ABOUT THE SPEAKER
Colin Camerer - Behavioral economist
Colin Camerer is a leading behavioral economist who studies the psychological and neural bases of choice and strategic decision-making.

Why you should listen

Colin Camerer focuses on brain behavior during decision making, strategizing and market trading. He is the Robert Kirby Professor of Behavioral Finance and Economics at the California Institute of Technology. A child prodigy in his youth, Camerer received a B.A. in quantitative studies from Johns Hopkins when he was just 17 and a PhD in decision theory from the University of Chicago Graduate School of Business when he was 22. Camerer's research departs from previous theory in that it does not assume the mind to be a rational and perfect system, but rather focuses on the limitations of everyday people when they play actual games, and seeks to predict how they will behave in situations that involve strategy. His studies focus on neurological findings from economic experiments in the lab (on humans -- and monkeys!) Camerer is the author of Behavioral Game Theory.

More profile about the speaker
Colin Camerer | Speaker | TED.com
TEDxCaltech

Colin Camerer: When you're making a deal, what's going on in your brain?

柯林.坎麥爾(Colin Camerer):神經科學、賽局理論、猴子

Filmed:
1,516,785 views

不論在合作或競爭的條件下,協商者的大腦產生了什麼變化?行為經濟學家柯林.坎麥爾(Colin Camerer)展示了一些可以讓我們預測別人思考方式的研究報告。並呈現出一個超乎預期的結果,顯示黑猩猩比人類優秀。(影片攝於 TEDxCalTech)
- Behavioral economist
Colin Camerer is a leading behavioral economist who studies the psychological and neural bases of choice and strategic decision-making. Full bio

Double-click the English transcript below to play the video.

00:12
I'm going to talk about the strategizing運籌帷幄 brain.
0
562
2253
我要說說大腦決策
00:14
We're going to use an unusual異常 combination組合 of tools工具
1
2815
2238
我們要用一些不尋常組合的工具
00:17
from game遊戲 theory理論 and neuroscience神經科學
2
5053
2013
從賽局理論到神經科學
00:19
to understand理解 how people interact相互作用 socially社交上 when value is on the line.
3
7066
2967
以瞭解利益糾葛時大眾的互動
00:22
So game遊戲 theory理論 is a branch of, originally本來, applied應用的 mathematics數學,
4
10033
3971
賽局理論源自應用數學
00:26
used mostly大多 in economics經濟學 and political政治 science科學, a little bit in biology生物學,
5
14004
2986
主要用在經濟和政治學
很少被應用在生物學上
00:28
that gives us a mathematical數學的 taxonomy分類 of social社會 life
6
16990
3289
它提供了關於社會行為的數學模型
00:32
and it predicts預測 what people are likely容易 to do
7
20279
2131
當人的行為會互相影響時
00:34
and believe others其他 will do
8
22410
1194
它可以用來預測
當人類行為會互相影響時
00:35
in cases where everyone's大家的 actions行動 affect影響 everyone大家 else其他.
9
23604
3390
人可能會採取的行動
00:38
That's a lot of things: competition競爭, cooperation合作, bargaining議價,
10
26994
3478
這包含很多事:競爭、合作、議價
00:42
games遊戲 like hide-and-seek捉迷藏, and poker撲克.
11
30472
3291
一些像是捉迷藏和撲克牌的遊戲
00:45
Here's這裡的 a simple簡單 game遊戲 to get us started開始.
12
33763
2239
我們用一個簡單的遊戲開始吧
00:48
Everyone大家 chooses a number from zero to 100,
13
36002
2162
每個人從 0 到 100 間選一個數字
00:50
we're going to compute計算 the average平均 of those numbers數字,
14
38164
2446
我們會計算這些數值的平均
00:52
and whoever's誰是 closest最近的 to two-thirds三分之二 of the average平均 wins a fixed固定 prize.
15
40610
4194
選擇最接近平均值 2/3 數值的人贏
00:56
So you want to be a little bit below下面 the average平均 number,
16
44804
2269
所以你會需要比平均值低一點的數值
00:59
but not too far below下面, and everyone大家 else其他 wants to be
17
47073
2236
但又不要太低
而且每個人都會想要
01:01
a little bit below下面 the average平均 number as well.
18
49309
1945
選比平均值低一些的數值
01:03
Think about what you might威力 pick.
19
51254
2579
想想你會選什麼
01:05
As you're thinking思維, this is a toy玩具 model模型 of something like
20
53833
3271
你可能會想:
這就像是股價上揚時
01:09
selling銷售 in the stock股票 market市場 during a rising升起 market市場. Right?
21
57104
2812
賣股票時機選擇的簡單模型
01:11
You don't want to sell too early, because you miss小姐 out on profits利潤,
22
59916
2225
你不想太早賣
因為你會減少獲利
01:14
but you don't want to wait too late晚了
23
62141
2180
但又不想太晚賣
01:16
to when everyone大家 else其他 sells塞爾斯, triggering觸發 a crash緊急.
24
64321
2401
因為大家都賣完後股價會崩盤
01:18
You want to be a little bit ahead of the competition競爭, but not too far ahead.
25
66722
2879
所以你會想超前大家一步,但別過頭
01:21
Okay, here's這裡的 two theories理論 about how people might威力 think about this,
26
69601
3613
好,對於大家的想法有兩種可能
01:25
and then we'll see some data數據.
27
73214
1396
然後我們看看數據
01:26
Some of these will sound聲音 familiar because you probably大概 are
28
74610
2191
有些聽起來可能會很熟悉
01:28
thinking思維 that way. I'm using運用 my brain theory理論 to see.
29
76801
3799
因為你就是這麼想
因為我是用大腦理論分析出來
01:32
A lot of people say, "I really don't know what people are going to pick,
30
80600
3110
許多人會說:
「我不知道大家會怎麼選,
01:35
so I think the average平均 will be 50."
31
83710
1673
但我知道平均值是 50。」
01:37
They're not being存在 really strategic戰略 at all.
32
85383
1911
這其中沒有策略思考成份
01:39
"And I'll pick two-thirds三分之二 of 50. That's 33." That's a start開始.
33
87294
3352
「我會選 50 的 2/3 ,也就是 33 。」這是個開始
01:42
Other people who are a little more sophisticated複雜的,
34
90646
1892
其他人比較世故一些
01:44
using運用 more working加工 memory記憶,
35
92538
1476
消耗大腦多一點運算能力
01:46
say, "I think people will pick 33 because they're going to pick a response響應 to 50,
36
94014
3904
說:「別人會因 50 而選 33 ,
01:49
and so I'll pick 22, which哪一個 is two-thirds三分之二 of 33."
37
97918
2968
所以我就選 22 , 也就是 33 的 2/3 。」
01:52
They're doing one extra額外 step of thinking思維, two steps腳步.
38
100886
2479
他們多想了一步,也就是兩步
01:55
That's better. And of course課程, in principle原理,
39
103365
2617
這更好,當然理論上
01:57
you could do three, four or more,
40
105982
1827
你可以多想三步、四步或更多
01:59
but it starts啟動 to get very difficult.
41
107809
1869
但是想越多越困難
02:01
Just like in language語言 and other domains, we know that it's hard for people to parse解析
42
109678
2592
就像在語言或其他領域
02:04
very complex複雜 sentences句子 with a kind of recursive遞歸 structure結構體.
43
112270
3634
要組織層疊的句子結構很困難
02:07
This is called a cognitive認知 hierarchy等級制度 theory理論, by the way.
44
115904
1734
順帶一題,這稱為認知層級理論
02:09
It's something that I've worked工作 on and a few少數 other people,
45
117638
2556
這是我和一些人研究的東西
02:12
and it indicates指示 a kind of hierarchy等級制度 along沿 with
46
120194
2220
這研究用來驗證
02:14
some assumptions假設 about how many許多 people stop at different不同 steps腳步
47
122414
2254
思考深度存在階層關係的假設
02:16
and how the steps腳步 of thinking思維 are affected受影響
48
124668
1884
和哪些有趣變因會影響思考深度
02:18
by lots of interesting有趣 variables變量 and variant變種 people, as we'll see in a minute分鐘.
49
126552
3696
這部份我們等下就會看到
02:22
A very different不同 theory理論, a much more popular流行 one, and an older舊的 one,
50
130248
3386
另一個不同的理論
也是較為早期、為人熟知的
02:25
due應有 largely大部分 to John約翰 Nash納什 of "A Beautiful美麗 Mind心神" fame名譽,
51
133634
3540
大部份歸因於約翰.納許的《美麗境界》
02:29
is what's called equilibrium平衡 analysis分析.
52
137174
2240
稱為均衡分析
02:31
So if you've ever taken採取 a game遊戲 theory理論 course課程 at any level水平,
53
139414
2454
如果你曾修過賽局理論的課
02:33
you will have learned學到了 a little bit about this.
54
141868
1713
你可能都學過一些
02:35
An equilibrium平衡 is a mathematical數學的 state in which哪一個 everybody每個人
55
143581
2855
「均衡」是對每個人都知道
02:38
has figured想通 out exactly究竟 what everyone大家 else其他 will do.
56
146436
2449
其他人會怎麼做的數學狀態
02:40
It is a very useful有用 concept概念, but behaviorally行為上,
57
148885
2007
這是一個有用的概念
02:42
it may可能 not exactly究竟 explain說明 what people do
58
150892
2003
但是首次應用在經濟領域
02:44
the first time they play these types類型 of economic經濟 games遊戲
59
152895
2735
或是其他現實世界的問題時
02:47
or in situations情況 in the outside world世界.
60
155630
2333
它卻無法確切解釋人的行為
02:49
In this case案件, the equilibrium平衡 makes品牌 a very bold膽大 prediction預測,
61
157963
2338
在這狀況下
「均衡」是一個大膽的假設:
02:52
which哪一個 is everyone大家 wants to be below下面 everyone大家 else其他,
62
160301
2860
每個人都會想比
其他人選的數值低
02:55
therefore因此 they'll他們會 play zero.
63
163161
2291
所以他們都會選 0
02:57
Let's see what happens發生. This experiment's實驗的 been doneDONE many許多, many許多 times.
64
165452
3009
讓我們來看結果
這是一個被重複多次的實驗
03:00
Some of the earliest最早 ones那些 were doneDONE in the '90s
65
168461
1883
我、羅斯瑪麗.納格爾及其它人
03:02
by me and Rosemarie羅斯瑪麗 Nagel內格爾 and others其他.
66
170344
2645
在 90 年代進行的早期實驗
03:04
This is a beautiful美麗 data數據 set of 9,000 people who wrote in
67
172989
2985
這是由 9 千人參加
公佈在 3 家報紙和雜誌的競賽後
03:07
to three newspapers報紙 and magazines雜誌 that had a contest比賽.
68
175974
2880
所蒐集到的資料
03:10
The contest比賽 said, send發送 in your numbers數字
69
178854
1814
競賽問題說:
「回傳你所選的數字,
03:12
and whoever is close to two-thirds三分之二 of the average平均 will win贏得 a big prize.
70
180668
3155
而任何最接近 2/3 平均值的人獲獎」
03:15
And as you can see, there's so much data數據 here, you can see the spikes釘鞋 very visibly明顯地.
71
183823
3088
如你所見,因為資料龐大
你可以清楚的看到一些高峰
03:18
There's a spike at 33. Those are people doing one step.
72
186911
3381
在 33 有一個高峰
這是只思考一步的人
03:22
There is another另一個 spike visible可見 at 22.
73
190292
2497
另外一個高峰是在 22
03:24
And notice注意, by the way, that most people pick numbers數字 right around there.
74
192789
2292
順帶一題
多數人選擇 33 和 22 附近的數值
03:27
They don't necessarily一定 pick exactly究竟 33 and 22.
75
195081
2510
他們不一定只選 33 或 22
03:29
There's something a little bit noisy嘈雜 around it.
76
197591
2056
所以在附近會有一些雜訊
03:31
But you can see those spikes釘鞋, and they're there.
77
199647
1478
但是你可以看到這些高峰確實存在
03:33
There's another另一個 group of people who seem似乎 to have
78
201125
1710
另外一些人
03:34
a firm公司 grip on equilibrium平衡 analysis分析,
79
202835
2075
對均衡理論有深入的瞭解
03:36
because they're picking選擇 zero or one.
80
204910
2395
因為他們選擇 0 或 1
03:39
But they lose失去, right?
81
207305
2089
但是他們輸了,對吧?
03:41
Because picking選擇 a number that low is actually其實 a bad choice選擇
82
209394
3192
因為如果其他人不懂均衡理論
03:44
if other people aren't doing equilibrium平衡 analysis分析 as well.
83
212586
2820
選擇低數值就不是個好策略
03:47
So they're smart聰明, but poor較差的.
84
215406
2112
所以他們聰明卻也貧窮
03:49
(Laughter笑聲)
85
217518
2088
(笑聲)
03:51
Where are these things happening事件 in the brain?
86
219606
1969
那大腦中發生了什麼事呢?
03:53
One study研究 by CoricelliCoricelli and Nagel內格爾 gives a really sharp尖銳, interesting有趣 answer回答.
87
221575
3875
柯里切利和納格爾
的一項研究給了明顯、有趣的答案
03:57
So they had people play this game遊戲
88
225450
1508
他們讓一些人玩這遊戲時
03:58
while they were being存在 scanned掃描 in an fMRI功能磁共振成像,
89
226958
2217
同時接受功能型核磁共振(fMRI)
04:01
and two conditions條件: in some trials試驗,
90
229175
2271
包含兩種狀況:在其中一組
04:03
they're told you're playing播放 another另一個 person
91
231446
1515
受試者被告知他們是
04:04
who's誰是 playing播放 right now and we're going to match比賽 up
92
232961
1588
和遊戲中的人玩
而我們最後會比較你們的作為
04:06
your behavior行為 at the end結束 and pay工資 you if you win贏得.
93
234549
2204
如果你贏了,就會得到獎勵
04:08
In the other trials試驗, they're told, you're playing播放 a computer電腦.
94
236753
1978
另外一組受試者
被告知他們是在電腦玩
04:10
They're just choosing選擇 randomly隨機.
95
238731
1634
受試者隨機分組
04:12
So what you see here is a subtraction減法
96
240365
2077
所以你們可以看到
04:14
of areas in which哪一個 there's more brain activity活動
97
242442
2750
當和對電腦玩的人比較時
04:17
when you're playing播放 people compared相比 to playing播放 the computer電腦.
98
245192
2976
對手是人的受試者
腦部活化程度較高
04:20
And you see activity活動 in some regions地區 we've我們已經 seen看到 today今天,
99
248168
1991
從圖上你可以看到
一些我們已知的區域
04:22
medial內側 prefrontal前額葉 cortex皮質, dorsomedial背內側, however然而, up here,
100
250159
3237
內側額葉皮層,事實上
在這裡的是背內側額葉皮層
04:25
ventromedial prefrontal前額葉 cortex皮質,
101
253396
1851
腹內前側額葉皮層
04:27
anterior前面的 cingulate扣帶, an area that's involved參與
102
255247
1354
和像是你在玩「老師說」遊戲時
04:28
in lots of types類型 of conflict衝突 resolution解析度, like if you're playing播放 "Simon西蒙 Says,"
103
256601
3637
處理大量衝突解決的前扣帶皮層
04:32
and also the right and left temporoparietal junction連接點.
104
260238
3814
以及左右邊的躡頂葉交界區
04:36
And these are all areas which哪一個 are fairly相當 reliably可靠 known已知
105
264052
2466
而這皆我們所熟知的區域
04:38
to be part部分 of what's called a "theory理論 of mind心神" circuit電路,
106
266518
2321
被納入一個名為「心靈原理」迴路
04:40
or "mentalizing心理化 circuit電路."
107
268839
1901
或稱為「心靈迴路」的一部份
04:42
That is, it's a circuit電路 that's used to imagine想像 what other people might威力 do.
108
270740
3378
這迴路是用來想像別人可能的行動
04:46
So these were some of the first studies學習 to see this
109
274118
2240
所以這部份早期的研究
04:48
tied in to game遊戲 theory理論.
110
276358
2033
和賽局理論有關
04:50
What happens發生 with these one-一- and two-step兩步 types類型?
111
278391
2240
但那些「想一步」和
「想兩步」的玩家又是如何呢?
04:52
So we classify分類 people by what they picked採摘的,
112
280631
2071
所以我們將受試者依其選擇區分
04:54
and then we look at the difference區別 between之間
113
282702
1667
然後我們再看看
04:56
playing播放 humans人類 versus playing播放 computers電腦,
114
284369
1975
對手是人或電腦
04:58
which哪一個 brain areas are differentially差異 active活性.
115
286344
1891
在腦部活化區域層面的不同
05:00
On the top最佳 you see the one-step一步 players玩家.
116
288235
1752
上方你可以看到「想一步」的玩家
05:01
There's almost幾乎 no difference區別.
117
289987
1520
它們幾乎沒有不同
05:03
The reason原因 is, they're treating治療 other people like a computer電腦, and the brain is too.
118
291507
2940
原因是他們將人視為電腦,
所以大腦也就如此反應
05:06
The bottom底部 players玩家, you see all the activity活動 in dorsomedial背內側 PFCPFC.
119
294447
4141
你可從下方玩家中
看到背內側額葉皮層的活動
05:10
So we know that those two-step兩步 players玩家 are doing something differently不同.
120
298588
2051
所以我們知道這些
「想兩步」的玩家有些不同
05:12
Now if you were to step back and say, "What can we do with this information信息?"
121
300639
3096
你可能會停下來想想
「我們從這些資訊中得到什麼?」
05:15
you might威力 be able能夠 to look at brain activity活動 and say,
122
303735
1821
你可能從大腦活動中判斷
05:17
"This person's人的 going to be a good poker撲克 player播放機,"
123
305556
1499
「這人會成為撲克牌高手」
05:19
or, "This person's人的 socially社交上 naive幼稚,"
124
307055
1929
或「這人是交際新手」
05:20
and we might威力 also be able能夠 to study研究 things
125
308984
1278
既然我們知道這個迴路的存在
05:22
like development發展 of adolescent青少年 brains大腦
126
310262
1598
我們也可以研究一些
05:23
once一旦 we have an idea理念 of where this circuitry電路 exists存在.
127
311860
3354
像是青少年腦部的發育
05:27
Okay. Get ready準備.
128
315214
2612
準備好囉
05:29
I'm saving保存 you some brain activity活動,
129
317826
2123
我為你們準備了些腦力激盪
05:31
because you don't need to use your hair頭髮 detector探測器 cells細胞.
130
319949
2810
你們不用擔心想太多會掉頭髮
05:34
You should use those cells細胞 to think carefully小心 about this game遊戲.
131
322759
2888
你們應該專注思考這個遊戲
05:37
This is a bargaining議價 game遊戲.
132
325647
1935
這是個協商遊戲
05:39
Two players玩家 who are being存在 scanned掃描 using運用 EEG腦電圖 electrodes電極
133
327582
2556
兩位被接上腦電波圖(EEG) 電極的玩家
05:42
are going to bargain討價還價 over one to six dollars美元.
134
330138
2877
正在進行一項由 1 到 6 元的協商
05:45
If they can do it in 10 seconds, they're going to actually其實 earn that money.
135
333015
2664
如果在 10 秒內完成,
最後就會得到這份金錢
05:47
If 10 seconds goes by and they haven't沒有 made製作 a deal合同, they get nothing.
136
335679
3040
如果未在 10 秒內完成協商,就得不到錢
05:50
That's kind of a mistake錯誤 together一起.
137
338719
1683
這是種兩人一起造成的錯誤
05:52
The twist is that one player播放機, on the left,
138
340402
2817
不同的地方在於左邊玩家
05:55
is informed通知 about how much on each trial審訊 there is.
139
343219
2688
確實知道每場協商的底價
05:57
They play lots of trials試驗 with different不同 amounts each time.
140
345907
2232
而他們會重複多次、
所包含的金額不同的協商
06:00
In this case案件, they know there's four dollars美元.
141
348139
2241
在這個例子中,他們知道是 4 元
06:02
The uninformed不知情 player播放機 doesn't know,
142
350380
1877
而一邊玩家不知道底價
06:04
but they know that the informed通知 player播放機 knows知道.
143
352257
2054
只知到他們的對手知道
06:06
So the uninformed不知情 player's玩家 challenge挑戰 is to say,
144
354311
2059
所以不知道底價的玩家會質疑
06:08
"Is this guy really being存在 fair公平
145
356370
1470
「對方真的很公平
06:09
or are they giving me a very low offer提供
146
357840
1854
或是他只給我很低的金額
06:11
in order訂購 to get me to think that there's only one or two dollars美元 available可得到 to split分裂?"
147
359694
3078
讓我以為我們真的
只有 1、2 塊錢可以分 ?」
06:14
in which哪一個 case案件 they might威力 reject拒絕 it and not come to a deal合同.
148
362772
3154
如此他們可能會拒絕對方,
而讓協議流產
06:17
So there's some tension張力 here between之間 trying to get the most money
149
365926
2950
所以這包含「想要的到最多的錢」
06:20
but trying to goad刺棒 the other player播放機 into giving you more.
150
368876
2573
和「如何促使對手給你更多錢」的衝突
06:23
And the way they bargain討價還價 is to point on a number line
151
371449
2330
他們議價方式是
06:25
that goes from zero to six dollars美元,
152
373779
1806
是從 0 到 6 這條線上指出一點
06:27
and they're bargaining議價 over how much the uninformed不知情 player播放機 gets得到,
153
375585
2978
他們是在協商「不知情玩家」該得多少
06:30
and the informed通知 player's玩家 going to get the rest休息.
154
378563
1585
而「知情玩家」會得到剩下的
06:32
So this is like a management-labor管理勞動 negotiation談判
155
380148
2575
這就好像顧主和勞工的協商
06:34
in which哪一個 the workers工人 don't know how much profits利潤
156
382723
2733
勞工不知道
06:37
the privately私自 held保持 company公司 has, right,
157
385456
2667
公司獲利經額的資訊
06:40
and they want to maybe hold保持 out for more money,
158
388123
2368
他們想要得到更多的錢
06:42
but the company公司 might威力 want to create創建 the impression印象
159
390491
1836
但是公司會想營造一個假象
06:44
that there's very little to split分裂: "I'm giving you the most that I can."
160
392327
2932
就是營收很少而且
「我已將能給的部份都給出去了」
06:47
First some behavior行為. So a bunch of the subject學科 pairs, they play face面對 to face面對.
161
395259
4231
首先看看行為,
一部份玩家座在長桌上、面對面協商
06:51
We have some other data數據 where they play across橫過 computers電腦.
162
399490
1836
我們也有一些
他們透過電腦協商的數據
06:53
That's an interesting有趣 difference區別, as you might威力 imagine想像.
163
401326
1738
你可以想像
兩者的差距十分有趣
06:55
But a bunch of the face-to-face面對面 pairs
164
403064
2202
長桌上面對面的協商
06:57
agree同意 to divide劃分 the money evenly every一切 single time.
165
405266
3693
會同意每次都均分金錢
07:00
Boring. It's just not interesting有趣 neurallyneurally.
166
408959
2906
無聊死了!
平均分配並不有趣
07:03
It's good for them. They make a lot of money.
167
411865
2514
這對他們很好,因為他們拿到很多錢
07:06
But we're interested有興趣 in, can we say something about
168
414379
2672
我們感興趣的是
07:09
when disagreements分歧 occur發生 versus don't occur發生?
169
417051
2536
如何解釋歧見是否發生
07:11
So this is the other group of subjects主題 who often經常 disagree不同意.
170
419587
2357
這是通常都達不成協議的一組
07:13
So they have a chance機會 of -- they bicker鬥嘴 and disagree不同意
171
421944
2768
他們有機會因歧見而爭執
07:16
and end結束 up with less money.
172
424712
1307
最後得到很少錢
07:18
They might威力 be eligible合格 to be on "Real真實 Housewives家庭主婦," the TV電視 show顯示.
173
426019
3917
他們協商情景
甚至可以改編成實境秀
07:21
You see on the left,
174
429936
1936
你看左側
07:23
when the amount to divide劃分 is one, two or three dollars美元,
175
431872
2664
當配額為 1 到 3 元時
07:26
they disagree不同意 about half the time,
176
434536
1648
他們約有 50% 會拒絕
07:28
and when the amount is four, five, six, they agree同意 quite相當 often經常.
177
436184
2192
而當配額為 4 到 6 元則會接受
07:30
This turns out to be something that's predicted預料到的
178
438376
1874
這就是我們所預期的
07:32
by a very complicated複雜 type類型 of game遊戲 theory理論
179
440250
2204
非常複雜的賽局理論
07:34
you should come to graduate畢業 school學校 at CalTech加州理工學院 and learn學習 about.
180
442454
2809
你應該到加州理工的研究所來學學
07:37
It's a little too complicated複雜 to explain說明 right now,
181
445263
2172
現在要解釋這太困難了
07:39
but the theory理論 tells告訴 you that this shape形狀 kind of should occur發生.
182
447435
3416
但這理論預期會出現這種趨勢
07:42
Your intuition直覺 might威力 tell you that too.
183
450851
2216
而你的直覺或許也是這樣告訴你
07:45
Now I'm going to show顯示 you the results結果 from the EEG腦電圖 recording記錄.
184
453067
2240
現在我要給你們
看腦電波圖(EEG)的紀錄
07:47
Very complicated複雜. The right brain schematic概要
185
455307
2353
這非常複雜
右邊的腦部模型
07:49
is the uninformed不知情 person, and the left is the informed通知.
186
457660
2863
是代表「不知情玩家」
而左邊則是「知情玩家」
07:52
Remember記得 that we scanned掃描 both brains大腦 at the same相同 time,
187
460523
2800
記得我們是同時監控兩個大腦
07:55
so we can ask about time-synced時間同步 activity活動
188
463323
2392
所以我們可以探討
在相同或相異大腦區域中
07:57
in similar類似 or different不同 areas simultaneously同時,
189
465715
3224
在一段時間內的活化狀況
08:00
just like if you wanted to study研究 a conversation會話
190
468939
2264
就像你想研究一段對話
08:03
and you were scanning掃描 two people talking to each other
191
471203
1936
你就同時掃描兩個人的腦部活動
08:05
and you'd expect期望 common共同 activity活動 in language語言 regions地區
192
473139
2360
而你會預期當他們正在對話時
08:07
when they're actually其實 kind of listening and communicating通信.
193
475499
2385
共同活化的是語言區
08:09
So the arrows箭頭 connect regions地區 that are active活性 at the same相同 time,
194
477884
3927
所以這個箭頭連接了同時活化的區域
08:13
and the direction方向 of the arrows箭頭 flows流動
195
481811
2040
而箭號方向從
08:15
from the region地區 that's active活性 first in time,
196
483851
2480
最先活化的區位指出去
08:18
and the arrowhead箭頭 goes to the region地區 that's active活性 later後來.
197
486331
3568
箭頭部份表示較後期活化的區域
08:21
So in this case案件, if you look carefully小心,
198
489899
2216
在這個例子中,如果你仔細看
08:24
most of the arrows箭頭 flow from right to left.
199
492115
1857
大部份的箭號是由右而左
08:25
That is, it looks容貌 as if the uninformed不知情 brain activity活動
200
493972
3280
看起來「不知情玩家」的
08:29
is happening事件 first,
201
497252
1959
腦部較先活化
08:31
and then it's followed其次 by activity活動 in the informed通知 brain.
202
499211
3515
接著才是「知情玩家」
08:34
And by the way, these were trials試驗 where their deals交易 were made製作.
203
502726
3692
順帶一提,
這是在達成協議的測試中出現的
08:38
This is from the first two seconds.
204
506418
1780
這是最初兩秒鐘所得到的數據
08:40
We haven't沒有 finished analyzing分析 this data數據,
205
508198
1980
我們還沒完成這部份的分析
08:42
so we're still peeking偷看 in, but the hope希望 is
206
510178
1900
所以這只是大概觀察到的狀況
08:44
that we can say something in the first couple一對 of seconds
207
512078
2564
但我們希望從前 1、2 秒鐘
08:46
about whether是否 they'll他們會 make a deal合同 or not,
208
514642
1723
就可以判斷他們是否能達成協議
08:48
which哪一個 could be very useful有用 in thinking思維 about avoiding避免 litigation訴訟
209
516365
2043
這可以用在避免
協商破裂,而走向進訴訟程序
08:50
and ugly醜陋 divorces離婚 and things like that.
210
518408
1928
例如:最後以不愉快的離婚收場
08:52
Those are all cases in which哪一個 a lot of value is lost丟失
211
520336
2883
在很多案例中
08:55
by delay延遲 and strikes罷工.
212
523219
2976
是因為延誤和罷工而減損的價值
08:58
Here's這裡的 the case案件 where the disagreements分歧 occur發生.
213
526195
2030
而這都是因為歧見
09:00
You can see it looks容貌 different不同 than the one before.
214
528225
2173
你會看到這和前一個例子不同
09:02
There's a lot more arrows箭頭.
215
530398
2249
上面有很箭號
09:04
That means手段 that the brains大腦 are synced已同步 up
216
532647
1511
這表示大腦正在同步運作
09:06
more closely密切 in terms條款 of simultaneous同時 activity活動,
217
534158
2552
更確切的說是同時間的活動
09:08
and the arrows箭頭 flow clearly明確地 from left to right.
218
536710
2010
而箭號由左指向右
09:10
That is, the informed通知 brain seems似乎 to be deciding決定,
219
538720
2242
這好像是「知情玩家」的大腦開使思考
09:12
"We're probably大概 not going to make a deal合同 here."
220
540962
2288
「這個可能是失敗的協商」的訊號
09:15
And then later後來 there's activity活動 in the uninformed不知情 brain.
221
543250
3225
後來這個活化模式也出現在「不知情玩家」
09:18
Next下一個 I'm going to introduce介紹 you to some relatives親戚們.
222
546475
2503
接著讓我來介紹我們的近親
09:20
They're hairy毛茸茸, smelly, fast快速 and strong強大.
223
548978
2261
牠們多毛、體味重、迅捷和強壯
09:23
You might威力 be thinking思維 back to your last Thanksgiving感恩.
224
551239
3190
你可能會想到你上個感恩節的裝扮
09:26
Maybe if you had a chimpanzee黑猩猩 with you.
225
554429
2693
如果你帶了知黑猩猩的話
09:29
Charles查爾斯 Darwin達爾文 and I and you broke打破 off from the family家庭 tree
226
557122
3461
查理.達爾文、你和我在 50 萬年前
09:32
from chimpanzees黑猩猩 about five million百萬 years年份 ago.
227
560583
2259
從演化樹上和黑猩猩分開
09:34
They're still our closest最近的 genetic遺傳 kin親屬.
228
562842
1968
遺傳上,是牠們仍是最接近我們的親戚
09:36
We share分享 98.8 percent百分 of the genes基因.
229
564810
1737
我們之間有 98.8 % 相同的基因
09:38
We share分享 more genes基因 with them than zebras斑馬 do with horses馬匹.
230
566547
2800
我們之間的共同基因
比斑馬和野馬間多的多
09:41
And we're also their closest最近的 cousin表姐.
231
569347
1717
我們也是牠們最接近的親戚
09:43
They have more genetic遺傳 relation關係 to us than to gorillas大猩猩.
232
571064
3002
牠們和我們的基因相似度
比和大猩猩更接近
09:46
So how humans人類 and chimpanzees黑猩猩 behave表現 differently不同
233
574066
2528
所以人類和黑猩猩行為的不同
09:48
might威力 tell us a lot about brain evolution演化.
234
576594
2455
可能提供我們瞭解腦部演化的線索
09:51
So this is an amazing驚人 memory記憶 test測試
235
579049
2601
這是日本名古屋靈長類研究中心
09:53
from Nagoya名古屋, Japan日本, Primate靈長類動物 Research研究 Institute研究所,
236
581650
2816
一個令人驚歎的記憶測試
09:56
where they've他們已經 doneDONE a lot of this research研究.
237
584466
1778
他們在那裡做了很多類似的研究
09:58
This goes back quite相當 a ways方法. They're interested有興趣 in working加工 memory記憶.
238
586244
2340
這已經是早期的研究,
他們對記憶形成很有興趣
10:00
The chimp黑猩猩 is going to see, watch carefully小心,
239
588584
1772
猩猩會有 200 毫秒的時間
10:02
they're going to see 200 milliseconds'毫秒“ exposure曝光
240
590356
2202
仔細去看
10:04
— that's fast快速, that's eight movie電影 frames
241
592558
1994
—— 這是很快的 8 個字卡 ——
10:06
of numbers數字 one, two, three, four, five.
242
594552
1951
數字從 1 到 5
10:08
Then they disappear消失 and they're replaced更換 by squares廣場,
243
596503
1998
當字卡消失後會出現方塊
10:10
and they have to press the squares廣場
244
598501
1755
牠們必須依據數字大小順序
10:12
that correspond對應 to the numbers數字 from low to high
245
600256
2321
按壓方塊
10:14
to get an apple蘋果 reward獎勵.
246
602577
1339
正確就能得到蘋果
10:15
Let's see how they can do it.
247
603916
4771
讓我們看看牠們表現如何
10:28
This is a young年輕 chimp黑猩猩. The young年輕 ones那些
248
616391
1493
這是隻年輕的黑猩猩
10:29
are better than the old ones那些, just like humans人類.
249
617884
2783
像人類一樣
年輕的表現比年老的好
10:32
And they're highly高度 experienced有經驗的, so they've他們已經 doneDONE this
250
620667
1592
而且牠們已經做了幾千次了
10:34
thousands數千 and thousands數千 of time.
251
622259
1432
可以說是駕輕就熟
10:35
Obviously明顯 there's a big training訓練 effect影響, as you can imagine想像.
252
623691
2884
你可以想像有很大一部份歸於訓練
10:39
(Laughter笑聲)
253
627928
1344
(笑聲)
10:41
You can see they're very blas布拉斯é and kind of effortless毫不費力.
254
629272
1935
你可以看到牠們
已經呈現玩膩時倦怠
10:43
Not only can they do it very well, they do it in a sort分類 of lazy way.
255
631207
3928
他們不只做的很好
而且是用一種慵懶的方式完成
10:47
Right? Who thinks you could beat擊敗 the chimps黑猩猩?
256
635135
3435
對嗎?誰認為自己可以打敗這些猩猩呢?
10:50
Wrong錯誤. (Laughter笑聲)
257
638570
1596
錯了 (笑聲)
10:52
We can try. We'll try. Maybe we'll try.
258
640166
2438
我們可以嘗試
或許我們真的會試試
10:54
Okay, so the next下一個 part部分 of this study研究
259
642604
2590
好的,這研究的另一部份
10:57
I'm going to go quickly很快 through通過
260
645194
1596
我會很快的帶過
10:58
is based基於 on an idea理念 of Tetsuro Matsuzawa松澤.
261
646790
2692
這是基於松澤綱野的想法
11:01
He had a bold膽大 idea理念 that -- what he called the cognitive認知 trade-off交易 hypothesis假設.
262
649482
3029
他有一個想法
稱為認知妥協假說
11:04
We know chimps黑猩猩 are faster更快 and stronger.
263
652511
1292
我們知道猩猩迅捷且強壯
11:05
They're also very obsessed痴迷 with status狀態.
264
653803
1680
牠們也對自身狀況非常執著
11:07
His thought was, maybe they've他們已經 preserved罐頭 brain activities活動
265
655483
2956
他認為,或許黑猩猩保留了
11:10
and they practice實踐 them in development發展
266
658439
2168
對牠們非常重要
11:12
that are really, really important重要 to them
267
660607
1851
用來協商條件和獲勝的腦部活動
11:14
to negotiate談判 status狀態 and to win贏得,
268
662458
2210
並在成長過程中練習
11:16
which哪一個 is something like strategic戰略 thinking思維 during competition競爭.
269
664668
2998
這就像是在競爭中的策略思考
11:19
So we're going to check that out
270
667666
1580
所以我們接著就去證實它
11:21
by having the chimps黑猩猩 actually其實 play a game遊戲
271
669246
2430
藉由讓黑猩猩玩個遊戲
11:23
by touching接觸 two touch觸摸 screens屏幕.
272
671676
2638
藉由點擊兩個觸控螢幕
11:26
The chimps黑猩猩 are actually其實 interacting互動 with each other through通過 the computers電腦.
273
674314
2440
猩猩真的可以借由電腦互相交流
11:28
They're going to press left or right.
274
676754
1608
他們可以選擇按「左」或「右」
11:30
One chimp黑猩猩 is called a matcher匹配.
275
678362
2113
其中一隻猩猩作為「協調者」
11:32
They win贏得 if they press left, left,
276
680475
1983
在兩隻猩猩都選同一邊時獲勝
11:34
like a seeker導引頭 finding發現 someone有人 in hide-and-seek捉迷藏, or right, right.
277
682458
3169
有點像是在玩捉迷藏
11:37
The mismatchermismatcher wants to mismatch不匹配.
278
685627
1228
「不諧調者」希望不一樣的組合
11:38
They want to press the opposite對面 screen屏幕 of the chimp黑猩猩.
279
686855
3076
牠們希望選擇和對手不同的選項
11:41
And the rewards獎勵 are apple蘋果 cube立方體 rewards獎勵.
280
689931
2544
而得到蘋果塊的獎賞
11:44
So here's這裡的 how game遊戲 theorists理論家 look at these data數據.
281
692475
2528
賽局理論學家是這樣看待這些數據的
11:47
This is a graph圖形 of the percentage百分比 of times
282
695003
1614
這圖的 X 軸表示
11:48
the matcher匹配 picked採摘的 right on the x-axisx軸,
283
696617
2618
「協調者」選擇「右」的比率
11:51
and the percentage百分比 of times they predicted預料到的 right
284
699235
1516
而 Y 軸表示
11:52
by the mismatchermismatcher on the y-axisy軸.
285
700751
2868
是牠們預測「不諧調者」會選「右」的比例
11:55
So a point here is the behavior行為 by a pair of players玩家,
286
703619
3200
所以上面的一點是由
兩位玩家共同決定的
11:58
one trying to match比賽, one trying to mismatch不匹配.
287
706819
2216
一個試著選擇一樣,
另一個希望有不同組合
12:01
The NENE square廣場 in the middle中間 -- actually其實 NENE, CHCH and QREQRE --
288
709035
3280
這個中間標示 NE 的正方形,
事實上包含 NE 、 CH 、 QRE
12:04
those are three different不同 theories理論 of Nash納什 equilibrium平衡, and others其他,
289
712315
2456
代表三種包含
納許均衡理論在內的三個理論
12:06
tells告訴 you what the theory理論 predicts預測,
290
714771
2424
對於牠們選擇的預測
12:09
which哪一個 is that they should match比賽 50-50,
291
717195
2272
牠們的選擇應該是 50 比 50
12:11
because if you play left too much, for example,
292
719467
2168
因為如果你選了太多左邊
12:13
I can exploit利用 that if I'm the mismatchermismatcher by then playing播放 right.
293
721635
2716
而我是「不諧調者」,
我就會傾向選右邊進行破壞
12:16
And as you can see, the chimps黑猩猩, each chimp黑猩猩 is one triangle三角形,
294
724351
2764
如你們所見,
每隻黑猩猩都以三角形表示
12:19
are circled盤旋 around, hovering徘徊 around that prediction預測.
295
727115
3847
三角形都坐落在預測位置的附近
12:22
Now we move移動 the payoffs收益.
296
730962
1774
然後我們改變獎勵條件
12:24
We're actually其實 going to make the left, left payoff付清 for the matcher匹配 a little bit higher更高.
297
732736
3443
我們要讓選擇「左」的
「協調者」獲得比較高的獎勵
12:28
Now they get three apple蘋果 cubes立方體.
298
736179
1520
現在牠們會得到三個蘋果塊
12:29
Game遊戲 theoretically理論上, that should actually其實 make the mismatcher'smismatcher的 behavior行為 shift轉移,
299
737699
2800
依據賽局理論
這會讓「不諧調者」行為改變
12:32
because what happens發生 is, the mismatchermismatcher will think,
300
740499
1590
因為「不諧調者」會想
12:34
oh, this guy's傢伙 going to go for the big reward獎勵,
301
742089
1810
牠的對手會選則獲利大的地方
12:35
and so I'm going to go to the right, make sure he doesn't get it.
302
743899
3065
所以我要選「右」,讓牠拿不到
12:38
And as you can see, their behavior行為 moves移動 up
303
746964
1665
而你們可以看到牠們的行為
12:40
in the direction方向 of this change更改 in the Nash納什 equilibrium平衡.
304
748629
3448
逐漸趨向納許均衡的預測
12:44
Finally最後, we changed the payoffs收益 one more time.
305
752077
2314
最後,我們再次改變獎勵條件
12:46
Now it's four apple蘋果 cubes立方體,
306
754391
1192
現在是四個蘋果塊
12:47
and their behavior行為 again moves移動 towards the Nash納什 equilibrium平衡.
307
755583
2172
他們的行為會更接近納許平衡
12:49
It's sprinkled around, but if you average平均 the chimps黑猩猩 out,
308
757755
2035
如果將所有黑猩猩的選擇平均
會出現一個高峰
12:51
they're really, really close, within .01.
309
759790
2190
它們會非常接近,間距在 0.1 之內
12:53
They're actually其實 closer接近 than any species種類 we've我們已經 observed觀察到的.
310
761980
3199
結果會比我們觀察的
任何一種物種都接近
12:57
What about humans人類? You think you're smarter聰明 than a chimpanzee黑猩猩?
311
765179
3193
那人類又會如何呢?
你們認為人類比黑猩猩聰明?
13:00
Here's這裡的 two human人的 groups in green綠色 and blue藍色.
312
768372
3567
這裡用藍、綠色表示兩組人類
13:03
They're closer接近 to 50-50. They're not responding響應 to payoffs收益 as closely密切,
313
771939
3744
他們選擇接近 50 比 50,
但對獎勵的反應和納許平衡相距較遠
13:07
and also if you study研究 their learning學習 in the game遊戲,
314
775683
1450
而且你如果研究
他們在遊戲中學習的能力
13:09
they aren't as sensitive敏感 to previous以前 rewards獎勵.
315
777133
1985
他們對之前獲得的獎勵較不敏感
13:11
The chimps黑猩猩 are playing播放 better than the humans人類,
316
779118
1364
黑猩猩玩得比人類好
13:12
better in the sense of adhering秉承 to game遊戲 theory理論.
317
780482
2423
是以接近賽局理論而言的好
13:14
And these are two different不同 groups of humans人類
318
782905
1419
而這兩組分別來自日本和非洲的人
13:16
from Japan日本 and Africa非洲. They replicate複製 quite相當 nicely很好.
319
784324
3196
他們的結果相互吻合
13:19
None沒有 of them are close to where the chimps黑猩猩 are.
320
787520
3235
但都不及黑猩猩
13:22
So here are some things we learned學到了 today今天.
321
790755
1755
所以今天我們學到了一些事情
13:24
People seem似乎 to do a limited有限 amount of strategic戰略 thinking思維
322
792510
1938
從大腦理論的角度
13:26
using運用 theory理論 of mind心神.
323
794448
1811
人類的策略思考似乎是有極限的
13:28
We have some preliminary初步 evidence證據 from bargaining議價
324
796259
1917
我們有些從議價得到的初步證據
13:30
that early warning警告 signs跡象 in the brain might威力 be used to predict預測
325
798176
2492
有些大腦活動可以作為
13:32
whether是否 there will be a bad disagreement異議 that costs成本 money,
326
800668
2226
可能造成協商破裂虧損的警訊
13:34
and chimps黑猩猩 are better competitors競爭對手 than humans人類,
327
802894
1840
從賽局理論看來
13:36
as judged判斷 by game遊戲 theory理論.
328
804734
2464
黑猩猩是比人類更好的競爭者
13:39
Thank you.
329
807198
1857
謝謝大家
13:41
(Applause掌聲)
330
809055
3573
(掌聲)
Translated by Hao-Wei Chang
Reviewed by Hermia Tsai

▲Back to top

ABOUT THE SPEAKER
Colin Camerer - Behavioral economist
Colin Camerer is a leading behavioral economist who studies the psychological and neural bases of choice and strategic decision-making.

Why you should listen

Colin Camerer focuses on brain behavior during decision making, strategizing and market trading. He is the Robert Kirby Professor of Behavioral Finance and Economics at the California Institute of Technology. A child prodigy in his youth, Camerer received a B.A. in quantitative studies from Johns Hopkins when he was just 17 and a PhD in decision theory from the University of Chicago Graduate School of Business when he was 22. Camerer's research departs from previous theory in that it does not assume the mind to be a rational and perfect system, but rather focuses on the limitations of everyday people when they play actual games, and seeks to predict how they will behave in situations that involve strategy. His studies focus on neurological findings from economic experiments in the lab (on humans -- and monkeys!) Camerer is the author of Behavioral Game Theory.

More profile about the speaker
Colin Camerer | Speaker | TED.com