Your words may predict your future mental health | Mariano Sigman

翻译人员: Jack Zhang 校对人员: 易帆 余

历史纪录可以让我们知道 古希腊人如何打扮、

We have historical records that allow us to know how the ancient Greeks dressed,

如何生活、

how they lived,

how they fought ...

如何打仗...

但他们如何思考呢？

but how did they think?

One natural idea is that the deepest aspects of human thought --

有一个很自然的方法就是， 去探索人类最深层的想法——

our ability to imagine,

我们的想像力、

to be conscious,

意识力、

to dream --

去梦想——

have always been the same.

是否是一样的。

Another possibility

另一种可能是，

is that the social transformations that have shaped our culture

去探索造就我们文化的社会变革，

这些变革也许就是 改变人类想法的主要因素。

may have also changed the structural columns of human thought.

We may all have different opinions about this.

对这一点，大家或许有不同的看法。

Actually, it's a long-standing philosophical debate.

实际上，这是一个存在已久的哲学辩论。

But is this question even amenable to science?

究竟这个问题是否可以 通过科学来处理？

Here I'd like to propose

我的建议是

that in the same way we can reconstruct how the ancient Greek cities looked

如同仅借由一些砖头， 我们得以重建希腊古都的外貌，

just based on a few bricks,

也可用同样的方式，

that the writings of a culture are the archaeological records,

借由一些文化作品， 比如考古纪录、

the fossils, of human thought.

化石，来了解人类的想法。

And in fact,

而实际上，

因为对人类的

doing some form of psychological analysis

of some of the most ancient books of human culture,

古老文化书籍做了一些心理分析，

Julian Jaynes came up in the '70s with a very wild and radical hypothesis:

朱利安 杰尼斯在70年代， 发表了一个相当大胆激进的假说：

that only 3,000 years ago,

他说，3000年前的人类，

humans were what today we would call schizophrenics.

是我们现在俗称的 “精神分裂症患者”。

And he made this claim

他会如此主张的原因是

based on the fact that the first humans described in these books

依据世界各地不同的传统及位置，

behaved consistently,

这些书籍里面

in different traditions and in different places of the world,

所描述的人类行为

as if they were hearing and obeying voices

似乎不约而同地都会服从

他们认为是从神袛

that they perceived as coming from the Gods,

那边传来的声音......

or from the muses ...

而如今，我们会称之为“幻听”。

what today we would call hallucinations.

And only then, as time went on,

随着时间的洗礼，

they began to recognize that they were the creators,

他们开始认知到 那些声音是他们自己创造的，

他们就是那些内在声音的主人。

the owners of these inner voices.

And with this, they gained introspection:

有了这样的认知， 他们学会了 “自省”：

一种反思自己想法的能力。

the ability to think about their own thoughts.

So Jaynes's theory is that consciousness,

所以杰尼斯对“意识”的理论就是，

至少现今我们觉察到的“意识”、

at least in the way we perceive it today,

where we feel that we are the pilots of our own existence --

感觉到我们能掌控 自我人生的感悟——

is a quite recent cultural development.

是相当近代的文化发展。

And this theory is quite spectacular,

这理论很有前瞻性，

but it has an obvious problem

但一个很明显的问题就是，

which is that it's built on just a few and very specific examples.

它是建立在极少又特殊的案例上。

所以问题是，

So the question is whether the theory

that introspection built up in human history only about 3,000 years ago

3000年来人类才建立起 自省能力的这个理论

can be examined in a quantitative and objective manner.

是否可以经得起量化且客观的考验。

And the problem of how to go about this is quite obvious.

至于要如何做的问题， 也是相当简单明了。

但我的意思并非，比如， 柏拉图有一天突然醒来写下

It's not like Plato woke up one day and then he wrote,

"Hello, I'm Plato,

“你好！我是柏拉图，

and as of today, I have a fully introspective consciousness."

我今天拥有完整的自省意识了。” 那样简单而已。

（笑声）

(Laughter)

And this tells us actually what is the essence of the problem.

而这鞥告诉我们，我们要找出 问题的本质是什么。

We need to find the emergence of a concept that's never said.

我们必须找到从来没有被 谈论过的概念。

The word introspection does not appear a single time

“自省”这个词，在我们研究的

in the books we want to analyze.

这些书本中从未出现过一次。

So our way to solve this is to build the space of words.

所以为了解决这个问题， 我们要建立一个字词的空间。

This is a huge space that contains all words

在这个大空间里， 包含了所有的词汇，

in such a way that the distance between any two of them

用这种方式可以衡量

is indicative of how closely related they are.

两个词语彼此之间的关联程度。

So for instance,

举个例子，

you want the words "dog" and "cat" to be very close together,

你会想，“狗”、“猫”是比较相关的词，

but the words "grapefruit" and "logarithm" to be very far away.

但“葡萄柚”和“对数” 就没什么关联了。

And this has to be true for any two words within the space.

而在这个空间里的任何 两个词都必须能以此衡量。

And there are different ways that we can construct the space of words.

而我们有很多方式 可以建立起这些字的空间架构，

One is just asking the experts,

方法一，只要请教专家就行了，

a bit like we do with dictionaries.

有点类似查字典。

Another possibility

另一个可行的方法是，

is following the simple assumption that when two words are related,

当两个字词出现关联性时， 去追踪它们的预设状况，

它们可能会出现在同一句、

they tend to appear in the same sentences,

in the same paragraphs,

同一段落、

in the same documents,

或同一文档中，

more often than would be expected just by pure chance.

比偶然出现频繁得多。

And this simple hypothesis,

在这个简单的前提下，

this simple method,

这个单纯且带有

with some computational tricks

运算技巧的方法

that have to do with the fact

在这个复杂且高维度的

空间中必须能充分发挥作用，

that this is a very complex and high-dimensional space,

turns out to be quite effective.

而事后证明，它相当有效。

向各位介绍一下，它多有效，

And just to give you a flavor of how well this works,

this is the result we get when we analyze this for some familiar words.

我们分析了一些经常用到的词语。

And you can see first

首先你可以看到，

that words automatically organize into semantic neighborhoods.

这些词语会自动地划分为 语义相近的相邻群组，

所以你可看到水果，身体部位，

So you get the fruits, the body parts,

电脑零件与科学术语等等。

the computer parts, the scientific terms and so on.

演算法也可以把我们要 整理的概念分门别类出来。

The algorithm also identifies that we organize concepts in a hierarchy.

So for instance,

举个例子，

你可以看到，科学的术语 被拆解成两个子类，

you can see that the scientific terms break down into two subcategories

of the astronomic and the physics terms.

分别是太空与物理的术语。

And then there are very fine things.

然后你会发现一件有趣的事。

For instance, the word astronomy,

举个例子，“天文学”这个词，

which seems a bit bizarre where it is,

它现在的位置看似不太对，

却的确在正确的位置上，

is actually exactly where it should be,

between what it is,

它应该介于科学与

天文学术语之间，

an actual science,

and between what it describes,

因为天文学是一门科学

the astronomical terms.

同时又包含了很多天文学术语。

我们可以持续寻找其它类似的情况。

And we could go on and on with this.

如果你盯着这些词一阵子，

Actually, if you stare at this for a while,

然后随机搭配连接一下这些词语，

and you just build random trajectories,

你会觉得好像自己在做诗。

you will see that it actually feels a bit like doing poetry.

那是因为在某种程度上，

And this is because, in a way,

walking in this space is like walking in the mind.

在这个空间里漫遊， 就像是在脑海中做诗一样。

最后，

And the last thing

is that this algorithm also identifies what are our intuitions,

演算法也能辨识出人类的直觉，

of which words should lead in the neighborhood of introspection.

并归纳到自省的词语范畴中。

So for instance,

举个例子，

words such as "self," "guilt," "reason," "emotion,"

比如“自我”、“內疚”、“理由”、“情绪”

are very close to "introspection,"

与“自省”的含义非常接近，

but other words,

但其它的词汇，

such as "red," "football," "candle," "banana,"

比如“红色”、“足球”、“蜡烛”、“香蕉”

就差很远了。

are just very far away.

所以一旦我们建立起 这样的词汇空间，

And so once we've built the space,

the question of the history of introspection,

有关于自省的历史，

or of the history of any concept

有关与任何概念的历史，

以前被认为是抽象 或是有点模糊的词汇，

which before could seem abstract and somehow vague,

becomes concrete --

都可以变成实实在在的

becomes amenable to quantitative science.

可以被量化的科学。

All that we have to do is take the books,

而我们要做的就是， 拿起这些书，

把它们数字化，

we digitize them,

and we take this stream of words as a trajectory

然后把这些词汇映射到

and project them into the space,

词汇空间里面，

and then we ask whether this trajectory spends significant time

然后我们问电脑， 这些词汇所经过的轨迹

花了多少时间才接近自省的概念。

circling closely to the concept of introspection.

And with this,

有了这些数据，

we could analyze the history of introspection

我们就可以分析古希腊传统中，

有关于自省的历史，

in the ancient Greek tradition,

因为我们拥有最完整的文字记录。

for which we have the best available written record.

So what we did is we took all the books --

所以我们先把这些书——

we just ordered them by time --

按照时间排列——

for each book we take the words

然后把每本书中的词汇都

and we project them to the space,

投射到词语空间里面，

and then we ask for each word how close it is to introspection,

然后我们问电脑，这些字词 与自省有多少的相关性，

再把它们平均起来。

and we just average that.

And then we ask whether, as time goes on and on,

然后，我们不断地问电脑问题，

these books get closer, and closer and closer

这些书就会越来越

接近自省的概念。

to the concept of introspection.

And this is exactly what happens in the ancient Greek tradition.

而这正是当时在古希腊所发生的事。

So you can see that for the oldest books in the Homeric tradition,

各位可以看到在 荷马时代最古老的书籍，

there is a small increase with books getting closer to introspection.

与自省的相关性只有一点点。

But about four centuries before Christ,

但在大约在公元前400年左右，

this starts ramping up very rapidly to an almost five-fold increase

这个数据却快速上涨至五倍，

of books getting closer, and closer and closer

这些书与自省的概念

to the concept of introspection.

越来越接近。

最棒的是，

And one of the nice things about this

is that now we can ask

我们可以问电脑，

whether this is also true in a different, independent tradition.

在不同的、独立的传统文化中， 是否也有一样的现象。

So we just ran this same analysis on the Judeo-Christian tradition,

所以，我们用同样的方法， 分析了传统犹太基督教的书籍，

也得到了类似的趋势。

and we got virtually the same pattern.

Again, you see a small increase for the oldest books in the Old Testament,

在最古老的旧约圣经中， 你可以看到它缓慢地增加，

之后在新约圣经中，

and then it increases much more rapidly

它在快速地增长。

in the new books of the New Testament.

大约公元400年，

And then we get the peak of introspection

圣人奥古斯丁的《忏悔录》中

in "The Confessions of Saint Augustine,"

about four centuries after Christ.

自省的词汇数量达到了最高峰。

And this was very important,

这个信息相当重要，

because Saint Augustine had been recognized by scholars,

因为圣人奥古斯丁已经被多位学者、

philologists, historians,

心理学家、历史学家公认为

as one of the founders of introspection.

是自省的创始人之一。

有些人认为他是现代心理学之父。

Actually, some believe him to be the father of modern psychology.

所以，我们演算法的优点

So our algorithm,

which has the virtue of being quantitative,

不仅可以量化，

of being objective,

而且客观，

当然速度也相当快——

and of course of being extremely fast --

几秒就可以跑完——

it just runs in a fraction of a second --

can capture some of the most important conclusions

并捕捉到使用传统方法 必须费长时间调查

才能抓到的一些重点。

of this long tradition of investigation.

And this is in a way one of the beauties of science,

这也是科学美好的地方之一，

which is that now this idea can be translated

它可以解读、归纳这想法，

and generalized to a whole lot of different domains.

然后广泛应用在许多不同的领域上。

So in the same way that we asked about the past of human consciousness,

或许最具挑战性的问题是，

maybe the most challenging question we can pose to ourselves

我们用电脑来分析过去的 自我意识发展的方法，

is whether this can tell us something about the future of our own consciousness.

是不是也可以告诉我们 自我意识的发展趋势呢？

To put it more precisely,

更确切地说，

我们现在说的话，

whether the words we say today

can tell us something of where our minds will be in a few days,

是否可以告诉我们接下来的几天、

in a few months

几个月或几年后，

or a few years from now.

我们的心智会达到什么情况。

And in the same way many of us are now wearing sensors

类似的，我们现在很多人 都使用穿戴式侦测器，

that detect our heart rate,

可以侦测我们的心跳、

our respiration,

呼吸、

our genes,

基因，

on the hopes that this may help us prevent diseases,

让我们可以预防疾病，

我们是否可以通过 监控和分析我们所说的话、

we can ask whether monitoring and analyzing the words we speak,

we tweet, we email, we write,

发的微博、邮件和书写的文字，

can tell us ahead of time whether something may go wrong with our minds.

来提前告诉我们，我们的心智 可能要发生问题了？

我跟我的兄弟，

And with Guillermo Cecchi,

who has been my brother in this adventure,

吉列尔莫 切基，

we took on this task.

扛起了这项任务。

And we did so by analyzing the recorded speech of 34 young people

我们纪录分析了 34 位年轻人的谈话。

who were at a high risk of developing schizophrenia.

他们曾是患精神分裂症的高风险人群。

And so what we did is, we measured speech at day one,

我们测量了他们第一天的谈话，

and then we asked whether the properties of the speech could predict,

然后问电脑，从他们的话中， 是否可以预测出，

within a window of almost three years,

未來三年內，

他们会不会患上精神错乱。

the future development of psychosis.

But despite our hopes,

但我们大失所望，

we got failure after failure.

一次又一次的失败。

There was just not enough information in semantics

没有足够的语义上的信息

to predict the future organization of the mind.

来预测未来的心智发展。

It was good enough

它有能力分辨

to distinguish between a group of schizophrenics and a control group,

精神病患者和健康人，

a bit like we had done for the ancient texts,

因为这有点像我们之前 做古文字的分析，

but not to predict the future onset of psychosis.

但没办法预测未来精神错乱的发病。

后来我们了解到，

But then we realized

that maybe the most important thing was not so much what they were saying,

也许最关键的不是他们说了什么，

而是他们怎么说。

but how they were saying it.

More specifically,

进一步说，

it was not in which semantic neighborhoods the words were,

不是他们说的话落在哪个 语义相近的群组里，

but how far and fast they jumped

而是他们说话的方式是否会在这几个

from one semantic neighborhood to the other one.

语义相近的群组里快速地跳来跳去。

And so we came up with this measure,

所以我们想出了一个

叫做“语义连贯性”的评估方法，

which we termed semantic coherence,

which essentially measures the persistence of speech within one semantic topic,

本质上就是评估谈话的持续性

within one semantic category.

是否会落在同一个 语义主题或类别上。

And it turned out to be that for this group of 34 people,

结果显示，刚刚的 34 位年轻人，

the algorithm based on semantic coherence could predict,

通过这个语义连贯性演算法，

预测谁会精神错乱的正确率

with 100 percent accuracy,

who developed psychosis and who will not.

达到了百分之百。

And this was something that could not be achieved --

目前临床上所有评估方式

not even close --

都无法达到、

with all the other existing clinical measures.

甚至无法接近这个数字。

And I remember vividly, while I was working on this,

在我做这项研究的时候， 清楚地记得一件事，

当时我坐在电脑前面，

I was sitting at my computer

and I saw a bunch of tweets by Polo --

看到保罗发的一些微博——

他是我之前在布宜诺斯艾利斯市 教书时的第一个学生，

Polo had been my first student back in Buenos Aires,

and at the time he was living in New York.

当时他住在纽约。

And there was something in this tweets --

我发现微博的内容不太对劲——

I could not tell exactly what because nothing was said explicitly --

我看不懂是什么， 因为他写得不太清楚——

但我有一种

but I got this strong hunch,

强烈的直觉，一定 有什么地方不对劲儿了。

this strong intuition, that something was going wrong.

So I picked up the phone, and I called Polo,

所以我立刻打电话给保罗，

没错，他当时感觉不太舒服。

and in fact he was not feeling well.

And this simple fact,

仅仅通过阅读

他微博的字里行间，

that reading in between the lines,

I could sense, through words, his feelings,

我就可以感受到他的精神健康状态，

阅读别人的用词 的确是个简单有效的帮助方式。

was a simple, but very effective way to help.

What I tell you today

今天我要告诉各位的是，

is that we're getting close to understanding

我们已经越来越能够理解

how we can convert this intuition that we all have,

如何把我们共有的，

that we all share,

共享的直觉

into an algorithm.

转换成演算法。

通过这样做，

And in doing so,

we may be seeing in the future a very different form of mental health,

未来我们也许可以看到一种 全然不同的精神健康模式，

based on objective, quantitative and automated analysis

是基于一种客观、 量化的方式来自动分析出

of the words we write,

我们所写的词汇，

of the words we say.

还有我们所说的话。

谢谢。

Gracias.

(Applause)

（掌声）