你的话可以预测你未来的心理健康

Your words may predict your future mental health

演讲者：Mariano Sigman

 

We have historical records that allow us to know how the ancient Greeks dressed, how they lived, how they fought ... but how did they think?

我们有历史记录，可以让我们 要知道古希腊人的穿着，他们如何生活，他们如何战斗...... 但是他们是怎么想的呢？

 

One natural idea is that the deepest aspects of human thought -- our ability to imagine, to be conscious, to dream -- have always been the same. Another possibility is that the social transformations that have shaped our culture may have also changed the structural columns of human thought.

一个自然的想法是，最深的 人类思想的各个方面——我们想象、有意识、做梦的能力——总是相同的。 另一种可能性是社会转型 塑造了我们的文化可能也发生了变化 人类思想的结构柱。

 

We may all have different opinions about this. Actually, it's a long-standing philosophical debate. But is this question even amenable to science?

我们可能都有不同的 对此的看法。 实际上，这是一个由来已久的 哲学辩论。 但这个问题是吗 甚至适合科学？

 

Here I'd like to propose that in the same way we can reconstruct how the ancient Greek cities looked just based on a few bricks, that the writings of a culture are the archaeological records, the fossils, of human thought.

在这里，我想提出，以同样的方式我们可以重建 古希腊城市的外观仅基于几块砖，即一种文化的著作 是人类思想的考古记录，化石。

 

footnote

footnote

And in fact, doing some form of psychological analysis of some of the most ancient books of human culture, Julian Jaynes came up in the '70s with a very wild and radical hypothesis: that only 3,000 years ago, humans were what today we would call schizophrenics. And he made this claim based on the fact that the first humans described in these books behaved consistently, in different traditions and in different places of the world, as if they were hearing and obeying voices that they perceived as coming from the Gods, or from the muses ... what today we would call hallucinations. And only then, as time went on, they began to recognize that they were the creators, the owners of these inner voices. And with this, they gained introspection: the ability to think about their own thoughts.

事实上，对一些最古老的人进行某种形式的心理分析 人类文化书籍，朱利安·杰恩斯（Julian Jaynes）在70年代问世 有一个非常疯狂和激进的假设：仅在 3，000 年前，人类就是今天的样子 我们称之为精神分裂症患者。 他提出这一主张是基于这样一个事实，即第一个 这些书中描述的人类在不同的传统中表现一致 在世界不同的地方，仿佛他们听到并服从他们所感知的声音 来自众神，或来自缪斯... 今天我们称之为幻觉。 直到那时，随着时间的推移，他们才开始认识到 他们是创造者，是这些内心声音的所有者。 有了这个，他们获得了内省：思考的能力 关于他们自己的想法。

 

So Jaynes's theory is that consciousness, at least in the way we perceive it today, where we feel that we are the pilots of our own existence -- is a quite recent cultural development. And this theory is quite spectacular, but it has an obvious problem which is that it's built on just a few and very specific examples. So the question is whether the theory that introspection built up in human history only about 3,000 years ago can be examined in a quantitative and objective manner.

所以杰恩斯的理论是意识，至少在我们今天感知它的方式中，我们觉得我们是飞行员。 我们自己的存在 - 是一个相当新的文化发展。 这个理论非常壮观，但它有一个明显的问题，那就是它只建立在几个基础上。 和非常具体的例子。 所以问题是，内省是否在人类身上建立起来的理论。 只有大约3年前的历史可以用定量来检验 和客观的态度。

 

And the problem of how to go about this is quite obvious. It's not like Plato woke up one day and then he wrote, "Hello, I'm Plato, and as of today, I have a fully introspective consciousness."

以及如何的问题 这样做是很明显的。 不像柏拉图有一天醒来 然后他写道：“你好，我是柏拉图，截至今天，我有 一个完全内省的意识。

 

And this tells us actually what is the essence of the problem. We need to find the emergence of a concept that's never said. The word introspection does not appear a single time in the books we want to analyze.

这实际上告诉我们 问题的本质是什么。 我们需要找到出现 一个从未说过的概念。 内省这个词 在我们想要分析的书中没有出现过一次。

 

So our way to solve this is to build the space of words. This is a huge space that contains all words in such a way that the distance between any two of them is indicative of how closely related they are. So for instance, you want the words "dog" and "cat" to be very close together, but the words "grapefruit" and "logarithm" to be very far away. And this has to be true for any two words within the space.

所以我们解决这个问题的方法 就是构建文字的空间。 这是一个巨大的空间 以这样的方式包含所有单词 在它们中的任何两个之间都表明如何 它们密切相关。 例如，你想要“狗”和“猫”这两个词 靠得很近，但“葡萄柚”和“对数”这两个词 离得很远。 这必须是真的 对于空间内的任意两个单词。

 

footnote

footnote

And there are different ways that we can construct the space of words. One is just asking the experts, a bit like we do with dictionaries. Another possibility is following the simple assumption that when two words are related, they tend to appear in the same sentences, in the same paragraphs, in the same documents, more often than would be expected just by pure chance. And this simple hypothesis, this simple method, with some computational tricks that have to do with the fact that this is a very complex and high-dimensional space, turns out to be quite effective.

并且有不同的方式 我们可以构建单词的空间。 一个是询问专家，有点像我们对字典所做的。 另一种可能性是遵循简单的假设 当两个单词相关时，它们往往出现在相同的句子、相同的段落、相同的文档中，比预期的更频繁 纯属偶然。 这个简单的假设，这个简单的方法，有一些计算技巧，与这是一个非常复杂的事实有关。 和高维空间，结果证明是相当有效的。

 

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. And you can see first that words automatically organize into semantic neighborhoods. So you get the fruits, the body parts, the computer parts, the scientific terms and so on.

只是为了给你一个味道 这有多好，这是我们在 我们分析一些熟悉的词。 你首先可以看到单词会自动组织 进入语义邻域。 所以你得到水果，身体部位，电脑部位， 科学术语等等。

 

footnote

footnote

The algorithm also identifies that we organize concepts in a hierarchy. So for instance, you can see that the scientific terms break down into two subcategories of the astronomic and the physics terms. And then there are very fine things. For instance, the word astronomy, which seems a bit bizarre where it is, is actually exactly where it should be, between what it is, an actual science, and between what it describes, the astronomical terms.

该算法还可以识别 我们在层次结构中组织概念。 例如，您可以看到科学术语 分为天文学和物理学术语的两个子类别。 然后有非常精细的东西。 例如，天文学这个词，它的位置似乎有点奇怪，实际上正是它应该在的地方，介于它是什么，一门真正的科学，以及它所描述的天文学术语之间。

 

And we could go on and on with this. Actually, if you stare at this for a while, and you just build random trajectories, you will see that it actually feels a bit like doing poetry. And this is because, in a way, walking in this space is like walking in the mind.

我们可以继续这样做。 其实，如果你盯着看 在一段时间内，你只是建立随机轨迹，你会发现它实际上感觉 有点像做诗。 这是因为，在某种程度上，在这个空间里行走。 就像在脑海中行走一样。

 

And the last thing is that this algorithm also identifies what are our intuitions, of which words should lead in the neighborhood of introspection. So for instance, words such as "self," "guilt," "reason," "emotion," are very close to "introspection," but other words, such as "red," "football," "candle," "banana," are just very far away.

最后一件事是，该算法还可以识别 我们的直觉是什么，哪些词应该引导 在内省的附近。 例如，诸如“自我”，“内疚”之类的词， “理性”、“情感”非常接近“内省”，但其他词，如“红色”、“足球”， “蜡烛”、“香蕉”就在很远的地方。

 

And so once we've built the space, the question of the history of introspection, or of the history of any concept which before could seem abstract and somehow vague, becomes concrete -- becomes amenable to quantitative science.

因此，一旦我们建立了空间，历史问题就出现了。 内省，或任何以前可能看起来很抽象的概念的历史 不知何故模糊，变得具体——变得适合定量科学。

 

 

All that we have to do is take the books, we digitize them, and we take this stream of words as a trajectory and project them into the space, and then we ask whether this trajectory spends significant time circling closely to the concept of introspection.

我们所要做的就是拿起书，我们把它们数字化，然后我们拿这个流。 单词作为轨迹并将它们投射到空间中，然后我们问这个轨迹是否 花费大量时间紧紧围绕概念 的内省。

 

 

And with this, we could analyze the history of introspection in the ancient Greek tradition, for which we have the best available written record. So what we did is we took all the books -- we just ordered them by time -- for each book we take the words and we project them to the space, and then we ask for each word how close it is to introspection, and we just average that. And then we ask whether, as time goes on and on, these books get closer, and closer and closer to the concept of introspection.

有了这个，我们可以分析 古希腊传统中的内省历史，我们拥有最好的 可用的书面记录。 所以我们做的是把所有的书都拿来—— 我们只是按时间排序—— 对于每本书，我们把单词投射到空间里，然后我们要求每个单词。 它与内省有多接近，我们只是平均这一点。 然后我们问， 随着时间的流逝，这些书越来越近， 越来越接近内省的概念。

 

And this is exactly what happens in the ancient Greek tradition. So you can see that for the oldest books in the Homeric tradition, there is a small increase with books getting closer to introspection. But about four centuries before Christ, this starts ramping up very rapidly to an almost five-fold increase of books getting closer, and closer and closer to the concept of introspection. And one of the nice things about this is that now we can ask whether this is also true in a different, independent tradition.

这正是发生的事情 在古希腊传统中。 所以你可以看到最古老的书 在荷马传统中，书籍略有增加 越来越接近内省。 但是在基督之前大约四个世纪，这种情况开始迅速增加。 近五倍的书籍越来越近， 越来越接近内省的概念。 这样做的好处之一是，现在我们可以问这是否也是真的。 在一个不同的、独立的传统中。

 

 

footnote

footnote

So we just ran this same analysis on the Judeo-Christian tradition, and we got virtually the same pattern. Again, you see a small increase for the oldest books in the Old Testament, and then it increases much more rapidly in the new books of the New Testament. And then we get the peak of introspection in "The Confessions of Saint Augustine," about four centuries after Christ. And this was very important, because Saint Augustine had been recognized by scholars, philologists, historians, as one of the founders of introspection. Actually, some believe him to be the father of modern psychology.

所以我们刚刚进行了同样的分析 在犹太教-基督教传统上，我们得到了几乎相同的模式。 同样，您会看到小幅增加 对于旧约中最古老的书卷，然后在新约的新书中增长得更快。 然后我们在“圣奥古斯丁的忏悔”中达到了内省的顶峰，大约在基督之后四个世纪。 这非常重要，因为圣奥古斯丁 被学者、语言学家、历史学家公认为内省的奠基人之一。 实际上，有些人认为他是 现代心理学之父。

 

 

So our algorithm, which has the virtue of being quantitative, of being objective, and of course of being extremely fast -- it just runs in a fraction of a second -- can capture some of the most important conclusions of this long tradition of investigation. And this is in a way one of the beauties of science, which is that now this idea can be translated and generalized to a whole lot of different domains.

所以我们的算法，它有优点 定量的，客观的，当然还有极快的——它只是在几分之一秒内运行——可以捕捉到一些最 这一悠久的调查传统的重要结论。 这在某种程度上是 科学的美妙之处之一，就是现在这个想法 可以翻译和概括为很多 不同的领域。

 

So in the same way that we asked about the past of human consciousness, maybe the most challenging question we can pose to ourselves is whether this can tell us something about the future of our own consciousness. To put it more precisely, whether the words we say today can tell us something of where our minds will be in a few days, in a few months or a few years from now.

所以就像我们问的那样 关于人类意识的过去，也许是最具挑战性的问题 我们可以对自己提出，这是否可以告诉我们一些事情 关于我们自己意识的未来。 更准确地说，我们今天说的话是否能告诉我们一些事情 几天后、几个月或几年后我们的思想将在哪里 。

 

And in the same way many of us are now wearing sensors that detect our heart rate, our respiration, our genes, on the hopes that this may help us prevent diseases, we can ask whether monitoring and analyzing the words we speak, we tweet, we email, we write, can tell us ahead of time whether something may go wrong with our minds. And with Guillermo Cecchi, who has been my brother in this adventure, we took on this task. And we did so by analyzing the recorded speech of 34 young people who were at a high risk of developing schizophrenia.

以同样的方式，我们中的许多人 现在戴着传感器来检测我们的心率、呼吸、基因，希望这可能会 帮助我们预防疾病，我们可以问是否监测 分析我们说的话，我们发推文，我们发电子邮件，我们写，可以提前告诉我们是否 我们的思想可能会出问题。 吉列尔莫·切奇（Guillermo Cecchi）是我在这次冒险中的兄弟，我们承担了这项任务。 我们通过分析 34名处于高风险中的年轻人的录音演讲 发展中的精神分裂症。

 

 

And so what we did is, we measured speech at day one, and then we asked whether the properties of the speech could predict, within a window of almost three years, the future development of psychosis. But despite our hopes, we got failure after failure. There was just not enough information in semantics to predict the future organization of the mind. It was good enough to distinguish between a group of schizophrenics and a control group, a bit like we had done for the ancient texts, but not to predict the future onset of psychosis.

所以我们所做的是， 我们在第一天测量了语音，然后我们询问属性是否 的演讲可以预测，在近三年的窗口内，精神病的未来发展。 但是，尽管我们抱有希望，我们还是一次又一次地失败。 只是不够 预测未来的语义学信息 心灵的组织。 区分一个群体就足够了 精神分裂症患者和对照组，有点像我们所做的那样 对于古代文本，但不预测未来 精神病发作。

 

But then we realized that maybe the most important thing was not so much what they were saying, but how they were saying it. More specifically, it was not in which semantic neighborhoods the words were, but how far and fast they jumped from one semantic neighborhood to the other one. And so we came up with this measure, which we termed semantic coherence, which essentially measures the persistence of speech within one semantic topic, within one semantic category.

但后来我们意识到，也许是最重要的事情 与其说他们在说什么，不如说他们怎么说。 更具体地说，它不是在哪个语义 这些词是邻里，但它们从一个语义邻域跳得有多远和多快 到另一个。 所以我们提出了这个度量，我们称之为语义连贯性，它基本上测量持久性。 一个语义主题内的语音，一个语义类别内的语音。

 

 

And it turned out to be that for this group of 34 people, the algorithm based on semantic coherence could predict, with 100 percent accuracy, who developed psychosis and who will not. And this was something that could not be achieved -- not even close -- with all the other existing clinical measures.

结果是 对于这组34人，基于语义的算法 连贯性可以100%准确地预测谁患上了精神病，谁不会。 这是一件事情 这是无法实现的 - 甚至无法接近 - 与所有其他 现有的临床措施。

 

 

footnote

footnote

And I remember vividly, while I was working on this, I was sitting at my computer and I saw a bunch of tweets by Polo -- Polo had been my first student back in Buenos Aires, and at the time he was living in New York. And there was something in this tweets -- I could not tell exactly what because nothing was said explicitly -- but I got this strong hunch, this strong intuition, that something was going wrong. So I picked up the phone, and I called Polo, and in fact he was not feeling well. And this simple fact, that reading in between the lines, I could sense, through words, his feelings, was a simple, but very effective way to help.

我清楚地记得， 当我在做这件事的时候，我坐在电脑前，我看到了一堆波罗的推文——波罗是我的第一个学生。 回到布宜诺斯艾利斯，当时 他住在纽约。 这条推文中有一些东西 - 我无法确切地说出是什么 因为什么都没有明确地说出来——但我有这种强烈的预感，这种强烈的直觉， 出事了。 于是我拿起电话， 我打电话给波罗，事实上他感觉不舒服。 这个简单的事实，在字里行间阅读，我能感觉到， 通过言语，他的感受，很简单，但非常 有效的帮助方式。

 

 

What I tell you today is that we're getting close to understanding how we can convert this intuition that we all have, that we all share, into an algorithm. And in doing so, we may be seeing in the future a very different form of mental health, based on objective, quantitative and automated analysis of the words we write, of the words we say.

我今天告诉你的是，我们正在得到 接近于理解我们如何转换这种直觉 我们都有，我们都分享，变成一个算法。 在这样做的过程中，我们可能会在未来看到 一种非常不同的心理健康形式，基于客观、定量 并自动分析我们写的单词，我们说的话。

 

Gracias.

谢谢。