TEDIndia 2009
Pawan Sinha: How brains learn to see
Pawan Sinha 之大腦如何學會看
Filmed:
Readability: 4.5
939,209 views
Pawan Sinha 詳細闡述他在大腦視覺系統發展的開創性研究。Sinha以及他的研究團隊為出生即看不見的孩子提供免費的視力恢復治療，並研究他們的大腦如何學習判讀視覺資料。這些工作也為神經科學、工程學，甚至自閉症提供了深入見解。
Pawan Sinha - Visual neuroscientist
Pawan Sinha researches how our brains interpret what our eyes see -- and uses that research to give blind children the gift of sight. Full bio
Pawan Sinha researches how our brains interpret what our eyes see -- and uses that research to give blind children the gift of sight. Full bio
Double-click the English transcript below to play the video.
00:15
If you are a blind child in India,
0
0
4000
如果您是在印度的盲童，
00:19
you will very likely have to contend with
1
4000
3000
您將很有可能必須面臨
00:22
at least two big pieces of bad news.
2
7000
3000
至少兩個重大的壞消息。
00:25
The first bad news
3
10000
2000
第一個壞消息是
00:27
is that the chances of getting treatment
4
12000
3000
得到治療的機會
00:30
are extremely slim to none,
5
15000
3000
幾乎渺茫至零，
00:33
and that's because most of the blindness
6
18000
2000
這是因為在這國家大多數的
00:35
alleviation programs in the country
7
20000
2000
盲人扶助專案
00:37
are focused on adults,
8
22000
2000
都只關注成人，
00:39
and there are very, very few hospitals
9
24000
3000
而只有非常非常少的醫院
00:42
that are actually equipped to treat children.
10
27000
3000
真正有能力治療病童。
00:46
In fact, if you were to be treated,
11
31000
4000
事實上，如果您接受治療，
00:51
you might well end up being treated
12
36000
3000
您可能最後是被
00:54
by a person who has no medical credentials
13
39000
3000
沒有醫療證書的人員治療
00:57
as this case from Rajasthan illustrates.
14
42000
3000
就像拉賈斯坦所揭發的這個實例。
01:00
This is a three-year-old orphan girl
15
45000
2000
這個三歲大孤女
01:02
who had cataracts.
16
47000
2000
患了白內障。
01:04
So, her caretakers took her
17
49000
2000
所以她的看護帶她
01:06
to the village medicine man,
18
51000
2000
到村醫那兒，
01:08
and instead of suggesting to the caretakers
19
53000
3000
然而村醫卻不是建議看護
01:11
that the girl be taken to a hospital,
20
56000
3000
帶小女孩去正規醫院，
01:14
the person decided to burn her abdomen
21
59000
2000
相反地那人決定用燒紅的鐵棒
01:16
with red-hot iron bars
22
61000
2000
炙燙小女孩的腹部
01:18
to drive out the demons.
23
63000
2000
以驅逐惡靈。
01:20
The second piece of bad news
24
65000
3000
接下來第二個壞消息
01:23
will be delivered to you
25
68000
2000
是從神經學家得來，
01:25
by neuroscientists, who will tell you
26
70000
3000
他將告訴您
01:28
that if you are older than four or five years of age,
27
73000
3000
如果您的年紀超過四或五歲
01:31
that even if you have your eye corrected,
28
76000
3000
即使您接受視力矯正
01:34
the chances of your brain learning how to see
29
79000
3000
您大腦學習如何解析所見的機會
01:37
are very, very slim --
30
82000
2000
也十分十分地渺茫
01:39
again, slim or none.
31
84000
3000
再一次地，渺茫或零。
01:42
So when I heard these two things,
32
87000
2000
所以當我知道這兩件事情
01:44
it troubled me deeply,
33
89000
2000
我同時因為個人
01:46
both because of personal reasons
34
91000
2000
以及科學因素
01:48
and scientific reasons.
35
93000
2000
深深感到困擾。
01:50
So let me first start with the personal reason.
36
95000
3000
讓我從個人因素開始說起吧
01:53
It'll sound corny, but it's sincere.
37
98000
3000
這聽起來像陳腔濫調，但是由衷的。
01:56
That's my son, Darius.
38
101000
2000
這是我兒子，Darius。
01:58
As a new father,
39
103000
2000
身為一個新手父親，
02:00
I have a qualitatively different sense
40
105000
4000
對於寶寶有多脆弱
02:04
of just how delicate babies are,
41
109000
3000
我們對他們有多少義務
02:07
what our obligations are towards them
42
112000
3000
還有我們能感覺到對一個孩子
02:10
and how much love
43
115000
2000
有多麼深的愛
02:12
we can feel towards a child.
44
117000
3000
有截然不同而美好的感受
02:15
I would move heaven and earth
45
120000
2000
我可能會為了治療Darius
02:17
in order to get treatment for Darius,
46
122000
3000
在天堂和現實間穿梭。
02:20
and for me to be told
47
125000
2000
對我而言，被告知
02:22
that there might be other Dariuses
48
127000
2000
別的地方還有更多像Darius這樣的孩子
02:24
who are not getting treatment,
49
129000
2000
並沒有得到治療
02:26
that's just viscerally wrong.
50
131000
3000
這直覺地就是個錯誤。
02:29
So that's the personal reason.
51
134000
2000
以上是我的個人因素。
02:31
Scientific reason is that this notion
52
136000
3000
科學方面
02:34
from neuroscience of critical periods --
53
139000
2000
神經科學有個關鍵時期之說
02:36
that if the brain is older
54
141000
3000
也就是，如果大腦年齡大於
02:39
than four or five years of age,
55
144000
2000
四歲或五歲，
02:41
it loses its ability to learn --
56
146000
2000
大腦就喪失學習能力這個觀點。
02:43
that doesn't sit well with me,
57
148000
2000
這無法說服我，
02:45
because I don't think that idea
58
150000
2000
因為我不認為這想法
02:47
has been tested adequately.
59
152000
3000
被充分地驗證過。
02:50
The birth of the idea is from
60
155000
2000
這想法是從
02:52
David Hubel and Torsten Wiesel's work,
61
157000
2000
David Hube和Torsten Wiesel的研究成果孕育而來，
02:54
two researchers who were at Harvard,
62
159000
2000
這兩位科學家曾在哈佛任職
02:56
and they got the Nobel Prize in 1981
63
161000
3000
並且在1981年因為他們視覺生理學
02:59
for their studies of visual physiology,
64
164000
2000
的研究得到諾貝爾獎，
03:01
which are remarkably beautiful studies,
65
166000
2000
那的確是傑出而漂亮的研究，
03:03
but I believe some of their work
66
168000
2000
但我相信其中的一部分成果
03:05
has been extrapolated
67
170000
2000
已被貿然地
03:07
into the human domain prematurely.
68
172000
2000
用於人類領域
03:09
So, they did their work with kittens,
69
174000
2000
他們用不同治療剝奪
03:11
with different kinds of deprivation regiments,
70
176000
2000
的小貓群組進行實驗
03:13
and those studies,
71
178000
2000
而這些可追溯到
03:15
which date back to the '60s,
72
180000
2000
六零年代的研究
03:17
are now being applied to human children.
73
182000
3000
現今被套用於人類孩子上。
03:20
So I felt that I needed to do two things.
74
185000
3000
所以我覺得我需要做兩件事。
03:23
One: provide care
75
188000
3000
一是提供
03:26
to children who are currently
76
191000
2000
目前被剝奪治療的孩子
03:28
being deprived of treatment.
77
193000
2000
醫療照護。
03:30
That's the humanitarian mission.
78
195000
2000
這是基於人道主義的使命。
03:32
And the scientific mission would be
79
197000
2000
而科學的使命則是
03:34
to test the limits
80
199000
2000
試探視覺可塑性
03:36
of visual plasticity.
81
201000
2000
的極限
03:38
And these two missions, as you can tell,
82
203000
3000
你可以看出，這兩個任務
03:41
thread together perfectly. One adds to the other;
83
206000
3000
完美地結合在一起，相輔相成。
03:44
in fact, one would be impossible without the other.
84
209000
3000
事實上是缺一不可。
03:49
So, to implement
85
214000
2000
為了執行
03:51
these twin missions,
86
216000
2000
這個雙生任務，
03:53
a few years ago, I launched Project Prakash.
87
218000
3000
幾年前我開始了Prakash計畫
03:56
Prakash, as many of you know,
88
221000
2000
Prakash，也許你們很多人都知道，
03:58
is the Sanskrit word for light,
89
223000
2000
是梵文中光的總稱
04:00
and the idea is that
90
225000
2000
我們的理念亦是如此
04:02
in bringing light into the lives of children,
91
227000
3000
將光明引入孩子們的生命中，
04:05
we also have a chance
92
230000
2000
同時也有
04:07
of shedding light on some of the
93
232000
2000
一窺某些
04:09
deepest mysteries of neuroscience.
94
234000
3000
神經科學奧秘的機會
04:12
And the logo -- even though it looks extremely Irish,
95
237000
3000
再看看這標誌，雖然它看起來很像幸運草
04:15
it's actually derived from
96
240000
2000
事實上它是源自
04:17
the Indian symbol of Diya, an earthen lamp.
97
242000
4000
一種印度陶燈 Diya 的象徵
04:21
The Prakash, the overall effort
98
246000
3000
Prakash整體工作內容
04:24
has three components:
99
249000
2000
包含三個部分，
04:26
outreach, to identify children in need of care;
100
251000
4000
宣傳理念、定義病童是否需要照護、醫藥治療，
04:30
medical treatment; and in subsequent study.
101
255000
3000
以及後續追蹤研究。
04:33
And I want to show you a short video clip
102
258000
3000
我想要在此分享一段節錄短片
04:36
that illustrates the first two components of this work.
103
261000
3000
來說明前兩個部份的工作。
04:41
This is an outreach station
104
266000
2000
這是一個由為盲童設立的學校
04:43
conducted at a school for the blind.
105
268000
3000
的宣傳站。
04:46
(Text: Most of the children are profoundly and permanently blind ...)
106
271000
5000
(字幕: 大多數的孩童嚴重且永久地失明...
04:51
Pawan Sinha: So, because this is a school for the blind,
107
276000
5000
Pawan Sinha: 因為這是間為盲人而辦的學校，
04:56
many children have permanent conditions.
108
281000
2000
許多盲童是永久地失明。
04:58
That's a case of microphthalmos,
109
283000
3000
這是一個小眼症的例子
05:01
which is malformed eyes,
110
286000
2000
也就是有畸形的眼睛，
05:03
and that's a permanent condition;
111
288000
2000
並且是永久性的。
05:05
it cannot be treated.
112
290000
2000
它沒有辦法被治癒。
05:07
That's an extreme of micropthalmos
113
292000
2000
這和極端的小眼症
05:09
called enophthalmos.
114
294000
2000
稱為眼球陷沒。
05:11
But, every so often, we come across children
115
296000
2000
但偶爾，我們遇見有
05:13
who show some residual vision,
116
298000
3000
殘存視力的孩童
05:16
and that is a very good sign
117
301000
3000
這是很好的徵兆
05:19
that the condition might actually be treatable.
118
304000
2000
顯示病情實際上可能是可治療的。
05:21
So, after that screening, we bring the children to the hospital.
119
306000
3000
於是經過篩檢，我們帶這些病童到醫院去。
05:24
That's the hospital we're working with in Delhi,
120
309000
2000
這是我們在德里工作的醫院，
05:26
the Schroff Charity Eye Hospital.
121
311000
3000
修福慈善眼科醫院
05:29
It has a very well-equipped
122
314000
2000
裏頭有非常裝備完善的
05:31
pediatric ophthalmic center,
123
316000
3000
小兒眼科中心，
05:35
which was made possible in part
124
320000
2000
這得以實現有部分
05:37
by a gift from the Ronald McDonald charity.
125
322000
4000
來自麥當勞叔叔慈善基金的贈與。
05:41
So, eating burgers actually helps.
126
326000
3000
所以吃漢堡的確可助一臂之力。
05:45
(Text: Such examinations allow us to improve
127
330000
2000
(字幕: 這樣的檢查允許我們增進
05:47
eye-health in many children, and ...
128
332000
2000
許多孩童的眼睛健康，並且
05:54
... help us find children who can participate in Project Prakash.)
129
339000
3000
幫助我們找到可參與Prakash計畫的孩子。)
05:57
PS: So, as I zoom in to the eyes of this child,
130
342000
2000
補充一下，當我放大這個孩子眼睛的影像，
05:59
you will see the cause of his blindness.
131
344000
3000
你將可以看到他失明的原因。
06:03
The whites that you see in the middle of his pupils
132
348000
3000
你看到在瞳孔中間白色部分
06:06
are congenital cataracts,
133
351000
3000
是先天性的白內障，
06:09
so opacities of the lens.
134
354000
2000
所以水晶體是不透明的。
06:11
In our eyes, the lens is clear,
135
356000
3000
我們的眼睛水晶體則是透明的。
06:14
but in this child, the lens has become opaque,
136
359000
2000
但是這個孩子的水晶體已經變得不透光，
06:16
and therefore he can't see the world.
137
361000
3000
因此他沒有辦法看見這世界。
06:19
So, the child is given treatment. You'll see shots of the eye.
138
364000
3000
所以這孩子接受治療。你所看到的是眼球置換手術。
06:22
Here's the eye with the opaque lens,
139
367000
2000
這是有不透明水晶體的眼球，
06:24
the opaque lens extracted
140
369000
2000
不透明的水晶體被吸取出
06:26
and an acrylic lens inserted.
141
371000
3000
然後置入丙烯酸人工水晶體。
06:29
And here's the same child
142
374000
2000
這是剛剛那孩子
06:31
three weeks post-operation,
143
376000
3000
在接受手術後三個星期，
06:34
with the right eye open.
144
379000
3000
使用右眼的情形。
06:40
(Applause)
145
385000
6000
(鼓掌)
06:46
Thank you.
146
391000
2000
謝謝大家。
06:48
So, even from that little clip, you can begin to get the sense
147
393000
3000
所以，即使從這小片段，你也能開始感覺到
06:51
that recovery is possible,
148
396000
2000
治癒是可能的，
06:53
and we have now
149
398000
2000
而我們已經
06:55
provided treatment to over 200 children,
150
400000
3000
對超過兩百位孩童施予治療
06:58
and the story repeats itself.
151
403000
2000
而成功復明故事不斷重演。
07:00
After treatment, the child
152
405000
2000
經過治療，孩子
07:02
gains significant functionality.
153
407000
3000
得到重要的視覺官能。
07:05
In fact, the story holds true
154
410000
3000
事實上，這情節即使對你知道的
07:08
even if you have a person who got sight
155
413000
2000
某個被剝奪視力多年的人
07:10
after several years of deprivation.
156
415000
2000
也同樣為真。
07:12
We did a paper a few years ago
157
417000
2000
我們幾年前撰寫了一篇報導
07:14
about this woman that you see on the right, SRD,
158
419000
4000
是有關你看到在右邊，得了SRD(漿液性網膜剝離)的女人，
07:18
and she got her sight late in life,
159
423000
2000
她在晚年復明，
07:20
and her vision is remarkable at this age.
160
425000
4000
而且在這年紀她的視力是相當好的。
07:24
I should add a tragic postscript to this --
161
429000
3000
我在這故事末加個悲劇附錄。
07:27
she died two years ago
162
432000
2000
她兩年前因為一場巴士意外
07:29
in a bus accident.
163
434000
2000
而去世了。
07:31
So, hers is just a truly inspiring story --
164
436000
4000
她的故事正是非常激勵人心的，
07:35
unknown, but inspiring story.
165
440000
3000
不為人知，但激勵人心的故事。
07:38
So when we started finding these results,
166
443000
2000
當我們開始發掘這些成果，
07:40
as you might imagine, it created quite a bit of stir
167
445000
3000
就像你想像的，它在科學以及普及的
07:43
in the scientific and the popular press.
168
448000
3000
出版物造成旋風。
07:46
Here's an article in Nature
169
451000
2000
這是在自然期刊側寫這工作
07:48
that profiled this work,
170
453000
2000
的一篇文章，
07:50
and another one in Time.
171
455000
2000
另一篇則是在時代雜誌。
07:52
So, we were fairly convinced -- we are convinced --
172
457000
2000
我們非常確信，我們確信
07:54
that recovery is feasible,
173
459000
2000
儘管長久地失去視力。
07:56
despite extended visual deprivation.
174
461000
3000
復明是可能的。
07:59
The next obvious question to ask:
175
464000
2000
下一個要問且顯而易見的問題是：
08:01
What is the process of recovery?
176
466000
3000
痊癒的過程如何發生？
08:04
So, the way we study that is,
177
469000
3000
所以，我們研究的方法是，
08:07
let's say we find a child who has light sensitivity.
178
472000
2000
假設我們找到一個對光線有反應的孩子
08:09
The child is provided treatment,
179
474000
2000
提供他治療，
08:11
and I want to stress that the treatment
180
476000
2000
我要強調這治療
08:13
is completely unconditional;
181
478000
2000
是完全無條件的。
08:15
there is no quid pro quo.
182
480000
2000
是無償的。
08:17
We treat many more children then we actually work with.
183
482000
3000
我們治療的孩童比參與計畫的更多。
08:20
Every child who needs treatment is treated.
184
485000
3000
每個需要醫療的孩子都被治療。
08:23
After treatment, about every week,
185
488000
2000
治療後，約每個星期一次，
08:25
we run the child
186
490000
2000
我們讓孩童
08:27
on a battery of simple visual tests
187
492000
3000
接受簡單視覺測試
08:30
in order to see how their visual skills
188
495000
2000
以了解他們看的技能
08:32
are coming on line.
189
497000
2000
是否如預期恢復。
08:34
And we try to do this for as long as possible.
190
499000
3000
我們試著盡可能長時間追蹤。
08:37
This arc of development
191
502000
2000
這些發展曲線，
08:39
gives us unprecedented
192
504000
2000
給了我們有關
08:41
and extremely valuable information
193
506000
2000
如何建構視覺
08:43
about how the scaffolding of vision
194
508000
2000
不同以往且十分有價值
08:45
gets set up.
195
510000
2000
的資訊
08:47
What might be the causal connections
196
512000
2000
早期發展的視覺技能和
08:49
between the early developing skills
197
514000
2000
較晚發展的兩者之間的
08:51
and the later developing ones?
198
516000
2000
因果關聯為何？
08:53
And we've used this general approach to study
199
518000
2000
我們已經用這套公認的研究方法來研究
08:55
many different visual proficiencies,
200
520000
3000
許多不同的視覺熟練程度，
08:58
but I want to highlight one particular one,
201
523000
4000
但我想要強調其中特別的一項，
09:02
and that is image parsing into objects.
202
527000
3000
那就是將圖像解析為物體的技能。
09:05
So, any image of the kind that you see on the left,
203
530000
2000
在這兒，左邊任何一種你所見的圖，
09:07
be it a real image or a synthetic image,
204
532000
3000
是真實影像還是合成的?
09:10
it's made up of little regions
205
535000
2000
這是由中間
09:12
that you see in the middle column,
206
537000
2000
那張有不同顏色、不同照度
09:14
regions of different colors, different luminances.
207
539000
3000
的區域合併起來做成的。
09:17
The brain has this complex task
208
542000
3000
大腦的複雜任務包括
09:20
of putting together, integrating,
209
545000
3000
組合、整合，
09:23
subsets of these regions
210
548000
2000
將這些區域統整
09:25
into something that's more meaningful,
211
550000
2000
成為更有意義的某樣物品，
09:27
into what we would consider to be objects,
212
552000
2000
是我們認知的實物，
09:29
as you see on the right.
213
554000
2000
就像右邊你所看到的那樣。
09:31
And nobody knows how this integration happens,
214
556000
2000
而沒有人知道這些整合工作如何發生。
09:33
and that's the question we asked with Project Prakash.
215
558000
4000
這就是我們在Prakash計畫中所問的問題。
09:37
So, here's what happens
216
562000
2000
這就是發生歷程
09:39
very soon after the onset of sight.
217
564000
3000
一旦開始建構視覺便變化快速。
09:42
Here's a person who had gained sight just a couple of weeks ago,
218
567000
3000
這個人他才在兩個星期前恢復視力，
09:45
and you see Ethan Myers, a graduate student from MIT,
219
570000
3000
再看到Ethan Myers，來自麻省理工的大學生，
09:48
running the experiment with him.
220
573000
3000
正在和他一起做實驗。
09:51
His visual-motor coordination is quite poor,
221
576000
4000
他的視覺運動協調十分地差，
09:55
but you get a general sense
222
580000
2000
但你大約可知
09:57
of what are the regions that he's trying to trace out.
223
582000
3000
他正試著描繪的區域為何
10:00
If you show him real world images,
224
585000
2000
如果向他展示真實世界的影像
10:02
if you show others like him real world images,
225
587000
3000
如果你向其他跟這個人相似的病友展示真實世界的影像，
10:05
they are unable to recognize most of the objects
226
590000
2000
他們不能辨識出大部分的物體
10:07
because the world to them is over-fragmented;
227
592000
3000
因為這樣的世界對他們來說太支離破碎。
10:10
it's made up of a collage, a patchwork,
228
595000
3000
他們所見是由不同顏色、照明的
10:13
of regions of different colors and luminances.
229
598000
2000
小區塊拼貼、補綴而成
10:15
And that's what's indicated in the green outlines.
230
600000
2000
而這就是綠色輪廓所表示的東西。
10:17
When you ask them,
231
602000
2000
當你要求他們，
10:19
"Even if you can't name the objects, just point to where the objects are,"
232
604000
3000
即使他們不能說出物體的名字，僅僅指出在哪裡就好，
10:22
these are the regions that they point to.
233
607000
2000
這些就是他們所指的地方。
10:24
So the world is this complex
234
609000
2000
所以世界是如此複雜的
10:26
patchwork of regions.
235
611000
2000
區塊拼貼。
10:28
Even the shadow on the ball
236
613000
2000
即使是球上面的陰影
10:30
becomes its own object.
237
615000
3000
也是一個獨立的物件。
10:33
Interestingly enough,
238
618000
2000
非常有趣的是，
10:35
you give them a few months,
239
620000
2000
你給他們幾個月的時間，
10:37
and this is what happens.
240
622000
3000
這就是後來所發生的事。
10:43
Doctor: How many are these?
241
628000
2000
醫生：這裡有多少個?
10:45
Patient: These are two things.
242
630000
2000
病人：這裡有兩個東西。
10:47
Doctor: What are their shapes?
243
632000
2000
醫生：他們的形狀是?
10:49
Patient: Their shapes ...
244
634000
2000
病人：他們的形狀...
10:51
This one is a circle,
245
636000
3000
這個是圓的，
10:54
and this
246
639000
2000
這個嘛
10:56
is a square.
247
641000
2000
是方的。
10:58
PS: A very dramatic transformation has come about.
248
643000
3000
附註：非常戲劇化的轉變已經發生。
11:01
And the question is:
249
646000
2000
而問題是：
11:03
What underlies this transformation?
250
648000
2000
這轉變是根據什麼？
11:05
It's a profound question,
251
650000
2000
這是個深奧的問題，
11:07
and what's even more amazing is how simple
252
652000
2000
而令人驚訝的是答案有
11:09
the answer is.
253
654000
2000
多麼地簡單。
11:11
The answer lies in motion
254
656000
2000
答案就是移動
11:13
and that's what I want to show you in the next clip.
255
658000
3000
這也是我想要在下個影片中呈現給大家的。
11:18
Doctor: What shape do you see here?
256
663000
2000
醫生: 你在這兒看到的是什麼形狀?
11:20
Patient: I can't make it out.
257
665000
3000
病患: 我無法回答這個問題。
11:28
Doctor: Now?
258
673000
2000
醫生: 現在呢?
11:31
Patient: Triangle.
259
676000
2000
病患: 三角形。
11:35
Doctor: How many things are these?
260
680000
3000
醫生: 這裡有多少東西?
11:48
Now, how many things are these?
261
693000
3000
現在，這裡有多少東西?
11:51
Patient: Two.
262
696000
2000
病患: 兩個。
11:53
Doctor: What are these things?
263
698000
2000
醫生: 這些東西是什麼呢?
11:56
Patient: A square and a circle.
264
701000
2000
病患: 一個方塊跟一個圓。
11:58
PS: And we see this pattern over and over again.
265
703000
3000
附帶說明，我們看到這個模式不斷重複。
12:01
The one thing the visual system needs
266
706000
3000
視覺系統所需
12:04
in order to begin parsing the world
267
709000
2000
解析世界的要件就是
12:06
is dynamic information.
268
711000
2000
動態資訊。
12:08
So the inference we are deriving from this,
269
713000
2000
從這結果和多次這類的實驗
12:10
and several such experiments,
270
715000
2000
我們推論，
12:12
is that dynamic information processing,
271
717000
2000
動態資訊的處理
12:14
or motion processing,
272
719000
2000
或動作資訊處裡，
12:16
serves as the bedrock for building
273
721000
2000
就像是建立後續複雜的
12:18
the rest of the complexity of visual processing;
274
723000
4000
視覺處理流程的基石。
12:22
it leads to visual integration
275
727000
2000
它成就了視覺整合
12:24
and eventually to recognition.
276
729000
3000
以及最終的辨認能力。
12:27
This simple idea has far reaching implications.
277
732000
3000
這簡單的概念在科學研究上有深遠的含意。
12:30
And let me just quickly mention two,
278
735000
3000
讓我快速地簡介其中兩個。
12:33
one, drawing from the domain of engineering,
279
738000
2000
一是從工程上來說，
12:35
and one from the clinic.
280
740000
2000
另一個是臨床上的意義。
12:37
So, from the perspective of engineering,
281
742000
2000
從工程的透視法，
12:39
we can ask: Goven that we know
282
744000
3000
我們能問，我們已知
12:42
that motion is so important for the human visual system,
283
747000
2000
動作對人類的視覺系統是如此重要，
12:44
can we use this as a recipe
284
749000
3000
我們可以利用這結果建立
12:47
for constructing machine-based vision systems
285
752000
3000
不需要程式設計師
12:50
that can learn on their own, that don't need to be programmed
286
755000
3000
設計程式就可以自我學習的
12:53
by a human programmer?
287
758000
2000
機械視覺系統嗎?
12:55
And that's what we're trying to do.
288
760000
2000
這就是我們正試著達成的。
12:57
I'm at MIT, at MIT you need to apply
289
762000
3000
我在麻省理工，在那兒你需要運用
13:00
whatever basic knowledge you gain.
290
765000
2000
任何你得到的基礎知識。
13:02
So we are creating Dylan,
291
767000
2000
所以我們創造了Dylan
13:04
which is a computational system
292
769000
2000
它是套有遠大目標
13:06
with an ambitious goal
293
771000
2000
的運算系統
13:08
of taking in visual inputs
294
773000
2000
可以接收的和一個人類孩童
13:10
of the same kind that a human child would receive,
295
775000
3000
相同的視覺輸入後，
13:13
and autonomously discovering:
296
778000
2000
自主地發現
13:15
What are the objects in this visual input?
297
780000
3000
這些視覺輸入有什麼東西。
13:18
So, don't worry about the internals of Dylan.
298
783000
3000
不用擔心Dylan裡面有什麼。
13:21
Here, I'm just going to talk about
299
786000
3000
在這裡我只想簡單提到
13:24
how we test Dylan.
300
789000
2000
我們如何測試Dylan。
13:26
The way we test Dylan is by giving it
301
791000
2000
我們測試它的方式是利用
13:28
inputs, as I said, of the same kind
302
793000
3000
給它輸入，像我先前說的，跟一個孩童，
13:31
that a baby, or a child in Project Prakash would get.
303
796000
3000
或是Prakash計畫中的孩子會得到的一樣。
13:34
But for a long time we couldn't quite figure out:
304
799000
3000
但是長久以來我們無法真正知道
13:37
Wow can we get these kinds of video inputs?
305
802000
3000
我們如何能的到這樣的視訊輸入。
13:41
So, I thought,
306
806000
2000
所以，我就想，
13:43
could we have Darius
307
808000
2000
我們是否能使Darius
13:45
serve as our babycam carrier,
308
810000
3000
像我們的嬰兒攝影機攜帶者，
13:48
and that way get the inputs that we feed into Dylan?
309
813000
3000
使我們得以將視訊輸入至Dylan。
13:51
So that's what we did.
310
816000
2000
這就是我們的方法。
13:53
(Laughter)
311
818000
7000
(眾笑)
14:00
I had to have long conversations with my wife.
312
825000
3000
我必須跟我妻子談很久。
14:03
(Laughter)
313
828000
5000
(眾笑)
14:08
In fact, Pam, if you're watching this,
314
833000
2000
事實上，Pam，如果你正看著，
14:10
please forgive me.
315
835000
2000
請原諒我。
14:13
So, we modified the optics of the camera
316
838000
4000
為了更生動模擬寶寶的視力
14:17
in order to mimic the baby's visual acuity.
317
842000
3000
我們修改了攝影機的光學元件。
14:20
As some of you might know,
318
845000
2000
也許你們有些人知道，
14:22
babyies are born pretty much legally blind.
319
847000
4000
寶寶剛出生的時候幾乎是看不見的。
14:26
Their acuity -- our acuity is 20/20;
320
851000
3000
寶寶看20英尺外的東西，
14:29
babies' acuity is like 20/800,
321
854000
3000
就像正常視力在800英尺看見的。
14:32
so they are looking at the world
322
857000
2000
所以他們是非常非常模糊地
14:34
in a very, very blurry fashion.
323
859000
3000
看著這個世界。
14:37
Here's what a baby-cam video looks like.
324
862000
3000
這就是嬰兒攝影機所見的影像。
14:41
(Laughter)
325
866000
9000
(眾大笑)
14:50
(Applause)
326
875000
3000
(鼓掌)
14:53
Thankfully, there isn't any audio
327
878000
2000
謝天謝地，這段影片
14:55
to go with this.
328
880000
3000
沒有任何聲音。
14:58
What's amazing is that working with such
329
883000
2000
令人驚訝的是，即使是這麼
15:00
highly degraded input,
330
885000
2000
高度劣化的影像輸入，
15:02
the baby, very quickly, is able
331
887000
2000
寶寶還是可以很快地
15:04
to discover meaning in such input.
332
889000
3000
知道影像的意義。
15:07
But then two or three days afterward,
333
892000
2000
經過兩三天，
15:09
babies begin to pay attention
334
894000
2000
寶寶開始把注意力放在
15:11
to their mother's or their father's face.
335
896000
2000
他們的爸爸或媽媽臉上。
15:13
How does that happen? We want Dylan to be able to do that,
336
898000
3000
這是如何發生的? 我們希望Dylan可以做到這樣。
15:16
and using this mantra of motion,
337
901000
3000
利用移動的奧秘，
15:19
Dylan actually can do that.
338
904000
2000
事實上Dylan可以做到，
15:21
So, given that kind of video input,
339
906000
3000
即使是那麼糟糕的影像輸入。
15:24
with just about six or seven minutes worth of video,
340
909000
3000
大概只有六或七分鐘可用的影片，
15:27
Dylan can begin to extract patterns
341
912000
3000
Dylan可以開始取得
15:30
that include faces.
342
915000
3000
包含臉部的圖樣。
15:33
So, it's an important demonstration
343
918000
2000
所以這是重要效果
15:35
of the power of motion.
344
920000
2000
的重要展示。
15:37
The clinical implication, it comes from the domain of autism.
345
922000
3000
臨床應用上，要從自閉症開始說起。
15:40
Visual integration has been associated with autism
346
925000
2000
視覺整合已經被好幾位研究學者證實
15:42
by several researchers.
347
927000
2000
和自閉症有關。
15:44
When we saw that, we asked:
348
929000
2000
當我們知道這結果，我們問：
15:46
Could the impairment in visual integration
349
931000
3000
是否動態資訊處理缺損
15:49
be the manifestation of something underneath,
350
934000
3000
就是視覺整合損傷
15:52
of dynamic information processing deficiencies in autism?
351
937000
3000
造成的自閉症的背後肇因
15:55
Because, if that hypothesis were to be true,
352
940000
3000
因為，若是這個假說成立，
15:58
it would have massive repercussions in our understanding
353
943000
3000
這將對我們了解
16:01
of what's causing the many different aspects
354
946000
2000
是什麼使得自閉症有不同方面的外顯特徵
16:03
of the autism phenotype.
355
948000
3000
有重要的回饋。
16:06
What you're going to see are
356
951000
2000
接下來的一段短片是
16:08
video clips of two children -- one neurotypical,
357
953000
3000
兩個孩子在玩乒乓球，一個非自閉症，
16:11
one with autism, playing Pong.
358
956000
2000
另一個有自閉症。
16:13
So, while the child is playing Pong, we are tracking where they're looking.
359
958000
3000
當孩子在玩壁球的時候，我們追蹤他們看著哪兒。
16:16
In red are the eye movement traces.
360
961000
3000
紅色的就是眼球移動的紀錄，
16:19
This is the neurotypical child, and what you see
361
964000
3000
這是非自閉症孩子的，而你所看見的是
16:22
is that the child is able to make cues
362
967000
2000
這個孩子可以利用
16:24
of the dynamic information
363
969000
2000
動態資訊給的暗示
16:26
to predict where the ball is going to go.
364
971000
2000
預測球會往那兒跑。
16:28
Even before the ball gets to a place,
365
973000
3000
即使在球到達之前，
16:31
the child is already looking there.
366
976000
3000
這孩子已經看著那邊。
16:34
Contrast this with a child
367
979000
2000
對照另一個玩著相同遊戲的
16:36
with autism playing the same game.
368
981000
2000
自閉症孩子。
16:38
Instead of anticipating,
369
983000
2000
這個孩子的視線總是
16:40
the child always follows where the ball has been.
370
985000
3000
跟著球到過的地方跑而非預測它。
16:43
The efficiency of the use
371
988000
2000
自閉症患者利用
16:45
of dynamic information
372
990000
2000
動態資訊的效率
16:47
seems to be significantly compromised in autism.
373
992000
3000
似乎明顯地被抑制。
16:51
So we are pursuing this line of work
374
996000
3000
所以我們持續進行這工作
16:54
and hopefully we'll have
375
999000
2000
並且希望我們將很快有
16:56
more results to report soon.
376
1001000
2000
更多結果可以報導。
16:58
Looking ahead, if you think of this disk
377
1003000
3000
未來，如果你把這圓盤看作
17:01
as representing all of the children
378
1006000
2000
我們目前為止
17:03
we've treated so far,
379
1008000
2000
治療過的孩子們的數量，
17:05
this is the magnitude of the problem.
380
1010000
2000
這就是問題的嚴重程度。
17:07
The red dots are the children we have not treated.
381
1012000
3000
這些紅點表示孩子我們尚未能治療的孩子。
17:10
So, there are many, many more children who need to be treated,
382
1015000
2000
所以，世界上還有更多的孩子需要治療，
17:12
and in order to expand the scope of the project,
383
1017000
3000
為了拓展這個計畫所及的區域，
17:15
we are planning on launching
384
1020000
2000
我們正計畫創建
17:17
The Prakash Center for Children,
385
1022000
2000
Prakash孩童中心，
17:19
which will have a dedicated pediatric hospital,
386
1024000
3000
它將有專門的小兒科醫院，
17:22
a school for the children we are treating
387
1027000
2000
一所為治療中的病童設立的學校，
17:24
and also a cutting-edge research facility.
388
1029000
2000
以及尖端研究設施。
17:26
The Prakash Center will integrate health care,
389
1031000
3000
Prakash中心將真正地
17:29
education and research in a way
390
1034000
2000
創造一個真正整合健康照護、
17:31
that truly creates the whole
391
1036000
2000
教育以及研究為一體
17:33
to be greater than the sum of the parts.
392
1038000
3000
比單純結合各部分更出色的機構。
17:36
So, to summarize: Prakash, in its five years of existence,
393
1041000
3000
在此，我做個小結，Prakash，在過去五年裡，
17:39
it's had an impact in multiple areas,
394
1044000
3000
已經對許多領域造成衝擊，
17:42
ranging from basic neuroscience
395
1047000
2000
範圍從基礎的神經科學裡
17:44
plasticity and learning in the brain,
396
1049000
2000
大腦的學習力及可塑性，
17:46
to clinically relevant hypotheses like in autism,
397
1051000
4000
到臨床相關的理論，如自閉症，
17:50
the development of autonomous machine vision systems,
398
1055000
3000
還有自主性機器視覺系統的發展，
17:53
education of the undergraduate and graduate students,
399
1058000
3000
大學生及研究生們的教育，
17:56
and most importantly in the alleviation
400
1061000
2000
最重要的，是減少了
17:58
of childhood blindness.
401
1063000
2000
盲童的數目。
18:00
And for my students and I, it's been
402
1065000
2000
而對我和我的學生來說，這計畫已是
18:02
just a phenomenal experience
403
1067000
2000
非比尋常的經驗，
18:04
because we have gotten to do interesting research,
404
1069000
4000
因為我們做著有趣的研究，
18:08
while at the same time
405
1073000
2000
同時也幫助
18:10
helping the many children that we have worked with.
406
1075000
2000
許多參與實驗的孩子們。
18:12
Thank you very much.
407
1077000
2000
非常感謝大家
18:14
(Applause)
408
1079000
2000
(鼓掌)
ABOUT THE SPEAKER
Pawan Sinha - Visual neuroscientistPawan Sinha researches how our brains interpret what our eyes see -- and uses that research to give blind children the gift of sight.
Why you should listen
At Pawan Sinha's MIT lab, he and his team spend their days trying to understand how the brain learns to recognize and use the patterns and scenes we see around us. To do this, they often use computers to model the processes of the human brain, but they also study human subjects, some of whom are seeing the world for the very first time and can tell them about the experience as it happens. They find these unusual subjects through the humanitarian branch of their research, Project Prakash.
Project Prakash sets up eye-care camps in some of the most habitually underserved regions of India, and gives free eye-health screenings to, since 2003, more than 700 functionally blind children. The children are then treated without charge, even if they do not fit the profile that would make them eligible for Sinha's research.
Sinha's eventual goal is to help 500 children each year; plans are under way for a center for visual rehabilitation in new Delhi. The special relationship that Sinha has created between research and humanitarianism promises to deliver on both fronts.
Pawan Sinha | Speaker | TED.com