引言与访谈说明

Curt Jaimungal: I just finished recording with Noam Chomsky. This is just an editor's note. His audio is poor and there's nothing that we can do. This tends to happen when someone changes location or if they're not well-versed enough to fix the issue because they're not familiar with computers enough. They don't know how to tweak the settings.

Curt Jaimungal: 我刚和诺姆·乔姆斯基录制完。这只是一个编者按。他的音频质量不佳，我们对此无能为力。当嘉宾更换录音地点，或者因为对电脑不够熟悉而无法解决技术问题时，这种情况时有发生。他们不知道如何调整设置。

Curt Jaimungal: However, there are great questions from Joscha Bach, from Professor Avi Loeb, from Karl Friston, from Philip Goff, from Tim Maudlin. So many great questions. So please endure. Persist past it.

Curt Jaimungal: 然而，我们有来自约沙·巴赫、阿维·勒布教授、卡尔·弗里斯顿、菲利普·戈夫、蒂姆·莫德林等人的精彩提问。问题都非常棒。所以请大家耐心听下去。

Curt Jaimungal: Donald Hoffman even has some questions about Space Time is Doomed. As usual, because Chomsky is not hearing well, I've printed out the questions for him beforehand. And so you'll hear me refer, okay, go to question 25, go to question 22, and so on. This is because we can't have a usual back and forth like it would with a regular TOE guest. They can hear me, a regular TOE guest, but Chomsky has a difficult time. So we have to stick straight to questions, answers, questions, answers. Okay, now on to the regular introduction.

Curt Jaimungal: 唐纳德·霍夫曼甚至提出了一些关于“时空注定消亡”的问题。像往常一样，因为乔姆斯基听力不好，我提前为他打印了问题。所以你会听到我说，好的，看第 25 个问题，看第 22 个问题，等等。这是因为我们无法像和常规 TOE（万物理论频道）嘉宾那样进行自然的来回交流。常规嘉宾能听清我，但乔姆斯基听起来很困难。所以我们必须严格按照“提问、回答、提问、回答”的模式进行。好了，现在进入常规介绍。

Curt Jaimungal: This is the eighth time that I've been lucky enough to speak to Professor Chomsky, the father of modern linguistics. And in keeping with the vein of TOE, we adhere to philosophy, consciousness, meaning, rather than, say, politics. There's a playlist in the description with each of Chomsky's theories of everything appearances cataloged below. And coming up in about three weeks, I'll be posting a three-hour behemoth compilation of the best of Noam's philosophical thoughts on TOE, with subtitles hard-coded in. If you'd like to volunteer some assistance to double check that the captions are indeed correct, because there's plenty of specialized terminology and names, then DM me on Twitter @TOEwithCurt. That is at T-O-E-W-I-T-H-C-U-R-T. Enjoy this eighth episode with Noam Chomsky.

Curt Jaimungal: 这是我第八次有幸与现代语言学之父乔姆斯基教授对话。为了保持 TOE 频道的风格，我们专注于哲学、意识、意义等话题，而非政治。描述区有一个播放列表，收录了乔姆斯基在 TOE 频道的所有访谈。大约三周后，我将发布一个长达三小时的精选合集，汇集诺姆在 TOE 上关于哲学的思考，并内嵌字幕。如果你愿意志愿协助校对字幕，确保其中大量的专业术语和人名准确无误，请在 Twitter 上私信我 @TOEwithCurt。账号是 T-O-E-W-I-T-H-C-U-R-T。请享受这第八期与诺姆·乔姆斯基的对话。

Curt Jaimungal: Thank you, Professor. Thank you for joining me for the eighth time.

Curt Jaimungal: 谢谢您，教授。感谢您第八次接受我的采访。

Noam Chomsky: Thank you. Pleased to be with you again.

诺姆·乔姆斯基: 谢谢你。很高兴再次与你交流。

[约沙·巴赫] 认知科学最重要的未解问题是什么？

Curt Jaimungal: So, Joscha Bach asks, what is the most important unsolved problem in cognitive science?

Curt Jaimungal: 好的，约沙·巴赫提问：认知科学（Cognitive Science：研究心智或思维信息处理过程的交叉学科）中最重要的未解问题是什么？

Noam Chomsky: Well, there are different criteria to determine what's the most important problem. For the working scientists, the most important problems unsolved are the ones right at the border of understanding, the ones that you have some grasp about, have an approach, but haven't really solved. Those are the ones you work on. Well, by that criterion, there's endless numbers of important unsolved problems, depending on what area you're in, you can enumerate any number of them. They won't be meaningful to others unfamiliar with your area of research.

诺姆·乔姆斯基: 嗯，确定什么是最重要的问题有不同的标准。对于一线科学家来说，最重要的未解问题是那些正处于理解边界上的问题，即你对它有一定把握，有研究方法，但尚未完全解决。这些就是你正在研究的问题。按照这个标准，有无数个重要的未解问题，具体取决于你所在的领域，你可以列举出任意数量，但对于不熟悉你研究领域的人来说，这些问题可能没有意义。

Noam Chomsky: I think what the problem may be, the question you're driving at is far-reaching problems, problems which have large consequences. In that case, there are also many, but the most important one that I think of is the problem of voluntary action. So, there's a famous state-of-the-art article on this, two of the leading researchers, Emilio Bizzi and Roberto Jamian, came out in Daedalus Journal of the American Academy a couple of years ago. And they went through what has been discovered about this problem, and what remains unsolved.

诺姆·乔姆斯基: 我认为你问题所指向的，可能是那些影响深远、后果重大的问题。在这种情况下，也有很多，但我能想到的最重要的问题是“自主行为”问题。关于这个问题，有一篇著名的前沿综述文章，由两位顶尖研究者埃米利奥·比齐和罗伯托·哈米安几年前发表在《代达罗斯》（美国文理科学院院刊）。他们梳理了关于这个问题已有的发现以及仍然未解的部分。

Noam Chomsky: And they put the question fancifully as they say at the end. They say, we have a grasp on how to deal with the puppet and the strings, but we can't say anything about the puppeteer, no matter how much they search, no matter what experiments they carry out, neural experiments and others. They can't figure out what lies behind my decision to, say, raise my finger, versus not to, or my decision to give one rather than another response to this question. That is a very far-reaching question. I think it's completely unresolved.

诺姆·乔姆斯基: 在文章结尾，他们用一个巧妙的比喻来描述这个问题。他们说，我们掌握了如何研究木偶和牵引它的线，但对于那个操纵木偶的人，我们却一无所知。无论他们做了多少研究，进行了多少神经实验或其他实验，他们都无法弄清楚我决定举起手指（而不是不举），或者我决定用一种方式而不是另一种方式回答这个问题的背后原因。这是一个影响极其深远的问题。我认为它完全没有得到解决。

Noam Chomsky: There are many, many scientists who feel that we have an answer. The answer is determined by complex physical laws and so on and so forth. Phenomena that are structures that are much too complex to deal with. That's a thesis that is vacuous, unverifiable. If it makes you feel comfortable to add it, okay. It's not doing anything. So I think that's a completely unsolved problem.

诺姆·乔姆斯基: 有很多科学家觉得我们已经有了答案，认为答案是由复杂的物理定律等等决定的，是由于现象的结构过于复杂而无法处理。这种论点是空洞的、无法验证的。如果你加上这个说法会让你心安，那也行，但它实际上没有任何解释力。所以我认为这是一个完全未解的问题。

[约沙·巴赫] 过去五年您对 AI 的看法有何更新？

Curt Jaimungal: We'll go to question 1C, still Joscha Bach. In which ways did you update your thinking on AI in the past five years? Basically, I haven't. The reason is that AI has not advanced in the last five years in ways which have anything to do with my thinking about it.

Curt Jaimungal: 我们来看问题 1C，仍然是约沙·巴赫提出的。在过去五年里，您对人工智能（AI：Artificial Intelligence）的看法在哪些方面有所更新？

Noam Chomsky: Basically, I haven't. The reason is that AI has not advanced in the last five years in ways which have anything to do with my thinking about it.

诺姆·乔姆斯基: 基本上没有。原因在于，过去五年人工智能的进步方式与我对其的思考毫无关联。

Noam Chomsky: It has advanced in many areas, but they just aren't related to my thinking about it. So for example, there are recent successes of the deep learning programs which have cast light, significant light, on serious problems of science like protein folding. The methods that are used are not of a kind that bear on AI. The preceding question asks whether the recent successes of machine learning surprised me. Well, that assumes that there are recent successes.

诺姆·乔姆斯基: 它在很多领域取得了进步，但这些进步与我的思考不相干。例如，最近深度学习（Deep Learning：机器学习的一个分支，模拟人脑进行数据分析）程序在蛋白质折叠等重大科学问题上取得了成功，带来了重要的启示。但所使用的方法与（我所关注的）人工智能无关。之前的问题问到机器学习的近期成功是否让我惊讶。嗯，这首先就假设了存在所谓的“近期成功”。

Noam Chomsky: I don't know of any in the areas that I work in, most of the language. There's lots of achievements, simple achievements that receive plenty of enthusiastic press commentary, but when you look into them, they're telling us basically nothing. So take the GPT series, which has received a lot of enthusiastic commentary, or Lambda, or others like it. I don't see what's successful about these.

诺姆·乔姆斯基: 在我工作的领域，主要是语言学领域，我不知道有任何成功。有很多成就，一些简单的成就获得了媒体热情洋溢的评论，但当你深入研究它们时，会发现它们基本上什么也没告诉我们。就拿 GPT 系列（Generative Pre-trained Transformer：一种大型语言模型架构）或 LaMDA（Language Model for Dialogue Applications：谷歌开发的对话式大语言模型）以及其他类似的模型来说，我看不出它们有什么成功之处。

Noam Chomsky: What has been shown is that if you toss a couple of supercomputers at 50 terabytes of data, you can find surface regularities, which if you put them together, delude people into thinking that the program is doing something. Doesn't sound very interesting to me. We know that the programs are not doing anything, because they fail the most elementary condition required for an explanation. Namely, that you say, why things are this way, and crucially, why they're not some other way.

诺姆·乔姆斯基: 它们所展示的不过是，如果你用几台超级计算机处理 50TB 的数据，你就能发现一些表面的规律性，把这些规律组合起来，就能骗得人们以为这个程序在“做”什么。这对我来说听起来没什么意思。我们知道这些程序什么也没做，因为它们不满足一个解释所需要的最基本条件。也就是说，你得能说明，为什么事情是这样，以及至关重要的是，为什么事情不是那样。

Noam Chomsky: Now, if I come to a physics conference, I've got a terrific theory, which includes all the laws of nature, the ones you've discovered, the ones you haven't yet discovered. And the theory is so simple, I can say it in just two words. Anything goes. I won't get any prizes. Doesn't say why things are not some other way. Well, that's what all of these programs do.

诺姆·乔姆斯基: 现在，如果我参加一个物理学会议，说我有一个了不起的理论，它包含了所有自然法则，包括已发现的和未发现的。这个理论非常简单，我可以用两个词概括：“什么都行”（Anything goes）。我不会因此获奖的。因为它没有解释为什么事情不是另一种样子。而所有这些程序做的就是这个。

Noam Chomsky: They work just as well for impossible systems as for possible ones. A system, if you give it a corpus, a massive corpus that violates the basic principles of language, it'll work fine, maybe better. So therefore, whatever it's doing, it's not telling us anything about language or cognition.

诺姆·乔姆斯基: 它们处理不可能的系统和处理可能的系统一样得心应手。如果你给它一个语料库，一个违反了语言基本原则的大规模语料库，它照样能运行得很好，甚至可能更好。因此，无论它在做什么，它都没有告诉我们任何关于语言或认知的事情。

Noam Chomsky: And it seems to me that all of the systems that I'm familiar with have that problem. So there's no successes that are surprising. There are very useful achievements. So for example, the life transcription, which I rely on because I'm hard of hearing, it's a fine achievement. But Google took the trouble to create it. But it's telling us nothing about the way humans interpret sentences. They don't surprise me. And there's been nothing to update my thinking. They're useful achievements, the 13-fold increase.

诺姆·乔姆斯基: 在我看来，我所熟悉的所有系统都存在这个问题。所以没有什么令人惊讶的成功。确实有一些非常有用的成就。例如，实时转写功能，因为我听力不好，我很依赖它，这是一项很好的成就。谷歌费心创造了它。但它并没有告诉我们任何关于人类如何理解句子的信息。它们没有让我感到惊讶。也没有任何东西足以更新我的想法。它们只是有用的成就而已，就像效率提高了 13 倍一样。

[约沙·巴赫] 当前的神经网络能否通向类人智能？

Curt Jaimungal: The next question is, brains and minds work very differently from today's neural networks. Do you think that today's approaches can be scaled and improved to reach human-like intelligence? Or do we need to divest from deep learning, go back to the drawing board and build something different from scratch if we want to get close to human performance?

Curt Jaimungal: 下一个问题是：大脑和心智的运作方式与今天的神经网络截然不同。您认为当今的方法可以通过扩展和改进达到类人智能吗？还是说，如果我们想接近人类的水平，就需要放弃深度学习，回到原点，从头开始构建一些完全不同的东西？

Noam Chomsky: No, I don't think the deep learning approach has anything to do with the way minds work. They're very useful. A lot of serious, sophisticated math and other major achievements, but it's about something different, not the way minds work. And part for the reasons I just mentioned, doesn't distinguish the way brains and minds work from the way they don't work. As long as you don't make that distinction, there's nothing said about how brains and minds work.

诺姆·乔姆斯基: 不，我不认为深度学习方法与心智的运作方式有任何关系。它们非常有用，涉及很多严肃、复杂的数学和其他重大成就，但它研究的是别的东西，不是心智的运作方式。部分原因就是我刚才提到的，它无法区分大脑和心智运作的方式与它们不运作的方式。只要你做不出这种区分，你就没有对大脑和心智如何运作说出任何有意义的东西。

Noam Chomsky: In the specific case of language, the one I looked at most closely, the problem is exactly what I mentioned. The systems like the GPT series and the others do just as well on non-languages as on languages. So they're very much like that imagine-talk-to-the-physics conference. It's not telling you anything about language by definition.

诺姆·乔姆斯基: 在语言这个我研究得最仔细的具体案例中，问题正是我所提到的。像 GPT 系列和其他系统，它们处理非语言数据和处理语言数据一样出色。所以它们很像那个去物理学会议上兜售“什么都行”理论的人。根据其定义，它根本没有告诉你任何关于语言的事情。

[约沙·巴赫] 意识对于实现人类水平的智能是否至关重要？

Curt Jaimungal: Okay, the next one, 1E. Is consciousness crucial to building systems that learn, think and create at a human level?

Curt Jaimungal: 好的，下一个问题，1E。对于构建能够在人类水平上学习、思考和创造的系统而言，意识是否至关重要？

Noam Chomsky: Consciousness is part of the way humans learn, think and create. Maybe not a very major part. Most of what our mental activities are, both in what's called learning, thinking and creating, is inaccessible to consciousness. So there's an awful lot going on that we're totally unaware of and can't become aware of. But consciousness enters at some level here and there. So if anything is going to think, learn and create the way humans do, it will have to involve consciousness.

诺姆·乔姆斯基: 意识是人类学习、思考和创造方式的一部分。也许不是一个非常主要的部分。我们大部分的心理活动，无论是在所谓的学习、思考还是创造中，都是意识无法触及的。因此，有大量的事情在发生，而我们完全没有意识到，也无法意识到。但意识在某些层面上时而介入。所以，如果有什么东西要像人类一样思考、学习和创造，它就必须涉及意识。

Noam Chomsky: Now when we say at a human level, that opens other questions. What's a human level? We have to answer that before we know the answer to this question. And I don't know of any characterization of at a human level other than saying, this is the way humans do it. Presumably at a human level includes things that are not the way humans do it, but somehow reach that level. But then we have to have a characterization of what the level is. I don't know of any.

诺姆·乔姆斯基: 当我们说“在人类水平上”时，这就引出了其他问题。“人类水平”是什么？在回答这个问题之前，我们必须先回答那个问题。除了说“这就是人类做事的方式”之外，我不知道对“人类水平”有任何其他的描述。大概“在人类水平上”也包括那些做事方式与人类不同，但以某种方式达到了那个水平的东西。但那样的话，我们就必须对这个“水平”有一个定义。我不知道有任何这样的定义。

[约沙·巴赫] 您仍然是一位“新神秘主义者”吗？

Curt Jaimungal: Okay, and number 1F. So the next one, are you a mysterian still in the sense that understanding mind and consciousness is not possible for us? Or do you feel there's a research direction that you can sketch out? First of all, I've never held that it's not possible for us.

Curt Jaimungal: 好的，问题 1F。下一个问题是，您仍然是一位“新神秘主义者”（Mysterian：一种哲学观点，认为人类心智在结构上无法解决某些难题，如意识问题），认为理解心智和意识对我们来说是不可能的吗？还是说您觉得可以勾勒出一个研究方向？

Noam Chomsky: First of all, I've never held that it's not possible for us. In fact, I think we can understand quite a lot about mind and consciousness. There are many questions that we have no answers to, even bad answers, and that have withstood inquiry for ages, still do. Like the one I mentioned in citing Bizzi and I know that, and Meiji and Jamie, is that within the realm of possible human understanding?

诺姆·乔姆斯基: 首先，我从未持有过“对我们来说不可能”的观点。事实上，我认为我们能够相当多地理解心智和意识。有很多问题我们没有答案，甚至连糟糕的答案都没有，这些问题经受了长久的探究，至今仍是如此。就像我之前引述比齐和杰米安时提到的那个问题，它是否在人类可能理解的范围之内？

Noam Chomsky: That goes back to an earlier question. Are humans organisms in the normal sense, or are they outside the range of organisms? I think humans are organisms. If so, they're going to be like other organisms. Their capacities have limits. The scope and the limits are related. They have to do with the internal, essential nature of the organism. So if you're, say, a rat, you can be trained to solve many kinds of mazes, which can't solve the prime number maze in which you have to turn right at every prime number option. And the reason is, prime number is just not within the cognitive capacity of a rat.

诺姆·乔姆斯基: 这又回到了一个更早的问题。人类是普通意义上的有机体，还是超出了有机体的范畴？我认为人类是有机体。如果是这样，他们就会像其他有机体一样。他们的能力有其极限。范围和极限是相关的，它们与有机体内在的、本质的属性有关。比如说，如果你是一只老鼠，你可以被训练解决多种迷宫，但无法解决一个“素数迷宫”，即你必须在每个素数选项处右转。原因在于，素数的概念根本不在老鼠的认知能力范围之内。

Noam Chomsky: There are many things that are not within our cognitive capacity. So I happen to live in the desert, out in my backyard are desert ants who have minuscule brains, but are capable of cognitive achievements that humans can't attain, like their navigational capacities. We can maybe duplicate it with instruments to an extent, but I can't find my way around the way an ant can. So that's a limit to my capacities.

诺姆·乔姆斯基: 同样，有很多事情也不在我们的认知能力范围之内。我碰巧住在沙漠里，我家后院有沙漠蚂蚁，它们的大脑微不足道，但却能完成人类无法企及的认知成就，比如它们的导航能力。我们或许可以用仪器在一定程度上复制这种能力，但我无法像蚂蚁那样找到路。所以这是我能力的一个极限。

Noam Chomsky: Do we have other kinds of limits? Well, if we're organisms, then comes the question everyone's interested in. Is something like the ability to make distinctions, whatever it is, is the explanation of that outside the range of our cognitive capacities? It's an empirical question. Is there a research direction you can sketch out? Sure. The one that's pursued, for example, by the Viet Cien, and Jamie, very good scientists, they try to find out they haven't been able to. Is it beyond their cognitive capacities? We simply don't know.

诺姆·乔姆斯基: 我们是否有其他类型的极限？嗯，如果我们是有机体，那么大家感兴趣的问题就来了。像做出区分的能力这样的东西，无论它是什么，对其的解释是否超出了我们的认知能力范围？这是一个经验性问题。有没有可以勾勒出的研究方向？当然有。就是比齐和杰米安这些非常优秀的科学家正在追求的方向。他们试图找出答案，但一直未能成功。这是否超出了他们的认知能力？我们根本不知道。

[唐纳德·霍夫曼] 时空注定消亡吗？

Curt Jaimungal: Question number two comes from Professor Donald Hoffman. 2a. What do you think about the claim that Nima Arkani-Hamed, David Gross, and other physicists are making that space-time is doomed, that we must look for new structures utterly beyond space-time and quantum theory that give rise to space-time and quantum theory?

Curt Jaimungal: 第二个问题来自唐纳德·霍夫曼教授。2a. 尼马·阿尔卡尼-哈米德、戴维·格罗斯等物理学家声称“时空注定消亡”，我们必须寻找超越时空和量子理论的全新结构，而时空和量子理论本身源自于这些新结构。您对这一主张有何看法？

Noam Chomsky: Of these three questions, a, b, and c, the first two, a and b, are utterly beyond my competence to answer. I just don't know enough about quantum theory, the arguments given, so I can't say anything about them.

诺姆·乔姆斯基: 在这三个问题 a、b、c 中，前两个问题 a 和 b 完全超出了我的能力范围。我对量子理论和相关的论证了解不够，所以无法对它们发表任何看法。

[唐纳德·霍夫曼] 放弃时空对心物问题有何影响？

Curt Jaimungal: Okay, so we'll go to 2c. What about what these discoveries of letting go of space-time might entail for the mind-body problem by Professor Donald Hoffman?

Curt Jaimungal: 好的，那我们来看 2c。唐纳德·霍夫曼教授提问：放弃时空这一发现，可能对心物问题（Mind-body problem：关于意识或心理事件与物理大脑之间关系的哲学难题）意味着什么？

Noam Chomsky: I have an opinion, for whatever it's worth, that they don't entail anything for the mind-body problem, because there is no such problem. There was a mind-body problem at one time.

诺姆·乔姆斯基: 我有一个看法，不管价值如何，我认为它们对心物问题没有任何影响，因为根本不存在所谓的心物问题。曾经有过一个心物问题。

Noam Chomsky: A classic exposition of it is Descartes, of course. Descartes argued, claimed in his physics that he was able to account for aspects of the world, including most of the properties of humans, by keeping within what was called at the time the mechanical philosophy. Philosophy just meant science. Mechanical science, it's basic conception of science that held from Galileo through Newton, namely that the world is a complex of the kind that's constructed by skilled artisans.

诺姆·乔姆斯基: 当然，其经典阐述者是笛卡尔。笛卡尔在他的物理学中论证并声称，他能够通过遵循当时的所谓“机械哲学”（Mechanical philosophy：将宇宙视为一部巨大、复杂的机器的科学观点）来解释世界的各个方面，包括人类的大部分属性。当时的“哲学”就是指科学。机械科学，是自伽利略到牛顿时期科学的基本观念，即世界是一个由技艺高超的工匠所构建的复杂集合体。

Noam Chomsky: At the time in Europe, there was a proliferation of complex objects, which are structures that were developed by skilled artisans, which amazed people in their conformity, in their similarity to what humans could do or what other animals could do. And the basic conception was, well, the world is just something created by an incomparably more skilled artisan, gears, levers, and so on. Descartes thought erroneously that his physical theory would account for almost all of nature within this framework. It wasn't accurate.

诺姆·乔姆斯基: 在那个时期的欧洲，涌现出大量复杂的物体，这些是由能工巧匠制造的结构，它们与人类或动物所能做到的事情惊人地相似，令人们叹为观止。当时的基本观念是，世界只不过是由一位技艺无比高超的工匠用齿轮、杠杆等创造出来的东西。笛卡尔错误地认为，他的物理理论能够在这个框架内解释几乎所有的自然现象。这是不准确的。

Noam Chomsky: In fact, one of Newton's contributions was to go through it, but it was understood by other scientists, Huygens and others, that Descartes' physics just didn't work. But he had an explanation for everything in those terms, then observed correctly that there are certain things that don't fall within mechanical explanation. One of his major examples, in fact, was pretty much what I said before, except he was talking about language, not lifting your finger. So, he said the capacity of humans in ordinary discussion, ordinary interchange, ordinary thought to construct new expressions of thought, which have never occurred before, and which are appropriate to situations but not caused by them.

诺姆·乔姆斯基: 事实上，牛顿的贡献之一就是审视了这一点，而惠更斯等其他科学家也明白，笛卡尔的物理学根本行不通。但笛卡尔用他的理论解释了一切，然后又正确地观察到，有些事情是无法用机械论来解释的。他的一个主要例子，其实和我之前说的差不多，只是他谈论的是语言，而不是举起手指。他说，人类在日常讨论、交流和思考中，能够构建出前所未有的思想表达，这些表达切合情境但并非由情境所引发。

Noam Chomsky: That's the system he was considering, certainly seems to be a normal property of human behavior. He argued, plausibly, that that doesn't fall within mechanical science. So, therefore, as a scientist, he postulated a new principle, res cogitans, thinking substance. Then you had a mind-body problem. How are these two substances related? That problem disappeared with Newton, who showed that there are no bodies in Descartes' sense. There's no physical entities in anything like the sense of early modern science.

诺姆·乔姆斯基: 他所考虑的这个系统，确实似乎是人类行为的一种正常属性。他很有说服力地论证说，这不属于机械科学的范畴。因此，作为一名科学家，他提出了一个新的原则，即“res cogitans”（思维实体：笛卡尔哲学中指代非物质的心灵或意识的术语）。于是，心物问题就产生了。这两种实体是如何关联的？这个问题随着牛顿的出现而消失了。牛顿证明了，笛卡尔意义上的“物体”是不存在的。近代早期科学意义上的物理实体根本不存在。

Noam Chomsky: Newton himself thought this result was completely absurd. He argued no person with any scientific understanding could accept it. In fact, he spent the rest of his life trying to overcome it. Finally, then his contemporaries like Huygens, Leibniz, notice what they were saying, that it wasn't that Newton's theories were unintelligible. They could understand the theories fine. It was the kind of world that he was postulating that was absurd. And in fact, it's in that context that Newton made his famous statement that he proposes no hypotheses. He was providing just a mathematical theory, not a physical theory.

诺姆·乔姆斯基: 牛顿本人认为这个结果是完全荒谬的。他认为任何有科学素养的人都无法接受它。事实上，他余生都在试图推翻它。最终，他的同代人如惠更斯、莱布尼茨，注意他们的说法，他们并不是说牛顿的理论难以理解，他们完全能理解这些理论，而是说牛顿所假设的那个世界是荒谬的。事实上，正是在这种背景下，牛顿发表了他著名的声明，即他“不作任何假设”。他提供的只是一个数学理论，而非一个物理理论。

Noam Chomsky: Finally, over time, Newton's mathematical theories were accepted, and scientists lost interest in whether the theory, if the theory was intelligible, whether what it described was intelligible. That question disappeared. As far as I can see, mind-body problems disappeared with it. We have no conception of body other than what our best theories postulate, which just as well includes mind. In fact, John Locke drew that conclusion right away. He said, for all we know, the organized matter just has properties we can't conceive of, like action, interaction without contact. So, it may also have properties like thinking.

诺姆·乔姆斯基: 最终，随着时间的推移，牛顿的数学理论被接受了，科学家们也就不再关心理论本身是否可理解，或者它所描述的世界是否可理解了。那个问题消失了。在我看来，心物问题也随之消失。我们对“物体”的概念，除了我们最好的理论所假设的之外，一无所知，而这些理论同样也包含了心智。事实上，约翰·洛克立刻就得出了这个结论。他说，据我们所知，有组织的物质就是拥有我们无法想象的属性，比如无需接触的相互作用。那么，它也可能拥有像思维这样的属性。

Noam Chomsky: He put this in a theological context, but we can drop that. Soon through the 18th century, very extensively, the problem of thinking matter, what properties of matter yield thinking where matter is whatever the world is constituted of. We still have basically no answer then. So, I just don't see how a mind-body problem, at least in the classical sense, can be formulated after Newton. I don't know of any other sense in which it can.

诺姆·乔姆斯基: 他把这一点放在神学背景下，但我们可以忽略这一点。很快，在整个 18 世纪，关于“会思考的物质”的问题被广泛讨论，即物质的何种属性产生了思维，这里的“物质”指的是构成世界的一切。对此我们至今基本上没有答案。所以，我实在看不出在牛顿之后，一个心物问题，至少在经典意义上的心物问题，如何还能被提出来。我也不知道它能以任何其他形式被提出。

[阿维·勒布] 人工智能会超越人类并建立自己的社会吗？

Curt Jaimungal: This one comes from Professor Avi Loeb. Imagine that we will develop sentient AI systems, and these will be able to interact with each other. Do you expect them to develop their own kinship, including a language that is more advanced than human languages, and that humans will not be able to follow? In such a case, they might be able to establish their own society of AI systems. We could always unplug them from their electric outlet, although some will view that as equivalent to murder.

Curt Jaimungal: 这个问题来自阿维·勒布教授。想象一下，我们将开发出有感知能力的人工智能系统，它们能够相互交流。您是否预期它们会发展出自己的亲缘关系，包括一种比人类语言更高级、人类无法理解的语言？在这种情况下，它们或许能够建立自己的人工智能系统社会。我们当然可以随时拔掉它们的电源，尽管有些人会认为这等同于谋杀。

Noam Chomsky: Well, to answer this question, we have to have a concept of sentient AI systems. There certainly are, if by sentient we mean capable of interacting with one another, responding to one another's states, and so on, and responding to features of the environment that are significant for choice of action, and so on. If that's what we mean by it, there are many sentient systems. Humans, animals, trees, so take trees. Trees are sentient in this sense.

诺姆·乔姆斯基: 嗯，要回答这个问题，我们首先需要一个“有感知能力的人工智能系统”的概念。如果“有感知能力”指的是能够相互作用、对彼此的状态作出反应，并对环境中对行动选择有重要意义的特征作出反应等等，那么当然存在这样的系统。如果这是我们的定义，那么有很多有感知能力的系统。人类、动物、树木——就拿树木来说。在这个意义上，树木是有感知能力的。

Noam Chomsky: If you have a forest, the trees interact with one another. They send signals to one another. They change what they do, dependent on the signals they receive from others. Of course, they respond to the environment. So, these seem to be sentient systems in any sense of the word that I know of. They do interact with each other. They don't develop their own kinship or language, and so on. The answer to whether sentient systems will do this is, well, obviously not, because plenty don't, just as animals don't.

诺姆·乔姆斯基: 如果你在一片森林里，树木会相互作用。它们互相发送信号。它们会根据从其他树木接收到的信号来改变自己的行为。当然，它们也对环境作出反应。所以，在我所知的任何词义上，这些似乎都是有感知能力的系统。它们确实相互作用。但它们没有发展出自己的亲缘关系或语言等等。所以，关于有感知能力的系统是否会这样做，答案是，嗯，显然不会，因为有很多系统就不会，就像动物一样。

Noam Chomsky: What about sentient AI systems? Well, that depends on how an AI system differs from natural systems. What's an AI system? Basically, an AI system is a program. A program is basically a theory that is presented in a notation that the computer can implement. So, we're basically asking, can we develop sentient theories? I'm not sure exactly what that means.

诺姆·乔姆斯基: 那么有感知能力的人工智能系统呢？这取决于人工智能系统与自然系统的区别。什么是人工智能系统？基本上，一个人工智能系统就是一个程序。一个程序基本上是一个用计算机可以执行的符号系统呈现的理论。所以，我们基本上是在问，我们能开发出有感知能力的理论吗？我不太确定这具体意味着什么。

Noam Chomsky: Maybe theories of sentience. Can we develop theories of sentience? Well, now we're back to the earlier questions. Is this within, to a certain extent, of course, we can. We have many such theories. Are they complete? No, they're not complete. But of course, we expect theories not to be complete. Is the possibility of completing them within human cognitive capacity? We never know. If we agree that we're organisms, if we're mysterians in that sense, yes, we're organisms, not angels, then could be, could not be. So, I don't see how to pursue the question any further for these reasons.

诺姆·乔姆斯基: 也许是关于“感知能力”的理论。我们能发展出关于感知能力的理论吗？嗯，现在我们又回到了之前的问题。这是否在……在一定程度上，当然可以。我们有很多这样的理论。它们是完备的吗？不，它们不完备。但我们当然也预料到理论是不完备的。将其完备化的可能性是否在人类的认知能力之内？我们永远不知道。如果我们同意我们是生物体，如果我们是那个意义上的“新神秘主义者”，是的，我们是生物体，不是天使，那么就有可能，也可能没有。所以，基于这些原因，我看不出如何进一步探讨这个问题。

[卡尔·弗里斯顿] 机器（即有感知的造物）会与我们交谈吗？

Curt Jaimungal: Question number four from Professor Karl Friston. Do you think machines will ever talk to each other or us? And let's define talk as posing questions and answers in the service of resolving uncertainty about the world or one another. What I'm getting at here is whether any sentient artifact can communicate with another sentient artifact, or whether two artifacts have to share a common ground or narrative that inherits from them having the same kind of phenotype and experiences.

Curt Jaimungal: 第四个问题来自卡尔·弗里斯顿教授。您认为机器会与彼此或与我们交谈吗？这里我们将“交谈”定义为：为了解决关于世界或彼此的不确定性而提出问题和给出答案。我想要探讨的是，任何有感知的造物都能与另一个有感知的造物交流吗？还是说，两个造物必须共享某种共同基础或叙事，而这种基础或叙事源自它们拥有同种的表现型（phenotype：生物体可观察的性状）和经历？

Noam Chomsky: Let's take trees again. There's sentient natural artifacts. They share a common ground and narrative in some sense, which they inherit, and they do interact with one another. So, there are natural objects that have this property. Well, let's go back to machines.

诺姆·乔姆斯基: 让我们再以树木为例。它们是有感知的自然造物。在某种意义上，它们共享着一种继承而来的共同基础和叙事，并且它们确实相互作用。所以，存在拥有这种属性的自然物体。好了，让我们回到机器上。

Noam Chomsky: First of all, remember what we all know. When in the framework that we're discussing, artificial intelligence, when we talk about machines, we don't mean the physical object. We mean the program. Alan Turing initiated the contemporary field of artificial intelligence with his famous 1950 paper on, Can Machines Think? He didn't mean, does a computer think? He meant, is there a program that thinks? He regarded that question, as you recall, as being too meaningless to deserve discussion, because we had no clear concept of thinking. He proposed some alternative concept, his imitation game.

诺姆·乔姆斯基: 首先，记住我们都知道的一点。在我们正在讨论的人工智能框架下，当我们谈论“机器”时，我们指的不是物理对象，而是程序。阿兰·图灵用他 1950 年的著名论文《机器能思考吗？》开创了当代人工智能领域。他的意思不是“计算机能思考吗？”，而是“是否存在一个能够思考的程序？”。你应该记得，他认为这个问题本身太没有意义，不值得讨论，因为我们对“思考”没有清晰的概念。他提出了一个替代概念，即他的“模仿游戏”。

Noam Chomsky: So, when we ask whether machines, sentient artifacts, will talk to each other, we're asking, are there programs that can interact with each other? I suppose so. Why not? If trees can do it, we can construct a program that can do it. They have to share some kind of common ground, or they won't be able to interact, but they do have experiences. They're affected by the environment. Same type of phenotype. Yes, in some sense, there's going to be something common about them.

诺姆·乔姆斯基: 所以，当我们问机器，即有感知的造物，是否会彼此交谈时，我们实际上是在问，是否存在可以相互作用的程序？我想是的。为什么不呢？如果树木能做到，我们就能构建一个能做到的程序。它们必须共享某种共同基础，否则无法互动，但它们确实有“经验”。它们受到环境的影响。同种表现型。是的，在某种意义上，它们之间必然会有某些共同之处。

Noam Chomsky: So, it seems to me, as far as I understand it, maybe I'm missing something. The question seems to me to be left at that level. Can we construct programs which will pose and answer questions? They won't resolve uncertainty about the world, because the programs have no concept of certainty about the world. It just doesn't arise for programs. People using the programs might find that they resolve some uncertainty about the world, just as the programs for protein folding did resolve some uncertainty about the world.

诺姆·乔姆斯基: 所以，就我所理解的而言，也许我遗漏了什么，这个问题似乎就停留在那个层面了。我们能构建出会提出和回答问题的程序吗？它们不会解决关于世界的不确定性，因为程序本身没有关于世界的“确定性”概念。这个概念对程序来说根本不存在。使用这些程序的人可能会发现它们解决了关于世界的某些不确定性，就像蛋白质折叠程序确实解决了关于世界的某些不确定性一样。

Noam Chomsky: Or, say, a theorem-proving program, which, say, used brute force to prove theorems, starting with the axioms, running through all possible proofs from shortest to longer, and you'll finally get a proof if it's a theorem. Well, that would relieve uncertainty about mathematics, but the program wouldn't know anything about it. We know about it. The next question, number five.

诺姆·乔姆斯基: 或者，比如说，一个定理证明程序，它用穷举法来证明定理，从公理出发，遍历所有可能的证明路径，从短到长，如果它是一个定理，最终你会得到一个证明。这会消除关于数学的不确定性，但程序对此一无所知。是我们知道了。下一个问题，第五个。

[蒂姆·莫德林] 古德曼的新归纳之谜在历史上是否准确？

Curt Jaimungal: Professor Tim Maudlin, he said, because I emailed him and asked him if he had any questions for you, actually, I just mentioned Chomsky in my graduate class. I said, because I heard this, that Chomsky was inspired to postulate universal grammar because of Goodman's new riddle of induction. That is, because that riddle that Chomsky thought there must be some a priori constraint on natural languages in order that they're learnable. I'm curious whether this is historically accurate, and if it is, do you still think universal grammar is the right way to express this constraint?

Curt Jaimungal: 蒂姆·莫德林教授说——我给他发邮件问他是否有问题想问您——他说：“实际上，我刚在我的研究生课上提到了乔姆斯基。我说，因为我听说，乔姆斯基是受了古德曼新归纳之谜（Goodman's new riddle of induction：一个哲学难题，挑战了我们对归纳推理和预测的理解）的启发，才提出了普遍语法（Universal Grammar：乔姆斯基提出的理论，认为所有人类语言共享一个潜在的、先天的语法结构）的假设。也就是说，正是因为这个谜题，乔姆斯基才认为自然语言必须存在某种先验的约束，才能保证其可学习性。我很好奇这在历史上是否准确？如果准确，您现在是否仍然认为普遍语法是表达这种约束的正确方式？”

Noam Chomsky: Well, I assume you know what Goodman's new riddle is. Put it very simply, he was asking which kinds of predicates are projectable in the sense that they fit into law-like statements, and we can use them to what's coming next. The famous example that he gave was the predicate grue, which is defined as meaning green before time t, where t has come time in the future, blue after time t. Then take a look at all the emeralds, where before time t, they're all green, they're all grue, but depending on whether we project green or grue, we'll make a different prediction for after time t. Then Goodman argued that grue is not projectable and green is projectable.

诺姆·乔姆斯基: 嗯，我假设你知道古德曼的新归纳之谜是什么。简单来说，他在问哪一类谓词是“可投射的”，即它们可以被纳入类似定律的陈述中，并用来预测未来。他给出的著名例子是谓词“grue”（绿蓝），其定义为在未来某个时间点 t 之前是绿色，在时间点 t 之后是蓝色。然后观察所有的祖母绿，在时间点 t 之前，它们既是绿色的，也是“绿蓝”的。但根据我们投射的是“绿色”还是“绿蓝”，我们对时间点 t 之后的预测就会不同。古德曼接着论证说，“绿蓝”是不可投射的，而“绿色”是可投射的。

Noam Chomsky: Actually, I was a student of his when he was working this out and discussing it in courses. That's basically...to answer the question, it had nothing to do with my own views on universal grammar. I frankly did not think that his explanation about the distinction of...he gave an argument to try to show that green was projectable and grue wasn't. He was never convinced that the argument went through.

诺姆·乔姆斯基: 实际上，当他在课程中研究和讨论这个问题时，我还是他的学生。基本上……回答这个问题，它与我自己的普遍语法观点毫无关系。坦白说，我并不认为他关于这个区别的解释……他给出了一个论证，试图证明“绿色”是可投射的而“绿蓝”不是。他自己也从未确信这个论证是成立的。

Noam Chomsky: I was, of course, familiar with Hume's argument, which Goodman discusses in his book, but not quite accurately. Hume's famous argument about induction is, as Hume put it, you have to assume that it's carried on because of what he called animal instinct. We have an instinct as animals that leads us to accept inductive reasoning in certain cases, not others. In a way, he spelled it out more carefully with some predicates and not others. And yes, I think that's the only answer to carrying out induction to the extent that we do some animal instinct.

诺姆·乔姆斯基: 我当然熟悉休谟的论证，古德曼在他的书中也讨论了，但不够准确。休谟关于归纳的著名论证是，如休谟所说，你必须假设归纳法之所以可行，是源于他所谓的“动物本能”。作为动物，我们有一种本能，引导我们在某些情况下接受归纳推理，而在另一些情况下则不然。在某种程度上，他用某些谓词而非其他谓词更仔细地阐述了这一点。是的，我认为对于我们进行归纳推理的程度而言，唯一的答案就是某种动物本能。

Noam Chomsky: But I don't think you need that insight in order to postulate universal grammar. It's just that there's no alternative. I mean, there was a time when I was a student, in fact, back in the 1940s, 1950s. It was assumed that language acquisition is a matter of training and habit. According to the leading American linguist, Leonard Bloomfield, philosophy of language took essentially the same view. Perhaps most influential Anglo-American philosopher of language in mind. For him, language was a complex of dispositions to respond, established in his picture by Scenario and Operative Condition. Those views fall by the wayside as soon as you take a look at the acquisition of language. It's nothing like the acquisition of languages based on what Hume called an animal instinct.

诺姆·乔姆斯基: 但我认为你不需要那个洞见来提出普遍语法的假设。只是因为没有其他选择。我的意思是，在我还是学生的时代，实际上是 20 世纪 40 年代、50 年代，人们普遍认为语言习得是训练和习惯的问题。根据美国顶尖语言学家伦纳德·布龙菲尔德的说法，语言哲学也持基本相同的观点。也许是最有影响力的英美语言哲学家也是如此。对他来说，语言是一系列反应倾向的复合体，在他的图景中由情景和操作性条件建立。但只要你看一眼语言习得的过程，这些观点就站不住脚了。它完全不同于基于休谟所谓的“动物本能”的语言习得。

Noam Chomsky: There's some capacity that a human infant has that a chimpanzee doesn't have, which enables the human infant to quickly acquire a language on the basis of scattered data, which for the chimp is just noise. That's got to be because of some animal instinct, if you like, some innate capacity. That's about as elementary as you can be. But if there's an innate capacity, there should be a theory of that innate capacity. The term universal grammar in the modern sense is just the name for that theory. So postulating universal grammar is elementary rationality. There's no alternative. Is it the right way to proceed expressing the constraints? I can't think what alternative there is. If you believe that there's something innate that distinguishes the infant from its pet chimp, which is certainly true, then this is what you have to postulate.

诺姆·乔姆斯基: 人类婴儿拥有一种黑猩猩所没有的能力，这种能力使婴儿能够基于零散的数据迅速习得一门语言，而对黑猩猩来说这些数据只是噪音。这必定是由于某种“动物本能”，如果你愿意这么说，即某种先天的能力。这是最基本的道理了。但如果存在一种先天的能力，就应该有一个关于这种先天能力的理论。现代意义上的“普遍语法”一词，正是这个理论的名称。所以，假设普遍语法的存在是基本的理性行为，没有其他选择。这是表达这种约束的正确方式吗？我想不出还有什么其他选择。如果你相信存在某种先天的东西，将婴儿与他的宠物黑猩猩区分开来——这肯定是事实——那么这就是你必须做的假设。

[菲利普·戈夫] 物质的内在本质与泛心论

Curt Jaimungal: Okay, question 17. So one, seven, 17. This is from Professor Philip Goff. We both agree with Russell that physics tells us what matter does, but fails to reveal its intrinsic nature. Therefore, the only thing we know directly about the intrinsic nature of matter is that some of it, i.e. the stuff in brains, is experiential. So either the matter outside of brains also has an experiential intrinsic nature, or it has some other property, completely unknown nature. Surely the former option, i.e. panpsychism, is to be preferred on grounds of simplicity. What do you make of this argument?

Curt Jaimungal: 好的，第 17 个问题。这是来自菲利普·戈夫教授。我们都同意罗素的观点，即物理学告诉我们物质做什么，但未能揭示其内在本质。因此，我们唯一直接了解的关于物质内在本质的事情是，其中一部分，即大脑中的物质，是具有经验性的。所以，要么大脑之外的物质也具有经验性的内在本质，要么它具有某种其他完全未知的性质。基于简单性原则，前者，也就是泛心论（Panpsychism：认为意识、心智或灵魂是宇宙基本且普遍存在的特征的哲学观点），显然是更可取的。您如何看待这个论证？

Noam Chomsky: Well, simplicity depends on the simplicity of a theory. It depends on the evidence that you're considering. Consider different evidence, different theories will be simpler or more difficult. So let's accept the basic framework generally accepted in contemporary philosophy within which this question arises. It goes back to a distinction, goes back to David Chalmers' distinction between the easy problems of consciousness and the hard problem.

诺姆·乔姆斯基: 嗯，简单性取决于理论的简单性。它取决于你所考虑的证据。考虑不同的证据，不同的理论就会显得或简单或复杂。那么，让我们接受当代哲学中普遍接受的、该问题赖以提出的基本框架。这要追溯到一个区别，追溯到大卫·查默斯关于意识的“简单问题”和“困难问题”的区别。

Noam Chomsky: The hard problem is supposed to be, what is it like to see a sunset? That's the hard problem. The easy problem is, what are the neurological, genetic, other properties that enter into being conscious of the sunset? I've never been particularly happy with that formulation, because it's not very clear to me what the hard problem is. If I'm asked, what is it like to see the sunset that you're looking at, I can give a very detailed answer. Red spot up here makes me feel happy, and go on and fit that in extensive detail. So there's no problem in describing what it's like to see that sunset. But if I'm asked, what is it like to see a sunset, I don't think that has an answer. At least I can't imagine what the answer is.

诺姆·乔姆斯基: 所谓的“困难问题”是：看到日落是什么感觉？这就是困难问题。而“简单问题”是：构成“意识到日落”这一过程的神经学、遗传学及其他属性是什么？我一直对这种提法不太满意，因为我不太清楚“困难问题”到底是什么。如果有人问我，“看到你正在看的那个日落是什么感觉”，我可以给出一个非常详细的回答。这里有个红点，让我感到快乐，然后我可以继续详细描述。所以描述看到那个特定日落的感觉是没有问题的。但如果有人问我，“看到日落是什么感觉”，我不认为这个问题有答案。至少我无法想象答案会是什么。

Noam Chomsky: And if a question doesn't have a formulable answer, it's not an authentic question. It's just a linguistic expression in interrogative form. It's kind of on a par with why do things happen? It has the form of a question, but it's not a real question. There's no conceivable answers. So the hard problem, as far as I can see, doesn't really exist. We're left with the easy problem, which is, of course, very far from easy. What's going on when I see the sunset? And I'm conscious of it, what I can describe in detail, what's going on? It's within that context that the question is raised.

诺姆·乔姆斯基: 如果一个问题没有一个可表述的答案，它就不是一个真正的问题。它只是一个疑问形式的语言表达。这有点像问“为什么事情会发生？”。它有问题的形式，但不是一个真正的问题。没有可以想见的答案。所以，在我看来，“困难问题”其实并不存在。我们剩下的是“简单问题”，当然，它远非简单。当我看到日落时，发生了什么？我意识到了它，我能详细描述它，到底发生了什么？正是在这个背景下，这个问题才被提出来。

Noam Chomsky: Now, one answer to it is all of nature has the property of consciousness. That's one possibility. Another possibility, another answer that is often given is there is no consciousness. It's an illusion stuck with either of those answers. I think we can expect to find an answer which will say that consciousness is a property of humans, other animals, but not the table on which my computer is resting. Is that a less simple answer than panpsychism? Depends on what our evidence is. So suppose someday we get real answers to the so-called easy problem.

诺姆·乔姆斯基: 现在，一个答案是，整个自然界都具有意识的属性。这是一种可能性。另一种可能性，一个经常被给出的答案是，根本没有意识，它是一种幻觉。如果在这两个答案中选择，我认为我们可以期望找到一个答案，它会说意识是人类、其他动物的属性，而不是我电脑所在的这张桌子的属性。这是否比泛心论更不简单？这取决于我们的证据是什么。假设有一天我们对所谓的“简单问题”得到了真正的答案。

Noam Chomsky: Suppose we find the neurological properties, genetic properties that enter into accounting for my moments of consciousness, and for what is not what I know without consciousness, like most of my mental activities. Suppose we find a way of distinguishing those and distinguishing humans from tables. Well, then the simplest argument, I think, will take into account all of these discoveries and information. And then I don't think panpsychism any longer is the simplest answer. It's an answer that simply postulates something for which there's no evidence that particles have this property, even though we already have evidence that only certain kinds of organization of whatever the world is constituted of, have this property. So simplicity depends on what you're accounting for. The simplicity of a theory depends on what it's dealing with. If we take the full range of evidence that we hope to have by making progress with the so-called easy problem, then I think we have simpler theories.

诺姆·乔姆斯基: 假设我们找到了能够解释我意识瞬间的神经学属性、遗传学属性，以及解释我无意识所知（像我大部分心理活动）的属性。假设我们找到了区分这两者以及区分人类和桌子的方法。那么，我认为，最简单的论证将会把所有这些发现和信息都考虑进去。到那时，我就不认为泛心论还是最简单的答案了。它只是一个简单地假设了某样东西的答案，而没有任何证据表明基本粒子具有这种属性，尽管我们已经有证据表明，只有特定组织形式的（无论世界由什么构成）才具有这种属性。所以，简单性取决于你要解释什么。一个理论的简单性取决于它处理的对象。如果我们把通过在所谓“简单问题”上取得进展而希望获得的全部证据都考虑进来，那么我想我们会得到更简单的理论。

[匿名] 您对说谎者悖论或渡鸦悖论有何解答？

Curt Jaimungal: Okay. 24. Do you have a resolution to the liar's paradox?

Curt Jaimungal: 好的。24. 您对说谎者悖论（Liar's Paradox：一个经典的逻辑悖论，如“这句话是假的”）有解决方案吗？

Noam Chomsky: No. Nothing to say about the paradoxes. That is all new. A lot has been said about them, but I have nothing to add.

诺姆·乔姆斯基: 没有。关于这些悖论我没什么可说的。这都是全新的领域。已经有很多关于它们的论述，但我没有什么可以补充的。

Curt Jaimungal: What about the raven paradox?

Curt Jaimungal: 那么渡鸦悖论（Raven Paradox：关于“所有渡鸦都是黑色的”这一陈述的归纳证据的悖论）呢？

Noam Chomsky: Same story. Sample's paradox. Again, I think they're reasonable answers, but I have nothing to add to them.

诺姆·乔姆斯基: 一样的。辛普森悖论。再说一次，我认为有合理的答案，但我没什么可补充的。

[匿名] 语言的公理与词语的意义

Curt Jaimungal: Okay. 26. So the next one. Each word in a dictionary is defined in terms of other words in the dictionary, which indicates that there are some fundamental concepts which have to be understood without reference to other words, like the axioms of a language, or that somehow the meaning of the words are inextricably tied to the relation that each word has to others. Are these the only two options?

Curt Jaimungal: 好的。26. 下一个问题。字典里的每个词都是用字典里的其他词来定义的，这表明要么存在一些无需参照其他词就能被理解的基本概念，就像语言的公理一样；要么词语的意义以某种方式与每个词同其他词的关系密不可分。这是仅有的两种选择吗？

Noam Chomsky: Well, we have to ask ourselves, are we talking about human language, or are we talking about formal systems that somebody constructed, like say, metamathematics or quantum theory? There are different answers in these two cases. In the case of a formal system that someone constructs, yes, there have to be some fundamental concepts in terms of which others are defined. That's true for metamathematics. It's true for physics to the extent that it's formalized, and so on. So then, in that context, yes, it's correct.

诺姆·乔姆斯基: 嗯，我们必须问自己，我们谈论的是人类语言，还是某人构建的形式系统，比如元数学或量子理论？在这两种情况下，答案是不同的。在某人构建的形式系统的情况下，是的，必须有一些基本概念，其他概念是根据它们来定义的。这对于元数学是真的。对于被形式化的物理学等也是如此。所以在那个背景下，是的，这是正确的。

Noam Chomsky: What about human language? Well, in human language, the question doesn't arise. Because words in the dictionary don't have definitions. When you look up the dictionary, say, most detailed dictionary, and find the OED, take a look at the definition of word there, it's nowhere near a definition. It's an explanation which gives a number of hints that enable you, with your rich intuitive understanding of language, to guess what the word means. But you're doing most of the work, your mind, the dictionary is doing almost nothing.

诺姆·乔姆斯基: 那么人类语言呢？在人类语言中，这个问题根本不会出现。因为字典里的词没有“定义”。当你查阅字典，即便是最详尽的字典，比如《牛津英语词典》（OED），看看那里对词语的定义，它远非一个严格的定义。它是一种解释，给出一些线索，让你凭借自己丰富的语言直觉去猜测词语的意思。但大部分工作是你，是你的心智在做，字典几乎什么也没做。

Noam Chomsky: You try to define the simplest words you can think of, book, table, river, cat, whatever you want, has a rich and complex meaning. One which, incidentally, is known to infants without experience, which poses another fundamental problem. How are infants able to acquire the rich and complex meaning of the most elementary words with only very few presentations, as has been shown by experimental studies? Serious problem. Go back to the first question, one of those deep ones, which we have no real answer to. But there are no definitions.

诺姆·乔姆斯基: 你试着去定义你能想到的最简单的词：书、桌子、河流、猫，任何你想要的词，它们都有着丰富而复杂的含义。顺便说一句，这些含义是毫无经验的婴儿就知道的，这又提出了另一个基本问题。实验研究已经表明，婴儿是如何仅凭极少数的接触就能获得最基本词汇的丰富而复杂的意义的？这是一个严肃的问题。回到第一个问题，那是我们没有真正答案的深层问题之一。但是，（在人类语言中）没有定义。

Noam Chomsky: There are only definitions for technical terms that are made up. So, if somebody introduces the term, tensor, let's say, they get the definition of it. But these are not the kinds of words that are just acquired in ordinary life. Those words don't have definitions. They are interrelated in all sorts of ways. But there's no particular reason to believe that some of them, I mean, to some extent, some are going to be more fundamental than others. But we have no reason to believe that there's some finite set, which is the basis, like an axiom system.

诺姆·乔姆斯基: 只有那些被创造出来的技术术语才有定义。所以，如果有人引入“张量”（tensor）这个术语，他们就会给出它的定义。但这些词不是日常生活中习得的那种词。那些词没有定义。它们以各种方式相互关联。但是，没有特别的理由相信其中一些……我的意思是，在某种程度上，有些词会比其他词更基本。但我们没有理由相信存在某个有限的集合，作为基础，就像一个公理系统一样。

Noam Chomsky: I might add here, just for those of you who are interested, that there is very interesting work dealing with the question of how we use our lexical items. What words are not really the right units, but the minimal meaning-bearing items. How do we use them? There's two different theories about this. One is that, let's call them words for simplicity. One is that the words are stored. We have a store of words in our mind. We access the words when we're trying to understand the sentence or the people's response. That's pretty much the standard theory.

诺姆·乔姆斯基: 在这里我或许可以补充一点，给感兴趣的人听。关于我们如何使用我们的词项，有一些非常有趣的研究工作。词语其实不是正确的单位，应该是最小的意义承载单元。我们如何使用它们？关于此有两种不同的理论。一种理论是，为简单起见我们称之为“词语”，词语是被存储的。我们头脑中有一个词库。当我们要理解句子或他人的回应时，我们就去提取这些词。这基本上是标准理论。

Noam Chomsky: There's an alternative, which is quite interesting, developed mostly by Alec Marantz, a linguist, a neuroscientist at NYU. He's argued that we don't store the words, we store the rules for creating words. And whether a word is actually used in the lexicon or not is kind of irrelevant. And he has quite interesting arguments, neurological, psycholinguistic, and others, arguing that the storage is actually just the generative rules, not the words themselves. So a possible word, which doesn't happen to be in our lexicon, say Blake, is treated the same way as one that isn't. That's quite an interesting scientific question.

诺姆·乔姆斯基: 还有一种替代理论，非常有趣，主要由纽约大学的语言学家、神经科学家亚历克·马兰茨提出。他认为我们不存储词语，而是存储创造词语的规则。一个词是否真正在词库中使用过，在某种程度上是无关紧要的。他提出了一些非常有趣的论证，涉及神经学、心理语言学等领域，认为我们实际存储的只是生成规则，而不是词语本身。所以一个可能的词，比如“blake”（一个虚构词），即使它碰巧不在我们的词库里，其处理方式也和在词库里的词一样。这是一个相当有趣的科学问题。

[匿名] 语言与延伸表现型之间有何关联？

Curt Jaimungal: Dawkins introduced the concept of extended phenotype. Is there any relation between language and extended phenotype?

Curt Jaimungal: 道金斯引入了延伸表现型（Extended Phenotype：指基因不仅影响其所在生物体的性状，还能通过生物体的行为影响外部环境，如河狸筑坝）的概念。语言与延伸表现型之间有何关联？

Noam Chomsky: It depends what we mean by language. Language is one of those informal terms of general usage, as much too imprecise to discuss. So like anything else in the sciences, when linguists try to discuss and understand something about language, or psychologists, or neurologists, they give a technical definition of language. So for example, if a physicist is asked, what's energy? What's work? What's spin? They're not going to give you the definition that's in ordinary language. There are no definitions. What they'll do is define the technical term, which is different from the one in ordinary language. And the same is true here.

诺姆·乔姆斯基: 这取决于我们所说的“语言”是什么意思。“语言”是日常用语中的一个非正式术语，太不精确以至于无法讨论。因此，就像科学中的任何其他事物一样，当语言学家、心理学家或神经学家试图讨论和理解关于语言的事情时，他们会给“语言”一个技术性的定义。例如，如果你问一个物理学家，什么是能量？什么是功？什么是自旋？他们不会给你日常语言中的定义。没有那种定义。他们会做的是定义这个技术术语，它不同于日常语言中的用法。这里也是一样。

Noam Chomsky: So we have to ask, what do we mean by language? Well, if we mean the technical notion that I would call I-language, internal language, then extended phenotypes have nothing to do with it. If we use the word language in a loose, figurative, common sense, then maybe extended phenotypes have something to do with it.

诺姆·乔姆斯基: 所以我们必须问，我们说的“语言”是什么意思？如果我们指的是我称之为“I-语言”（I-language：内在语言，指个体头脑中心理和生理上的语言知识系统），那么延伸表现型与它无关。如果我们以一种宽松的、比喻性的、常识性的方式使用“语言”这个词，那么也许延伸表现型与它有些关系。

[匿名] 我们对图灵测试的执着是否是行为主义的遗留影响？

Curt Jaimungal: Is our obsession with the Turing test a holdover of the influence of behaviorism? Does a behavior, however convincing, ever decisively imply consciousness? Should we be using some other test other than the Turing test?

Curt Jaimungal: 我们对图灵测试（Turing Test：由图灵提出的一种测试机器是否能表现出与人等价或无法区分的智能的方法）的执着，是否是行为主义（Behaviorism：一种心理学流派，主张心理学只应研究可观察的行为）影响的遗留物？一种行为，无论多么有说服力，是否能决定性地意味着意识的存在？我们是否应该使用图灵测试之外的其他测试？

Noam Chomsky: Well, first of all, I might mention as a historical fact that the Turing test actually goes back to the 17th century, after Descartes' postulation of a second substance for things like the creative use of language. His associates, mainly Géraud de Cordemoy, took the obvious next step and said, can we discover experimental tests that will tell us whether a creature that looks like us has this property or not? Then he went through various things, which are kind of like Turing's imitation game. So that's basically the Turing test. But there it was for a fundamental scientific reason, to find out a property of an organism.

诺姆·乔姆斯基: 嗯，首先，我作为一个历史事实提一下，图灵测试实际上可以追溯到 17 世纪，在笛卡尔为像语言的创造性使用这样的事物假设了第二种实体之后。他的同事，主要是热罗·德·科德穆瓦，迈出了显而易见的下一步，问道：我们能否找到实验性的测试，来判断一个看起来像我们的生物是否具有这种属性？然后他列举了各种方法，有点像图灵的模仿游戏。所以那基本上就是图灵测试。但在当时，它有一个根本的科学目的：找出有机体的一种属性。

Noam Chomsky: As I said, that kind of disappeared with Newton. If we can test behaviors with the Turing's imitation game, does that imply consciousness? No, not. You could make up a program that can do it, which has no consciousness.

诺姆·乔姆斯基: 正如我所说，那种探究随着牛顿的出现而消失了。如果我们能用图灵的模仿游戏来测试行为，这是否意味着意识？不，不是。你可以编写一个能通过测试的程序，但它没有任何意识。

Noam Chomsky: Should we be using some other test? Depends what we're trying to achieve. I don't see any particular purpose in trying to understand, to determine whether a machine is doing something, machine meaning program is doing something like thinking. But if we regard it as a significant question, we use whatever test fits our particular definition of thinking.

诺姆·乔姆斯基: 我们应该使用其他测试吗？这取决于我们想实现什么目标。我看不出试图去理解、去判断一个机器——也就是程序——是否在做类似思考的事情有什么特别的目的。但如果我们认为这是一个重要的问题，那么我们就应该使用任何符合我们对“思考”的特定定义的测试。

[匿名] 不明确的数学陈述和物理概念能有意义吗？

Curt Jaimungal: Okay, question number 30. With colorless green ideas sleep furiously, you showed that you can have syntactic sense without semantic sense. Outside of idioms and common phrases, can something make semantic sense without being syntactic? So if so, can this apply to math or physics? That is, it seems like in math and physics, as soon as you don't have syntactic sense, then the whole equation, the mathematical sentence, is not well posed, and thus is meaningless. Can you imagine cases where we have ill-defined mathematical statements that have meaning?

Curt Jaimungal: 好的，第 30 个问题。您用“无色的绿色思想狂怒地睡着”这个例子表明，一个句子可以有句法意义而没有语义意义。那么，在习语和常用短语之外，有没有可能一个东西有语义意义而没有句法意义？如果可以，这能应用于数学或物理学吗？也就是说，在数学和物理学中，似乎一旦没有了句法意义，整个方程、整个数学句子就是不适定的，从而没有意义。您能想象存在定义不明确但却有意义的数学陈述的案例吗？

Noam Chomsky: Many examples in the history of mathematics. So take, for example, there was, take just something as simple as arithmetic or geometry. I mean, arithmetic and geometry were studied highly productively, literally for millennia, before the concepts were well-defined. Now, there wasn't a well-defined, there weren't good, serious in the modern sense, definition of basic concepts of arithmetic until about the 1860s. Dedekind and Peano gave well-defined in the concepts of arithmetic for some. Geometry wasn't formalized until Hilbert, around 1900.

诺姆·乔姆斯基: 在数学史上有很多例子。举个例子，就拿算术或几何这样简单的东西来说。我的意思是，在这些概念被明确定义之前，算术和几何已经被卓有成效地研究了数千年。现代意义上对算术基本概念的明确、严谨的定义直到大约 19 世纪 60 年代才出现。戴德金和皮亚诺为一些算术概念给出了明确的定义。而几何学直到 1900 年左右才由希尔伯特完成形式化。

Noam Chomsky: But of course, it had been used forever, productively and effectively. And it's not until the formalization that they're well-defined. Kind of a famous case is the concept of limit. If you go back to the 17th century, calculus, Newton, Leibniz's calculus, were based on some intuitive concept of limit. In fact, there's a long debate in the history of mathematics as to whether Newton's proofs actually involve an equivocation in the use of zero.

诺姆·乔姆斯基: 但当然，它一直被富有成效地、有效地使用着。直到形式化之后，它们才被明确定义。一个著名的例子是“极限”的概念。如果你回到 17 世纪的微积分，牛顿、莱布尼茨的微积分都是基于某种直观的极限概念。事实上，数学史上有一个长期的争论，即牛顿的证明是否在“零”的使用上存在歧义。

Noam Chomsky: Is it the case that in one line, Newton used zero to mean zero, and another line, he used it to mean as small as possible, he had no theory of limit. Well, long debate about that. But what happened apparently, histories of mathematics say, I don't know much about this myself, I'm just relying on the history, that British mathematicians worked hard on trying to sharpen and clarify the notions, which actually wasn't really done until the mid-19th century. Continental mathematicians basically ignored it. Gauss, the other, Euler, the other great mathematicians, and they developed most of classical analysis based on ill-defined notions, while the British mathematicians for the most part were stymied because they didn't really have a clear notion. Well, so yes, you can certainly imagine such cases. In fact, there are plenty of them.

诺姆·乔姆斯基: 是不是在某一行中，牛顿用零表示零，而在另一行中，他用它表示尽可能小？他没有极限理论。关于这一点有很长的争论。但显然发生的是——据数学史记载，我自己对此了解不多，只是依赖史料——英国数学家努力试图使这些概念更精确、更清晰，这项工作直到 19 世纪中叶才真正完成。而欧洲大陆的数学家基本上忽略了这个问题。高斯、欧拉以及其他伟大的数学家，他们在定义不明确的概念之上发展了大部分经典分析学，而大多数英国数学家则因为没有清晰的概念而停滞不前。所以，是的，你当然可以想象这样的案例。事实上，有很多这样的例子。

[诺曼·怀尔德伯格] 是否有语言学证据支持人类普遍的先天数学观念？

Curt Jaimungal: There were two additional questions, one anonymous and the other by Professor Norman Wildberger, which we didn't get to, so we got to over email. Professor Norman Wildberger says, here's a question I would like to ask Professor Chomsky. Is there any linguistic evidence for universal, or almost universal, innate mathematical notions in humans? Maybe notions is not the right word here. If not, how does one frame a meaningful question in this direction?

Curt Jaimungal: 还有两个额外的问题，一个是匿名的，另一个来自诺曼·怀尔德伯格教授，我们没来得及在访谈中讨论，所以通过邮件进行了交流。诺曼·怀尔德伯格教授说：“我想向乔姆斯基教授提一个问题。是否存在语言学上的证据，表明人类具有普遍的、或近乎普遍的先天数学观念？也许‘观念’这个词在这里不完全准确。如果没有，我们如何在这个方向上构建一个有意义的问题？”

Noam Chomsky (via computer voice): Definitely. Knowledge of arithmetic is a classic case, the topic of a famous debate between Darwin and Wallace. Both recognized that knowledge of arithmetic is universal, though it plainly couldn't have been selected. Wallace suggested some new force operating in evolution. Darwin was unwilling to accept that. It plainly must be innate. There's no way to learn that the integers go on forever. There's some reason to believe that it may have the same origins as the language faculty. The simplest language would have a single lexical item and the simplest rule, internal merge, that yields the successor function and the rudiments of addition. From there it's not hard to go on to addition. So whatever rewiring of the brain that yielded recursive enumeration, apparently unique to humans, could have provided the basis for both language and arithmetic.

诺姆·乔姆斯基 (通过电脑语音): 当然有。算术知识就是一个典型的例子，也是达尔文和华莱士之间著名辩论的主题。两人都认识到算术知识是普遍的，尽管它显然不可能是通过自然选择形成的。华莱士提出，演化中存在某种新的力量在起作用。达尔文不愿接受这一点。算术知识显然必须是先天的。你无法通过学习得知整数是无限的。有理由相信，它可能与语言能力同源。最简单的语言会有一个词项和最简单的规则——“内向合并”（Internal Merge），这能产生后继函数和加法的雏形。从那里发展到加法并不困难。因此，无论是什么样的大脑重接产生了递归枚举能力——这显然是人类独有的——它都可能为语言和算术提供了共同的基础。

Noam Chomsky (via computer voice): As for mathematics generally, that's another matter. There's a famous comment of Kronecker's that God created the natural numbers and man created the rest, which translates as arithmetic is innate but human creativity, whatever that is, created the set theory and on from there.

诺姆·乔姆斯基 (通过电脑语音): 至于一般的数学，那就是另一回事了。克罗内克有一句名言：“上帝创造了自然数，其余都是人做的工作”，这可以解读为：算术是先天的，但人类的创造力——无论那是什么——创造了集合论以及之后的一切。

[匿名] “语法”一词的广延性

Curt Jaimungal: And the second question is how far does grammar extend? People talk about the grammar of narratives and stories or the grammar of video games, such as common tropes and patterns ingrained in players that are unconscious and widely used across almost each video game. Is this a generalization of grammar that bears only a passing resemblance to the linguistic concept of grammar? Or is there some common neurological, even psychological, basis between these types of quote unquote grammars?

Curt Jaimungal: 第二个问题是，“语法”能延伸到多远？人们谈论叙事和故事的语法，或者电子游戏的语法，比如那些根植于玩家潜意识中、在几乎所有游戏中都广泛使用的常见套路和模式。这是否只是对“语法”一词的泛化，与语言学中的语法概念只有表面的相似之处？还是说，在这些所谓的“语法”之间，存在某种共同的神经学，甚至是心理学基础？

Noam Chomsky (via computer voice): There's no clear answer. We can extend terms to broader uses like quote unquote the language of the stars. It's not arbitrary, but there are no clear principles. The term grammar, without any such metaphorical extensions, is itself much too imprecise to use in studying language. So it's given a technical meaning, just as is done in the sciences generally.

诺姆·乔姆斯基 (通过电脑语音): 没有明确的答案。我们可以将术语扩展到更广泛的用途，比如所谓的“星辰的语言”。这种扩展不是任意的，但没有明确的原则。“语法”这个词本身，即便不进行任何此类隐喻性扩展，也过于不精确，无法用于研究语言。因此，就像在一般科学中所做的那样，它被赋予了一个技术性的含义。

Curt Jaimungal: I'm sorry, but I afraid I have a appointment in two minutes that I have to go to. Thank you. Thank you, Professor.

Curt Jaimungal: 对不起，恐怕我两分钟后有个约会，必须得走了。谢谢您。谢谢您，教授。

Curt Jaimungal: The podcast is now finished. If you'd like to support conversations like this, then do consider going to patreon.com slash c u r t j a i m u n g a l. That is Curt Jaimungal. It's support from the patrons and from the sponsors that allow me to do this full-time. Every dollar helps tremendously. Thank you.

Curt Jaimungal: 播客到此结束。如果您想支持这样的对话，请考虑访问 patreon.com/curtjaimungal。也就是 Curt Jaimungal。正是来自 Patreon 会员和赞助商的支持，才让我能够全职做这件事。每一分钱都帮助巨大。谢谢大家。