1
00:00:00,040 --> 00:00:02,460
The following content is
provided under a Creative

2
00:00:02,460 --> 00:00:03,870
Commons license.

3
00:00:03,870 --> 00:00:06,910
Your support will help MIT
OpenCourseWare continue to

4
00:00:06,910 --> 00:00:10,560
offer high quality educational
resources for free.

5
00:00:10,560 --> 00:00:13,460
To make a donation or view
additional materials from

6
00:00:13,460 --> 00:00:16,071
hundreds of MIT courses, visit
MIT OpenCourseWare at

7
00:00:16,071 --> 00:00:17,321
ocw.mit.edu.

8
00:00:27,930 --> 00:00:30,930
PROFESSOR: So let's talk about
memory today and the fragile

9
00:00:30,930 --> 00:00:31,840
power of memory.

10
00:00:31,840 --> 00:00:33,790
If memory didn't work, we
wouldn't remember anything.

11
00:00:33,790 --> 00:00:34,650
We wouldn't learn anything.

12
00:00:34,650 --> 00:00:37,010
So we know it has to
work reasonably.

13
00:00:37,010 --> 00:00:40,420
But it has a number of fragile
aspects we'll talk about

14
00:00:40,420 --> 00:00:43,080
where, of course, we
forget things.

15
00:00:43,080 --> 00:00:45,220
But on top of that, it turns out
there's ways in which our

16
00:00:45,220 --> 00:00:49,890
memories get less accurate and
distorted, are not actual

17
00:00:49,890 --> 00:00:52,730
records of our experience but
rather interpretations or our

18
00:00:52,730 --> 00:00:54,150
memory of our memory.

19
00:00:54,150 --> 00:00:56,870
And we'll talk about experiments
that reveal that.

20
00:00:56,870 --> 00:00:59,120
So to a large extent, we'll
talk about what do we know

21
00:00:59,120 --> 00:01:02,300
from scientific research about
why we remember some of our

22
00:01:02,300 --> 00:01:05,670
lives and why we forget so
many other things that we

23
00:01:05,670 --> 00:01:07,590
experience.

24
00:01:07,590 --> 00:01:11,070
And a version of the way people
sometimes think about

25
00:01:11,070 --> 00:01:13,950
it is in what sense is memory
like a camera, that every

26
00:01:13,950 --> 00:01:17,050
experience you have is stored
in your mind somewhere?

27
00:01:17,050 --> 00:01:20,040
We all have the experience,
perhaps, that something will

28
00:01:20,040 --> 00:01:21,990
remind you of something you
haven't thought about for

29
00:01:21,990 --> 00:01:24,145
days, weeks, or years.

30
00:01:24,145 --> 00:01:26,210
And you go, oh, well they must
be all stored there.

31
00:01:26,210 --> 00:01:28,080
I just need the right
cue or the right

32
00:01:28,080 --> 00:01:30,410
path to get that memory.

33
00:01:30,410 --> 00:01:31,920
So that's one version
of memory,

34
00:01:31,920 --> 00:01:33,830
everything is in us recorded.

35
00:01:33,830 --> 00:01:34,830
Sometimes we can get to it.

36
00:01:34,830 --> 00:01:36,520
Sometimes we can't.

37
00:01:36,520 --> 00:01:38,730
Another version is the way to
think about your memory.

38
00:01:38,730 --> 00:01:40,160
It's kind of like
a punch bowl.

39
00:01:40,160 --> 00:01:43,100
You put in some milk.

40
00:01:43,100 --> 00:01:44,780
And then you put in
some orange juice

41
00:01:44,780 --> 00:01:46,100
and swirl it around.

42
00:01:46,100 --> 00:01:48,850
And then you put in some
Coca-Cola and swirl it around.

43
00:01:48,850 --> 00:01:51,970
And it's not a desirable drink
that you have in your bowl.

44
00:01:51,970 --> 00:01:56,070
But you can no more get the
specific milk, or the specific

45
00:01:56,070 --> 00:01:58,480
orange juice, or the
specific Coke.

46
00:01:58,480 --> 00:01:59,930
Because they're all
mixed together.

47
00:01:59,930 --> 00:02:01,520
And is memory more like that?

48
00:02:01,520 --> 00:02:03,630
As life goes on, things
get mixed in.

49
00:02:03,630 --> 00:02:06,890
And the raw memory is never
recoverable from the mixture

50
00:02:06,890 --> 00:02:10,580
of subsequent experiences,
thoughts, and feelings.

51
00:02:10,580 --> 00:02:12,920
So we'll bring the same
perspective a little bit.

52
00:02:12,920 --> 00:02:15,930
Bottom-up memory might be really
the video camera if we

53
00:02:15,930 --> 00:02:19,190
had it, an actual recording of
what we hear, what we see,

54
00:02:19,190 --> 00:02:21,470
what we experience out
there in the world.

55
00:02:21,470 --> 00:02:23,950
Top-down processes are the
mental things we have in our

56
00:02:23,950 --> 00:02:29,110
minds that let us interpret
what's out there, concepts,

57
00:02:29,110 --> 00:02:30,750
expectations.

58
00:02:30,750 --> 00:02:34,510
Also, it turns out, you'll see
that subsequent experiences go

59
00:02:34,510 --> 00:02:37,790
back and change memories
from prior times.

60
00:02:37,790 --> 00:02:40,430
And which of these things, the
bottom-up original experience,

61
00:02:40,430 --> 00:02:44,120
or the top-down interpretation
grows over time to constitute

62
00:02:44,120 --> 00:02:45,780
your memory?

63
00:02:45,780 --> 00:02:49,770
So we can have a very simple
model of memory,

64
00:02:49,770 --> 00:02:51,620
a flash for a moment.

65
00:02:51,620 --> 00:02:53,300
I'll give you an example of
how people test this, a

66
00:02:53,300 --> 00:02:56,670
sensory memory, very
tiny moments only.

67
00:02:56,670 --> 00:02:59,140
Short-term memory that you can
rehearse in some cases.

68
00:02:59,140 --> 00:03:01,260
Somebody tells you a number, And
you can repeat it over and

69
00:03:01,260 --> 00:03:01,680
over again.

70
00:03:01,680 --> 00:03:03,160
You have that capacity.

71
00:03:03,160 --> 00:03:04,840
And then long-term memory,
things that

72
00:03:04,840 --> 00:03:06,560
last for a long time.

73
00:03:06,560 --> 00:03:08,300
You can define these things
many different ways.

74
00:03:08,300 --> 00:03:11,290
When we talk about the brain
organization of human memory,

75
00:03:11,290 --> 00:03:13,560
we'll say that short-term
memory is

76
00:03:13,560 --> 00:03:15,810
probably, lasts seconds.

77
00:03:15,810 --> 00:03:18,180
And everything after that
is long-term memory.

78
00:03:18,180 --> 00:03:20,170
Now, of course, if you just
experienced something, you'd

79
00:03:20,170 --> 00:03:21,450
remember it better.

80
00:03:21,450 --> 00:03:24,290
If it's a week, a month, a year,
five years from now,

81
00:03:24,290 --> 00:03:25,680
most likely less better.

82
00:03:25,680 --> 00:03:27,080
So time passes.

83
00:03:27,080 --> 00:03:29,560
But we think almost all of
that is in the context of

84
00:03:29,560 --> 00:03:32,800
long-term memory capacities
of your mind.

85
00:03:32,800 --> 00:03:35,200
So here's a kind
of experiment.

86
00:03:35,200 --> 00:03:36,755
We'll do a couple of exercises
at your seat.

87
00:03:36,755 --> 00:03:38,240
But this I'm just going
to give you a feeling.

88
00:03:38,240 --> 00:03:40,180
Because this would require
millisecond

89
00:03:40,180 --> 00:03:41,280
timing to do correctly.

90
00:03:41,280 --> 00:03:43,915
But the way they approached this
sensory memory, the flash

91
00:03:43,915 --> 00:03:47,340
of memory, just capturing for a
moment what's out there, is

92
00:03:47,340 --> 00:03:48,040
something like this.

93
00:03:48,040 --> 00:03:50,160
So imagine your job is
to report back the

94
00:03:50,160 --> 00:03:53,290
letters that you see.

95
00:03:53,290 --> 00:03:56,056
OK.

96
00:03:56,056 --> 00:03:59,960
Yeah, you're getting
a number of them.

97
00:03:59,960 --> 00:04:03,420
So you just did what's called
a whole report method.

98
00:04:03,420 --> 00:04:06,830
Tell me everything you saw.

99
00:04:06,830 --> 00:04:11,240
They did another approach,
which is present letters.

100
00:04:11,240 --> 00:04:15,900
And then just as it disappeared
did we get a tone,

101
00:04:15,900 --> 00:04:19,279
a high, medium, or low tone.

102
00:04:19,279 --> 00:04:22,270
And that would tell them which
row to report, the top row,

103
00:04:22,270 --> 00:04:24,680
the middle row, or the bottom
row depending on the tone they

104
00:04:24,680 --> 00:04:27,610
heard just after the flash,
immediately after the flash.

105
00:04:27,610 --> 00:04:27,990
Does make sense?

106
00:04:27,990 --> 00:04:28,380
OK.

107
00:04:28,380 --> 00:04:28,960
It's gone.

108
00:04:28,960 --> 00:04:31,470
But the tone is telling you
which one to report.

109
00:04:31,470 --> 00:04:34,180
And the striking finding
in this was this.

110
00:04:34,180 --> 00:04:36,940
If they showed you 12 letters
like you just saw, and said,

111
00:04:36,940 --> 00:04:39,840
tell me everything you see,
people could give you about

112
00:04:39,840 --> 00:04:40,680
four letters back.

113
00:04:40,680 --> 00:04:43,730
Maybe you had that experience,
about four letters accurately.

114
00:04:43,730 --> 00:04:45,980
If they gave you the letters,
they flashed them, then you've

115
00:04:45,980 --> 00:04:47,400
got the tone.

116
00:04:47,400 --> 00:04:50,500
For one row only, you've got
about three letters correct.

117
00:04:50,500 --> 00:04:51,270
Now this is interesting.

118
00:04:51,270 --> 00:04:52,790
The letters are gone.

119
00:04:52,790 --> 00:04:54,830
But it's kind of like as
if the photograph of

120
00:04:54,830 --> 00:04:55,745
your memory is fading.

121
00:04:55,745 --> 00:04:59,050
But if you put your attention
somewhere, you can still pull

122
00:04:59,050 --> 00:05:01,690
it back, but just a
little bit of it.

123
00:05:01,690 --> 00:05:04,030
Because here you get
four out of 12.

124
00:05:04,030 --> 00:05:04,970
Here you get three out of 12.

125
00:05:04,970 --> 00:05:08,780
Almost the same because your
attention can zoom in on that

126
00:05:08,780 --> 00:05:12,820
fading mental image of
what you just saw.

127
00:05:12,820 --> 00:05:14,040
So much is sensed.

128
00:05:14,040 --> 00:05:17,640
But attention only selects a
little bit to be remembered.

129
00:05:17,640 --> 00:05:22,290
So that's this very first
moment of memory.

130
00:05:22,290 --> 00:05:24,800
One of the more impressive lines
of research is the next

131
00:05:24,800 --> 00:05:27,910
moment, short-term memory,
keeping in mind something in

132
00:05:27,910 --> 00:05:29,280
your mind briefly.

133
00:05:29,280 --> 00:05:31,980
And one of the most striking
aspects of human short-term

134
00:05:31,980 --> 00:05:35,040
memory is limited in how
much it can hold.

135
00:05:35,040 --> 00:05:38,210
And across a tremendous range of
materials, words, letters,

136
00:05:38,210 --> 00:05:41,900
shapes, tones, and so on, you
can hold approximately 7 plus

137
00:05:41,900 --> 00:05:45,200
or minus 2 chunks of
information, limited

138
00:05:45,200 --> 00:05:47,410
short-term memory.

139
00:05:47,410 --> 00:05:50,430
So now I need somebody who's
willing, at their

140
00:05:50,430 --> 00:05:54,450
chair, to do something.

141
00:05:54,450 --> 00:05:56,250
I'm going to read to you
numbers, and ask you

142
00:05:56,250 --> 00:05:57,110
to repeat them back.

143
00:05:57,110 --> 00:05:59,370
But your eyes will need
to be closed.

144
00:05:59,370 --> 00:06:00,835
And it'll be fairly
challenging.

145
00:06:08,120 --> 00:06:10,080
All right, thank
you very much.

146
00:06:10,080 --> 00:06:10,840
OK.

147
00:06:10,840 --> 00:06:11,920
Close your eyes.

148
00:06:11,920 --> 00:06:12,410
And here we go.

149
00:06:12,410 --> 00:06:12,920
Ready?

150
00:06:12,920 --> 00:06:14,550
Keep your eyes closed
until we're done.

151
00:06:14,550 --> 00:06:15,500
Repeat after me.

152
00:06:15,500 --> 00:06:16,721
6, 1.

153
00:06:16,721 --> 00:06:18,350
AUDIENCE: We're repeating
as you say it or after?

154
00:06:18,350 --> 00:06:18,880
PROFESSOR: When I'm done.

155
00:06:18,880 --> 00:06:19,920
I'm sorry, yeah.

156
00:06:19,920 --> 00:06:20,410
Ready?

157
00:06:20,410 --> 00:06:22,300
6, 1, 9.

158
00:06:22,300 --> 00:06:22,800
4.

159
00:06:22,800 --> 00:06:23,590
Repeat.

160
00:06:23,590 --> 00:06:24,430
AUDIENCE: 6, 1, 9, 4.

161
00:06:24,430 --> 00:06:25,050
PROFESSOR: Right.

162
00:06:25,050 --> 00:06:26,210
Ready?

163
00:06:26,210 --> 00:06:29,560
3, 7, 8, 5, 2, repeat.

164
00:06:29,560 --> 00:06:31,495
AUDIENCE: 3, 7, 8, 5, 2.

165
00:06:31,495 --> 00:06:32,270
PROFESSOR: Great.

166
00:06:32,270 --> 00:06:34,100
Ready?

167
00:06:34,100 --> 00:06:38,580
9, 6, 5, 2, 8, 3, repeat.

168
00:06:38,580 --> 00:06:40,030
AUDIENCE: 9, 6, 5, 2, 8, 3.

169
00:06:40,030 --> 00:06:40,942
PROFESSOR: Great.

170
00:06:40,942 --> 00:06:42,750
Ready?

171
00:06:42,750 --> 00:06:46,720
4, 2, 6, 9, 8, 5, 1.

172
00:06:46,720 --> 00:06:47,300
repeat.

173
00:06:47,300 --> 00:06:49,730
AUDIENCE: 4, 2, 6, 9, 8, 5, 1.

174
00:06:49,730 --> 00:06:50,450
PROFESSOR: Very good.

175
00:06:50,450 --> 00:06:52,310
Ready?

176
00:06:52,310 --> 00:06:55,870
8, 1, 6, 3, 7, 2, 4, 9.

177
00:06:55,870 --> 00:06:56,500
Repeat.

178
00:06:56,500 --> 00:06:59,860
AUDIENCE: 8, 1, 6,
7, 4, 2, 1, 9.

179
00:06:59,860 --> 00:07:01,000
PROFESSOR: OK, perfect.

180
00:07:01,000 --> 00:07:01,460
OK your eyes.

181
00:07:01,460 --> 00:07:02,200
Thank you very much.

182
00:07:02,200 --> 00:07:04,620
You did terrific.

183
00:07:04,620 --> 00:07:07,410
You got four right, five right,
six right, seven right,

184
00:07:07,410 --> 00:07:08,070
eight right, you got
them mostly right.

185
00:07:08,070 --> 00:07:09,550
You just reversed one or two.

186
00:07:09,550 --> 00:07:11,820
Seven plus or minus two, it
doesn't matter how smart you

187
00:07:11,820 --> 00:07:14,360
are, how advanced you are.

188
00:07:14,360 --> 00:07:16,090
There used to be some confusions
about this.

189
00:07:16,090 --> 00:07:18,150
Because they would say kids in
different countries have

190
00:07:18,150 --> 00:07:19,580
different limits.

191
00:07:19,580 --> 00:07:22,960
And they'd say our country is
smarter than your country, or

192
00:07:22,960 --> 00:07:24,320
our educational system
is better than

193
00:07:24,320 --> 00:07:25,400
your educational system.

194
00:07:25,400 --> 00:07:29,860
It turned out that all of that
got equated around the world

195
00:07:29,860 --> 00:07:32,560
if you counted up the
number of syllables

196
00:07:32,560 --> 00:07:33,470
that are in the digits.

197
00:07:33,470 --> 00:07:36,010
The real unit of memory
here are syllables.

198
00:07:36,010 --> 00:07:39,590
So in English, most of the
numbers are one syllable.

199
00:07:39,590 --> 00:07:42,340
1, 2, 3, 4, 5, 6 are
all one syllable.

200
00:07:42,340 --> 00:07:43,560
7 gets two.

201
00:07:43,560 --> 00:07:44,300
8 is one syllable.

202
00:07:44,300 --> 00:07:45,480
9 is one syllable.

203
00:07:45,480 --> 00:07:48,810
In some other languages, the
words that go with the digits

204
00:07:48,810 --> 00:07:51,120
are two syllables.

205
00:07:51,120 --> 00:07:53,500
So countries that have two
syllables, they have lower

206
00:07:53,500 --> 00:07:56,000
spans not because they were
dumber or had a worse

207
00:07:56,000 --> 00:07:57,150
education system.

208
00:07:57,150 --> 00:07:59,940
It's because the real unit
is number of syllables.

209
00:07:59,940 --> 00:08:00,910
Because you're hearing it.

210
00:08:00,910 --> 00:08:03,210
It's language.

211
00:08:03,210 --> 00:08:06,140
So around the world, it's the
fundamental capacity of the

212
00:08:06,140 --> 00:08:07,960
human mind is that
you're limited to

213
00:08:07,960 --> 00:08:09,340
these kinds of stuff.

214
00:08:09,340 --> 00:08:11,200
So how do we get around that?

215
00:08:11,200 --> 00:08:16,010
Why are we not always barely
able to do anything with very

216
00:08:16,010 --> 00:08:17,200
small amounts of information?

217
00:08:17,200 --> 00:08:20,710
So it's just a constraint
of our minds.

218
00:08:20,710 --> 00:08:23,400
It's because background prior
knowledge, things you know

219
00:08:23,400 --> 00:08:25,690
before you have to learn
something, has a

220
00:08:25,690 --> 00:08:27,120
very powerful effect.

221
00:08:27,120 --> 00:08:28,830
It allows you for the
information to

222
00:08:28,830 --> 00:08:30,300
be retained in memory.

223
00:08:30,300 --> 00:08:32,600
But it has a double-edged sword,
which is kind of a

224
00:08:32,600 --> 00:08:33,870
theme a lot today.

225
00:08:33,870 --> 00:08:36,190
As much as it helps you remember
things, it can fool

226
00:08:36,190 --> 00:08:38,120
you into misremembering
things.

227
00:08:38,120 --> 00:08:40,330
And let me give you a
concrete example.

228
00:08:40,330 --> 00:08:45,000
So you can put chess displays
in front of people who are

229
00:08:45,000 --> 00:08:47,340
either beginning chess players,
have played for a

230
00:08:47,340 --> 00:08:50,000
while, or are masters.

231
00:08:50,000 --> 00:08:51,290
And they're not going to play.

232
00:08:51,290 --> 00:08:52,650
They're going to look at it.

233
00:08:52,650 --> 00:08:55,660
The pieces will be swept
to the side.

234
00:08:55,660 --> 00:08:58,010
And then they have to put all
the pieces back out of memory

235
00:08:58,010 --> 00:09:00,050
into the same location where
they just saw it.

236
00:09:00,050 --> 00:09:01,660
They know that's going
to happen.

237
00:09:01,660 --> 00:09:02,700
So it's a memory test.

238
00:09:02,700 --> 00:09:03,350
Here's the pieces.

239
00:09:03,350 --> 00:09:04,160
Take a look.

240
00:09:04,160 --> 00:09:06,310
The examiner sweeps the
pieces to the site.

241
00:09:06,310 --> 00:09:08,310
And you put them back
as well as you can.

242
00:09:08,310 --> 00:09:09,000
You see it.

243
00:09:09,000 --> 00:09:09,870
It gets swept to the side.

244
00:09:09,870 --> 00:09:12,920
And you reproduce it from memory
what you just saw.

245
00:09:12,920 --> 00:09:19,040
And here's the performance of
master players and beginner

246
00:09:19,040 --> 00:09:21,320
players in this case, so people
who have played a lot

247
00:09:21,320 --> 00:09:23,630
of chess and people who have
played very little chess.

248
00:09:23,630 --> 00:09:27,810
It's better to be high, the
number of correct pieces.

249
00:09:27,810 --> 00:09:30,760
And there's two kinds
of displays.

250
00:09:30,760 --> 00:09:32,060
In one kind of displays
is what we

251
00:09:32,060 --> 00:09:33,870
call the normal displays.

252
00:09:33,870 --> 00:09:35,340
Those are ones that
could really have

253
00:09:35,340 --> 00:09:36,160
happened in the game.

254
00:09:36,160 --> 00:09:40,190
They were kind of reasonable
configurations of pieces.

255
00:09:40,190 --> 00:09:43,680
The random displays are pieces
all over the place in a

256
00:09:43,680 --> 00:09:46,070
configuration that you're not
likely to see in a real game

257
00:09:46,070 --> 00:09:48,380
of chess, equal number
of pieces.

258
00:09:48,380 --> 00:09:49,500
And look what happens.

259
00:09:49,500 --> 00:09:53,890
The best performance are master
players who see a real

260
00:09:53,890 --> 00:09:56,320
normal chess configuration.

261
00:09:56,320 --> 00:09:58,120
Here are the players
who are beginners.

262
00:09:58,120 --> 00:10:01,830
And look at the worst
performances, the chess

263
00:10:01,830 --> 00:10:04,690
masters who see a
random set up.

264
00:10:04,690 --> 00:10:07,088
So what do you guess
is happening?

265
00:10:07,088 --> 00:10:08,040
Yeah?

266
00:10:08,040 --> 00:10:09,990
AUDIENCE: People who
play chess a lot

267
00:10:09,990 --> 00:10:11,802
memorize games and positions.

268
00:10:11,802 --> 00:10:14,366
So they're used to seeing
those patterns.

269
00:10:14,366 --> 00:10:17,238
And when they see a random
configuration, they can't

270
00:10:17,238 --> 00:10:18,230
recreate it as well.

271
00:10:18,230 --> 00:10:18,440
PROFESSOR: Right.

272
00:10:18,440 --> 00:10:20,590
So the people who have played
chess a lot have a lot of

273
00:10:20,590 --> 00:10:22,800
configurations in their mind,
background knowledge.

274
00:10:22,800 --> 00:10:25,790
They use that background
knowledge of what a chess

275
00:10:25,790 --> 00:10:27,030
array means.

276
00:10:27,030 --> 00:10:29,440
Somebody's in trouble,
somebody's winning, something

277
00:10:29,440 --> 00:10:31,220
interesting is happening
in this corner.

278
00:10:31,220 --> 00:10:34,080
That knowledge helps them escape
the boundaries of seven

279
00:10:34,080 --> 00:10:36,280
plus or minus two.

280
00:10:36,280 --> 00:10:39,910
And they do a lot better than
the beginners for whom it

281
00:10:39,910 --> 00:10:40,890
doesn't mean much.

282
00:10:40,890 --> 00:10:44,220
It's just a bunch of stuff out
there hard to remember.

283
00:10:44,220 --> 00:10:47,220
But what happens here when
the pieces are random?

284
00:10:47,220 --> 00:10:51,670
The chess masters are making
mistakes because they're

285
00:10:51,670 --> 00:10:54,930
putting the pieces in the places
they ought to be rather

286
00:10:54,930 --> 00:10:57,010
than the places they perversely

287
00:10:57,010 --> 00:10:58,410
were randomly located.

288
00:10:58,410 --> 00:11:02,290
If you're a chess player, you
don't like that random board.

289
00:11:02,290 --> 00:11:04,950
It doesn't feel right.

290
00:11:04,950 --> 00:11:06,500
But they're trying
to be accurate.

291
00:11:06,500 --> 00:11:09,410
But now they're misplacing their
memory for what they

292
00:11:09,410 --> 00:11:12,730
actually saw with their
mental model of what

293
00:11:12,730 --> 00:11:14,820
ought to be out there.

294
00:11:14,820 --> 00:11:17,890
Now the players who are not so
good, you don't see much of a

295
00:11:17,890 --> 00:11:19,430
difference between
these lines.

296
00:11:19,430 --> 00:11:21,170
They don't understand
when it's random.

297
00:11:21,170 --> 00:11:22,320
They don't understand
when it's organized.

298
00:11:22,320 --> 00:11:23,860
It's just a bunch of pieces.

299
00:11:23,860 --> 00:11:27,310
So the double-edge of background
knowledge.

300
00:11:27,310 --> 00:11:30,600
Background knowledge lets you
escape the bounds of limited

301
00:11:30,600 --> 00:11:35,060
short-term memory but at a cost
of sometimes substituting

302
00:11:35,060 --> 00:11:36,880
in for the actual experience
you have.

303
00:11:40,130 --> 00:11:41,730
So we'll try this.

304
00:11:41,730 --> 00:11:44,480
And people use the word
"chunking" as the way in which

305
00:11:44,480 --> 00:11:47,480
you use your background
knowledge to get big pieces of

306
00:11:47,480 --> 00:11:49,960
information stored that are
better than you think for

307
00:11:49,960 --> 00:11:51,100
short-term memory limits.

308
00:11:51,100 --> 00:11:52,820
So here's a quick one.

309
00:11:52,820 --> 00:11:55,085
I need a volunteer actually.

310
00:11:55,085 --> 00:11:56,335
It's not too bad.

311
00:11:58,814 --> 00:12:00,620
Wow, tough volunteer day.

312
00:12:00,620 --> 00:12:01,650
You're mad at me
about the exam.

313
00:12:01,650 --> 00:12:01,840
OK.

314
00:12:01,840 --> 00:12:03,350
Thank you.

315
00:12:03,350 --> 00:12:06,430
OK, ready?

316
00:12:06,430 --> 00:12:07,568
What were the letters?

317
00:12:07,568 --> 00:12:09,320
AUDIENCE: [INAUDIBLE].

318
00:12:09,320 --> 00:12:09,710
PROFESSOR: OK.

319
00:12:09,710 --> 00:12:11,486
How about these ones?

320
00:12:11,486 --> 00:12:13,500
[LAUGHTER]

321
00:12:13,500 --> 00:12:15,580
PROFESSOR: The same letters,
it's just that

322
00:12:15,580 --> 00:12:16,540
now you see the chunks.

323
00:12:16,540 --> 00:12:19,130
And you can map them onto
long-term memory more easily.

324
00:12:19,130 --> 00:12:20,330
Plus I showed them longer.

325
00:12:20,330 --> 00:12:24,840
Anybody else want to try
something like this?

326
00:12:24,840 --> 00:12:26,070
Thank you.

327
00:12:26,070 --> 00:12:27,320
Here we go.

328
00:12:30,020 --> 00:12:30,700
What letters do you see?

329
00:12:30,700 --> 00:12:32,912
Ready?

330
00:12:32,912 --> 00:12:35,750
AUDIENCE: [INAUDIBLE].

331
00:12:35,750 --> 00:12:36,230
PROFESSOR: OK.

332
00:12:36,230 --> 00:12:38,690
Now I'm going to let your
background knowledge triumph.

333
00:12:38,690 --> 00:12:39,940
Ready?

334
00:12:44,720 --> 00:12:45,970
It's the same letters.

335
00:12:48,520 --> 00:12:51,590
Same letters, all it is, is your
background knowledge is

336
00:12:51,590 --> 00:12:54,230
useless, your knowledge of
words, your knowledge of the

337
00:12:54,230 --> 00:12:55,575
alphabet, useless.

338
00:12:55,575 --> 00:12:58,390
It's the same information in a
sense, same number of letters,

339
00:12:58,390 --> 00:12:59,180
same identity of letters.

340
00:12:59,180 --> 00:13:00,765
But now your background
knowledge is powerful.

341
00:13:03,940 --> 00:13:07,680
I need somebody else,
one more volunteer.

342
00:13:07,680 --> 00:13:08,960
Thank you.

343
00:13:08,960 --> 00:13:10,020
I'm going to show
you a sentence.

344
00:13:10,020 --> 00:13:11,050
This is meant to be hard,
the first one.

345
00:13:11,050 --> 00:13:11,500
Don't worry about it.

346
00:13:11,500 --> 00:13:12,160
OK?

347
00:13:12,160 --> 00:13:14,950
Here we go.

348
00:13:14,950 --> 00:13:16,421
Can you tell me what
they were?

349
00:13:16,421 --> 00:13:18,730
AUDIENCE: Leaf, paper,
steer, car, beach.

350
00:13:18,730 --> 00:13:20,615
PROFESSOR: That's pretty good.

351
00:13:20,615 --> 00:13:23,890
Let me show you another
equal number of words.

352
00:13:23,890 --> 00:13:24,130
Ready?

353
00:13:24,130 --> 00:13:29,050
Go OK, what is it?

354
00:13:29,050 --> 00:13:32,410
AUDIENCE: While I was walking
around-- while I was walking--

355
00:13:32,410 --> 00:13:33,270
PROFESSOR: You would
do better.

356
00:13:33,270 --> 00:13:36,620
OK It's hard to pull
these things off.

357
00:13:36,620 --> 00:13:39,440
So 13 words, people would
do pretty miserably.

358
00:13:39,440 --> 00:13:41,860
Although you did very well,
a lot better at this.

359
00:13:41,860 --> 00:13:43,810
Because, again, you can use your
background knowledge of

360
00:13:43,810 --> 00:13:44,840
what does a sentence mean?

361
00:13:44,840 --> 00:13:46,030
What's the ergative syntax?

362
00:13:46,030 --> 00:13:47,580
All your background knowledge
can be applied.

363
00:13:47,580 --> 00:13:51,010
Again, the power of background
knowledge is immense.

364
00:13:51,010 --> 00:13:53,000
I can tell you that if you go
work somewhere, or if you're

365
00:13:53,000 --> 00:13:55,800
working in a lab now, or if
you're working in any place,

366
00:13:55,800 --> 00:13:58,320
you might be impressed by people
who are older than you,

367
00:13:58,320 --> 00:14:00,830
and what they know, and what
they can learn and pick up.

368
00:14:00,830 --> 00:14:02,580
It's background knowledge,
background knowledge,

369
00:14:02,580 --> 00:14:04,220
background knowledge.

370
00:14:04,220 --> 00:14:07,860
It lets you quickly understand
an idea, because you can

371
00:14:07,860 --> 00:14:11,150
interpret it from prior ideas
that you know, quickly read a

372
00:14:11,150 --> 00:14:13,950
paper, because you can slot it
into prior things you know.

373
00:14:13,950 --> 00:14:17,850
Background knowledge is
incredibly powerful to let you

374
00:14:17,850 --> 00:14:20,350
discover the signal and get
a lot of the specific

375
00:14:20,350 --> 00:14:22,706
information in any kind of
information out there.

376
00:14:26,140 --> 00:14:27,806
Tyler, I left with
you this time.

377
00:14:27,806 --> 00:14:28,560
Sorry.

378
00:14:28,560 --> 00:14:31,420
So we're going to do a
quick memory test.

379
00:14:31,420 --> 00:14:32,500
You will do it at
your own seat.

380
00:14:32,500 --> 00:14:34,300
It's voluntary.

381
00:14:34,300 --> 00:14:38,400
But I will ask you to share a
little bit of information.

382
00:14:38,400 --> 00:14:40,160
This is for everybody now,
not for one person to

383
00:14:40,160 --> 00:14:43,000
be put on the spot.

384
00:14:43,000 --> 00:14:48,600
So what I'm going to do is read
you a list of words, wait

385
00:14:48,600 --> 00:14:49,720
until the end.

386
00:14:49,720 --> 00:14:52,460
And when I say, recall,
write down the

387
00:14:52,460 --> 00:14:54,560
words from your memory.

388
00:14:54,560 --> 00:14:55,910
So I'll wait for a moment
for you to get a

389
00:14:55,910 --> 00:14:57,985
pencil or pen or something.

390
00:15:18,800 --> 00:15:20,050
OK.

391
00:15:23,000 --> 00:15:26,410
I see a few people still
getting stuff.

392
00:15:26,410 --> 00:15:27,210
OK, ready?

393
00:15:27,210 --> 00:15:27,840
Here comes the word.

394
00:15:27,840 --> 00:15:29,350
Just listen, don't write.

395
00:15:29,350 --> 00:15:30,505
That would be too easy.

396
00:15:30,505 --> 00:15:32,070
I'll say recall, and then
write them down.

397
00:15:32,070 --> 00:15:32,880
Here we go.

398
00:15:32,880 --> 00:15:39,590
Mailbox, sardine, shotgun,
peacock, credit, detail,

399
00:15:39,590 --> 00:15:44,610
flicker, airline, spinach,
clarinet.

400
00:15:44,610 --> 00:15:45,860
Recall.

401
00:16:06,570 --> 00:16:07,820
OK.

402
00:16:09,800 --> 00:16:11,050
Here are the answers.

403
00:16:13,160 --> 00:16:15,970
As you look at this list, put
your hand up if you've got all

404
00:16:15,970 --> 00:16:29,870
10, 9, 8, 7, 6, 5, 4, 3, 2.

405
00:16:29,870 --> 00:16:30,400
Thank you.

406
00:16:30,400 --> 00:16:30,950
All right.

407
00:16:30,950 --> 00:16:33,830
So think about this
for a moment.

408
00:16:33,830 --> 00:16:35,400
We said seven plus
or minus two.

409
00:16:35,400 --> 00:16:38,990
When the words are presented in
a random, unorganized list

410
00:16:38,990 --> 00:16:40,820
like this, you're pretty much
back to short-term memory

411
00:16:40,820 --> 00:16:43,790
limits in many senses, not
completely, but in many ways.

412
00:16:43,790 --> 00:16:45,860
So if you were to tell a person
on the street, I mean,

413
00:16:45,860 --> 00:16:48,110
I have to say two things about
you guys, which is certainly

414
00:16:48,110 --> 00:16:50,780
true and complimentary.

415
00:16:50,780 --> 00:16:55,210
Most of you, if you're somewhere
between 18 and 25,

416
00:16:55,210 --> 00:16:56,850
you're at your peak memory
capacity for the

417
00:16:56,850 --> 00:16:58,100
rest of your life.

418
00:17:00,680 --> 00:17:03,080
You will never do better
for rote learning then

419
00:17:03,080 --> 00:17:03,780
you are right now.

420
00:17:03,780 --> 00:17:05,300
There's a ton of research
that shows that.

421
00:17:05,300 --> 00:17:07,710
You will never do better
for rote learning

422
00:17:07,710 --> 00:17:09,470
than about 18 to 25.

423
00:17:09,470 --> 00:17:12,150
After that it's downhill.

424
00:17:12,150 --> 00:17:13,510
You get other benefits.

425
00:17:13,510 --> 00:17:14,730
Salaries go up a little bit.

426
00:17:14,730 --> 00:17:15,400
You get authority.

427
00:17:15,400 --> 00:17:18,150
That's a nice one.

428
00:17:18,150 --> 00:17:20,500
But rote memory, you're
at the top.

429
00:17:20,500 --> 00:17:23,960
Not only are you at the top,
but you are an incredibly

430
00:17:23,960 --> 00:17:26,690
selected academically
achieving group.

431
00:17:26,690 --> 00:17:28,450
You're MIT undergraduates.

432
00:17:28,450 --> 00:17:32,630
You're incredibly selected to be
the top of the top, in many

433
00:17:32,630 --> 00:17:36,730
senses, for learning between 18
to 25 as far as MIT could

434
00:17:36,730 --> 00:17:39,200
possibly figure this out.

435
00:17:39,200 --> 00:17:41,750
You are our hope
for the future.

436
00:17:41,750 --> 00:17:46,190
If you guys can't fix global
warming, we're in trouble.

437
00:17:46,190 --> 00:17:48,270
So you can't get 10 words?

438
00:17:48,270 --> 00:17:49,660
Come on.

439
00:17:49,660 --> 00:17:50,710
We have big problems to solve.

440
00:17:50,710 --> 00:17:54,660
It's a capacity of the human
mind that we just can't do

441
00:17:54,660 --> 00:17:56,020
about more than this.

442
00:17:56,020 --> 00:17:59,170
It's just amazing how limited
rote memory is

443
00:17:59,170 --> 00:18:00,080
without other things.

444
00:18:00,080 --> 00:18:01,920
But there's a little bit
more than we can tell.

445
00:18:01,920 --> 00:18:05,520
How many of you got mailbox
and sardine?

446
00:18:05,520 --> 00:18:06,690
A fair number of hands.

447
00:18:06,690 --> 00:18:07,150
OK.

448
00:18:07,150 --> 00:18:09,090
How many of you got spinach
and clarinet,

449
00:18:09,090 --> 00:18:10,150
the last two words?

450
00:18:10,150 --> 00:18:12,650
How many of you got
credit and detail?

451
00:18:12,650 --> 00:18:13,710
Way fewer hands.

452
00:18:13,710 --> 00:18:16,300
And that's what happens under a
well-controlled experiment.

453
00:18:16,300 --> 00:18:17,510
So this is correct.

454
00:18:17,510 --> 00:18:18,890
High is correct.

455
00:18:18,890 --> 00:18:22,750
This is what order the word
was in the first word you

456
00:18:22,750 --> 00:18:24,430
heard, say the fifth.

457
00:18:24,430 --> 00:18:26,610
This experiment is 15 words.

458
00:18:26,610 --> 00:18:28,880
You can see that.

459
00:18:28,880 --> 00:18:30,200
Here's their performance.

460
00:18:30,200 --> 00:18:31,910
Two things are noteworthy.

461
00:18:31,910 --> 00:18:35,170
People do best for the first
couple of words.

462
00:18:35,170 --> 00:18:38,170
And sometimes they did best for
the last couple of words.

463
00:18:38,170 --> 00:18:39,540
So here's three delays.

464
00:18:39,540 --> 00:18:42,920
These delays are zero
seconds what we did.

465
00:18:42,920 --> 00:18:45,020
You get the list, you
immediately write them down

466
00:18:45,020 --> 00:18:46,180
and recall them verbally.

467
00:18:46,180 --> 00:18:47,950
That's zero, immediate.

468
00:18:47,950 --> 00:18:51,300
Or I wait 10 seconds and
then say, now recall.

469
00:18:51,300 --> 00:18:53,850
Or I wait 1/2 minute and
say, now recall.

470
00:18:53,850 --> 00:18:55,740
Now memory for the first two
items is what's called a

471
00:18:55,740 --> 00:18:56,920
primacy effect.

472
00:18:56,920 --> 00:18:58,170
It doesn't matter.

473
00:18:58,170 --> 00:19:00,500
In all three conditions, those
are the best two remembered

474
00:19:00,500 --> 00:19:04,400
words, on average, for most
of the experiment.

475
00:19:04,400 --> 00:19:07,410
And people think that's a
signature of long-term memory.

476
00:19:07,410 --> 00:19:08,650
You can get those one
or two words.

477
00:19:08,650 --> 00:19:10,540
And then you start to
get overwhelmed

478
00:19:10,540 --> 00:19:12,490
in long-term memory.

479
00:19:12,490 --> 00:19:16,460
Here, you only get the boost for
the last couple of words

480
00:19:16,460 --> 00:19:17,620
if there's zero second delay.

481
00:19:17,620 --> 00:19:20,700
If there's 10 seconds or 30
seconds delay, no boost.

482
00:19:20,700 --> 00:19:23,260
So people interpret this recency
effect, superior

483
00:19:23,260 --> 00:19:25,610
memory for the last couple of
words, as a signature of

484
00:19:25,610 --> 00:19:28,360
short-term memory that
lasts only seconds.

485
00:19:28,360 --> 00:19:29,790
And so in that simple
experiment, you can see

486
00:19:29,790 --> 00:19:32,020
long-term memory influences,
short-term memory influences.

487
00:19:32,020 --> 00:19:33,090
And short-term is
very short-term.

488
00:19:33,090 --> 00:19:34,340
It's just seconds.

489
00:19:37,300 --> 00:19:39,850
Now Ebbinghaus is one of the
founders of experimental

490
00:19:39,850 --> 00:19:40,630
studies of memory.

491
00:19:40,630 --> 00:19:43,310
And he was heroic.

492
00:19:43,310 --> 00:19:44,420
He would just train himself.

493
00:19:44,420 --> 00:19:46,160
He would use nonsense syllables,
because he wanted

494
00:19:46,160 --> 00:19:49,050
to get rid of a lot of
background stuff, rote memory.

495
00:19:49,050 --> 00:19:52,100
And he would learn thousands of
these and then test himself

496
00:19:52,100 --> 00:19:54,650
and score himself honestly.

497
00:19:54,650 --> 00:19:56,460
He did the bulk of the
work on himself.

498
00:19:56,460 --> 00:20:00,050
And he discovered something
that's pretty simple, but that

499
00:20:00,050 --> 00:20:01,180
has had a huge effect.

500
00:20:01,180 --> 00:20:03,790
So how well do you do since you
learned something all in

501
00:20:03,790 --> 00:20:06,170
long-term memory if you're
tested immediately?

502
00:20:06,170 --> 00:20:08,940
100% if it's small list
and you studied a lot.

503
00:20:08,940 --> 00:20:11,470
20 minutes later, an hour later,
nine hours later, you

504
00:20:11,470 --> 00:20:15,150
can see the steep forgetting
up until about 10 hours.

505
00:20:15,150 --> 00:20:16,875
And then it sort of
hangs in there.

506
00:20:16,875 --> 00:20:20,200
The forgetting curve of
long-term memory, immediately

507
00:20:20,200 --> 00:20:21,650
afterwards, we're pretty good.

508
00:20:21,650 --> 00:20:23,700
Then there's a lot of forgetting
in 20 minutes, and

509
00:20:23,700 --> 00:20:26,570
for a day, and then things
hang in there.

510
00:20:26,570 --> 00:20:32,210
This is a shaped asymptotic
curve of forgetting.

511
00:20:32,210 --> 00:20:35,460
OK now here's another
mission for you.

512
00:20:35,460 --> 00:20:36,240
You need a pencil.

513
00:20:36,240 --> 00:20:39,690
Or just your fingers will
work fine either way.

514
00:20:39,690 --> 00:20:40,830
You're going to see
a list of words.

515
00:20:40,830 --> 00:20:42,070
Some will be in uppercase.

516
00:20:42,070 --> 00:20:43,430
Some will be in lowercase.

517
00:20:43,430 --> 00:20:44,470
Capital letters, uppercase.

518
00:20:44,470 --> 00:20:45,070
All right.

519
00:20:45,070 --> 00:20:48,750
Your job as you see them from
top to bottom is to tap your

520
00:20:48,750 --> 00:20:52,970
left hand for each word that's
in the capital letter and your

521
00:20:52,970 --> 00:20:55,610
right hand for each word that's
a lowercase letter.

522
00:20:55,610 --> 00:20:56,480
Is that OK?

523
00:20:56,480 --> 00:20:59,260
So if the first work is capital,
you tap left. if the

524
00:20:59,260 --> 00:21:00,660
second word is lowercase,
you tap right.

525
00:21:00,660 --> 00:21:03,630
So you just tap, tap, tap until
you get to the bottom.

526
00:21:03,630 --> 00:21:04,430
And then it will go.

527
00:21:04,430 --> 00:21:04,720
You already?

528
00:21:04,720 --> 00:21:05,360
You ready to help me out?

529
00:21:05,360 --> 00:21:05,740
Here we go.

530
00:21:05,740 --> 00:21:06,850
Thank you very much.

531
00:21:06,850 --> 00:21:08,100
Here we go.

532
00:21:14,350 --> 00:21:14,890
Excellent.

533
00:21:14,890 --> 00:21:15,940
All right.

534
00:21:15,940 --> 00:21:17,514
What were words?

535
00:21:17,514 --> 00:21:22,910
[LAUGHTER]

536
00:21:22,910 --> 00:21:25,090
PROFESSOR: You might be
impressed that even though

537
00:21:25,090 --> 00:21:26,710
they were right in front of you,
and even though you were

538
00:21:26,710 --> 00:21:28,292
looking whether they were upper
and lowercase case so

539
00:21:28,292 --> 00:21:32,800
you really saw them, if you
don't bring the relevant

540
00:21:32,800 --> 00:21:34,830
background knowledge to
remembering it, which is the

541
00:21:34,830 --> 00:21:37,510
meaning of words most
the time, our

542
00:21:37,510 --> 00:21:39,660
memory is pretty hideous.

543
00:21:39,660 --> 00:21:40,150
All right.

544
00:21:40,150 --> 00:21:40,760
try another one.

545
00:21:40,760 --> 00:21:44,030
Now the trick is gone
a little bit.

546
00:21:44,030 --> 00:21:45,060
This is a little bit
of a set up.

547
00:21:45,060 --> 00:21:47,280
But I can tell you it's
scientifically true.

548
00:21:47,280 --> 00:21:48,810
So now it's the same idea.

549
00:21:48,810 --> 00:21:49,770
You can see words.

550
00:21:49,770 --> 00:21:53,580
You can read the word, if it's a
living thing like a dog, tap

551
00:21:53,580 --> 00:21:54,120
the left hand.

552
00:21:54,120 --> 00:21:56,140
If it's a nonliving thing like
a chair, the right hand.

553
00:21:56,140 --> 00:21:56,980
Ready?

554
00:21:56,980 --> 00:21:57,520
You did awesome.

555
00:21:57,520 --> 00:21:57,770
Thanks.

556
00:21:57,770 --> 00:21:59,020
Here we go.

557
00:22:05,690 --> 00:22:06,940
OK.

558
00:22:09,110 --> 00:22:11,470
Does it feel like you would
remember the words better?

559
00:22:11,470 --> 00:22:14,150
Does it feel like you would?

560
00:22:14,150 --> 00:22:14,530
OK.

561
00:22:14,530 --> 00:22:16,040
So here's what we did.

562
00:22:16,040 --> 00:22:17,850
The first list, is you
thought about the

563
00:22:17,850 --> 00:22:19,350
appearance of the word.

564
00:22:19,350 --> 00:22:21,830
In memory, that's called shallow
encoding, just if it's

565
00:22:21,830 --> 00:22:22,640
upper and lowercase.

566
00:22:22,640 --> 00:22:25,150
And typically, that leads
to pretty poor memory.

567
00:22:25,150 --> 00:22:28,260
If you think about the meaning
of the word, we think that

568
00:22:28,260 --> 00:22:29,210
optimizes memory.

569
00:22:29,210 --> 00:22:30,340
People would call it
deep encoding.

570
00:22:30,340 --> 00:22:31,350
And it leads to better memory.

571
00:22:31,350 --> 00:22:33,480
And in a well-controlled
experiment, now here we had a

572
00:22:33,480 --> 00:22:34,220
certain order.

573
00:22:34,220 --> 00:22:35,490
And you knew there
was a trick.

574
00:22:35,490 --> 00:22:37,340
But if you had a well-controlled
experiment,

575
00:22:37,340 --> 00:22:38,630
here's what you'd find.

576
00:22:38,630 --> 00:22:41,950
If you just think about the
appearance of the word, here's

577
00:22:41,950 --> 00:22:44,830
how well you do, pretty
low if you see the

578
00:22:44,830 --> 00:22:45,880
word once or twice.

579
00:22:45,880 --> 00:22:48,090
And if we look over here at the
meaning of the word you,

580
00:22:48,090 --> 00:22:52,270
can see one thought about the
meaning of the word is better

581
00:22:52,270 --> 00:22:55,870
than two presentations thinking
about the appearance.

582
00:22:55,870 --> 00:22:57,690
And two thoughts about the
meaning of the word, and your

583
00:22:57,690 --> 00:22:58,970
memory's really excellent.

584
00:22:58,970 --> 00:23:00,768
Yeah, question?

585
00:23:00,768 --> 00:23:02,205
AUDIENCE: So if you're forced
to think about--

586
00:23:02,205 --> 00:23:04,918
get confused by the uppercase
and the lowercase and the

587
00:23:04,918 --> 00:23:06,040
living versus the non-living--

588
00:23:06,040 --> 00:23:07,760
PROFESSOR: Were they confusing
to do one then the other?

589
00:23:07,760 --> 00:23:08,908
It was a little bit, right?

590
00:23:08,908 --> 00:23:11,150
AUDIENCE: Is that supposed to
make you remember it better?

591
00:23:11,150 --> 00:23:13,230
You have to think that
hard about it?

592
00:23:13,230 --> 00:23:14,480
PROFESSOR: No.

593
00:23:16,930 --> 00:23:19,340
The question is, is the
confusion of doing two

594
00:23:19,340 --> 00:23:20,670
different tests help you out?

595
00:23:20,670 --> 00:23:22,650
So, roughly speaking, the answer
is to the extent it

596
00:23:22,650 --> 00:23:24,780
draws your attention more
to the word, yes.

597
00:23:24,780 --> 00:23:25,730
To the extent you're
sitting there

598
00:23:25,730 --> 00:23:27,230
going, OK, was it uppercase?

599
00:23:27,230 --> 00:23:27,980
Or was it living?

600
00:23:27,980 --> 00:23:29,690
This is weirding me out.

601
00:23:29,690 --> 00:23:31,990
It's going to hurt you, because
your mind is filled

602
00:23:31,990 --> 00:23:33,810
with those thoughts
not looking at

603
00:23:33,810 --> 00:23:35,060
what's in front of you.

604
00:23:37,710 --> 00:23:43,320
In these demonstrations, they're
a little cloogy.

605
00:23:43,320 --> 00:23:45,450
Here's a fun one, the
effective context.

606
00:23:45,450 --> 00:23:47,230
So this is how they did
the experiment.

607
00:23:47,230 --> 00:23:48,070
You can do it different ways.

608
00:23:48,070 --> 00:23:48,780
But this is a fun one.

609
00:23:48,780 --> 00:23:50,420
They actually went
in the United

610
00:23:50,420 --> 00:23:52,140
Kingdom to the shoreline.

611
00:23:52,140 --> 00:23:54,810
And they had people put on
these old style diving

612
00:23:54,810 --> 00:23:56,350
helmets, the old big helmets.

613
00:23:56,350 --> 00:23:58,550
And you go under the water with
the big hose in front of

614
00:23:58,550 --> 00:24:00,240
you to your air source.

615
00:24:00,240 --> 00:24:03,030
And they had them learn word
lists, like you were looking

616
00:24:03,030 --> 00:24:06,175
at word lists, on the
land or underwater.

617
00:24:08,860 --> 00:24:11,230
And then they were tested
for this on the

618
00:24:11,230 --> 00:24:12,780
land or in the water.

619
00:24:12,780 --> 00:24:16,100
And the core message was, you
do better remembering

620
00:24:16,100 --> 00:24:19,350
information if you're tested in
the same situation in which

621
00:24:19,350 --> 00:24:20,420
you learned it.

622
00:24:20,420 --> 00:24:24,430
So what we see here is here's
the words learned underwater.

623
00:24:24,430 --> 00:24:26,730
You're better if you're tested
underwater instead of

624
00:24:26,730 --> 00:24:27,990
going on the land.

625
00:24:27,990 --> 00:24:30,040
If you learned the word on the
land, you're better off if you

626
00:24:30,040 --> 00:24:32,590
test on the land than if
you go underwater.

627
00:24:32,590 --> 00:24:36,710
The context in which you learned
something helps you.

628
00:24:36,710 --> 00:24:39,960
Studying underwater and tested
under water, you do better

629
00:24:39,960 --> 00:24:43,240
than studying underwater and
going on land or vice versa.

630
00:24:43,240 --> 00:24:47,690
Being in the same room, same
time of day, all those things

631
00:24:47,690 --> 00:24:49,760
help you a little bit to
bring a memory back.

632
00:24:49,760 --> 00:24:52,300
Because those are parts of the
memory in the context in which

633
00:24:52,300 --> 00:24:54,250
you experience it.

634
00:24:54,250 --> 00:24:56,010
Another approach that people
have taken, this is about

635
00:24:56,010 --> 00:24:56,430
organization.

636
00:24:56,430 --> 00:24:59,070
They'll present a set of words
like this and ask people to

637
00:24:59,070 --> 00:25:00,090
try and remember them.

638
00:25:00,090 --> 00:25:02,170
It's a pretty big list.

639
00:25:02,170 --> 00:25:06,930
But if you take the same words
and do that, same words people

640
00:25:06,930 --> 00:25:09,460
remember a lot better,
because there's an

641
00:25:09,460 --> 00:25:11,880
organization that helps you.

642
00:25:11,880 --> 00:25:14,420
Your background knowledge of
what is a career, what is

643
00:25:14,420 --> 00:25:18,820
housework, what is a food all
contribute to helping your

644
00:25:18,820 --> 00:25:22,260
memory for these items
be better.

645
00:25:22,260 --> 00:25:24,400
So let's think about
forgetting.

646
00:25:24,400 --> 00:25:25,730
We know something
from last year.

647
00:25:25,730 --> 00:25:26,320
We forgot it.

648
00:25:26,320 --> 00:25:27,900
There's a least two ideas that
people have said, what's the

649
00:25:27,900 --> 00:25:30,150
core reason we forget things?

650
00:25:30,150 --> 00:25:31,390
One of them is like passive.

651
00:25:31,390 --> 00:25:34,920
Well, I haven't used
it in a while.

652
00:25:34,920 --> 00:25:35,930
It's just not available
anymore.

653
00:25:35,930 --> 00:25:37,020
Because time has passed.

654
00:25:37,020 --> 00:25:37,410
It's not been used.

655
00:25:37,410 --> 00:25:38,480
It's not available.

656
00:25:38,480 --> 00:25:40,520
That's passive forgetting.

657
00:25:40,520 --> 00:25:43,190
The other way to think about why
we forget is not that it's

658
00:25:43,190 --> 00:25:45,560
a passive loss of knowledge.

659
00:25:45,560 --> 00:25:50,100
But that experiences you have
after that moment when you

660
00:25:50,100 --> 00:25:53,560
learned something, will go
back and write over, and

661
00:25:53,560 --> 00:25:56,230
muddy, and confuse your memory
for the original event.

662
00:25:56,230 --> 00:25:59,505
It's active interference of
subsequent experience.

663
00:25:59,505 --> 00:26:02,020
It's not passive withering.

664
00:26:02,020 --> 00:26:05,770
Other stuff is getting in your
head that's writing over and

665
00:26:05,770 --> 00:26:07,810
modifying the original memory.

666
00:26:07,810 --> 00:26:09,410
And the original memory
is never there again.

667
00:26:12,150 --> 00:26:14,320
And people have talked about it
in two different directions

668
00:26:14,320 --> 00:26:15,560
that you can think about this.

669
00:26:15,560 --> 00:26:18,490
One of them is what you call
proactive interference where

670
00:26:18,490 --> 00:26:22,050
things from before mess up your
learning on something.

671
00:26:22,050 --> 00:26:25,620
Or retroactive interference, you
already have the memory.

672
00:26:25,620 --> 00:26:29,085
And now new things mess you up
and write over the memory.

673
00:26:29,085 --> 00:26:32,150
So memory is threatened by
what came before they

674
00:26:32,150 --> 00:26:35,440
experience or what happens after
the experience In the

675
00:26:35,440 --> 00:26:38,060
way that it interferes with
accuracy of memory.

676
00:26:38,060 --> 00:26:40,190
So here's a fun experiment.

677
00:26:40,190 --> 00:26:40,810
It's just two subjects.

678
00:26:40,810 --> 00:26:42,560
But there's a lot of other
studies that looked at this.

679
00:26:42,560 --> 00:26:45,600
This is the original
demonstration where people had

680
00:26:45,600 --> 00:26:47,040
to remember syllables.

681
00:26:47,040 --> 00:26:50,190
And they taught them the
syllables just before they

682
00:26:50,190 --> 00:26:54,180
went to sleep or as
they woke up.

683
00:26:54,180 --> 00:26:57,050
And in each case, they were
tested, for example, some

684
00:26:57,050 --> 00:26:57,800
number of hours later.

685
00:26:57,800 --> 00:26:59,240
Let's say eight hours later.

686
00:26:59,240 --> 00:27:01,490
And their idea in the study--
and there's one wrinkle I'll

687
00:27:01,490 --> 00:27:03,160
mention now that we
know things--

688
00:27:03,160 --> 00:27:07,380
their idea was when you go to
sleep, there's no retroactive

689
00:27:07,380 --> 00:27:08,100
interference.

690
00:27:08,100 --> 00:27:09,550
Because you're not learning
anything else.

691
00:27:09,550 --> 00:27:10,240
You're not experiencing.

692
00:27:10,240 --> 00:27:10,850
You're not talking.

693
00:27:10,850 --> 00:27:11,380
You're not reading.

694
00:27:11,380 --> 00:27:13,810
So when you wake up in the
morning, nothing has messed up

695
00:27:13,810 --> 00:27:15,580
your memory from when you
just went to sleep.

696
00:27:15,580 --> 00:27:16,720
Does that make sense?

697
00:27:16,720 --> 00:27:19,110
If you study in the morning,
now you go and

698
00:27:19,110 --> 00:27:19,720
you have your day.

699
00:27:19,720 --> 00:27:21,000
You go to classes you
talk to people.

700
00:27:21,000 --> 00:27:21,850
You read things.

701
00:27:21,850 --> 00:27:25,800
All those new experiences can go
back and mess up the memory

702
00:27:25,800 --> 00:27:26,810
you acquired to start with.

703
00:27:26,810 --> 00:27:28,540
So in both cases, you get the
memory to start with.

704
00:27:28,540 --> 00:27:30,400
But in one case, you're sleeping
for eight hours.

705
00:27:30,400 --> 00:27:31,830
It's not messed up.

706
00:27:31,830 --> 00:27:32,980
In the other case, you were
out in the world.

707
00:27:32,980 --> 00:27:34,370
Lots of new things are
happening that are

708
00:27:34,370 --> 00:27:36,350
retroactively messing
up the memory.

709
00:27:39,810 --> 00:27:42,280
And you can see the people who
slept after who slept after

710
00:27:42,280 --> 00:27:44,330
they learned, when they woke up,
did a lot better than the

711
00:27:44,330 --> 00:27:47,290
people who walked around
during the day.

712
00:27:47,290 --> 00:27:51,570
And that thought can be one of a
huge line of research saying

713
00:27:51,570 --> 00:27:54,800
that one of the reasons that
memories become available to

714
00:27:54,800 --> 00:27:58,970
us is not simply passive
withering, but active other

715
00:27:58,970 --> 00:28:02,940
mental experiences that block
right over and change the

716
00:28:02,940 --> 00:28:03,383
original memory.

717
00:28:03,383 --> 00:28:03,910
Question, yeah.

718
00:28:03,910 --> 00:28:08,002
AUDIENCE: Studying at night, and
they going to sleep, does

719
00:28:08,002 --> 00:28:11,236
just having those memories
hanging around in your brain

720
00:28:11,236 --> 00:28:12,210
strengthen them?

721
00:28:12,210 --> 00:28:13,180
Because you're going to
wake up and do stuff--

722
00:28:13,180 --> 00:28:13,490
PROFESSOR: Yeah.

723
00:28:13,490 --> 00:28:15,290
So now there's a second thing.

724
00:28:15,290 --> 00:28:17,380
The question is, is there
something good about going to

725
00:28:17,380 --> 00:28:18,220
sleep after studying?

726
00:28:18,220 --> 00:28:21,480
And the answer is, there's
a lot of evidence now in

727
00:28:21,480 --> 00:28:25,360
animals, and increasingly in
humans, that sleep is a period

728
00:28:25,360 --> 00:28:29,030
that fixes memories from
the prior day.

729
00:28:29,030 --> 00:28:31,320
And if you don't get the right
kind of sleep or something

730
00:28:31,320 --> 00:28:34,215
like that, those memories from
the prior day don't seem to

731
00:28:34,215 --> 00:28:36,610
get as well fixed into
long-term memory.

732
00:28:36,610 --> 00:28:38,250
So this is mixing those
two issues.

733
00:28:38,250 --> 00:28:40,250
These people have two advantages
now we think.

734
00:28:40,250 --> 00:28:42,450
One advantage they have is
they're not having new

735
00:28:42,450 --> 00:28:45,010
experiences during the night
that are messing up

736
00:28:45,010 --> 00:28:46,790
the memory they got.

737
00:28:46,790 --> 00:28:49,030
And they're having this
consolidation

738
00:28:49,030 --> 00:28:50,440
period during sleep.

739
00:28:50,440 --> 00:28:53,360
People have long studied
what's the use of sleep

740
00:28:53,360 --> 00:28:55,060
besides making you not sleepy?

741
00:28:55,060 --> 00:28:57,920
And one of the most fascinating
things-- and

742
00:28:57,920 --> 00:29:00,170
there's just more and more
evidence for this--

743
00:29:00,170 --> 00:29:02,310
it's a period where because our
brain is kind of quiet in

744
00:29:02,310 --> 00:29:05,760
many ways, it seems to work a
lot on fixing memories from

745
00:29:05,760 --> 00:29:09,910
that day into a better
long-term unit.

746
00:29:09,910 --> 00:29:11,800
So those people have two
advantages by our current

747
00:29:11,800 --> 00:29:13,650
understanding.

748
00:29:13,650 --> 00:29:14,730
So that's retroactive
interference.

749
00:29:14,730 --> 00:29:17,760
Let me give you an example of
proactive interference.

750
00:29:17,760 --> 00:29:18,800
And here's how they study it.

751
00:29:18,800 --> 00:29:19,490
We won't to try it.

752
00:29:19,490 --> 00:29:21,480
But let me give you
an example.

753
00:29:21,480 --> 00:29:25,090
They would have you hear
three words like

754
00:29:25,090 --> 00:29:26,080
banana, peach, and apple.

755
00:29:26,080 --> 00:29:27,000
They're all related.

756
00:29:27,000 --> 00:29:30,250
Then they'd have you count aloud
from 138 backward by

757
00:29:30,250 --> 00:29:32,330
sevens for a few moments.

758
00:29:32,330 --> 00:29:33,340
They say, what were the words?

759
00:29:33,340 --> 00:29:34,550
And you give what you can.

760
00:29:34,550 --> 00:29:36,360
Then they give you a plum,
apricot, and lime, count

761
00:29:36,360 --> 00:29:38,060
backwards by sevens.

762
00:29:38,060 --> 00:29:39,520
Melon, lemon, we can see
all of these are

763
00:29:39,520 --> 00:29:41,450
from the same category.

764
00:29:41,450 --> 00:29:44,290
And then they switch
the category.

765
00:29:44,290 --> 00:29:46,550
So we said, typically,
background knowledge helps.

766
00:29:46,550 --> 00:29:48,730
But here's the way it can
get weirdly not helpful.

767
00:29:48,730 --> 00:29:50,910
So here's the performance
of healthy young

768
00:29:50,910 --> 00:29:51,870
people doing this.

769
00:29:51,870 --> 00:29:52,930
So this is percent correct.

770
00:29:52,930 --> 00:29:53,370
Good is high.

771
00:29:53,370 --> 00:29:54,880
The first list, they're
really good.

772
00:29:54,880 --> 00:29:56,570
Occasionally they forget
one of the three words.

773
00:29:56,570 --> 00:29:58,320
But looks what's happening as
they go along for the second

774
00:29:58,320 --> 00:29:59,590
list and the third list.

775
00:29:59,590 --> 00:30:02,720
They're getting worse
and worse.

776
00:30:02,720 --> 00:30:04,510
That's proactive interference.

777
00:30:04,510 --> 00:30:07,310
The experience in the first
list is messing you up to

778
00:30:07,310 --> 00:30:09,190
learn the new information
in the second list.

779
00:30:09,190 --> 00:30:12,130
And most of the errors people
make are substitutions.

780
00:30:12,130 --> 00:30:15,540
They go plum, apricot, apple.

781
00:30:15,540 --> 00:30:18,450
By the third list, they're
going melon, lime, apple.

782
00:30:18,450 --> 00:30:19,770
Do you see what's happening?

783
00:30:19,770 --> 00:30:23,900
Because they're related, because
they're related, now

784
00:30:23,900 --> 00:30:26,370
you're getting confused.

785
00:30:26,370 --> 00:30:28,810
If I heard the word grape
recently, was that in the

786
00:30:28,810 --> 00:30:29,490
first group?

787
00:30:29,490 --> 00:30:30,740
I'm not reporting that
one right now.

788
00:30:30,740 --> 00:30:33,100
Was it in the one before that
or the one I'm supposed to

789
00:30:33,100 --> 00:30:34,430
report right now?

790
00:30:34,430 --> 00:30:38,240
The relations among them mess
you up because it's hard to

791
00:30:38,240 --> 00:30:39,480
tell when you heard them.

792
00:30:39,480 --> 00:30:40,550
And they're all related.

793
00:30:40,550 --> 00:30:42,610
And just to convince you that
it's true, because you might

794
00:30:42,610 --> 00:30:45,520
say well, maybe you're just
getting tired like list, and

795
00:30:45,520 --> 00:30:47,120
list, and counting.

796
00:30:47,120 --> 00:30:48,400
Then they give you a new list.

797
00:30:48,400 --> 00:30:49,570
And look what happens.

798
00:30:49,570 --> 00:30:52,110
If it's fruits, still fruits,
you're still stuck.

799
00:30:52,110 --> 00:30:54,740
But if it's another list like
vegetables, flowers, and

800
00:30:54,740 --> 00:30:56,770
meats, or professions,
very different.

801
00:30:56,770 --> 00:30:59,080
Your memory zooms
right back up.

802
00:30:59,080 --> 00:31:02,650
Because all the interference you
had here about fruits is

803
00:31:02,650 --> 00:31:06,290
no longer relevant for learning
a set of professions.

804
00:31:06,290 --> 00:31:07,590
And it's kind of
graded, right?

805
00:31:07,590 --> 00:31:11,340
We think the flowers as more
related to fruits kind of.

806
00:31:11,340 --> 00:31:14,730
And professions is very
unrelated, doctor, lawyer,

807
00:31:14,730 --> 00:31:16,160
programmer.

808
00:31:16,160 --> 00:31:18,450
The more unrelated it is, the
more your memory goes right

809
00:31:18,450 --> 00:31:19,360
back to where it was.

810
00:31:19,360 --> 00:31:20,710
Because the proactive
interference

811
00:31:20,710 --> 00:31:21,960
is no longer happening.

812
00:31:24,610 --> 00:31:29,490
Another thing that researchers
have focused on as we read

813
00:31:29,490 --> 00:31:33,000
things, what do we focus on?

814
00:31:33,000 --> 00:31:37,650
So here's an example where they
read people sentences.

815
00:31:37,650 --> 00:31:41,160
And then they tested them either
for the meaning or for

816
00:31:41,160 --> 00:31:42,790
the style of the sentence.

817
00:31:42,790 --> 00:31:44,890
Usually when we read stuff,
we're focused on the meaning,

818
00:31:44,890 --> 00:31:46,090
not the style.

819
00:31:46,090 --> 00:31:47,110
Style, we might notice it.

820
00:31:47,110 --> 00:31:48,580
But usually, what
am I reading?

821
00:31:48,580 --> 00:31:49,830
What's the point?

822
00:31:53,040 --> 00:31:55,460
If they weren't unwarned,
they read the sentence.

823
00:31:55,460 --> 00:31:56,230
Then you're tested.

824
00:31:56,230 --> 00:31:58,790
It's much better for meaning
than for style.

825
00:31:58,790 --> 00:32:00,630
Because we automatically
go for meaning.

826
00:32:00,630 --> 00:32:03,120
That's what we go for when
we read by and large.

827
00:32:03,120 --> 00:32:04,630
When we think about poetry
or something,

828
00:32:04,630 --> 00:32:05,300
that might be different.

829
00:32:05,300 --> 00:32:07,430
But reading typical text,
what's the meaning?

830
00:32:07,430 --> 00:32:09,630
We strip away, almost,
the style.

831
00:32:09,630 --> 00:32:12,320
We want to get to the content.

832
00:32:12,320 --> 00:32:15,450
And they're terrible if they're
asked about the style

833
00:32:15,450 --> 00:32:17,760
or the specific organization of
the sentence, the word by

834
00:32:17,760 --> 00:32:18,950
word organization.

835
00:32:18,950 --> 00:32:20,480
But, of course, if they
focus on it, they

836
00:32:20,480 --> 00:32:21,870
can do a lot better.

837
00:32:21,870 --> 00:32:24,000
So this is just saying
that we naturally go

838
00:32:24,000 --> 00:32:25,610
for gist and meaning.

839
00:32:25,610 --> 00:32:28,130
And we try to throw away a lot
of the specific particulars

840
00:32:28,130 --> 00:32:29,510
like the way the sentence
was organized, the

841
00:32:29,510 --> 00:32:32,160
style of the writing.

842
00:32:32,160 --> 00:32:32,960
OK.

843
00:32:32,960 --> 00:32:35,240
So now I'd like to do a
demonstration that will work

844
00:32:35,240 --> 00:32:36,140
better for some than others.

845
00:32:36,140 --> 00:32:38,660
But here we go.

846
00:32:38,660 --> 00:32:40,740
So here we go.

847
00:32:40,740 --> 00:32:42,250
I'm going to read you
a list of words.

848
00:32:42,250 --> 00:32:46,010
And then I'm going to ask you
for a couple different words

849
00:32:46,010 --> 00:32:47,635
and ask you if they
were on the list.

850
00:32:47,635 --> 00:32:48,060
OK.

851
00:32:48,060 --> 00:32:48,560
Here we go.

852
00:32:48,560 --> 00:32:49,930
Ready?

853
00:32:49,930 --> 00:32:57,410
Door, glass, pane, shade, ledge,
sill, house, open,

854
00:32:57,410 --> 00:33:03,760
curtain, frame, view, breeze,
sash, screen, shutter.

855
00:33:03,760 --> 00:33:05,300
OK, was the word glass?

856
00:33:05,300 --> 00:33:07,020
AUDIENCE: Yes.

857
00:33:07,020 --> 00:33:08,720
PROFESSOR: Was the word
potato on the list?

858
00:33:08,720 --> 00:33:09,480
AUDIENCE: No.

859
00:33:09,480 --> 00:33:11,026
PROFESSOR: Was the word
shade on the list?

860
00:33:11,026 --> 00:33:11,780
AUDIENCE: Yes.

861
00:33:11,780 --> 00:33:13,870
PROFESSOR: Was the word
car on the list?

862
00:33:13,870 --> 00:33:14,590
AUDIENCE: No.

863
00:33:14,590 --> 00:33:15,836
PROFESSOR: Was the word
window on the list?

864
00:33:15,836 --> 00:33:16,172
AUDIENCE: Yes.

865
00:33:16,172 --> 00:33:16,790
No.

866
00:33:16,790 --> 00:33:16,930
PROFESSOR: OK.

867
00:33:16,930 --> 00:33:18,310
How many people said window?

868
00:33:18,310 --> 00:33:20,110
Put your hands up.

869
00:33:20,110 --> 00:33:22,060
I heard a lot of window.

870
00:33:22,060 --> 00:33:22,750
All right.

871
00:33:22,750 --> 00:33:24,230
You're not helping
me out here.

872
00:33:24,230 --> 00:33:27,690
Smart undergraduates like you,
at the peak of your memory,

873
00:33:27,690 --> 00:33:29,440
will about half the time
say yes to window.

874
00:33:29,440 --> 00:33:31,460
You saw this demonstration
before, or

875
00:33:31,460 --> 00:33:33,260
you heard it before.

876
00:33:33,260 --> 00:33:36,890
The way they compose these lists
is they ask students to

877
00:33:36,890 --> 00:33:38,890
say what's the first word you
think of that goes with door?

878
00:33:38,890 --> 00:33:40,200
The second word?

879
00:33:40,200 --> 00:33:42,050
And the first word that goes
with door, the most common

880
00:33:42,050 --> 00:33:44,410
one, they'll put over here.

881
00:33:44,410 --> 00:33:46,440
That's the trick word,
the lure word.

882
00:33:46,440 --> 00:33:48,850
And then these are other words
that people think kind of goes

883
00:33:48,850 --> 00:33:50,160
with the word door.

884
00:33:50,160 --> 00:33:54,530
So if you hear this list, and
you're tested for did you hear

885
00:33:54,530 --> 00:33:56,510
the word window or see the
word window-- it doesn't

886
00:33:56,510 --> 00:33:58,180
matter whether you see
it or hear it--

887
00:33:58,180 --> 00:34:01,720
about half the time
undergraduates will say yes, I

888
00:34:01,720 --> 00:34:02,910
heard it or saw it.

889
00:34:02,910 --> 00:34:04,710
Because they're getting
the gist of the list.

890
00:34:04,710 --> 00:34:07,420
This is all about door
and window stuff.

891
00:34:07,420 --> 00:34:09,949
So when the word window appears,
they have a false or

892
00:34:09,949 --> 00:34:12,150
illusory memory.

893
00:34:12,150 --> 00:34:14,230
The rate of false or illusory
memories goes

894
00:34:14,230 --> 00:34:16,929
up with age as well.

895
00:34:16,929 --> 00:34:18,580
For every decade that will
pass, you'll add

896
00:34:18,580 --> 00:34:21,110
a few more of those.

897
00:34:21,110 --> 00:34:24,659
But even undergraduates,
typically, about half the time

898
00:34:24,659 --> 00:34:26,929
they'll falsely recall they
heard the word window.

899
00:34:26,929 --> 00:34:30,400
Again, the idea is here's
this double-edged sword.

900
00:34:30,400 --> 00:34:34,540
It's about words that go
with doors and windows.

901
00:34:34,540 --> 00:34:35,420
That's the gist.

902
00:34:35,420 --> 00:34:36,050
That's the main idea.

903
00:34:36,050 --> 00:34:38,270
When you get the word window,
you falsely believe that you

904
00:34:38,270 --> 00:34:39,310
saw it or heard it.

905
00:34:39,310 --> 00:34:43,830
So that's the way that people
try in the laboratory to

906
00:34:43,830 --> 00:34:46,830
create false memories that are
experimentally testable.

907
00:34:46,830 --> 00:34:49,150
We'll talk about real life
ones in a moment.

908
00:34:49,150 --> 00:34:53,320
So we said all these things make
it feel like memory is

909
00:34:53,320 --> 00:34:56,350
more the punch bowl metaphor
than the camera.

910
00:34:56,350 --> 00:34:58,840
Because it's all swished
together, the proactive stuff,

911
00:34:58,840 --> 00:35:02,600
the interpretation, the
retroactive changes that are

912
00:35:02,600 --> 00:35:06,500
created in memory are all
altering the memory based on

913
00:35:06,500 --> 00:35:11,170
the interpretation or based
on subsequent experience.

914
00:35:11,170 --> 00:35:15,200
So let me stop for one moment
to tell you, OK,

915
00:35:15,200 --> 00:35:16,360
so memory is shaky.

916
00:35:16,360 --> 00:35:18,930
Wouldn't it be cool to have
truly photographic memory?

917
00:35:18,930 --> 00:35:20,590
I mean, that would be awesome
as a student.

918
00:35:20,590 --> 00:35:22,460
It wouldn't be bad in life.

919
00:35:22,460 --> 00:35:25,020
There's only been one
well-described case of a

920
00:35:25,020 --> 00:35:27,610
person who seems to have truly
photographic memory.

921
00:35:27,610 --> 00:35:29,280
You hear various stories.

922
00:35:29,280 --> 00:35:32,495
Most of the people who are
things like memory performers,

923
00:35:32,495 --> 00:35:35,040
or there's memory Olympics and
things like that, they train

924
00:35:35,040 --> 00:35:37,760
the heck out of themselves to
use different methods that can

925
00:35:37,760 --> 00:35:38,670
enhance your memory.

926
00:35:38,670 --> 00:35:40,560
And they can do amazing
things.

927
00:35:40,560 --> 00:35:43,160
But it's all hard work.

928
00:35:43,160 --> 00:35:45,720
It's all hard work as far
as we understand.

929
00:35:45,720 --> 00:35:48,390
But here's a guy who was
different, studied by Luria, a

930
00:35:48,390 --> 00:35:51,440
psychologist in Russia,
about 100 years ago.

931
00:35:51,440 --> 00:35:53,810
So this is a guy who was a
reporter at a newspaper.

932
00:35:53,810 --> 00:35:57,980
And he would go out and come
back and type up the report of

933
00:35:57,980 --> 00:36:00,140
what he had seen at some event
for the newspaper.

934
00:36:00,140 --> 00:36:03,730
And the editor noticed that
this guy never took notes.

935
00:36:03,730 --> 00:36:05,300
He came back and just
started typing.

936
00:36:05,300 --> 00:36:06,750
And the editor said, you've
got to take notes.

937
00:36:06,750 --> 00:36:08,210
Because you won't know exactly
what happened,

938
00:36:08,210 --> 00:36:09,750
who said what exactly.

939
00:36:09,750 --> 00:36:10,850
You'll make mistakes.

940
00:36:10,850 --> 00:36:11,380
The guy said, no.

941
00:36:11,380 --> 00:36:12,630
I'm always perfect.

942
00:36:12,630 --> 00:36:15,270
The editor said, OK.

943
00:36:15,270 --> 00:36:17,010
We're going to teach you
a lesson about this

944
00:36:17,010 --> 00:36:19,160
just to show you.

945
00:36:19,160 --> 00:36:20,290
He got some material.

946
00:36:20,290 --> 00:36:22,070
He said, everybody
come around.

947
00:36:22,070 --> 00:36:23,950
He's going to embarrass
the guy.

948
00:36:23,950 --> 00:36:24,730
I'm going to read
you something.

949
00:36:24,730 --> 00:36:25,940
You tell me every word I said.

950
00:36:25,940 --> 00:36:28,020
He read him some gibberish.

951
00:36:28,020 --> 00:36:31,550
And the guy repeated every
word perfectly.

952
00:36:31,550 --> 00:36:33,490
This guy truly had photographic
memory.

953
00:36:33,490 --> 00:36:36,210
And kind of interestingly, when
they said, wow, you have

954
00:36:36,210 --> 00:36:37,030
photographic memory.

955
00:36:37,030 --> 00:36:37,640
He didn't go, yes.

956
00:36:37,640 --> 00:36:38,090
I'm awesome.

957
00:36:38,090 --> 00:36:39,380
I'm a superhero.

958
00:36:39,380 --> 00:36:42,420
He said, doesn't everybody?

959
00:36:42,420 --> 00:36:46,150
He was not aware that most of us
don't remember every single

960
00:36:46,150 --> 00:36:50,590
thing that we study easily and
for the rest of our lives.

961
00:36:50,590 --> 00:36:55,720
So he would be presented rows
of digits like these, study

962
00:36:55,720 --> 00:36:59,630
them for three minutes, and he
could recall them days, weeks,

963
00:36:59,630 --> 00:37:01,130
and months later perfectly.

964
00:37:04,370 --> 00:37:06,810
This is almost the only case
we have well-described and

965
00:37:06,810 --> 00:37:07,890
well-analyzed.

966
00:37:07,890 --> 00:37:10,900
So it's really rare.

967
00:37:10,900 --> 00:37:13,350
He would say when he had to
think back at a particular

968
00:37:13,350 --> 00:37:15,523
list, like tell me the one from
last year, or tell me the

969
00:37:15,523 --> 00:37:17,750
one from two years ago, he would
have to think where he

970
00:37:17,750 --> 00:37:19,790
was and reinstate.

971
00:37:19,790 --> 00:37:21,760
So that in his mind's eye, he
would see the room or hear the

972
00:37:21,760 --> 00:37:24,600
voice of the person
reading it to him.

973
00:37:24,600 --> 00:37:27,480
But even 15 years later,
perfect memory

974
00:37:27,480 --> 00:37:28,610
for a list like this.

975
00:37:28,610 --> 00:37:31,350
Wouldn't you like to have
that right now?

976
00:37:31,350 --> 00:37:32,600
I wouldn't mind it either.

977
00:37:35,300 --> 00:37:36,630
He also was synesthetic.

978
00:37:36,630 --> 00:37:39,450
That is, when he heard things in
one modality, it triggered

979
00:37:39,450 --> 00:37:42,290
perception like experience
is another modality.

980
00:37:42,290 --> 00:37:46,280
He could change his resting
pulse from 70 to 100.

981
00:37:46,280 --> 00:37:49,370
He could change his temperature
by two degrees by

982
00:37:49,370 --> 00:37:50,260
force of will.

983
00:37:50,260 --> 00:37:51,510
He's an unusual person.

984
00:37:55,240 --> 00:37:57,950
We would love to discover
another person like this in

985
00:37:57,950 --> 00:37:58,720
the place of the planet.

986
00:37:58,720 --> 00:38:00,785
We have not seen another
one well-documented.

987
00:38:00,785 --> 00:38:02,170
Probably there is.

988
00:38:02,170 --> 00:38:05,630
But they're sitting somewhere
in some village remembering

989
00:38:05,630 --> 00:38:07,880
everything perfectly going, I
thought you all remembered

990
00:38:07,880 --> 00:38:10,810
everything perfectly.

991
00:38:10,810 --> 00:38:11,930
So you could think, wow,
that's awesome.

992
00:38:11,930 --> 00:38:13,900
That's a nice superpower
to have.

993
00:38:13,900 --> 00:38:15,370
But he was not a happy person.

994
00:38:15,370 --> 00:38:16,550
This is kind of an interesting
question.

995
00:38:16,550 --> 00:38:19,390
Because he complained that when
somebody spoke to him

996
00:38:19,390 --> 00:38:21,990
like a family member, what would
happen was he would be

997
00:38:21,990 --> 00:38:25,390
flooded by a word with some
memory for some list or some

998
00:38:25,390 --> 00:38:28,730
event last week, last month,
or 10 years ago.

999
00:38:28,730 --> 00:38:31,720
His family members said, you're
not listening to me.

1000
00:38:31,720 --> 00:38:34,940
Yeah, because every word
triggered back or brought back

1001
00:38:34,940 --> 00:38:37,310
a flood of memory.

1002
00:38:37,310 --> 00:38:40,850
He got so miserable that things
in his environment

1003
00:38:40,850 --> 00:38:44,590
would trigger perfect floods
of memory into his mind--

1004
00:38:44,590 --> 00:38:47,320
and this is truly
like a novel--

1005
00:38:47,320 --> 00:38:49,540
at one point, in order to get
rid of memories-- and this did

1006
00:38:49,540 --> 00:38:52,007
not work-- he wrote them down on
a piece of paper and threw

1007
00:38:52,007 --> 00:38:55,080
them into the fire thinking
somehow that would work.

1008
00:38:55,080 --> 00:38:59,350
So this is one of these Twilight
Zone, O. Henry kinds

1009
00:38:59,350 --> 00:39:03,430
of stories where you think
it's awesome to have

1010
00:39:03,430 --> 00:39:04,710
photographic-like memory.

1011
00:39:04,710 --> 00:39:08,060
But, in fact, it seems to have
blocked, in many ways, his

1012
00:39:08,060 --> 00:39:09,810
development as a person.

1013
00:39:09,810 --> 00:39:12,660
Once he realized he had such
a rare memory, he became a

1014
00:39:12,660 --> 00:39:15,190
professional memory performer.

1015
00:39:15,190 --> 00:39:17,790
And then all kinds of jobs and
finally retreated into a

1016
00:39:17,790 --> 00:39:20,390
countryside kind of unhappy,
treating people with herbs

1017
00:39:20,390 --> 00:39:21,690
with his wife and son.

1018
00:39:21,690 --> 00:39:23,870
He did not have a happy,
awesome life.

1019
00:39:23,870 --> 00:39:26,140
I don't know if in this world
he would do better?

1020
00:39:26,140 --> 00:39:29,810
Because you could get on TV
shows and become a celebrity

1021
00:39:29,810 --> 00:39:31,060
or something if you
could do this.

1022
00:39:33,390 --> 00:39:36,260
So what's the problem with
photographic-like

1023
00:39:36,260 --> 00:39:38,220
memory do you think?

1024
00:39:38,220 --> 00:39:40,980
Why is it that we kind of wish
we had it at certain moments?

1025
00:39:40,980 --> 00:39:42,230
But what's the drawback?

1026
00:39:44,750 --> 00:39:48,660
Well the drawback seems to be,
imagine if you had in your

1027
00:39:48,660 --> 00:39:50,290
head a photograph
of everything.

1028
00:39:50,290 --> 00:39:54,440
You would have so much stuff
that you wouldn't be analyzing

1029
00:39:54,440 --> 00:39:57,000
it for the basic point
of what it's about.

1030
00:39:57,000 --> 00:39:59,470
So we just said, the risk of
analyzing things for their

1031
00:39:59,470 --> 00:40:03,590
basic point, their basic gist,
is that we want to get the

1032
00:40:03,590 --> 00:40:04,240
bottom message.

1033
00:40:04,240 --> 00:40:05,850
And we give up a lot
of the particulars.

1034
00:40:05,850 --> 00:40:07,190
It's not like a photograph.

1035
00:40:07,190 --> 00:40:10,290
It's like a very small abstract
or a note about a

1036
00:40:10,290 --> 00:40:12,110
photograph that's
in your mind.

1037
00:40:12,110 --> 00:40:14,090
But that's a very powerful way
to approach the world.

1038
00:40:14,090 --> 00:40:15,890
Because then all your background
knowledge is

1039
00:40:15,890 --> 00:40:18,580
instantly available to help you
interpret what's going on.

1040
00:40:18,580 --> 00:40:21,150
Because you're using lots of
connections and very few

1041
00:40:21,150 --> 00:40:22,250
particulars.

1042
00:40:22,250 --> 00:40:25,660
Too many particulars are as
burdensome as probably having

1043
00:40:25,660 --> 00:40:28,160
too many generalities.

1044
00:40:28,160 --> 00:40:30,410
Let me talk a minute about
flashbulb memories.

1045
00:40:30,410 --> 00:40:32,525
So this is a kind
of funny thing.

1046
00:40:32,525 --> 00:40:36,280
You might have the experience
in your own life of very

1047
00:40:36,280 --> 00:40:40,070
salient moments, big moments in
your life with your family,

1048
00:40:40,070 --> 00:40:42,690
other things you're doing in
your life, where you say, I'll

1049
00:40:42,690 --> 00:40:43,960
never forget this moment.

1050
00:40:43,960 --> 00:40:46,590
Because just everything
is so vivid for me.

1051
00:40:46,590 --> 00:40:47,840
People call this flashbulb.

1052
00:40:51,290 --> 00:40:54,100
We can never know how accurate
your flashbulb memory is.

1053
00:40:54,100 --> 00:40:56,340
I'll never forget when I got
into MIT, the moment.

1054
00:40:56,340 --> 00:40:58,150
Because I was wearing
my lucky socks.

1055
00:40:58,150 --> 00:40:59,080
And it was sunny outside.

1056
00:40:59,080 --> 00:41:01,070
And you might be right,
or you might be wrong.

1057
00:41:01,070 --> 00:41:06,210
So there's a slightly gruesome
research enterprise which

1058
00:41:06,210 --> 00:41:10,190
takes public events that are
very emotional for people.

1059
00:41:10,190 --> 00:41:13,070
So some of them will be very
historical for you somewhere

1060
00:41:13,070 --> 00:41:14,500
in your lifetime.

1061
00:41:14,500 --> 00:41:18,300
John F. Kennedy's assassination,
the Challenger

1062
00:41:18,300 --> 00:41:21,550
blowing up as a high school
teacher was going up to be the

1063
00:41:21,550 --> 00:41:23,600
first high school teacher
to do that.

1064
00:41:23,600 --> 00:41:27,940
The O.J. Simpson verdict, were
you even alive for this?

1065
00:41:30,760 --> 00:41:31,700
I can tell you that got an

1066
00:41:31,700 --> 00:41:33,110
unbelievable amount of interest.

1067
00:41:33,110 --> 00:41:34,320
Because I'm from Buffalo.

1068
00:41:34,320 --> 00:41:35,540
That's where O.J. Simpson
played football.

1069
00:41:35,540 --> 00:41:37,350
So I have a particular
interest.

1070
00:41:37,350 --> 00:41:40,500
It was like everything you
wanted in a TV show.

1071
00:41:40,500 --> 00:41:41,870
Tragically, two people died.

1072
00:41:41,870 --> 00:41:43,990
Because O.J. Simpson is a super
famous football player,

1073
00:41:43,990 --> 00:41:46,390
there was huge debate about
whether he had murdered his

1074
00:41:46,390 --> 00:41:48,840
wife and another person
in the house.

1075
00:41:48,840 --> 00:41:50,290
And there was an exact
moment when they

1076
00:41:50,290 --> 00:41:52,030
said, get on your TVs.

1077
00:41:52,030 --> 00:41:55,000
Here comes a verdict.

1078
00:41:55,000 --> 00:41:56,250
Or the 9/11 attack.

1079
00:41:58,620 --> 00:42:01,230
So here's this funny
research business.

1080
00:42:01,230 --> 00:42:03,050
What people do is they go.

1081
00:42:03,050 --> 00:42:05,460
And as soon as these horrible
things happened, they ask

1082
00:42:05,460 --> 00:42:08,440
people to write down where were
you, what were you doing,

1083
00:42:08,440 --> 00:42:10,895
how did you learn about it, what
were your feelings, what

1084
00:42:10,895 --> 00:42:12,970
were the feelings of the people
around you, and what

1085
00:42:12,970 --> 00:42:15,540
did you do in the next hour?

1086
00:42:15,540 --> 00:42:17,430
You can't be prepared for
this other than wait

1087
00:42:17,430 --> 00:42:18,800
for something horrible.

1088
00:42:18,800 --> 00:42:21,050
Everybody says, I'll never
forget where I was when 9/11

1089
00:42:21,050 --> 00:42:22,160
happened or things like that.

1090
00:42:22,160 --> 00:42:23,860
People feel that way.

1091
00:42:23,860 --> 00:42:27,730
And they certainly remember that
day more in some sense

1092
00:42:27,730 --> 00:42:30,410
than a typical quiet day.

1093
00:42:30,410 --> 00:42:31,360
But here's what they do.

1094
00:42:31,360 --> 00:42:33,680
They test them on their
own answers.

1095
00:42:33,680 --> 00:42:35,800
So you gave the information
just a little bit after.

1096
00:42:35,800 --> 00:42:37,470
Presumably, it's roughly
correct.

1097
00:42:37,470 --> 00:42:40,760
If you're tested a year later,
yes, you remember that day

1098
00:42:40,760 --> 00:42:42,320
better than other days.

1099
00:42:42,320 --> 00:42:44,770
But it's full of errors.

1100
00:42:44,770 --> 00:42:46,520
It's full of errors.

1101
00:42:46,520 --> 00:42:48,470
And people are way
overconfident.

1102
00:42:48,470 --> 00:42:51,030
They say, I know where
I was a year later.

1103
00:42:51,030 --> 00:42:52,620
And I know I was with
my friends.

1104
00:42:52,620 --> 00:42:54,050
And you say, a year
ago, you wrote

1105
00:42:54,050 --> 00:42:55,110
down you were by yourself.

1106
00:42:55,110 --> 00:42:56,680
Oops.

1107
00:42:56,680 --> 00:42:59,530
But I'm pretty sure I was
wearing my raincoat.

1108
00:42:59,530 --> 00:43:01,430
And you go, oops, you
wrote down you

1109
00:43:01,430 --> 00:43:02,330
were wearing a sweater.

1110
00:43:02,330 --> 00:43:03,520
Oops.

1111
00:43:03,520 --> 00:43:05,240
Your memory is way higher.

1112
00:43:05,240 --> 00:43:07,990
But it's full of inaccuracies,
even though people

1113
00:43:07,990 --> 00:43:09,490
are sure it's not.

1114
00:43:09,490 --> 00:43:13,280
They're sure that's
not a inaccurate.

1115
00:43:13,280 --> 00:43:16,430
So the next horrible public
event that happens, there will

1116
00:43:16,430 --> 00:43:18,810
be a bunch of memory researchers
running around.

1117
00:43:18,810 --> 00:43:20,300
And they'll be saying,
where were you?

1118
00:43:20,300 --> 00:43:20,800
What were you doing?

1119
00:43:20,800 --> 00:43:21,520
What were you feeling?

1120
00:43:21,520 --> 00:43:24,150
Because we need that to be able
to verify whether your

1121
00:43:24,150 --> 00:43:27,060
memory for emotionally intense
experience is accurate or not

1122
00:43:27,060 --> 00:43:28,310
and how long it lasts.

1123
00:43:33,030 --> 00:43:35,770
A couple of experiments
and then a video.

1124
00:43:35,770 --> 00:43:37,970
So here's eyewitness
testimony.

1125
00:43:37,970 --> 00:43:41,320
There's nothing more dramatic in
court on the TV shows or in

1126
00:43:41,320 --> 00:43:43,470
real life, if you have to
be around these things.

1127
00:43:43,470 --> 00:43:45,500
Where the person says,
who was a person who

1128
00:43:45,500 --> 00:43:46,430
murdered your neighbor?

1129
00:43:46,430 --> 00:43:47,760
And you go, that's the woman.

1130
00:43:47,760 --> 00:43:48,720
That's the man.

1131
00:43:48,720 --> 00:43:50,990
And the jury's looking.

1132
00:43:50,990 --> 00:43:53,950
So how accurate is eyewitness
testimony?

1133
00:43:53,950 --> 00:43:55,490
Of course it has to be
somewhat accurate.

1134
00:43:55,490 --> 00:43:57,290
We're not living in a crazy
world where people are

1135
00:43:57,290 --> 00:43:59,160
randomly imagining things.

1136
00:43:59,160 --> 00:44:01,730
But how accurate is it really?

1137
00:44:01,730 --> 00:44:03,350
So here's a kind of an
experiment they did.

1138
00:44:03,350 --> 00:44:04,900
This is work from Elizabeth
Loftus who's a

1139
00:44:04,900 --> 00:44:06,130
leader in this area.

1140
00:44:06,130 --> 00:44:07,950
She would have people
just see slides--

1141
00:44:07,950 --> 00:44:10,170
this is before computers,
basically--

1142
00:44:10,170 --> 00:44:12,160
where a sports car would come
to an intersection, would

1143
00:44:12,160 --> 00:44:15,080
turn, and would hit a pedestrian
in the slides.

1144
00:44:15,080 --> 00:44:16,580
So it's not the drama
of really seeing it.

1145
00:44:16,580 --> 00:44:18,390
It's a stage thing.

1146
00:44:18,390 --> 00:44:23,100
And it would be either a yield
sign or a stop sign just

1147
00:44:23,100 --> 00:44:26,630
before the car turns that
you would clearly see.

1148
00:44:26,630 --> 00:44:28,360
And let's pretend you were in a
condition where you saw the

1149
00:44:28,360 --> 00:44:29,470
yield sign.

1150
00:44:29,470 --> 00:44:32,360
Then afterwards, they would say,
did another car pass the

1151
00:44:32,360 --> 00:44:34,510
red Datsun while it was stopped
at the yield sign?

1152
00:44:34,510 --> 00:44:35,880
That's consistent,
the world yield.

1153
00:44:35,880 --> 00:44:38,870
Or did another car pass the
Datsun while it was stopped at

1154
00:44:38,870 --> 00:44:39,650
the stop sign?

1155
00:44:39,650 --> 00:44:42,750
See you slipped in
the word stopped?

1156
00:44:42,750 --> 00:44:43,950
Or in neutral, you don't
say anything.

1157
00:44:43,950 --> 00:44:46,540
And now you say, which
slide did you see?

1158
00:44:46,540 --> 00:44:48,290
You show them the actual
slide with a stop

1159
00:44:48,290 --> 00:44:49,890
sign or a yield sign.

1160
00:44:49,890 --> 00:44:50,740
You saw the yield sign.

1161
00:44:50,740 --> 00:44:53,140
But here's what happens.

1162
00:44:53,140 --> 00:44:57,280
If you saw the yield sign, and
you are shown this yield sign,

1163
00:44:57,280 --> 00:44:59,440
you do pretty well,
consistent.

1164
00:44:59,440 --> 00:45:03,630
Look what happens if you
saw the yield sign.

1165
00:45:03,630 --> 00:45:07,040
You've got the question, did
you see a stop sign?

1166
00:45:07,040 --> 00:45:10,160
And now you're tested, what
did you actually see?

1167
00:45:10,160 --> 00:45:11,190
You're pretty much wrong.

1168
00:45:11,190 --> 00:45:13,820
You're pretty much saying,
I saw a stop sign.

1169
00:45:13,820 --> 00:45:15,280
That's retroactive
interference.

1170
00:45:15,280 --> 00:45:16,970
You saw the yield sign.

1171
00:45:16,970 --> 00:45:19,130
You were asked about
a stop sign.

1172
00:45:19,130 --> 00:45:22,150
And now you believe you
saw a stop sign.

1173
00:45:22,150 --> 00:45:25,040
That's why it's so important
how police or other people

1174
00:45:25,040 --> 00:45:27,500
interrogate witnesses.

1175
00:45:27,500 --> 00:45:30,920
Because if they slip in certain
words, people have a

1176
00:45:30,920 --> 00:45:32,630
hard time going back to
the original memory.

1177
00:45:32,630 --> 00:45:34,260
They'll have a memory that's
a mixture of the original

1178
00:45:34,260 --> 00:45:39,240
experience and the questions
asked about that experience.

1179
00:45:39,240 --> 00:45:42,310
So people say, oh that's just
a laboratory experiment.

1180
00:45:42,310 --> 00:45:44,140
In real life, that
wouldn't happen.

1181
00:45:44,140 --> 00:45:47,390
So they did this experiment.

1182
00:45:47,390 --> 00:45:49,600
Jim, a graduate student,
reminded his younger brother

1183
00:45:49,600 --> 00:45:54,450
Chris that they lost Chris in
a mall when Chris was five,

1184
00:45:54,450 --> 00:45:57,230
and an older tallish man
brought him back.

1185
00:45:57,230 --> 00:45:58,845
So that would be a pretty
traumatic thing for a

1186
00:45:58,845 --> 00:46:01,130
five-year-old to be lost in
a mall and have an old

1187
00:46:01,130 --> 00:46:03,660
man take you back.

1188
00:46:03,660 --> 00:46:05,410
Certainly for the parents,
you're going where's the kid?

1189
00:46:05,410 --> 00:46:05,900
Where's the kid?

1190
00:46:05,900 --> 00:46:11,400
So two days later, Chris says,
the younger kid, yeah.

1191
00:46:11,400 --> 00:46:13,130
I was with you guys
for a second.

1192
00:46:13,130 --> 00:46:15,750
And then I went over to look at
the toy store, the KB Toys.

1193
00:46:15,750 --> 00:46:16,490
And then we got lost.

1194
00:46:16,490 --> 00:46:17,170
And I was looking around.

1195
00:46:17,170 --> 00:46:19,180
I thought, I'm in trouble now.

1196
00:46:19,180 --> 00:46:20,100
And then I thought it
was never going to

1197
00:46:20,100 --> 00:46:20,960
see my family again.

1198
00:46:20,960 --> 00:46:22,150
I was really scared.

1199
00:46:22,150 --> 00:46:24,060
And this old man, I think he
was wearing a blue flannel,

1200
00:46:24,060 --> 00:46:24,660
came up to me.

1201
00:46:24,660 --> 00:46:25,250
He was kind of old.

1202
00:46:25,250 --> 00:46:26,500
He was kind of bald.

1203
00:46:29,380 --> 00:46:30,630
Chris was never lost.

1204
00:46:33,880 --> 00:46:36,510
Jim, the graduate student,
being mean to his younger

1205
00:46:36,510 --> 00:46:39,170
brother, but for good science
reasons, told

1206
00:46:39,170 --> 00:46:40,320
him this slight story.

1207
00:46:40,320 --> 00:46:42,310
He didn't give him any
of these details.

1208
00:46:42,310 --> 00:46:44,320
He just said remember when you
were five years old and got

1209
00:46:44,320 --> 00:46:48,230
lost in the mall, and the old
tall man brought you back?

1210
00:46:48,230 --> 00:46:51,560
But what's happening is our
memories of when we're kids

1211
00:46:51,560 --> 00:46:53,340
are pretty vague.

1212
00:46:53,340 --> 00:46:55,970
Chris is sort of mixing in some
real things that happened

1213
00:46:55,970 --> 00:46:57,290
in the story.

1214
00:46:57,290 --> 00:46:59,520
And he's fabricating.

1215
00:46:59,520 --> 00:47:00,870
He's not trying to get
away with anything.

1216
00:47:00,870 --> 00:47:02,350
He's just creating this
false memory.

1217
00:47:05,140 --> 00:47:05,850
And he said, oh no.

1218
00:47:05,850 --> 00:47:08,110
We want it even more
lifelike than that.

1219
00:47:08,110 --> 00:47:09,820
So here's the experiment
they did.

1220
00:47:09,820 --> 00:47:13,540
They had 120 students see
an ad for Bugs Bunny at

1221
00:47:13,540 --> 00:47:14,970
Disneyland to evaluate the ad.

1222
00:47:14,970 --> 00:47:15,840
So you're brought into a room.

1223
00:47:15,840 --> 00:47:19,090
They say, we want to market test
whether this is a good ad

1224
00:47:19,090 --> 00:47:20,340
for Disneyland.

1225
00:47:24,710 --> 00:47:27,100
All these people had been to
Disneyland some years ago with

1226
00:47:27,100 --> 00:47:29,050
kids in real life.

1227
00:47:29,050 --> 00:47:31,540
Then they said, by
the way we're

1228
00:47:31,540 --> 00:47:32,400
thinking about Bugs Bunny.

1229
00:47:32,400 --> 00:47:33,580
When you were there,
did you meet him?

1230
00:47:33,580 --> 00:47:36,060
Did you shake his hand?

1231
00:47:36,060 --> 00:47:37,960
Now, what's the big trick
about Bugs Bunny and

1232
00:47:37,960 --> 00:47:38,180
Disneyland?

1233
00:47:38,180 --> 00:47:40,820
I saw some smiles right away.

1234
00:47:40,820 --> 00:47:42,880
Bugs Bunny is one of the few
cartoon characters that many

1235
00:47:42,880 --> 00:47:46,050
kids know that's not a Disney
character and, by law, cannot

1236
00:47:46,050 --> 00:47:48,940
be found at Disneyland
or Disney World.

1237
00:47:48,940 --> 00:47:55,240
But one out of three college
students remembered seeing

1238
00:47:55,240 --> 00:47:58,150
Bugs Bunny at Disney World or
Disneyland when they were a

1239
00:47:58,150 --> 00:48:01,890
kid after they had just seen
some promotional material

1240
00:48:01,890 --> 00:48:03,710
showing Bugs Bunny
shaking hands.

1241
00:48:06,380 --> 00:48:07,660
Because they had been there.

1242
00:48:07,660 --> 00:48:08,710
They don't remember stuff.

1243
00:48:08,710 --> 00:48:09,020
OK.

1244
00:48:09,020 --> 00:48:12,130
Bugs Bunny, sure.

1245
00:48:12,130 --> 00:48:16,640
How potent is this way that we
mix up new information with

1246
00:48:16,640 --> 00:48:17,500
the original experience.

1247
00:48:17,500 --> 00:48:20,260
So we can't tell apart the
original memory from the mixed

1248
00:48:20,260 --> 00:48:22,350
up new relevant information.

1249
00:48:22,350 --> 00:48:24,450
I mean, this would not have
worked, of course, if they

1250
00:48:24,450 --> 00:48:25,960
show you King Kong.

1251
00:48:25,960 --> 00:48:27,590
You'd be like, there's no
way I saw King Kong.

1252
00:48:27,590 --> 00:48:29,680
It has to be slightly
credible.

1253
00:48:32,190 --> 00:48:34,740
But it's a pretty big difference
in some cases.

1254
00:48:34,740 --> 00:48:38,470
So here's one example
in real life.

1255
00:48:38,470 --> 00:48:39,220
This is a real case.

1256
00:48:39,220 --> 00:48:41,940
In a dark, October night, a
man picks up on Pacific

1257
00:48:41,940 --> 00:48:45,042
Highway, south of Seattle,
hitchhiker.

1258
00:48:45,042 --> 00:48:47,210
It turns into an
isolated road.

1259
00:48:47,210 --> 00:48:49,040
The man brutally rapes
her and leaves her by

1260
00:48:49,040 --> 00:48:49,670
the side of the road.

1261
00:48:49,670 --> 00:48:52,420
24 hours later, she looks at
an array of photos, and she

1262
00:48:52,420 --> 00:48:53,850
picks out a man who
was convicted.

1263
00:48:53,850 --> 00:48:57,020
Because she says, that's the
man who brutally raped me.

1264
00:48:57,020 --> 00:48:59,150
A few months later, another
man was arrested

1265
00:48:59,150 --> 00:49:00,110
for a series of rapes.

1266
00:49:00,110 --> 00:49:01,600
She sees his picture in
the newspaper, and

1267
00:49:01,600 --> 00:49:03,240
she realizes oops.

1268
00:49:03,240 --> 00:49:07,740
Sadly, this is the man who
attacked and raped me.

1269
00:49:07,740 --> 00:49:10,620
He's released from jail.

1270
00:49:10,620 --> 00:49:12,580
By then, he's lost his money,
his job, his fiance, his

1271
00:49:12,580 --> 00:49:12,885
reputation.

1272
00:49:12,885 --> 00:49:14,730
He spends four years
pursuing a lawsuit.

1273
00:49:14,730 --> 00:49:17,480
He dies 11 days prior
to the trial.

1274
00:49:17,480 --> 00:49:20,280
10 months later, his estate gets
$2.8 million because the

1275
00:49:20,280 --> 00:49:23,140
police weren't very careful how
they did the photo lineup.

1276
00:49:23,140 --> 00:49:26,150
It turns out if you look at the
photo lineup, his was a

1277
00:49:26,150 --> 00:49:29,950
slightly outlier picture
of the six pictures.

1278
00:49:29,950 --> 00:49:31,320
So the woman went
for that guy.

1279
00:49:31,320 --> 00:49:32,720
So when she's accusing
him in court, she's

1280
00:49:32,720 --> 00:49:35,220
not trying to lie.

1281
00:49:35,220 --> 00:49:37,920
what's she mixing up?

1282
00:49:37,920 --> 00:49:41,030
Her memory for the photo array
versus for memory for the

1283
00:49:41,030 --> 00:49:42,040
original event.

1284
00:49:42,040 --> 00:49:43,720
Those two have gotten
blended in the

1285
00:49:43,720 --> 00:49:44,635
punch bowl of her memory.

1286
00:49:44,635 --> 00:49:45,520
Does that make sense?

1287
00:49:45,520 --> 00:49:47,580
Perfectly with good intention.

1288
00:49:50,160 --> 00:49:53,690
And there's 80,000 trials
you estimate the rely on

1289
00:49:53,690 --> 00:49:54,710
eyewitness testimony.

1290
00:49:54,710 --> 00:49:57,120
And many, many of those
are mistaken.

1291
00:49:57,120 --> 00:49:59,190
It's just human nature to
mistake them, especially

1292
00:49:59,190 --> 00:50:01,830
because trials are not boom
one minute later.

1293
00:50:01,830 --> 00:50:05,600
There's lots of discussion,
lots of interviews with

1294
00:50:05,600 --> 00:50:07,610
policeman, or prosecutors,
or lawyers.

1295
00:50:07,610 --> 00:50:10,510
Lots of things happen from the
original crime moment until

1296
00:50:10,510 --> 00:50:11,990
that moment in court.

1297
00:50:11,990 --> 00:50:14,460
Every discussion, and thought,
and feeling you have from the

1298
00:50:14,460 --> 00:50:18,240
original experience blends in
with the original experience.

1299
00:50:18,240 --> 00:50:19,800
And it can be hard to
pull those apart.

1300
00:50:22,660 --> 00:50:26,050
Here's another sort
of weird example.

1301
00:50:26,050 --> 00:50:28,150
A woman at home in Australia,
she's brutally

1302
00:50:28,150 --> 00:50:28,980
raped in her home.

1303
00:50:28,980 --> 00:50:30,790
She identifies Donald Thompson--
now you'll see why

1304
00:50:30,790 --> 00:50:32,960
this is a bit ironic
and a weird story--

1305
00:50:32,960 --> 00:50:36,050
a renowned Australian
psychologist who studies

1306
00:50:36,050 --> 00:50:39,630
eyewitness memory problems.

1307
00:50:39,630 --> 00:50:42,970
It turns out he was on
television at the moment that

1308
00:50:42,970 --> 00:50:43,840
she was being raped.

1309
00:50:43,840 --> 00:50:45,340
And, obviously, she's in a very

1310
00:50:45,340 --> 00:50:46,840
disturbed difficult moment.

1311
00:50:46,840 --> 00:50:50,590
And so she conflates the face on
the TV with the face of the

1312
00:50:50,590 --> 00:50:53,460
actual horrible perpetrator.

1313
00:50:53,460 --> 00:50:54,530
So he has a good alibi.

1314
00:50:54,530 --> 00:50:55,410
He's on TV somewhere.

1315
00:50:55,410 --> 00:50:59,410
But in a perfectly good way,
she's trying to be correct.

1316
00:50:59,410 --> 00:51:02,720
These things get jumbled
together.

1317
00:51:02,720 --> 00:51:04,240
So you're going to see
it in one minute, a

1318
00:51:04,240 --> 00:51:06,000
film of Susan Nason.

1319
00:51:06,000 --> 00:51:08,990
This is an amazing story.

1320
00:51:08,990 --> 00:51:09,320
Sorry.

1321
00:51:09,320 --> 00:51:10,630
You'll see the film of
Eileen Franklin.

1322
00:51:10,630 --> 00:51:12,090
But Susan Nason was an
eight-year-old girl.

1323
00:51:12,090 --> 00:51:16,626
She was missing and found
murdered in October 31, 1969.

1324
00:51:16,626 --> 00:51:20,820
20 years later, Eileen
Franklin-- you'll see later--

1325
00:51:20,820 --> 00:51:23,940
remembers that her father
murdered Susan Nason.

1326
00:51:23,940 --> 00:51:25,790
She testifies against him,
and he's found guilty.

1327
00:51:25,790 --> 00:51:26,870
And he goes to jail.

1328
00:51:26,870 --> 00:51:28,670
He's been subsequently
released.

1329
00:51:28,670 --> 00:51:30,720
Because she says I was
with him when he

1330
00:51:30,720 --> 00:51:33,370
murdered my little friend.

1331
00:51:33,370 --> 00:51:35,950
But she doesn't remember
continuously.

1332
00:51:35,950 --> 00:51:38,990
She remembers it because a
therapist has had her think

1333
00:51:38,990 --> 00:51:40,140
about her past.

1334
00:51:40,140 --> 00:51:42,560
And all of a sudden, what she
feels like was a repressed

1335
00:51:42,560 --> 00:51:44,160
memory comes back.

1336
00:51:44,160 --> 00:51:47,720
So we're going to show a
movie about this topic.

1337
00:51:47,720 --> 00:51:50,090
It's your next paper topic.

1338
00:51:50,090 --> 00:51:51,560
What's the status of memory?

1339
00:51:51,560 --> 00:51:54,640
So let's separate two things,
because it's really important.

1340
00:51:54,640 --> 00:52:00,930
Cases where women or men--
more often women--

1341
00:52:00,930 --> 00:52:05,920
remember being abused, sexually
abused, remember it

1342
00:52:05,920 --> 00:52:06,580
all the way through.

1343
00:52:06,580 --> 00:52:07,970
That may not like to
think about it.

1344
00:52:07,970 --> 00:52:09,970
But they remember it all
the way through.

1345
00:52:09,970 --> 00:52:11,520
That's not what we're
talking about.

1346
00:52:11,520 --> 00:52:15,030
What we're talking about are the
rare cases where a man or

1347
00:52:15,030 --> 00:52:17,100
a woman says, I haven't
thought about this.

1348
00:52:17,100 --> 00:52:19,510
I feel like I've repressed
this for 20,

1349
00:52:19,510 --> 00:52:21,710
or 30, or 40 years.

1350
00:52:21,710 --> 00:52:25,250
And now, more often than not
because of working with maybe

1351
00:52:25,250 --> 00:52:27,550
a psychologist or clinical
psychologist to help me with

1352
00:52:27,550 --> 00:52:32,540
problems I have, I recover
my repressed memory.

1353
00:52:32,540 --> 00:52:33,780
And it's a brutal thing.

1354
00:52:33,780 --> 00:52:36,980
Because you can imagine on the
one hand, you have a young

1355
00:52:36,980 --> 00:52:40,410
woman accusing her father, or
uncle, or the father we'll

1356
00:52:40,410 --> 00:52:44,020
say, of doing something that we
think is terrible, which is

1357
00:52:44,020 --> 00:52:47,830
sexually abusing her, the
biggest failure of trust you

1358
00:52:47,830 --> 00:52:51,710
can have in a father,
a huge violation.

1359
00:52:51,710 --> 00:52:54,200
So he's horrible
if she's right.

1360
00:52:54,200 --> 00:52:56,260
But if her memory's inaccurate,
what's more

1361
00:52:56,260 --> 00:52:58,650
horrible than an innocent man
being accused by his own

1362
00:52:58,650 --> 00:53:02,310
daughter of sexually
abusing her because

1363
00:53:02,310 --> 00:53:04,080
her memory's incorrect.

1364
00:53:04,080 --> 00:53:06,380
So what we'll show you
two films briefly.

1365
00:53:18,390 --> 00:53:19,670
So the father went to jail.

1366
00:53:22,350 --> 00:53:26,380
The lawyers who were defending
the father wanted to tell the

1367
00:53:26,380 --> 00:53:29,250
jurors the following thing,
which in every account you

1368
00:53:29,250 --> 00:53:30,500
read is accurate.

1369
00:53:33,320 --> 00:53:36,430
That everything that she
described about the death of

1370
00:53:36,430 --> 00:53:40,110
that girl, Susan Nason, every
detail she described was in

1371
00:53:40,110 --> 00:53:42,830
the newspaper accounts
of the time.

1372
00:53:42,830 --> 00:53:48,330
And that no detail that she
provided was verifiable and

1373
00:53:48,330 --> 00:53:50,860
different than newspapers had
reported at the time.

1374
00:53:54,080 --> 00:53:56,100
One of the things that people
like to do, investigators, is

1375
00:53:56,100 --> 00:53:59,555
have the person know something
that wasn't publicly released.

1376
00:53:59,555 --> 00:54:02,530
And everything that she spoke
about in regards to the little

1377
00:54:02,530 --> 00:54:05,450
girl's murder, Susan Nason's
was publicly available

1378
00:54:05,450 --> 00:54:08,400
information that had been on
television or in newspapers.

1379
00:54:08,400 --> 00:54:10,820
And the judge originally
ruled that the jurors

1380
00:54:10,820 --> 00:54:13,190
could not be told that.

1381
00:54:13,190 --> 00:54:15,890
And then later on, they
retried the father.

1382
00:54:15,890 --> 00:54:17,190
And the jurors were told that.

1383
00:54:17,190 --> 00:54:20,770
And the Father was
let go from jail.

1384
00:54:20,770 --> 00:54:22,450
So there was this huge
pendulum swing.

1385
00:54:22,450 --> 00:54:24,520
Because, originally, as these
repressed memory cases were

1386
00:54:24,520 --> 00:54:27,450
brought, people wondered, well,
why would the woman lie

1387
00:54:27,450 --> 00:54:28,750
about such a horrible thing?

1388
00:54:28,750 --> 00:54:29,880
It's so horrible.

1389
00:54:29,880 --> 00:54:31,960
Of course she must be
telling the truth.

1390
00:54:31,960 --> 00:54:36,040
And of course the father must
be lying to protect himself.

1391
00:54:36,040 --> 00:54:38,570
But now, especially in the
context of repressed

1392
00:54:38,570 --> 00:54:40,190
memories-- again, very different
than continuing

1393
00:54:40,190 --> 00:54:43,140
memories, the ones that are
recovered after decades--

1394
00:54:43,140 --> 00:54:45,665
there's a suspicion that some
of them might be inaccurate.

1395
00:54:48,636 --> 00:54:52,250
And some psychologists working
with some patients might be

1396
00:54:52,250 --> 00:54:54,810
having patients think about
possible sources of their

1397
00:54:54,810 --> 00:54:55,850
difficulty.

1398
00:54:55,850 --> 00:54:58,560
And these memories get
falsely constructed.

1399
00:54:58,560 --> 00:55:00,670
And the hard part, and the part
that you'll write your

1400
00:55:00,670 --> 00:55:02,650
paper about, is knowing
which is which.

1401
00:55:02,650 --> 00:55:04,600
Because it's heartrending.

1402
00:55:04,600 --> 00:55:06,260
It's heartrending to have
a woman who was

1403
00:55:06,260 --> 00:55:09,600
abused be not believed.

1404
00:55:09,600 --> 00:55:12,040
It's heartrending for a father
who was not an abuser to be

1405
00:55:12,040 --> 00:55:14,190
accused of being an abuser.

1406
00:55:14,190 --> 00:55:16,570
And it's incredibly hard to
have direct evidence about

1407
00:55:16,570 --> 00:55:19,740
who's telling the truth and
who's not in these kinds of

1408
00:55:19,740 --> 00:55:25,000
cases that are so far from when
the crime occurred or

1409
00:55:25,000 --> 00:55:28,710
might have occurred, and where
there's such a long period of

1410
00:55:28,710 --> 00:55:30,150
nondiscussion about it
because it's been

1411
00:55:30,150 --> 00:55:31,140
repressed for decades.

1412
00:55:31,140 --> 00:55:32,110
So you'll look at that.

1413
00:55:32,110 --> 00:55:33,370
It's a big task.

1414
00:55:33,370 --> 00:55:34,460
It takes you into courts.

1415
00:55:34,460 --> 00:55:37,760
It takes you into clinical
complexities of people.