20,630
He thinks it's his idea.

203
00:11:20,630 --> 00:11:22,940
He does not, he's being a --

204
00:11:22,940 --> 00:11:25,550
probably not, being a slimeball
to say I'm going to

205
00:11:25,550 --> 00:11:28,470
steal their ideas.

206
00:11:28,470 --> 00:11:31,750
It's, you actually forget
where it comes from.

207
00:11:31,750 --> 00:11:34,960
And in the absence of a clear
memory of where the idea came

208
00:11:34,960 --> 00:11:39,070
from, if I've got an idea in
my head about my research,

209
00:11:39,070 --> 00:11:40,470
gee, I wonder where
that came from.

210
00:11:40,470 --> 00:11:41,320
It came from --

211
00:11:41,320 --> 00:11:42,990
I think they come from
my shower head.

212
00:11:42,990 --> 00:11:44,490
Because all the good
ideas I have

213
00:11:44,490 --> 00:11:45,450
always come in the shower.

214
00:11:45,450 --> 00:11:48,710
But, most people tend to think
it comes from them.

215
00:11:48,710 --> 00:11:51,620
I just cite the shower.

216
00:11:51,620 --> 00:11:55,960
But this is actually a very good
argument, by the way, for

217
00:11:55,960 --> 00:11:59,450
taking copious notes in a lab
book when you work in a.

218
00:11:59,450 --> 00:12:03,730
Lab Because if it ever gets to
the point of an argument, you

219
00:12:03,730 --> 00:12:06,740
want to be able to say, look,
in this dated entry from

220
00:12:06,740 --> 00:12:09,170
October 5th, 10,000 and --

221
00:12:09,170 --> 00:12:15,830
2004, it might take you a long
time to get your doctorate.

222
00:12:15,830 --> 00:12:17,070
here's the idea.

223
00:12:17,070 --> 00:12:19,250
I wrote down this idea, and
then I wrote down this

224
00:12:19,250 --> 00:12:22,950
tear-stained comment where you
told me I was a doofus.

225
00:12:22,950 --> 00:12:25,940
But, anyway, I really do
think this was my idea.

226
00:12:25,940 --> 00:12:28,210
And I ought to be an author
on the paper that you just

227
00:12:28,210 --> 00:12:30,290
submitted about it.

228
00:12:30,290 --> 00:12:31,750
Anyway, it's a serious
problem.

229
00:12:31,750 --> 00:12:36,760
Another place where that shows
up is there are recurring

230
00:12:36,760 --> 00:12:40,750
scandals in the sort of academic
book-publishing world

231
00:12:40,750 --> 00:12:44,810
where some well-respected author
is proven to have taken

232
00:12:44,810 --> 00:12:47,650
a chunk out of somebody
else's book.

233
00:12:47,650 --> 00:12:49,880
Or something very close to a
chunk out of somebody else

234
00:12:49,880 --> 00:12:50,870
book, and publishing it.

235
00:12:50,870 --> 00:12:52,360
Some people are doing
that because they're

236
00:12:52,360 --> 00:12:55,190
sloppy, slimy people.

237
00:12:55,190 --> 00:12:57,400
Other people are doing
it because you

238
00:12:57,400 --> 00:12:59,730
read a lot of stuff.

239
00:12:59,730 --> 00:13:01,490
Maybe you took good notes,
maybe you didn't.

240
00:13:01,490 --> 00:13:05,450
And then when you start writing,
this clever phrase

241
00:13:05,450 --> 00:13:07,380
comes to you -- or maybe
just a pedestrian

242
00:13:07,380 --> 00:13:08,690
phrase comes to mind.

243
00:13:08,690 --> 00:13:10,050
But a phrase comes to mind.

244
00:13:10,050 --> 00:13:13,590
And you simply have forgotten
the source of that phrase.

245
00:13:13,590 --> 00:13:15,670
Which is that you got it
from somebody else's

246
00:13:15,670 --> 00:13:17,400
writing, and then --

247
00:13:17,400 --> 00:13:21,370
it's a strong argument for
making sure that you keep very

248
00:13:21,370 --> 00:13:23,830
clear track of where you're
getting your material from.

249
00:13:23,830 --> 00:13:26,590
And it's part of why we beat
so hard on the notion of

250
00:13:26,590 --> 00:13:31,980
referencing in papers in
a course like this.

251
00:13:31,980 --> 00:13:35,050
That's what [? crypt ?]

252
00:13:35,050 --> 00:13:39,600
amnesia is, in any case.

253
00:13:39,600 --> 00:13:42,230
So, those are failures
of source monitoring.

254
00:13:42,230 --> 00:13:46,590
It is also possible, to move
on to the next one, to have

255
00:13:46,590 --> 00:13:49,980
memories that feel
like memories but

256
00:13:49,980 --> 00:13:53,480
just are not true.

257
00:13:53,480 --> 00:13:56,600
Again, Liz Loftus has done some
of the most interesting

258
00:13:56,600 --> 00:13:59,910
research on this.

259
00:13:59,910 --> 00:14:03,450
One of her experiments, another
one of these car crash

260
00:14:03,450 --> 00:14:06,130
experiments, remember the
experiment of, how fast did

261
00:14:06,130 --> 00:14:10,220
the car smash, or bump, or tap,
or whatever it was, into

262
00:14:10,220 --> 00:14:11,130
the other car?

263
00:14:11,130 --> 00:14:12,420
Well, here's another one.

264
00:14:12,420 --> 00:14:17,150
You watch another one of these
videos of a car getting into

265
00:14:17,150 --> 00:14:19,310
an accident.

266
00:14:19,310 --> 00:14:21,740
And that part of the problem is
that the car went straight

267
00:14:21,740 --> 00:14:23,660
through a yield sign.

268
00:14:23,660 --> 00:14:27,810
Later, you are given information
that suggests that

269
00:14:27,810 --> 00:14:29,950
it might have been
a stop sign.

270
00:14:29,950 --> 00:14:32,760
Can do this in a variety of
ways, including ways that tell

271
00:14:32,760 --> 00:14:35,750
you that that is bad
information.

272
00:14:35,750 --> 00:14:39,040
Another source monitoring
problem.

273
00:14:39,040 --> 00:14:43,390
But, given later information
that it was a stop sign, not a

274
00:14:43,390 --> 00:14:47,990
yield sign, people will
incorrectly remember it --

275
00:14:47,990 --> 00:14:50,350
not everybody, but you will get
a percentage of people --

276
00:14:50,350 --> 00:14:54,780
who will now misremember the
original scene as having had a

277
00:14:54,780 --> 00:14:56,690
stop sign in it.

278
00:14:56,690 --> 00:15:01,250
If you ask what color was the
sign, they'll say red, even

279
00:15:01,250 --> 00:15:05,090
though a yield sign would be
yellow, and was yellow in the

280
00:15:05,090 --> 00:15:07,700
in the actual video
that you saw.

281
00:15:07,700 --> 00:15:11,820
So you've recreated, you've
now put together the true

282
00:15:11,820 --> 00:15:15,080
memory with some false
later information.

283
00:15:15,080 --> 00:15:17,760
And it's corrupted
that true memory.

284
00:15:17,760 --> 00:15:22,990
Now, that's of obviously some
interest in terms of getting

285
00:15:22,990 --> 00:15:25,480
testimony in court and
things like that.

286
00:15:25,480 --> 00:15:28,350
The place where this has become
extremely controversial

287
00:15:28,350 --> 00:15:37,530
and a real problem, is the
issue, is it possible to

288
00:15:37,530 --> 00:15:42,380
recover memories that
were suppressed?

289
00:15:42,380 --> 00:15:44,140
Suppose something really
bad happens to you

290
00:15:44,140 --> 00:15:44,960
when you're a kid.

291
00:15:44,960 --> 00:15:49,760
Is it possible to suppress that
memory and not have it

292
00:15:49,760 --> 00:15:51,720
available to you until
some later point.

293
00:15:51,720 --> 00:15:57,000
At which point you now
remember this memory.

294
00:15:57,000 --> 00:15:58,000
Is that possible?

295
00:15:58,000 --> 00:16:00,890
The evidence suggests that it
may indeed be possible to not

296
00:16:00,890 --> 00:16:02,420
think about something for
a very long time.

297
00:16:02,420 --> 00:16:05,030
Have it bubble up into
your memory.

298
00:16:05,030 --> 00:16:09,350
Sadly, the evidence is also
strong that it is possible for

299
00:16:09,350 --> 00:16:13,810
stuff to bubble up into your
memory that isn't true.

300
00:16:13,810 --> 00:16:17,000
Under the right circumstances.

301
00:16:17,000 --> 00:16:21,600
If what's bubbling up into
memory is memories of, for

302
00:16:21,600 --> 00:16:24,210
instance, childhood sexual
abuse, for which there were a

303
00:16:24,210 --> 00:16:28,120
large number of criminal cases
about a decade ago, now, it's

304
00:16:28,120 --> 00:16:31,350
a really important question
to know whether

305
00:16:31,350 --> 00:16:33,210
these are real memories.

306
00:16:33,210 --> 00:16:38,800
And it is deeply unclear that
there's any nice, neat way to

307
00:16:38,800 --> 00:16:39,810
know the answer.

308
00:16:39,810 --> 00:16:42,940
Now, look, obviously you can't,
one of the reasons it's

309
00:16:42,940 --> 00:16:46,200
very unclear, is you can't
do the real experiment.

310
00:16:46,200 --> 00:16:51,030
You can't take a little kid,
abuse the little kid and say,

311
00:16:51,030 --> 00:16:52,530
let's come back 20
years and see if

312
00:16:52,530 --> 00:16:53,510
they recover the memory.

313
00:16:53,510 --> 00:16:57,490
I mean, it doesn't take a genius
to figure out that

314
00:16:57,490 --> 00:16:58,720
that's not going to work.

315
00:16:58,720 --> 00:17:06,140
But, Liz Loftus has a very nice
experiment that points in

316
00:17:06,140 --> 00:17:12,840
the direction of this
sort of problem.

317
00:17:12,840 --> 00:17:18,290
Here's what she did: you go and
interview families about,

318
00:17:18,290 --> 00:17:20,270
let's take one specific
example.

319
00:17:20,270 --> 00:17:23,100
Getting lost in a mall.

320
00:17:23,100 --> 00:17:30,740
Did so-and-so, the designated
subject.

321
00:17:30,740 --> 00:17:31,870
You are --

322
00:17:31,870 --> 00:17:33,490
AUDIENCE: Olga.

323
00:17:33,490 --> 00:17:33,730
PROFESSOR: Olga.

324
00:17:33,730 --> 00:17:35,010
Did you ever get lost
in the mall?

325
00:17:35,010 --> 00:17:36,140
AUDIENCE: [UNINTELLIGIBLE]

326
00:17:36,140 --> 00:17:36,580
PROFESSOR: OK, good.

327
00:17:36,580 --> 00:17:37,870
We won't ask her that.

328
00:17:37,870 --> 00:17:40,920
But we'll ask her family
whether Olga ever

329
00:17:40,920 --> 00:17:42,750
got lost in the mall.

330
00:17:42,750 --> 00:17:44,300
And they'll say, no, no,
no, it never happened.

331
00:17:44,300 --> 00:17:45,560
OK, good.

332
00:17:45,560 --> 00:17:46,550
Got a brother or sister?

333
00:17:46,550 --> 00:17:47,190
AUDIENCE: Yeah.

334
00:17:47,190 --> 00:17:47,410
PROFESSOR: Good.

335
00:17:47,410 --> 00:17:48,910
What have you got?

336
00:17:48,910 --> 00:17:49,210
AUDIENCE: Brother.

337
00:17:49,210 --> 00:17:50,580
PROFESSOR: Brother.

338
00:17:50,580 --> 00:17:54,030
So now we call her brother aside
here, and say, we're

339
00:17:54,030 --> 00:17:56,100
going to do a little
experiment here.

340
00:17:56,100 --> 00:18:00,660
Go ask Olga if she remembers
the time that she

341
00:18:00,660 --> 00:18:03,160
got lost in the mall.

342
00:18:03,160 --> 00:18:06,430
And you do this with a lot of,
Olga might look at her brother

343
00:18:06,430 --> 00:18:08,430
and say, no.

344
00:18:08,430 --> 00:18:11,260
And that's the end of that
particular experiment.

345
00:18:11,260 --> 00:18:15,700
But some percentage of people,
given brother who seems to

346
00:18:15,700 --> 00:18:19,100
think that this happened at
some point, will, yeah, I

347
00:18:19,100 --> 00:18:19,980
remember that.

348
00:18:19,980 --> 00:18:20,760
Well, tell me about it.

349
00:18:20,760 --> 00:18:23,660
Well, you know how
this story goes.

350
00:18:23,660 --> 00:18:26,060
Even if you were never lost in
the mall, we could ask Olga.

351
00:18:26,060 --> 00:18:28,170
She could generate the story
pretty successfully.

352
00:18:28,170 --> 00:18:30,710
It'll be something like, well,
you know, I was there with my

353
00:18:30,710 --> 00:18:31,780
parents, right?

354
00:18:31,780 --> 00:18:35,660
And I was looking at these toys,
like, in the window.

355
00:18:35,660 --> 00:18:38,610
And then I followed these,
I was little,

356
00:18:38,610 --> 00:18:39,150
I was really little.

357
00:18:39,150 --> 00:18:40,400
So you don't get lost
in the mall when

358
00:18:40,400 --> 00:18:41,720
you're 18 or something.

359
00:18:41,720 --> 00:18:44,010
So, I was little.

360
00:18:44,010 --> 00:18:45,530
I'm following these
legs around.

361
00:18:45,530 --> 00:18:47,990
And it turns out to
be the wrong legs.

362
00:18:47,990 --> 00:18:50,430
I thought it was my mom and
it was somebody else.

363
00:18:50,430 --> 00:18:54,810
And I started to cry, and then
this big man came and he took

364
00:18:54,810 --> 00:18:57,810
me to the security guys.

365
00:18:57,810 --> 00:19:02,400
And they announced it over the
PA system, that I was lost.

366
00:19:02,400 --> 00:19:05,070
And my mother came back --

367
00:19:05,070 --> 00:19:08,700
people fill in really good,
detailed stories of this.

368
00:19:08,700 --> 00:19:13,810
That are apparently invented
out of -- well, not out of

369
00:19:13,810 --> 00:19:15,600
really whole cloth, I suppose.

370
00:19:15,600 --> 00:19:20,010
They're invented out of the
understanding that you have of

371
00:19:20,010 --> 00:19:23,560
the possibility of being
lost in the mall.

372
00:19:23,560 --> 00:19:25,900
And the childhood fear of
being lost in the mall.

373
00:19:25,900 --> 00:19:27,920
But it's not a real memory.

374
00:19:27,920 --> 00:19:29,120
Feels like a real memory.

375
00:19:29,120 --> 00:19:32,700
Subjects in Loftus's
experiments, confronted later

376
00:19:32,700 --> 00:19:37,320
with the reality that this was
all a setup, were often quite

377
00:19:37,320 --> 00:19:38,760
surprised --

378
00:19:38,760 --> 00:19:41,200
Sure I was never lost
in the mall?

379
00:19:41,200 --> 00:19:43,410
No, says Olga's brother.

380
00:19:43,410 --> 00:19:46,460
Liz Loftus just told me
to tell you that.

381
00:19:46,460 --> 00:19:49,020
And, oh, ah, hm, that's
a little weird.

382
00:19:49,020 --> 00:19:52,930
But, the point is it is possible
to generate things

383
00:19:52,930 --> 00:19:57,460
that feel like memories
that are not memories.

384
00:19:57,460 --> 00:19:58,480
Yes.

385
00:19:58,480 --> 00:19:58,991
AUDIENCE: What about

386
00:19:58,991 --> 00:20:04,606
remembering dreams, supposing
something happened in a dream

387
00:20:04,606 --> 00:20:07,159
that could happen, and you
remember that as a [INAUDIBLE]

388
00:20:09,700 --> 00:20:11,850
PROFESSOR: If you get routinely
confused about what

389
00:20:11,850 --> 00:20:16,460
happens in your dreams and
what happens in reality,

390
00:20:16,460 --> 00:20:20,880
that's known in the trade as a
failure of reality testing.

391
00:20:20,880 --> 00:20:25,440
And is not considered a
good psychiatric sign.

392
00:20:25,440 --> 00:20:30,950
But, sure, those sorts of
things are possible.

393
00:20:30,950 --> 00:20:36,090
Particularly when you end up
having desperately realistic

394
00:20:36,090 --> 00:20:38,880
kinds of dreams, the sorts of
things where these, sort of,

395
00:20:38,880 --> 00:20:40,700
reality testing things do.

396
00:20:40,700 --> 00:20:44,410
Some dirt is, how many people,
I'll probably ask this again

397
00:20:44,410 --> 00:20:45,730
when we talk about
sleep and dreams.

398
00:20:45,730 --> 00:20:46,670
But, never mind.

399
00:20:46,670 --> 00:20:49,330
How many people have had
the dream of their

400
00:20:49,330 --> 00:20:51,760
alarm clock going off?

401
00:20:51,760 --> 00:20:55,650
Dreamed that they turned off
their alarm clock and got up.

402
00:20:55,650 --> 00:21:02,310
Only to then discover,
oh, man, I'm in bed.

403
00:21:02,310 --> 00:21:05,580
And I'm supposed to be in
my psych recitation or

404
00:21:05,580 --> 00:21:06,970
something like that.

405
00:21:06,970 --> 00:21:11,150
Or, at least tried to give
that a story to their TA.

406
00:21:11,150 --> 00:21:15,370
So, yeah, there are possible
sources of

407
00:21:15,370 --> 00:21:18,350
confusion like that.

408
00:21:18,350 --> 00:21:25,300
The last item on this list of
sources of possible confusion

409
00:21:25,300 --> 00:21:27,860
I think, says something like
becoming plastic again.

410
00:21:27,860 --> 00:21:30,740
This is new stuff which I will
come to momentarily.

411
00:21:30,740 --> 00:21:32,170
AUDIENCE: [UNINTELLIGIBLE]

412
00:21:32,170 --> 00:21:33,160
PROFESSOR: That's OK.

413
00:21:33,160 --> 00:21:36,072
AUDIENCE: Is there a way to
distinguish between these

414
00:21:36,072 --> 00:21:39,010
false memories or if they've
actually occurred?

415
00:21:39,010 --> 00:21:43,120
[UNINTELLIGIBLE]

416
00:21:43,120 --> 00:21:45,636
If you don't know them
[UNINTELLIGIBLE]

417
00:21:45,636 --> 00:21:46,813
comes and says, I think I was
sexually abused as a child, is

418
00:21:46,813 --> 00:21:48,460
there a way to determine,
is this, like --

419
00:21:51,960 --> 00:21:55,930
PROFESSOR: There is, at the
present, time no ironclad way

420
00:21:55,930 --> 00:21:57,500
to know that.

421
00:21:57,500 --> 00:22:02,080
I mean, sometimes you can go and
get evidence -- there's no

422
00:22:02,080 --> 00:22:04,970
ironclad way to know simply
from the memory testimony.

423
00:22:04,970 --> 00:22:08,060
There may be ways to find out by
talking to other witnesses

424
00:22:08,060 --> 00:22:09,340
or something like that.

425
00:22:09,340 --> 00:22:12,640
There have been various efforts
to look, for instance,

426
00:22:12,640 --> 00:22:17,110
at brain activity and see
whether or not there's a

427
00:22:17,110 --> 00:22:18,850
difference in the pattern.

428
00:22:18,850 --> 00:22:20,550
So, you do this at the lab.

429
00:22:20,550 --> 00:22:24,260
Create a false memory and a real
memory, and see if you

430
00:22:24,260 --> 00:22:29,140
can see the difference somehow
in a FMRI scan or

431
00:22:29,140 --> 00:22:31,110
something of that sort.

432
00:22:31,110 --> 00:22:33,720
There are hints that maybe
you can get there.

433
00:22:33,720 --> 00:22:35,860
It's like lie detectors.

434
00:22:35,860 --> 00:22:37,710
A fascinating topic for
a whole lecture

435
00:22:37,710 --> 00:22:39,320
that I'm not giving.

436
00:22:39,320 --> 00:22:45,830
There is no ironclad way of
telling a lie from the truth

437
00:22:45,830 --> 00:22:47,590
by wiring anybody up.

438
00:22:47,590 --> 00:22:51,290
Regardless of what the
FBI tells you.

439
00:22:51,290 --> 00:22:53,360
Actually, I shouldn't be telling
you this, because the

440
00:22:53,360 --> 00:22:55,780
only reason that these things
work is because

441
00:22:55,780 --> 00:22:57,240
people think they work.

442
00:22:57,240 --> 00:22:59,630
If you think that the lie
detector's really going to

443
00:22:59,630 --> 00:23:03,100
work, then if you're lying, you
tend to get kind of more

444
00:23:03,100 --> 00:23:05,800
nervous than if you don't
think, than if

445
00:23:05,800 --> 00:23:06,820
you're telling the truth.

446
00:23:06,820 --> 00:23:08,580
And that's what they
can monitor.

447
00:23:08,580 --> 00:23:13,290
But a really good pathological
liar, or a well-trained agent

448
00:23:13,290 --> 00:23:16,220
is apparently pretty good
at getting past

449
00:23:16,220 --> 00:23:18,660
your average lie detector.

450
00:23:18,660 --> 00:23:21,860
Lots of people, you can get
serious defense money by

451
00:23:21,860 --> 00:23:24,130
having a brilliant new idea
about how to tell truth from

452
00:23:24,130 --> 00:23:27,160
lie and real memory
from false memory.

453
00:23:27,160 --> 00:23:29,130
There's a lot of interest
in doing that.

454
00:23:29,130 --> 00:23:35,760
But, no ironclad way of doing
it at the present.

455
00:23:35,760 --> 00:23:39,660
Makes for great sci-fi novels,
though, right what's going to

456
00:23:39,660 --> 00:23:43,250
happen when we finally figure
out that we can actually stick

457
00:23:43,250 --> 00:23:46,150
this thing on your head and
tell what your memories,

458
00:23:46,150 --> 00:23:49,180
whether your memories
are true or not.

459
00:23:49,180 --> 00:23:52,810
But at the moment it's
still sci-fi.

460
00:23:52,810 --> 00:23:59,000
Not quite sci-fi, but still very
new, is the indication.

461
00:23:59,000 --> 00:24:01,350
Remember, I talked about
consolidation last time.

462
00:24:01,350 --> 00:24:04,730
This process of making a fragile
memory into something

463
00:24:04,730 --> 00:24:06,950
that will last for
years and years.

464
00:24:06,950 --> 00:24:09,620
There is evidence at least
from rats in a fairly

465
00:24:09,620 --> 00:24:15,820
restricted range of studies,
that when you recall a memory,

466
00:24:15,820 --> 00:24:19,630
when you bring it back out of,
say, long-term memory into

467
00:24:19,630 --> 00:24:23,290
that working memory, that it
becomes plastic again.

468
00:24:23,290 --> 00:24:26,510
And that perhaps some of the
distortions in memory can

469
00:24:26,510 --> 00:24:28,100
arise at that point.

470
00:24:28,100 --> 00:24:29,940
Because the memory
is again plastic.

471
00:24:29,940 --> 00:24:32,720
It needs to be reconsolidated
in order to

472
00:24:32,720 --> 00:24:35,770
be, in effect, restored.

473
00:24:35,770 --> 00:24:38,780
So you bring it out.

474
00:24:38,780 --> 00:24:43,730
It's now vulnerable again in
ways that it wasn't before you

475
00:24:43,730 --> 00:24:44,480
remembered it.

476
00:24:44,480 --> 00:24:49,230
My guess is that if you come
back to the course ten years

477
00:24:49,230 --> 00:24:52,590
from now, that there'll
be much more to

478
00:24:52,590 --> 00:24:53,540
be said about that.

479
00:24:53,540 --> 00:24:58,490
That's new and interesting.

480
00:24:58,490 --> 00:25:00,680
But the general point should
be clear enough.

481
00:25:00,680 --> 00:25:05,000
Which is that recall out of
memory, retrieval out of

482
00:25:05,000 --> 00:25:07,270
memory, is an active
reconstruction.

483
00:25:07,270 --> 00:25:10,580
It has multiple pitfalls in it
that might cause what you are

484
00:25:10,580 --> 00:25:15,120
pulling out as a memory
to be distorted in

485
00:25:15,120 --> 00:25:20,120
some fashion or other.

486
00:25:20,120 --> 00:25:24,020
Now, the second part, the larger
part, of what I want to

487
00:25:24,020 --> 00:25:30,500
talk about today is the sense
in which we are not simple

488
00:25:30,500 --> 00:25:35,150
logic engines in the way that
our computers, most

489
00:25:35,150 --> 00:25:36,380
of the time, are.

490
00:25:36,380 --> 00:25:42,880
Now, one way to think about
that is to think about a

491
00:25:42,880 --> 00:25:47,400
distinction between what's
called narrative thought and

492
00:25:47,400 --> 00:25:50,290
propositional thought.

493
00:25:50,290 --> 00:25:53,110
Propositional thought, and
this runs parallel.

494
00:25:53,110 --> 00:25:55,510
Narrative thought and
episodic memory.

495
00:25:55,510 --> 00:25:57,780
Propositional thought
and semantic memory.

496
00:25:57,780 --> 00:25:59,640
They travel together.

497
00:25:59,640 --> 00:26:07,100
So, propositional thought is the
sort of thought that your

498
00:26:07,100 --> 00:26:09,550
computer would be good at.

499
00:26:09,550 --> 00:26:17,530
Moving symbols around, is x
bigger than y, and problem set

500
00:26:17,530 --> 00:26:20,900
kind of thinking is mostly
propositional thought.

501
00:26:20,900 --> 00:26:25,040
Narrative thought is
thought that has a

502
00:26:25,040 --> 00:26:28,580
story-like aspect to it.

503
00:26:28,580 --> 00:26:31,740
Perhaps an episodic memory
type aspect to it.

504
00:26:31,740 --> 00:26:41,440
So, an example of narrative
thought might be do you want

505
00:26:41,440 --> 00:26:49,090
to go to the party that is
happening on Saturday night at

506
00:26:49,090 --> 00:26:52,120
Alpha Beta Gamma house, or
something like that.

507
00:26:52,120 --> 00:26:53,420
How do you figure that out?

508
00:26:53,420 --> 00:26:55,950
Well you probably don't
do some symbolic

509
00:26:55,950 --> 00:26:57,620
cost-benefit analysis.

510
00:26:57,620 --> 00:27:02,440
You probably imagine
yourself there.

511
00:27:02,440 --> 00:27:04,080
Would I be having fun?

512
00:27:04,080 --> 00:27:06,460
Who's going to be there.

513
00:27:06,460 --> 00:27:07,640
You'd run a scenario.

514
00:27:07,640 --> 00:27:10,300
And that's the sort of
narrative thought.

515
00:27:10,300 --> 00:27:14,860
Turns out that narrative thought
has a power in our own

516
00:27:14,860 --> 00:27:21,320
decision-making that is capable
of running across,

517
00:27:21,320 --> 00:27:24,560
running over, the simple
facts of the situation.

518
00:27:24,560 --> 00:27:28,790
Even when we know the
facts correctly.

519
00:27:28,790 --> 00:27:32,350
Risk assessment is one
example of this.

520
00:27:32,350 --> 00:27:33,910
So, forced choice.

521
00:27:33,910 --> 00:27:35,040
You gotta pick one
of the other.

522
00:27:35,040 --> 00:27:38,640
Don't just sit on
your hands here.

523
00:27:38,640 --> 00:27:44,390
Which makes you more nervous:
taking a ride in the car or

524
00:27:44,390 --> 00:27:45,370
flying in an airplane?

525
00:27:45,370 --> 00:27:47,810
How many vote for riding
in the car?

526
00:27:47,810 --> 00:27:51,580
How many vote for flying
an airplane.

527
00:27:51,580 --> 00:27:55,470
As usual half the people can't
figure out that forced choice,

528
00:27:55,470 --> 00:27:57,650
pick one or the other, means
you have to raise your hand

529
00:27:57,650 --> 00:28:00,140
for one of these two options.

530
00:28:00,140 --> 00:28:02,730
At least if you're
moderately alert.

531
00:28:02,730 --> 00:28:06,140
But anyway, plane
wins big-time.

532
00:28:06,140 --> 00:28:10,760
How many people, let's just ask
the propositional thought,

533
00:28:10,760 --> 00:28:13,200
or the semantic memory nugget.

534
00:28:13,200 --> 00:28:17,120
Which is more dangerous?

535
00:28:17,120 --> 00:28:20,650
Flying in a car,
that's only --

536
00:28:24,350 --> 00:28:27,170
that's the Harry
Potter version.

537
00:28:27,170 --> 00:28:29,010
Flying a plane or
driving a car.

538
00:28:29,010 --> 00:28:31,140
How many vote for plane?

539
00:28:31,140 --> 00:28:32,830
How many vote for car?

540
00:28:32,830 --> 00:28:36,950
Everybody knows the
proposition here.

541
00:28:36,950 --> 00:28:41,500
It's riskier to, you can measure
this in all sorts of

542
00:28:41,500 --> 00:28:42,710
different ways.

543
00:28:42,710 --> 00:28:44,780
Per mile, per time.

544
00:28:44,780 --> 00:28:47,570
It's it's riskier traveling
in a car than

545
00:28:47,570 --> 00:28:51,880
traveling in a plane.

546
00:28:51,880 --> 00:28:53,880
But people are much more
nervous about planes.

547
00:28:53,880 --> 00:28:54,700
Why is that?

548
00:28:54,700 --> 00:28:55,650
There are multiple reasons.

549
00:28:55,650 --> 00:29:00,790
But perhaps one of the driving
ones is that the narrative

550
00:29:00,790 --> 00:29:05,220
surrounding planes, and
particularly surrounding plane

551
00:29:05,220 --> 00:29:08,940
crashes is sufficiently vivid
that it overrides your

552
00:29:08,940 --> 00:29:12,870
knowledge of the mere
facts of the matter.

553
00:29:12,870 --> 00:29:17,590
If and when a plane sadly goes
down, that's big news.

554
00:29:17,590 --> 00:29:19,630
A lot of people tend
to get killed.

555
00:29:19,630 --> 00:29:23,920
It makes big news in the
papers and on TV.

556
00:29:23,920 --> 00:29:25,360
Highly salient.

557
00:29:25,360 --> 00:29:32,760
The fact that there's a steady
drumbeat of car crashes,

558
00:29:32,760 --> 00:29:36,740
unless it happens to be very
local to you, doesn't have any

559
00:29:36,740 --> 00:29:38,270
particular impact on you.

560
00:29:38,270 --> 00:29:41,490
And the raw statistic
doesn't much matter.

561
00:29:41,490 --> 00:29:44,640
In the same way, this turns
out to be true in course

562
00:29:44,640 --> 00:29:47,890
catalog land.

563
00:29:47,890 --> 00:29:52,390
So that the MIT course
evaluation guide, when it

564
00:29:52,390 --> 00:29:58,370
comes out, has vast piles
of cool statistics.

565
00:29:58,370 --> 00:30:00,060
What people thought of this
particular course.

566
00:30:00,060 --> 00:30:00,650
And then it tends --

567
00:30:00,650 --> 00:30:02,170
I haven't seen the
most recent one.

568
00:30:02,170 --> 00:30:06,550
But typically it has a paragraph
of comments.

569
00:30:06,550 --> 00:30:10,000
One of my favorites over the
years, so-and-so is to the

570
00:30:10,000 --> 00:30:16,990
teaching of physics as Einstein
is to the National

571
00:30:16,990 --> 00:30:18,240
Football League.

572
00:30:21,120 --> 00:30:23,360
That's a very salient
bit of narrative.

573
00:30:23,360 --> 00:30:26,510
And it turns out that those
sort of comments have much

574
00:30:26,510 --> 00:30:28,870
more impact than all
the statistics.

575
00:30:28,870 --> 00:30:33,430
In spite of the fact that the
nice rational you knows that

576
00:30:33,430 --> 00:30:37,550
one person's brilliant catty
comment, or brilliantly

577
00:30:37,550 --> 00:30:40,910
positive comment, is presumably
much less valuable

578
00:30:40,910 --> 00:30:43,860
than the statistical average
across 300 students.

579
00:30:43,860 --> 00:30:46,170
But the narrative, again,
overwhelms.

580
00:30:46,170 --> 00:30:51,800
You can watch this on Friday,
when you watch the debate.

581
00:30:51,800 --> 00:30:59,510
You can be guaranteed that
somebody is going to, it'll

582
00:30:59,510 --> 00:31:01,880
probably be either Kerry
or Bush, actually.

583
00:31:01,880 --> 00:31:09,990
That Kerry or Bush will take
some boring statistics and

584
00:31:09,990 --> 00:31:11,610
personalize it.

585
00:31:11,610 --> 00:31:12,850
They do this all the time.

586
00:31:12,850 --> 00:31:17,040
It's become a sort of it a
trope, a standard item in the

587
00:31:17,040 --> 00:31:20,730
State of the Union message where
the President always put

588
00:31:20,730 --> 00:31:22,480
a couple of heroes.

589
00:31:22,480 --> 00:31:24,930
They probably are, but, you
know, heroes up in the balcony

590
00:31:24,930 --> 00:31:27,860
there so that he can point
to them and say,

591
00:31:27,860 --> 00:31:28,770
look at this guy.

592
00:31:28,770 --> 00:31:29,420
Look at that guy.

593
00:31:29,420 --> 00:31:35,630
And in the debate it'll be,
there'll be some discussion of

594
00:31:35,630 --> 00:31:37,360
employment statistics.

595
00:31:37,360 --> 00:31:40,350
And presumably this
would be Kerry.

596
00:31:40,350 --> 00:31:42,240
Because he'll be beating
on Bush about it.

597
00:31:42,240 --> 00:31:44,960
I don't care about what you
say about the economy

598
00:31:44,960 --> 00:31:46,250
recovering or whatever.

599
00:31:46,250 --> 00:31:50,450
What I care about is
poor little Olga.

600
00:31:50,450 --> 00:31:54,270
She got lost in the mall, and
then her father lost her job

601
00:31:54,270 --> 00:31:58,680
as a mall security guy because
of the decisions that your

602
00:31:58,680 --> 00:32:01,770
government, your administration
made.

603
00:32:01,770 --> 00:32:05,580
And what am I going to say
to poor little Olga.

604
00:32:05,580 --> 00:32:08,820
Why?

605
00:32:08,820 --> 00:32:11,430
It's very sad about poor little
Olga, who's now been

606
00:32:11,430 --> 00:32:14,340
lost in the mall and she's
never going to sit in the

607
00:32:14,340 --> 00:32:16,520
front again.

608
00:32:16,520 --> 00:32:22,500
But there are, like, 260 million
people, whatever it is

609
00:32:22,500 --> 00:32:23,660
in the country at the moment.

610
00:32:23,660 --> 00:32:27,030
It's very sad about poor
little whoever it is.

611
00:32:27,030 --> 00:32:30,530
What's really important are the
broad economic statistics.

612
00:32:30,530 --> 00:32:35,030
Broad economic statistics are
wonderful propositional

613
00:32:35,030 --> 00:32:36,590
reasoning kinds of stuff.

614
00:32:36,590 --> 00:32:38,510
But are really boring.

615
00:32:38,510 --> 00:32:40,000
Unless you're an economist.

616
00:32:40,000 --> 00:32:42,690
But poor little Olga
is really riveting.

617
00:32:42,690 --> 00:32:44,390
And poor little Olga
is what you'll hear

618
00:32:44,390 --> 00:32:47,800
about in the debates.

619
00:32:47,800 --> 00:32:52,240
We should do a debriefing
about this on Monday.

620
00:32:52,240 --> 00:32:54,340
If anybody listens to the debate
and hears any of these

621
00:32:54,340 --> 00:32:56,540
little tidbits, send me email.

622
00:32:59,620 --> 00:33:01,500
In case I miss it.

623
00:33:01,500 --> 00:33:03,810
And then we can we about this.

624
00:33:03,810 --> 00:33:07,680
Now, this can get studied
in the lab.

625
00:33:07,680 --> 00:33:08,430
Here, take a look.

626
00:33:08,430 --> 00:33:12,720
On the handout, there's a thing
called the framing demo.

627
00:33:12,720 --> 00:33:15,410
Says that there's an
unusual flu coming.

628
00:33:15,410 --> 00:33:18,880
It's expected to kill 600 people
in the U.S., and there

629
00:33:18,880 --> 00:33:20,190
are two things you can do.

630
00:33:20,190 --> 00:33:24,740
There's Program A and Program
B. Take a quick look, decide

631
00:33:24,740 --> 00:33:29,200
on Program A or Program B, and
then we'll collect a little

632
00:33:29,200 --> 00:33:31,040
data and talk about it here.

633
00:33:37,190 --> 00:33:41,770
Don't cheat off your
neighbor's test.

634
00:33:41,770 --> 00:33:42,230
OK.

635
00:33:42,230 --> 00:33:46,060
Everybody manage to come
up with an answer here?

636
00:33:46,060 --> 00:33:46,910
OK.

637
00:33:46,910 --> 00:33:49,340
So, let's see for starters.

638
00:33:49,340 --> 00:33:51,470
How many people went for A?

639
00:33:54,210 --> 00:33:55,820
How many people went for B?

640
00:33:58,460 --> 00:34:01,600
How many people don't know
what they're going for?

641
00:34:01,600 --> 00:34:03,380
OK, well it looks about
evenly divided.

642
00:34:03,380 --> 00:34:04,330
Boy, that's really boring.

643
00:34:04,330 --> 00:34:07,270
Well, it would be really boring,
except that there are

644
00:34:07,270 --> 00:34:11,620
two versions of the
handout out there.

645
00:34:11,620 --> 00:34:19,040
So, Version One of the handout
says that if Program A is

646
00:34:19,040 --> 00:34:22,440
adopted, uh-oh, I don't know
which Version One says.

647
00:34:22,440 --> 00:34:25,680
Well, it doesn't matter which
I call Version One, does it?

648
00:34:25,680 --> 00:34:31,120
Version One says, 400
people will die.

649
00:34:31,120 --> 00:34:39,030
Version Two says 200 people
will be saved.

650
00:34:41,770 --> 00:34:43,870
Did I get the language
right there?

651
00:34:43,870 --> 00:34:47,000
What you will notice it, if the
net is 600 here, these are

652
00:34:47,000 --> 00:34:48,560
identical, right?

653
00:34:48,560 --> 00:34:51,530
There's no difference except in
the way that it's phrased.

654
00:34:51,530 --> 00:34:56,780
Or the way the question
is framed.

655
00:34:56,780 --> 00:35:01,140
There's a similar change
to the second option,

656
00:35:01,140 --> 00:35:03,270
B in the die --

657
00:35:03,270 --> 00:35:07,360
in this person, it's described
as a 1/3 chance that no-one

658
00:35:07,360 --> 00:35:12,390
will die and a 2/3 chance that
600 hundred people will die.

659
00:35:12,390 --> 00:35:14,960
Do you know about the notion
of expected value?

660
00:35:14,960 --> 00:35:18,910
You can calculate the expected
value of that option, which

661
00:35:18,910 --> 00:35:34,540
will be 1/3 chance that no-one
will die and a 2/3 chance that

662
00:35:34,540 --> 00:35:38,420
600 people will die.

663
00:35:38,420 --> 00:35:42,750
And then that's going to be 400
people dying, again, here.

664
00:35:42,750 --> 00:35:48,010
So all the options are
mathematically identical.

665
00:35:48,010 --> 00:35:52,900
Now, how many people have
Version One, the 400 hundred

666
00:35:52,900 --> 00:35:55,310
people will die thing?

667
00:35:55,310 --> 00:35:58,680
OK, how many people have the
200 people will die thing?

668
00:35:58,680 --> 00:35:59,660
OK, good.

669
00:35:59,660 --> 00:36:02,020
So, it's roughly
evenly divided.

670
00:36:02,020 --> 00:36:05,050
Now, what I want to do is, I
want to get the A and B for

671
00:36:05,050 --> 00:36:06,700
these two different versions.

672
00:36:06,700 --> 00:36:10,170
If you have the 400 people will
die version, how many

673
00:36:10,170 --> 00:36:14,300
people opted for Option A?

674
00:36:14,300 --> 00:36:17,370
How many people opted
for Option B?

675
00:36:17,370 --> 00:36:22,510
OK, so B beats A, and actually
it beat A big time here.

676
00:36:22,510 --> 00:36:26,210
If you have the 200 people will
be saved version, how

677
00:36:26,210 --> 00:36:29,030
many people went for A?

678
00:36:29,030 --> 00:36:31,260
How many people went for B?

679
00:36:31,260 --> 00:36:35,990
OK, in that case A beats B
not quite as massively.

680
00:36:35,990 --> 00:36:40,710
But, clearly this manipulation
has flipped the answer.

681
00:36:40,710 --> 00:36:44,260
It ain't math that's doing it,
because the math is the same.

682
00:36:44,260 --> 00:36:48,750
What's changing here is what
Kahneman and Tversky called

683
00:36:48,750 --> 00:36:51,270
the framing of the problem.

684
00:36:51,270 --> 00:36:53,650
But really it's the narrative
around the problem.

685
00:36:53,650 --> 00:36:59,280
What happens is, you read Option
A. 400 people will die.

686
00:36:59,280 --> 00:37:01,100
That's terrible.

687
00:37:01,100 --> 00:37:04,665
Well, the second possibility in
this case, the B is, woah,

688
00:37:04,665 --> 00:37:07,050
there's something we could do.

689
00:37:07,050 --> 00:37:11,240
That probabilistically might
save some of those people.

690
00:37:11,240 --> 00:37:12,090
That sounds OK.

691
00:37:12,090 --> 00:37:13,730
I'll opt for this.

692
00:37:13,730 --> 00:37:18,240
As opposed to this one, 200
people will be saved.

693
00:37:18,240 --> 00:37:20,430
Well, that's better
than nothing.

694
00:37:20,430 --> 00:37:23,580
And the alternative, B, here.

695
00:37:23,580 --> 00:37:27,690
You read the part that says,
everybody might die.

696
00:37:27,690 --> 00:37:30,240
Hm, doesn't sound too good.

697
00:37:30,240 --> 00:37:32,080
And so you flip it.

698
00:37:32,080 --> 00:37:34,980
So what you're doing here is
the way that you ask the

699
00:37:34,980 --> 00:37:38,400
question changes the
answer you get.

700
00:37:38,400 --> 00:37:42,370
Well, we can go back to the
politics game here just fine.

701
00:37:42,370 --> 00:37:44,700
Pollsters do this
all the time.

702
00:37:44,700 --> 00:37:47,990
The neutral pollsters
desperately try not to do it.

703
00:37:47,990 --> 00:37:50,270
But campaign polls, when they're
trying to get the

704
00:37:50,270 --> 00:37:57,590
correct answer out, who would
you rather have for President,

705
00:37:57,590 --> 00:38:03,855
a strong leader who has ruled
the free world with a firm

706
00:38:03,855 --> 00:38:08,890
hand for four years, or a
flip-flopping senator from a

707
00:38:08,890 --> 00:38:11,650
liberal East Coast state?

708
00:38:11,650 --> 00:38:14,620
Or you could ask the question,
who would you like to have as

709
00:38:14,620 --> 00:38:19,220
your President for the next four
years, a guy who barely

710
00:38:19,220 --> 00:38:21,440
made it out of Yale and has
driven the country into a

711
00:38:21,440 --> 00:38:26,140
ditch, or a guy who actually got
decent grades at Yale and

712
00:38:26,140 --> 00:38:30,100
is a national war hero and
has really good hair?

713
00:38:34,040 --> 00:38:41,160
You can move the question
around, by the way, and not

714
00:38:41,160 --> 00:38:42,260
just in polling.

715
00:38:42,260 --> 00:38:44,960
But the way you frame issues.

716
00:38:44,960 --> 00:38:47,650
I mean, talk about
framing issues.

717
00:38:47,650 --> 00:38:55,650
If you frame the war in Iraq as
part of the War on Terror,

718
00:38:55,650 --> 00:38:58,490
and an integral part of the War
on Terror asking whether

719
00:38:58,490 --> 00:39:01,370
you want to keep doing that, is
different than taking the

720
00:39:01,370 --> 00:39:07,400
same thing and calling it a
distraction from the War on

721
00:39:07,400 --> 00:39:09,530
Terror, for example.

722
00:39:09,530 --> 00:39:12,250
The facts remain sort
of similar.

723
00:39:12,250 --> 00:39:14,640
It's harder to do the experiment
in a nice, clean,

724
00:39:14,640 --> 00:39:15,690
controlled --

725
00:39:15,690 --> 00:39:18,360
here, we can make all the
math come out right.

726
00:39:18,360 --> 00:39:21,460
In a political debate, there's
an argument about the actual

727
00:39:21,460 --> 00:39:22,710
facts, too, of course.

728
00:39:22,710 --> 00:39:25,825
But the way that you present the
facts, the way you frame

729
00:39:25,825 --> 00:39:30,820
the facts, is designed to get
the answer that you want out

730
00:39:30,820 --> 00:39:33,440
of those facts.

731
00:39:33,440 --> 00:39:36,140
There's a beautiful round of
this in today's paper, because

732
00:39:36,140 --> 00:39:39,540
there's the report of the
whatever they're called, Iraq

733
00:39:39,540 --> 00:39:42,620
Commission that was looking for
weapons and didn't find

734
00:39:42,620 --> 00:39:43,400
any weapons.

735
00:39:43,400 --> 00:39:47,470
And if you read the Kerry camp
statements about what that

736
00:39:47,470 --> 00:39:50,990
means, and what the Bush
camp says it means.

737
00:39:50,990 --> 00:39:54,020
The Bush camp is very big on
intention this morning.

738
00:39:54,020 --> 00:39:59,600
He really, the report that
Saddam wanted weapons.

739
00:39:59,600 --> 00:40:03,970
And so the Bush camp is busy
saying, he wanted weapons and

740
00:40:03,970 --> 00:40:05,470
he would have gotten them
as soon as he could.

741
00:40:05,470 --> 00:40:10,240
And the Kerry camp is looking at
the same facts and saying,

742
00:40:10,240 --> 00:40:11,180
we all want stuff.

743
00:40:11,180 --> 00:40:14,180
Man, he didn't have anything.

744
00:40:14,180 --> 00:40:14,520
So.

745
00:40:14,520 --> 00:40:16,970
Anyway, you can have
lots of fun Friday.

746
00:40:16,970 --> 00:40:19,680
You can watch them
do this one, too.

747
00:40:19,680 --> 00:40:21,380
All right, so.

748
00:40:21,380 --> 00:40:25,340
The narrative that you
put around a story.

749
00:40:25,340 --> 00:40:27,380
The narrative that you put
around the story does not

750
00:40:27,380 --> 00:40:29,220
influence your computer.

751
00:40:29,220 --> 00:40:34,820
You can tell your computer about
1/3 times 600 plus 2/3

752
00:40:34,820 --> 00:40:37,430
times minus 600, any
way you like.

753
00:40:37,430 --> 00:40:40,650
You can use big font, you can
use little font, you can use

754
00:40:40,650 --> 00:40:43,010
you want, it's going to come
out with the same answer.

755
00:40:43,010 --> 00:40:45,960
Not true of the way that you
think about these things.

756
00:40:45,960 --> 00:40:51,420
You're going to get different
answers depending on how you

757
00:40:51,420 --> 00:40:53,440
frame the question.

758
00:40:53,440 --> 00:40:56,850
Let me do a different example.

759
00:40:56,850 --> 00:40:58,840
That's not my handout, I got to
see what's actually on the

760
00:40:58,840 --> 00:41:00,870
handout here.

761
00:41:00,870 --> 00:41:03,800
So, right underneath that,
underneath the bit about

762
00:41:03,800 --> 00:41:05,800
framing, there's another demo.

763
00:41:05,800 --> 00:41:10,240
I will, in advance, promise you
that this is the same on

764
00:41:10,240 --> 00:41:12,410
all the handouts.

765
00:41:12,410 --> 00:41:14,980
Because otherwise you're going
to sit there trying to figure

766
00:41:14,980 --> 00:41:16,180
out what's the weird thing.

767
00:41:16,180 --> 00:41:18,570
So, this is the same on
all the handouts.

768
00:41:18,570 --> 00:41:22,580
I'm also aware that I have six
slots between most probable

769
00:41:22,580 --> 00:41:23,620
and least probable.

770
00:41:23,620 --> 00:41:27,040
And only five options
to go in there.

771
00:41:27,040 --> 00:41:29,510
Deal with it.

772
00:41:29,510 --> 00:41:31,040
Here's what you've got to do.

773
00:41:31,040 --> 00:41:32,340
You've got this guy Henry.

774
00:41:32,340 --> 00:41:34,080
Henry is a short, slim man.

775
00:41:34,080 --> 00:41:35,240
He likes to read poetry.

776
00:41:35,240 --> 00:41:36,630
He's been active in
environmental

777
00:41:36,630 --> 00:41:37,960
and feminist causes.

778
00:41:37,960 --> 00:41:41,230
And Henry is a what?

779
00:41:41,230 --> 00:41:44,920
And so there are these
five options here.

780
00:41:44,920 --> 00:41:48,280
What you want to do is rank
order them from most likely to

781
00:41:48,280 --> 00:41:49,530
least likely.

782
00:41:53,880 --> 00:41:56,150
And then we'll talk
about that a bit.

783
00:41:56,150 --> 00:42:00,110
Let's see how that --

784
00:42:03,140 --> 00:42:07,220
No, you can't copy off
your neighbor.

785
00:42:07,220 --> 00:42:08,470
You know Henry.

786
00:42:16,140 --> 00:42:17,850
I should have brought
Henry in.

787
00:42:17,850 --> 00:42:21,210
Just grabbed some character
and said, this is Henry.

788
00:42:21,210 --> 00:42:21,730
All right.

789
00:42:21,730 --> 00:42:24,080
How many people need
more time for this?

790
00:42:24,080 --> 00:42:26,450
OK, a couple of people
still working on it.

791
00:42:35,480 --> 00:42:36,210
All right.

792
00:42:36,210 --> 00:42:42,250
Well, we'll just go up.

793
00:42:42,250 --> 00:42:44,410
Let's look at a couple
of particular

794
00:42:44,410 --> 00:42:47,440
comparisons on here.

795
00:42:50,370 --> 00:42:56,440
How many people put down, let's
look at A and B. Whoops.

796
00:42:56,440 --> 00:43:00,310
That looks propositional
thought.

797
00:43:04,920 --> 00:43:06,470
All gone.

798
00:43:06,470 --> 00:43:09,540
OK, let's look at A and
B, for example.

799
00:43:09,540 --> 00:43:14,420
How many people said A was
more probable than B?

800
00:43:14,420 --> 00:43:17,550
How many people said B was
more probable than A?

801
00:43:17,550 --> 00:43:18,060
All right.

802
00:43:18,060 --> 00:43:19,840
B wins.

803
00:43:19,840 --> 00:43:28,630
And if I've got the right one
here, so ivy beats truck

804
00:43:28,630 --> 00:43:31,550
big-time, right?

805
00:43:31,550 --> 00:43:33,180
All right.

806
00:43:33,180 --> 00:43:38,130
How many truck drivers are
there in the country?

807
00:43:38,130 --> 00:43:38,620
AUDIENCE: A lot.

808
00:43:38,620 --> 00:43:39,380
PROFESSOR: A lot.

809
00:43:39,380 --> 00:43:41,130
10 to the --

810
00:43:41,130 --> 00:43:44,170
I don't know the answer
to this, but?

811
00:43:44,170 --> 00:43:45,760
5?

812
00:43:45,760 --> 00:43:48,200
4 is only a thousand.

813
00:43:48,200 --> 00:43:48,960
AUDIENCE: 5.

814
00:43:48,960 --> 00:43:53,260
PROFESSOR: 5 is probably-- it
may be 6, but let's go with 5.

815
00:43:53,260 --> 00:43:55,250
OK.

816
00:43:55,250 --> 00:43:58,030
How many Ivy League classics
professors are

817
00:43:58,030 --> 00:43:59,280
there in the country?

818
00:44:04,900 --> 00:44:07,120
How many Ivy League
schools are there?

819
00:44:07,120 --> 00:44:07,740
AUDIENCE: Eight.

820
00:44:07,740 --> 00:44:09,670
PROFESSOR: Eight to ten,
something like that.

821
00:44:09,670 --> 00:44:11,840
Maybe, if you're lucky,
there's ten classics

822
00:44:11,840 --> 00:44:13,110
professors at each.

823
00:44:13,110 --> 00:44:16,320
So that's going to get
you to 10 to the 2nd.

824
00:44:16,320 --> 00:44:21,630
Is it a thousand times more
likely that some random guy

825
00:44:21,630 --> 00:44:26,290
that you meet, who happens to be
-- well, let's phrase this

826
00:44:26,290 --> 00:44:26,890
a different way.

827
00:44:26,890 --> 00:44:30,790
If you took all, whatever this
is, a hundred thousand truck

828
00:44:30,790 --> 00:44:31,800
drivers in the country.

829
00:44:31,800 --> 00:44:32,760
I'm sure there are
more than that.

830
00:44:32,760 --> 00:44:34,490
But let's just suppose there
are only a hundred thousand

831
00:44:34,490 --> 00:44:36,710
truck drivers in the country.

832
00:44:36,710 --> 00:44:42,940
Are there fewer than a hundred
of them who might be, for

833
00:44:42,940 --> 00:44:46,330
instance, short, poetry-loving
feminists?

834
00:44:49,780 --> 00:44:53,090
And that assumes that every
one of the Ivy League

835
00:44:53,090 --> 00:44:56,480
professors is a short --

836
00:44:56,480 --> 00:45:00,120
there's no Ivy Leagues
professors who are big, burly

837
00:45:00,120 --> 00:45:02,640
guys with tattoos and stuff.

838
00:45:02,640 --> 00:45:04,870
Or women, for that matter.

839
00:45:04,870 --> 00:45:09,110
Probably half of them are women,
so this is already --

840
00:45:09,110 --> 00:45:11,520
What you've done here is you've
ignored what's known as

841
00:45:11,520 --> 00:45:13,130
the base rate.

842
00:45:13,130 --> 00:45:20,000
All else being equal, it's more
likely that, if I just

843
00:45:20,000 --> 00:45:24,630
say, here's Henry, is it more
likely that Henry is a truck

844
00:45:24,630 --> 00:45:26,620
driver or an Ivy League
classics professor?

845
00:45:26,620 --> 00:45:28,950
It's way, way, way more likely
that he's a truck driver.

846
00:45:28,950 --> 00:45:31,300
In fact, it's so much more
likely that he's a truck

847
00:45:31,300 --> 00:45:34,240
driver that it's probably
more likely.

848
00:45:34,240 --> 00:45:35,990
It's almost undoubtedly
more likely.

849
00:45:35,990 --> 00:45:40,140
Even for this description of
Henry the amazing feminist,

850
00:45:40,140 --> 00:45:42,630
whatever he was.

851
00:45:42,630 --> 00:45:45,740
You see, this is what's known
as the base rate fallacy.

852
00:45:45,740 --> 00:45:50,700
People do not tend to take into
account the base rate in

853
00:45:50,700 --> 00:45:53,700
the population when they're
thinking about this.

854
00:45:53,700 --> 00:45:56,660
I'm asking how likely it is,
and you're answering a

855
00:45:56,660 --> 00:45:58,070
different question.

856
00:45:58,070 --> 00:46:04,830
You're answering the question,
how typical is Henry of your

857
00:46:04,830 --> 00:46:11,010
image, of your stereotype, of
a truck driver versus of an

858
00:46:11,010 --> 00:46:14,010
Ivy League professor?

859
00:46:14,010 --> 00:46:17,490
And realize that even if
-- let's suppose your

860
00:46:17,490 --> 00:46:19,370
stereotype is right.

861
00:46:19,370 --> 00:46:23,700
What's the chance that Henry
could be a truck driver?

862
00:46:23,700 --> 00:46:25,610
Maybe it's only 1 in 100.

863
00:46:25,610 --> 00:46:27,690
All right, if it's 1 in
100, there's still 10

864
00:46:27,690 --> 00:46:29,820
to the 3rd of them.

865
00:46:29,820 --> 00:46:32,970
And we're still swamping
the population here.

866
00:46:32,970 --> 00:46:38,040
So, failing to take into account
the base rate is going

867
00:46:38,040 --> 00:46:40,490
to lead you to the wrong
answer there's.

868
00:46:40,490 --> 00:46:44,620
Let's look at another
piece of these data.

869
00:46:44,620 --> 00:46:46,810
Let's look at --

870
00:46:52,440 --> 00:47:01,850
Is it more likely, A and E. A,
E. How many people said that A

871
00:47:01,850 --> 00:47:05,070
was more likely than E?

872
00:47:05,070 --> 00:47:07,980
How many people said that E
was more likely than A?

873
00:47:07,980 --> 00:47:09,820
Yeah, well this is this
is Kahneman got the

874
00:47:09,820 --> 00:47:10,940
Nobel Prize, you see.

875
00:47:10,940 --> 00:47:12,090
Because he said that, too.

876
00:47:12,090 --> 00:47:13,060
Well, he didn't say that.

877
00:47:13,060 --> 00:47:14,950
He said, that's what
you would say.

878
00:47:14,950 --> 00:47:15,930
Or he found, anyway.

879
00:47:15,930 --> 00:47:19,560
So E is much more likely.

880
00:47:19,560 --> 00:47:24,880
Remember, like, third grade
set theory stuff?

881
00:47:24,880 --> 00:47:29,120
Let's make the little
picture here.

882
00:47:29,120 --> 00:47:30,180
All right.

883
00:47:30,180 --> 00:47:34,960
The set of all truck
drivers, right?

884
00:47:34,960 --> 00:47:36,960
The set of, I can't even
remember what E is.

885
00:47:36,960 --> 00:47:43,370
E's like Mensa Ivy League truck
drivers or something.

886
00:47:43,370 --> 00:47:51,740
It's some small little set,
it's a subset of A.

887
00:47:51,740 --> 00:47:57,340
So, for starters it's a little
unlikely that the chance of

888
00:47:57,340 --> 00:48:00,280
being in here could be
greater than the

889
00:48:00,280 --> 00:48:04,010
chance of being in here.

890
00:48:04,010 --> 00:48:06,120
The best it's going
to be is equal.

891
00:48:06,120 --> 00:48:07,370
In some fashion.

892
00:48:09,760 --> 00:48:10,390
Here's Henry.

893
00:48:10,390 --> 00:48:13,580
I'm meeting some random Henry.

894
00:48:13,580 --> 00:48:16,680
What's the chance
that he's in E?

895
00:48:16,680 --> 00:48:20,190
It's just not --

896
00:48:20,190 --> 00:48:22,900
Again, you're answering a
question that says something

897
00:48:22,900 --> 00:48:26,710
about typical rather
than probable.

898
00:48:26,710 --> 00:48:30,270
And you're ignoring what you
would perfectly well know, of

899
00:48:30,270 --> 00:48:34,050
course, if we drew it out
in set theory terms.

900
00:48:34,050 --> 00:48:40,150
You can't be more likely that
something is in here, if the

901
00:48:40,150 --> 00:48:43,960
larger set includes that.

902
00:48:43,960 --> 00:48:45,880
That doesn't work really well.

903
00:48:45,880 --> 00:48:52,200
So, people are very
bad at base rates.

904
00:48:52,200 --> 00:48:53,950
Is this a bad --

905
00:48:53,950 --> 00:48:59,960
is the purpose of this lecture
to say we're all morons here?

906
00:48:59,960 --> 00:49:00,950
No, not really.

907
00:49:00,950 --> 00:49:05,520
The faults we have, the problems
that we have in the

908
00:49:05,520 --> 00:49:11,340
way that we reason about
problems like this, reflect --

909
00:49:11,340 --> 00:49:16,170
What Kahneman and Tversky talked
about were, if I'm

910
00:49:16,170 --> 00:49:19,420
spelling it right, heuristics.

911
00:49:19,420 --> 00:49:24,400
Sort of, mental shortcuts
that do a certain

912
00:49:24,400 --> 00:49:28,000
amount of work for us.

913
00:49:28,000 --> 00:49:31,340
Our job isn't actually out there
in the world to figure

914
00:49:31,340 --> 00:49:38,050
out set theory problems.

915
00:49:38,050 --> 00:49:42,210
Our problem is to get home
safely at night.

916
00:49:42,210 --> 00:49:45,770
So, you're walking down
the street at night.

917
00:49:45,770 --> 00:49:49,110
And it's a dark kind
of street.

918
00:49:49,110 --> 00:49:53,390
And you see this big
guy with a stick.

919
00:49:53,390 --> 00:49:57,370
And he's walking down the same
side of the street as you.

920
00:49:57,370 --> 00:49:59,570
Do you cross the street?

921
00:49:59,570 --> 00:50:02,910
Well, if you're in a Kahneman
and Tversky kind of experiment

922
00:50:02,910 --> 00:50:05,170
you say, I know what I'm
supposed to do here.

923
00:50:05,170 --> 00:50:08,090
I'm supposed to think hard
about the base rates.

924
00:50:08,090 --> 00:50:11,500
Is it more likely, how many
people are there in the world

925
00:50:11,500 --> 00:50:16,280
who are bad, evil, people with
sticks that are going to,

926
00:50:16,280 --> 00:50:18,020
like, hit me.

927
00:50:18,020 --> 00:50:21,560
Versus, how many people are
there who are, like, guys

928
00:50:21,560 --> 00:50:24,370
coming home from a softball
game with their bat, or

929
00:50:24,370 --> 00:50:26,620
something like that?

930
00:50:26,620 --> 00:50:29,150
And they're nice
people, mostly.

931
00:50:29,150 --> 00:50:31,200
Or at least even if they're bad,
evil people, they're not

932
00:50:31,200 --> 00:50:33,080
really going to hit me.

933
00:50:33,080 --> 00:50:36,840
The answer is, there's a very
small population of people out

934
00:50:36,840 --> 00:50:41,890
there with sticks wanting
to hit you.

935
00:50:41,890 --> 00:50:44,580
But, in the occasion that you
get that wrong, it's a really

936
00:50:44,580 --> 00:50:47,660
bad mistake to make.

937
00:50:47,660 --> 00:50:54,880
And so, a mental shortcut that
says not, what's the base rate

938
00:50:54,880 --> 00:50:59,300
here, but is this typical of
a situation that could be

939
00:50:59,300 --> 00:51:02,120
dangerous, sort of, flipping
it around.

940
00:51:02,120 --> 00:51:04,200
Could this be a bad thing.

941
00:51:04,200 --> 00:51:08,110
If I make a mistake and he's a
really a nice person, so you

942
00:51:08,110 --> 00:51:10,110
might feel a little hurt that I
crossed to the other side of

943
00:51:10,110 --> 00:51:10,660
the street.

944
00:51:10,660 --> 00:51:11,520
What's the big deal.

945
00:51:11,520 --> 00:51:14,160
If he's a guy with a bat and a
nail in it and he's going to

946
00:51:14,160 --> 00:51:15,740
poke me in the head,
and terrible things

947
00:51:15,740 --> 00:51:17,080
are going to happen.

948
00:51:17,080 --> 00:51:19,610
I'm not going to worry about
the base rates here.

949
00:51:19,610 --> 00:51:27,180
And your cognitive hardware
seems to have been set up

950
00:51:27,180 --> 00:51:29,720
under those sort of constraints
more than it was

951
00:51:29,720 --> 00:51:37,150
set up under what might be
optimal public policy solving

952
00:51:37,150 --> 00:51:38,320
constraints.

953
00:51:38,320 --> 00:51:45,440
So, this is why you can then run
a political campaign where

954
00:51:45,440 --> 00:51:48,370
one side or the other spends
a lot of time conjuring up

955
00:51:48,370 --> 00:51:51,190
images of bad guys
with sticks, of

956
00:51:51,190 --> 00:51:53,330
some variety or other.

957
00:51:53,330 --> 00:51:56,470
Because you hear about
the bad guy with

958
00:51:56,470 --> 00:51:58,160
the stick often enough.

959
00:51:58,160 --> 00:52:01,020
And if this guy promises he's
going to save you from the bad

960
00:52:01,020 --> 00:52:03,740
guy with the stick, well, you'd
better go with that.

961
00:52:03,740 --> 00:52:06,290
At least, that's an
appealing thought.

962
00:52:06,290 --> 00:52:07,860
And it works for them.

963
00:52:07,860 --> 00:52:15,350
It works as a political
technique.

964
00:52:15,350 --> 00:52:19,140
All right, this is all -- who
knows about Henry the, some

965
00:52:19,140 --> 00:52:21,270
weird truck driver that
I invented or

966
00:52:21,270 --> 00:52:22,240
something like that?

967
00:52:22,240 --> 00:52:31,010
How about a realm where things
really ought to be logical and

968
00:52:31,010 --> 00:52:32,150
mathematical.

969
00:52:32,150 --> 00:52:34,410
Which would be money.

970
00:52:34,410 --> 00:52:36,130
I mean, if there was ever going
to be something that

971
00:52:36,130 --> 00:52:39,920
ought to just work out in terms
of the math, you would

972
00:52:39,920 --> 00:52:41,960
think it would be money.

973
00:52:41,960 --> 00:52:45,570
And this is, of course, the
reason why there is no Nobel

974
00:52:45,570 --> 00:52:47,630
Prize in psychology.

975
00:52:47,630 --> 00:52:49,730
But there is a Nobel
Prize in economics.

976
00:52:49,730 --> 00:52:53,340
And that's what Kahneman,
in fact, won.

977
00:52:53,340 --> 00:52:57,350
Kaheman and Tversky would have
won it, but Tversky,

978
00:52:57,350 --> 00:52:59,810
unfortunately, died young.

979
00:52:59,810 --> 00:53:03,720
And you don't win the Nobel
Prize posthumously.

980
00:53:03,720 --> 00:53:07,190
So, Kahneman and Tversky also
looked at the way that people

981
00:53:07,190 --> 00:53:08,040
reason about money.

982
00:53:08,040 --> 00:53:11,560
This is something that
goes back long before

983
00:53:11,560 --> 00:53:13,810
Kahneman and Tversky.

984
00:53:13,810 --> 00:53:16,600
And, I can perhaps illustrate
one of the older

985
00:53:16,600 --> 00:53:18,530
bits with an example.

986
00:53:18,530 --> 00:53:24,440
Let's suppose that you're going
to buy a toy car for

987
00:53:24,440 --> 00:53:27,430
your cousin.

988
00:53:27,430 --> 00:53:30,110
You like your cousin, just in
case you were wondering here.

989
00:53:30,110 --> 00:53:32,610
So you're buying this toy
car for your cousin.

990
00:53:32,610 --> 00:53:41,560
And there are two toy cars
out there at the moment.

991
00:53:41,560 --> 00:53:45,800
They are, for present purposes,
functionally

992
00:53:45,800 --> 00:53:49,440
equivalent on all dimensions
of kid

993
00:53:49,440 --> 00:53:51,990
loveliness or something.

994
00:53:51,990 --> 00:54:01,950
There's the midnight blue, I
don't know, Model PT roadster

995
00:54:01,950 --> 00:54:06,470
or something from Toys R Us.

996
00:54:06,470 --> 00:54:17,570
And the midnight green Model
PT roadster from Kids Are

997
00:54:17,570 --> 00:54:18,820
Toys, or something.

998
00:54:27,830 --> 00:54:30,460
So, the difference
is the color.

999
00:54:30,460 --> 00:54:32,790
The green color is cooler.

1000
00:54:32,790 --> 00:54:38,420
You know your cousin actually
likes the green one better.

1001
00:54:38,420 --> 00:54:39,980
It's a bit cooler.

1002
00:54:39,980 --> 00:54:46,130
Anyway, the midnight blue one
costs $12 and the midnight

1003
00:54:46,130 --> 00:54:49,610
green one costs $40.

1004
00:54:49,610 --> 00:54:53,360
How many people are going
to buy one of them.

1005
00:54:53,360 --> 00:54:54,830
You have to pick one, again.

1006
00:54:54,830 --> 00:54:58,050
How many pick the blue one?

1007
00:54:58,050 --> 00:55:00,110
How many people pick
the green one?

1008
00:55:00,110 --> 00:55:00,490
OK.

1009
00:55:00,490 --> 00:55:03,520
You guys have nice cousins,
that's good.

1010
00:55:03,520 --> 00:55:08,690
All right, the blue
one wins big-time.

1011
00:55:08,690 --> 00:55:09,900
All right.

1012
00:55:09,900 --> 00:55:14,280
You now have magically
graduated from MIT.

1013
00:55:14,280 --> 00:55:16,570
Congratulations.

1014
00:55:16,570 --> 00:55:18,160
You got the job.

1015
00:55:18,160 --> 00:55:22,180
You're now going
to buy the car.

1016
00:55:22,180 --> 00:55:23,880
And so, you're going to buy --

1017
00:55:23,880 --> 00:55:27,030
you're going to buy a
real PT roadster.

1018
00:55:27,030 --> 00:55:31,790
and And they come in two
different colors.

1019
00:55:31,790 --> 00:55:33,820
Amazingly, for today's purposes,
they come in

1020
00:55:33,820 --> 00:55:35,260
midnight blue and
midnight green.

1021
00:55:35,260 --> 00:55:41,120
The midnight blue one
costs $30,012.

1022
00:55:41,120 --> 00:55:48,460
The midnight green one
costs $30,040.

1023
00:55:48,460 --> 00:55:50,860
You like the green one better.

1024
00:55:50,860 --> 00:55:54,430
How many people are going
to buy the green one?

1025
00:55:54,430 --> 00:55:58,190
OK, so you graduated from
MIT with this brain.

1026
00:55:58,190 --> 00:56:00,220
It's the same.

1027
00:56:00,220 --> 00:56:02,060
So, obviously it's now
going very much

1028
00:56:02,060 --> 00:56:02,950
in the other direction.

1029
00:56:02,950 --> 00:56:06,920
It's the same $28.

1030
00:56:06,920 --> 00:56:08,540
What's your problem here?

1031
00:56:08,540 --> 00:56:10,790
AUDIENCE: [UNINTELLIGIBLE]

1032
00:56:13,630 --> 00:56:15,220
PROFESSOR: You were
going to --

1033
00:56:15,220 --> 00:56:18,813
AUDIENCE: You're already paying
over $30,000 so the $20

1034
00:56:18,813 --> 00:56:21,380
just seems very trivial.

1035
00:56:21,380 --> 00:56:22,440
PROFESSOR: Yeah.

1036
00:56:22,440 --> 00:56:23,570
It's the same $20.

1037
00:56:23,570 --> 00:56:25,820
This is absolutely
right, of course.

1038
00:56:25,820 --> 00:56:27,823
But yeah.

1039
00:56:27,823 --> 00:56:29,073
AUDIENCE: [INAUDIBLE]

1040
00:56:35,140 --> 00:56:36,660
PROFESSOR: The toy car is
probably going to last you

1041
00:56:36,660 --> 00:56:38,020
least as long as the roadster.

1042
00:56:42,130 --> 00:56:45,340
I keep trying to fine-tune this
to make this, you're not

1043
00:56:45,340 --> 00:56:47,620
going to buy 15 cars
here, and --

1044
00:56:47,620 --> 00:56:50,560
We're assuming a one-time
purchase here, a one-time

1045
00:56:50,560 --> 00:56:52,670
purchase here.

1046
00:56:52,670 --> 00:56:57,630
And they're each going to last
three years, or 100,000 miles,

1047
00:56:57,630 --> 00:56:59,040
or whichever comes first,
or something.

1048
00:56:59,040 --> 00:56:59,320
Yeah.

1049
00:56:59,320 --> 00:57:00,570
AUDIENCE: [UNINTELLIGIBLE]

1050
00:57:04,370 --> 00:57:07,750
PROFESSOR: That much
more money?

1051
00:57:07,750 --> 00:57:08,850
All right, all right.

1052
00:57:08,850 --> 00:57:11,470
Would your answer have changed
radically if I say,

1053
00:57:11,470 --> 00:57:14,010
congratulations you have
just graduated.

1054
00:57:14,010 --> 00:57:17,240
It's also your cousin's
birthday.

1055
00:57:17,240 --> 00:57:19,930
You'd all of a sudden be --

1056
00:57:19,930 --> 00:57:23,020
Boy, there must me sort of
a U-shaped function.

1057
00:57:23,020 --> 00:57:27,990
Because by the time you get to
a parent or something, ask

1058
00:57:27,990 --> 00:57:31,060
your parents, who graduated
a while ago.

1059
00:57:31,060 --> 00:57:35,550
Hey, Mommy, can I have the
midnight green car, it only

1060
00:57:35,550 --> 00:57:39,930
costs three times what
the other one cost.

1061
00:57:39,930 --> 00:57:42,520
Oh, actually I just sort of
gave away the answer.

1062
00:57:42,520 --> 00:57:44,770
It's the three times issue.

1063
00:57:44,770 --> 00:57:46,930
And you were alluding
to this already.

1064
00:57:46,930 --> 00:57:51,310
That money is perceived, even
though it's obviously the same

1065
00:57:51,310 --> 00:57:57,030
$28, that people think about
money in ratio terms and not

1066
00:57:57,030 --> 00:57:57,890
in absolute terms.

1067
00:57:57,890 --> 00:58:02,040
This was actually picked up
by Bernoulli a couple of

1068
00:58:02,040 --> 00:58:03,860
hundred years ago.

1069
00:58:03,860 --> 00:58:04,980
I don't know which Bernoulli.

1070
00:58:04,980 --> 00:58:08,740
It turns out that there were
buckets of Bernoullis.

1071
00:58:08,740 --> 00:58:13,040
Five Bernoullis, and they
were all geniuses.

1072
00:58:13,040 --> 00:58:24,720
So, but one of the Bernoullis
basically asked, what is the

1073
00:58:24,720 --> 00:58:26,610
psychological value of money?

1074
00:58:26,610 --> 00:58:29,810
Actually, in Kahneman and
Tversky language, that's often

1075
00:58:29,810 --> 00:58:34,420
called the utility of money as
a function of actual money.

1076
00:58:34,420 --> 00:58:41,460
And what Bernoulli asserted was
that the utility of money,

1077
00:58:41,460 --> 00:58:44,910
the psychological value of
the money, basically

1078
00:58:44,910 --> 00:58:49,030
went up with log money.

1079
00:58:49,030 --> 00:58:56,140
So it's basically preserving
ratios and not the absolute.

1080
00:58:56,140 --> 00:59:01,030
It's not preserving the
absolute values.

1081
00:59:01,030 --> 00:59:06,520
Which, by the way, when you go
out to buy that first car when

1082
00:59:06,520 --> 00:59:08,590
you have all this money that
you're apparently going to

1083
00:59:08,590 --> 00:59:10,550
have when you graduate.

1084
00:59:10,550 --> 00:59:16,110
The guy selling you the car
knows all about this.

1085
00:59:16,110 --> 00:59:24,650
And he knows that, do you want
the plain vanilla one, or do

1086
00:59:24,650 --> 00:59:28,360
you want the PT roadster
with the really cool

1087
00:59:28,360 --> 00:59:30,940
racing stripe on it.

1088
00:59:30,940 --> 00:59:37,000
And this is part of our cool new
graduate package, of stuff

1089
00:59:37,000 --> 00:59:42,490
that cost us $3.95 and we'll
crank the price up by $500

1090
00:59:42,490 --> 00:59:46,080
here, because on $30,000,
who can tell, right?

1091
00:59:46,080 --> 00:59:48,250
And he knows.

1092
00:59:48,250 --> 00:59:51,530
He knows that it's the same
$28 here and here.

1093
00:59:51,530 --> 00:59:54,560
And his job is to get
as many of those $28

1094
00:59:54,560 --> 00:59:57,790
out of you as possible.

1095
00:59:57,790 --> 01:00:01,770
And will take advantage
of exactly this fact.

1096
01:00:01,770 --> 01:00:05,360
If somebody said, I'll put a
little stripe on your toy car

1097
01:00:05,360 --> 01:00:09,440
for an extra $28, well
it's a smaller car.

1098
01:00:09,440 --> 01:00:13,220
I'll put two stripes on it
because it's a little car.

1099
01:00:13,220 --> 01:00:15,830
You'll say, get away from me.

1100
01:00:15,830 --> 01:00:20,290
But just listen to yourself go
for the options when you go

1101
01:00:20,290 --> 01:00:25,790
off and buy that
car eventually.

1102
01:00:25,790 --> 01:00:27,270
All right.

1103
01:00:27,270 --> 01:00:30,150
let us take a momentary break.

1104
01:00:30,150 --> 01:00:33,570
And then I will say a word more
about the oddities of

1105
01:00:33,570 --> 01:00:35,476
money here.

1106
01:00:35,476 --> 01:01:42,510
[RANDOM NOISE]

1107
01:01:42,510 --> 01:01:43,760
OK.

1108
01:01:47,530 --> 01:02:04,690
Let us contemplate a little more
of the way in which the

1109
01:02:04,690 --> 01:02:06,970
-- now, the reason this
is important.

1110
01:02:06,970 --> 01:02:10,080
The reason that this is the sort
of stuff that eventually

1111
01:02:10,080 --> 01:02:14,360
gets -- that where working
this out is worth the

1112
01:02:14,360 --> 01:02:20,050
attention of the Nobel committee
in economics is that

1113
01:02:20,050 --> 01:02:25,680
economists, for a long time,
had this notion, had a

1114
01:02:25,680 --> 01:02:29,300
psychological theory, in effect,
of what people were

1115
01:02:29,300 --> 01:02:30,140
doing economically.

1116
01:02:30,140 --> 01:02:35,040
And it was the notion of a
rational consumer, a rational

1117
01:02:35,040 --> 01:02:38,280
person who's making decisions
that were in a sense

1118
01:02:38,280 --> 01:02:40,020
propositional.

1119
01:02:40,020 --> 01:02:44,590
That they were doing the math to
the best of their ability,

1120
01:02:44,590 --> 01:02:46,550
and working things out
on that basis.

1121
01:02:46,550 --> 01:02:50,930
And this the Kahneman and
Tversky program of research

1122
01:02:50,930 --> 01:02:54,740
tells you is, the very
interesting constraints on

1123
01:02:54,740 --> 01:02:56,630
that rationality.

1124
01:02:56,630 --> 01:03:00,730
The ways in which it sort of
systematically deviates from

1125
01:03:00,730 --> 01:03:03,360
what your rational computer
might think about the same

1126
01:03:03,360 --> 01:03:04,510
kind of question.

1127
01:03:04,510 --> 01:03:07,040
So, consider the following.

1128
01:03:07,040 --> 01:03:13,310
Let's think about doing
some gambling games.

1129
01:03:13,310 --> 01:03:15,000
Where this is a one-shot game.

1130
01:03:15,000 --> 01:03:16,700
You only get a chance to
play this game once.

1131
01:03:16,700 --> 01:03:17,520
I got a coin.

1132
01:03:17,520 --> 01:03:19,910
It's a fair coin.

1133
01:03:19,910 --> 01:03:28,090
If I flip it and it comes up
heads, I give you a penny.

1134
01:03:28,090 --> 01:03:31,250
If it comes up tails,
you give me a penny.

1135
01:03:31,250 --> 01:03:33,950
How many people are willing
to play that game?

1136
01:03:38,420 --> 01:03:39,200
Few people.

1137
01:03:39,200 --> 01:03:42,090
AUDIENCE: [UNINTELLIGIBLE]

1138
01:03:42,090 --> 01:03:42,540
PROFESSOR: No, no.

1139
01:03:42,540 --> 01:03:43,080
One shot.

1140
01:03:43,080 --> 01:03:45,470
One time.

1141
01:03:45,470 --> 01:03:49,490
The whole game is
this one penny.

1142
01:03:49,490 --> 01:03:54,720
How many people are willing to
play one round of this game?

1143
01:03:54,720 --> 01:03:58,270
AUDIENCE: Give me the
quarter instead.

1144
01:03:58,270 --> 01:04:00,360
PROFESSOR: Oh, she wants
to play for real money.

1145
01:04:00,360 --> 01:04:01,450
OK.

1146
01:04:01,450 --> 01:04:02,480
That's fine.

1147
01:04:02,480 --> 01:04:04,340
Let's do that instead.

1148
01:04:04,340 --> 01:04:06,080
So.

1149
01:04:06,080 --> 01:04:08,130
Most people are willing
to play that, though.

1150
01:04:08,130 --> 01:04:10,990
By this point in the lecture a
variety of people are starting

1151
01:04:10,990 --> 01:04:14,320
to wonder what is that I'm
revealing about myself if I

1152
01:04:14,320 --> 01:04:17,660
say I'll do this.

1153
01:04:17,660 --> 01:04:23,280
So, the way you write that,
you could write that as,

1154
01:04:23,280 --> 01:04:29,870
you've got a 0.5 chance of
winning a penny and a 0.5

1155
01:04:29,870 --> 01:04:32,410
chance of losing a penny.

1156
01:04:32,410 --> 01:04:35,990
And the expected value
of this is 0.5 --

1157
01:04:35,990 --> 01:04:37,950
you can figure that
out, it's 0.

1158
01:04:37,950 --> 01:04:39,230
It's a fair game.

1159
01:04:39,230 --> 01:04:42,810
OK, let's play it again.

1160
01:04:42,810 --> 01:04:44,250
We'll flip, we'll
play for her.

1161
01:04:44,250 --> 01:04:45,540
Flip the coin.

1162
01:04:45,540 --> 01:04:48,860
Heads, I give you $100.

1163
01:04:48,860 --> 01:04:51,130
Tails, you give me $100.

1164
01:04:51,130 --> 01:04:52,380
How many people want to play?

1165
01:04:56,080 --> 01:04:58,340
Your computer doesn't care much
about this, because the

1166
01:04:58,340 --> 01:05:00,760
expected value remains
0, right?

1167
01:05:00,760 --> 01:05:04,024
What's the problem?

1168
01:05:04,024 --> 01:05:05,274
AUDIENCE: [UNINTELLIGIBLE]

1169
01:05:08,490 --> 01:05:10,080
PROFESSOR: People
are risk averse.

1170
01:05:10,080 --> 01:05:13,120
Yeah, that's certainly one.

1171
01:05:13,120 --> 01:05:15,340
That's what you find here.

1172
01:05:15,340 --> 01:05:20,520
But why don't you once
want to play?

1173
01:05:20,520 --> 01:05:21,770
AUDIENCE: [UNINTELLIGIBLE]

1174
01:05:26,630 --> 01:05:29,610
PROFESSOR: No, the expected
value is just a thing defined

1175
01:05:29,610 --> 01:05:31,530
in math land or statistics
land.

1176
01:05:31,530 --> 01:05:35,230
It's zero on this.

1177
01:05:35,230 --> 01:05:38,180
But the two, you could either
win or lose, it's true.

1178
01:05:38,180 --> 01:05:39,950
You're not going to
come out at 0.

1179
01:05:39,950 --> 01:05:43,080
But if we played it for
everybody in the classroom, if

1180
01:05:43,080 --> 01:05:45,830
I had a lot of $100 bills.

1181
01:05:45,830 --> 01:05:47,250
Gee, I wouldn't want
to play that game.

1182
01:05:47,250 --> 01:05:47,570
Yeah.

1183
01:05:47,570 --> 01:05:51,698
AUDIENCE: The expected utility
of playing is less than the

1184
01:05:51,698 --> 01:05:55,310
expected utility
of not playing.

1185
01:05:55,310 --> 01:05:55,920
PROFESSOR: Hmm.

1186
01:05:55,920 --> 01:05:57,400
That sounded cool.

1187
01:06:00,440 --> 01:06:02,730
But I'm not being quick enough
to figure this out.

1188
01:06:02,730 --> 01:06:04,550
What did you --

1189
01:06:04,550 --> 01:06:07,630
AUDIENCE: I'm saying that
winning $100 is less good than

1190
01:06:07,630 --> 01:06:07,700
losing $100.

1191
01:06:07,700 --> 01:06:07,880
PROFESSOR: Thank you.

1192
01:06:07,880 --> 01:06:11,460
That's exactly right.

1193
01:06:11,460 --> 01:06:14,140
It turns out that this
curve is not

1194
01:06:14,140 --> 01:06:16,800
symmetrical around the origin.

1195
01:06:16,800 --> 01:06:21,570
That negative money, the utility
of negative money,

1196
01:06:21,570 --> 01:06:25,340
tends to be somewhat steeper
and more linear

1197
01:06:25,340 --> 01:06:28,010
than positive money.

1198
01:06:28,010 --> 01:06:33,380
With the result that, exactly
what the gentleman said.

1199
01:06:33,380 --> 01:06:41,720
That losing, that the absolute
value of losing $100 on this

1200
01:06:41,720 --> 01:06:47,720
scale is greater than
the absolute

1201
01:06:47,720 --> 01:06:51,150
value of gaining $100.

1202
01:06:51,150 --> 01:06:55,080
I mean, one way to intuit this
is, there's plenty of room for

1203
01:06:55,080 --> 01:06:56,940
you to gain $100.

1204
01:06:56,940 --> 01:06:58,220
You can do that all the time.

1205
01:06:58,220 --> 01:07:02,770
Do you have $100 handy that
you can afford to lose?

1206
01:07:02,770 --> 01:07:04,470
Maybe not.

1207
01:07:04,470 --> 01:07:06,720
It would just hurt a lot more.

1208
01:07:06,720 --> 01:07:08,060
And if we decide --

1209
01:07:08,060 --> 01:07:10,490
And now we can --

1210
01:07:10,490 --> 01:07:12,440
There are some versions of this
-- it's not that you'll

1211
01:07:12,440 --> 01:07:14,260
never play a game for
this kind of stakes.

1212
01:07:14,260 --> 01:07:17,600
I mean, if I say I'll
flip a coin.

1213
01:07:17,600 --> 01:07:22,170
Heads, you give me a penny,
tails, I give you $100, how

1214
01:07:22,170 --> 01:07:24,980
many people want to play?

1215
01:07:24,980 --> 01:07:26,780
And the other ones didn't
figure out what I said,

1216
01:07:26,780 --> 01:07:29,700
presumably.

1217
01:07:29,700 --> 01:07:34,200
So, somewhere between that
extreme and the even game is

1218
01:07:34,200 --> 01:07:37,060
the point at which we
would measure how

1219
01:07:37,060 --> 01:07:38,030
risk averse you are.

1220
01:07:38,030 --> 01:07:40,160
It'd be different for
different people.

1221
01:07:40,160 --> 01:07:46,110
I mean, would you play if heads,
one time play, heads

1222
01:07:46,110 --> 01:07:48,870
you give me $50, tails
I give you $100?

1223
01:07:48,870 --> 01:07:52,260
Now, some people would
be willing to play.

1224
01:07:52,260 --> 01:07:53,600
He's good, he's got $50 handy.

1225
01:07:53,600 --> 01:07:55,250
Or do you have a question?

1226
01:07:55,250 --> 01:07:55,840
No, you were just --

1227
01:07:55,840 --> 01:07:56,000
AUDIENCE: [UNINTELLIGIBLE]

1228
01:07:56,000 --> 01:07:57,700
198.

1229
01:07:57,700 --> 01:07:58,910
PROFESSOR: 198, OK.

1230
01:07:58,910 --> 01:08:01,040
He's got -- talk to him later.

1231
01:08:01,040 --> 01:08:05,000
He's got more money than he
knows what to do with.

1232
01:08:05,000 --> 01:08:09,510
But, in any case, there will
be a balance point

1233
01:08:09,510 --> 01:08:12,090
at which your --

1234
01:08:12,090 --> 01:08:14,950
where the loss and the
gain balance out.

1235
01:08:14,950 --> 01:08:18,050
And then you'd be presumably
willing to pay.

1236
01:08:18,050 --> 01:08:23,000
Strange things happen
at the extremes.

1237
01:08:23,000 --> 01:08:28,430
And that's the basis for a
willingness to play lotteries.

1238
01:08:28,430 --> 01:08:39,290
So, if you have a wager that
says, heads you give me $1,

1239
01:08:39,290 --> 01:08:47,360
tails, I give you $10 million,
but the probability of coming

1240
01:08:47,360 --> 01:08:55,980
up tails equal 10 to the minus,
I don't know, 80 or

1241
01:08:55,980 --> 01:08:57,230
something like that.

1242
01:08:59,620 --> 01:09:02,820
I'm not getting the right
numbers, particularly, for a

1243
01:09:02,820 --> 01:09:03,530
state lottery.

1244
01:09:03,530 --> 01:09:04,520
But it's not --

1245
01:09:04,520 --> 01:09:06,000
it's that kind of thing.

1246
01:09:06,000 --> 01:09:08,580
The reason people run state
lotteries is because they want

1247
01:09:08,580 --> 01:09:09,190
your dollars.

1248
01:09:09,190 --> 01:09:14,310
Not because they're interested
in giving you a pile of money.

1249
01:09:14,310 --> 01:09:17,730
So, even so by the time you get
these very extreme kinds

1250
01:09:17,730 --> 01:09:23,080
of gambles, as long as the cost
is sort of minimal down

1251
01:09:23,080 --> 01:09:26,030
here and the potential payoff is
huge, for instance, you're

1252
01:09:26,030 --> 01:09:28,710
willing to pay radically unfair
games that you would

1253
01:09:28,710 --> 01:09:37,890
never play in the range of, in
the middle kind of range.

1254
01:09:37,890 --> 01:09:47,020
If I tell you, well all right,
we can make this just a

1255
01:09:47,020 --> 01:09:48,450
tenfold difference.

1256
01:09:48,450 --> 01:09:51,210
So if we play a game that --

1257
01:09:51,210 --> 01:09:52,500
you can figure this out.

1258
01:09:52,500 --> 01:09:54,370
At the extremes you're
willing to risk $1

1259
01:09:54,370 --> 01:09:55,620
to make $100 million.

1260
01:09:57,920 --> 01:09:59,800
Even if the odds are
against you.

1261
01:09:59,800 --> 01:10:02,210
In a more moderate game,
you would not be

1262
01:10:02,210 --> 01:10:04,650
willing to do that.

1263
01:10:04,650 --> 01:10:07,940
And it's that sort of reasoning,
or lack thereof,

1264
01:10:07,940 --> 01:10:12,750
that allows lotteries
to make their money.

1265
01:10:12,750 --> 01:10:14,040
Even --

1266
01:10:14,040 --> 01:10:18,310
it's probably also confounded
there by the fact it it's not

1267
01:10:18,310 --> 01:10:21,900
clear that the population as a
whole really understands that

1268
01:10:21,900 --> 01:10:23,580
this is a sucker's game.

1269
01:10:23,580 --> 01:10:26,450
That the state is actually in
the business of taking your

1270
01:10:26,450 --> 01:10:30,150
money from you, not just
redistributing the wealth.

1271
01:10:30,150 --> 01:10:35,220
But even people who perfectly
well understand that their

1272
01:10:35,220 --> 01:10:38,690
chances are less than even of
winning, even if they were to

1273
01:10:38,690 --> 01:10:42,890
buy 10 to the 80th tickets,
continue to buy lottery

1274
01:10:42,890 --> 01:10:43,950
tickets and continue to.

1275
01:10:43,950 --> 01:10:46,250
It's a strange way to raise
money for your state, but

1276
01:10:46,250 --> 01:10:49,190
never mind.

1277
01:10:49,190 --> 01:10:55,330
So, money is like other
aspects of reasoning.

1278
01:10:55,330 --> 01:11:00,660
Is subject to the sorts of
narrative stories that say, it

1279
01:11:00,660 --> 01:11:07,360
hurts more to lose $100 than
it feels good to gain $100.

1280
01:11:07,360 --> 01:11:12,990
The sort of narrative that is
beyond just the simple math of

1281
01:11:12,990 --> 01:11:15,810
the situation.

1282
01:11:15,810 --> 01:11:17,530
Let me --

1283
01:11:17,530 --> 01:11:20,590
OK, I have time to do
one more example.

1284
01:11:20,590 --> 01:11:23,250
Which is an interestingly
illustrative one.

1285
01:11:23,250 --> 01:11:24,950
And involves these cards.

1286
01:11:24,950 --> 01:11:26,800
Which I believe I put
on the handout.

1287
01:11:26,800 --> 01:11:30,060
This is known as the Wason
selection task, because good

1288
01:11:30,060 --> 01:11:32,740
old Wason developed it.

1289
01:11:32,740 --> 01:11:34,820
And, oh look, the rules
are properly

1290
01:11:34,820 --> 01:11:36,340
described on the handout.

1291
01:11:36,340 --> 01:11:39,910
You've got four cards here.

1292
01:11:39,910 --> 01:11:43,940
Each card has a letter on
one side and a number

1293
01:11:43,940 --> 01:11:45,080
on the other side.

1294
01:11:45,080 --> 01:11:48,140
You know that.

1295
01:11:48,140 --> 01:11:53,750
What you want to know is,
and here's the rule.

1296
01:11:53,750 --> 01:12:03,260
The rule is, if the card has a
vowel on one side, then it has

1297
01:12:03,260 --> 01:12:06,740
an odd number on
the other side.

1298
01:12:06,740 --> 01:12:10,480
That's the rule that
we want to check.

1299
01:12:10,480 --> 01:12:15,710
Which cards, what's the minimum
and full set of cards

1300
01:12:15,710 --> 01:12:20,750
that you need to flip to check
whether that rule pertains for

1301
01:12:20,750 --> 01:12:22,360
this set of cards.

1302
01:12:22,360 --> 01:12:23,910
That's the question.

1303
01:12:23,910 --> 01:12:27,040
And so we'll vote on each
card as we go along.

1304
01:12:27,040 --> 01:12:32,400
How many people want
to flip E?

1305
01:12:32,400 --> 01:12:32,750
All right.

1306
01:12:32,750 --> 01:12:36,980
Lots and lots of people want
to flip E. How many people

1307
01:12:36,980 --> 01:12:39,850
want to flip the S?

1308
01:12:39,850 --> 01:12:41,510
Hm.

1309
01:12:41,510 --> 01:12:43,960
Couple of people want to flip
the S. How many people want to

1310
01:12:43,960 --> 01:12:45,210
flip the 7?

1311
01:12:47,510 --> 01:12:49,120
A bunch of people want
to flip the 7.

1312
01:12:49,120 --> 01:12:52,660
How many people want
to flip the 4?

1313
01:12:52,660 --> 01:12:57,770
About the same bunch of people
want to flip the 4.

1314
01:12:57,770 --> 01:12:59,320
And some of the same --

1315
01:12:59,320 --> 01:13:02,090
I probably, well, it's too
complicated to collect all the

1316
01:13:02,090 --> 01:13:07,230
details of how many cards people
think you need to flip.

1317
01:13:07,230 --> 01:13:09,160
The answer is, you need
to flip two and

1318
01:13:09,160 --> 01:13:12,010
only two cards here.

1319
01:13:12,010 --> 01:13:15,250
And before I tell you which ones
it is, I should say that

1320
01:13:15,250 --> 01:13:19,780
like many of the demos in this
lecture, this is the sort of

1321
01:13:19,780 --> 01:13:23,750
thing that drives MIT
students nuts.

1322
01:13:23,750 --> 01:13:26,380
Because they -- it's a cognition
lecture, it's a

1323
01:13:26,380 --> 01:13:27,125
logic lecture.

1324
01:13:27,125 --> 01:13:28,660
It's got, like, mathy
things in it.

1325
01:13:28,660 --> 01:13:29,550
And it's logic.

1326
01:13:29,550 --> 01:13:30,810
And that's why I'm here.

1327
01:13:30,810 --> 01:13:33,530
And I'm getting them all weird
and wrong, and, and, and, and,

1328
01:13:33,530 --> 01:13:35,730
and I'm going to flunk
all my courses.

1329
01:13:35,730 --> 01:13:38,140
And I'm going to die,
or something.

1330
01:13:38,140 --> 01:13:41,970
I mean.

1331
01:13:41,970 --> 01:13:45,210
I should note that one of the
characteristics of the Wason

1332
01:13:45,210 --> 01:13:48,010
selection task is that everybody
gets it wrong.

1333
01:13:48,010 --> 01:13:50,810
I don't remember if it's Wason
in originally who did it.

1334
01:13:50,810 --> 01:13:55,280
But he tried this out on
logicians and they blow it.

1335
01:13:55,280 --> 01:13:59,200
And if you're feeling, like,
depressed after this, you take

1336
01:13:59,200 --> 01:14:07,430
this to the calculus TA
next hour, and see

1337
01:14:07,430 --> 01:14:08,480
if she can do it.

1338
01:14:08,480 --> 01:14:13,880
So, as people correctly
figure, you've got

1339
01:14:13,880 --> 01:14:16,300
to pull this guy.

1340
01:14:16,300 --> 01:14:19,280
Because if it's got an even
number on the other side

1341
01:14:19,280 --> 01:14:20,380
you're dead.

1342
01:14:20,380 --> 01:14:21,690
And you don't have to
do anything because

1343
01:14:21,690 --> 01:14:22,540
we don't know --

1344
01:14:22,540 --> 01:14:27,220
there's no assertion about
what non-vowels are.

1345
01:14:27,220 --> 01:14:29,480
You don't have to flip him.

1346
01:14:29,480 --> 01:14:32,570
Because it doesn't
say, it says --

1347
01:14:32,570 --> 01:14:34,333
here, let's do this in
good logic terms.

1348
01:14:34,333 --> 01:14:37,840
Which for mysterious reasons
always talks about P. The

1349
01:14:37,840 --> 01:14:46,080
assertion is, if P,
then Q. Then odd.

1350
01:14:46,080 --> 01:14:50,010
If P then Q. If you
know Q, who cares.

1351
01:14:50,010 --> 01:14:54,330
It doesn't say, if Q then P. So
this could be an S without

1352
01:14:54,330 --> 01:14:56,910
violating the rule.

1353
01:14:56,910 --> 01:15:01,270
This guy, on the other hand,
is not Q. This guy's not P.

1354
01:15:01,270 --> 01:15:04,290
Everybody figured out we don't
care about not-P. Not-Q is the

1355
01:15:04,290 --> 01:15:08,390
other one that you
do need to check.

1356
01:15:08,390 --> 01:15:12,290
Because if this has an E
on the other side, the

1357
01:15:12,290 --> 01:15:15,320
statement is false.

1358
01:15:15,320 --> 01:15:16,120
The rule is false.

1359
01:15:16,120 --> 01:15:20,040
So you've got to check for an
E on the other side of this.

1360
01:15:20,040 --> 01:15:24,212
So, there are now three people
here who say, I've got it,

1361
01:15:24,212 --> 01:15:24,400
I've got it.

1362
01:15:24,400 --> 01:15:26,540
And good for you.

1363
01:15:26,540 --> 01:15:27,080
That's nice.

1364
01:15:27,080 --> 01:15:31,150
People routinely do very
badly on this.

1365
01:15:31,150 --> 01:15:37,610
And the interesting question is,
is why -- well, actually,

1366
01:15:37,610 --> 01:15:39,720
I'm not sure that's an
interesting question.

1367
01:15:39,720 --> 01:15:40,710
Why you do that.

1368
01:15:40,710 --> 01:15:48,970
The interesting fact is that
people do very well on other

1369
01:15:48,970 --> 01:15:52,260
versions of exactly
this problem.

1370
01:15:52,260 --> 01:15:59,490
So let's set up another version
of this problem.

1371
01:15:59,490 --> 01:16:05,510
We've got somebody with a --

1372
01:16:05,510 --> 01:16:06,760
a beer.

1373
01:16:09,120 --> 01:16:13,520
We've got somebody
with a soda.

1374
01:16:13,520 --> 01:16:17,490
We've got an 18-year-old.

1375
01:16:17,490 --> 01:16:20,120
And we've got a 22-year-old.

1376
01:16:20,120 --> 01:16:28,530
The rule is, you have
to be 22 to drink.

1377
01:16:28,530 --> 01:16:29,420
Well, 21.

1378
01:16:29,420 --> 01:16:33,100
You have to be over
21 to drink.

1379
01:16:33,100 --> 01:16:36,500
Which of these cards do
we need to check?

1380
01:16:36,500 --> 01:16:40,170
So do we need to
check this one?

1381
01:16:40,170 --> 01:16:41,260
Yeah, sure, right.

1382
01:16:41,260 --> 01:16:45,240
If this sucker's an 18-year-old,
he's going down.

1383
01:16:45,240 --> 01:16:47,900
Do we need to check this one?

1384
01:16:47,900 --> 01:16:49,390
We don't care about that.

1385
01:16:49,390 --> 01:16:51,590
Do we need to check this one?

1386
01:16:51,590 --> 01:16:52,110
Yeah.

1387
01:16:52,110 --> 01:16:53,920
We need to check this one?

1388
01:16:53,920 --> 01:16:55,990
No, we don't care about that.

1389
01:16:55,990 --> 01:16:59,610
People are perfect at
this, essentially.

1390
01:16:59,610 --> 01:17:02,010
Sometimes if I phrase it right
you can manage to get yourself

1391
01:17:02,010 --> 01:17:02,650
a little confused.

1392
01:17:02,650 --> 01:17:05,360
So if you got it wrong don't sit
there and say, I'll never

1393
01:17:05,360 --> 01:17:06,610
drink again.

1394
01:17:13,830 --> 01:17:17,460
So, people are perfectly
good at this sort

1395
01:17:17,460 --> 01:17:20,970
of real world example.

1396
01:17:20,970 --> 01:17:22,740
It turns out that they're
not good at

1397
01:17:22,740 --> 01:17:24,290
all real world examples.

1398
01:17:24,290 --> 01:17:29,050
I could, I won't bother trying
to generate another real world

1399
01:17:29,050 --> 01:17:30,720
example here.

1400
01:17:30,720 --> 01:17:35,700
Leda Cosmides has argued that
what people are good about

1401
01:17:35,700 --> 01:17:38,850
reasoning about is cheating.

1402
01:17:38,850 --> 01:17:42,510
Is somebody getting goodies that
they're not supposed to

1403
01:17:42,510 --> 01:17:48,810
Get Is this 18-year-old getting
a drink when I can't

1404
01:17:48,810 --> 01:17:50,950
get a drink yet?

1405
01:17:50,950 --> 01:17:53,930
I'm going to keep good
track of him.

1406
01:17:53,930 --> 01:17:59,250
And the guy drinking this beer
looks kind of young to me.

1407
01:17:59,250 --> 01:18:07,160
Whereas, that four -- yeah,
nobody cares about that.

1408
01:18:07,160 --> 01:18:09,700
And if you reframe the problem,
there are plenty of

1409
01:18:09,700 --> 01:18:13,360
ways to reframe this, to put
things like, oh, there's a

1410
01:18:13,360 --> 01:18:16,950
version about, if you're in
Seattle it must be raining.

1411
01:18:16,950 --> 01:18:18,220
Examples like that.

1412
01:18:18,220 --> 01:18:20,370
People bomb that all
over the place.

1413
01:18:20,370 --> 01:18:23,850
There's nothing at
stake there.

1414
01:18:23,850 --> 01:18:26,880
Cosmides and a variety of the
evolutionary psych people

1415
01:18:26,880 --> 01:18:33,390
argue that your cognitive
capabilities have been shaped

1416
01:18:33,390 --> 01:18:35,890
by the problems that you
need to solve out

1417
01:18:35,890 --> 01:18:36,950
there in the world.

1418
01:18:36,950 --> 01:18:39,640
Is somebody getting goodies
they shouldn't be getting?

1419
01:18:39,640 --> 01:18:41,110
Is the guy with the
stick going to

1420
01:18:41,110 --> 01:18:42,430
hit me with the stick?

1421
01:18:42,430 --> 01:18:43,350
That kind of thing.

1422
01:18:43,350 --> 01:18:45,770
It would be lovely
if that was true.

1423
01:18:45,770 --> 01:18:46,610
It might be true.

1424
01:18:46,610 --> 01:18:50,820
The problem is that the research
subsequent to this

1425
01:18:50,820 --> 01:18:52,900
has made the issue
more complicated.

1426
01:18:52,900 --> 01:18:57,690
It is not -- there are cheating
examples that people

1427
01:18:57,690 --> 01:19:01,350
have cooked up that bomb.

1428
01:19:01,350 --> 01:19:03,730
And there are non-cheating
examples that people

1429
01:19:03,730 --> 01:19:05,620
successfully manage to solve.

1430
01:19:05,620 --> 01:19:08,610
But what does seem to be pretty
clear is that we are

1431
01:19:08,610 --> 01:19:12,110
not built to solve abstract
problems.

1432
01:19:12,110 --> 01:19:16,250
That's why you have to show
up at MIT to learn

1433
01:19:16,250 --> 01:19:18,290
that kind of stuff.

1434
01:19:18,290 --> 01:19:21,660
Nobody goes to MIT to
take courses in

1435
01:19:21,660 --> 01:19:25,580
carding guys at bars.

1436
01:19:25,580 --> 01:19:26,890
You come with that.

1437
01:19:26,890 --> 01:19:29,610
You can figure that
out for yourself.

1438
01:19:29,610 --> 01:19:32,710
It's when you have
-- unfortunately.

1439
01:19:32,710 --> 01:19:36,510
The problem of running the
country turns out to be

1440
01:19:36,510 --> 01:19:40,610
probably a lot more like
this than like this.

1441
01:19:40,610 --> 01:19:43,620
But the decision is
made like this.

1442
01:19:43,620 --> 01:19:45,410
So, everybody go watch
the debate and

1443
01:19:45,410 --> 01:19:47,230
tell me what you find.