1
00:00:00,530 --> 00:00:02,960
The following content is
provided under a Creative

2
00:00:02,960 --> 00:00:04,370
Commons license.

3
00:00:04,370 --> 00:00:07,410
Your support will help MIT
OpenCourseWare continue to

4
00:00:07,410 --> 00:00:11,060
offer high quality educational
resources for free.

5
00:00:11,060 --> 00:00:13,960
To make a donation or view
additional materials from

6
00:00:13,960 --> 00:00:19,790
hundreds of MIT courses, visit
MIT OpenCourseWare at

7
00:00:19,790 --> 00:00:21,040
ocw.mit.edu.

8
00:00:34,700 --> 00:00:36,110
PROFESSOR: Good afternoon.

9
00:00:36,110 --> 00:00:38,780
Today we're going to go
beyond the way we've

10
00:00:38,780 --> 00:00:39,480
spoken about before.

11
00:00:39,480 --> 00:00:41,180
We've talked about how
psychologists measure the

12
00:00:41,180 --> 00:00:44,690
mind, how neuroscientists
interrogate the brain.

13
00:00:44,690 --> 00:00:47,080
And so now, we'll talk about the
first human faculty that

14
00:00:47,080 --> 00:00:48,790
we'll consider in a series
of faculties

15
00:00:48,790 --> 00:00:50,460
through the course, Vision--

16
00:00:50,460 --> 00:00:52,450
How We See.

17
00:00:52,450 --> 00:00:55,015
And for today's lecture, I got
help from Melissa Troyer, one

18
00:00:55,015 --> 00:00:56,790
of your teaching assistants.

19
00:00:56,790 --> 00:00:59,060
So, one of the things we can
think about is what are the

20
00:00:59,060 --> 00:01:00,180
purposes of vision?

21
00:01:00,180 --> 00:01:03,080
What do we use sight for?

22
00:01:03,080 --> 00:01:04,440
And we'll talk about that.

23
00:01:04,440 --> 00:01:06,695
Problems that the visual
system has to overcome.

24
00:01:06,695 --> 00:01:09,530
What are the obstacles for
vision to work accurately and

25
00:01:09,530 --> 00:01:10,700
efficiently?

26
00:01:10,700 --> 00:01:13,130
And how does the brain
organization of vision serve

27
00:01:13,130 --> 00:01:17,160
the human capacity to see?

28
00:01:17,160 --> 00:01:20,740
So human perceptual abilities
are amazing, as are perceptual

29
00:01:20,740 --> 00:01:22,680
abilities of many species
in different areas.

30
00:01:22,680 --> 00:01:26,450
But humans can detect a candle
30 miles away on

31
00:01:26,450 --> 00:01:28,190
a dark, clear night.

32
00:01:28,190 --> 00:01:30,120
They can detect cochlear
displacements

33
00:01:30,120 --> 00:01:31,400
that's in your ear--

34
00:01:31,400 --> 00:01:34,500
equal to the width of
a hydrogen atom.

35
00:01:34,500 --> 00:01:37,240
You can taste 1 teaspoon of
sugar, even when it's mixed in

36
00:01:37,240 --> 00:01:39,200
two gallons of water.

37
00:01:39,200 --> 00:01:41,900
You can smell a drop of perfume
diffused into the

38
00:01:41,900 --> 00:01:44,200
space of a three-bedroom
apartment.

39
00:01:44,200 --> 00:01:47,260
So the human capacity to
perceive the world in a highly

40
00:01:47,260 --> 00:01:49,750
sensitive way is remarkable.

41
00:01:49,750 --> 00:01:52,130
We're going to focus on vision
specifically today.

42
00:01:52,130 --> 00:01:55,110
And think, what do we
use vision for?

43
00:01:55,110 --> 00:01:59,030
Well, how does it serve us in
dealing with the world?

44
00:01:59,030 --> 00:02:02,680
And researchers have thought
about two major reasons that

45
00:02:02,680 --> 00:02:04,350
we see, two purposes
of vision.

46
00:02:04,350 --> 00:02:06,800
One of them is object
recognition.

47
00:02:06,800 --> 00:02:08,550
Recognizing things
in the world.

48
00:02:08,550 --> 00:02:11,200
People we know, words
we can read,

49
00:02:11,200 --> 00:02:13,750
chairs, tables, animals--

50
00:02:13,750 --> 00:02:15,300
all kinds of things that
we see out there.

51
00:02:15,300 --> 00:02:18,190
And we need to know who they are
or what they are in order

52
00:02:18,190 --> 00:02:20,040
to operate in the world.

53
00:02:20,040 --> 00:02:23,050
And the second one is a little
bit different, navigation--

54
00:02:23,050 --> 00:02:24,020
getting around in the world.

55
00:02:24,020 --> 00:02:25,590
When you're running, when you're
jumping, when you're

56
00:02:25,590 --> 00:02:28,440
reaching to grab something,
you're not usually much

57
00:02:28,440 --> 00:02:31,470
interested in what something
is, as where you're moving,

58
00:02:31,470 --> 00:02:33,310
where you need to move
to avoid something

59
00:02:33,310 --> 00:02:34,510
being thrown at you.

60
00:02:34,510 --> 00:02:36,160
So two big purposes
of vision--

61
00:02:36,160 --> 00:02:38,930
what things are and where they
are, so you can navigate

62
00:02:38,930 --> 00:02:41,540
around quickly in the world.

63
00:02:41,540 --> 00:02:44,570
And when we think about purposes
of vision, for what

64
00:02:44,570 --> 00:02:48,720
things are or where they are,
psychologists and people

65
00:02:48,720 --> 00:02:51,040
working in computer vision have
recognized a series of

66
00:02:51,040 --> 00:02:54,360
problems that our visual
system has to solve.

67
00:02:54,360 --> 00:02:55,990
We have to link unique images.

68
00:02:55,990 --> 00:02:58,310
Because when we see something,
it'll almost always look

69
00:02:58,310 --> 00:03:01,500
different from one perceptual
moment to the next.

70
00:03:01,500 --> 00:03:04,000
We can see the same person,
for example, but under

71
00:03:04,000 --> 00:03:06,180
different conditions of
illumination, darkness or

72
00:03:06,180 --> 00:03:08,440
lightness or shadows, the angle
we see them at, the

73
00:03:08,440 --> 00:03:11,160
distance we see them at, the
expression on their face-- a

74
00:03:11,160 --> 00:03:13,140
smile or frown--

75
00:03:13,140 --> 00:03:15,690
shadows that might be being
cast, occlusion, a thing

76
00:03:15,690 --> 00:03:17,380
blocking part of their face.

77
00:03:17,380 --> 00:03:19,760
And yet, we can recognize a
person very well under all

78
00:03:19,760 --> 00:03:20,550
these circumstances.

79
00:03:20,550 --> 00:03:23,630
So what arrives in our
eye is different from

80
00:03:23,630 --> 00:03:25,150
all of these views.

81
00:03:25,150 --> 00:03:27,600
But what we interpret them
is all the same.

82
00:03:27,600 --> 00:03:29,250
That's a person I know.

83
00:03:29,250 --> 00:03:30,230
The same thing with letters.

84
00:03:30,230 --> 00:03:32,550
We can see letters in all kinds
of fonts and all kinds

85
00:03:32,550 --> 00:03:35,420
of handwriting, and yet we can
interpret what they are in a

86
00:03:35,420 --> 00:03:36,670
uniform way.

87
00:03:36,670 --> 00:03:37,440
Or body.

88
00:03:37,440 --> 00:03:39,810
We can see bodies of all kinds
of different positions.

89
00:03:39,810 --> 00:03:41,440
But our visual system
lets us know right

90
00:03:41,440 --> 00:03:43,210
away that it's a person.

91
00:03:43,210 --> 00:03:45,530
And so psychologists have tried
to formalize these kinds

92
00:03:45,530 --> 00:03:47,990
of problems to ones
of equivalents.

93
00:03:47,990 --> 00:03:50,550
So something is the same thing
at different views.

94
00:03:50,550 --> 00:03:51,440
Generalization--

95
00:03:51,440 --> 00:03:53,770
something has a very
different shape.

96
00:03:53,770 --> 00:03:56,320
Or impoverished input, where
there's poor lighting or

97
00:03:56,320 --> 00:03:57,690
something's complicated.

98
00:03:57,690 --> 00:04:00,210
And you still figure
out what it is.

99
00:04:00,210 --> 00:04:02,430
For getting around in the world
or where things are in

100
00:04:02,430 --> 00:04:04,640
the world, you're moving
through space

101
00:04:04,640 --> 00:04:05,420
and tracking things.

102
00:04:05,420 --> 00:04:07,310
And there's many sources
of movement.

103
00:04:07,310 --> 00:04:08,630
Your eyes are constantly
moving.

104
00:04:08,630 --> 00:04:09,440
Your head is moving.

105
00:04:09,440 --> 00:04:10,670
Your body is moving.

106
00:04:10,670 --> 00:04:11,880
And the world is moving.

107
00:04:11,880 --> 00:04:13,890
And you have to sort
all those out.

108
00:04:13,890 --> 00:04:14,460
Run quickly.

109
00:04:14,460 --> 00:04:17,380
Jump [INAUDIBLE]

110
00:04:17,380 --> 00:04:19,370
world spatially.

111
00:04:19,370 --> 00:04:20,399
And so here's some examples.

112
00:04:20,399 --> 00:04:23,490
Here's a shape constancy
problem.

113
00:04:23,490 --> 00:04:25,020
Here's two different
versions of a cat.

114
00:04:25,020 --> 00:04:27,320
It's trivial for you to know
that even though these things

115
00:04:27,320 --> 00:04:29,790
have different shapes, they
represent the same thing.

116
00:04:29,790 --> 00:04:32,310
But that's an achievement of
your visual system, to easily

117
00:04:32,310 --> 00:04:35,690
make that equivalence across
two very different shapes.

118
00:04:35,690 --> 00:04:38,430
And we know, from computer
vision, how hard these

119
00:04:38,430 --> 00:04:40,580
problems are.

120
00:04:40,580 --> 00:04:43,780
No computer vision system sees
with anything like the

121
00:04:43,780 --> 00:04:47,110
generalized ability of
a very young child.

122
00:04:47,110 --> 00:04:50,820
So our visual systems are so
brilliant that psychologists

123
00:04:50,820 --> 00:04:53,990
and vision scientists really
interrogate our visual system

124
00:04:53,990 --> 00:04:56,060
to figure out, how
do we do it.

125
00:04:56,060 --> 00:04:58,410
If people didn't do this, we'd
think it would be impossible

126
00:04:58,410 --> 00:05:01,250
to accomplish at all.

127
00:05:01,250 --> 00:05:03,640
Here's another example that's
easily solved by people-- not

128
00:05:03,640 --> 00:05:05,940
so easily solved by machines.

129
00:05:05,940 --> 00:05:09,080
If you see something like a car
at different angles, we

130
00:05:09,080 --> 00:05:11,790
know that's the same thing,
pretty easily, even though,

131
00:05:11,790 --> 00:05:14,800
again, the information that
arrives to our eyes is

132
00:05:14,800 --> 00:05:16,710
radically different.

133
00:05:16,710 --> 00:05:19,170
And the same thing can be said
of something like all these

134
00:05:19,170 --> 00:05:21,500
different versions
of the letter C.

135
00:05:21,500 --> 00:05:23,220
Radically different shapes.

136
00:05:23,220 --> 00:05:25,550
But we can readily interpret
almost all them-- this one

137
00:05:25,550 --> 00:05:26,740
won't be too tough--

138
00:05:26,740 --> 00:05:28,410
as the letter C.

139
00:05:28,410 --> 00:05:31,260
So our visual system does
this brilliantly.

140
00:05:31,260 --> 00:05:34,140
And also you get very
impoverished views.

141
00:05:34,140 --> 00:05:36,740
A key might be covered by
something, clothes on a hook.

142
00:05:36,740 --> 00:05:38,430
There might be a
book over here.

143
00:05:38,430 --> 00:05:41,240
There might be a fire hydrant
over here or a person.

144
00:05:41,240 --> 00:05:44,160
And yet, even though you get
only partial information, it's

145
00:05:44,160 --> 00:05:46,790
relatively trivial to discover,
for you, what it is

146
00:05:46,790 --> 00:05:48,820
you're looking at.

147
00:05:48,820 --> 00:05:51,770
Or you can see things like
you've never seen before.

148
00:05:51,770 --> 00:05:56,900
The pink elephant, this odd
picture, this striped ...

149
00:05:56,900 --> 00:06:01,960
you've never seen them before
but it's not really ...

150
00:06:01,960 --> 00:06:04,610
So again, your visual system
is just so brilliant at

151
00:06:04,610 --> 00:06:08,110
matching what it sees with what
you know to figure out

152
00:06:08,110 --> 00:06:11,070
what's in front of you.

153
00:06:11,070 --> 00:06:14,960
So today we're going to talk
about how our minds and brains

154
00:06:14,960 --> 00:06:16,730
allow us to see the
world vision.

155
00:06:16,730 --> 00:06:19,920
And we'll talk about it in three
parts; how philosophers

156
00:06:19,920 --> 00:06:23,620
and psychologists have thought
about the idea of seeing; how

157
00:06:23,620 --> 00:06:27,320
when information first enters
your eye-- your retina--

158
00:06:27,320 --> 00:06:31,190
which are set of layers in the
back of the eye; how that

159
00:06:31,190 --> 00:06:34,110
retina's already organized to
help you see the world; and

160
00:06:34,110 --> 00:06:37,940
how that information is then
transported into the cortex,

161
00:06:37,940 --> 00:06:40,930
where higher-level processing
lets you know where you are

162
00:06:40,930 --> 00:06:43,480
and what you're looking at.

163
00:06:43,480 --> 00:06:46,760
So the ideas about how we learn
to see the world have

164
00:06:46,760 --> 00:06:47,840
their roots in philosophy.

165
00:06:47,840 --> 00:06:51,690
For example, John Locke is
associated with the idea of

166
00:06:51,690 --> 00:06:54,110
the blank slate or
tabula rosa.

167
00:06:54,110 --> 00:06:57,350
He said, "Let us suppose the
mind to be a white paper void

168
00:06:57,350 --> 00:06:59,640
of any characters, without
any ideas.

169
00:06:59,640 --> 00:07:01,360
How comes it to be furnished?

170
00:07:01,360 --> 00:07:04,050
Whence comes it by that vast
store, which the busy and

171
00:07:04,050 --> 00:07:06,920
boundless fancy of man has
painted on it with an almost

172
00:07:06,920 --> 00:07:09,960
endless variety?" Where do we
get the information for how

173
00:07:09,960 --> 00:07:11,780
the world works when
we look at it?

174
00:07:11,780 --> 00:07:13,810
And his answer to that
was experience.

175
00:07:13,810 --> 00:07:17,120
That we notice things around us,
and experience drives our

176
00:07:17,120 --> 00:07:18,940
understanding of the world.

177
00:07:18,940 --> 00:07:21,870
And that position that there's
a blank slate and we learn

178
00:07:21,870 --> 00:07:24,420
objectively from the world
around us has been called the

179
00:07:24,420 --> 00:07:25,870
Objectivist view.

180
00:07:25,870 --> 00:07:28,350
And maybe, that we see the world
that we build up through

181
00:07:28,350 --> 00:07:31,410
experience, knowledge about how
to accurately represent

182
00:07:31,410 --> 00:07:33,190
what's around us.

183
00:07:33,190 --> 00:07:36,330
The alternative you has
emphasized how much our mind

184
00:07:36,330 --> 00:07:39,640
organizes what we see, instead
of what's out there itself.

185
00:07:39,640 --> 00:07:42,390
And so Gestalt psychologists or
Gestalt views on perception

186
00:07:42,390 --> 00:07:44,820
say, it's not what's out there
in the world, but the way our

187
00:07:44,820 --> 00:07:48,710
mind organizes and believes
what's out there.

188
00:07:48,710 --> 00:07:50,670
And these two things
can come together.

189
00:07:50,670 --> 00:07:52,650
So for example, when you think
about how you listen to a

190
00:07:52,650 --> 00:07:57,420
piano, music comes from
piano the same way a

191
00:07:57,420 --> 00:07:58,880
pianist strikes chords.

192
00:07:58,880 --> 00:08:01,600
There's things out there in
the world that you see.

193
00:08:01,600 --> 00:08:03,900
The piano has certain
limitations.

194
00:08:03,900 --> 00:08:05,570
It can sound like somethings
but not others.

195
00:08:05,570 --> 00:08:06,590
It can't be a clarinet.

196
00:08:06,590 --> 00:08:09,240
There's constraints in the world
on what's possible to

197
00:08:09,240 --> 00:08:11,740
see what's out there.

198
00:08:11,740 --> 00:08:14,190
And our percepts are evoked
by nature in this way.

199
00:08:14,190 --> 00:08:16,420
But they are personal and
not a copy of nature.

200
00:08:16,420 --> 00:08:19,010
In the words, we each,
individually, figure out

201
00:08:19,010 --> 00:08:20,950
what's going on there by
our own experiences and

202
00:08:20,950 --> 00:08:22,450
interpretation.

203
00:08:22,450 --> 00:08:25,350
And they're exactly that-- an
interpretation, not a simple

204
00:08:25,350 --> 00:08:28,770
mirroring or copy of nature.

205
00:08:28,770 --> 00:08:30,560
So I'm going to try and convince
you of something that

206
00:08:30,560 --> 00:08:33,190
might not be obvious to people
when they first think about

207
00:08:33,190 --> 00:08:35,470
it, that vision is an
interpretation of the

208
00:08:35,470 --> 00:08:36,289
world around us.

209
00:08:36,289 --> 00:08:39,940
It's not simply a video camera
or mirror, it's an active

210
00:08:39,940 --> 00:08:41,630
calculation of what's
out there.

211
00:08:41,630 --> 00:08:45,760
And we see, through inference,
what we believe that we see.

212
00:08:45,760 --> 00:08:46,930
And that visual illusions--

213
00:08:46,930 --> 00:08:49,180
and you've seen some before, and
I'll show you some again--

214
00:08:49,180 --> 00:08:52,810
are not just illusions to trick
people and confuse them.

215
00:08:52,810 --> 00:08:55,850
But they're demonstrations of a
gap between what's actually

216
00:08:55,850 --> 00:08:59,490
out there in the world and
how we interpret it.

217
00:08:59,490 --> 00:09:01,480
Most of the time we don't
have illusions.

218
00:09:01,480 --> 00:09:02,770
We don't walk into walls.

219
00:09:02,770 --> 00:09:06,030
We don't misidentify people
we know commonly.

220
00:09:06,030 --> 00:09:08,170
Because illusions are rare,
because our minds and brains

221
00:09:08,170 --> 00:09:11,110
have evolved to interpret the
world that's out there, and

222
00:09:11,110 --> 00:09:14,310
the system works so brilliantly
coupled with the

223
00:09:14,310 --> 00:09:15,970
environment that we actually
live in, that we

224
00:09:15,970 --> 00:09:16,840
don't think about it.

225
00:09:16,840 --> 00:09:18,690
So the person on the street
says, well, what

226
00:09:18,690 --> 00:09:20,500
do I need to see?

227
00:09:20,500 --> 00:09:22,250
I open my eyes.

228
00:09:22,250 --> 00:09:25,200
But that's because our visual
system works so brilliantly,

229
00:09:25,200 --> 00:09:27,000
effortlessly, and
automatically.

230
00:09:27,000 --> 00:09:29,410
It's such a calculating machine
that that's why it

231
00:09:29,410 --> 00:09:31,820
feels so simple.

232
00:09:31,820 --> 00:09:34,820
So here's one, small example.

233
00:09:34,820 --> 00:09:37,960
And I'll show you a number of
visual illusions that show us

234
00:09:37,960 --> 00:09:39,960
how we're constantly
interpreting even the most

235
00:09:39,960 --> 00:09:41,060
simple thing.

236
00:09:41,060 --> 00:09:43,060
So for example, here's
two people.

237
00:09:43,060 --> 00:09:45,390
And we're not worried that this
is one of the shortest

238
00:09:45,390 --> 00:09:48,220
men you've ever seen, because we
know, through prospective,

239
00:09:48,220 --> 00:09:50,260
this person's standing
back there.

240
00:09:50,260 --> 00:09:52,940
Even though, if they were
standing right next to

241
00:09:52,940 --> 00:09:55,710
somebody, you'd worry about
the size of that person.

242
00:09:55,710 --> 00:09:57,910
So we're constantly interpreting
information by

243
00:09:57,910 --> 00:10:00,660
the context in which
we see it.

244
00:10:00,660 --> 00:10:03,800
Here's another example of the
choice of interpretations.

245
00:10:03,800 --> 00:10:04,900
You can see a vase.

246
00:10:04,900 --> 00:10:07,580
Or you can see two faces here. ,
Usually it's hard for people

247
00:10:07,580 --> 00:10:08,930
to see them both
simultaneously.

248
00:10:08,930 --> 00:10:10,580
They tend to flip
back and forth.

249
00:10:10,580 --> 00:10:11,390
But you can choose the

250
00:10:11,390 --> 00:10:14,620
interpretation of what you see.

251
00:10:14,620 --> 00:10:17,190
So let me pick one example where
people have studied a

252
00:10:17,190 --> 00:10:20,390
lot, which is the issue of
brightness constancy.

253
00:10:20,390 --> 00:10:23,140
So here's the problem for
the visual system.

254
00:10:23,140 --> 00:10:26,380
The ambient brightness, the
brightness around us, varies

255
00:10:26,380 --> 00:10:28,290
in staggering ratios.

256
00:10:28,290 --> 00:10:30,570
When we're outdoors in the
sun-- and even that's not

257
00:10:30,570 --> 00:10:31,110
constant, right?

258
00:10:31,110 --> 00:10:32,150
Sometimes it's a bright day.

259
00:10:32,150 --> 00:10:33,860
Sometimes it's a cloudy day.

260
00:10:33,860 --> 00:10:36,670
Sometimes we're indoors under
bright conditions or less

261
00:10:36,670 --> 00:10:37,300
bright conditions.

262
00:10:37,300 --> 00:10:40,330
So huge changes in the
ambient illumination.

263
00:10:40,330 --> 00:10:43,520
Or in shadows-- we'll make
it even more complicated.

264
00:10:43,520 --> 00:10:45,790
That means in a purely
measurement sense, in a purely

265
00:10:45,790 --> 00:10:49,810
physics sense, a piece of coal
in sunlight reflects 10 times

266
00:10:49,810 --> 00:10:51,570
as much light as snow
in the shade.

267
00:10:51,570 --> 00:10:53,410
So think about that
for a moment.

268
00:10:53,410 --> 00:10:55,030
You're not disturbed if
you see a piece of

269
00:10:55,030 --> 00:10:56,400
coal inside or outside.

270
00:10:56,400 --> 00:10:59,730
They both look, to you, like
black, dark pieces of coal.

271
00:10:59,730 --> 00:11:02,270
And snow looks, pretty much, the
same to you if it's inside

272
00:11:02,270 --> 00:11:03,500
or outside.

273
00:11:03,500 --> 00:11:05,990
But if you're visual system
worked like a simple,

274
00:11:05,990 --> 00:11:10,100
objective light meter, you would
conclude that the piece

275
00:11:10,100 --> 00:11:13,270
of coal in the sunlight is a
much brighter thing-- in an

276
00:11:13,270 --> 00:11:14,490
objective sense--

277
00:11:14,490 --> 00:11:16,200
than snow in the shade.

278
00:11:16,200 --> 00:11:18,770
And yet, you never make that
conclusion, because you're

279
00:11:18,770 --> 00:11:22,210
constantly, automatically,
without effort, adjusting

280
00:11:22,210 --> 00:11:25,350
mentally for the ambient
light condition.

281
00:11:25,350 --> 00:11:27,800
So we use brightness all the
time to help us figure out is

282
00:11:27,800 --> 00:11:29,750
that a piece of coal
or sunlight?

283
00:11:29,750 --> 00:11:33,410
But that brightness perception
is constantly adjusted to

284
00:11:33,410 --> 00:11:36,430
interpret brightness in the
context we're sitting in.

285
00:11:36,430 --> 00:11:41,000
And perceptual psychologists
have worked on how people

286
00:11:41,000 --> 00:11:42,020
calculate this.

287
00:11:42,020 --> 00:11:46,040
So here's an example of a dark
letter T. Again, the black T

288
00:11:46,040 --> 00:11:48,460
would be 10 times brighter
outdoors than

289
00:11:48,460 --> 00:11:52,130
white paper is indoors.

290
00:11:52,130 --> 00:11:53,600
And yet, you're never
bothered by that.

291
00:11:53,600 --> 00:11:56,500
You can read a newspaper or
something else indoors or

292
00:11:56,500 --> 00:11:58,980
outdoors with equal these,
because people are calculating

293
00:11:58,980 --> 00:12:00,890
the ratio of the ambient
lighting--

294
00:12:00,890 --> 00:12:04,010
lower inside, much brighter
outside, from the sun--

295
00:12:04,010 --> 00:12:06,780
and using that information to
make inferences about the

296
00:12:06,780 --> 00:12:10,630
relative brightness of a
stimulus they're looking at.

297
00:12:10,630 --> 00:12:13,870
And there's all kinds of
examples or illusions that

298
00:12:13,870 --> 00:12:15,130
show us this.

299
00:12:15,130 --> 00:12:18,965
So for example, here, you detect
a white tile in the

300
00:12:18,965 --> 00:12:21,320
shade and a gray tile
in brightness.

301
00:12:21,320 --> 00:12:22,970
And that's how you
interpret it.

302
00:12:22,970 --> 00:12:25,120
But that's because you're taking
already into account--

303
00:12:25,120 --> 00:12:26,960
without even thinking about
it for a moment--

304
00:12:26,960 --> 00:12:29,560
the fact that this is in the
shadow, and this is in the

305
00:12:29,560 --> 00:12:31,340
bright light.

306
00:12:31,340 --> 00:12:33,400
You're right in your
interpretation.

307
00:12:33,400 --> 00:12:37,540
But if we show a constant
background, you could see that

308
00:12:37,540 --> 00:12:41,340
those two are actually identical
in brightness.

309
00:12:41,340 --> 00:12:45,090
The same thing-- you could take
a look at this cube and

310
00:12:45,090 --> 00:12:47,670
this cube as part of these
Rubik's cubes.

311
00:12:47,670 --> 00:12:48,680
This looks orange.

312
00:12:48,680 --> 00:12:50,370
This looks brown.

313
00:12:50,370 --> 00:12:52,400
You're adjusting for light,
as you make those

314
00:12:52,400 --> 00:12:53,640
interpretations.

315
00:12:53,640 --> 00:12:56,110
And if you make the background
constant, you can see those

316
00:12:56,110 --> 00:12:58,470
are identical in color
and luminance.

317
00:12:58,470 --> 00:13:00,550
So this shows you that we're
constantly making these

318
00:13:00,550 --> 00:13:01,770
interpretations.

319
00:13:01,770 --> 00:13:05,470
Only these tricks of visual
illusion make us realize that

320
00:13:05,470 --> 00:13:07,790
our inferences are occasionally
wrong.

321
00:13:07,790 --> 00:13:10,540
But they're incredibly right
almost all the time.

322
00:13:10,540 --> 00:13:13,790
So here's another example
of taking a local

323
00:13:13,790 --> 00:13:15,220
luminance into account.

324
00:13:15,220 --> 00:13:18,180
So typically, people see
this as a lighter gray.

325
00:13:18,180 --> 00:13:19,570
And this is a darker gray.

326
00:13:19,570 --> 00:13:22,090
Again, they're using the
contrast that surrounds it to

327
00:13:22,090 --> 00:13:23,860
make that interpretation.

328
00:13:23,860 --> 00:13:26,330
Because if we give it a constant
background, you can

329
00:13:26,330 --> 00:13:28,530
see that in an objective sense
the two are identical.

330
00:13:31,730 --> 00:13:34,570
People use edges as a very
powerful source of information

331
00:13:34,570 --> 00:13:35,540
to make these inferences.

332
00:13:35,540 --> 00:13:37,240
And you can see that
in one moment.

333
00:13:37,240 --> 00:13:40,300
Because this boundary, here,
is used to make judgments

334
00:13:40,300 --> 00:13:43,670
about the relative brightness
around an object.

335
00:13:43,670 --> 00:13:45,890
So you just do this.

336
00:13:45,890 --> 00:13:48,850
The boundary disappears, and
the two things look equally

337
00:13:48,850 --> 00:13:50,920
gray, one above the other.

338
00:13:50,920 --> 00:13:54,960
Or no less spectacular, but
measured example, here's two

339
00:13:54,960 --> 00:13:57,530
different shades of gray.

340
00:13:57,530 --> 00:13:59,560
And if you're in a light meter,
objectively looking at

341
00:13:59,560 --> 00:14:02,990
how much light is reflected from
this, here that changes.

342
00:14:02,990 --> 00:14:05,490
This looks almost the
same as this.

343
00:14:05,490 --> 00:14:07,220
But really, there's just
a manipulation at the

344
00:14:07,220 --> 00:14:07,750
boundaries.

345
00:14:07,750 --> 00:14:08,780
Again, here's a light meter.

346
00:14:08,780 --> 00:14:09,720
Here's a boundary.

347
00:14:09,720 --> 00:14:12,280
So these two sides are
really identical.

348
00:14:12,280 --> 00:14:15,130
It's just been manipulated
by the boundary.

349
00:14:15,130 --> 00:14:17,950
And you can see that, because
if you get rid of that

350
00:14:17,950 --> 00:14:19,570
boundary, now you can
see that these two

351
00:14:19,570 --> 00:14:21,040
sides are equally bright.

352
00:14:23,680 --> 00:14:27,100
Another example, again, of how
we're constantly using local

353
00:14:27,100 --> 00:14:28,650
information about brightness
to make

354
00:14:28,650 --> 00:14:30,500
judgments about objects.

355
00:14:30,500 --> 00:14:32,610
Here, you see a picture
of a woman.

356
00:14:32,610 --> 00:14:33,530
Here's her forehead.

357
00:14:33,530 --> 00:14:35,880
Here's her darker hair.

358
00:14:35,880 --> 00:14:38,980
But in fact, that's a judgment
in interpretation.

359
00:14:38,980 --> 00:14:41,920
Because if we cover up the
surrounding information, the

360
00:14:41,920 --> 00:14:43,470
hair is identical to the face.

361
00:14:46,020 --> 00:14:48,510
Here's some other illusions,
not all about brightness.

362
00:14:48,510 --> 00:14:51,750
But, again, they're just
reminders that what we see is

363
00:14:51,750 --> 00:14:53,360
constantly a calculation and an

364
00:14:53,360 --> 00:14:55,730
interpretation of many factors.

365
00:14:55,730 --> 00:14:56,995
So here's a fun spiral.

366
00:14:56,995 --> 00:15:00,510
You can see it zooming around,
going from inner tightness and

367
00:15:00,510 --> 00:15:02,860
unwinding peripherally.

368
00:15:02,860 --> 00:15:07,030
And so we can add
a dot to that.

369
00:15:07,030 --> 00:15:09,475
And now, we can let this travel
around to the middle.

370
00:15:12,510 --> 00:15:15,910
You can see it's not quite
making the middle.

371
00:15:15,910 --> 00:15:18,440
And we can try that again.

372
00:15:18,440 --> 00:15:21,560
Because in fact, in this
picture, every line, every

373
00:15:21,560 --> 00:15:22,700
circle is a circle.

374
00:15:22,700 --> 00:15:24,310
And none of them are connected,
one to the other.

375
00:15:24,310 --> 00:15:26,980
They just appear that way.

376
00:15:26,980 --> 00:15:30,940
This illusion, that Richard
Gregory describes, was

377
00:15:30,940 --> 00:15:34,120
inspired because he saw a wall
in a cafe that looked like it

378
00:15:34,120 --> 00:15:36,900
had not been well assembled
by the person building it.

379
00:15:36,900 --> 00:15:39,870
As if they had been consuming
something from the cafe as

380
00:15:39,870 --> 00:15:41,780
they built the wall.

381
00:15:41,780 --> 00:15:46,650
And bizarrely, having all kinds
of lines that are not

382
00:15:46,650 --> 00:15:47,980
lined up, as you'd expect in

383
00:15:47,980 --> 00:15:49,480
well-constructed tiles on a wall.

384
00:15:49,480 --> 00:15:53,260
But in fact, every line, here,
is perfectly straight.

385
00:15:53,260 --> 00:15:56,830
And you're interpreting the
black and white edges to

386
00:15:56,830 --> 00:16:00,770
confuse you about the lining
up across the horizontal.

387
00:16:00,770 --> 00:16:02,110
This is very striking.

388
00:16:02,110 --> 00:16:06,260
Here's Rubik's cubes that,
again, now they're adding one

389
00:16:06,260 --> 00:16:08,020
more story into this,
which is color.

390
00:16:08,020 --> 00:16:10,970
But we use that often in every
day sight, of course, to make

391
00:16:10,970 --> 00:16:13,020
conclusions about things.

392
00:16:13,020 --> 00:16:15,670
And here, you can see the
difference between, for

393
00:16:15,670 --> 00:16:19,410
example, this yellow cube
and this brown cube.

394
00:16:23,320 --> 00:16:26,320
What did I want to do?

395
00:16:26,320 --> 00:16:27,710
You can see the difference
between this brown cube--

396
00:16:27,710 --> 00:16:30,170
sorry, let me get my
illusion correct--

397
00:16:30,170 --> 00:16:34,400
and this darker cube, lighter
brown and darker brown.

398
00:16:34,400 --> 00:16:36,480
But if we cover those up,
those are identical.

399
00:16:36,480 --> 00:16:39,250
And they're gray.

400
00:16:39,250 --> 00:16:41,710
Here's another one that shows
you it's not just the color

401
00:16:41,710 --> 00:16:43,680
itself but the spatial
location.

402
00:16:43,680 --> 00:16:48,410
So here's a pink circle, near
the middle of this display.

403
00:16:48,410 --> 00:16:51,110
And here's a more orange one.

404
00:16:51,110 --> 00:16:55,780
So we're just simply going to
expand those out and get it

405
00:16:55,780 --> 00:16:57,800
bigger and bigger.

406
00:16:57,800 --> 00:17:01,030
And as it's moving
across your eye--

407
00:17:01,030 --> 00:17:03,655
even though, objectively, it
remains pink and orange, or

408
00:17:03,655 --> 00:17:05,530
the things will lead you
to you interpret that--

409
00:17:05,530 --> 00:17:08,440
as it gets big enough, the color
difference disappears to

410
00:17:08,440 --> 00:17:10,660
your perception.

411
00:17:10,660 --> 00:17:12,400
And sometimes you can
get-- and this is a

412
00:17:12,400 --> 00:17:14,430
very complicated one--

413
00:17:14,430 --> 00:17:15,609
illusions--

414
00:17:15,609 --> 00:17:17,604
I don't know if it's
working for you--

415
00:17:17,604 --> 00:17:21,200
that these things are actually
turning, even though none of

416
00:17:21,200 --> 00:17:22,450
them are moving at all.

417
00:17:26,390 --> 00:17:26,690
OK.

418
00:17:26,690 --> 00:17:29,370
And one last, simple example.

419
00:17:29,370 --> 00:17:31,180
Here's yellow on gray.

420
00:17:31,180 --> 00:17:32,680
Here's light gray on yellow.

421
00:17:32,680 --> 00:17:34,490
But in fact, these X's
are identical.

422
00:17:34,490 --> 00:17:36,170
And you can see that when
they meet here.

423
00:17:36,170 --> 00:17:41,160
So again, we're constantly
interpreting color, shape,

424
00:17:41,160 --> 00:17:43,720
brightness, by things that
are surrounding it.

425
00:17:43,720 --> 00:17:47,070
And everything is being the
product of an interpretation,

426
00:17:47,070 --> 00:17:50,530
not what's out there in an
objective sense of self.

427
00:17:50,530 --> 00:17:52,920
And so, if we think about this
Objectivist view that that

428
00:17:52,920 --> 00:17:56,390
can't be right, that the limit
of this objective view would

429
00:17:56,390 --> 00:17:58,180
be that everybody sees
everything differently all the

430
00:17:58,180 --> 00:18:00,140
time, that would be a
confusing world of

431
00:18:00,140 --> 00:18:00,830
hallucinations.

432
00:18:00,830 --> 00:18:02,590
We don't live there
either, mostly.

433
00:18:02,590 --> 00:18:05,020
So we think it's the because
there's a relationship between

434
00:18:05,020 --> 00:18:06,150
these, or a synthesis.

435
00:18:06,150 --> 00:18:08,330
We perceive only within limits
of our nervous system.

436
00:18:08,330 --> 00:18:10,900
There's a way that
we see the world.

437
00:18:10,900 --> 00:18:13,750
But the way that our nervous
system computes and makes

438
00:18:13,750 --> 00:18:16,590
inferences about the world
reflects many properties of

439
00:18:16,590 --> 00:18:19,900
the world that are efficient
and accurate.

440
00:18:19,900 --> 00:18:23,460
So how do we get information
into our brains to see?

441
00:18:23,460 --> 00:18:26,130
So, if you were to design a
system that was hopelessly

442
00:18:26,130 --> 00:18:29,220
failed, the first thing you
would do is think the visual

443
00:18:29,220 --> 00:18:31,090
system is going to fail.

444
00:18:31,090 --> 00:18:33,810
The first thing it does is it
takes objects in the world and

445
00:18:33,810 --> 00:18:35,800
it flips them upside down-- and
how they're perceived in

446
00:18:35,800 --> 00:18:38,070
the eye, the front of the eye,
the back of the eye.

447
00:18:38,070 --> 00:18:39,510
Here's the retina.

448
00:18:39,510 --> 00:18:42,860
And the first place in your
brain that sees something is

449
00:18:42,860 --> 00:18:44,370
in the back of your eye.

450
00:18:44,370 --> 00:18:46,190
You'd think you would put it
in the front of your eye.

451
00:18:46,190 --> 00:18:48,510
And information bounces
back into the

452
00:18:48,510 --> 00:18:49,280
deepest part of the eye.

453
00:18:49,280 --> 00:18:52,930
And then it comes out and leads
through the optic nerve.

454
00:18:52,930 --> 00:18:54,890
You also know, from prior
classes, that the world is

455
00:18:54,890 --> 00:18:58,070
organized by left and right
visual fields, instead of the

456
00:18:58,070 --> 00:19:02,090
way we intuitively think
about the world.

457
00:19:02,090 --> 00:19:03,840
Let's take a closer look
at the retina.

458
00:19:03,840 --> 00:19:07,980
These layers of cells in your
eye that begin vision.

459
00:19:07,980 --> 00:19:10,900
So light comes in and bounces
through the back.

460
00:19:10,900 --> 00:19:14,350
And here are the rods and cones
where vision begins.

461
00:19:14,350 --> 00:19:17,030
The cones are primarily
near the fovea.

462
00:19:17,030 --> 00:19:18,655
The rods are primarily
peripheral.

463
00:19:21,950 --> 00:19:26,750
And I'm reminding you, again, as
the optic nerve leaves, it

464
00:19:26,750 --> 00:19:28,020
goes through the optic chiasm.

465
00:19:28,020 --> 00:19:30,350
The fibers get sorted out by
whether they're having

466
00:19:30,350 --> 00:19:33,250
information from the left or
right side of the world, and

467
00:19:33,250 --> 00:19:34,520
then enter the cortex
that way.

468
00:19:37,020 --> 00:19:42,830
Within the retina, within your
eye, here is a graphic of the

469
00:19:42,830 --> 00:19:43,810
different kinds of cells.

470
00:19:43,810 --> 00:19:46,340
Here's what the cells
actually look like.

471
00:19:48,840 --> 00:19:54,900
And here's a blow up of these
remarkable cones, these larger

472
00:19:54,900 --> 00:20:00,620
entities and the rods that you
can see are very small, but

473
00:20:00,620 --> 00:20:02,300
many of them.

474
00:20:02,300 --> 00:20:04,890
And the big difference between
rods and cones is that the

475
00:20:04,890 --> 00:20:07,660
rods are receptive to
information at only one

476
00:20:07,660 --> 00:20:08,400
wavelength.

477
00:20:08,400 --> 00:20:09,760
They're effectively
color blind.

478
00:20:09,760 --> 00:20:10,940
We'll come back to that.

479
00:20:10,940 --> 00:20:13,910
But that's a good wavelength
to pick up information, for

480
00:20:13,910 --> 00:20:15,980
example, in low light.

481
00:20:15,980 --> 00:20:19,570
The cones are selective for
blue or red or green.

482
00:20:19,570 --> 00:20:22,350
That's the stuff of
seeing color.

483
00:20:22,350 --> 00:20:24,990
But it doesn't operate too well,
unless you have a pretty

484
00:20:24,990 --> 00:20:28,430
well illuminated situation.

485
00:20:28,430 --> 00:20:32,880
And the rods and cones have
a striking physical

486
00:20:32,880 --> 00:20:33,960
organization.

487
00:20:33,960 --> 00:20:36,930
So here's the distribution
of cones for color.

488
00:20:36,930 --> 00:20:40,880
Pretty much, all of them in the
fovea in the middle part

489
00:20:40,880 --> 00:20:42,280
of the retina.

490
00:20:42,280 --> 00:20:45,450
And then the rods are spread
out peripherally.

491
00:20:45,450 --> 00:20:50,050
So there's this big difference
in where they're located.

492
00:20:50,050 --> 00:20:53,260
And another big difference is
that the cones, the ones that

493
00:20:53,260 --> 00:20:57,860
respond to color, and that are
in the middle of the fovea,

494
00:20:57,860 --> 00:21:01,130
have an almost one-to-one
relationship between the

495
00:21:01,130 --> 00:21:04,070
receptors for color information
and the neurons

496
00:21:04,070 --> 00:21:05,730
that leave the eye with
information that

497
00:21:05,730 --> 00:21:07,350
will go into the brain.

498
00:21:07,350 --> 00:21:09,420
The rods have a many-to-one
relationship.

499
00:21:09,420 --> 00:21:13,510
Many rods are contributing for a
combined calculation of some

500
00:21:13,510 --> 00:21:16,040
kind to a single neuron
that leaves the brain.

501
00:21:16,040 --> 00:21:18,620
So you can already see, at the
very first moment of reception

502
00:21:18,620 --> 00:21:21,330
from the rods and from the cons,
very different kinds of

503
00:21:21,330 --> 00:21:26,430
processes are being begun,
as you begin to see.

504
00:21:26,430 --> 00:21:29,180
Two ideas have been very helpful
for understanding how

505
00:21:29,180 --> 00:21:30,810
vision is organized
in the brain--

506
00:21:30,810 --> 00:21:33,640
our receptive fields
and retinotopy.

507
00:21:33,640 --> 00:21:36,760
Receptive fields are, simply,
an area of external space.

508
00:21:36,760 --> 00:21:39,560
It's a physical spot
in space in which a

509
00:21:39,560 --> 00:21:40,885
stimulus activates a neuron.

510
00:21:40,885 --> 00:21:44,290
So neurons, early in vision,
will respond to specific

511
00:21:44,290 --> 00:21:46,850
locations in space when you're
looking straightforward.

512
00:21:46,850 --> 00:21:48,460
One neuron might cover
this area.

513
00:21:48,460 --> 00:21:51,880
Another neuron might cover
that area and so forth.

514
00:21:51,880 --> 00:21:55,085
Retinotopy refers to topographic
or spatial maps of

515
00:21:55,085 --> 00:21:56,540
visual space across
a restricted

516
00:21:56,540 --> 00:21:58,100
region of the brain.

517
00:21:58,100 --> 00:22:02,240
And so, retinotopy keeps these
receptive fields lined up.

518
00:22:02,240 --> 00:22:05,000
So if the neuron is seeing this
spot, this retinotopic

519
00:22:05,000 --> 00:22:07,330
map, and so the next spot
stays next to it.

520
00:22:07,330 --> 00:22:08,750
And the next spot stays
next to it.

521
00:22:08,750 --> 00:22:11,730
So you can reconstruct what
you see, based on a lot of

522
00:22:11,730 --> 00:22:16,480
very local glances at
the environment.

523
00:22:16,480 --> 00:22:19,670
And then it travels from the
eye, through the optic nerve,

524
00:22:19,670 --> 00:22:22,060
into something called the
lateral geniculate nucleus.

525
00:22:22,060 --> 00:22:24,760
The nucleus is simply a
collection of cells.

526
00:22:24,760 --> 00:22:26,670
This is part of the thalamus.

527
00:22:26,670 --> 00:22:29,490
And from there, the fibers will
travel into the cortex to

528
00:22:29,490 --> 00:22:31,920
begin conscious visual
perception.

529
00:22:31,920 --> 00:22:34,200
And you have one on the left
and one in the right of the

530
00:22:34,200 --> 00:22:36,930
lateral geniculate nucleon.

531
00:22:36,930 --> 00:22:38,440
And if you look at
this blown up--

532
00:22:38,440 --> 00:22:40,430
this is a sample from a monkey,
but the human one

533
00:22:40,430 --> 00:22:41,850
looks very similar.

534
00:22:41,850 --> 00:22:46,960
You can see 1, 2, 3, 4, 5,
6 layers of cells in each

535
00:22:46,960 --> 00:22:49,180
lateral geniculate nucleus.

536
00:22:49,180 --> 00:22:51,060
The first two look a little
bit different.

537
00:22:51,060 --> 00:22:55,470
If you look carefully, they're
comprised of larger cells.

538
00:22:55,470 --> 00:22:58,870
The other four are comprised
of smaller cells.

539
00:22:58,870 --> 00:23:01,350
So people describe the cell
layer, made up of larger

540
00:23:01,350 --> 00:23:02,710
cells, as magnocellular--

541
00:23:02,710 --> 00:23:03,890
big cells.

542
00:23:03,890 --> 00:23:07,620
And the other four layers as
parvocellular, or small cells.

543
00:23:07,620 --> 00:23:11,950
The mangocellular layers tend
to have large cells.

544
00:23:11,950 --> 00:23:14,190
A lot of their information comes
from the rods, the black

545
00:23:14,190 --> 00:23:17,650
and white elements, good for
seeing in low vision.

546
00:23:17,650 --> 00:23:19,430
They have large receptive
fields.

547
00:23:19,430 --> 00:23:21,380
They cover relatively
large patches of the

548
00:23:21,380 --> 00:23:22,460
world for each neuron.

549
00:23:22,460 --> 00:23:25,820
Three times larger than the
parvocellular ones, the

550
00:23:25,820 --> 00:23:29,610
neurons fire rapidly,
transiently, are color blind.

551
00:23:29,610 --> 00:23:33,460
They're good in low contrast,
low lighting.

552
00:23:33,460 --> 00:23:35,430
The parvocellular ones
are smaller.

553
00:23:35,430 --> 00:23:37,260
They get a lot of their
input from the cones.

554
00:23:37,260 --> 00:23:39,750
They have smaller receptive
fields.

555
00:23:39,750 --> 00:23:42,770
They have slow, sustained
responses.

556
00:23:42,770 --> 00:23:45,010
They don't turn on and
off very quickly.

557
00:23:45,010 --> 00:23:46,150
They're wavelength sensitive.

558
00:23:46,150 --> 00:23:48,170
So they can respond to color.

559
00:23:48,170 --> 00:23:50,690
They operate best when things
are very bright.

560
00:23:50,690 --> 00:23:52,240
They're unique to primates,
and there are 10

561
00:23:52,240 --> 00:23:53,280
times more of them.

562
00:23:53,280 --> 00:23:55,420
But again, even before you get
to the cortex, there's this

563
00:23:55,420 --> 00:23:58,510
huge organization between
two modes or pathways of

564
00:23:58,510 --> 00:24:00,040
perception, so you
can construct the

565
00:24:00,040 --> 00:24:02,150
world as you see it.

566
00:24:02,150 --> 00:24:04,770
And then we've gone over
before that as you

567
00:24:04,770 --> 00:24:06,310
leave the nuclei --

568
00:24:06,310 --> 00:24:08,310
the lateral geniculate nuclei--
then the next fiber

569
00:24:08,310 --> 00:24:11,570
pathway takes you into the
primary visual cortex in the

570
00:24:11,570 --> 00:24:12,820
back of your brain.

571
00:24:14,990 --> 00:24:18,950
Now, vision is so important for
primates and people, that

572
00:24:18,950 --> 00:24:22,370
when people estimate how much of
the neocortex is devoted to

573
00:24:22,370 --> 00:24:25,520
different modalities, a common
estimate is that about half of

574
00:24:25,520 --> 00:24:29,000
the brain is primarily
devoted to vision--

575
00:24:29,000 --> 00:24:31,700
11% to touch, 3% to audition.

576
00:24:31,700 --> 00:24:33,380
These are ballpark estimates.

577
00:24:33,380 --> 00:24:38,430
But it shows you how visual we
are with our fellow primates.

578
00:24:38,430 --> 00:24:42,080
There's many specialized
areas within the

579
00:24:42,080 --> 00:24:43,260
human visual cortex.

580
00:24:43,260 --> 00:24:47,390
They estimated some years ago it
was 32 distinct areas, each

581
00:24:47,390 --> 00:24:50,450
performing a different tasks,
lots of specialization.

582
00:24:50,450 --> 00:24:54,610
And another fascinating aspect
of vision is proliferation as

583
00:24:54,610 --> 00:24:58,970
we go up into higher stations
of visual processing.

584
00:24:58,970 --> 00:25:02,440
So in one lateral geniculate
nucleus in your brain-- and

585
00:25:02,440 --> 00:25:03,960
you have two of them, one on
the left and one on the

586
00:25:03,960 --> 00:25:06,780
right-- you have about
a million neurons.

587
00:25:06,780 --> 00:25:10,210
They will send information that
will be interpreted by

588
00:25:10,210 --> 00:25:14,280
about 250 million neurons
in your visual cortex.

589
00:25:14,280 --> 00:25:17,460
Those will communicate about
400 million neurons in the

590
00:25:17,460 --> 00:25:19,450
next level of processing.

591
00:25:19,450 --> 00:25:22,120
And finally, there'll be
something like 1.3 billion

592
00:25:22,120 --> 00:25:23,580
visual cortical neurons
in the brain,

593
00:25:23,580 --> 00:25:26,030
overall, in a gross estimate.

594
00:25:26,030 --> 00:25:28,610
It's as if to discover what
you're seeing, you're having

595
00:25:28,610 --> 00:25:31,620
larger and larger populations
of neurons unpacking the

596
00:25:31,620 --> 00:25:34,520
mystery of what initially
came into the brain very

597
00:25:34,520 --> 00:25:38,010
impoverished information, that's
expanded to begin to

598
00:25:38,010 --> 00:25:41,760
understand what you need to
perceive a face or a word or a

599
00:25:41,760 --> 00:25:43,160
physical movement.

600
00:25:43,160 --> 00:25:47,230
So a single lateral geniculate
neuron, it could be estimated,

601
00:25:47,230 --> 00:25:51,900
will take 600 cortical neurons
to interpret what this lateral

602
00:25:51,900 --> 00:25:53,870
geniculate neuron has
been exposed to.

603
00:25:56,850 --> 00:25:59,380
Here's this area in the visual
cortex, where the information

604
00:25:59,380 --> 00:26:02,185
arrives first into your brain
and primary visual cortex.

605
00:26:04,970 --> 00:26:09,150
And when we think about cortex
in perception, in several

606
00:26:09,150 --> 00:26:10,900
modalities, one of the striking
things is how it's

607
00:26:10,900 --> 00:26:13,670
organized to achieve
certain goals.

608
00:26:13,670 --> 00:26:15,910
So here, we're going to switch
for a moment to the parts of

609
00:26:15,910 --> 00:26:17,690
your brain that move
your body--

610
00:26:17,690 --> 00:26:21,890
motor cortex or that have
your sense of touch.

611
00:26:21,890 --> 00:26:24,740
And when people stimulate these
and look at what they're

612
00:26:24,740 --> 00:26:25,380
related to.

613
00:26:25,380 --> 00:26:27,310
For example, in patients
undergoing treatment for

614
00:26:27,310 --> 00:26:30,080
epilepsy or an animal study--
support this where you can do

615
00:26:30,080 --> 00:26:32,330
much more expensive
experimentation--

616
00:26:32,330 --> 00:26:36,320
a very striking thing occurs,
which is the number of neurons

617
00:26:36,320 --> 00:26:38,880
devoted to different parts of
your body are radically

618
00:26:38,880 --> 00:26:41,680
different than your
actual body size.

619
00:26:41,680 --> 00:26:44,390
So this funny looking,
so-called homunculus

620
00:26:44,390 --> 00:26:46,930
represents how many neurons,
for example, in your motor

621
00:26:46,930 --> 00:26:50,590
cortex are devoted to different
parts of your body.

622
00:26:50,590 --> 00:26:53,640
So it turns out that we do
incredible things with our

623
00:26:53,640 --> 00:26:57,050
hands, so we devote, it seems
like, a tremendous amount of

624
00:26:57,050 --> 00:26:57,960
this area to the hand.

625
00:26:57,960 --> 00:26:59,210
That's what that
looks so large.

626
00:27:01,850 --> 00:27:04,030
There's a reason we do
handshaking and handwriting

627
00:27:04,030 --> 00:27:07,140
and typing with our hands
and not with our hips.

628
00:27:07,140 --> 00:27:09,260
Because although our hips are
physically larger than our

629
00:27:09,260 --> 00:27:11,750
hand, we devote an incredibly
small number of

630
00:27:11,750 --> 00:27:13,320
neurons to do that.

631
00:27:13,320 --> 00:27:16,630
So it's a trick of the brain,
given a certain limited size,

632
00:27:16,630 --> 00:27:19,970
to blow up its representation
of what needs to be done

633
00:27:19,970 --> 00:27:21,270
brilliantly.

634
00:27:21,270 --> 00:27:23,910
And to reduce its representation
of parts of the

635
00:27:23,910 --> 00:27:26,420
body that it doesn't have
to do too much with.

636
00:27:26,420 --> 00:27:27,290
So what's blown up?

637
00:27:27,290 --> 00:27:28,840
Your hand?

638
00:27:28,840 --> 00:27:32,130
Things to do with your face,
so that you can speak?

639
00:27:32,130 --> 00:27:34,220
Your tongue, because that has to
accomplish a lot, in terms

640
00:27:34,220 --> 00:27:36,810
of speech, in terms of eating.

641
00:27:36,810 --> 00:27:39,540
And things like your toes
and ankle are gypped of

642
00:27:39,540 --> 00:27:41,900
representation, because you
don't do that much that's

643
00:27:41,900 --> 00:27:42,640
brilliance.

644
00:27:42,640 --> 00:27:44,510
There's a very clever strategy,
by the brain, to

645
00:27:44,510 --> 00:27:48,760
devote the amount of neural
resources to make that part of

646
00:27:48,760 --> 00:27:51,600
your body accomplish either
very complicated things or

647
00:27:51,600 --> 00:27:53,850
very simple things.

648
00:27:53,850 --> 00:27:59,440
And if people on the outside
looked like this--

649
00:27:59,440 --> 00:28:01,010
this is what they
would look like.

650
00:28:01,010 --> 00:28:04,490
Giant hands, giant head, and
a very shriveled up trunk.

651
00:28:04,490 --> 00:28:07,410
So that's not what we look like,
but that's the way our

652
00:28:07,410 --> 00:28:10,820
brain represents us, our bodies,
so that we could have

653
00:28:10,820 --> 00:28:14,540
fantastically complicated
control of our hands and

654
00:28:14,540 --> 00:28:18,960
things having to do with speech
around our mouth.

655
00:28:18,960 --> 00:28:20,670
It's different for different
modalities.

656
00:28:20,670 --> 00:28:22,900
For hearing, things
are organized

657
00:28:22,900 --> 00:28:26,220
by frequency, tonotopy.

658
00:28:26,220 --> 00:28:29,350
For vision, things are organized
in two ways.

659
00:28:29,350 --> 00:28:31,050
Spatial relations
are maintained.

660
00:28:31,050 --> 00:28:33,650
They have to be, so you know,
as different neurons saw

661
00:28:33,650 --> 00:28:36,500
different receptive fields,
how's that all kept up?

662
00:28:36,500 --> 00:28:39,520
Here's a kind of a rough
experiment, where a monkey

663
00:28:39,520 --> 00:28:42,730
saw-- before it was, as they
politely say, sacrificed--

664
00:28:42,730 --> 00:28:44,480
a display like this.

665
00:28:44,480 --> 00:28:46,520
This is the brain of the
monkey, flattened out.

666
00:28:46,520 --> 00:28:49,620
And you can see it has this
representation, like this.

667
00:28:49,620 --> 00:28:51,240
Now you could say, well,
that's how we see.

668
00:28:51,240 --> 00:28:52,100
We line everything up.

669
00:28:52,100 --> 00:28:54,240
But this is not the part of our
brain that we associate

670
00:28:54,240 --> 00:28:56,130
with conscious perception
at all.

671
00:28:56,130 --> 00:28:59,880
But if you didn't keep the
spatial information aligned

672
00:28:59,880 --> 00:29:02,530
correctly, you could never
interpret things, in the end,

673
00:29:02,530 --> 00:29:04,830
correctly that go together, like
different parts of a hand

674
00:29:04,830 --> 00:29:07,190
or different letters
within a word.

675
00:29:07,190 --> 00:29:10,470
In humans, it's also in
the back of the brain.

676
00:29:10,470 --> 00:29:12,000
It also keeps spatial
information.

677
00:29:12,000 --> 00:29:14,800
But what do humans do,
and other primates?

678
00:29:14,800 --> 00:29:18,710
They greatly expand the central
part of vision.

679
00:29:18,710 --> 00:29:21,940
So here's the central part of
vision, the so-called fovea,

680
00:29:21,940 --> 00:29:24,560
which is a small part of
the visual display.

681
00:29:24,560 --> 00:29:27,240
Here is everything else that's
in peripheral vision.

682
00:29:27,240 --> 00:29:28,960
So this is dark green,
dark purple.

683
00:29:28,960 --> 00:29:31,690
But when it comes to the
representation in the visual

684
00:29:31,690 --> 00:29:34,260
cortex, this small area
of what we see

685
00:29:34,260 --> 00:29:35,095
becomes pretty large.

686
00:29:35,095 --> 00:29:37,110
It's way overrepresented.

687
00:29:37,110 --> 00:29:42,010
So just as the motor cortex over
represents the hand, the

688
00:29:42,010 --> 00:29:45,750
visual cortex over represents
the cone

689
00:29:45,750 --> 00:29:49,380
foveal part of the brain.

690
00:29:49,380 --> 00:29:51,960
So they can do much more
examination, much more

691
00:29:51,960 --> 00:29:55,460
computation on that small
foveated area.

692
00:29:55,460 --> 00:29:57,980
So that's why it matters,
tremendously, where people put

693
00:29:57,980 --> 00:29:58,870
their eyes.

694
00:29:58,870 --> 00:30:01,270
Where they put their eyes is
where neuromachinery is

695
00:30:01,270 --> 00:30:04,050
dedicated to do its most
brilliant analysis.

696
00:30:04,050 --> 00:30:06,590
And peripheral regions, which
are large, in vision, get

697
00:30:06,590 --> 00:30:08,570
relatively small
representation.

698
00:30:08,570 --> 00:30:11,640
All we do, in the periphery,
is notice if something is

699
00:30:11,640 --> 00:30:13,930
whizzing at our head.

700
00:30:13,930 --> 00:30:15,655
And for that, we don't need
to be that sophisticated.

701
00:30:18,580 --> 00:30:21,330
Now, what do neurons
communicate in

702
00:30:21,330 --> 00:30:22,410
primary visual cortex?

703
00:30:22,410 --> 00:30:24,510
Well, this is the, sort of,
seminal Nobel Prize-winning

704
00:30:24,510 --> 00:30:27,240
work from Hubel and Wiesel, at
Harvard, who discovered that

705
00:30:27,240 --> 00:30:32,130
what neurons respond to in
primary visual cortex is the

706
00:30:32,130 --> 00:30:34,850
orientation of a piece of
something like this.

707
00:30:34,850 --> 00:30:37,955
So here's a neuron that loves
a little bit of a stimulus.

708
00:30:37,955 --> 00:30:39,340
You notice that disorientation?

709
00:30:39,340 --> 00:30:43,200
Here's each of these lines
as of the neuron firing.

710
00:30:43,200 --> 00:30:46,960
It generalizes if a bar is close
in orientation, it will

711
00:30:46,960 --> 00:30:48,820
still like it, somewhat.

712
00:30:48,820 --> 00:30:51,980
And if the bar moves away from
its preferred orientation, it

713
00:30:51,980 --> 00:30:53,330
won't respond and all.

714
00:30:53,330 --> 00:30:55,920
So these neurons are coding
something about local

715
00:30:55,920 --> 00:30:57,830
orientation of little lines.

716
00:30:57,830 --> 00:30:59,820
Those little lines will be
assembled later on in the

717
00:30:59,820 --> 00:31:03,420
brain to represent a letter,
a word, a book, a face, a

718
00:31:03,420 --> 00:31:04,540
chair, and so on.

719
00:31:04,540 --> 00:31:07,240
But these neurons simply know
they're seeing a line of this

720
00:31:07,240 --> 00:31:10,020
angle or that angle that will
be assembled later on for an

721
00:31:10,020 --> 00:31:12,840
entire conscious percept.

722
00:31:12,840 --> 00:31:15,940
And then that information will
go from primary visual cortex,

723
00:31:15,940 --> 00:31:19,210
in two pathways, that have two
quite different properties,

724
00:31:19,210 --> 00:31:22,020
but turn out to be the
super-information processing

725
00:31:22,020 --> 00:31:25,220
pathways of your brain
and my brain.

726
00:31:25,220 --> 00:31:27,390
So where do we first learn
about this fundamental

727
00:31:27,390 --> 00:31:30,510
organization of how we see into
two giant highways of

728
00:31:30,510 --> 00:31:31,880
information processing?

729
00:31:31,880 --> 00:31:34,270
Well, the first studies where we
should study's in monkeys.

730
00:31:34,270 --> 00:31:36,600
And I'll show you work
in humans as well.

731
00:31:36,600 --> 00:31:39,430
And the major discovery was
this, that in our brains,

732
00:31:39,430 --> 00:31:42,220
there's a so-called where
pathway that goes up in the

733
00:31:42,220 --> 00:31:46,010
brain into the parietal cortex
and a what pathway that goes

734
00:31:46,010 --> 00:31:48,730
into temporal lobe
or lower cortex.

735
00:31:48,730 --> 00:31:52,190
And if you make lesions in these
different regions, one

736
00:31:52,190 --> 00:31:54,900
form of vision is spared
and one is impaired

737
00:31:54,900 --> 00:31:55,910
in selective ways.

738
00:31:55,910 --> 00:31:57,660
So how did they discover this?

739
00:31:57,660 --> 00:31:59,590
So again, here's the
big picture.

740
00:31:59,590 --> 00:32:02,960
Early visual processing
information a what pathway to

741
00:32:02,960 --> 00:32:06,480
know what objects are a
face, a word, a chair.

742
00:32:06,480 --> 00:32:09,400
Or where things are, that
you might run to,

743
00:32:09,400 --> 00:32:12,410
grab, or jump over.

744
00:32:12,410 --> 00:32:13,280
So here was the experiment.

745
00:32:13,280 --> 00:32:14,820
It was pretty simple.

746
00:32:14,820 --> 00:32:16,460
Here were two food wells.

747
00:32:16,460 --> 00:32:19,840
And monkeys that were eager to
get food would get rewarded,

748
00:32:19,840 --> 00:32:23,260
if they would pick the
correct food well.

749
00:32:23,260 --> 00:32:25,790
They didn't see where the food
was that was hidden, but they

750
00:32:25,790 --> 00:32:27,930
would get to pick one
or the other.

751
00:32:27,930 --> 00:32:32,700
And in one task they would have
to pick which food well

752
00:32:32,700 --> 00:32:34,540
is closer to the stimulus.

753
00:32:34,540 --> 00:32:35,680
That's a where task.

754
00:32:35,680 --> 00:32:36,970
Where is this located?

755
00:32:36,970 --> 00:32:38,570
That's the piece of information
that will tell me

756
00:32:38,570 --> 00:32:40,140
where to get my food.

757
00:32:40,140 --> 00:32:41,850
And you can see, here's the
performance of the animals,

758
00:32:41,850 --> 00:32:42,420
their errors.

759
00:32:42,420 --> 00:32:43,360
So it's good to be low.

760
00:32:43,360 --> 00:32:44,800
They are eager to
get their food.

761
00:32:44,800 --> 00:32:48,370
But if the animal had a
surgical lesion in the

762
00:32:48,370 --> 00:32:51,180
parietal cortex, the where
pathway, they were very poor

763
00:32:51,180 --> 00:32:52,810
at performing this task.

764
00:32:52,810 --> 00:32:55,740
As if they couldn't appreciate
the spatial relation between

765
00:32:55,740 --> 00:32:59,360
where the cylinder was located
and the food well.

766
00:32:59,360 --> 00:33:01,370
On the other hand, monkeys who
had lesions made in the

767
00:33:01,370 --> 00:33:05,720
temporal cortex were poor when
they had to make this task.

768
00:33:05,720 --> 00:33:07,270
They would see two different
objects.

769
00:33:07,270 --> 00:33:09,760
And they were taught that
one always meant that's

770
00:33:09,760 --> 00:33:10,400
where the food is.

771
00:33:10,400 --> 00:33:12,350
Let's pretend it's
the cylinder.

772
00:33:12,350 --> 00:33:14,000
So in order to know the correct
answer, they would

773
00:33:14,000 --> 00:33:17,200
have to say, OK, is it next to
this object or this object?

774
00:33:17,200 --> 00:33:18,480
That's a what discrimination.

775
00:33:18,480 --> 00:33:20,450
Is it a cylinder or a cube?

776
00:33:20,450 --> 00:33:22,670
And now, the temporal lobe
lesions were the ones that

777
00:33:22,670 --> 00:33:24,130
affected performance.

778
00:33:24,130 --> 00:33:26,850
So it's as if one part of the
brain tells you where things

779
00:33:26,850 --> 00:33:27,740
are, in vision--

780
00:33:27,740 --> 00:33:28,810
parietal lobe.

781
00:33:28,810 --> 00:33:30,610
And one tells you
what they are--

782
00:33:30,610 --> 00:33:32,030
the temporal lobe.

783
00:33:32,030 --> 00:33:35,170
So that's lesion studies
in monkeys.

784
00:33:35,170 --> 00:33:38,170
But there's very interesting
things about when we look at

785
00:33:38,170 --> 00:33:41,110
what cells communicate in these
two pathways that makes

786
00:33:41,110 --> 00:33:44,140
sense in what we understand to
be the goals of these systems.

787
00:33:44,140 --> 00:33:46,710
So for example, in the where
pathway, in the parietal

788
00:33:46,710 --> 00:33:50,780
cortex, many neurons pick up
information from the fovea,

789
00:33:50,780 --> 00:33:52,580
the central part of vision.

790
00:33:52,580 --> 00:33:57,340
But the majority are sensitive
to stimuli in the periphery.

791
00:33:57,340 --> 00:33:59,240
So if you imagine, a good
thing to know if you're

792
00:33:59,240 --> 00:34:00,840
running, is in the periphery.

793
00:34:00,840 --> 00:34:04,020
Are there things coming at your
head, things to avoid?

794
00:34:04,020 --> 00:34:06,610
You might grab something here,
if you need to grab it there.

795
00:34:06,610 --> 00:34:09,030
You would want to have a lot of
peripheral information for

796
00:34:09,030 --> 00:34:10,909
spatial things around you.

797
00:34:10,909 --> 00:34:14,679
And in fact, if a monkey was
looking at this spot, which is

798
00:34:14,679 --> 00:34:17,610
away, entirely, from the
stimulus, it would respond to

799
00:34:17,610 --> 00:34:20,409
both a large and small stimulus
in the periphery

800
00:34:20,409 --> 00:34:21,690
quite strongly.

801
00:34:21,690 --> 00:34:24,120
So these neurons are responding
to a pretty broad

802
00:34:24,120 --> 00:34:27,480
range of space where things
could happen.

803
00:34:27,480 --> 00:34:31,360
Neurons in the what pathway
have almost their entire

804
00:34:31,360 --> 00:34:33,870
responsiveness in the fovea.

805
00:34:33,870 --> 00:34:35,750
Because you know what something
is, if you're

806
00:34:35,750 --> 00:34:37,469
reading a word and
looking at it.

807
00:34:37,469 --> 00:34:39,360
If you're looking at a face,
they're figuring out

808
00:34:39,360 --> 00:34:40,460
who the person is.

809
00:34:40,460 --> 00:34:43,586
So it's only responding, these
kinds of neurons, to

810
00:34:43,586 --> 00:34:44,940
information in the center.

811
00:34:44,940 --> 00:34:47,500
But it has some very interesting
properties, even

812
00:34:47,500 --> 00:34:49,360
at the level of singular
neurons.

813
00:34:49,360 --> 00:34:50,650
So here's a depiction of an

814
00:34:50,650 --> 00:34:52,110
experiment, now, from a monkey.

815
00:34:52,110 --> 00:34:53,889
You're looking at a
single neuron or

816
00:34:53,889 --> 00:34:55,429
small group of neurons.

817
00:34:55,429 --> 00:34:57,370
And you present something
like a hand.

818
00:34:57,370 --> 00:35:00,170
And you can see that those
neurons really like the hand.

819
00:35:00,170 --> 00:35:03,850
Now, don't forget, for a primary
visual cortex, it just

820
00:35:03,850 --> 00:35:04,390
sees lines.

821
00:35:04,390 --> 00:35:06,430
It doesn't think about
the entire objects.

822
00:35:06,430 --> 00:35:08,740
But by the time you get to the
higher levels of vision, it's

823
00:35:08,740 --> 00:35:09,840
encoding an entire object.

824
00:35:09,840 --> 00:35:11,870
And then, you can do experiments
that basically

825
00:35:11,870 --> 00:35:15,530
interrogate, what does this
neuron discover in the world?

826
00:35:15,530 --> 00:35:17,270
What is it interested in?

827
00:35:17,270 --> 00:35:20,890
So if you turn the hand over,
it's still very interested.

828
00:35:20,890 --> 00:35:22,900
You show it a less
detailed hand.

829
00:35:22,900 --> 00:35:23,790
It's still very interested.

830
00:35:23,790 --> 00:35:26,900
Maybe a bit less, but it's
still very interested.

831
00:35:26,900 --> 00:35:29,190
A hand, this way, where all
the spatial relations have

832
00:35:29,190 --> 00:35:31,990
changed still recognizes
a hand.

833
00:35:31,990 --> 00:35:34,920
It still recognizes a hand with
a bit less enthusiasm.

834
00:35:34,920 --> 00:35:38,660
But now, a mitten, which kind
of looks like a hand.

835
00:35:38,660 --> 00:35:40,200
That barely counts
as a hand at all.

836
00:35:40,200 --> 00:35:43,800
That neuron has lost interest
in coding that as a hand.

837
00:35:43,800 --> 00:35:45,170
And you're showed other
things that have some

838
00:35:45,170 --> 00:35:47,100
similarity in shape.

839
00:35:47,100 --> 00:35:48,520
Because these all have
four elements in

840
00:35:48,520 --> 00:35:49,920
them, like four fingers.

841
00:35:49,920 --> 00:35:51,230
This neuron's not responding
at all.

842
00:35:51,230 --> 00:35:53,970
It's not that it has four
things or lines.

843
00:35:53,970 --> 00:35:55,290
It's a hand.

844
00:35:55,290 --> 00:35:56,390
And then, you could even
worry about things.

845
00:35:56,390 --> 00:35:57,940
Well, maybe this, because
it's from a person.

846
00:35:57,940 --> 00:36:01,450
But if they see the face, this
neuron doesn't care.

847
00:36:01,450 --> 00:36:03,400
So it's nothing about
the gross shape

848
00:36:03,400 --> 00:36:04,540
or that it's a person.

849
00:36:04,540 --> 00:36:08,500
This neuron is specialized,
foveally, for spotting a hand.

850
00:36:08,500 --> 00:36:10,580
And not only that, these kinds
of neurons already have the

851
00:36:10,580 --> 00:36:14,080
properties we described that are
big problems for vision.

852
00:36:14,080 --> 00:36:17,100
So here's a neuron that's
responding to an upright face.

853
00:36:19,860 --> 00:36:23,020
And then it's responding if the
face is turned on its side

854
00:36:23,020 --> 00:36:24,050
pretty well.

855
00:36:24,050 --> 00:36:26,670
The face is turned upside down,
a little less well, but

856
00:36:26,670 --> 00:36:28,680
it's still responding
to a face.

857
00:36:28,680 --> 00:36:31,290
If the face is distant,
it's still responding.

858
00:36:31,290 --> 00:36:33,120
If the face changes expression,
it's still

859
00:36:33,120 --> 00:36:34,704
responding.

860
00:36:34,704 --> 00:36:37,450
And if we put a green filter on
it, so you have a Martian

861
00:36:37,450 --> 00:36:38,690
person, it's still responding.

862
00:36:38,690 --> 00:36:41,980
So all kinds of ways that a
face changes in the world,

863
00:36:41,980 --> 00:36:43,170
this neuron is still firing.

864
00:36:43,170 --> 00:36:46,520
And interestingly, this neuron
was firing for one of the

865
00:36:46,520 --> 00:36:47,790
experimenters.

866
00:36:47,790 --> 00:36:50,430
But it wasn't too interested
in the other experimenter.

867
00:36:50,430 --> 00:36:53,170
So it's a neuron that's
generalizing all the different

868
00:36:53,170 --> 00:36:55,260
views that you might
have of a person.

869
00:36:55,260 --> 00:36:57,930
But it's responding to one
neuron versus another.

870
00:36:57,930 --> 00:37:00,940
And there's been a lot
of fun in monkey

871
00:37:00,940 --> 00:37:01,840
neurophysiology stuff.

872
00:37:01,840 --> 00:37:05,840
There's a so-called Jennifer
Aniston cell, where

873
00:37:05,840 --> 00:37:07,740
researchers have discovered
neurons, in monkeys, that seem

874
00:37:07,740 --> 00:37:10,280
to respond to Jennifer Aniston,
for whatever reason.

875
00:37:10,280 --> 00:37:14,760
And other specific
famous people.

876
00:37:14,760 --> 00:37:17,690
Although, not particularly
meaningful to the primates.

877
00:37:17,690 --> 00:37:20,920
And of course, in human brain
imaging, you can do tasks that

878
00:37:20,920 --> 00:37:23,045
emphasize what an object
is or where it is.

879
00:37:23,045 --> 00:37:25,505
And corresponding to the monkey
work, we see activation

880
00:37:25,505 --> 00:37:29,040
in this where pathway towards
the parietal cortex.

881
00:37:29,040 --> 00:37:30,450
If you have to make a
spatial judgment,

882
00:37:30,450 --> 00:37:32,030
where things are located.

883
00:37:32,030 --> 00:37:34,760
Or this lower temporal
lobe, what pathway.

884
00:37:34,760 --> 00:37:37,135
You have to make a judgment
about what object you're

885
00:37:37,135 --> 00:37:38,550
looking at.

886
00:37:38,550 --> 00:37:41,880
So that lines up in intact
humans very much, with what we

887
00:37:41,880 --> 00:37:46,410
see in primates with
invasive studies.

888
00:37:46,410 --> 00:37:49,710
So we're going to end today
by discussing two kinds of

889
00:37:49,710 --> 00:37:54,480
lesions in humans and some
spectacular impairments in the

890
00:37:54,480 --> 00:37:55,590
what or where pathways.

891
00:37:55,590 --> 00:37:58,360
So let me get you ready for
these films for a moment.

892
00:37:58,360 --> 00:38:02,480
So one film you're going to see
is the patient who has a

893
00:38:02,480 --> 00:38:05,950
great injury to the parts of
the brain in the parietal

894
00:38:05,950 --> 00:38:09,490
cortex that serve our
where system.

895
00:38:09,490 --> 00:38:12,470
So this is so-called Balint's
syndrome, in the Neurology.

896
00:38:12,470 --> 00:38:14,400
These are bilateral injuries.

897
00:38:14,400 --> 00:38:15,910
That means on both sides
of the brain, in

898
00:38:15,910 --> 00:38:17,780
the parietal region.

899
00:38:17,780 --> 00:38:19,830
They're pretty good at knowing
what something is, because the

900
00:38:19,830 --> 00:38:22,570
temporal lobe pathway, the
what pathway, is intact.

901
00:38:22,570 --> 00:38:24,480
But they have problems in
knowing where things are,

902
00:38:24,480 --> 00:38:26,930
reaching for things, where to
put their gaze, estimating

903
00:38:26,930 --> 00:38:29,090
distances, and navigation,
in general.

904
00:38:29,090 --> 00:38:32,440
So I'll show you an example
of a patient like this.

905
00:38:32,440 --> 00:38:35,640
And then the converse, a patient
who has big problems

906
00:38:35,640 --> 00:38:36,410
in the what system.

907
00:38:36,410 --> 00:38:39,970
And we'll start with that.

908
00:38:39,970 --> 00:38:42,530
And I'll tell you just one last
bit of information to

909
00:38:42,530 --> 00:38:43,100
think about.

910
00:38:43,100 --> 00:38:45,140
That the what and where
distinction, sometimes, is not

911
00:38:45,140 --> 00:38:48,940
as complete in every way
as you might imagine.

912
00:38:48,940 --> 00:38:51,140
And so, sometimes people have
said, really, the where system

913
00:38:51,140 --> 00:38:55,060
is better described as not
where things are, but the

914
00:38:55,060 --> 00:38:58,770
information you need for
physical action in the world,

915
00:38:58,770 --> 00:38:59,830
which is very close to where.

916
00:38:59,830 --> 00:39:00,720
But it's a little
bit different.

917
00:39:00,720 --> 00:39:04,280
And let me give you the kind
of what type of information

918
00:39:04,280 --> 00:39:07,320
that the where system still
seems to have in it.

919
00:39:07,320 --> 00:39:10,580
So here's a patient with
extensive damage in the what

920
00:39:10,580 --> 00:39:12,870
system, in the lower parts
of the brain and

921
00:39:12,870 --> 00:39:14,580
posteriorly for vision.

922
00:39:14,580 --> 00:39:17,300
Terrible perception of shapes
and orientation of shapes.

923
00:39:17,300 --> 00:39:21,220
Very bad what ability.

924
00:39:21,220 --> 00:39:24,300
And she was asked, then, to
reach with an envelope to a

925
00:39:24,300 --> 00:39:27,790
slot at different
orientations.

926
00:39:27,790 --> 00:39:30,820
And what's striking is even
though if her hand was a

927
00:39:30,820 --> 00:39:33,670
distance away-- so if she would
have to think, what is

928
00:39:33,670 --> 00:39:35,260
the angle I would need
to approach it with--

929
00:39:35,260 --> 00:39:36,320
she was terrible.

930
00:39:36,320 --> 00:39:39,550
When she actually moved her hand
towards the slot, she was

931
00:39:39,550 --> 00:39:40,910
surprisingly good.

932
00:39:40,910 --> 00:39:43,640
So let me show you
the feeling.

933
00:39:43,640 --> 00:39:47,790
So if this patient would have
to put her hand so that this

934
00:39:47,790 --> 00:39:50,320
letter would fit into
this slot--

935
00:39:50,320 --> 00:39:56,780
she was very poor at spatial
relations and knowing what

936
00:39:56,780 --> 00:39:57,650
things were--

937
00:39:57,650 --> 00:39:59,450
she couldn't line
them up at all.

938
00:39:59,450 --> 00:40:01,370
But now, they said, well, please
put your hand out and

939
00:40:01,370 --> 00:40:02,500
just stick it in.

940
00:40:02,500 --> 00:40:05,977
And as her hand approaches it,
she changes orientations and

941
00:40:05,977 --> 00:40:07,280
is correct.

942
00:40:07,280 --> 00:40:10,320
Because in order to do things,
like to grab something--

943
00:40:10,320 --> 00:40:10,780
think about it.

944
00:40:10,780 --> 00:40:13,760
When you grab something, like a
pencil or a cup, you need a

945
00:40:13,760 --> 00:40:15,260
little bit of knowledge
to know what it is.

946
00:40:15,260 --> 00:40:18,230
You wouldn't grab things
similarly, whether they could

947
00:40:18,230 --> 00:40:21,320
spill easily or be pointy
or painful.

948
00:40:21,320 --> 00:40:24,550
A little bit what information
helps you guide even something

949
00:40:24,550 --> 00:40:27,790
like a simple reaching, which
is a special task.

950
00:40:27,790 --> 00:40:29,640
So it's not so much, maybe,
that the where

951
00:40:29,640 --> 00:40:32,050
system is only where.

952
00:40:32,050 --> 00:40:34,600
But it has just enough
information to guide actions

953
00:40:34,600 --> 00:40:37,780
in the world, which has a tiny
bit of information about what

954
00:40:37,780 --> 00:40:39,600
things are that you're
jumping over.

955
00:40:39,600 --> 00:40:41,240
If you're going to run into
something, is it likely to be

956
00:40:41,240 --> 00:40:42,490
soft or hard.