字幕列表 影片播放
I do two things:
譯者: Bill Hsiung 審譯者: Calvin Chun-yu Chan
I design mobile computers and I study brains.
我有兩個職業。我設計行動電腦,而且我研究大腦。
Today's talk is about brains and -- (Audience member cheers)
今天的演講與大腦有關,
Yay! I have a brain fan out there.
耶,看來今天聽眾中有人是大腦迷。
(Laughter)
(笑聲)
If I could have my first slide,
如果我的投影片已經準備好了,
you'll see the title of my talk and my two affiliations.
你將會看到今天的演講主題及我的兩個所屬機構,
So what I'm going to talk about is why we don't have a good brain theory,
今天我將要談的是 — 為什麼我們沒有一個好的大腦理論,
why it is important that we should develop one
為什麼發展大腦理論如此重要,還有,我們能利用這個理論做什麼?
and what we can do about it.
我將會嘗試在廿分鐘內完成全部的主題。我參與兩家公司。
I'll try to do all that in 20 minutes.
你們大多數是因為我在 Palm 及 Handspring 的工作而認識我的,
I have two affiliations.
但是我同時也經營一個非營利性的科學研究機構
Most of you know me from my Palm and Handspring days,
它位於加州門洛帕克,叫做「紅木神經科學研究所」,
but I also run a nonprofit scientific research institute
我們專攻理論神經科學相關的研究,
called the Redwood Neuroscience Institute in Menlo Park.
我們對研究大腦新皮層如何運作有興趣。
We study theoretical neuroscience and how the neocortex works.
我將談談這一方面。
I'm going to talk all about that.
我將我的另一個生活面(電腦生活)做成了一張投影片,你現在可以看到。
I have one slide on my other life, the computer life,
我在過去的廿年間參與了一些產品的開發,
and that's this slide here.
從第一台筆記型電腦到首批平板電腦等等,
These are some of the products I've worked on over the last 20 years,
最新的一個產品是 Treo,
starting from the very original laptop
我們將會繼續電子產品的開發。
to some of the first tablet computers
我之所以會參與這一行主要是因為我相信行動運算
and so on, ending up most recently with the Treo,
是個人運算產品的未來,而我試著藉由開發這些產品
and we're continuing to do this.
來讓世界更美好。
I've done this because I believe mobile computing
但是我必須承認,這一切都是個意外。
is the future of personal computing,
我其實本來一點都沒有打算要開發這些產品
and I'm trying to make the world a little bit better
而且在我事業剛剛開始的時候我還決定
by working on these things.
我不要從事電腦相關產業。
But this was, I admit, all an accident.
但在我告訴你這個故事之前,我必須告訴你
I really didn't want to do any of these products.
我某天從網路上看到的一張關於 graffiti 輸入法照片的故事。
Very early in my career
當時我在網上尋找 graffiti 的照片,那是一種輸入法程式語言,
I decided I was not going to be in the computer industry.
然後我發現一個網站,它是為一群老師們所架設的,你知道的,
Before that, I just have to tell you
利用 script 來控制黑板上的跑馬燈,
about this picture of Graffiti I picked off the web the other day.
他們網站內容竟然包含 graffiti,我對此感到很抱歉。
I was looking for a picture for Graffiti that'll text input language.
(笑聲)
I found a website dedicated to teachers who want to make script-writing things
當我還年輕,剛剛從工學院畢業的時候,
across the top of their blackboard,
我是康乃爾 79 年畢業班,我決定去 Intel 工作。
and they had added Graffiti to it, and I'm sorry about that.
我在電腦業奮鬥了三個月之後,
(Laughter)
我愛上了另一個東西,我說:「我入錯行了」,
So what happened was,
因為我愛上了大腦。
when I was young and got out of engineering school at Cornell in '79,
這不是真的大腦。這是大腦的描繪圖。
I went to work for Intel and was in the computer industry,
我已經記不清當初是如何開始的了,
and three months into that, I fell in love with something else.
在我腦海中只有一個鮮明的回憶。
I said, "I made the wrong career choice here,"
1979 年九月,新一期的科學美國人出刊
and I fell in love with brains.
那是一期談論大腦的特刊。非常的棒。
This is not a real brain.
那是有史以來最棒的一期雜誌之一。那期刊物中談論神經、
This is a picture of one, a line drawing.
發育、疾病以及視力等等所有的
And I don't remember exactly how it happened,
跟大腦相關且你會感興趣的主題。真的非常令人印象深刻。
but I have one recollection, which was pretty strong in my mind.
而人會得到一種錯誤的印象,那就是我們已經非常了解我們的大腦了。
In September of 1979,
但是那一期的最後一篇文章是由發現 DNA 結構而成名的法蘭西斯•克里克所撰寫。
Scientific American came out with a single-topic issue about the brain.
今天,如果我沒記錯的話,剛好是發現 DNA 結構五十週年紀念日。
It was one of their best issues ever.
他寫了一個故事,主要是告訴我們:
They talked about the neuron, development, disease, vision
這個嘛~這些研究都很棒,可是你知道嗎?
and all the things you might want to know about brains.
我們對大腦一點都不了解
It was really quite impressive.
沒有人知道大腦是如何運作的,
One might've had the impression we knew a lot about brains.
所以別相信其他人告訴你的事情。
But the last article in that issue was written by Francis Crick of DNA fame.
這是從文章中摘錄下來的一句話。他說:「這裡顯著缺乏的是,」
Today is, I think, the 50th anniversary of the discovery of DNA.
他是一個非常有禮的英國紳士,「我們會注意到可以用來解釋這些研究
And he wrote a story basically saying, this is all well and good,
的廣泛概念架構是明顯地不足的。」
but you know, we don't know diddly squat about brains,
我認為他用「架構」一詞用得非常洽當。
and no one has a clue how they work,
他並沒有說我們連一個理論都沒有。他所說得是,
so don't believe what anyone tells you.
我們連如何開始建立理論都不知道該如何下手 —
This is a quote from that article, he says:
我們連個架構都沒有。
"What is conspicuously lacking" -- he's a very proper British gentleman --
如果你想要引用湯瑪斯•孔恩的說法,我們處在一個前典範的時代。
"What is conspicuously lacking is a broad framework of ideas
因此我愛上這個領域了,然後說:看看,
in which to interpret these different approaches."
我們已經知道這麼多關於腦的知識。這會有多難?
I thought the word "framework" was great.
而且這是個可以一輩子鑽研的題目。我認為我能對世界做出一點貢獻,
He didn't say we didn't have a theory.
因此我嘗試著離開電腦業,轉行到腦科學研究領域。
He says we don't even know how to begin to think about it.
首先,我跑去麻省理工裡的一間人工智慧實驗室,
We don't even have a framework.
我說,嘿,我也想要建造智能機器,
We are in the pre-paradigm days, if you want to use Thomas Kuhn.
但是我覺得達到這個目標前必須要先能了解大腦是如何運作的。
So I fell in love with this.
然而他們說,喔,你並不需要知道那個。
I said, look: We have all this knowledge about brains -- how hard can it be?
我們只需要設計電腦程式,不需要做其他不相干的事。
It's something we can work on in my lifetime; I could make a difference.
我再說,不,你們真的應該研究大腦。他們說,喔,你知道嗎?
So I tried to get out of the computer business, into the brain business.
你錯了。然後我說,不,你才錯了,所以當然我沒被錄取。
First, I went to MIT, the AI lab was there.
(笑聲)
I said, I want to build intelligent machines too,
但我有點失望 — 因為我還年輕,但幾年以後我又嘗試了一次
but I want to study how brains work first.
這次是在加州,我跑去柏克萊。
And they said, "Oh, you don't need to do that.
然後我說,我要從生物方面開始著手。
You're just going to program computers, that's all.
所以我被錄取了,進入了生物物理博士班。然後我心想,太棒了,
I said, you really ought to study brains.
我現在開始研究大腦了,然後我說,好的,我想要鑽研理論。
They said, "No, you're wrong."
但他們告訴我,喔,不,你不能研究關於腦的理論。
I said, "No, you're wrong," and I didn't get in.
你不想做那個的。沒有人會給你經費支持你做這種研究。
(Laughter)
身為一個研究生,你不能這麼做。所以我又說了,我的老天,
I was a little disappointed -- pretty young --
我非常沮喪。我說,但我能在這方面有所成就。
but I went back again a few years later,
所以我唯一能做的是,我回到了電腦業
this time in California, and I went to Berkeley.
然後說,好吧,我將留下來工作一段時間,做出一番成就。
And I said, I'll go in from the biological side.
然後我就開始設計出所有這些電子產品。
So I got in the PhD program in biophysics.
(笑聲)
I was like, I'm studying brains now. Well, I want to study theory.
我告訴自己,我在這邊待四年,賺些錢,
They said, "You can't study theory about brains.
我會成家,變得更成熟些,
You can't get funded for that.
同時也許神經科學領域也會發展得成熟一點。
And as a graduate student, you can't do that."
好吧,我花了超過四年的時間。時光飛逝,已經 16 年了。
So I said, oh my gosh.
但是我終於在研究大腦了,而我將會跟你們談談我的研究。
I was depressed; I said, but I can make a difference in this field.
為什麼我們應該要有一個好的大腦理論?
I went back in the computer industry
人們為了千百種不同的理由研究科學。
and said, I'll have to work here for a while.
其中一個理由 — 最基本的理由 — 是我們想要了解事物。
That's when I designed all those computer products.
人類是好奇的,我們只是想要獲取新知而已,你了解嗎?
(Laughter)
為什麼我們要研究螞蟻?不為什麼,只因為它很有趣。
I said, I want to do this for four years, make some money,
也許我們能從中學到新知,但是研究本身既有趣又吸引人。
I was having a family, and I would mature a bit,
但有時,科學有一些其他的屬性
and maybe the business of neuroscience would mature a bit.
而這些屬性會讓它額外的吸引人。
Well, it took longer than four years. It's been about 16 years.
有時候科學能夠讓我們更加認識自己,
But I'm doing it now, and I'm going to tell you about it.
它會讓我們知道我們是誰。
So why should we have a good brain theory?
雖然這極少發生,如你所知演化學說是一例,哥白尼也做到了,
Well, there's lots of reasons people do science.
它們徹底地改變了我們對自己身份地位上的認知。
The most basic one is, people like to know things.
但是最基本的,我們代表著我們的大腦。我的大腦正在和你的交談著。
We're curious, and we go out and get knowledge.
雖然我們的身體隨時陪伴著我們,但是是我的腦在和你的腦交談。
Why do we study ants? It's interesting.
所以如果我們想要了解我們到底是誰,我們是如何感覺、理解事物,
Maybe we'll learn something useful, but it's interesting and fascinating.
我們真的需要了解大腦是什麼。
But sometimes a science has other attributes
另一方面,有時科學
which makes it really interesting.
能對社會利益、科技、
Sometimes a science will tell something about ourselves;
商業,各式各樣領域做出極大的貢獻。這也是其中之一,
it'll tell us who we are.
因為當我們了解大腦是如何運作之後,我們將能夠
Evolution did this and Copernicus did this,
建造智慧機器,我相信整體來說,這會是件好事,
where we have a new understanding of who we are.
這將會對社會有極大助益
And after all, we are our brains. My brain is talking to your brain.
就如同基礎科技一般。
Our bodies are hanging along for the ride,
所以,為什麼我們沒有一個好的大腦理論?
but my brain is talking to your brain.
而且人們研究大腦的歷史已經有百來年了。
And if we want to understand who we are and how we feel and perceive,
那麼,讓我們先來看看普通科學領域的狀況。
we need to understand brains.
這是普通科學領域。
Another thing is sometimes science leads to big societal benefits, technologies,
普通科學領域中的理論與實作家呈現一個良好的平衡。
or businesses or whatever.
因此當理論學者說,嗯,我認為事情是這般這般,
This is one, too, because when we understand how brains work,
然後實驗科學家說,不,你錯了。
we'll be able to build intelligent machines.
然後就像這樣一直反覆來回,對吧?
That's a good thing on the whole,
這方法對物理適用。對地理適用。但這些是普通科學領域,
with tremendous benefits to society,
神經科學看起來是什麼樣子?這就是神經科學的狀況。
just like a fundamental technology.
我們的數據累積得比山還高,解剖學、生理學和行為學的數據。
So why don't we have a good theory of brains?
你無法想像我們對大腦的枝微末節了解得如何透徹。
People have been working on it for 100 years.
今年 (2003) 的神經科學研討會共有 28,000 人參加,
Let's first take a look at what normal science looks like.
每一個都在研究大腦。
This is normal science.
太多資訊。但沒有理論。在上層的這一塊是如此的微小,搖搖欲墜。
Normal science is a nice balance between theory and experimentalists.
而且理論在神經科學中尚未扮演任何重要的角色。
The theorist guy says, "I think this is what's going on,"
這真可恥。為什麼會這樣?
the experimentalist says, "You're wrong."
如果你問神經科學家,為什麼會是這種狀況?
It goes back and forth, this works in physics, this in geology.
一開始他們都會承認此事。但如果你接著問,他們會說,
But if this is normal science, what does neuroscience look like?
這個嘛,有很多的原因使我們沒有一個好的大腦理論。
This is what neuroscience looks like.
有些人會說,呃,我們還沒有足夠的數據,
We have this mountain of data,
我們還需要更多資訊,還有很多我們不知道的事。
which is anatomy, physiology and behavior.
我才剛剛告訴你們,我們有的數據多到你們的腦袋都裝不下。
You can't imagine how much detail we know about brains.
我們擁有如此多的資訊;我們不知道如何開始整理這些資訊。
There were 28,000 people who went to the neuroscience conference this year,
再有更多資訊又能怎樣?
and every one of them is doing research in brains.
也許我們會幸運的發現某些寶藏,但我不這麼認為。
A lot of data, but no theory.
這其實只是因為我們沒有理論這個事實所導致的症狀罷了。
There's a little wimpy box on top there.
我們不需要更多數據 — 我們需要一個好理論。
And theory has not played a role in any sort of grand way
有時候某些人會回答另一個說法,因為大腦是如此複雜,
in the neurosciences.
我們還需要 50 年的研究。
And it's a real shame.
我甚至好像聽到 Chris 昨天才說了類似的話。
Now, why has this come about?
我不確定你說了什麼,Chris,但好像是類似
If you ask neuroscientists why is this the state of affairs,
— 大腦是宇宙中最複雜的事物之一。這不是真的。
first, they'll admit it.
你比你的大腦還要複雜。腦只是你身體的一部分。
But if you ask them, they say,
並且,雖然大腦看起來非常複雜,
there's various reasons we don't have a good brain theory.
但是我們常認為我們所不了解的事物是複雜的。
Some say we still don't have enough data,
總是這樣子的。我們能夠說的只是,這個嘛,
we need more information, there's all these things we don't know.
我的新皮層,大腦中我感興趣的部份,有三百億個細胞。
Well, I just told you there's data coming out of your ears.
但你知道嗎?它非常、非常的規則。
We have so much information, we don't even know how to organize it.
事實上,它看起來像是同一個東西不斷的重複、重複再重複。
What good is more going to do?
它不像看起來般如此複雜。所以這不是問題。
Maybe we'll be lucky and discover some magic thing, but I don't think so.
某些人說,大腦無法了解大腦。
This is a symptom of the fact that we just don't have a theory.
非常具有禪意。呼,是吧 —
We don't need more data, we need a good theory.
(笑聲)
Another one is sometimes people say,
聽起來很有道理,但為什麼?我是說,真的有道理嗎?
"Brains are so complex, it'll take another 50 years."
大腦只不過是一堆細胞。你能了解你的肝臟呀。
I even think Chris said something like this yesterday, something like,
肝臟中也有很多細胞,對吧?
it's one of the most complicated things in the universe.
所以,你知道,我不覺得這有什麼問題。
That's not true -- you're more complicated than your brain.
最後,某些人會說,那麼,你知道,
You've got a brain.
我不覺得我是一堆細胞,你能理解嗎?我有意識。
And although the brain looks very complicated,
我能累積經驗,我生活在世界中,類似這些話。
things look complicated until you understand them.
我不可能只是一堆細胞。是的,你知道,
That's always been the case.
人們總是相信生物體內存在某種「生命力」,
So we can say, my neocortex, the part of the brain I'm interested in,
我們現在知道這一點都不是事實。
has 30 billion cells.
這一點都沒有事實根據,好吧,除了人們不想相信
But, you know what? It's very, very regular.
細胞可以做到人們平日在做的事情。
In fact, it looks like it's the same thing repeated over and over again.
因此,如果某些人們落入形而上學二元論的泥淖中,
It's not as complex as it looks. That's not the issue.
一些很聰明的人也不例外,但是我們可以駁斥他們的所有說法。
Some people say, brains can't understand brains.
(笑聲)
Very Zen-like. Woo.
不,我將要告訴你們還有別的,
(Laughter)
而且非常基本,就是我下面要說的這句話:
You know, it sounds good, but why? I mean, what's the point?
我們沒有一個好的大腦理論的另一個理由是,
It's just a bunch of cells. You understand your liver.
我們被一種直觀的、根深蒂固的
It's got a lot of cells in it too, right?
但是錯誤的假設所蒙蔽,因此一直無法找到問題的答案。
So, you know, I don't think there's anything to that.
我們所相信的某些事情,雖然表面上很顯而易見,但是它是錯的。
And finally, some people say,
事實上,科學界的歷史中已經發生過同樣的事情,而在我告訴你以前,
"I don't feel like a bunch of cells -- I'm conscious.
我要先跟你談談科學界的歷史。
I've got this experience, I'm in the world.
你們看看其他的科學革命,
I can't be just a bunch of cells."
這邊,我們來談談太陽系,那是哥白尼的貢獻,
Well, people used to believe there was a life force to be living,
達爾文的演化還有魏格納的板塊構造論。
and we now know that's really not true at all.
他們都與大腦科學有很多共通之處。
And there's really no evidence,
首先,他們有很多無法解釋的數據,一堆數據。
other than that people just disbelieve that cells can do what they do.
但是當他們有了理論之後,這些數據變得容易處理的多。
So some people have fallen into the pit of metaphysical dualism,
偉大的心靈總是會遭遇許多困難,那些極端、極端聰明的人們。
some really smart people, too, but we can reject all that.
我們現在並不比他們當時聰明。
(Laughter)
思考問題是極端困難的,
No, there's something else,
但一旦你想通了,事情就會得容易理解得多。
something really fundamental, and it is:
我女兒能夠了解這三個理論
another reason why we don't have a good brain theory
至少了解他們的基本架構,而那時她只是個幼稚園學童而已。
is because we have an intuitive, strongly held but incorrect assumption
因此,這並沒有這麼難,就像這樣,這是蘋果,這是柳丁,
that has prevented us from seeing the answer.
你知道的,地球在公轉,類似的這種東西。
There's something we believe that just, it's obvious, but it's wrong.
最後,另一件事是答案始終在那邊,
Now, there's a history of this in science and before I tell you what it is,
但是我們卻因為錯誤而明顯的假設而忽略了它,這就是問題所在。
I'll tell you about the history of it in science.
問題就是這個直觀且根深蒂固的認知是錯的。
Look at other scientific revolutions --
拿太陽系的例子來說,地球自轉的概念
the solar system, that's Copernicus,
還有地球表面以每小時幾千英哩的速度在轉動著,
Darwin's evolution, and tectonic plates, that's Wegener.
不用說還有地球本身以幾百萬英哩的時速在太陽系中移動著。
They all have a lot in common with brain science.
這真是瘋了。我們都知道地球並沒有在動。
First, they had a lot of unexplained data. A lot of it.
你覺得你有在以千哩的時速移動嗎?
But it got more manageable once they had a theory.
當然沒有。你知道,當有人說,
The best minds were stumped -- really smart people.
地球在太空中自轉,而太空是如此之大,
We're not smarter now than they were then;
然後他們會把你關起來,這就是當時他們所做的事。
it just turns out it's really hard to think of things,
(笑聲)
but once you've thought of them, it's easy to understand.
所以這是直觀且顯而易見的。現在,我們談談演化…
My daughters understood these three theories,
發生在演化上的情形是一樣的。我們教導孩子,嗯,聖經上說,
in their basic framework, in kindergarten.
你知道的,上帝創造了所有生命,貓是貓,狗是狗,
It's not that hard -- here's the apple, here's the orange,
人是人,樹木是樹木,他們是不變的。
the Earth goes around, that kind of stuff.
諾亞奉命將他們放到方舟內,如此這般。而且,你知道,
Another thing is the answer was there all along,
事實上,如果你相信演化,我們都來自同一個祖先,
but we kind of ignored it because of this obvious thing.
則我們和大廳裡那些植物有共同的祖先。
It was an intuitive, strongly held belief that was wrong.
這是演化告訴我們的。並且它是真的。儘管有點難令人相信。
In the case of the solar system,
板塊構造論也遭遇類似情形,不是嗎?
the idea that the Earth is spinning,
所有的山嶽與大陸都飄浮在地球的表面,
the surface is going a thousand miles an hour,
你相信嗎?這真的一點都不合邏輯。
and it's going through the solar system at a million miles an hour --
所以什麼是我說的關於大腦直觀但是不正確的假設,
this is lunacy; we all know the Earth isn't moving.
並使我們不能真正的了解大腦?
Do you feel like you're moving a thousand miles an hour?
現在我將要告訴你們,而且它將會看起來正確無誤不容懷疑,
If you said Earth was spinning around in space and was huge --
但這就是我想要說明的,不是嗎?然後我將會作一番論述
they would lock you up, that's what they did back then.
為什麼你們另一個假設也是錯的。
So it was intuitive and obvious. Now, what about evolution?
這個直觀且明顯的事情就是:智能可以藉由
Evolution, same thing.
行為來界定,
We taught our kids the Bible says God created all these species,
我們擁有智能乃是因為我們行事的方法
cats are cats; dogs are dogs; people are people; plants are plants;
還有我們展現智慧的行為,但是我要告訴你們這是錯的。
they don't change.
智能其實應該是由預測能力來界定的。
Noah put them on the ark in that order, blah, blah.
接下來的幾張投影片,我將解釋我的論點,
The fact is, if you believe in evolution, we all have a common ancestor.
給你們一個可以了解它的意義的例子。這裡有一個系統。
We all have a common ancestor with the plant in the lobby!
工程師喜歡這樣看待系統。科學家也喜歡這樣看待系統。
This is what evolution tells us. And it's true. It's kind of unbelievable.
他們說,嗯,這個箱子裡面有某種東西,然後我們有輸入跟輸出。
And the same thing about tectonic plates.
研究人工智慧的人說,我知道,箱子裡的東西是可編程的電腦
All the mountains and the continents
因為它和腦是對等的,我們將會給它一些輸入訊號
are kind of floating around on top of the Earth.
然後我們可以讓它做些事情,產生行為。
It doesn't make any sense.
然後艾倫•涂林訂定了涂林測驗,這個測驗基本上是說,
So what is the intuitive, but incorrect assumption,
如果某物的行為可以表現得跟人一模一樣,我們知道它有智能。
that's kept us from understanding brains?
對於智能本質上的一個行為標準,
I'll tell you. It'll seem obvious that it's correct. That's the point.
這個假設佔據了我們的想法很長的一段時間。
Then I'll make an argument why you're incorrect on the other assumption.
但是事實上,我稱之為真實智慧。
The intuitive but obvious thing is:
真實智慧是建築在其它東西上。
somehow, intelligence is defined by behavior;
我們藉由一序列的模式來體驗這個世界,我們儲存這些模式,
we're intelligent because of how we do things
我們也會回憶這些模式。當我們回憶時,我們會將現實與記憶中的
and how we behave intelligently.
模式對照,並且我們無時無刻不在預測下一刻。
And I'm going to tell you that's wrong.
這是永恆的標準。有一個關於我們的外在標準大概是這樣的,
Intelligence is defined by prediction.
我們了解這個世界嗎?我正在做預測嗎?等等這些。
I'm going to work you through this in a few slides,
你們現在都顯示出智慧,但是你們並沒有在做任何事。
and give you an example of what this means.
也許你剛剛正在搔癢,或者挖鼻孔,
Here's a system.
我不知道,但是你現在並沒有在做任何事,
Engineers and scientists like to look at systems like this.
但是你是有智慧的,你了解我在說什麼。
They say, we have a thing in a box. We have its inputs and outputs.
因為你有智慧而且你聽得懂英文,
The AI people said, the thing in the box is a programmable computer,
你知道這句話最後一個 — (沉默)
because it's equivalent to a brain.
字是什麼。
We'll feed it some inputs and get it to do something, have some behavior.
這個字會自己顯現,你無時無刻不在做類似這種的預測。
Alan Turing defined the Turing test, which essentially says,
所以,我要說的是,
we'll know if something's intelligent if it behaves identical to a human --
這個永恆的預測是我們大腦新皮層的訊號輸出。
a behavioral metric of what intelligence is
不知怎麼的,預測最終導致智能行為。
that has stuck in our minds for a long time.
這裡我來解釋它是如何發生的。讓我們先從非智能大腦開始看起。
Reality, though -- I call it real intelligence.
其實我不贊成稱之為非智能大腦,這種原始的大腦也是我們的一部分,
Real intelligence is built on something else.
所以下面我們稱之為非哺乳動物的腦,例如爬蟲類,
We experience the world through a sequence of patterns,
所以我說,就鱷魚吧,我們拿鱷魚來當例子。
and we store them, and we recall them.
鱷魚擁有一些非常複雜的感知能力。
When we recall them, we match them up against reality,
牠有非常好的視覺、聽覺、觸覺等等。
and we're making predictions all the time.
一張嘴一隻鼻子。牠擁有非常複雜的行為。
It's an internal metric; there's an internal metric about us,
牠可以奔跑、躲藏。牠擁有恐懼與情緒。牠能將你吃了,你知道吧。
saying, do we understand the world, am I making predictions, and so on.
牠可以攻擊。牠可以做各種事。
You're all being intelligent now, but you're not doing anything.
但是我們不認為鱷魚智力很高,跟人類一點都不能相比。
Maybe you're scratching yourself, but you're not doing anything.
但是牠已經擁有如此複雜的行為了。
But you're being intelligent; you're understanding what I'm saying.
在演化過程中,到底發生了什麼事?
Because you're intelligent and you speak English,
在哺乳類的演化過成中首先,
you know the word at the end of this
我們開始發展出所謂的新皮層。
sentence.
我將在這邊用此來表示新皮層,
The word came to you; you make these predictions all the time.
用這個建基於原始大腦上方的方塊來表示。
What I'm saying is,
新皮層就是一層新的組織。一層覆蓋在你大腦上方的新組織。
the internal prediction is the output in the neocortex,
如果你不知道,它就是你頭裡面最外層那個充滿皺摺的東西,
and somehow, prediction leads to intelligent behavior.
因為它不合身且被胡亂地塞在你的腦袋裡,所以它充滿了皺摺。
Here's how that happens:
(笑聲)
Let's start with a non-intelligent brain.
不,我說真的,真的是這樣。它大約跟張桌巾一般大小。
I'll argue a non-intelligent brain, we'll call it an old brain.
它並不合身,所以它充滿皺摺。看看在這邊我是怎麼畫它的。
And we'll say it's a non-mammal, like a reptile,
原始大腦仍然在那邊。你還擁有著與鱷魚相似的腦。
say, an alligator; we have an alligator.
是真的。那是你原始情緒的腦。
And the alligator has some very sophisticated senses.
就是那些東西,所有你會有的直覺反應。
It's got good eyes and ears and touch senses and so on,
而在那個上方。我們有一個稱為新皮層的記憶系統。
a mouth and a nose.
而這個記憶系統座落在大腦感知區的上方。
It has very complex behavior.
所以當感官訊號輸入進來並刺激了原始大腦,
It can run and hide. It has fears and emotions. It can eat you.
它開始往更上層的新皮層傳遞。而新皮層只是將之記憶下來。
It can attack. It can do all kinds of stuff.
它待在那邊說,呃,我將要把正在發生的事情全部記下來,
But we don't consider the alligator very intelligent,
我去了哪裡,我見了哪些人,我聽到了什麼東西,如此這般。
not in a human sort of way.
到了未來,當它再次見到類似的東西,
But it has all this complex behavior already.
處於類似或者同樣的環境下,
Now in evolution, what happened?
它就會重播。它會開始重播。
First thing that happened in evolution with mammals
喔,我到過這裡。當你上次在這裡的時候,
is we started to develop a thing called the neocortex.
接下來發生了這件事。它能讓你對未來產生預測。
I'm going to represent the neocortex by this box on top of the old brain.
它能讓你,就是它提供你腦部信號回饋,
Neocortex means "new layer." It's a new layer on top of your brain.
他們能讓你了解即將會發生的事,
It's the wrinkly thing on the top of your head
能讓你聽到一句話的最後一個「字」,即使我還沒說出口。
that got wrinkly because it got shoved in there and doesn't fit.
就是這種給原始大腦的回饋
(Laughter)
能夠讓你做出更多有智慧的決定。
Literally, it's about the size of a table napkin
這是我這次演講中最重要的一張投影片,因此我會再花點時間來解釋。
and doesn't fit, so it's wrinkly.
所以,每次當你說,喔,我能預測到這些事情。
Now, look at how I've drawn this.
就像如果你是一隻迷宮中的老鼠,然後你認識了這個迷宮,
The old brain is still there.
下一次當你在迷宮中的時候,你會做一樣的事情,
You still have that alligator brain. You do. It's your emotional brain.
但是突然間,你變聰明了
It's all those gut reactions you have.
因為你會說,喔,我認得這個迷宮,我知道該往哪邊走,
On top of it, we have this memory system called the neocortex.
我曾經到過這裡,我能夠預見未來。這就是智慧在做的事。
And the memory system is sitting over the sensory part of the brain.
在人身上,換句話說,這適用於所有哺乳動物,
So as the sensory input comes in and feeds from the old brain,
同樣適用於其他哺乳動物,但在人類身上,這個額外重要。
it also goes up into the neocortex.
在人身上,我們事實上發展出了新皮層的前段部份
And the neocortex is just memorizing.
稱為新皮層前緣。自然界在這邊耍了一個小手段。
It's sitting there saying, I'm going to memorize all the things going on:
它複製了後緣部份,後段的感知部份,
where I've been, people I've seen, things I've heard, and so on.
然後把它放來前面。
And in the future, when it sees something similar to that again,
因此人類很特殊的在腦前段也有此相同的構造,
in a similar environment, or the exact same environment,
但是我們使用它來控制運動功能。
it'll start playing it back: "Oh, I've been here before,"
所以現在我們能夠策劃非常複雜的運動計畫,和類似的事情。
and when you were here before, this happened next.
我沒有時間詳細解說所有的這些東西,但是如果你們想要了解大腦是如何運作的,
It allows you to predict the future.
你們必須了解上一段我所解釋的哺乳動物新皮層運作的原理,
It literally feeds back the signals into your brain;
它是如何的使我們具有儲存模式和進行預測的能力。
they'll let you see what's going to happen next,
現在讓我給你們一些關於預測的實例。
will let you hear the word "sentence" before I said it.
我已經說過那個關於「字」的例子了。在音樂中,
And it's this feeding back into the old brain
如果你曾經聽過一首歌,如果你之前聽過 Jill 唱這些歌,
that will allow you to make more intelligent decisions.
當她唱歌時,下一個音符就已經躍進你的耳朵了 —
This is the most important slide of my talk, so I'll dwell on it a little.
當你一邊在聽歌的時候,你一邊在預期著。如果是一張音樂專輯,
And all the time you say, "Oh, I can predict things,"
當一首歌結束,下一首歌會自動在你腦海中浮現。
so if you're a rat and you go through a maze, and you learn the maze,
而且這種事情一直不斷的在發生。你一直在做這些預測。
next time you're in one, you have the same behavior.
我聽過一個稱作「變更的門」的思想實驗。
But suddenly, you're smarter; you say, "I recognize this maze,
這個思想實驗指出,如果你在家裏有一個門,
I know which way to go; I've been here before; I can envision the future."
當你在這裡聽演講的時候,我去更動它,我找了一個人
That's what it's doing.
在這時候回到你家,任意對那扇門做變更,
This is true for all mammals --
他們將把你們的門把移動約兩寸的距離。
in humans, it got a lot worse.
然後當你今晚回到家的時候,你將會把你的手伸出,
Humans actually developed the front of the neocortex,
然後你將會碰到門把,就在這時,你會注意到
called the anterior part of the neocortex.
門把的位置不對了,然後你會驚覺,哇,有事情發生了。
And nature did a little trick.
你仍然需要一兩秒來思考到底發生了什麼事,但是一定有什麼不一樣。
It copied the posterior, the back part, which is sensory,
我可以任意更動你的門把。
and put it in the front.
我可以使它變大或變小,我可以由黃銅改成鍍銀,
Humans uniquely have the same mechanism on the front,
我可以將門把改為門桿。我可以改變你的門本身,為它上色,
but we use it for motor control.
或者加上窗戶。我有一千種以上的方法來變更你的門,
So we're now able to do very sophisticated motor planning, things like that.
然後在你開門的兩秒內,
I don't have time to explain, but to understand how a brain works,
你將會注意到某些變更的存在。
you have to understand how the first part of the mammalian neocortex works,
你沒辦法藉由工程學來完成這件事,人工智慧的解決途徑是,
how it is we store patterns and make predictions.
建立一個門的資料庫。它擁有所有這些與門相關的特性表。
Let me give you a few examples of predictions.
然後當你走到門前時,你知道,讓我們按照表來一個個檢查這些項目。
I already said the word "sentence."
門、門、門、你知道的、顏色,你知道我想說什麼嗎?
In music, if you've heard a song before,
我們不是這麼做的。你的大腦不是這樣運作的。
when you hear it, the next note pops into your head already --
你的大腦事實上是一直在做預測
you anticipate it.
預測在你的環境中將會發生什麼事。
With an album, at the end of a song, the next song pops into your head.
當我把我的手放上這張桌子,我會預期感覺到我的手停止。
It happens all the time, you make predictions.
當我走路時,每一步,即使只差了 1/8 英吋,
I have this thing called the "altered door" thought experiment.
我也會察覺某些事情不一樣了。
It says, you have a door at home;
你持續的在對周遭的環境做預測。
when you're here, I'm changing it --
我將簡短的談談視覺。這是一張女人的照片。
I've got a guy back at your house right now, moving the door around,
當你看著人時,你的眼睛大約會以
moving your doorknob over two inches.
每秒兩至三次的頻率移動。
When you go home tonight, you'll put your hand out, reach for the doorknob,
你不自覺,可是你的眼睛是不停的在移動著。
notice it's in the wrong spot
因此當你在看某人的臉時,
and go, "Whoa, something happened."
一般來說你會從一隻眼睛看到另一隻眼睛,再從眼睛到鼻子到嘴巴。
It may take a second, but something happened.
現在,當你的眼睛在對方眼睛間移動的時候,
I can change your doorknob in other ways --
如果一個鼻子出現在那邊,
make it larger, smaller, change its brass to silver, make it a lever,
你會在本來應該出現眼睛的地方看到鼻子,
I can change the door; put colors on, put windows in.
然後你會像,喔,天呀,你知道 —
I can change a thousand things about your door
(笑聲)
and in the two seconds you take to open it,
這個人不太對勁。
you'll notice something has changed.
而這是因為你一直在做預測。
Now, the engineering approach, the AI approach to this,
你不是只是往那邊看,然後說:我現在看到什麼東西?
is to build a door database with all the door attributes.
一個鼻子,那沒什麼。不,你會預期你將看到的東西。
And as you go up to the door, we check them off one at time:
(笑聲)
door, door, color ...
無時無刻。最後,讓我們來想想我們是如何做智力測驗的。
We don't do that. Your brain doesn't do that.
我們用預測能力來測驗它。下一個字是什麼,對吧?
Your brain is making constant predictions all the time
這個之於這個等於那個之於那個。這個序列的下一個數字是什麼?
about what will happen in your environment.
這是一個物體的三視圖。
As I put my hand on this table, I expect to feel it stop.
第四面可能是什麼?這就是我們測驗智力的方法。全部都跟預測能力有關。
When I walk, every step, if I missed it by an eighth of an inch,
那麼大腦理論的配方到底是什麼?
I'll know something has changed.
首先,我們必須要有正確的架構。
You're constantly making predictions about your environment.
而這個架構是記憶架構,
I'll talk about vision, briefly.
而不是計算或是行為架構。是一個記憶架構。
This is a picture of a woman.
你如何儲存並回憶這些序列與模式?一個時間與空間的模式。
When we look at people, our eyes saccade over two to three times a second.
然後,如果在那個架構中,你有一群好的理論學者。
We're not aware of it, but our eyes are always moving.
現在的生物學家通常不是好的理論學者。
When we look at a face, we typically go from eye to eye to nose to mouth.
並不是總是這樣,但是通常是,生物學沒有建夠好理論的歷史習慣。
When your eye moves from eye to eye,
我能找到最好的工作夥伴是物理學家,
if there was something else there like a nose,
工程師和數學家,他們習於演算思維模式。
you'd see a nose where an eye is supposed to be and go, "Oh, shit!"
然後他們必須學習解剖學和生理學。
(Laughter)
你必須使這些理論在解剖層面上也是非常真實的。
"There's something wrong about this person."
任何人當他跳出來告訴你他們關於大腦運行的理論
That's because you're making a prediction.
但是不能解釋這些事情如何在腦內發生
It's not like you just look over and say, "What am I seeing? A nose? OK."
還有腦內的連結關係是什麼,這就不是一個理論。
No, you have an expectation of what you're going to see.
這就是我們在紅木神經科學研究所進行的研究。
Every single moment.
我希望我能有更多時間來告訴你們,我們已經在這方面有了驚人的進步,
And finally, let's think about how we test intelligence.
而我預期未來還能再回到這裡演講,
We test it by prediction: What is the next word in this ...?
因此也許在不久的將來我將能有機會再次跟你們談談。
This is to this as this is to this. What is the next number in this sentence?
我真的非常、非常興奮。這絕對不需要再五十年。
Here's three visions of an object. What's the fourth one?
因此大腦理論究竟看起來會是什麼樣子?
That's how we test it. It's all about prediction.
首先,它會是一個關於記憶的理論。
So what is the recipe for brain theory?
跟電腦記憶體不一樣。它一點都不會像是電腦記憶體。
First of all, we have to have the right framework.
會非常、非常的不同。它會是這些非常高維模式
And the framework is a memory framework,
的記憶,就跟你從眼睛看到的東西一般。
not a computational or behavior framework,
它會是序列的記憶。
it's a memory framework.
你不能學習或是回憶序列外的任何事物。
How do you store and recall these sequences of patterns?
一首歌必須按照時間的順序來聽,
It's spatiotemporal patterns.
你也必須按照時間順序來播放。
Then, if in that framework, you take a bunch of theoreticians --
然後這些順序就會自動被相關連在一起重播,因此如果我看到某些東西,
biologists generally are not good theoreticians.
聽到某些東西,它讓我回一起相關的事物,然後就會自動重播。
Not always, but generally, there's not a good history of theory in biology.
它是自動重播。然後對於未來所將輸入訊息的預測是我們所希望的輸出。
I've found the best people to work with are physicists,
像我提過的,這個理論必須是生物學正確的。
engineers and mathematicians,
它必須能被測試,然且你必須能夠建造它。
who tend to think algorithmically.
如果你不能建造它,你就是不了解它。因此,最後一張投影片。
Then they have to learn the anatomy and the physiology.
這最終會產生什麼結果?我們能夠真的建造出智能機器嗎?
You have to make these theories very realistic in anatomical terms.
絕對可以。而且它會和一般人們所想的不同。
Anyone who tells you their theory about how the brain works
我認為這無疑的會發生。
and doesn't tell you exactly how it's working
首先,它會被建造,我們將會用矽建出這個東西。
and how the wiring works --
跟我們用來建造以矽為原料的電腦記憶體同樣的技術,
it's not a theory.
我們在這邊也同樣可以使用。
And that's what we do at the Redwood Neuroscience Institute.
但是它們會是非常不同種類的記憶體。
I'd love to tell you we're making fantastic progress in this thing,
然後我們將會將這些記憶體連結上感應器,
and I expect to be back on this stage sometime in the not too distant future,
這些感應器將會經歷真實世界的即時數據,
to tell you about it.
然後這些東西將會認識它們的環境。
I'm really excited; this is not going to take 50 years.
而且你將會看到的第一批成品應該非常不可能會長得像個機器人。
What will brain theory look like?
不是因為機器人沒有用而且人們可以建造機器人。
First of all, it's going to be about memory.
但是機器人的部份是最難的部份。那是原始的大腦。非常的難。
Not like computer memory -- not at all like computer memory.
這個新的腦袋要比原始腦袋簡單一些。
It's very different.
所以我們將建造的第一個東西將會是不需要太多機器人特徵的東西。
It's a memory of very high-dimensional patterns,
所以你將不會看到 C-3PO。
like the things that come from your eyes.
你可能會比較常看到類似,例如,智慧車
It's also memory of sequences:
真的能了解交通狀況和駕駛
you cannot learn or recall anything outside of a sequence.
而且能夠解讀某些方向燈在閃的車輛過半分鐘後
A song must be heard in sequence over time,
也許即將轉彎,如此這般的事情。
and you must play it back in sequence over time.
(笑聲)
And these sequences are auto-associatively recalled,
我們也可以設計智慧型保全系統。
so if I see something, I hear something, it reminds me of it,
任何我們需要動用到腦力,但是不會執行太多機械動作的場合。
and it plays back automatically.
這些將會是首先發生的情況。
It's an automatic playback.
但是最終,沒什麼是不可能的。
And prediction of future inputs is the desired output.
我不知道這將會發展的如何。
And as I said, the theory must be biologically accurate,
我知道許多發明微處理器的人
it must be testable and you must be able to build it.
如果你問他們,他們知道他們是在從事一些非常重要的事情,
If you don't build it, you don't understand it.
但是他們不知道將會發生什麼事。
One more slide.
他們不能預測到手機、網路等等這些事情的發生。
What is this going to result in?
他們只知道像,嘿,他們將要建造計算機
Are we going to really build intelligent machines?
和交通號誌燈。但是這將會很重要。
Absolutely. And it's going to be different than people think.
同樣的道理,大腦理論和這些記憶體
No doubt that it's going to happen, in my mind.
將會是非常基礎的科技,而且會
First of all, we're going to build this stuff out of silicon.
在未來的一百年內帶來非常不可思議的改變。
The same techniques we use to build silicon computer memories,
我最興奮的是我們將會如何將它們應用到科學研究上。
we can use here.
我想我的時間已經到了,我超時了,所以我將要結束這次演講
But they're very different types of memories.
就在這裡結束。
And we'll attach these memories to sensors,
and the sensors will experience real-live, real-world data,
and learn about their environment.
Now, it's very unlikely the first things you'll see are like robots.
Not that robots aren't useful; people can build robots.
But the robotics part is the hardest part. That's old brain. That's really hard.
The new brain is easier than the old brain.
So first we'll do things that don't require a lot of robotics.
So you're not going to see C-3PO.
You're going to see things more like intelligent cars
that really understand what traffic is, what driving is
and have learned that cars with the blinkers on for half a minute
probably aren't going to turn.
(Laughter)
We can also do intelligent security systems.
Anytime we're basically using our brain but not doing a lot of mechanics --
those are the things that will happen first.
But ultimately, the world's the limit.
I don't know how this will turn out.
I know a lot of people who invented the microprocessor.
And if you talk to them,
they knew what they were doing was really significant,
but they didn't really know what was going to happen.
They couldn't anticipate cell phones and the Internet
and all this kind of stuff.
They just knew like, "We're going to build calculators
and traffic-light controllers.
But it's going to be big!"
In the same way, brain science and these memories
are going to be a very fundamental technology,
and it will lead to unbelievable changes in the next 100 years.
And I'm most excited about how we're going to use them in science.
So I think that's all my time -- I'm over,
and I'm going to end my talk right there.