TED2015
Laura Schulz: The surprisingly logical minds of babies
羅拉·舒爾茨: 嬰兒令人驚訝的邏輯思維
Filmed:
Readability: 3.8
1,888,975 views
嬰兒如何從零開始，快速地學到這麼多的事情? 在這個有趣和引用了很多實驗的講座中，認知科學家羅拉·舒爾茨分享嬰兒如何在懂得說話前，已經能用令人驚訝的邏輯思維作出決定。
Laura Schulz - Cognitive scientist
Developmental behavior studies spearheaded by Laura Schulz are changing our notions of how children learn. Full bio
Developmental behavior studies spearheaded by Laura Schulz are changing our notions of how children learn. Full bio
Double-click the English transcript below to play the video.
00:12
Mark Twain summed up
what I take to be
what I take to be
0
835
2155
馬克．吐溫用一句妙語概括了，
我認為是認知科學的一個最根本問題。
我認為是認知科學的一個最根本問題。
00:14
one of the fundamental problems
of cognitive science
of cognitive science
1
2990
3120
00:18
with a single witticism.
2
6110
1710
00:20
He said, "There's something
fascinating about science.
fascinating about science.
3
8410
3082
他說:「科學的有趣之處在於，
00:23
One gets such wholesale
returns of conjecture
returns of conjecture
4
11492
3228
一個人可從微不足道的事得出了偉大的猜想。」
00:26
out of such a trifling
investment in fact."
investment in fact."
5
14720
3204
00:29
(Laughter)
6
17924
1585
(笑聲)
00:32
Twain meant it as a joke,
of course, but he's right:
of course, but he's right:
7
20199
2604
馬克．吐溫當然只是開玩笑，但他是對的。
00:34
There's something
fascinating about science.
fascinating about science.
8
22803
2876
科學有其有趣之處。
00:37
From a few bones, we infer
the existence of dinosuars.
the existence of dinosuars.
9
25679
4261
從幾塊骨頭，我們推測了恐龍的存在;
00:42
From spectral lines,
the composition of nebulae.
the composition of nebulae.
10
30910
3871
從譜線得出了星雲的成份;
00:47
From fruit flies,
11
35471
2938
從果蠅得出了遺傳的機制;
00:50
the mechanisms of heredity,
12
38409
2943
00:53
and from reconstructed images
of blood flowing through the brain,
of blood flowing through the brain,
13
41352
4249
以及從血液流入大腦的重建影像，
00:57
or in my case, from the behavior
of very young children,
of very young children,
14
45601
4708
在我的研究則是從幼兒的行為中，
01:02
we try to say something about
the fundamental mechanisms
the fundamental mechanisms
15
50309
2829
我們嘗試解釋人類認知的基本機制。
01:05
of human cognition.
16
53138
1618
01:07
In particular, in my lab in the Department
of Brain and Cognitive Sciences at MIT,
of Brain and Cognitive Sciences at MIT,
17
55716
4759
我在麻省理工大腦及認知科學系實驗室中，
01:12
I have spent the past decade
trying to understand the mystery
trying to understand the mystery
18
60475
3654
花了過去十年研究一個謎團，
01:16
of how children learn so much
from so little so quickly.
from so little so quickly.
19
64129
3977
就是兒童如何從零開始，
快速地學到那麼多的東西。
快速地學到那麼多的東西。
01:20
Because, it turns out that
the fascinating thing about science
the fascinating thing about science
20
68666
2978
科學令人著迷之處，
01:23
is also a fascinating
thing about children,
thing about children,
21
71644
3529
亦正是孩子令人著迷的地方。
01:27
which, to put a gentler
spin on Mark Twain,
spin on Mark Twain,
22
75173
2581
回應馬克．吐溫的話，
01:29
is precisely their ability
to draw rich, abstract inferences
to draw rich, abstract inferences
23
77754
4650
那就是孩子從零碎和離亂的訊息中，
能夠得出豐富而抽象的推論的能力。
能夠得出豐富而抽象的推論的能力。
01:34
rapidly and accurately
from sparse, noisy data.
from sparse, noisy data.
24
82404
4661
01:40
I'm going to give you
just two examples today.
just two examples today.
25
88355
2398
我將舉出兩個例子:
01:42
One is about a problem of generalization,
26
90753
2287
一個是關於廣義化的問題，
01:45
and the other is about a problem
of causal reasoning.
of causal reasoning.
27
93040
2850
另一個則是關於因果推理的。
01:47
And although I'm going to talk
about work in my lab,
about work in my lab,
28
95890
2525
雖然我將會談及我實驗室的研究，
01:50
this work is inspired by
and indebted to a field.
and indebted to a field.
29
98415
3460
但這個研究的靈感是來自一個領域，
01:53
I'm grateful to mentors, colleagues,
and collaborators around the world.
and collaborators around the world.
30
101875
4283
一個我要感謝世界各地的導師、
同事和工作夥伴付出的領域。
同事和工作夥伴付出的領域。
01:59
Let me start with the problem
of generalization.
of generalization.
31
107308
2974
讓我先談談廣義化的問題。
02:02
Generalizing from small samples of data
is the bread and butter of science.
is the bread and butter of science.
32
110652
4133
歸納數據樣本在科學上是不可或缺的，
02:06
We poll a tiny fraction of the electorate
33
114785
2554
如我們調查一部分的選民，
02:09
and we predict the outcome
of national elections.
of national elections.
34
117339
2321
然後預測國家大選的結果。
02:12
We see how a handful of patients
responds to treatment in a clinical trial,
responds to treatment in a clinical trial,
35
120240
3925
我們觀察一小撮病人在臨床試驗中的反應，
02:16
and we bring drugs to a national market.
36
124165
3065
然後把藥物帶入市場，
02:19
But this only works if our sample
is randomly drawn from the population.
is randomly drawn from the population.
37
127230
4365
但只有在整個人口中隨機抽樣才可行。
02:23
If our sample is cherry-picked
in some way --
in some way --
38
131595
2735
當我們刻意挑選樣本，
02:26
say, we poll only urban voters,
39
134330
2072
如我們只調查城市中的選民，
02:28
or say, in our clinical trials
for treatments for heart disease,
for treatments for heart disease,
40
136402
4388
又或在治療心臟病的臨床試驗中，
02:32
we include only men --
41
140790
1881
我們只研究男性，
02:34
the results may not generalize
to the broader population.
to the broader population.
42
142671
3158
這樣的結果便不能代表整個人口。
02:38
So scientists care whether evidence
is randomly sampled or not,
is randomly sampled or not,
43
146479
3581
因此科學家著緊抽樣的方法是否隨機。
02:42
but what does that have to do with babies?
44
150060
2015
但這又跟嬰兒有甚麼關係?
02:44
Well, babies have to generalize
from small samples of data all the time.
from small samples of data all the time.
45
152585
4621
嬰兒在任何時候都要歸納數據樣本，
02:49
They see a few rubber ducks
and learn that they float,
and learn that they float,
46
157206
3158
當他們看到幾隻橡皮鴨，
並知道它們浮在水面。
並知道它們浮在水面。
02:52
or a few balls and learn that they bounce.
47
160364
3575
又或見到幾個皮球，
並知道它們能彈跳。
並知道它們能彈跳。
02:55
And they develop expectations
about ducks and balls
about ducks and balls
48
163939
2951
從中他們建立對橡膠鴨和皮球的概念，
02:58
that they're going to extend
to rubber ducks and balls
to rubber ducks and balls
49
166890
2716
並將這概念延伸至日後會見到的
所有橡膠鴨和皮球。
所有橡膠鴨和皮球。
03:01
for the rest of their lives.
50
169606
1879
03:03
And the kinds of generalizations
babies have to make about ducks and balls
babies have to make about ducks and balls
51
171485
3739
嬰兒對橡膠鴨和皮球的這種概括，
03:07
they have to make about almost everything:
52
175224
2089
他們會運用在每一件事上:
03:09
shoes and ships and sealing wax
and cabbages and kings.
and cabbages and kings.
53
177313
3917
鞋子、船、封蠟、捲心菜和皇帝。
03:14
So do babies care whether
the tiny bit of evidence they see
the tiny bit of evidence they see
54
182200
2961
因此嬰兒留意這些細節能否代表整體。
03:17
is plausibly representative
of a larger population?
of a larger population?
55
185161
3692
03:21
Let's find out.
56
189763
1900
我們一起看看吧。
03:23
I'm going to show you two movies,
57
191663
1723
我將讓你看兩段短片，
03:25
one from each of two conditions
of an experiment,
of an experiment,
58
193386
2462
這兩段短片分別代表實驗的兩個情況。
03:27
and because you're going to see
just two movies,
just two movies,
59
195848
2438
由於你將看到兩段短片，
03:30
you're going to see just two babies,
60
198286
2136
你只會看到兩個嬰兒，
03:32
and any two babies differ from each other
in innumerable ways.
in innumerable ways.
61
200422
3947
而這兩個嬰兒在很多地方都是不同的。
03:36
But these babies, of course,
here stand in for groups of babies,
here stand in for groups of babies,
62
204369
3051
但這兩個嬰兒將代表更大的群組，
03:39
and the differences you're going to see
63
207420
1895
你將看到的不同之處則代表
嬰兒行為中的平均差異。
嬰兒行為中的平均差異。
03:41
represent average group differences
in babies' behavior across conditions.
in babies' behavior across conditions.
64
209315
5195
03:47
In each movie, you're going to see
a baby doing maybe
a baby doing maybe
65
215160
2583
在每一段短片中你會見到嬰兒
在做些他們正常會做的事。
在做些他們正常會做的事。
03:49
just exactly what you might
expect a baby to do,
expect a baby to do,
66
217743
3460
03:53
and we can hardly make babies
more magical than they already are.
more magical than they already are.
67
221203
4017
嬰兒本身已是十分神奇的，
03:58
But to my mind the magical thing,
68
226090
2010
但對我來說他們的神奇之處，
04:00
and what I want you to pay attention to,
69
228100
2089
也是我想你們留意的地方，
04:02
is the contrast between
these two conditions,
these two conditions,
70
230189
3111
就是這兩種情況之間的分別。
04:05
because the only thing
that differs between these two movies
that differs between these two movies
71
233300
3529
因為這兩段短片唯一不同的地方，
04:08
is the statistical evidence
the babies are going to observe.
the babies are going to observe.
72
236829
3466
正是嬰兒將要觀察的資料。
04:13
We're going to show babies
a box of blue and yellow balls,
a box of blue and yellow balls,
73
241425
3183
我們把一些藍色和黃色的球給嬰兒看。
04:16
and my then-graduate student,
now colleague at Stanford, Hyowon Gweon,
now colleague at Stanford, Hyowon Gweon,
74
244608
4620
權孝媛當時是我的學生，
現在則是史丹佛大學的同事。
現在則是史丹佛大學的同事。
04:21
is going to pull three blue balls
in a row out of this box,
in a row out of this box,
75
249228
3077
她將拿出三個藍色的球，
04:24
and when she pulls those balls out,
she's going to squeeze them,
she's going to squeeze them,
76
252305
3123
而當她拿出這些球時，
她會把球擠一下，
她會把球擠一下，
04:27
and the balls are going to squeak.
77
255428
2113
讓這些球發出吱吱聲。
04:29
And if you're a baby,
that's like a TED Talk.
that's like a TED Talk.
78
257541
2763
這對於嬰兒來說就像TED一樣，
04:32
It doesn't get better than that.
79
260304
1904
是件很美好的事。
04:34
(Laughter)
80
262208
2561
（笑聲）
04:38
But the important point is it's really
easy to pull three blue balls in a row
easy to pull three blue balls in a row
81
266968
3659
從一個裝滿藍色球的箱中，
抽出三個藍色球是件很容易的事。
抽出三個藍色球是件很容易的事。
04:42
out of a box of mostly blue balls.
82
270627
2305
04:44
You could do that with your eyes closed.
83
272932
2060
你閉上眼睛也能做到，
04:46
It's plausibly a random sample
from this population.
from this population.
84
274992
2996
這就像隨機抽樣。
04:49
And if you can reach into a box at random
and pull out things that squeak,
and pull out things that squeak,
85
277988
3732
因此當你可以在箱中隨機地抽出
能吱吱叫的物件時，
能吱吱叫的物件時，
04:53
then maybe everything in the box squeaks.
86
281720
2839
也許箱中所有物件都能吱吱叫，
04:56
So maybe babies should expect
those yellow balls to squeak as well.
those yellow balls to squeak as well.
87
284559
3650
所以嬰兒可能會假設黃色球也能吱吱叫。
05:00
Now, those yellow balls
have funny sticks on the end,
have funny sticks on the end,
88
288209
2519
但這些黃色球都有一根棒，
05:02
so babies could do other things
with them if they wanted to.
with them if they wanted to.
89
290728
2857
所以嬰兒可用它們做些不同的事，
05:05
They could pound them or whack them.
90
293585
1831
他們可以拍打或搖動這些球。
05:07
But let's see what the baby does.
91
295416
2586
就讓我們看看這嬰兒會做甚麼。
05:12
(Video) Hyowon Gweon: See this?
(Ball squeaks)
(Ball squeaks)
92
300548
3343
(影片) 權孝媛: 看看這個。
(球發出吱吱聲)
(球發出吱吱聲)
05:16
Did you see that?
(Ball squeaks)
(Ball squeaks)
93
304531
3045
看到這個嗎?
(球發出吱吱聲)
(球發出吱吱聲)
05:20
Cool.
94
308036
3066
很酷吧!
05:24
See this one?
95
312706
1950
看看這個。
05:26
(Ball squeaks)
96
314656
1881
(球發出吱吱聲)
05:28
Wow.
97
316537
2653
哇!
05:33
Laura Schulz: Told you. (Laughs)
98
321854
2113
羅拉·舒爾茨: 早就說了。 (笑聲)
05:35
(Video) HG: See this one?
(Ball squeaks)
(Ball squeaks)
99
323967
4031
(影片) 孝媛: 看到這個嗎?
(球發出吱吱聲)
(球發出吱吱聲)
05:39
Hey Clara, this one's for you.
You can go ahead and play.
You can go ahead and play.
100
327998
4619
克拉拉, 這個是給你的, 你拿去玩吧。
05:51
(Laughter)
101
339854
4365
(笑聲)
05:56
LS: I don't even have to talk, right?
102
344219
2995
羅拉: 我不用解釋, 對吧?
05:59
All right, it's nice that babies
will generalize properties
will generalize properties
103
347214
2899
嬰兒把藍色球的特性套用到黃色球上。
06:02
of blue balls to yellow balls,
104
350113
1528
嬰兒從模仿我們中學習，這是很神奇的，
06:03
and it's impressive that babies
can learn from imitating us,
can learn from imitating us,
105
351641
3096
06:06
but we've known those things about babies
for a very long time.
for a very long time.
106
354737
3669
但我們早就知道嬰兒能這樣做。
06:10
The really interesting question
107
358406
1811
有趣的地方是當把一樣的東西給嬰兒看時，
甚麼事會發生。
甚麼事會發生。
06:12
is what happens when we show babies
exactly the same thing,
exactly the same thing,
108
360217
2852
我們能肯定這是完全一樣的，
因我們有個秘密的空間，
因我們有個秘密的空間，
06:15
and we can ensure it's exactly the same
because we have a secret compartment
because we have a secret compartment
109
363069
3611
06:18
and we actually pull the balls from there,
110
366680
2110
從中我們抽出這些球。
06:20
but this time, all we change
is the apparent population
is the apparent population
111
368790
3478
但這次我們改變了抽樣的母體。
06:24
from which that evidence was drawn.
112
372268
2902
06:27
This time, we're going to show babies
three blue balls
three blue balls
113
375170
3553
這次我們在一個裝滿黃色球的箱中，
抽出三個藍色球給嬰兒看。
抽出三個藍色球給嬰兒看。
06:30
pulled out of a box
of mostly yellow balls,
of mostly yellow balls,
114
378723
3384
想想甚麼事會發生?
06:34
and guess what?
115
382107
1322
06:35
You [probably won't] randomly draw
three blue balls in a row
three blue balls in a row
116
383429
2840
你大概不能隨機地在裝滿黃色球的箱中，
連續抽出三個藍色球，
連續抽出三個藍色球，
06:38
out of a box of mostly yellow balls.
117
386269
2484
06:40
That is not plausibly
randomly sampled evidence.
randomly sampled evidence.
118
388753
3747
因此這很可能不是隨機抽樣。
06:44
That evidence suggests that maybe Hyowon
was deliberately sampling the blue balls.
was deliberately sampling the blue balls.
119
392500
5123
這反映了孝媛可能是刻意抽出藍色球，
06:49
Maybe there's something special
about the blue balls.
about the blue balls.
120
397623
2583
可能這些藍色球是特別的，
06:52
Maybe only the blue balls squeak.
121
400846
2976
可能只有藍色球能吱吱叫。
06:55
Let's see what the baby does.
122
403822
1895
一起看看這嬰兒會做甚麼。
06:57
(Video) HG: See this?
(Ball squeaks)
(Ball squeaks)
123
405717
2904
(影片) 孝媛: 看看這個。
(球發出吱吱聲)
(球發出吱吱聲)
07:02
See this toy?
(Ball squeaks)
(Ball squeaks)
124
410851
2645
看到這個玩具嗎? (球發出吱吱聲)
07:05
Oh, that was cool. See?
(Ball squeaks)
(Ball squeaks)
125
413496
5480
哇, 這很酷, 看到嗎? (球發出吱吱聲)
07:10
Now this one's for you to play.
You can go ahead and play.
You can go ahead and play.
126
418976
4394
這個是給你的, 你拿去玩吧。
07:18
(Fussing)
(Laughter)
(Laughter)
127
426074
6347
(不耐煩的)
(笑聲)
(笑聲)
07:26
LS: So you just saw
two 15-month-old babies
two 15-month-old babies
128
434901
2748
羅拉: 你剛剛看到兩個15月大的嬰兒，
07:29
do entirely different things
129
437649
1942
按他們觀察到樣本出現的機率，
而做出完全不同的事。
而做出完全不同的事。
07:31
based only on the probability
of the sample they observed.
of the sample they observed.
130
439591
3599
07:35
Let me show you the experimental results.
131
443190
2321
一起看看實驗的結果，
07:37
On the vertical axis, you'll see
the percentage of babies
the percentage of babies
132
445511
2764
垂直軸代表在每一個情況中，
有多少百分比的嬰兒擠壓球。
有多少百分比的嬰兒擠壓球。
07:40
who squeezed the ball in each condition,
133
448275
2530
你可看見嬰兒在樣本和整體一致時，
07:42
and as you'll see, babies are much
more likely to generalize the evidence
more likely to generalize the evidence
134
450805
3715
07:46
when it's plausibly representative
of the population
of the population
135
454520
3135
比刻意挑選的樣本，
07:49
than when the evidence
is clearly cherry-picked.
is clearly cherry-picked.
136
457655
3738
較會歸納他們看到的特徵。
07:53
And this leads to a fun prediction:
137
461393
2415
因此這帶出一個有趣的預測。
07:55
Suppose you pulled just one blue ball
out of the mostly yellow box.
out of the mostly yellow box.
138
463808
4868
假設你在一個裝滿黃色球的箱中
只拿出一個藍色球，
只拿出一個藍色球，
08:00
You [probably won't] pull three blue balls
in a row at random out of a yellow box,
in a row at random out of a yellow box,
139
468896
3869
當然你很難隨機地連續抽出三個藍色球，
08:04
but you could randomly sample
just one blue ball.
just one blue ball.
140
472765
2455
但你可以只用一個藍色球作樣本，
08:07
That's not an improbable sample.
141
475220
1970
這不一定是個不可行的樣本。
08:09
And if you could reach into
a box at random
a box at random
142
477190
2224
當你隨機抽出一個會吱吱叫的東西時，
08:11
and pull out something that squeaks,
maybe everything in the box squeaks.
maybe everything in the box squeaks.
143
479414
3987
可能箱中所有的東西都會吱吱叫，
08:15
So even though babies are going to see
much less evidence for squeaking,
much less evidence for squeaking,
144
483875
4445
因此雖然嬰兒會看到較少吱吱叫的例子，
08:20
and have many fewer actions to imitate
145
488320
2242
而且在只抽出一個球的情況下，
他們會有較少的動作去模仿，
他們會有較少的動作去模仿，
08:22
in this one ball condition than in
the condition you just saw,
the condition you just saw,
146
490562
3343
08:25
we predicted that babies themselves
would squeeze more,
would squeeze more,
147
493905
3892
但我們預計會有更多嬰兒擠壓球。
08:29
and that's exactly what we found.
148
497797
2894
這正是我們發現的結果。
08:32
So 15-month-old babies,
in this respect, like scientists,
in this respect, like scientists,
149
500691
4411
因此15月大的嬰兒在這方面就像科學家，
08:37
care whether evidence
is randomly sampled or not,
is randomly sampled or not,
150
505102
3088
他們留意抽樣的方法是否隨機，
08:40
and they use this to develop
expectations about the world:
expectations about the world:
151
508190
3507
並以此建立對事物的概念:
08:43
what squeaks and what doesn't,
152
511697
2182
甚麼會吱吱叫而甚麼不會，
08:45
what to explore and what to ignore.
153
513879
3145
甚麼需要探索而甚麼可忽略。
08:50
Let me show you another example now,
154
518384
2066
現在讓我給你們看看另一個例子，
08:52
this time about a problem
of causal reasoning.
of causal reasoning.
155
520450
2730
這次是關於因果推理的。
08:55
And it starts with a problem
of confounded evidence
of confounded evidence
156
523180
2439
每人都要面對這個問題，
08:57
that all of us have,
157
525619
1672
因為我們都是這世界的一部份。
08:59
which is that we are part of the world.
158
527291
2020
09:01
And this might not seem like a problem
to you, but like most problems,
to you, but like most problems,
159
529311
3436
這看似不是一個問題，
但和其他問題一樣，
但和其他問題一樣，
09:04
it's only a problem when things go wrong.
160
532747
2337
事情會出狀況。
09:07
Take this baby, for instance.
161
535464
1811
以這個嬰兒為例，
09:09
Things are going wrong for him.
162
537275
1705
所有事都出了問題，
09:10
He would like to make
this toy go, and he can't.
this toy go, and he can't.
163
538980
2271
他想開動這個玩具，但他做不到。
09:13
I'll show you a few-second clip.
164
541251
2529
我會讓你看一段幾秒的影片。
09:21
And there's two possibilities, broadly:
165
549340
1920
這有兩個可能的原因，
09:23
Maybe he's doing something wrong,
166
551260
2634
可能是他做錯了一些事，
09:25
or maybe there's something
wrong with the toy.
wrong with the toy.
167
553894
4216
又或是那個玩具有些問題。
09:30
So in this next experiment,
168
558110
2111
因此在這個實驗中，
09:32
we're going to give babies
just a tiny bit of statistical data
just a tiny bit of statistical data
169
560221
3297
我們會給嬰兒們少許資料。
09:35
supporting one hypothesis over the other,
170
563518
2582
這些資料會傾向支持其中一個可能性，
09:38
and we're going to see if babies
can use that to make different decisions
can use that to make different decisions
171
566100
3455
我們將研究這些嬰兒能否運用這些資料，
09:41
about what to do.
172
569555
1834
而作出不同的決定。
09:43
Here's the setup.
173
571389
2022
這個實驗是這樣的:
09:46
Hyowon is going to try to make
the toy go and succeed.
the toy go and succeed.
174
574071
3030
孝媛嘗試開動那個玩具並成功了，
09:49
I am then going to try twice
and fail both times,
and fail both times,
175
577101
3320
而我的兩次嘗試都失敗了，
09:52
and then Hyowon is going
to try again and succeed,
to try again and succeed,
176
580421
3112
之後孝媛再嘗試，並再次成功了。
09:55
and this roughly sums up my relationship
to my graduate students
to my graduate students
177
583533
3172
這就像我和我的學生在使用新科技的情況。
09:58
in technology across the board.
178
586705
2835
10:02
But the important point here is
it provides a little bit of evidence
it provides a little bit of evidence
179
590030
3292
重要的是這提供了少許的資料，
10:05
that the problem isn't with the toy,
it's with the person.
it's with the person.
180
593322
3668
這反映玩具並沒有問題，而是人的問題。
10:08
Some people can make this toy go,
181
596990
2350
有些人可以開動這玩具，
10:11
and some can't.
182
599340
959
有些人則不能。
10:12
Now, when the baby gets the toy,
he's going to have a choice.
he's going to have a choice.
183
600799
3413
當這嬰兒拿到玩具時，他要作一個選擇。
10:16
His mom is right there,
184
604212
2188
他的母親在旁，
10:18
so he can go ahead and hand off the toy
and change the person,
and change the person,
185
606400
3315
所以他可以把玩具交給母親，
換另一人嘗試。
換另一人嘗試。
10:21
but there's also going to be
another toy at the end of that cloth,
another toy at the end of that cloth,
186
609715
3158
同時在毛巾上有另一個玩具，
10:24
and he can pull the cloth towards him
and change the toy.
and change the toy.
187
612873
3552
所以他也可以把玩具拉向自己，
換另一個玩具。
換另一個玩具。
10:28
So let's see what the baby does.
188
616425
2090
一起看看嬰兒會怎樣做。
10:30
(Video) HG: Two, three. Go!
(Music)
(Music)
189
618515
4183
(影片) 孝媛: 二、三、開始!
(音樂)
(音樂)
10:34
LS: One, two, three, go!
190
622698
3131
羅拉: 一、二、三、開始!
10:37
Arthur, I'm going to try again.
One, two, three, go!
One, two, three, go!
191
625829
7382
亞瑟，讓我再試一次，
一、二、三、開始!
一、二、三、開始!
10:45
YG: Arthur, let me try again, okay?
192
633677
2600
孝媛: 亞瑟，讓我再試吧，好嗎?
10:48
One, two, three, go!
(Music)
(Music)
193
636277
4550
一、二、三、開始!
(音樂)
(音樂)
10:53
Look at that. Remember these toys?
194
641583
1883
看看這裡，記得這些玩具嗎?
10:55
See these toys? Yeah, I'm going
to put this one over here,
to put this one over here,
195
643466
3264
看到嗎? 對，我會把這個放在這裡，
10:58
and I'm going to give this one to you.
196
646730
2062
把另一個給你。
11:00
You can go ahead and play.
197
648792
2335
你拿去玩吧。
11:23
LS: Okay, Laura, but of course,
babies love their mommies.
babies love their mommies.
198
671213
4737
羅拉: 你或許會說嬰兒都愛他們的母親，
11:27
Of course babies give toys
to their mommies
to their mommies
199
675950
2182
因此當玩具出現問題時，
嬰兒自然會把它交給母親。
嬰兒自然會把它交給母親。
11:30
when they can't make them work.
200
678132
2030
因此，問題在於當我們稍微改變資料時，
甚麼事會發生。
甚麼事會發生。
11:32
So again, the really important question
is what happens when we change
is what happens when we change
201
680162
3593
11:35
the statistical data ever so slightly.
202
683755
3154
11:38
This time, babies are going to see the toy
work and fail in exactly the same order,
work and fail in exactly the same order,
203
686909
4087
這次，嬰兒將看到這玩具
按同一次序成功運作和失敗，
按同一次序成功運作和失敗，
11:42
but we're changing
the distribution of evidence.
the distribution of evidence.
204
690996
2415
但我們改變了資料的分佈。
11:45
This time, Hyowon is going to succeed
once and fail once, and so am I.
once and fail once, and so am I.
205
693411
4411
這次孝媛和我各有一次成功和一次失敗，
11:49
And this suggests it doesn't matter
who tries this toy, the toy is broken.
who tries this toy, the toy is broken.
206
697822
5637
這代表誰人嘗試都沒有分別，
那件玩具是壞的，
那件玩具是壞的，
11:55
It doesn't work all the time.
207
703459
1886
它不是每次都能運作的。
11:57
Again, the baby's going to have a choice.
208
705345
1965
同樣地，嬰兒要作出一個選擇，
11:59
Her mom is right next to her,
so she can change the person,
so she can change the person,
209
707310
3396
她的母親在旁，所以她可換另一人嘗試，
12:02
and there's going to be another toy
at the end of the cloth.
at the end of the cloth.
210
710706
3204
同時另一個玩具就在毛巾上。
12:05
Let's watch what she does.
211
713910
1378
看看她會怎樣做。
12:07
(Video) HG: Two, three, go!
(Music)
(Music)
212
715288
4348
(影片) 孝媛: 二、三、開始!
(音樂)
(音樂)
12:11
Let me try one more time.
One, two, three, go!
One, two, three, go!
213
719636
4984
讓我再試一次，
一、二、三、開始!
一、二、三、開始!
12:17
Hmm.
214
725460
1697
嗯...
12:19
LS: Let me try, Clara.
215
727950
2692
羅拉: 讓我試試吧，克拉拉。
12:22
One, two, three, go!
216
730642
3945
一、二、三、開始!
12:27
Hmm, let me try again.
217
735265
1935
嗯...讓我再試試。
12:29
One, two, three, go!
(Music)
(Music)
218
737200
5670
一、二、三、開始!
(音樂)
(音樂)
12:35
HG: I'm going
to put this one over here,
to put this one over here,
219
743009
2233
孝媛: 我把這個放在這裡，
12:37
and I'm going to give this one to you.
220
745242
2001
這個則交給你，
12:39
You can go ahead and play.
221
747243
3597
你拿去玩吧。
12:58
(Applause)
222
766376
4897
(掌聲)
13:04
LS: Let me show you
the experimental results.
the experimental results.
223
772993
2392
羅拉: 看看這個實驗的結果，
13:07
On the vertical axis,
you'll see the distribution
you'll see the distribution
224
775385
2475
在垂直軸上，你會看到在每種情況下，
嬰兒作出不同選擇的分佈。
嬰兒作出不同選擇的分佈。
13:09
of children's choices in each condition,
225
777860
2577
你會發現他們作的選擇是
基於他們觀察到的資料。
基於他們觀察到的資料。
13:12
and you'll see that the distribution
of the choices children make
of the choices children make
226
780437
4551
13:16
depends on the evidence they observe.
227
784988
2787
13:19
So in the second year of life,
228
787775
1857
因此當他們兩歲時，
13:21
babies can use a tiny bit
of statistical data
of statistical data
229
789632
2577
嬰兒已經可以運用細微的資料，
13:24
to decide between two
fundamentally different strategies
fundamentally different strategies
230
792209
3367
在兩個完全不同的選項中作出決定:
13:27
for acting in the world:
231
795576
1881
13:29
asking for help and exploring.
232
797457
2743
尋求幫忙或自行探索。
13:33
I've just shown you
two laboratory experiments
two laboratory experiments
233
801700
3434
我剛才讓你們看了兩個實驗，
13:37
out of literally hundreds in the field
that make similar points,
that make similar points,
234
805134
3691
在這領域中有數千個得出相同結果的實驗。
13:40
because the really critical point
235
808825
2392
當中反映的重點是，
13:43
is that children's ability
to make rich inferences from sparse data
to make rich inferences from sparse data
236
811217
5108
兒童擁有充分解讀零碎資訊的能力，
13:48
underlies all the species-specific
cultural learning that we do.
cultural learning that we do.
237
816325
5341
這超出了所有文化的學習方式。
13:53
Children learn about new tools
from just a few examples.
from just a few examples.
238
821666
4597
孩子從少數的例子便能學到新技能，
13:58
They learn new causal relationships
from just a few examples.
from just a few examples.
239
826263
4717
他們從少數的例子便能領略到新的因果關係，
14:03
They even learn new words,
in this case in American Sign Language.
in this case in American Sign Language.
240
831928
4871
他們甚至能學到新的生字，如美國手語。
14:08
I want to close with just two points.
241
836799
2311
我會提出兩個重點作總結。
14:12
If you've been following my world,
the field of brain and cognitive sciences,
the field of brain and cognitive sciences,
242
840050
3688
如果你近年有留意大腦和認知科學領域，
14:15
for the past few years,
243
843738
1927
你會聽到三個重要的概念。
14:17
three big ideas will have come
to your attention.
to your attention.
244
845665
2415
14:20
The first is that this is
the era of the brain.
the era of the brain.
245
848080
3436
第一，現在是大腦的時代。
14:23
And indeed, there have been
staggering discoveries in neuroscience:
staggering discoveries in neuroscience:
246
851516
3669
的確，神經科學近來有不少驚人的發現，
14:27
localizing functionally specialized
regions of cortex,
regions of cortex,
247
855185
3436
例如標記了大腦皮層負責不同功能的位置、
14:30
turning mouse brains transparent,
248
858621
2601
製造出透明的老鼠大腦、
14:33
activating neurons with light.
249
861222
3776
以及利用光線啟動神經元。
14:36
A second big idea
250
864998
1996
第二個重要的概念是，
14:38
is that this is the era of big data
and machine learning,
and machine learning,
251
866994
4104
現在是大數據和機器學習的時代，
14:43
and machine learning promises
to revolutionize our understanding
to revolutionize our understanding
252
871098
3141
而機器學習能徹底改變我們對任何事的理解，
14:46
of everything from social networks
to epidemiology.
to epidemiology.
253
874239
4667
從社交網站到流行病學。
14:50
And maybe, as it tackles problems
of scene understanding
of scene understanding
254
878906
2693
當機器學習能理解埸合和處理自然語言時，
14:53
and natural language processing,
255
881599
1993
也許我們能藉此了解人類的認知。
14:55
to tell us something
about human cognition.
about human cognition.
256
883592
3324
14:59
And the final big idea you'll have heard
257
887756
1937
最後一個你會聽過的重要概念是，
15:01
is that maybe it's a good idea we're going
to know so much about brains
to know so much about brains
258
889693
3387
我們將對大腦有很深入的認識，
並能掌握大數據，而這很可能是件好事。
並能掌握大數據，而這很可能是件好事。
15:05
and have so much access to big data,
259
893080
1917
15:06
because left to our own devices,
260
894997
2507
因為相比機器而言，
15:09
humans are fallible, we take shortcuts,
261
897504
3831
人類易犯錯誤，我們會走捷徑，
15:13
we err, we make mistakes,
262
901335
3437
我們會做錯，
15:16
we're biased, and in innumerable ways,
263
904772
3684
我們在很多方面都有偏見，
15:20
we get the world wrong.
264
908456
2969
我們會有錯誤的理解。
15:24
I think these are all important stories,
265
912843
2949
我認為這都是重要的，
15:27
and they have a lot to tell us
about what it means to be human,
about what it means to be human,
266
915792
3785
因為這反映了人類的特質，
15:31
but I want you to note that today
I told you a very different story.
I told you a very different story.
267
919577
3529
但我今天想帶出事情的另一面。
15:35
It's a story about minds and not brains,
268
923966
3807
這是關於思維而非大腦的，
15:39
and in particular, it's a story
about the kinds of computations
about the kinds of computations
269
927773
3006
尤其是人類獨有的運算能力，
15:42
that uniquely human minds can perform,
270
930779
2590
這牽涉了豐富、有條理的知識，
15:45
which involve rich, structured knowledge
and the ability to learn
and the ability to learn
271
933369
3944
以及從少量的數據和例子中學習的能力。
15:49
from small amounts of data,
the evidence of just a few examples.
the evidence of just a few examples.
272
937313
5268
15:56
And fundamentally, it's a story
about how starting as very small children
about how starting as very small children
273
944301
4299
再者，這是關於我們如何從幼童，
16:00
and continuing out all the way
to the greatest accomplishments
to the greatest accomplishments
274
948600
4180
一路發展至成為文化中偉大的成就，
16:04
of our culture,
275
952780
3843
16:08
we get the world right.
276
956623
1997
我們能正確地理解這個世界。
16:12
Folks, human minds do not only learn
from small amounts of data.
from small amounts of data.
277
960433
5267
大家，
人腦不只是懂得從少量的數據中學習。
人腦不只是懂得從少量的數據中學習。
16:18
Human minds think
of altogether new ideas.
of altogether new ideas.
278
966285
2101
人腦能想到新的主意。
16:20
Human minds generate
research and discovery,
research and discovery,
279
968746
3041
人腦能創造出研究和發明。
16:23
and human minds generate
art and literature and poetry and theater,
art and literature and poetry and theater,
280
971787
5273
人腦能創作藝術、文學、寫詩和戲劇。
16:29
and human minds take care of other humans:
281
977070
3760
人腦可照顧其他人，
16:32
our old, our young, our sick.
282
980830
3427
包括年老的、年輕的、患病的，
16:36
We even heal them.
283
984517
2367
我們甚至能治癒他們。
16:39
In the years to come, we're going
to see technological innovations
to see technological innovations
284
987564
3103
在未來，我們將會看到
超乎現在能想像的科技發展，
超乎現在能想像的科技發展，
16:42
beyond anything I can even envision,
285
990667
3797
16:46
but we are very unlikely
286
994464
2150
但在我或你們的一生中，
我們不太可能目睹比得上嬰兒運算能力的機器。
我們不太可能目睹比得上嬰兒運算能力的機器。
16:48
to see anything even approximating
the computational power of a human child
the computational power of a human child
287
996614
5709
16:54
in my lifetime or in yours.
288
1002323
4298
16:58
If we invest in these most powerful
learners and their development,
learners and their development,
289
1006621
5047
假如我們投資在最厲害的學習者和其發展身上，
17:03
in babies and children
290
1011668
2917
在嬰兒和兒童身上、
17:06
and mothers and fathers
291
1014585
1826
在母親和父親身上、
17:08
and caregivers and teachers
292
1016411
2699
在照顧者和老師身上，
17:11
the ways we invest in our other
most powerful and elegant forms
most powerful and elegant forms
293
1019110
4170
如同我們投資在最厲害的科技、工程和設計上時，
17:15
of technology, engineering and design,
294
1023280
3218
17:18
we will not just be dreaming
of a better future,
of a better future,
295
1026498
2939
我們不只是夢想有個更好的將來，
17:21
we will be planning for one.
296
1029437
2127
而是在計劃一個更好的將來。
17:23
Thank you very much.
297
1031564
2345
謝謝。
17:25
(Applause)
298
1033909
3421
(掌聲)
17:29
Chris Anderson: Laura, thank you.
I do actually have a question for you.
I do actually have a question for you.
299
1037810
4426
克里斯·安德森: 羅拉, 謝謝你，
我其實想問你一個問題。
我其實想問你一個問題。
17:34
First of all, the research is insane.
300
1042236
2359
首先，這項研究真是太瘋狂了。
17:36
I mean, who would design
an experiment like that? (Laughter)
an experiment like that? (Laughter)
301
1044595
3725
我的意思是，有誰會想到這些實驗? (笑聲)
17:41
I've seen that a couple of times,
302
1049150
1790
我見過很多類似的實驗，
17:42
and I still don't honestly believe
that that can truly be happening,
that that can truly be happening,
303
1050940
3222
但我仍然覺得難以置信，
17:46
but other people have done
similar experiments; it checks out.
similar experiments; it checks out.
304
1054162
3158
儘管很多人做了類似的實驗，而事實的確如此。
17:49
The babies really are that genius.
305
1057320
1633
這些嬰兒根本是天才。
17:50
LS: You know, they look really impressive
in our experiments,
in our experiments,
306
1058953
3007
羅拉: 在實驗中這看似很神奇，
17:53
but think about what they
look like in real life, right?
look like in real life, right?
307
1061960
2652
但想想在現實生活中是怎樣的，對嗎?
17:56
It starts out as a baby.
308
1064612
1150
一出世時，他只是個嬰兒，
17:57
Eighteen months later,
it's talking to you,
it's talking to you,
309
1065762
2007
但18個月後他開始說話，
17:59
and babies' first words aren't just
things like balls and ducks,
things like balls and ducks,
310
1067769
3041
而嬰兒最初說的話不只是物件，
如皮球和鴨子，
如皮球和鴨子，
18:02
they're things like "all gone,"
which refer to disappearance,
which refer to disappearance,
311
1070810
2881
他們更能表達「不見了」的概念，
18:05
or "uh-oh," which refer
to unintentional actions.
to unintentional actions.
312
1073691
2283
又或是以「哎喲」表達無心之失。
18:07
It has to be that powerful.
313
1075974
1562
這必須是那麼厲害的，
18:09
It has to be much more powerful
than anything I showed you.
than anything I showed you.
314
1077536
2775
這必須比我剛才展示的還要厲害。
18:12
They're figuring out the entire world.
315
1080311
1974
嬰兒在弄清楚整個世界，
18:14
A four-year-old can talk to you
about almost anything.
about almost anything.
316
1082285
3144
一個四歲的小孩幾乎懂得說所有東西。
18:17
(Applause)
317
1085429
1601
(掌聲)
18:19
CA: And if I understand you right,
the other key point you're making is,
the other key point you're making is,
318
1087030
3414
克里斯: 如果我沒錯的話，
你想指出的另一個重點是，
你想指出的另一個重點是，
18:22
we've been through these years
where there's all this talk
where there's all this talk
319
1090444
2754
這些年來，我們都聽說
我們的腦袋是不可信和會出錯的，
我們的腦袋是不可信和會出錯的，
18:25
of how quirky and buggy our minds are,
320
1093198
1932
18:27
that behavioral economics
and the whole theories behind that
and the whole theories behind that
321
1095130
2867
行為經濟學和其他新理論都指出我們不是理性的。
18:29
that we're not rational agents.
322
1097997
1603
18:31
You're really saying that the bigger
story is how extraordinary,
story is how extraordinary,
323
1099600
4216
但你指出了我們的腦袋是非凡的，
18:35
and there really is genius there
that is underappreciated.
that is underappreciated.
324
1103816
4944
我們一直忽略了我們的腦袋是多麼神奇。
18:40
LS: One of my favorite
quotes in psychology
quotes in psychology
325
1108760
2070
羅拉: 我最喜愛的心理學名言之一，
18:42
comes from the social
psychologist Solomon Asch,
psychologist Solomon Asch,
326
1110830
2290
來自社會心理學家所羅門·阿希，
18:45
and he said the fundamental task
of psychology is to remove
of psychology is to remove
327
1113120
2807
他說心理學首要的任務是
去除那些毋需証明的事物的面紗。
去除那些毋需証明的事物的面紗。
18:47
the veil of self-evidence from things.
328
1115927
2626
18:50
There are orders of magnitude
more decisions you make every day
more decisions you make every day
329
1118553
4551
每天你作出大大小小的決定去理解這個世界。
18:55
that get the world right.
330
1123104
1347
18:56
You know about objects
and their properties.
and their properties.
331
1124451
2132
你知道不同物件及其特性，
18:58
You know them when they're occluded.
You know them in the dark.
You know them in the dark.
332
1126583
3029
即使被覆蓋和在黑暗中你也知道。
19:01
You can walk through rooms.
333
1129612
1308
你能在空間中行走。
19:02
You can figure out what other people
are thinking. You can talk to them.
are thinking. You can talk to them.
334
1130920
3532
你能猜到別人在想甚麼，你能和別人交談。
你能探索空間，你明白數字。
19:06
You can navigate space.
You know about numbers.
You know about numbers.
335
1134452
2230
你明白因果關係，
你懂得分辨是非。
你懂得分辨是非。
19:08
You know causal relationships.
You know about moral reasoning.
You know about moral reasoning.
336
1136682
3022
你毫不費力便能做到，
所以我們不會察覺，
所以我們不會察覺，
19:11
You do this effortlessly,
so we don't see it,
so we don't see it,
337
1139704
2356
但這就是我們理解這個世界的方法，
19:14
but that is how we get the world right,
and it's a remarkable
and it's a remarkable
338
1142060
2912
這是個神奇而又難以理解的成就。
19:16
and very difficult-to-understand
accomplishment.
accomplishment.
339
1144972
2318
克里斯: 我相信在坐有人認為科技正急速發展，
19:19
CA: I suspect there are people
in the audience who have
in the audience who have
340
1147290
2628
19:21
this view of accelerating
technological power
technological power
341
1149918
2238
他們可能不認同你說電腦
不能做到三歲小孩能做到的事。
不能做到三歲小孩能做到的事。
19:24
who might dispute your statement
that never in our lifetimes
that never in our lifetimes
342
1152156
2958
19:27
will a computer do what
a three-year-old child can do,
a three-year-old child can do,
343
1155114
2618
但可以肯定的是，無論在甚麼場合，
19:29
but what's clear is that in any scenario,
344
1157732
3248
19:32
our machines have so much to learn
from our toddlers.
from our toddlers.
345
1160980
3770
嬰兒有很多地方值得我們的機器學習。
19:38
LS: I think so. You'll have some
machine learning folks up here.
machine learning folks up here.
346
1166230
3216
羅拉: 我同意。有些人認同機器學習。
19:41
I mean, you should never bet
against babies or chimpanzees
against babies or chimpanzees
347
1169446
4203
我的意思是，你不應將嬰兒和黑猩猩跟科技比較，
19:45
or technology as a matter of practice,
348
1173649
3645
因為這不是數量上的不同，
19:49
but it's not just
a difference in quantity,
a difference in quantity,
349
1177294
4528
19:53
it's a difference in kind.
350
1181822
1764
而是性質上的不同。
19:55
We have incredibly powerful computers,
351
1183586
2160
我們有十分厲害的電腦，
19:57
and they do do amazingly
sophisticated things,
sophisticated things,
352
1185746
2391
它們能做到複雜的事情，
20:00
often with very big amounts of data.
353
1188137
3204
和處理大量的資料。
20:03
Human minds do, I think,
something quite different,
something quite different,
354
1191341
2607
我認為人類的腦袋做的事是不同的，
20:05
and I think it's the structured,
hierarchical nature of human knowledge
hierarchical nature of human knowledge
355
1193948
3895
人類的知識是有系統和條理分明的，
20:09
that remains a real challenge.
356
1197843
2032
這對機器仍然是一個挑戰。
20:11
CA: Laura Schulz, wonderful
food for thought. Thank you so much.
food for thought. Thank you so much.
357
1199875
3061
克里斯: 勞拉·舒爾茨，十分精彩。謝謝。
20:14
LS: Thank you.
(Applause)
(Applause)
358
1202936
2922
羅拉: 謝謝。
(掌聲)
(掌聲)
ABOUT THE SPEAKER
Laura Schulz - Cognitive scientistDevelopmental behavior studies spearheaded by Laura Schulz are changing our notions of how children learn.
Why you should listen
MIT Early Childhood Cognition Lab lead investigator Laura Schulz studies learning in early childhood. Her research bridges computational models of cognitive development and behavioral studies in order to understand the origins of inquiry and discovery.
Working in play labs, children’s museums, and a recently-launched citizen science website, Schultz is reshaping how we view young children’s perceptions of the world around them. Some of the surprising results of her research: before the age of four, children expect hidden causes when events happen probabilistically, use simple experiments to distinguish causal hypotheses, and trade off learning from instruction and exploration.
Laura Schulz | Speaker | TED.com