#325 – 迈克尔·莱文：生物学、生命、外星生命、进化、胚胎发生与异种机器人

以下是与迈克尔·莱文的对话，他是我交谈过的最引人入胜且才华横溢的生物学家之一。他和他在塔夫茨大学的实验室致力于研究理解与控制生物系统中复杂模式形成的新方法。世界级人工智能研究员安德烈·卡帕西是最初向我介绍迈克尔·莱文工作的人。我提到这点是因为这两人让我意识到，生物学对人工智能有很多启示，而人工智能或许也能为生物学提供许多洞见。现在，快速用两秒钟介绍一下本期赞助商。

The following is a conversation with Michael Levin, one of the most fascinating and brilliant biologists I've ever talked to. He and his lab at Tufts University works on novel ways to understand and control complex pattern formation in biological systems. Andre Karpathy, a world class AI researcher, is the person who first introduced me to Michael Levin's work. I bring this up because these two people make me realize that biology has a lot to teach us about AI, and AI might have a lot to teach us about biology. And now, a quick two second mention of each sponsor.

详情请查看描述栏，这是支持本播客的最佳方式。我们有Henson剃须刀提供优质剃须体验，Eight Sleep负责电商业务，Element提供便携电解质，Insight Tracker用于生物追踪。朋友们，请明智选择。现在进入完整广告环节——我们绝不会在中途插播广告。

Check them out in the description. It's the best way to support this podcast. We got Henson Shaving for a great razor and shave, Eight Sleep for ecommerce, Element for on the go electrolytes, and Insight Tracker for biological tracking. Choose wisely, my friends. And now onto the full ad reads, never ads in the middle.

那种插播广告很糟糕。我尽量让这里的广告变得有趣，或许值得一听，主要是因为很多内容其实与赞助商无关，更像是受赞助商启发创作的。但如果你实在要跳过，请务必看看赞助商信息。正是因为他们，我才能制作这个播客，所以请以任何方式支持他们。

Those suck. I tried to make these ads here interesting, so they're worth listening to perhaps, mostly because a lot of them don't have anything to do with a sponsor. They're almost like inspired by the sponsor. But if you still must skip, please check out the sponsors. They're the reason I'm able to do this podcast, so please support them in any way you can.

我喜欢他们的产品，或许你也会喜欢。本节目由Henson剃须刀赞助播出，这家家族经营的航空航天制造商将精密工程带入你的剃须体验。'工程'这个词就足够打动我——虽然我差点要说这让我有点兴奋，但在广告里这么说太不合适了，我绝不会那样讲话。

I enjoy their stuff. Maybe you will too. This show is brought to you by Henson Shaving, a family owned aerospace manufacturer bringing precision engineering to your shaving experience. You had me at engineering. If there's any word a person can state to me that, you know, I was gonna say turns me on a little bit, but that would be very inappropriate to do an ad read, so I would never say the thing like that.

总之，无论是材料工程、机械工程还是土木工程，任何形式的工程都让我着迷。他们使用航空级CNC机床，因此能制造出刀片仅伸出0.0013英寸（比人类头发还薄）的金属剃须刀。正是这种精密工艺让你能获得安全、干净、顺滑的剃须体验。访问hensonshaving.com/lex选择剃须刀并使用代码lex，即可免费获得100片刀片。

But anyway, I find engineering just in all of its forms, materials engineering, mechanical engineering, civil engineering, all of it. I love it. They're using aerospace grade CNC machines and because of that are able to make metal razors that extend just point zero zero one three inches, which is less than the thickness of a human hair. And the precision of it is the reason you can have a safe, clean, smooth shave. Check them out at hensonshaving.com/lex to pick your razor and use code lex, and you'll get 100 free blades with your razor.

必须同时加入100片刀片套装和剃须刀才能享受折扣。本期节目还由Eight Sleep及其新款Pod 3床垫赞助播出。它是我快乐的源泉——小憩总能抚慰我心中的焦虑、不安、悲伤与忧郁。

You must add both the 100 blade pack and the razor for the discount to apply. This episode is also brought to you by Eight Sleep and its new pod three mattress. It is a source of happiness for me. Naps bring joy to my heart. They put to rest all the anxiety, the uncertainty, the sadness, the melancholy that I have on my heart.

让这些情绪逐渐消散，如同凤凰从午睡的灰烬中重生。凉爽的床面搭配温暖的毯子，简直完美。通常20-30分钟的小憩就能治愈所有烦恼。无论我对世界有多少不确定、恐惧或焦虑，甚至是不安全感，一次小憩就能化解，这或许就是生活馈赠的礼物。

They let that fade and the phoenix rises from the ashes of the nap. And a cool surface of the bed with a warm blanket, I just love it. It's usually gonna be about twenty to thirty minute nap, cures all ills. I can have so much uncertainty, so much fear, so much anxiety about the world, and just just a little nap or insecurity, all of it. A nap can cure it, and maybe that's a gift.

也许这是化学赐予我的礼物。我很幸运没有遭受那种可能让你持续多日、数周甚至数月的化学性抑郁。对我来说，小睡能治愈很多，这很美妙。我是说，睡眠真的非常重要。所以我强烈推荐八小时睡眠。

Maybe that's a chemical gift for me. I'm so fortunate not to suffer from sort of chemical depression that can hold you down for many days, weeks, or months. For me, a nap can cure so much, and it's beautiful. I mean, sleep is so so important. So I highly recommend eight Sleep.

正如我所说，它一直是我许多快乐的源泉。快去8sleep.com/lex看看并享受特别优惠。本期节目还由Element电解质饮料粉赞助，拼写是l-m-n-t。我喜欢他们的西瓜盐口味，但最近我也在尝试巧克力薄荷味——我对这个口味打了个问号。

It's been, like I said, a source of a lot of happiness for me. Check it out and get special savings when you go to 8sleep.com/lex. This episode is also brought to you by Element electrolyte drink mix, spelled l m n t. I like their watermelon salt one, but lately, I've also been consuming some of the chocolate mint. I have a question mark on that.

我不记得它本应是什么味道，但尝起来确实如此。虽然美味，但本质上完全不同——营养成份相同，但精神体验截然不同。西瓜盐口味更多是我想给水加点‘冲击力’，同时确保为我的低碳生活补充足够的钠钾镁。

I don't remember if that's what that is, but it's what it tastes like, And it's delicious, but it's a different it's a kind of a different thing. Nutritionally, it's the same thing, but spiritually, it's a different thing. So watermelon salt is more what I'm just trying to add a little, like, kick to my water. Right? And making sure I'm getting all the sodium potassium magnesium right for the for the low carb stuff I'm doing.

而巧克力口味则带有——我不愿说是甜味——更像是巧克力特有的复杂层次感。它更像是让我放松的饮品，就像一天结束时的甜点。购买任意产品即可获赠免费试用装。

But the chocolate one has a little I don't wanna say sweetness, but a little more complexity in the way chocolate does. Right? And so I feel like it's the thing that relaxes me a bit more. It's like a dessert at the end of the day. Anyway, get a sample pack for free with any purchase.

登录drinkelement.com/lex即可尝试。本期节目还由我使用的生物数据追踪服务InsideTracker赞助。他们提供多种方案，多数包含血液检测，通过机器学习算法结合你的血液数据、DNA数据和健身追踪数据，为饮食生活方式提供优化建议。正如约翰·梅尔曾说：你的身体是座奇境。

Try it at drinkelement.com/lex. This show is also brought to you by InsideTracker, a service I use to track the biological data for my body. They have a bunch of plans. Most include a blood test that gives you a bunch of information that is then used by machine learning algorithms to use your blood data, DNA data, and fitness tracker data to make recommendations for positive diet and lifestyle changes. Your body, as John Mayer has once said, is a wonderland.

它充满数据，不断向你发送信号。我们身体的奇妙之处在于它能整合这些信号维持健康。如今我们开发的医疗工具和饮食方案能扩展身体机能，但它们需要了解这些信号。

It's full of data. It's it's full of signals that it's sending to you. And the incredible aspect of our body is that it's able to integrate those signals to then maintain health. Right? Now we're creating a lot of medical tools and diet tools and all that kind of stuff that's able to extend the capabilities of our body, but they need it needs to know about those signals.

必须读懂身体发出的信号。我确信这将是未来医疗健康的发展方向。限时登录insidetracker.com/lex享受特别优惠。这里是Lex Friedman播客，请通过简介中的赞助商链接支持我们。

It needs to know about the signals the body sends. It's obvious to me that this is the future of health and medicine and all that kind of stuff. Get special savings for a limited time when you go to insidetracker.com/lex. This is a Lex Friedman podcast. To support it, please check out our sponsors in the description.

现在，亲爱的朋友们，有请迈克尔·莱文。胚胎发生是从单个细胞构建人体的过程。我认为这是地球上最不可思议的现象之一——从单个胚胎发展而来。那么这个过程是如何运作的呢？

And now, dear friends, here's Michael Levin. Embryogenesis is the process of building the human body from a single cell. I think it's one of the most incredible things that exists on Earth from a single embryo. So how does this process work?

是的，这确实是个不可思议的过程。我认为这可能是最神奇的过程，其中最根本且有趣的是它展示了我们每个人都经历了从所谓的纯物理到心智的旅程。因为我们都是以单个静止未受精卵开始生命，本质上就是一袋化学物质，你会说这就是化学和物理。而九个月加上若干年后，你就有了一个具有高级认知、偏好和内在生命的有机体。胚胎发生告诉我们，从物理到心智的转变是渐进且平滑的。

Yeah. It is it is an incredible process. I think it's maybe the most magical process there is and I think one of the most fundamentally interesting things about it is that it shows that each of us takes the journey from so called just physics to mind, right, because we all start life as a single quiescent unfertilized oocyte and it's basically a bag of chemicals and you look at that and you say okay this is chemistry and physics. And then nine months and some years later you have an organism with high level cognition and preferences and inner life and so on. And what embryogenesis tells us is that that transformation from physics to mind is gradual, it's smooth.

并不存在某个特殊时刻会突然'轰'地一声让你从物理跃迁到真正认知，这种情况不会发生。通过这个过程我们可以看到整个宇宙最大的谜题——如何从物质中产生心智。

There is no special place where a lightning bolt says boom, now you've gone from physics to true cognition, that doesn't happen. And so we can see in this process that the whole mystery, you know, the biggest mystery of the of the universe basically, how you get mind from matter.

所谓的'纯物理'。那么魔法究竟在哪里？我们如何从DNA编码的信息中创造出物理现实？

From just physics, in quotes. Yeah. So where's the magic into the thing? How do we get from information encoded in DNA and make physical reality out of that information?

如果我们要把DNA纳入讨论，我认为很重要的一点是要明白DNA编码的是生命的硬件。DNA包含了每个细胞所拥有的微观层面硬件指令——所有蛋白质、信号因子、离子通道等细胞拥有的精巧硬件都来自DNA。其余则存在于所谓的通用规律中，包括数学定律、计算法则、物理规律等各类不在DNA直接编码中的有趣内容。

So one of the things that I think is really important if we're going to bring in DNA into this picture is to think about the fact that what DNA encodes is the hardware of life. DNA contains the instructions for the kind of micro level hardware that every cell gets to play with. So all the proteins, all the signaling factors, the ion channels, the cool little pieces of hardware that cells have, that's what's in the DNA. The rest of it is in so called generic laws. These are laws of mathematics, these are laws of computation, these are laws of physics, all kinds of interesting things that are not directly in the DNA.

我总是给'纯物理'打引号，因为我认为根本不存在所谓的'纯物理'。用二元分类来看待这些问题——这是物理，这是真实认知，这只是假装——这种思维方式会让我们陷入困境。我们必须理解这是个连续统，需要研究尺度定律。这方面有很多值得探讨的有趣观点。

And that process, you know I think the reason I always put just physics in quotes is because I don't think there is such a thing as just physics. I think that thinking about these things in binary categories like this is physics, this is true cognition, this is as if, it's only faking, these kinds of things, I think that's what gets us in trouble. I think that we really have to understand that it's a continuum, We have to work up the scaling, the laws of scaling, and we can we can certainly talk about that. There's a lot of really interesting thoughts to be had there.

所以物理与信息是深度整合的。DNA并非独立存在，它在某种意义上始终与所处环境的物理定律相融合。

So the physics is deeply integrated with the information. So the DNA doesn't exist on its own. The DNA is integrated as in in some sense in response to the the the laws of physics at every the laws of the environment it exists in.

是的。环境以及宇宙的法则。我的意思是，关键在于一旦进化发现了某种机器，如果物理实现得当，某种程度上——这很难用语言描述，因为我们还没有合适的词汇——但这很像柏拉图式的概念：只要机器存在，它就会自动带来有趣的事物，你无需从零开始进化，因为物理法则已经免费提供给你了。举个特别蠢的例子，如果你想进化出一个特定的三角形，你可以进化第一个角，再进化第二个角，第三个角根本不需要进化——你早就知道它该是多少度了。

Yeah. The environment and also the laws of the universe. I mean, the thing about the the is that it's once evolution discovers a certain kind of machine, that if the physical implementation is appropriate, it's sort of and this is hard to talk about because we don't have a good vocabulary for this yet, but it's a very kind of platonic notion that if the machine is there, it pulls down interesting things that you do not have to evolve from scratch because the laws of physics give it to you for free. So just as a really stupid example, if you're trying to evolve a particular triangle, you can evolve the first angle and you evolve the second angle, you don't need to evolve the third. You know what it is already.

为什么你会知道？这是特定空间几何学免费赠予的礼物，你自然知道那个角必须是多少度。如果你进化出一个离子通道（离子通道本质上就是晶体管，它们是电压门控的电流导体），你立刻就能使用真值表，获得逻辑功能。你不需要进化逻辑功能，不需要进化真值表，DNA里也不必编码这些——它们都是免费获得的。有了NAND门你就能构建任何想要的东西，这也是免费获得的。

Now why do you know? That's a gift for free from geometry in a particular space, you know what that angle has to be. And if you evolve an ion channel, which is ion channels are basically transistors, right, they're voltage gated current conductances. If you evolve that ion channel, you immediately get to use things like truth tables, get logic functions. You don't have to evolve the logic function, you don't have to evolve a truth table, doesn't have to be in the DNA, you get it for free, right, and the fact that if you have NAND gates you can build anything you want, you get that for free.

你只需要进化出第一步——那个能让你与这些法则耦合的初始小机器。还有粘附法则等诸多法则，这些都是遗传建立的硬件与后天形成的软件（也就是完成所有计算、认知等功能的生理软件）之间的相互作用，是DNA信息与物理计算法则等要素之间真实的交织。

All you have to evolve is that first step, that first little machine that enables you to couple to those laws. And there's laws of adhesion and many other things and this is all that interplay between the hardware that's set up by the genetics and the software that's made, right, the physiological software that basically does all the computation and the cognition and everything else is a real interplay between the information and the DNA and the laws of physics of computation and so on.

那么是否可以这样说：就像数学法则是被发现的、它们潜藏在宇宙结构之中那样，生物学法则某种程度上也是被发现的？

So is it fair to say just like this idea that the laws of mathematics are discovered, they're latent within the fabric of the universe in that same way the laws of biology are kind of discovered?

没错。我认为这完全正确——虽然可能不是主流观点，但我认为切中要害。是的。

Yeah. Think that's absolutely and and it's it's probably not a popular view, but I I think that's right on the money. Yeah.

我觉得这是个非常深刻的观点。那么胚胎发生就是揭示这些法则的过程，是具现化、显现这些法则的过程。你并不是在构建这些法则。对。

Well, I think that's a really deep idea. Then embryogenesis is the process of revealing Yeah. Of embodying, of manifesting these laws. You're not building the laws. Yep.

你只是在创造揭示它们的能力。

You're just creating the capacity to reveal.

是的。我认为，再次强调，这绝非分子生物学的标准观点，但我认为这完全正确。我举个简单例子。我们最近关于异种机器人的研究——我们从早期青蛙胚胎中取出一些皮肤细胞，本质上是在探究它们的可塑性。

Yes. I think, again, not the standard view of molecular biology by any means, but I think that's right on the money. I'll give you a simple example. Some of our latest work with these xenobots, right? So what we've done is to take some skin cells off of an early frog embryo and basically ask about their plasticity.

如果给这些细胞机会在不同环境中重启其多细胞性，它们会怎么做？因为观察胚胎发育时你可能会假设这个过程极其可靠，对吧？它非常稳健。但这恰恰掩盖了它最有趣的特征。我们已经习以为常了。

If we give you a chance to reboot your multicellularity in a different context, what would you do? Because what you might assume by looking at, the thing about embryogenesis is that it's super reliable, right? It's very robust. And that really obscures some of its most interesting features. We get used to it.

我们习惯了橡子长成橡树、蛙卵变成青蛙，觉得还能变成什么？这就是标准答案。但现实是，当你观察这些皮肤细胞时不禁要问：它们究竟会做什么？它们原本只会作为被动的二维外层防止细菌侵入胚胎。然而实验表明，若移除胚胎其他部分，让这些皮肤细胞独处时——

We get used to the fact that acorns make oak trees and frog eggs make frogs and we say, what else is it going make? That's what it makes, that's a standard story. But the reality is, and so you look at these skin cells and say, well what do they know how to do? Well they know how to be passive boring two dimensional outer layer keeping the bacteria from getting into the embryo, that's what they know how to do. Well it turns out that if you take these skin cells and you remove the rest of the embryo, so you remove all of the rest of the cells and you say well you're by yourself now, what do you want to do?

它们会形成微型多细胞生物体在培养皿中游走，展现出惊人能力：穿越迷宫、独立或协作完成各种行为。它们实现了冯·诺依曼自我复制的构想——当培养皿中撒入松散细胞时，这些微型体会聚集细胞形成新代异种机器人，堪称能利用环境材料自我复制的机器。这些特性完全超出对青蛙基因组的预期。

So what they do is they form this little, this multi little creature that runs around the dish, they have all kinds of incredible capacities. They navigate through mazes, they have various behaviors that they do both independently and together. They they have a basically they implement Von Neumann's dream of self replication, because if you sprinkle a bunch of loose cells into the dish, what they do is they run around, they collect those cells into little piles, they they sort of mush them together until those little piles become the next generation of xenobots. So you've got this machine that builds copies of itself from loose material in its environment. None of this are things that you would have expected from the frog genome.

实际上这些细胞基因组是野生型，未经过任何基因编辑或纳米材料添加。我们通过减法实现了工程改造：移除那些迫使细胞分化为皮肤细胞的其他细胞后，发现它们的默认行为模式其实是成为异种机器人——在胚胎内其他细胞类型的压制下才被迫成为皮肤细胞。这就引出了你刚才提出的深刻问题——

In fact the genome is wild type, there's nothing wrong with their genetics. Nothing has been added, no nanomaterials, no genomic editing, nothing. And so what we have done there is engineer by subtraction. What you've done is you've removed the other cells that normally basically bully these cells into being skin cells, and you find out that what they really want to do is to be their default behavior is to be a xenobody, but in vivo, in the embryo, they get told to be skinned by these other cell types. So now here comes this really interesting question that you just posed.

当追问蝌蚪和青蛙形态的起源时，标准答案是数百万年自然选择塑造出适应蛙类环境的特定体型。但异种机器人的形态从何而来？自然界从未存在过异种机器人，也从未有过相关选择压力。这些细胞在新环境中仅用48小时就演化出全新原始生命形态，具备运动式自我复制等青蛙和蝌蚪都不具备的能力。

When you ask where does the form of the tadpole and the frog come from, the standard answer is, well, it's it's it's selection. So over over millions of years, right, it's been shaped to to produce the specific body with that's fit for froggy environments. Where does the shape of the xenobot come from? There's never been any xenobots, there's never been selection to be a good xenobot. These cells find themselves in the new environment, in forty eight hours they figure out how to be an entirely different proto organism with new capacities like kinematic self replication, that's not how frogs or tadpoles replicate.

我们完全阻断了它们常规的繁殖途径。短短几天内，这些小家伙就找到了生物圈中前所未有的全新繁殖方式。

We've made it impossible for them to replicate their normal way. Within a couple days these guys find a new way of doing it that's not done anywhere else in the biosphere.

好吧，实际上，让我们退一步来定义什么是异种机器人？

Well actually, let's step back and define what are xenobots?

异种机器人是一种自我组装的小型原始生物体，同时也是生物机器人。这两者并不矛盾，它兼具两者的特性。

So a xenobot is a self assembling little proto organism. It's also a biological robot. Those things are not distinct. It's a member of both How

其中有多少是生物特性？又有多少是机器人特性？

much is a biology? How much is a robot?

目前来看，大部分属于生物特性，因为我们正在探索这些细胞及细胞集合体的自然行为。这项工作的重要部分是与佛蒙特大学Josh Bongard团队的合作，他们是从事人工智能的计算机科学家，已能运用模拟进化的方法研究如何操控这些细胞——通过信号刺激而非改写DNA（即不改变硬件而是改变经验信号）。比如移除或添加细胞，用不同方式刺激它们以改变行为。虽然这是尚未发表的未来研究方向，但我们正在开发各种新方法来重编程它们的行为。但在重编程之前，必须首先理解它们与生俱来的能力。

At this point, most of it is biology because what we're doing is we're discovering natural behaviors of these the cells and also of the cell collectives. Now one of the really important parts of this was that we're working together with Josh Bongard's group University of Vermont, they're computer scientists who do AI and they've basically been able to use an evolutionary, a simulated evolution approach to ask how can we manipulate these cells, give them signals, not rewire their DNA, so not hardware but experience signals. So can we remove some cells, can we add some cells, can we poke them in different ways to get them to do other things. So in the future there's going be, you know, we're we're now and this is this is future unpublished work, but we're doing all sorts of interesting ways to reprogram them to new behaviors. But before you can start to reprogram these things, you have to understand what their innate capacities are.

明白了。这意味着未来你们会对其进行工程化编程。某种程度上，机器人的定义就包含人为工程改造的部分，而非自然演化。

Okay. So that means engineering, programming, you're engineering them in in in the future. And in some sense, the the definition of a robot is something you in part engineer Yeah. And and first Yes. Versus evolve.

其实这个定义本身就很模糊。某种意义上，我们体内的许多生物组织也可以算作某种机器人。

I mean, it it's such a fuzzy definition anyway. In some sense, many of the organisms within our body are kinds of robots. Yes.

是的。

Yes.

我认为机器人是个奇怪的界限，因为我们往往把机器人视为异类。我想未来会有类似机器人权利运动的时刻，不过这个我们或许可以稍后再谈。当然。那么，我们该如何深入探讨呢？如何构建一个异种机器人？

And I think robots is a weird line because it's we tend to see robots as the other. I think there will be a time in the future when there's going to be something akin to the civil rights movements for robots, but we'll talk about that later perhaps. Sure. Anyway, so how do you can we just linger on it? How do you build a xenobot?

我们在这里讨论什么？它是从何处起源，又如何成长为辉煌的异种机器人的？

What are we talking about here? From from whence does it start and how does it become the glorious xenobot?

是的。先退一步说，很多人纠结的一点是认为工程学需要新的DNA电路或纳米材料。关键在于，我们正从使用被动材料的传统工程学转向新时代——木材金属这类材料，你唯一能指望的就是它们保持形状，仅此而已。它们不会主动作为，工程师需要完全操控它们的所有行为。

Yeah. So just to take one step back, one of the things that a lot of people get stuck on is they say, well, engineering requires new DNA circuits or it requires new nanomaterials. Thing is we are now moving from old school engineering which use passive materials, Things like wood metal, things like this, that basically the only thing you could depend on is that they were going to keep their shape, that's it. They don't do anything else. It's on you as an engineer to make them do everything they're going to do.

接着出现了活性材料，现在是计算材料。这是全新时代，这些是具有能动性的材料。你现在是在与基底材料协作，因为材料本身就有其意图。这些细胞拥有数十亿年的进化史，它们有自己的目标。

And then there were active materials and now computational materials. This is a whole new era, these are agential materials. This is you're you're now collaborating with your substrate because your material has an agenda. These cells have, you know, billions of years of evolution. They have goals.

它们有偏好，不会乖乖待在你放置的位置。

They have preferences. They're not just going to sit where you put them.

这太滑稽了，你居然得说服你的材料保持——

That's hilarious that that you have to talk your material into keeping

安全。没错，完全正确。

it safe. That's it. That is exactly right.

就像待在那里不动一样。这就像是试图用一堆猫之类的东西组织出某种形状。

Is exactly Stay there. It's like getting a bunch of cats or something and trying to organize a shape out of them.

有趣的是我们想到一块儿去了，因为在一篇刚被《自然·生物工程》接受的论文里，我有个图表就是用乐高积木搭塔与用狗搭塔的对比。对吧？所以想想两者的区别。

It's funny we're on the same page here because in a paper, this is this is currently just been accepted in Nature Bioengineering. One of the figures I have is building a tower out of Legos versus dogs. Right? Yeah. So think about the difference.

对吧？用乐高搭建时你能完全控制结构走向，但要是有人碰倒就全完了。用狗的话你没法直接堆叠它们，它们不会保持那种状态。但好消息是如果训练得当，即使被推倒它们也会重新站好。所以没关系。

Right? If you build out of Legos, you have full control over where it's gonna go, but if somebody knocks it over, it's game over. With the dogs, you cannot just come and stack them, they're not gonna stay that way. But the good news is that if you train them, then somebody knocks it over, they'll get right back up. So it's all right.

所以作为工程师，你真正需要知道的是：能依赖这个东西做什么。这才是关键。很多人从构成或起源定义机器人，比如设计还是进化之类的。我觉得这些都没用。我认为工程师需要知道的是：当我不在身边微观管理时，能在多大程度上依赖它完成任务？

So as an engineer, what you really want to know is what can they depend on this thing to do. Right? That's really you know, a lot of people have definitions of robots as far as what they're made of or how they got here, you know, design versus evolve, whatever. I don't think any of that is useful. I think I think as an engineer, what you want to know is how much can I depend on this thing to do when I'm not around to micromanage it?

我能给予这个东西多大程度的依赖？它有多少自主性？这决定了你采用什么技术手段。是用微观管理精确定位，还是训练它？

What level of dependency can I can I give this thing? How much agency does it have? Which then tells you what techniques do you use. So do you use micromanagement, like you put everything where it goes? Do you train it?

是给它信号指令？还是说服它行动？明白吗？你的基质有多智能？现在我们正进入这个与具有能动性的材料合作的领域。

Do you give it signals? Do you try to convince it to do things? Right? How much, you know, how intelligent is your substrate? And so now we're moving into this this area where you're you're you're working with agential materials.

这是协作关系，不是那种老派做法。

That's a collaboration. That's not that's not old old style.

你用的词是什么？'Agential'（主体性）吗？

What's the word you're using? Agential?

对，就是'Agential'（主体性）。

Agential. Yeah.

这具体是什么意思？

What's that mean?

源于'Agency'（能动性）这个概念。简单来说，材料具有某种程度的能动性——当然不是人类级别的，但它们确实存在偏好、目标、记忆能力，能基于记忆进行未来预测。比如细胞就不同程度地具备所有这些特性。

Agency. It comes from the word agency. So so basically the material has agency, meaning that it has some level of, obviously not human level, but some level of preferences, goals, memories, ability to remember things to compute into the future, meaning anticipate, you know, when you're working with cells they have all of that to some to various degrees.

让材料真正拥有自主意识，这到底是赋能还是限制呢？

Is that empowering or limiting, having material as a mind of its own literally?

我认为是双刃剑。一方面这会带来挑战：如果你用线性思维预测结果，会不断遭遇意外，因为生物行为不符合线性预期。但另一方面又极具解放性——再生医学的突破正需利用这点，让材料自主完成你无法微观操控的事。举个简单例子：想让老鼠表演钻圈，你可以选择微观操控每个神经元（虽然未来或许能实现），或者直接训练老鼠。

I think it's both, right? So it raises difficulties because it means that if you're using the old mindset which is a linear kind of extrapolation of what's going to happen, you're going to be surprised and shocked all the time because biology does not do what we linearly expect materials to do. On the other hand it's massively liberating and so in the following way, I've argued that advances in regenerative medicine require us to take advantage of this because what it means is that you can get the material to do things that you don't know how to micromanage. So just as a simple example, if you had a rat and you wanted this rat to do a circus trick, put a ball in a little hoop, you can do it the micromanagement way which is try to control every neuron and try to play the thing like a puppet, right, and maybe someday that'll be possible, maybe. Or you can train the rat.

这就是为什么人类在毫无神经科学知识的数千年里，仍能驯化动物。一旦认识到系统的能动性层级，就能采用合适方法。了解动机的奖惩机制、认知水平、行为偏好后，你探索的是远比微观操控更流畅的解决方案。再生医学中，无论是让手臂再生还是修复眼部细胞缺陷，难道真要实时操控数万基因？还是找到'在此处构建手臂'这样的高级模块化指令？毕竟构建手臂的指令本就存在。

And this is why humanity for thousands of years before we knew any neuroscience, we had no idea what's between the ears of any animal, we were able to train these animals. Because once you recognize the level of agency of a certain system you can use appropriate techniques. If you know the currency of motivation, reward and punishment, you know how smart it is, you know what kinds of things it likes to do, you are searching a much more, much smoother, much nicer problem space than if you try to micromanage the thing. And in regenerative medicine when you're trying to get let's say an arm to grow back or an eye to repair a cell birth defect or something, do you really want to be controlling tens of thousands of genes at each point to try to micromanage it, or do you want to find the high level modular controls that say build an arm here? You already know how to build an arm.

你之前做过，现在再做一次。所以我认为这既是挑战也是机遇。它既困难，又促使我们开发新的工程方法，同时具有巨大的赋能潜力。

You did it before and do it again. So that's I I think it's it's both. It's both difficult and it challenges us to develop new ways of engineering, and it's it's hugely empowering.

好的。那么具体怎么做呢？沿用猫狗的比喻，我猜你需要识别出'狗'并处理掉'猫'。因为你知道，就像老话说的'赶猫难'是个问题，所以你可能能训练狗。我怀疑你永远无法训练猫。就算能训练，你也永远无法信任它们。

Okay. So how do you do I mean, maybe sticking with the metaphor of dogs and cats, I presume you have to figure out the find the dogs and dispose of the cats. Because, you know, it's like the old herding cats is is an issue, so you may be able to train dogs. I suspect you will not be able to train cats. Or if you do, you're never gonna be able to trust them.

那么有没有方法能判断哪种材料适合被'驱赶'？是通过实验室工作还是通过模拟？

So is there a way to figure out which material is amenable to herding? Is it in the lab work or is it in simulation?

目前主要在实验室进行，因为我们的模拟还无法捕捉生物学中最有趣和强大的部分。我们擅长模拟的是前馈涌现类现象，比如细胞自动机——设定简单规则后让每个代理或细胞按规则发展，就会产生复杂结果，比如蚁群算法这类。困难在于这类过程极难逆向推导，这是个非常棘手的反问题。

Right now it's largely in the lab because we our simulations do not capture yet the most interesting and powerful things about biology. The simulation what we're pretty good at simulating are feed forward emergent types of things, right, so cellular automata, if you have simple rules and you sort of roll those forward for every agent or every cell in the simulation and complex things happen, you know ant colony or algorithms, things like that. We're good at that and that's fine. The difficulty with all of that is that it's incredibly hard to reverse. So this is a really hard inverse problem, right?

如果你观察白蚁群建造单烟囱结构时心想'我想要双烟囱'，如何修改白蚁行为规则来实现？或者看到细胞发育成某种生物形态时想修复先天缺陷，如何调控所有底层蛋白质相互作用？从目标解剖结构反推硬件规则通常是无解的。所以目前主要靠实验，我们需要理解生物学如何运用自上而下的控制机制。

If you look at a bunch of termites and they make a thing with a single chimney and you say, well I like it but I'd like two chimneys. How do you change the rules of behavior free termites so they make two chimneys, right? Or if you say, here are a bunch of cells that are creating this kind of organism, I don't think that's optimal, I'd like to repair that birth defect, how do you control all the individual low level rules, all the protein interactions and everything else. Rolling it back from the anatomy that you want to the low level hardware rules is in general intractable, it's an inverse problem, it's generally not solvable. So right now it's mostly in the lab because what we need to do is we need to understand how biology uses top down controls.

关键不是自下而上的涌现，而是目标导向的'测试-操作-退出'循环——在新空间（非基因表达空间，比如解剖结构空间）进行误差最小化。举个简单例子：蝾螈断肢后能精确再生缺失部分并适时停止。再生最神奇之处就在于这个停止机制——它知道何时形成了完整的蝾螈前肢。

So the idea is not bottom up emergence, but the idea of things like goal directed test operate exit kinds of loops where it's basically an error minimization function over a new space, not a space of gene expression, but for example a space of anatomy. So just as a simple example, if you have a salamander and it's got an arm, can amputate that arm anywhere along the length, it will grow exactly what's needed and then it stops. That's the most amazing thing about regeneration is that it stops, it knows when to stop. When does it stop? It stops when a correct salamander arm has been completed.

这说明存在胺类终末分析机制：生物体必须知道正确肢体的形态。它能评估当前形状，计算与目标形态的差值，然后持续进行重塑生长直至完成。我们在实验室利用涡虫和蛙胚验证了这个稳态循环——一旦掌握这个机制，就能干预这个稳态循环。比如可以问：形态记忆如何存储？能否植入错误记忆让细胞构建不同结构？

So that tells you that's a that's a amines ends kind of analysis where it has to know what the correct limb is supposed to look like, right? So it has a way to ascertain the current shape, it has a way to measure that delta from what shape it's supposed to be, and then it will keep taking actions, meaning remodeling and growing and everything else until that's complete. Once So you know that, and we've taken advantage of this in the lab to do some really wild things with both planaria and frog embryos and so on, once you know that you can start playing with that homeostatic cycle. You can ask for example, well how does it remember what the correct shape is and can we mess with that memory? Can we give it a false memory of what the shape should be and let the cells build something else?

或者我们可以干扰测量设备，对吧？这样你就能得到那种... 这个想法本质上是将认知神经科学和行为科学中的大量方法和概念应用到那些曾被视作无生命的材料上。如果你说'我的细胞想做这个，我的细胞想做那个'，在课堂上你会因为拟人化而被训斥。但我认为这是个重大错误，它让我们错失了大量潜在能力。

Or can we mess with the measurement apparatus, right? So it gives you those kinds of so the idea is to basically appropriate a lot of the approaches and concepts from cognitive neuroscience and behavioral science into things that previously were taken to be dumb materials and you you get yelled at in class for being anthropomorphic if you said well my cells want to do this and my cells want to do that. And I think I think that's a that's a major mistake that leaves a ton of capabilities on the table.

所以把生物系统看作具有记忆、几乎类似认知能力的存在。但想想看，蝾螈手臂的重建过程没有独裁者指挥，这多么不可思议？就像细胞自动机系统，所有个体工作者各行其是。那么那个自上而下的控制信号究竟来自哪里？

So thinking about biological systems as things that have memory, have almost something like cognitive ability. But I mean, how incredible is it, you know, that the salamander arm is being rebuilt not with a dictator? It's kinda like the cellular automata system. All the individual workers are doing their own thing. So where's that top down signal that does the control coming from?

比如，你怎么能找到它？是啊。为什么它会停止生长？它怎么知道形状？它怎么记住形状的？

Like, how can you find it? Yeah. Like, why does it stop growing? How does it know the shape? How does it have memory of the shape?

它又是怎么告诉大家'哇哇哇，慢点，我们完成了'的？

And how does it tell everybody to be like, woah woah woah, slow down, we're done?

首先要思考的是，根本不存在所谓的中央独裁者。因为在这类科学中，万物都由部件组成。尽管我们感觉自己是个统一的中枢智能和认知点，但我们实际上是一袋神经元，对吧？所有智能都是集体智能。这点很重要，因为很多人认为存在'真实智能'（比如我）和'集体智能'（比如蚂蚁、鸟群、白蚁等），可能适合也可能不适合将它们视为个体。很多人对此持怀疑态度。但你必须意识到，根本不存在这种不可分割的智能钻石——那种不由部件组成的单一中心。我们都是由部件组成的。

So the first thing to think about I think is that there are no examples anywhere of a central dictator because in this kind of science because everything is made of parts and so we even though we feel as a unified central sort of intelligence and kind of point of cognition, we are a bag of neurons, right? All intelligence is collective intelligence. This is important to kind of about because a lot of people think, okay, there's real intelligence, like me, and then there's collective intelligence, which is ants and flocks of birds and termites and things like that, and maybe it's appropriate to think of them as an individual and maybe it's not, and a lot of people are skeptical about about that and so on. But you've to realize that we are not there's there's no such thing as this, like, indivisible diamond of intelligence that's like this one central thing that's not made of parts. We are all made of parts.

所以如果你相信——我认为这很难否认——我们确实拥有集中的目标和偏好，会计划会行动等等，那么你就已经承认细胞集合体能够做到这些，因为我们就是细胞集合体。这是无法回避的。就我们而言，我们所做的是在三维世界中导航，并产生行为。

And so if if you believe, which I think is is is hard to to get around, that that that we in fact have a centralized set of goals and preferences and we plan and we do things and so on, you are already committed to the fact that a collection of cells is able to do this, because we are a collection of cells. There's no getting around that. In our case, what we do is we navigate the three-dimensional world, and we have behavior.

这简直让我震惊，因为我们只是细胞的集合。哦，是啊。所以当我移动这只手臂时，我感觉自己是这个动作的中央独裁者。但实际上有太多事情在同时发生。

This is blowing my mind right now because we are just a collection of cells. Oh, yeah. Yeah. So when I'm moving this arm, I feel like I'm the central dictator of that action. But there's a lot of stuff going on.

比如，这里所有的细胞都在以某种方式协作，它们正从中枢神经系统接收信号，这很有趣。嗯，甚至

Like, all all the cells here are collaborating in some They're interesting getting signal from the central nervous system. Well, even

中枢神经系统这个名称其实有误导性，因为它并不真正位于中心。再说一次，它其实是...

the central nervous system is is misleadingly named because it isn't really central. Again, it's it's what what

它只是一堆细胞。

It's a bunch of cells.

就是一堆细胞而已。我是说，所有东西都是这样，对吧？世界上不存在单一不可分割的智能体。我们所见的一切实例都是某种集体存在。只是我们习以为常了。

It's just a bunch of cells. I mean, all of right? There are no there are no singular indivisible intelligences anywhere. We are all every every example that we've ever seen is is a collective of some of something. It's just that we're used to it.

我们习惯了这种认知，明白吗？我们习惯了把某个东西看作单一实体，但其实不是。放大看就知道——你看到的是一群四处活动的细胞。

We're used to that, you know, we're used to, okay. This thing is kind of a single thing, but it's really not. You zoom in. You know what you see. You see a bunch of cells running around.

那么是否存在某种统一性...我们话题跳来跳去，但你提到的生物电信号与生物化学信号，化学与电的关系...或许生命不是细胞间的对抗，而是像交响乐团演奏，最终呈现的音乐才是主宰者。

And so Is there some unifying I mean, we're jumping around, but that's something that you look as the the the bioelectrical signal versus the biochemical, the the chemistry, the electricity. Maybe the life isn't that versus the cells. It's the there's there's an orchestra playing, and the resulting music is the dictator.

这个比喻不错。这是丹尼斯·诺布尔的观点。他有两本很好的书专门讨论这个音乐类比。所以我觉得...我很喜欢这个说法。

That's not bad. Dennis that's Dennis Noble's kind of view of things. He has he has two really good books where he talks about this musical analogy. Right? So so I think that's that's that's I I like it.

我喜欢它。

I like it.

但这有错吗？

Is it wrong though?

我不认为是错的。我不觉得这是错的。我真的不认为这是错的。我认为关键在于我们必须接受一个事实：一个真正、真正完善的认知智能仍然可以由部件构成。这些部件必须如此，事实上也必须如此。

I don't think it's no. I don't think it's wrong. I I don't I don't think it's wrong. I think I think the important thing about it is that we have to come to grips with the fact that a true a true proper cognitive intelligence can still be made of parts. Those things are and in fact it has to be.

我觉得这真的很遗憾，但我经常看到这种情况。当你拥有这样一个集体时，无论是机器人群体，还是细胞、神经元之类的集合体，一旦我们对其运作机制有所了解——比如发现'哦，要执行这个动作，信息是通过这种化学机制处理的'——人们立刻就会说'那这不是真正的认知，只是物理现象'。

I think it's a real shame, but I see this all the time. When you have when you have a collective like this, whether it be a a group of robots or a, you know, a collection of cells or neurons or whatever, as soon as as soon as we gain some insight into how it works, right, meaning that, oh, see. In order to take this action, here's the information that got processed via this chemical mechanism or whatever. Immediately people say, oh, well then that's not real cognition. That's just physics.

我认为这种观点从根本上就是错误的。如果你放大观察任何事物，能看到什么？当然只会看到物理现象。底层还能有什么？不会是魔法粉末，只能是物理和化学反应。但这并不减损一个神奇的事实：通过特定方式组织这些物理化学反应（尤其是生物电，我对此非常着迷），就能涌现出具有目标、偏好、记忆和预期的集体智能——这些属性不属于任何子单元。

And I think this is this is fundamentally flawed because if you zoom into anything, what are you going to see? Of course you're just going to see physics. What else could be underneath, right? It's not going be fairy dust, it's going to be physics and chemistry. But that doesn't take away from the magic of the fact that there are certain ways to arrange that physics and chemistry, and in particular the bioelectricity, which which I like a lot, to give you an emergent collective with goals and preferences and memories and anticipations that do not belong to any of the subunits.

所以我们现在探讨的——后续可以讨论胚胎发生过程中的表现——本质上是在探索'大写的自我'的起源。我们人类是一种自我，还存在许多其他类型的自我。关于自我如何产生又如何统一，我们能讲述许多精彩的故事。

So I think what we're getting into here, and we can talk about how this happens during embryogenesis and so on, what we're getting into is the origin of the of of a self, you know, with a big with a capital s. So we ourselves, there are many other kinds of selves, and we can tell some really interesting stories about where selves come from and how they become unified.

是啊。这是不是第一个——至少人类倾向于认为这是'大写的自我'最初诞生的层级？但我们真的不愿把人类文明或地球本身看作一个活的生命体。没错，这种想法会让我们非常不适。

Yeah. Is this the first or at least humans tend to think that this is the the level of which the self with a capital s is first born. But and we really don't wanna see human civilization or earth itself as one living organism. Yeah. That's very uncomfortable to us.

是的。没错。但确实。自我诞生于何处？

It is. Yeah. But is yeah. Where's the self born?

我们必须超越那个阶段。我喜欢做的是，嗯，我会给你讲两个关于这个的简短故事。喜欢倒推。所以不同于，如果你开始并说，好吧，这是一个草履虫，你看到它，知道它是一种单细胞生物，你看到它做各种事情，人们会说，肯定有某种化学解释来说明它是如何做到的，所以那不是真正的认知。对吧？

We have to grow up past that. So what I like to do is well, I'll I'll tell you two quick stories about that. Like to roll backwards. So so as opposed to so if you start and you say, okay, here's a paramecium and you see it you know, it's a single cell organism, you see it doing various things, and people will say, I'm sure there's some chemical story to be told about how it's doing it, so that's not true cognition. Right?

人们会争论这一点。我喜欢倒推。比如说，让我们同意，你和我，坐在这里，如果有真正的认知，我们就是例子。因为如果有真正的认知，我们就是例子。现在让我们慢慢倒推。

And people will argue about that. I I like to work it backwards. Say, let's let's let's agree that you and I, as as we sit here, are examples of true cognition, if anything. Because if there's anything that's true cognition, we are we are examples of it. Now let's just roll back slowly.

对吧？所以你倒推到你还是个小孩的时候，习惯做任何事情，然后一天天倒推，最终你或多或少变成了那个草履虫，甚至更低级，对吧，作为一个未受精的卵母细胞。据我所知，没有人能提出一个令人信服的离散步骤，说明我的认知能力在哪个点消失。生物学并没有提供任何具体的步骤，它是极其平滑、缓慢和连续的。所以我认为这种认为它神奇地出现在某个点，然后人类有了其他地方不存在的真正自我的观点，我认为它与我们所知的进化、发育生物学的一切相违背，这些都是缓慢的连续体。

Right? So you roll back to the time when you were a small child and used to doing whatever, and then just sort of day by day you roll back, and eventually you become more or less that paramecium, and you're sort of even below that, right, as an unfertilized oocyte. So no one has, to my knowledge, no one has come up with any convincing discrete step at which my cognitive powers disappear. Just doesn't biology doesn't offer any specific step, it's incredibly smooth and slow and continuous. And so I think this idea that it just sort of magically shows up at one point and then humans have true selves that don't exist elsewhere, I think it runs against everything we know about evolution, everything we know about developmental biology, these are all slow continua.

另一个非常重要的故事是胚胎的来源。想一下，羊膜胚胎，这是人类、鸟类等，哺乳动物和鸟类等。想象一个扁平的细胞盘，可能有5万个细胞，当你得到一个受精卵，比如说你从农场买一个受精卵，那个卵会有大约5万个细胞在一个扁平盘里，看起来像一个小飞盘。在那个扁平盘里，会发生的是，有一组细胞会变得特殊，它会告诉其他细胞，我要成为头部，你们不要成为头部。

And the other really important story I want to tell is where embryos come from. So think about this for a second. Amniotic embryo, so this is humans, birds and so on, mammals and birds and so on. Imagine a flat disc of cells, so there's maybe 50,000 cells, and in that, so when you get an egg from fertilized let's say you buy a fertilized egg from a farm, That egg will have about 50,000 cells in a flat disk, it looks like a little tiny little frisbee. And in that flat disk what will happen is there will be one set of cells will become special and it will tell all the other cells I'm going to be the head, you guys don't be the head.

它会放大对称性破缺的放大，你得到一个胚胎，有神经组织和一些其他东西形成。现在你说，好吧，我有一个卵和一个胚胎，就是这样，还能是什么？但现实是，我曾经做过所有这些研究，如果你用一根小针在那个胚盘上划一下，让细胞暂时无法交流，它会愈合，但暂时它们无法交流。会发生的是，两个区域都会决定它们可以成为胚胎，然后会有两个，当它们愈合时，它们会成为连体双胞胎，你可以制造两个，三个，甚至更多。所以里面有多少个自我，这个问题只有在整个过程完成后才能回答。

So it'll amplify symmetry breaking amplification, you get one embryo, there's a neural tissue and some other stuff forms. Now you say, okay, I had one egg and one embryo and there you go, what else could it be? Well the reality is, and I used to, I did all of this as a grad student, if you take a little needle and you make a scratch in that blastoderm, in that disc such that the cells can't talk to each other for a while, it heals up, but for a while they can't talk to each other. What will happen is that both regions will decide that they can be the embryo, and there will be two of them, and then when they heal up they become conjoint twins, and you can make two, you can make three, you can make lots. So the question of how many selves are in there cannot be answered until it's actually played all the way through.

不一定是只有一个。可以有很多。所以你拥有这个媒介，这个未分化的——我相信在心理学上也有类似的术语，但我不知道确切的术语——你拥有这个潜能的海洋。你有这些成千上万的细胞，一些个体会从中形成，通常是一个，有时是零个，有时是几个，它们从这些细胞中形成，因为这些细胞的一个区域组织成一个集体，这个集体会有目标，单个细胞没有的目标。比如，一个肢体，制造一只眼睛，多少只眼睛？

It isn't necessarily that there's just one. There can be many. So what you have is you have this medium, this this undifferentiated I'm sure there's a there's a psychological version of this somewhere that I don't know the proper terminology, but you have this you have this list like ocean of potentiality. You have these thousands of cells and some number of individuals are going to be formed out of it, usually one, sometimes zero, sometimes several, and they form out of these cells because a region of these cells organizes into a collective that will have goals, goals that individual cells don't have. For example, a limb, make an eye, how many eyes?

确切地说就是两个。单个细胞并不知道眼睛是什么，也不知道你应该有多少只眼睛，但细胞集体知道。集体拥有个体细胞所没有的目标、记忆和预期。正是这种边界的建立，以及维持和追求特定目标的能力，构成了自我意识的起源。

Well exactly two. So individual cells don't know what an eye is, they don't know how many eyes you're supposed to have, but the collective does. The collective has goals and memories and anticipations that the individual cells don't. And that that the establishment of that boundary with its own ability to maintain to to pursue certain goals, that's the origin of of selfhood.

但‘我’这个概念是否包含在那些目标中？它们是否命中注定？比如，它们是在发现那个目标吗？从原核生物到真核细胞的进化过程中，究竟是在哪个环节发现了这一点？然后它们开始形成群体，当你形成特定群体时，你会让理解这件事听起来如此棘手。

But I is that goal in there somewhere? Were they always destined? Like, are they discovering that goal? Like, where the hell did evolution discover this when you went from the prokaryotes to eukaryotic cells? And then they started making groups, and when you make a certain group, you make a you you make it sound such a tricky thing to try to understand.

你说得好像细胞们并非聚在一起制定了目标，而是它们聚集的行为本身揭示了一直存在的目标。那个目标的可能性始终存在。

You make it sound like the cells didn't get together and came up with a goal, but the very act of them getting together revealed the goal that was always there. There was always that potential for that goal.

首先要说的是，这里的问题远比确定的答案多得多。明白吗？我告诉你的一切都是最前沿的发展动态，我们中没人知道确切答案。这是我的观点：我认为进化——我不认为进化会针对特定问题（即特定环境）产生解决方案。

So the first thing to say is that there are way more questions here than than certainties. Okay? So everything I'm telling you is is cutting edge developing, you know, stuff, so it's not as if any of us know the answer to this. Here's my opinion on this. I think what evolution I don't think that evolution produces solutions to specific problems, in other words specific environments.

比如某种青蛙能很好地适应蛙类环境。我认为进化产生的是问题解决机器，它们能在不同空间解决问题——不仅是三维空间，这回到我们之前讨论的内容。我们的大脑是进化后期的产物，这个系统能通过向肌肉发出指令在三维空间中追求目标。这个系统从何而来？它源自更古老的进化系统，那时细胞群通过指令调控细胞行为——细胞移动、分裂、死亡、转化为不同类型，在形态空间中导航（即所有可能解剖结构的空间）。

Like here's a frog that can live well in a froggy environment. I think what evolution produces is problem solving machines that will solve problems in different spaces, so not just three-dimensional space, this goes back to what we were talking about before. We, the brain is evolutionarily a late development, it's a system that is able to pursue goals in three-dimensional space by giving commands to muscles. Where did that system come from? That system evolved from a much more ancient, evolutionarily much more ancient system, where collections of cells gave instructions for cell behaviors, meaning cells move to divide, to die, to change into different cell types, to navigate morphospace, the space of anatomies, the space of all possible anatomies.

在那之前，细胞在转录空间中导航（所有可能基因表达的空间），更早之前则在代谢空间中。我认为进化创造的是擅长用各种技巧在不同空间导航的硬件——其中很多技巧肯定能用于自动驾驶汽车、机器人等领域。关键在于它们导航时并不需要完全理解空间本质，事实上它们根本不知道空间是什么。对吧？

And before that, cells were navigating transcriptional space, which is a space of all possible gene expressions, and before that metabolic space. So what evolution has done, think, is produced hardware that is very good at navigating different spaces using a bag of tricks, right, which I'm sure many of them we can steal for autonomous vehicles and robotics and various things. And what happens is that they navigate these spaces without a whole lot of commitment to what the space is. In fact, they don't know what the space is. Right?

可以说我们都是缸中之脑。每个细胞都无从知晓。对吧？每个细胞都是其他细胞的外部环境。对吧？

We are all brains in a vat, so to speak. Every cell does not know. Right? Every cell is some other cell's external environment. Right?

那么，你与外部世界之间的边界究竟在哪里？你其实并不清楚。对吧？每一个细胞集合都必须从头开始探索这一点。进化要求所有这些元素去理解它们是什么、拥有哪些效应器和传感器、在何处划定自我与外界的界限才有意义。你必须从零构建这一切——这种自我创生过程，正是定义'自我'边界的本质。

So where does that border between you you and the outside world, you don't really know where that is. Right? Every every collection of cells has to figure that out from scratch. And the fact that evolution requires all of these things to figure out what they are, what effectors they have, what sensors they have, where does it make sense to draw a boundary between me and the outside world. The fact that you have to build all that from scratch, this autopoiesis, is what defines the border of A self.

生物学采用了一种多尺度能力架构，意味着每个层级都有其目标。分子网络有目标，细胞有目标，组织、器官、群体亦是如此。正是这些层级间的相互作用，使生物能够以新方式解决问题，例如异种机器人等领域。正如你所说，细胞确实在不断探索新的存在方式，但同时进化也在塑造这一切。因此，进化堪称是这种主体性生物工程的大师。

Now biology uses like a multi scale competency architecture, meaning that every level has goals. So molecular networks have goals, cells have goals, tissues, organs, colonies, and it's the interplay of all of those that enable biology to solve problems in new ways, for example in xenobots and various other things. This is, you know, it's it's exactly as you said. In many ways, the cells are discovering new ways of being, but at the same time, evolution certainly shapes all this. So so evolution is very good at this agential bioengineering.

对吧？当进化在探索动物或植物等生物的新存在方式时，有时是通过改变硬件——比如蛋白质结构等实现的。但更多时候并非改变硬件，而是改变细胞间传递的信号。这就像我们工程师所做的：通过信号、经验和刺激来说服细胞执行各种功能。生物学必须这样做，因为它面对的不是一张白纸。

Right? When when evolution is discovering a new way of being an animal, an animal or a plant or something, sometimes it's by changing the hardware, you know, protein changing protein structure and so on. But much of the time it's not by changing the hardware, it's by changing the signals that the cells give to each other. It's doing what we as engineers do, which is try to convince the cells to do various things by using signals, experiences, stimuli. That's what biology does, it has to, because it's not dealing with a blank slate.

要知道，每次当进化试图构建一个生物体时，它面对的不是任人摆布的新鲜材料。这些材料已有自己的行为倾向，因此最有效的适应方式就是找到能说服细胞执行特定功能的信号。对吧？

Every time, as you know, if you're evolution and you're trying to make an organism, you're not dealing with a passive material that is fresh and you have to specify. It already wants to do certain things, so the easiest way to do that search to find whatever is going be adaptive is to find the signals that are going to convince the cells to do various things. Right?

你的观点是进化同时在软件和硬件层面运作，而在软件层面操作更为简便高效。

Your sense is that evolution operates both in the software and the hardware, and it's just easier and more efficient to operate in the software.

是的。但我要补充，我认为这种区分并不绝对。应该说是一个连续谱系：你可以改变特定蛋白质使其酶功能发生变化，从而影响代谢等，这会对适应性产生影响；或者你也可以改变基因组中大量非结构性的信号信息——即细胞间交流的时机与方式，这可能对问题解决机制产生巨大改变。

Yes. And I should also say, I don't think the distinction is sharp. Other words, I think it's a continuum, but I it's a meaningful distinction where you can make changes to a particular protein and now the enzymatic function is different and it metabolizes differently and whatever, and that will have implications for fitness. Or you can change the huge amount of information in the genome that isn't structural at all, it's signaling. It's when and how do cells say certain things to each other, and that can have massive changes as far as how it's gonna solve problems.

这种多层级能力架构的概念确实令人惊叹。进化构建的这种层级体系——我不知道该归功于谁。当我看到人类官僚体系的无能时就更困惑了：进化究竟如何做到让每个层级都只保留最优解？

I mean, this idea of multihierarchical competence architecture, which is incredible to think about. So this hierarchy that evolution builds, I don't know who's responsible for this. I also see the incompetence of bureaucracies of humans when they get together. So how the hell does evolution build this? Where at every level, only the best get to stick around.

他们不知怎地就能在不明全局的情况下完成工作。是的。然后还有那些管理者，他们以某种方式处理更大的事务，或者说你现在可以把一小群细胞抽象为一个器官之类的，而这个器官在整个身体的背景下发挥更大的作用。这是如何构建的？你能提供些关于这种层级能力架构构建的直觉吗？

They somehow figure out how to do their job without knowing the bigger picture. Yeah. And then there's, the bosses that do the bigger thing somehow, or that you can now abstract away the small group of cells as a as an organ or something, and then that organ does something bigger in the context of the full body or something like this. How is that built? Is there some intuition you can kind of provide of how that's constructed, that hierarchical competence architecture?

我喜欢这个说法。'能力'。单是'能力'这个词在这种语境下就很酷，因为每个人都在某种程度上擅长自己的工作。

I love that. Competence. Just the word competence is is pretty cool in this context because everybody's good at their job somehow.

没错。关键在于，能力的另一个美妙之处在于——我在这方面的核心信念是：工程学才是看待这一切的正确视角，因为它能让你摆脱主观术语。人们谈论意识这个那个，那些东西很难定义，人们会从哲学角度争论不休。

Yep. You know, what's really key, and and the other nice thing about competency is that so so my my central belief in all of this is that engineering is the right perspective on all of this stuff because it gets you away from subjective terms. Know, people talk about sentience and this and that. Those things are very hard to define. People argue about them philosophically.

我认为像'能力'、'目标追求'这样的工程学术语在实践中极其有用，因为你看到就能明白。如果它能帮助你构建系统——如果我选对层级，说这个东西具有某种能力水平，比如像恒温器或更高级的恒温器，或是其他各种复杂系统——只要能帮助我控制、预测和构建这类系统，那就无需多言，不必再争论哲学问题。我喜欢能力的这种可量化特性，你必须明确声明'在什么方面有能力'。如果我说它有目标，问题就是目标是什么以及如何确认？我会说：因为每次偏离特定状态时，它都会消耗能量回归该状态——这就是目标，我们可以量化并客观看待它。

I think that engineering terms like competency, like pursuit of goals, right, all of these things are empirically incredibly useful because you know it when you see it. And if it helps you build, right, if I can pick the right level, I say this thing has, I believe this is x level of competency, I think it's like a thermostat, or I think it's like a better thermostat, or I think it's a you know various other kinds of, know, different kinds of complex systems. If that helps me to control and predict and build such systems, then that's all there is to say, there's no more philosophy to argue So I like competency in that way because you can quantify, you have to, in fact you have to, you have to make a claim competent at what, and then or if I say, if I tell you it has a goal, the question is what's the goal and how do you know? And I say well because every time I deviated from this particular state that's what it spends energy to get back to. That's the goal, and we can quantify it and we can be objective about it.

我们通常不这么思考。我有个演讲叫《为什么机器人不会得癌症？》，原因在于除了少数例外，我们的架构都是由一堆愚钝的部件组成，寄望于组合后能产生智能。但生物不同，每个层级都有其意图，最终结果是各层级内外合作与竞争的共同产物。例如胚胎发育时，你的组织器官就在相互竞争。

So we're not used to thinking about this. Give a talk sometimes called why don't robots get cancer? And the reason robots don't get cancer is because generally speaking, with a few exceptions, our architectures have been, you've got a bunch of dumb parts and you hope that if you put them together the overlying machine will have some intelligence and do something rather, but the individual parts don't care, they don't have an agenda. Biology isn't like that, every level has an agenda and the final outcome is the result of cooperation and competition both within and across levels. So for example during embryogenesis your tissues and organs are competing with each other.

这其实是发育过程中非常重要的一环——它们相互竞争是有原因的，不只是互助，更是在争夺信息资源和有限的新陈代谢条件。回到你另一点：相比人类某些努力，这看起来效率极高。但要注意，这里每个层级都会扭曲下级的选择空间，使底层部件看不见全局几何形态（我用的是相对论的术语，空间确实被弯曲了）。高层级扭曲选择空间后，底层只需沿着浓度梯度行动，它们无需也不可能知晓全局。但只要空间弯曲得当，它们按局部最优行动时，实际上就在执行你的意志。

It's actually a really important part of development, there's a reason they compete with each other, they're not all just of helping each other, they're also competing for information, for metabolic, for limited metabolic constraints. But to get back to your other point, which is this seems like really efficient and good and so on compared to some of our human efforts, we also have to keep in mind that what happens here is that each level bends the option space for the level beneath, so that your parts basically they don't see the the geometry, so I'm using, and I think I take this seriously terminology from relativity, right, where the space is literally bent. So the option space is deformed by the higher levels so that the lower levels, all they really have to do is go down their concentration gradient. They don't have to, in fact they don't, they can't know what the big picture is. But if you bend the space just right, if they do what locally seems right, they end up doing your bidding.

它们最终会在更高维度实现最优。反过来说，由于组件擅长完成本职工作，你作为高层级就无需计算底层控制细节，只需弯曲空间。你不知道也不关心它们具体如何实现。举个超级简单的例子：

They end up doing things that are optimal in higher space. Conversely, because the components are good at getting their job done, you as the higher level don't need to try to compute all the low level controls. All you're doing is bending the space. You don't know or care how they're going to do it. I'll give you a super simple example.

在蝌蚪研究中我们发现，蝌蚪需要变成青蛙，要从蝌蚪头部转变为青蛙头部，必须重新排列面部结构。眼睛需要前移，下颌要突出，鼻孔位置改变——所有部位都在移动。过去认为，由于所有蝌蚪和青蛙看起来都一样，只要每个部位都按正确方向和距离移动，就能得到标准青蛙形态。于是我们决定验证这个假设——我认为这个系统实际上比想象得更智能，我们是怎么做的呢？

In the tadpole, we found that okay, so so tadpoles need to become frogs and to become to go from a tadpole head to a frog head, you have to rearrange the face. So the eyes have to move forward, the jaws have to come out, the nostrils move, like everything moves. It used to be thought that because all tadpoles look the same and all frogs look the same, if you just remember if every piece just moves in the right direction the right amount, then you get your you get your frog right. So we decided to to test. We I had this hypothesis that I thought I thought actually the system is probably more intelligent than that, so what did we do?

我们培育了所谓的'毕加索蝌蚪'。这些蝌蚪的所有器官都错位了：眼睛长在后脑勺，下颌歪在侧面，完全混乱排列。但猜猜它们变成了什么？它们发育成了相当正常的青蛙，因为所有器官会沿着新路径移动重组，直到形成正确的青蛙面部结构后停止。这让我们联想到进化过程。

We made what we call Picasso tadpoles. So these are so everything is scrambled, the eyes are on the back of the head, the jaws are off to the side, everything is scrambled. Well, guess what they make? They make pretty normal frogs because all the different things move around in novel paths configurations until they get to the correct froggy sort of frog face configuration, then they stop. So the thing about that is now imagine evolution, right?

假设发生某种突变——就像所有突变那样——会同时产生多种影响。可能带来益处，但也可能让嘴巴歪到一边。如果没有这种多尺度适应能力（你能明白这个逻辑），生物体就会死亡——适应度归零因为它无法进食，也就永远无法探索这个突变的其他益处。只能等待出现不改变嘴部位置的新突变，这极其困难。这样适应度地形会变得异常崎岖，进化将耗费永恒时间。

So you make some sort of mutation and it does, like every mutation, it does many things. So something good comes of it, but also it moves your mouth off to the side. Now if there wasn't this multi scale competency, you can see where this is going, if there wasn't this multi scale competency the organism would be dead, your fitness is zero because you can't eat, and you would never get to explore the other beneficial consequences of that mutation. You'd have to wait until you find some other way of doing it without moving the mouth, that's really hard. So the fitness landscape would be incredibly rugged, evolution would take forever.

这个机制运作良好的关键原因在于：即便出现异常也不用担心，嘴巴最终会找到正确位置。这样就能继续探索突变的其他可能性。这意味着所有原本有害的突变现在都变成中性，因为各部位的自主调节能力弥补了各种异常。发育过程中的噪音、环境变异等因素都被这种能力所抵消。这很神奇，对吧？

The reason it works, one of the reasons it works so well, is because you do that, no worries, the mouth will find its way where belongs, right? So now you get to explore. So what that means is that all of these mutations that otherwise would be deleterious are now neutral, because the competency of the parts make up for all kinds of things. So all the noise of development, all the variability in the environment, all these things, the competency of the parts makes up for it. So that's fantastic, right?

这一切都很棒。但将之与人类行为对比时还需注意：每个组成部分在各自维度都有目标，通常很少考虑其他层级的利益。举个简单例子：作为复杂系统的你会去练柔术或攀岩，即便会磨破手掌细胞，但整个系统（你）感到快乐。

That's all great. The only other thing to remember when we compare this to human efforts is this. Every component has its own goals in various spaces, usually with very little regard for the welfare of the other levels. So so as a simple example, you know, you as a as a complex system, you will go out and you will do, you know, jujitsu or whatever. You'll have some go you have to go rock climbing and scrape a bunch of cells off your hands, and then you're happy as a system.

对吧？你归来时达成了目标感到愉悦，但那些细胞已经死亡消失了。

Right? You come back and you've you've accomplished some goals and you're really happy. Those cells are dead. They're gone. Right?

你考虑过那些细胞的感受吗？并没有。你只是有些淤青罢了。

Did you think about those cells? Not really. Right? You had some you had some bruising, alps.

自私的混蛋。没错。

Selfish SOB. Right.

就是这样。所以需要记住的是，我们从历史中知道，仅仅成为一个集体是不够的，因为这个集体目标相对于个体福祉而言，是一个极其开放的问题。

That's it. And so and so that's the thing to remember is that, you know, and we know this from from history is that is that just being a collective isn't enough because what the goals of that collective will be relative to the welfare of the individual parts is a massively open question.

为达目的不择手段。我告诉你，斯大林确实有所领悟。不。

Justify the means. I'm telling you, Stalin was onto something. No.

这就是危险所在。

So that's the danger.

但我们可以精确地...这正是我们人类面临的危险，我们必须构建这样的伦理体系：在这个体系下，我们不认真对待完整的生物学机制并将其应用于世界运行方式——这是我们划下的一条有趣界线。在某种意义上，我们拒绝了这个造就我们的世界。当我们构建人类社会时，这个国家建立在'人人生而平等'的理念上，这是个多么迷人的想法。

But we can exactly. That's the danger of for us humans, we have to construct ethical systems under which we don't take seriously the full mechanism of biology and apply it to the way the world functions, which is which is an interesting line we've drawn. The world that built us is the one we reject in some sense Yeah. When we construct human societies. The idea that this country was founded on that all men are created equal, That's such a fascinating idea.

就像你在与自然对抗。你在说，这里还有比等级能力架构更重要的东西。是的。但你说了很多有趣的观点。

It's like you're fighting against nature. You're saying, well, there's something bigger here than Yeah. A hierarchical competency architecture. Yeah. And but there's so many interesting things you said.

从算法角度看，扭曲选择空间的行为确实非常深刻。因为如果你观察当今AI系统的构建方式，就像我说的，一个拥有机器人等的大系统有目标，它会越来越擅长优化和实现那个目标。但如果生物学构建了一个层级系统，其中每个部分都在进行计算和实现目标——不仅如此，它还显得有些愚蠢。在扭曲的选择空间里，它只是做着最容易的事情。

So from an algorithmic perspective, the act of bending the option space, That's really that's really profound. Because if you look at the way AI systems are built today, there's a big system, like I said, with robots and has a goal, and it gets better and better at optimizing that goal, at accomplishing that goal. But if biology built a hierarchical system where everything is doing computation and everything is accomplishing the goal, not only that, it's kind of dumb. You know? With the with the limited with the bent option space, it's just doing the thing that's the easiest thing for Yep.

从某种意义上说，是的。这种方式让你能够像叠乌龟一样，层层堆叠系统，最终形成极其智能的整体。

In some sense. Yeah. And somehow, that allows you to have turtles on top of turtles, literally, dump systems on top of dump systems that as a whole creates something incredibly smart.

没错。每个系统在其特定问题领域都具备一定程度的智能。比如细胞会在生理和转录空间解决问题，我可以举些有趣的例子。而集体则在解剖空间解决问题，比如形成生物体、生长血管等。整个身体又在解决其他层面的问题。

Yeah. I mean, every system is has some degree of intelligence in its own problem domain. So cells will have problems they're trying to solve in physiological space and transcriptional space, and then I can give you some cool examples of that. But the collective is trying to solve problems in anatomical space, right, and forming a creature and growing your blood vessels and so on. And then the collect the the the the whole body is solving yet other problems.

它们可能涉及社交空间、语言空间、三维空间等等。群体的解题领域甚至可能是金融空间之类的。当今大多数AI的主要差异在于架构的扁平化，以及它们是由外部构建的——它们的边界和...（省略重复词）很大程度上我们的技术都是这样：造出的机器明确知道自己的传感器、效应器，以及与外界的界限。

They may be in social space and linguistic space and three-dimensional space and and who knows, you know, the group might be solving problems in in, you know, I don't know, some sort of financial space or something. So one of the major differences with with most with most AIs today is is a, the the kind of flatness of those architecture, but also of the fact that they are constructed from outside. Their their borders and their you know, so so if you for so to a large extent, and of course, are counterexamples now, but but to a large extent, our technology had been such that you create a machine or a robot, it knows what its sensors are. It knows what its effectors are. It knows the boundary between it and the outside world.

这些都是外部赋予的。而生物学是从零开始构建的。机器人学中最经典的例子是Josh Bongard 2006年的研究：他制造的机器人起初并不知道自身形态，像婴儿般笨拙探索，通过假设建立认知。

All of this is given from the outside. Biology constructs this from scratch. Now the best example of this that that originally in in robotics was actually Josh Bongard's work in 2006 where he made these these robots that did not know their shape to start with. So like a baby, they sort of floundered around. They made some hypotheses.

比如'我这样移动，也许我有轮子或六条腿'，它们会建立模型最终学会爬行。

Well, I did this and I moved in this way. Well, maybe I'm whatever. Maybe I have wheels or maybe I have six legs or whatever. Right? And they would make a model and then eventually it would crawl around.

这确实很棒，属于自创生的一部分。但我们可以更进一步——有些人正在研究，我们也在探索——即回溯更基础的问题：你甚至不知道自己的传感器在哪，分不清自身与外部世界的边界。

So that's I mean, that's really good. That's part of the auto poesis. But we can go a step further and some people are doing this and then we're sort of working on some of this too, is this idea that let's even go back further. You don't even know what sensors you have. You don't know where you end and the outside world begins.

你仅有的只是主动推理这类机制——试图最小化意外感。你面临代谢限制，没有无限能量，也没有充足时间思考所有想思考的问题。

All you have is is certain things like active inference, meaning you're trying to minimize surprise. Right? You have some metabolic constraints. You don't have all the energy you need. You don't have all the time in the world to to to think about everything you want to think about.

这意味着你不能成为微观还原主义者。要知道，面对所有这些涌入的数据，你必须进行粗粒度处理，比如把这一堆信息统称为‘猫’，把那些信息标记为‘我不想跌落的桌沿’。我不需要了解微观状态，我需要知道的是如何最优地划分我的世界——顺便说一句，那边那个东西，就是我。

So that means that you can't afford to be a micro reductionist. You know, all this data coming in, you have to coarse grain it and say, going take all this stuff, I'm going to call that a cat. I'm going take all this, I'm going to call that the edge of the table I don't want to fall off of. And I don't want to know anything about the microstates. What I want to know is what is the optimal way to cut up my world, and by the way this thing over here, that's me.

之所以说那是我，是因为我对它的控制力远超过其他事物。于是你开始构建自我：先对外部世界建模，再向内转，开始建立自我模型。对吧？这立刻涉及能动性与控制问题——在代谢约束下（即能量有限），你必须高效运作，这迫使你开始讲述关于粗粒度行为主体的叙事。

And the reason that's me is because I have more control over this than I have over any of this other stuff, And so now you can begin to write. So that self construction, that figuring out making models of the outside world and then turning that inwards and starting to make a model of yourself. Right? Which immediately starts to get into issues of of agency and control because in order to if if you are under metabolic constraints, meaning you don't have the energy, right, that all the energy in the world, you have to be efficient, that immediately forces you to start telling stories about coarse grained agents that do things. Right?

你不可能像拉普拉斯恶魔那样计算所有可能状态，必须进行粗粒度处理，将事物归类为需要规避或接近的行为主体——无论是配偶、食物还是其他。就在最基础的简单生物开始对行为主体建模时，自由意志模型的雏形就此诞生。因为你将周围世界视为具有能动性，然后反观自身：等等，我也有能动性？我能做出行动？

You don't have the energy to like Laplace's demon, you know, calculate every every possible state that's going to happen, you have to you have to coarse grain and you have to say that is the kind of creature that does things, either things that I avoid or things that I will go towards, that's a mate or food or whatever it's going be. And so right at the base of simple, very simple organisms starting to make models of agents doing things, that is the origin of models of free will basically. Right? Because you see the world around you as having agency, and then you turn that on yourself and you say, wait, I have agency too. I can I do things?

对吧？接着你开始决定自己的行为。所有这些都可以用一个模型解释：它们都源于早期对自我定义的迫切需求，以及在能量最优化空间内采取行动的必要性。

Right? And and then you make decisions about what you're going to do. So all of this one one model is to view all of those kinds of things as being driven by that early need to determine what you are and to do so and to then take actions in the most energetically efficient space possible.

对吧？所以当你试图简化环境、构建美好叙事时，自由意志就显现了。

Right? So free will emerges when you try to simplify, tell a nice narrative about your environment.

我认为这个解释非常合理。是的。

I think that's very plausible. Yeah.

你认为自由意志是幻觉。那么你其实是在暗示它是一种有用的‘认知捷径’。

You think free will is an illusion. So so you're kind of implying that it's a useful hack.

嗯，我要说两点。第一点是我认为，非常合理地说，任何自我组织或自我构建的生物体或代理，无论是否具有生物性，在能量约束下都会相信自由意志。我们稍后会讨论它是否真正拥有自由意志，但我认为这必然会导致一种将自我与外界视为具有能动性的观点。我认为这是不可避免的。

Well, I'll say two things. The first thing is I think I think it's very plausible to say that any organism that self or any agent that self whether it's biological or or not, any agent that self constructs under energy constraints is going to believe in free will. We'll we'll get to whether it has free will momentarily, but but I think but I think what what it definitely drives is a view of yourself and the outside world as an agential view. I think that's inescapable.

所以即使对原始生物来说也是如此。

So that's true for even primitive organisms.

我认为是的。不过显然需要按比例缩小理解——它们不具备我们这种复杂的元认知能力，无法进行长期规划或思考自由意志等问题，但这种能动性感知...

I think so. I think this Now, now they don't have Now, obviously you have to scale down, right? So they don't have the kinds of complex metacognition that we have so they can do long term planning and thinking about free will and so on, but but The sense

对于完成任务（无论简单还是复杂）确实非常有用。

of agency is really useful to accomplish those tasks, simple or complicated.

没错。适用于各种空间维度，不仅是明显的三维空间。人类非常擅长探测三维世界中以中等速度运动的中等大小物体，对吧？

That's right. In in all kinds of spaces, not just in in obvious three-dimensional space. I mean, we're very good. The the thing is humans are very good at detecting of medium sized objects moving at medium speeds in the three-dimensional world. Right?

我们看到保龄球和老鼠时，能立刻分辨差异并知道该如何应对，对吧？

We see a bowling ball and we see a mouse and we immediately know what the difference is, right, and how we're going

主要是你能吃或被吃的东西。

Mostly things you can eat or get eaten by.

是的，是的。那是我们的训练集对吧？从小时候起，你的训练集就是这段有限经历中的视觉数据。但想象一下，如果我们从出生起就天生能感知血液化学变化——就像视觉一样能直接感受血液化学，拥有高带宽连接，能感知所有器官活动，比如胰腺、肝脏等等。

Yeah. Yeah. That's our training set, right? From the time you're little your training set is visual data on this like little chunk of your experience. But imagine if from the time that we were born we had innate senses of your blood chemistry, If you could feel your blood chemistry the way you can see, right, you had a high bandwidth connection and you could feel your blood chemistry and you could see, you could sense all the things that your organs were doing, so your pancreas, your liver, all the things.

如果我们具备这种能力，就能在生理领域非常擅长检测智能。我们会清楚各个器官在处理预期刺激时展现的智能水平。但现实是我们在这方面非常糟糕，甚至很多人认为谈论其他领域的智能是疯狂的，因为我们只理解运动层面的智能。

If we had that, we would be very good at detecting intelligence in physiological space. We would know the level of intelligence that our various organs were deploying to deal with things that were coming to anticipate, stimulate, to you know, but we're just terrible at that. In fact people don't even, you you talk about intelligence in these other spaces and a lot of people think that's just crazy because all we're all we know is motion.

我们确实能获取这些信息，所以从进化角度完全有可能构造出能感知这些的有机体

We do have access to that information, so it's it's actually possible that so evolution could if we wanted to construct an organism that's able to perceive

毫无疑问。

Most certainly.

血液在体内流动的方式。就像你见到老友时会说'嘿，最近怎么样？老婆孩子都好？'那样，你也能感受到与肝脏的连接。

The flow of blood through your body. The way you see an old friend and say, yo. What's up? How's the wife and the kids? In that same way, you would see you would feel like a connection to the liver.

对，对。我觉得...

Yeah. Yeah. I think, you know

或许还能感知别人的肝脏？还是仅限于自己的？毕竟你无法接触他人的肝脏。

Maybe other people's liver or no? Just your own? Because you don't have access to other people's liver.

还没有，但你可以想象一些非常有趣的关联，对吧？不过有些

Not yet, but you could imagine some really interesting connection. Right? But some

性选择？比如，哦，那女孩的肝脏真不错。好吧，这很酷。她血液流动的方式，血液的动态非常有趣。这很新颖。

Sexual selection? Like, oh, that girl's got a nice liver. Well, that's cool. The the the way her her blood flows, the the dynamics of the blood is very interesting. It's novel.

我从未见过这样的。

I've never seen one of those.

但你知道，这正是我们试图半吊子评估的东西——当我们通过面部对称性等来判断美的时候。那正是对这种情况的半吊子评估。是的，正是如此。因为如果你的细胞不能充分合作保持生物体的对称性

But, you know, that's that's exactly what we're we're trying to half ass when we when we judge judgment of of beauty by facial symmetry and so on. That's that's a half assed assessment of exactly that. Yeah. Of exactly that. Because if your cells could not cooperate enough to keep your your organism symmetrical

是啊。

Yeah.

你知道，你可以推断出还有什么地方不对劲。对吧？就像，那是非常基础的。

You know, you can make some inferences about what else is wrong. Right? Like, that's a that's a very, you know, that's a very basic.

有意思。是啊。所以在某种深层意义上，我们实际上就是在做这个。我们试图推断这个生物系统的健康状况——我们用'健康'这个词，但本质上是在看这个生物系统的功能性如何，这样我就能与之结合并繁衍后代？

Interesting. Yeah. So that in some deep sense, actually, that is what we're doing. We're trying to infer how health we use the word healthy, but, basically, how functional is this biological system I'm looking at so I can hook up with that one and make offspring?

是的，是的。那么，他们的基因组学未来能给我提供哪些可能有用的硬件呢？

Yeah. Yeah. Well, what kind of hardware might their genomics give me that that might be useful in the future?

我在想为什么进化没有给我们更高分辨率的信号。比如，为什么孔雀会有羽毛这种展示？这看起来不像是一个低带宽的选择信号。

I wonder why evolution didn't give us a higher resolution signal. Like, why the whole peacock thing with the feathers? It doesn't seem it's a very low bandwidth signal for selection.

我会没事的。而且我对这些东西并不是专家。

I'm gonna alright. And I'm not an expert on on this stuff.

关于孔雀吗？

On peacocks?

嗯，你知道，不是。明白吗？但我会尝试解释原因。我认为这是因为这是一场军备竞赛。你看，你不想让所有人都了解你的一切。

Well, you know, no. You know? But but I'll take a stab at at the reason. I think that it's because it's an arms race. You see, you don't want everybody to know everything about you.

所以我认为，事实上这场军备竞赛还有另一个有趣的部分，如果你仔细想想，最具适应性、最易进化的系统是那种具有最高层次自上而下控制的系统。如果很容易对一群细胞说'再长一根手指'，而不是'好吧，这里有一万个基因表达变化需要你改变手指'，对吧？具有良好自上而下控制且拥有记忆的系统——顺便说，我们需要回到那个问题，关于记忆存储位置等我没回答的问题——一个能运用所有这些的系统确实具有高度进化能力，这很棒。但猜猜看，它也极易被寄生虫、各类作弊者、同种个体所劫持。

So I think that as much as as much as and and in fact, there's another interesting part of this arms race, which is if you think about this, the most adaptive evolvable system is one that has the most level of top down control. If it's really easy to say to a bunch of cells make another finger versus okay, here's 10,000 gene expression changes that you need to do to change your finger, right? The system with good top down control that has memory, we need to get back to that, by the way, that's a question I neglected to answer about where the memory is and so on. A system that uses all of that is really highly evolvable and that's fantastic. But guess what, it's also highly subject to hijacking by parasites, by cheaters of various kinds, by conspecifics.

就像我们发现的那样，这又回到了模式记忆的故事上。在这些涡虫身上，有一种细菌寄生，这种细菌能影响涡虫会长出多少个头，因为它劫持了那个控制系统，能制造一种化学物质，基本上与计算你应该有多少个头的系统对接，它们能让涡虫长出两个头。所以你可以想象，如果你...你希望自己的各个部分能相互理解，但又不希望太容易被理解，因为那样就太容易被控制了。因此我认为，正是这种对立压力阻止了我们成为那种超高带宽的存在——让我们只需看某人一眼就能完全了解对方

Like we we found that and that goes back to the story of the pattern memory, in these planaria, there's a bacterium that lives on these planaria, that bacterium has an input into how many heads the worm is going have because it hijacks that control system and it's able to make a chemical that basically interfaces with the system that calculates how many heads you're supposed to have and they can make them have two heads. And so you can imagine that if you are two so you want to be understandable for your own parts to understand each other, but you don't want be too understandable because you'll be too easily controllable. And so I think that that my guess is that that that that opposing pressure keeps us from being a super high bandwidth kind of thing where we can just look at somebody and know know

这一切就像是生物版的德州扑克。对，你亮出部分牌，同时隐藏其他牌。嗯。这其中包含虚张声势等各种策略。

everything about So it's a kind of biological game of Texas hold them. Yeah. You're showing some cards and you're hiding other cards. Mhmm. And there's part of it and there's bluffing and there's all that.

可能有些物种会过度使用虚张声势的策略。孔雀大概就属于这类。我记得有本关于鸟类美丽进化的书，但不确定是我读过还是写过书评——达尔文是不是讨论过这个？

Then there's probably whole species that would do way too much bluffing. That's probably where peacocks fall. I there there's a there's a book that I don't remember if I read or or if I if I wrote if I read summaries of the book, but it's about evolution of beauty and birds. Where is that from? Is that a book, or does Richard Dawkins talk about it?

但本质上，某些物种开始过度追求美丽。不，不是过度，而是出于某种原因单纯选择美丽。这个观点其实有据可依——我现在逐渐想起来了。

But, basically, there's some species start to, like, over select for beauty. Not over select. They just, for some reason, select for beauty. There is a case to be made. Actually, now I'm starting to remember.

我记得达尔文本人就论证过可以仅凭美丽进行选择。没错，美丽发展到某个程度就不再代表任何深层...正是这样...生物学真相，而开始为美而美。

I think Darwin himself made a case that you can select based on beauty alone. So Yeah. That beauty there's a point where beauty doesn't represent some underlying That's right. Biological truth. You start to select for for beauty itself.

更深层的问题是：美丽是否具有进化价值？这个想法很有趣——我们能否完全脱离生物本质，单纯欣赏某种抽象概括？让我再回忆下，这个概念非常有意思：细胞集合如何形成记忆？生物系统如何存储信息？

And I think the the deep question is there is some is there some evolutionary value to beauty? But it's an interesting kind of thought that this can we deviate completely from the deep biological truth to actually appreciate some kind of the the summarization in itself. Let me get back to memory because this is a really interesting idea. How do a collection of cells remember anything? How do biological systems remember anything?

这种记忆机制与我们人类认知引擎中的记忆有何相似之处？

How is that akin to the kind of memory we think of humans as having within our big cognitive engine?

没错。理解生物电的切入点就是思考：神经元和大脑用来运行惊人问题解决能力的这些精妙电路从何而来？它们不可能凭空出现，必然是从更古老的细胞网络能力演化而来——比如控制形态空间和身体结构的网络。

Yeah. One of the ways to start thinking about bioelectricity is to ask ourselves where did neurons and all these cool tricks that the brain uses to run these amazing problem solving abilities on and basically an electrical network. Right, where did that come from? It didn't just evolve, you know, appear out of nowhere, it must have evolved from something. And what it evolved from was a much more ancient ability of cells to form networks to solve other kinds of problems, for example to navigate morphospace, to control the body shape.

因此，神经元的所有组成部分——离子通道、神经递质机制、电突触等——其存在时间远早于大脑，甚至早于神经元本身，事实上比多细胞生物的出现还要古老。就连细菌生物膜也是如此，加州大学圣地亚哥分校关于细菌生物膜中类脑动态的研究非常精彩。进化很早就发现，电网络在记忆存储、远距离信息整合以及图像识别等各类优化任务中表现卓越，这些功能远在大脑出现之前就已存在。

And so all of the components of neurons, so ion channels, neurotransmitter machinery, electrical synapses, all this stuff is way older than brains, way older than neurons, in fact older than multicellularity. And so it was already that even even bacterial biofilms, there's some beautiful work from UCSD on on on brain like dynamics in bacterial biofilms. So evolution figured out very early on that electrical networks are amazing at having memories, at integrating information across distance, at different kinds of optimization tasks, you know, image recognition and so on, long before there were brains.

能否请您退一步或重新解释一下？什么是生物电？什么是生物化学？什么是电网络？我认为生物学界大多将化学物质视为使整个系统运作的信号机制。

Can you actually step back or return to it? What is bioelectricity? What is biochemistry? What is what are electrical networks? I think a lot of the biology community focuses on the chemicals as the signaling mechanisms that make the whole thing work.

您在很大程度上——如果我理解有误请指正——独树一帜地专注于生物电研究，即利用电力进行信号传递。可能还存在机械...

You have, I think, to to a large degree uniquely, maybe you can correct me on that, have focused on the bioelectricity, which is using electricity for signaling. There's also probably mechanical

当然。确实如此。

Sure. Sure.

生物力学在敲门。那么区别究竟是什么？什么是电网络？

Biomechanical knocking on the door. So what what what's the difference and what's an electrical network?

是的。我想首先公正地承认前人的贡献。早在1903年，甚至十九世纪末期，人们就已经开始思考电现象在生命中的重要性。我绝非第一个强调电重要性的人。三十年代、四十年代，再到七八十年代和九十年代，曾涌现过几波生物电研究的先驱，他们在这个领域做出了惊人成就。

Yeah. So I want to make sure and and kind of give credit where credit is due. So so as far back as nineteen o three and probably late eighteen hundreds already, people were thinking about the importance of electrical phenomena in life. So I'm for sure not the first person to stress the importance of electricity. People there waves of research in the thirties, in the forties, and then again in the kind of seventies, eighties and nineties of sort of the pioneers of bioelectricity who did some amazing work on all this.

我认为我们的新贡献在于跳出了传统认知——稍后我会解释什么是生物电——不再将其视为理解生理和发育过程中需要关注的物理现象，而是开始将其视作一种特殊的计算层级，这种层级能让我们触及组织的底层认知，将发育生物物理学与计算认知等概念相结合。这才是我们的创新之处。但生物电的讨论由来已久，接下来我会给出定义：单个细胞具有细胞膜，膜上存在称为离子通道的蛋白质，这些蛋白质允许带电分子（如钾、钠、氯离子）在特定条件下进出细胞。当这些离子失衡时，膜两侧就会形成电压梯度。所有活细胞都会努力维持特定的跨膜电压差，并为此消耗大量能量。

I think what we've done that's new is to step away from this idea that, and I'll describe what the bioelectricity is, to step away from the idea that well here's another piece of physics that you need to keep track of to understand physiology and development, and to really start looking at this as saying no this is a privileged computational layer that gives you access to the actual cognition of the tissue, of basal cognition, merging that developmental biophysics with ideas in cognition of computation and so on, think that's what we've done, that's new. But people have been talking about bioelectricity for a really long time and so I'll define that. So what happens is that if you have a single cell, a cell has a membrane, in that membrane are proteins called ion channels and those proteins allow charged molecules, potassium, sodium, chloride, to go in and out under certain circumstances. And when there's an imbalance of those ions there becomes a voltage gradient across that membrane. And so all cells, all living cells, try to hold a particular kind of voltage difference across the membrane and they spend a lot of energy to do so.

当你现在，那是一个单细胞。当你有多个细胞时，相邻的细胞可以通过多种方式互相传递它们的电压状态，其中一种就是所谓的间隙连接，它基本上就像一个小型潜艇舱口，只是对接起来，对吧，离子可以从一侧流向另一侧，反之亦然。

When you now, so that's a single cell. When you have multiple cells, the cells sitting next to each other, they can communicate their voltage state to each other via a number of different ways, but one of them is this thing called a gap junction, which is basically like a little submarine hatch that just kind of docks, right, and the ions from one side can flow to the other side and vice versa.

这进化出来不是很不可思议吗？这不是很疯狂吗？因为这东西原本并不存在。

Isn't it incredible that this evolved? Isn't isn't that wild? Because that didn't exist.

没错。这必须是进化而来的。它

Correct. This had to be this had to be evolved. It

必须被发明出来。

had to be invented.

正是如此。

That's right.

有人在海洋中发明了电。希望这个被发明出来。

Somebody invented electricity in the in the ocean. Wanted this to get invented.

是的。所以，我是说，这确实令人难以置信。发现间隙连接的人，沃纳·洛温斯坦，我去拜访过他，他那时已经非常老了。一个人类？是他发现了它们。

Yeah. So so so, I mean, it's an it it is it is incredible. The guy who discovered gap junctions, Werner Lowenstein, I visited him, he was he was really old. A human being? He discovered them.

我在想这个问题，因为真正发现它们的人可能生活在40亿年前。说得好。所以你是把功劳归于应得之人。我只是...

Was wondering what because who really discovered them lived probably 4,000,000,000 years ago. Good point. So you're you're give credit where creditors do. I'm He just

他重新发现了间隙连接，但大约二十年前我在伍兹霍尔拜访他时，他告诉我他正在写一本书。遗憾的是他去世了，我想这本书最终未能完成。他当时在写关于间隙连接与意识的书，那本该是本了不起的著作——因为间隙连接是魔法般的，我稍后解释原因。关键在于，无论是离子通道还是间隙连接，许多都具有电压敏感性。这种电压敏感的电流传导就是晶体管原理，一旦实现这点，你就能触及数学真理的柏拉图领域里晶体管的所有精妙功能。当细胞形成网络时，它们不仅能相互交流，还能传递电压差异信号。对神经科学家来说这是老生常谈，大脑中的动作电位就是这种电信号。

he he rediscovered he rediscovered gap junctions, but when I visited him in in Woods Hole maybe twenty years ago now, he told me that he was writing, and unfortunately he he he passed away, and I think this this book never got written. He was writing a book on on gap junctions and consciousness, and I think it would have been an incredible book because gap junctions are magic, I'll explain why in a minute, what happens is that, just imagine, the thing about both these ion channels and these gap junctions is that many of them are themselves voltage sensitive. So that's a voltage sensitive current conductance, that's a transistor, and as soon as you've invented one, immediately you now get access to, from this platonic space of mathematical truths, you get access to all of the cool things that transistors do. So now when you have a network of cells, not only do they talk to each other, but they can send messages to each other and the differences of voltage can propagate. Now to neuroscientists this is his old hat because you see this in the brain, this action potential is the electricity.

现在有令人惊叹的成像技术，比如用斑马鱼这类透明生物做实验时，你能直接观察到它做觅食决策时的神经放电，非常神奇。其实你全身细胞时刻都在进行类似活动，只是速度慢得多。神经元能做到的事，体内其他细胞基本也都能做到，只是时间尺度不同。当大脑思考三维空间的问题时，胚胎细胞则在解决解剖学空间的问题——它们会有类似'我们该长几根手指？'这样的记忆。

You can, they these awesome movies where you can take a zebra, like a transparent animal like a zebrafish, you can literally look down and you can see all the firings as the fish is like making decisions about what to eat and things like this, right, it's amazing. Well your whole body is doing that all the time, just much slower. So there are very few things that neurons do that other cells, that all the cells in your body don't do. They all do very similar things, just on a much slower timescale, and whereas your brain is thinking about how to solve problems in three-dimensional space, the cells in an embryo are thinking about how to solve problems in anatomical space. They're trying to have memories like hey how many fingers are we supposed to have?

现在有几根？要怎么从现状达成目标？这就是它们思考的问题。间隙连接的魔力在于——想象远古时代两个细胞的情景：

Well how many do we have now? What do we do to get from here to there? That's the kind of problems they're thinking about. And the reason that gap junctions are magic is, imagine, right, from earliest time. Here are two cells.

这个细胞怎么通讯？最简单的方式是释放化学信号，飘过去撞击那个细胞的受体。由于信号来自外部，接收细胞能明确识别这是外来信息。无论传递什么内容，它都知道'这不是我的信息'，可以选择信任、忽略或处理，但清楚其外部来源。

This cell, how can they communicate? Well the simple version is this cell could send a chemical signal, it floats over and it hits a receptor on this cell. Because it comes from outside, this cell can very easily tell that that came from outside. Whatever information is coming, that's not my information, that information is coming from the outside. I can trust it, I can ignore it, I can do various things with it, whatever, but I know it comes from the outside.

现在想象两个细胞通过间隙连接相连。假如这个细胞受刺激产生钙离子峰，信号通过间隙连接传到那个细胞。这个信号没有归属元数据，接收细胞无法辨别来源，因为信号形态与其自身记忆产生的信号完全一致。间隙连接在某种程度上抹去了数据的归属信息，当两个个体共享记忆却无法区分记忆归属时，就形成了意识融合的雏形。

Now imagine instead that you have two cells with a gap junction between them. Something happens, let's say this cell gets poked, there's a calcium spike, and the calcium spike or whatever small molecule signal propagates through the gap junction to this cell. There's no ownership metadata on that signal. This cell does not know now that it came from outside because it looks exactly like its own memories would have looked like of whatever had happened, right? So gap junctions to some extent wipe ownership information on data, which means that if I can't if if you and I are sharing memories and we can't quite tell who the memories belong to, that's the beginning of a mind meld.

这就是认知升级的起点——从'你'和'我'转变为'我们'的共同体。

That's the beginning of a scale up of cognition from here's me and here's you to, no, now there's just us.

所以他们强制实施了一种集体智能，那就是间隙连接。

So they enforce a collective intelligence That's gap junctions.

没错。这有帮助。这只是开始。远非全部，但确实是个开端。

That's right. It helps. It's the beginning. It's not the whole story by any means, but it's the start.

系统的状态存储在哪里？

Where is state stored of the system?

有

There are

部分存储在间隙连接本身中吗？是在细胞里？那里

in part in the gap junctions themselves? Is it in the cells? There

和生物学中常见的情况一样，这涉及许多层次。有化学网络，比如基因调控网络，或者任何化学通路，其中不同化学物质相互激活和抑制，它们可以存储记忆。从动力系统的角度看，它们能存储记忆，能进入难以被打破的稳定状态，一旦进入，就形成了对已发生事件的永久或半永久记忆。还有细胞骨架结构，它们通过物理构型存储记忆。

are many, many layers to this as always in biology. So there are chemical networks. For example gene regulatory networks, or basically any kind of chemical pathway where different chemicals activate and repress each other, they can store memories. So in a dynamical system sense they can store memories, they can get into stable states that are hard to pull them out of, so that becomes once they get in, that's a memory, a permanent memory or semi permanent memory of something that's happened. There are cytoskeletal structures, right, that are physically they store they store memories in in physical configuration.

还有像触发器这样的电记忆，不存在物理变化对吧？比如我给学生们展示触发器例子，它存储0或1的原因不是硬件部件移动了，而是因为电流在装置一侧循环。如果我短暂地将另一侧保持高电压，它就会翻转状态，但硬件本身没有任何移动。信息是以稳定的动力学形式存在的，即使用X光照射也无法判断是0还是1，因为只能看到硬件位置，看不到系统的能量状态。所以也存在以完全相同方式维持的生物电状态，本质上就像易失性内存，存在于系统的电气状态中。

There are electrical memories like flip flops where there is no physical right? So so if you look, I I show my students this example as a flip flop, and the reason that it stores a zero or one is not because some some piece of the hardware moved, it's because there's a there's a cycling of the current in one side of the thing. If I come over and I hold, you know, I hold the other side to a to a high voltage for for, you know, a brief period of time, it flips over and now it's here, but the hard none of the hardware moved. The information is in a a stable dynamical sense, and if you were to x-ray the thing, you couldn't tell me if it was zero or one, because all you would see is where the hardware is, you wouldn't see the the energetic state of the system. So there are also so there are bioelectrical states that are held in that exact way, like like like volatile RAM basically, like in the in the electrical state of the

非常类似于计算机中存储记忆的不同方式。比如有RAM（随机存取存储器），还有硬盘。

very akin to the different ways the memory stored in a computer. So there's there's RAM. There's hard drives.

你可以做这样的类比，对吧？我觉得有趣的是，基于生物学原理，我们可以发展出更复杂的——我认为我们可以修订一些计算机工程方法，因为生物学中有许多我们尚未实现的精妙机制。但这个类比大体上是成立的，在很多方面都说得通。

You can make that mapping. Right? So I think the interesting thing is that based on the biology, we can have a more sophisticated you know, I I think we can revise some of our some of our computer engineering methods because there are some interesting things that biology does that we haven't done yet. But but you can but that map but that mapping is not bad. I mean, I think it in many ways.

是啊。我在想，因为计算机科学的根基是正确性验证的理念，我们编程追求完美可靠。而对未知条件的适应性和鲁棒性并不那么重要——而这正是生物学的强项。所以我不知道未来该如何从计算机系统过渡到生物系统。

Yeah. I wonder, because I mean, the way we build computers, at the root of computer science is the idea of proof of correctness, the way we we program things to be perfect, reliable. You know, this idea of resilience and robustness to unknown conditions is not as important. So that's what biology is really good at. So I don't know what kind of systems I don't know how we go from a computer to a biological system in the future.

没错。我认为生物学的本质就是在信息极其有限的情况下快速做出关键决策。你必须立即行动，风险极高，而你掌握的信息远不足以做到完美。所以生物学甚至不会试图追求完美或绝对正确。

Yeah. I think that, you know, the thing about biology is all about making really important decisions really quickly on very limited information. I mean, that's what biology is all about. You have to act, you have to act now, the stakes are very high and you don't know most of what you need to know to be perfect. So there's not even an attempt to be to be perfect or to get it right in any sense.

只有像主动推理、最小化意外、优化某些效率这类原则在指导整个运作过程。

There are just things like active inference, minimize surprise, optimize some some efficiency and and and some things like this that that guides the whole the whole business.

我私下也提过，安德烈·卡帕西是你作品的粉丝，他除了其他成就外还写了个很棒的博客。他提出了'软件2.0'的概念（不确定是否他首创），即在人工神经网络配置空间中进行编程。这是否预示着人类编程的未来方向——我们将减少Python式的编程，转而更多地调整类似生物系统的超参数，观察其运行并通过反馈循环让系统自我修正？

I mentioned too offline that somebody who's a fan of your work is Andre Kapathi, and he's amongst many things also writes occasionally a great blog. He came up with this idea, I don't know if he coined the term, but of software two point o, where the programming is done in the space of configuring these artificial neural networks. Is there some sense in which that would be the future of programming for us humans where we're less doing, like, Python like programming and more how would you how would that look like? But, basically, doing the hyperparameters of something akin to a biological system and watching it go and keeping adjusting it and creating some kind of feedback loop within the system so it corrects itself. Yeah.

然后我们观察它逐步实现既定目标。这就是你在《自然》论文中描述的'智能体'（dogs）的愿景吗？

And then we watch it over time accomplish the goals we wanted to accomplish. Is that kind of the the dream of the the dogs that you described in the nature paper?

是的。我的意思是，你刚才描述的正是我们在再生医学领域所做的努力，可以称之为一种躯体精神病学。其核心理念是不试图进行微观管理。想想当今许多药物的局限性，我们试图在分子通路的层面进行干预。

Yeah. I mean, that that's what you just painted is a is a very good description of our efforts at regenerative medicine as a kind of somatic psychiatry. So the idea is that you're not trying to micromanage. I mean, think about the limitations of a lot of the medicines today. We try to interact down at the level of pathways.

对吧？所以我们试图微观管理它。问题是什么？其中一个问题是，除了抗生素外，几乎所有药物一旦停用，问题就会立刻复发。你根本没有真正解决问题。

Right? So we're trying to micromanage it. What's the problem? Well, one problem is that for almost every medicine other than antibiotics, once you stop it, the problem comes right back. You haven't fixed anything.

你只是在处理症状。实际上除了抗生素外，你并没有真正治愈任何问题。这是第一个问题。第二个问题是会产生大量副作用，因为你在最底层进行干预。对吧？

You were addressing symptoms. You weren't actually curing anything again except for antibiotics. That's one problem. The other problem is you have massive amount of side effects because you were trying to interact at the lowest level. It's right?

就像我要...你知道，我试图通过改变铜的熔点来编程这台电脑。也许你可以这样做，但天哪，在硬件层面编程实在太困难了。所以我们开始认识到——顺便说这回到了你之前提到的观点——我们可以感知自身内部状态。那些练习这类技术的人...

It's like I'm gonna, you know, I'm gonna I'm gonna try to program this computer by changing the the melting point of copper. Like maybe you can do things that way, but my god, it's hard to to to program at the you're right at the at the hardware level. So what what I think we're we're we're starting to understand is that and and by the way, this goes back to what you were saying before about that that we could have access to our internal state. Right? So people who practice that kind of stuff.

对吧？比如瑜伽、生物反馈等。这些实践者都会不约而同地说：身体具有智慧等等。这两类观点完全吻合，因为它们本质上是正确的。一旦从这个角度思考，你就会意识到更好的控制点并不总是在最底层。

Right? So yoga and and and biofeedback and those. Those are all the people that uniformly will say things like, well, the body has an intelligence and this and that. Like those two sets overlap perfectly because because that's exactly right. Because once you start thinking about it that way you realize that the better locus of control is not always at the lowest level.

这就是为什么我们不用电烙铁编程的原因，对吧？我们应该利用现有的高级智能，这意味着要弄清楚：你的哪些组织具有学习能力？能学习什么？为什么某些药物服用一段时间后会失效？是习惯化吗？我们能否理解化学通路中的习惯化、敏感化、联想学习等现象？我认为当我们开始关注目标状态和子系统智能，而不是把所有治疗都当作从化学层面自下而上的微观管理时，我们将彻底改变用药和医疗的方式。

This is why we don't all program with a soldering iron, right? We take advantage of the high level intelligences that are there, which means trying to figure out okay, which of your tissues can learn, what can they learn, why is it that certain drugs stop working after you take them for a while, whether it's habituation, right, and so can we understand habituation, sensitization, associative learning, these kinds of things in chemical pathways. We're going to have a completely different way, I think, we're going have a completely different way of using drugs and of medicine in general when we start focusing on the goal states and on the intelligence of our subsystems as opposed to treating everything as if the only path was micromanagement from chemistry upwards.

说到这个躯体精神病学的概念——什么是体细胞？它们如何形成利用生物电来存储记忆的网络？体细胞的基本定义是什么？

Well, you speak to this idea of somatic psychiatry? What are somatic cells? How do they form networks that use bioelectricity to have memory and all those kinds of things. What are somatic cells, like basics here?

体细胞就是指你身体里的细胞，'体'就是身体的意思，对吧？体细胞其实就是——我甚至没有特别区分体细胞和干细胞之类的。基本上就是你体内所有的细胞，不仅仅是神经元，而是你身体里的所有细胞。它们在胚胎发生和再生过程中会形成电网络，这些网络部分功能是处理关于我们当前形态和目标形态的信息。我怎么知道这个？因为我可以举几个例子，一个例子是我们开始研究时用的涡虫。

Somatic cells just means the cells of your body, so much just means body, right? Somatic cells are just the I'm not even specifically making a distinction between somatic cells and stem cells or anything like that. Basically all the cells in your body, not just neurons, but all the cells in your body, They form electrical networks during embryogenesis, during regeneration, what those networks are doing in part is processing information about what our current shape is and what the goal shape is. Now how do I know this? Because I can give you a couple of examples, one example is when we started studying this we said okay, here's a planarian.

涡虫是一种扁形虫，通常有一个头和一条尾巴，关于涡虫有几个惊人的事实，但基本上它们——我认为涡虫几乎能解答所有生命深层次问题。首先，它们类似我们的祖先，具有真正的对称性，拥有真正的大脑，不像蚯蚓那样，它们是更高级的生命形式。它们有多种内脏器官，但体型很小，大约1到2厘米。它们有头有尾。第一点是涡虫是永生的，不会衰老，不存在'老涡虫'这个概念。

A planarian is a flatworm, it has one head and one tail normally, and the amazing, there's several amazing things about planaria, but basically they kind of, I think I think planaria hold the answer to pretty much every deep question of life. For one thing, they're similar to our ancestors, so they have true symmetry, they have a true brain, they're not like earthworms, they're you know, they're much more advanced lifeform. They have lots of different internal organs, but they're these little, they're about maybe two centimeters, a centimeter to two in size. They have a head and a tail. And the first thing is planaria are immortal, so they do not age, there's no such thing as an old planarian.

这就直接证明了那些关于寿命受热力学限制的理论是错误的，并不是说时间久了所有东西都会退化，不，涡虫可以持续生存——想想它们存在多久了，四亿年对吧。所以我们实验室里的涡虫实际上与四亿年前的涡虫在物质连续性上是相连的。

So that right there tells you that these theories of thermodynamic limitations on lifespan are wrong, it's not that well over time everything degrades, no, planaria can keep it going for probably, you know, how long have they been around, four hundred million years, right. So these are the actual so the planaria in our lab are actually in physical continuity with planaria that were here 400,000,000 ago.

所以本质上存在活了那么久的涡虫。'物质连续性'是什么意思？

So there's planaria that have lived that long, essentially. What does he mean physical continuity?

因为它们会分裂成两半。它们的繁殖方式就是分裂。涡虫的后半部分会抓住培养皿，前半部分脱离，然后把自己撕成两半。

Because because what they do is they split in half. The way they reproduce is they split in half. So so the planaria, the back the back end grabs the petri dish, the front end takes off, and they they rip themselves in half.

但从某种意义上说，这不也是一种物理延续吗？就像你也是某种物理延续？

But isn't isn't in some sense where, like, you are a physical continuation?

是的。区别在于我们会经历单细胞瓶颈期（即卵子阶段），而它们不会。我是说...有些涡虫确实会...

Yes. Except that except that we go through a bottleneck of one cell, which is the egg. They do not. I mean, can. There are certain planaria there.

明白了。所以我们经历了一个非常严苛的压缩过程，而它们不需要。

Got it. So we go through a very ruthless compression process and they don't.

是的。就像自动编码器，你知道的，那种

Yes. Like an autoencoder, you know, which

是

is

被压缩成一个细胞然后再恢复。而这些家伙直接把自己撕成两半，然后每一半都能...最神奇的是它们还能再生。你可以把它们切成碎片。记录是托马斯·亨特·摩根在1976年2月左右创造的。每一片都能重新长成一条完整的小虫子。

sort of squashed down to one cell and then back out. These these guys just tear themselves in half and then each and then and so the other amazing thing about them is they regenerate. So you can cut them into pieces. The record is I think 02/1976 or something like that by Thomas Hunt Morgan. And each piece regrows a perfect little worm.

它们完全清楚每一部分缺失什么、需要做什么。实际上，如果你把它切成两半，在它生长出另一半时，原始组织的体积会缩小，这样当新的小头长出来时，它们就是成比例的。完美的比例。如果饿着它们就会缩小，再喂食又会膨胀，它们对解剖结构的控制简直不可思议。

They know exactly every piece knows exactly what's missing, what needs to happen. In fact in fact, if you chop it in half, as it grows the other half, the original the original's tissue shrinks so that when the new tiny head shows up, they're proportional. So it perfect proportion. If you starve them they shrink, if you feed them again they expand, their control, their anatomical control is just insane.

有人把它们切成了200多块？

Somebody cut them into over 200 pieces?

对，是托马斯·亨特·摩根做的。

Yeah, Thomas Hunt Morgan did.

科学标签。

Hashtag science.

是的，太神奇了。没错。可能还不止这些。我是说，那时候还没有抗生素。我打赌他因为感染失去了一些。

Yep, Amazing. Yeah. And maybe more. I mean, didn't have antibiotics back then. I bet he lost some due to infection.

我打赌实际上比那还要多。我赌你能做到的远不止这些。

I bet I bet it's actually more than that. I bet you could do more than that.

人类做不到那样。

Humans can't do that.

嗯，是的。我是说，再次强调，确实如此，除了可能...

Well, yes. I mean, again, true except Maybe

在胚胎层面是可以的。

you can at the embryonic level.

嗯，关键就在这里，对吧？所以我每次谈到这个都会说，记住虽然从一小片组织再生出完整涡虫很神奇，但有一半人类能从单个细胞发育出完整身体，对吧？所以发育过程其实你可以把发育看作再生的一个例子。

Well, that's the thing, right? So I tell when I talk about this, I say just remember that as amazing as it is to grow a whole planarian from a tiny fragment, half of the human population can grow a full body from one cell, Right? So so development is really you can look at development as a as a just an example of regeneration.

是啊。想想我们要讨论再生医学，但某种程度上可以想象，五百年后会是怎样的温暖场景，那时我可能直接就能重新长出一只手。

Yeah. To think we'll talk about regenerative medicine, but there's some sense what would be like that warm in like five hundred years where I I can just go regrow a hand.

没错。考虑到时间因素，大型组织的生长确实需要时间，但目前来说...是的，我也这么认为。

Yep. I with with a given time, it takes time to grow large things, but For now. Yeah. I think so. I think

你可能会问为什么不加速这个过程？哦，因为生物学自有其节奏？

You can probably ex why not accelerate? Oh, biology takes its time?

我不会断言任何事不可能，但我确实不知道加速这些过程的方法。我认为这是可能的，我们终将实现再生，只是目前尚无良策。

I'm not going to say anything is impossible, but I don't know of a way to accelerate these processes. I think it's possible. I think we are going to be regenerative, but I don't know of a way

让再生变得迅速。我在想几个世纪后的人们可能会说：'他们当年居然要等一周才能再生出手，就像微波炉发明前——等等，那种在吐司上放芝士烤的东西叫什么来着？我只知道很好吃。'

to make it fast. I could just think people from a few centuries from now would be like, well, they have to they used to have to wait a week for the hand to regrow. It's like when the microwave was invented. You can you can toast your what's that called when you put a cheese on a toast? It's delicious is all I know.

我一时想不起来了。总之...好吧。那么涡虫，我们为什么要讨论这种神奇的生物？它们身上藏着生命之谜对吧？

I'm I'm blanking. Any anywho. Alright. So planaria, why were we talking about the magical planaria that they have the mystery of life?

没错。我们讨论涡虫不仅因为它们永生不死，不仅能再生身体每个部位，更关键的是它们几乎不会得癌症——这点的重要性我们稍后可以探讨。它们很聪明，能学习能被训练。最神奇的是，如果你训练一只涡虫后切掉它的头，尾部再生的全新大脑仍保留着原始记忆。

Yeah. So so the reason we're talking about planaria is not only are they immortal, okay, not only do they regenerate every part of their body, they generally don't get cancer, right, which we can talk about why that's important. They're smart, they can learn things, you can train them. And it turns out that if you train a planarian and then cut their heads off, the tail will regenerate a brand new brain that still remembers the original information.

它们体内是否存在生物电网络？是的。所以它们的体细胞形成了一个网络，这就是你所说的真正大脑吗？真正大脑需要满足什么条件？

Do they have a bioelectrical network going on or no? Yes. So their somatic cells are forming a network, and that's that's what you mean by a true brain? What what's the requirement for a true brain?

和其他事物一样，这是一个连续谱系，但真正的大脑具有某些特征，比如神经元局部密度能指导行为。

Like everything else, it's a continuum, but but but a true brain has certain characteristics as far as the density, like a localized density of neurons that guides behavior.

在头部。如果你

In the head. If you

砍掉它们的头，尾巴就什么都不会做，只会呆在那里直到新的大脑再生。它们拥有和你我相同的所有神经递质。但这就是我们在此讨论它们的原因。比如涡虫，你切掉头部，切掉尾部，剩下中间片段。这个中间片段必须生成一个头和一个尾。

cut their head off, tail doesn't do anything, it just sits there until the new brain regenerates. They have all the same neurotransmitters that you and I have. But here's why we're talking about them in this context. So here's your planaria, you cut off the head, you cut off the tail, you have a middle fragment. That middle fragment has to make one head and one tail.

它如何知道各自生成多少以及位置在哪？为什么不会颠倒？于是我们做了个简单实验，假设存在一个体细胞电网络能记住正确模式，它正在回忆那个记忆并按该模式重建。我们用了一种可视化这些细胞电活动的方法——这是脑电检测技术的变体。结果发现那个片段具有非常特殊的电模式。

How does it know how many of each to make and where do they go? How come it doesn't switch? So so we did a very simple thing, and we said, okay, let's let's make the hypothesis that there's a somatic electrical network that remembers the correct pattern, and that what it's doing is is recalling that memory and building to that pattern. So what we did was we used a way to visualize electrical activity in these cells, right, it's a variant of what people use to look for electricity in the brain. And we saw that that fragment has a very particular electrical pattern.

当我们开发出这项技术后，你能直接看到它。那种特殊电模式会显示头和尾的位置。明白吗？你能直接观察到。于是我们决定验证这个电模式是否真是控制头尾生长的记忆。

You can literally see it once once we developed the technique. It has a very particular electrical pattern that shows you where the head and the tail goes. Right? You can you can just see it. And then we said, well now let's test the idea that that's a memory that actually controls where the head and the tail goes.

让我们改变那个模式。本质上就是植入虚假记忆。可以通过多种方式实现，比如使用靶向离子通道的药物——选择这些药物后，就能将原本一头一尾的电模式改成双头模式。你只是在修改网络中的电信息。

Let's change that pattern. So basically, incept a false memory. And so what you can do is you can do that in many different ways. One way is with drugs that target ion channels to say and so you pick these drugs and you say, okay, I'm going to do it so that instead of this one head one tail electrical pattern, you have a two headed pattern. You're just editing the electrical information in network.

当你这样做时，猜猜细胞会构建什么？它们会构建出一条双头蠕虫，最酷的是这完全无需基因改变，我们没动过基因组，基因组完全是野生型。但更神奇的是，当你把这些双头动物再次切碎时，某些碎片仍会继续生成双头动物。这种信息、这种记忆、这种电路不仅存储着头部数量的信息，不仅会利用这些信息指导细胞如何再生，还会将其永久保存。一旦重置，它就会持续保持。

When you do that, guess what the cells build? They build a two headed worm, and the coolest thing about it, now no genetic changes, we haven't touched the genome, the genome is totally wild type. But the amazing thing about it is that when you take these two headed animals and you cut them into pieces again, some of those pieces will continue to make two headed animals. So that information, that memory, that electrical circuit, only does it hold the information for how many heads, not only does it use that information to tell the cells what to do to regenerate, but it stores it. Once you've reset it, it keeps.

我们还能逆转这个过程，把双头动物变回单头。这里有几个耐人寻味的发现对理解基因组等事物意义重大。想象我拿着这只双头动物——顺便说，当它们通过分裂繁殖时，后代仍是双头。假设我把它们扔进查尔斯河，一百年后科学家来取样时会惊呼：这里既有单头形态也有双头形态！

And we can go back, we can take a two headed animal and put it back to one headed. So now imagine so there's a couple of interesting things here that have implications for understanding what genomes and things like that. Imagine I take this two headed animal oh, and by the way, when they reproduce, when they tear themselves in half, you still get two headed animals. So imagine I take them and I throw them in the Charles River over here. So a hundred years later, some scientists come along and they scoop up some samples and they go, oh, here's a single headed form and a two headed form.

哇，物种形成事件！酷，我们测序基因组看看发生了什么。结果基因组完全一致，没有任何异常。这就回到你最初的问题：身体蓝图究竟从何而来？

Wow, a speciation event. Cool, let's sequence the genome and see what happened. The genomes are identical. There's nothing wrong with the genome. So if you ask the question, how does so this goes back to your very first question, is where do body plans come from?

涡虫如何知道自己该长几个头？你可能会说DNA，但事实证明：DNA产生的硬件默认设置是单头。就像计算器每次开机都显示零，但它其实是可编程的——一旦被改写，下次就会显示其他内容。同理，你可以制造单头、双头甚至无头涡虫。

How does the planarian know how many heads it's supposed to have? Now it's interesting because you could say DNA, but what what what as it turns out, the DNA produces a piece of hardware that by default says one head. The way that when you turn on a calculator, by default it's a zero every single time, right, when you turn it on it just says zero. But it's a programmable calculator as it turns out. So once you've changed that, next time it won't say zero, it'll say something else, and the same thing here, so you can make one headed, two headed, you can make no headed worms.

我们还做过其他类似实验，造出些非常古怪的结构。必须强调硬件与软件的区分至关重要：硬件是基础，没有正确硬件就无法形成记忆生理结构，但硬件并不完全决定信息内容。这些信息能被我们、细菌或寄生虫重新编程。更惊人的是：多数动物体细胞突变不会遗传给后代，但涡虫通过分裂繁殖——四亿年来它们保留了所有不致命的突变。

We've done some other things along these lines, some other really weird constructs. So question, so again it's really important, the hardware software distinction is really important because the hardware is essential, because without proper hardware you're never going to get to the right physiology of having that memory, but once you have it, it doesn't fully determine what the information is going to be. You can have other information in there and it's reprogrammable by us, by bacteria, by various parasites probably, things like that. The other amazing thing about these planarians, think about this, most animals when we get a mutation in our bodies, our children don't inherit it, right? So you could go on, you could run around for fifty, sixty years getting mutations, your children don't have those mutations because we go through the egg stage.

涡虫通过分裂繁殖，因此四亿年来它们保留了所有不致死细胞的突变。这些涡虫的身体被称为混倍体，每个细胞染色体数量都可能不同。它们的基因组看起来像肿瘤般混乱，却拥有地球最强的再生能力——尽管基因组一团糟，解剖结构却稳如磐石。

Planaria tear themselves in half and that's how they reproduce. So for four hundred million years they keep every mutation that they've had that doesn't kill the cell that it's in. So when you look at these planaria their bodies are what's called myxaploid, meaning that every cell might have a different number of chromosomes. They look like a tumor. If you look at the genome is an incredible mess because they accumulate all this stuff, and yet their body structure is they are the best regenerators on the planet, their anatomy is rock solid even though their genome is all kinds of crap.

这简直是个悖论：我们总说基因组决定身体，为什么基因组最差的动物反而拥有最精准的解剖控制、最强抗癌能力和再生能力？我们刚刚开始理解基因组决定的硬件与...（顺便说，几个月前我似乎有点明白原因了）但这仍是个重大谜题。

So this is kind of a scandal, right, that when we learn what are genomes, genomes determine your body, why does the animal with the worst genome have the best anatomical control, the most cancer resistant, the most regenerative. Really we're just beginning to start to understand this relationship between the genomically determined hardware and by the way just as a couple of months ago I think I now somewhat understand why this is, but it's really a major puzzle.

这确实给先天与后天之争带来了难题，因为通常你会把电力归为后天因素，硬件归为先天因素，而现在它们却奇怪地纠缠在一起，这种混杂状态还会代代相传。

I mean, that really throws a wrench into the whole nature versus nurture because you usually associate electricity with the with the nurture and the hardware within nature, and it's there's just this weird integrated mess Yeah. That propagates through generations.

是的，这要流动得多，也复杂得多。你可以想象这里的演化过程——就像这种具有多尺度能力的动物，它的组织也具备某种程度的多尺度适应性。

Yeah. It's much more fluid. It's much more complex. You can imagine what's happening here is just imagine the evolution of an animal like this, that multiscale, this goes back to this multiscale competency. Imagine that you have an animal where its tissues have some degree of multi scale competency.

比如我们在蝌蚪实验中看到的，把眼睛移植到尾巴上它仍能视物，这种可塑性令人惊叹。当这样的生物面临自然选择时，进化无法判断其适应性是源于优秀基因，还是虽然基因一般但靠能力弥补的。这意味着生物能力越强，选择机制就越难识别最优基因组——能力掩盖了基因信息。

So for example, like we saw in the tadpole, if you put an eye on its tail they can still see out of that eye, right? There's incredible plasticity. So if you have an animal and it comes up for selection and the fitness is quite good, evolution doesn't know whether the fitness is good because the genome was awesome or because the genome was kind of junky, but the competency made up for it, right, and things kind of ended up good. So what that means is that the more competency you have, the harder it is for selection to pick the best genomes. It hides information.

对吧？于是演化就会把主要精力放在提升能力上，因为基因组越来越难以观测。我认为涡虫身上就发生了这种失控现象：所有进化资源都投入到了算法上，导致基因组质量下降也无法修正，最终存活下来的都是那些无论基因多差都能造出完好蠕虫的算法。这自然降低了对保持优质基因的选择压力。

Right? And so that means that so what happens, you know, evolution basically starts all the hard work is being done to increase the competency because it's harder and harder to see the genomes. So I think in Planaria what happened is that there's this runaway phenomenon where all the effort went into the algorithm such that we know you've got a crappy genome, we can't clean up the genome, we can't keep track of it, so 's going to happen is what survives are the algorithms that can create a great worm no matter what the genome is. So everything went into the algorithm, which of course then reduces the pressure on keeping a clean genome. So this idea of right?

不同生物的这种特性程度各异。关键在于把能量投入算法而非过度依赖先验——生物系统本质如此，进化不会过分拘泥过去，它构建的是问题解决机器，而非对历史经验的精确复现。

And different animals have this in different levels. But this idea of putting energy into an algorithm that does not overtrain on priors, right? It can't assume, I mean I think biology is this way in general, evolution doesn't take the past too seriously because it makes these basically problem solving machines as opposed to like exactly what, you know, to deal with exactly what happened last time.

对，是问题解决而非记忆回溯。需要一点记忆，但更侧重解决问题的能力。

Yeah, problem solving versus memory recall. Mhmm. So a little memory, but a lot of problem solving.

我想是的。多数情况下确实如此。解决问题的能力。

I think so. Yeah. In many cases. Yeah. Problem solving.

是的。

Yep.

我是说，这类系统能够被构建出来真是令人难以置信，尤其是它们与我们在人工智能领域构建问题解决系统的方式形成鲜明对比。回到异种机器人这个话题。我不确定我们是否描述过异种机器人是如何构建的，抱歉。你们有一篇题为《生物机器人：一个新兴跨学科领域的视角》的论文。开头你提到'异种机器人'这个词有些争议。

I mean, it's incredible that those kinds of systems are able to be constructed, especially how much they contrast with the way we build problem solving systems in the AI world. Back to Xenobots. I'm not sure if we ever described how Xenobots Sorry. Are built, but, mean, you have a paper titled biological robots, perspectives on an emerging interdisciplinary field. In the beginning, you you mentioned that the word xenobots is, like, controversial.

你们用'异种机器人'这个词会惹上麻烦吗？还是说人们不喜欢这个词？你们用'异种机器人'而不是'生物机器人'是想故意制造话题吗？我不太清楚。

Do you guys get in trouble for using xenobots? Or what? Do people not like the word xenobots? Are you trying to be provocative with the word xenobots versus biological robots? I don't know.

这个...是的。这里面有什么我们应该知道的戏剧性情节吗？

This Yeah. Is there some drama that we should be aware of?

确实有些小风波。我认为争议主要源于人们对术语含义的固有观念，这些观念在很多情况下已经完全落后于科学现状，更不用说未来发展了——这些概念在未来几十年内必将被淘汰。比如你问任何人，包括很多生物学界人士（他们坚持要严格区分生物体和机器人），'什么是机器人？'他们会说机器人是从工厂生产出来的。

There's a little bit of drama. I think I think the drama is basically related to people having very fixed ideas about what terms mean, and I think in many cases these ideas are completely out of date with with where science is now, and for sure, they're they're out of date with what's going to be I mean, these these these concepts are not going to survive the next couple of decades. So if you ask a person, and including, you know, a lot of people in biology who kind of want to keep a sharp distinction between biologicals and robots. Say, what's a robot? Well, a robot, it comes out of a factory.

它是人造的，很无趣，你能预测它的所有行为，由金属和其他无机材料构成。而生物体是神奇的，会自主生长等等。

It's made by humans. It is boring. It is a meaning that you can predict everything it's going to do. It's made of metal and certain other inorganic materials. Living organisms are magical, they arise and so on.

这些区分标准...我认为这些区分从来就不合理，未来将彻底失去意义。我们写过几篇论文，其中一篇是我和Josh Baumgard合著的，我们直接挑战了这些术语，指出这些二元分类是基于技术发展和想象力局限形成的非本质性标准，过去或许成立，但现在必须摒弃。我们称之为'异种机器人'——'xeno'源自非洲爪蟾（xenopus laevis），即构成这些机器人的蛙类细胞。但我们认为这是生物机器人技术的范例，因为一旦我们掌握与这些细胞沟通并操控其输入的方法，就能让它们构建任何我们想要的形态，这才是真正的机器人技术。

There's these distinctions. I think these distinctions I think were never good, but they're going to be completely useless going forward. So part of, there's a couple of papers, that's one paper and there's another one that Josh Baumgard and I wrote where we really attack the terminology, and we say these binary categories are based on very non essential kind of surface limitations of technology and imagination that were true before, but they've got to go. And we call them xenobots. So xeno for xenopus laevis, it's the frog that these guys are made of, but we think it's an example of a bio bot technology because ultimately once we understand how to communicate and manipulate the inputs to these cells, we will be able to get them to build whatever we want them to build, and that's robotics.

对吧？这是对具有实用目的的机器进行理性构建。我绝对认为这是一个机器人平台，而一些生物学家并不这么看。

Right? It's it's the rational construction of machines that have useful purposes. I I I absolutely think that this is a robotics platform, whereas some biologists don't.

但它的构建方式让所有不同组件都在进行各自的计算。就像我们一直在讨论的那样。所以你试图对其实施自上而下的控制

But it's built in a way that all the different components are doing their own computation. So in a way that we've been talking about. So you're trying to do top down control on that

正是这个系统。没错。而且未来所有这些都将融合，因为在某个阶段我们会加入合成生物电路，对吧，新的转录电路让它们执行新功能。当然我们会加入一些这类设计，但在最初几篇论文中我们刻意避开了这些——后续还有几篇我认为会非常重磅的论文即将发表——因为我们想展示原始细胞的构成。因为如果你对它们进行过度改造，比如加入新的转录因子和代谢机制等，人们就会说'好吧，这是你们改造的结果'。

That's exactly system. Correct. And and in the future, all of this will will will merge together because, of at some point we're going to throw in synthetic biology circuits, right, new transcriptional circuits to get them to do new things. Of course we'll throw some of that in, but we specifically stayed away from all of that because in the first few papers, and there's some more coming down the pike that are I think going be pretty dynamite, that we want to show what the native cells are made of. Because what happens is if you engineer the heck out of them, right, if we were to put in new transcription factors and some new metabolic machinery and whatever, people will say, well okay, you engineered this and you made it do whatever and fine.

我和整个团队想展示的是生物的可塑性、智慧与本质。在你开始以那种方式操控硬件之前，它能展现出哪些令人惊讶的特性？

I wanted to show, and and and the whole team wanted to show the plasticity and the intelligence and the biology. What does it do that's surprising before you even start manipulating the hardware in that way?

没错。不要过度控制它。让生物系统的全部美感自然绽放。为什么选择非洲爪蟾（Xenopus levus levus）？

Yeah. Don't try to over control the thing. Let it flourish. The the full beauty of the biological system. Why is Xenopus levus levus?

这个词怎么发音？就是那种青蛙。

How do you pronounce it? The frog.

非洲爪蟾（Xenopus levus）。对。这是一种非常

Xenopus levus. Yeah. Yeah. It's a very

这只青蛙很受欢迎吗？

popular this frog?

我想是从五十年代就开始使用了。它非常方便，因为你可以...你知道，我们把成蛙养在这个精致的青蛙栖息地里。它们产卵，一次能产下成千上万的卵。卵就在你眼前发育。

It's been used since, I think, the fifties. It's just very convenient because you can you you know, we we keep the adults in this in this very fine frog habitat. They lay eggs. They lay tens of thousands of eggs at a time. The eggs develop right in front of your eyes.

这是最神奇的现象——通常如果用老鼠或兔子做实验，你看不到早期发育阶段，因为胚胎都在母体内。而这个实验里，所有过程都在室温培养皿中进行。你能看到受精卵不断分裂再分裂，所有器官逐渐形成。科研界为此开发了大量工具来观察和干预这个过程，比如研究先天缺陷、神经生物学和癌症免疫学。

It's the most magical thing you see because normally, you know, if you were to deal with mice or rabbits or whatever, you don't see the early stages, right, because everything's inside the mother. Here, everything's in a petri dish at room temperature. So you have an egg, it's fertilized, and you can just watch it divide and divide and divide and all the organs form, you can just see it. And at that point the community has developed lots of different tools for understanding what's going on and also for manipulating it, right? So people use it for understanding birth defects and neurobiology and cancer immunology also.

所以整个胚胎发育过程都在培养皿里完成？没错。看起来太酷了，有相关视频吗？

So you get the whole embryogenesis in the petri dish? Yep. That's so cool to watch. Is there videos of this?

当然有，网上有很多精彩视频。哺乳动物胚胎也很有意思——比如单卵双胞胎就是哺乳动物胚胎被切成两半后形成的。不会得到两个半身，而是两个完整个体，因为这属于再生现象。

Oh, Yeah. There's amazing videos online. I mean mammalian embryos are super cool too. For example, monozygotic twins are what happens when you cut a mammalian embryo in half. You don't get two half bodies, you get two perfectly normal bodies because it's a regeneration event.

对吧？发育本质上就是一种再生过程。

Right? Development is just the kind of regeneration really.

为什么选这种青蛙？只是因为五十年代就开始用它做实验...

And why this particular frog? It's just because they were doing it in the fifties and

它在实验室里繁殖得很好，你知道，很容易饲养且繁殖力强，几十年来人们基本上一直在开发相关工具。也有人使用其他蛙类，但必须说明的是，异种机器人本质上与青蛙无关——我不能透露太多，因为这些发现尚未发表和经过同行评审——但我们已用与青蛙完全无关的其他材料制造出了异种机器人。这不是青蛙特有的现象。我们最初选择青蛙只是因其便利性，但这种可塑性并非源于它们的蛙类特性。

It breeds well in, you know, in it's easy to raise in in the laboratory and it's very prolific, and all the tools basically for decades people have been developing tools. There's other pieces, some people use other frogs, but I have to say, this is important, xenobots are fundamentally not anything about frogs, so I I can't say too much about this because it's not published and peer reviewed yet, but we've made Xenobots out of other things that have nothing to do with frogs. It's this is not a frog phenomenon. This is we we started with frog because it's so convenient, but this this this plasticity is not a frog, you know, it's not related to the fact that they're frogs.

如果你亲吻它会发生什么？会变成王子吗？不会。或者公主？到底是哪种？

What happens when you kiss it? Does it turn to a prince? No. Or a princess? Which way?

王子。对。王子。

Prince. Yeah. Prince.

应该是王子。没错。这个实验我们好像还没做过。就算做过，我也...

It should be a prince. Yeah. That's an experiment that I don't believe we've done. And if we have, I don't

我就不参与了。这个项目我能牵头负责吗？好吧。酷。这些细胞是如何协调的？

I won't collaborate. I can I can take on the lead on that effort? Okay. Cool. How does the cells coordinate?

我们聚焦胚胎发生过程来看：最初是一个细胞，它分裂时并不需要严格控制每个子细胞的分化方向。

Let's focus in on just the embryogenesis. So there's one cell. So it divides. Doesn't have to be very careful about what each cell starts doing once they divide.

是的。

Yes.

就像，是的。当有三个人时，就像是联合创始人什么的。比如，好吧，慢点说。你负责这个。他们什么时候开始专业化，又是如何协调这种专业化的？

And like Yeah. When there's three of them, it's like the cofounders or whatever. Like, well, like, slow down. You're responsible for this. When do they become specialized, and how do they coordinate that specialization?

所以这是发育生物学的基础科学。这方面有很多已知知识，但我要告诉你的是，我认为最关键的部分在于：确实，谁负责什么很重要。然而——回到我之前说的生物学不太拘泥于过去的观点——我的意思是它不会假定所有事物都按预期发展，明白吗？这里有个例子：这是上世纪40年代的老实验，想象一只蝾螈，它有些通向肾脏的小管，就是这种微型管道。

So so this is the the basic science of developmental biology. There's a lot known about all of that, but but what what I'll I'll tell you, I mean, what I think is the kind of the most important part, which is, yes, it's very important who does what. However, because going back to this issue of why I made this claim that biology doesn't take the past too seriously, and what I mean by that is it doesn't assume that everything is the way it's expected to be, right? And here's an example of that. This was done, this was an old experiment, going back to the 40s, but basically imagine, it's a newt, a salamander, it's got these little tubules that go to the kidneys, this little tube.

截取那段管道的横截面，你会看到8到10个细胞协同形成了这个微型管道的截面。神奇的是，你可以干扰早期细胞分裂让细胞变得巨大——可以改变它们的大小，强制调整尺寸。当细胞大小不同时，整体管径仍保持不变。所以截面可能只有4-5个细胞，甚至3个，直到细胞大到单个细胞自我卷曲就能形成相同宏观结构，但分子机制已完全不同。

Take a cross section of that tube you see eight to 10 cells that have cooperated to make this little tube in cross section. So one amazing thing you can do is you can mess with very early cell division to make the cells gigantic, bigger. You can make them different sizes, can force them to be different sizes. So if you make the cells different sizes the whole nude is still the same size. So if you take a cross section through that tubule, instead of eight to 10 cells you might have four or five, or you might have three, until you make the cells so enormous that one single cell wraps around itself and and gives you that same large scale structure, but a completely different molecular mechanism.

现在不再是细胞间通讯形成管道，而是一个细胞利用细胞骨架自我弯曲。想想这意味着什么：为了维持宏观解剖特征，会调用不同的分子机制——这就是自上而下控制。设想你是个裸细胞正在构建胚胎，如果固执认定分工必须固定，那当细胞变大时就会完蛋。生命对部件尺寸、DNA含量等变化有着惊人容忍度——它是高度互操作的，插入电极或纳米材料仍能运作。

So now instead of cell to cell communication to make a tubule, instead of that it's one cell using the cytoskeleton to bend itself around. So think about what that means, in the service of a large scale, talk about top down control, in the service of a large scale anatomical feature different molecular mechanisms get called up. So now think about this, you're a nude cell and trying to make an embryo, if you had a fixed idea of who was supposed to do what, you'd be screwed because now your cells are gigantic, nothing would work. There's an incredible tolerance for changes in the size of the parts, in the amount of DNA in those parts, all sorts of stuff. Can the life is highly interoperable, you can put electrodes in there, you can put weird nanomaterials, it still works.

这就是问题解决能力的体现，对吧？即使在环境变化时也能达成目标。这正是智能的标志——威廉·詹姆斯将智能定义为通过不同手段达成相同目标的能力。眼前就是例证：手段完全不同，目标始终如一。

It's it's this is that problem solving action, right? It's able to do what it needs to do even when circumstances change. That is, you know, the hallmark of intelligence, right? William James defined intelligence as the ability to get to the same goal by different means. That's this, you get to the same goal by completely different means.

我之所以讲这些，是想说明：细胞确实需要各司其职，但它们对意外状况有惊人容忍度，仍能完成任务。这些都不是硬连接的——有些生物可能是，比如秀丽隐杆线虫，每个细胞都有编号，所有个体细胞数量位置完全相同，存在精确的发育图谱。

And so why am I bringing this up, is just to say that yeah, it's important for the cells to do the right stuff, but they have incredible tolerances for things not being what you expect and to still get their job done. So all of these things are not hardwired, there are organisms that might be hardwired, for example the nematode C. Elegans, in that organism every cell is numbered, meaning that every C. Elegans has exactly the same number of cells as every other C. Elegans, they're all in the same place, all divide, there's literally a map of how it works.

这类系统更像是模具压制的。但绝大多数生物都具有惊人的可塑性。

In that sort of system it's much more cookie cutter, But but most most organisms are incredibly plastic in that way.

对你来说，整个发育生物学过程有什么特别神奇之处吗？你能描述一下吗？因为你刚才提到了。它们非常擅长完成它们需要做的工作目标，那种能力层面的东西，但你能从一个细胞得到一整个活生生的生物体。没错。

Is there something particularly magical to you about the whole developmental biology process? Is there something you could say? Because you just said it. They're very good at accomplishing the goal of the job they need to do, the competency thing, but you get freaking organism from one cell. Yep.

我的意思是，要凭直觉理解整个过程非常非常困难，甚至思考如何逆向工程这个过程都很难。

It's like I mean, it's very hard hard to intuit that whole process, to even think about reverse engineering that process.

没错。难到我经常让学生做这个思想实验：想象你缩小到一个细胞的尺度，身处胚胎中央环顾四周。细胞四处活动，有些在死亡，每次观察时大多数生物体的细胞数量似乎都不同。我想如果你不知道胚胎发育是什么，你根本不会意识到眼前的一切最终总会形成相同的生物。

Right. Very hard to the point where I often just imagine, I I I sometimes ask my students to do this thought experiment. Imagine you were you were shrunk down to the scale of a single cell, and you were in the middle of an embryo and you were looking around at what's going on. And the cells running around, some cells are dying, know, every time you look it's kind of a different number of cells for most organisms and so on. I think that if you didn't know what embryonic development was, you would have no clue that what you're seeing is always going to make the same thing.

更不用说知道那是什么了。即便拥有完整的基因组信息，我们也无法判断它们究竟在构建什么——但就连猜测'哇，所有这些活动的结局总会形成相同结果'都令人难以置信。

Never mind knowing what that is. Never mind being able to say, even with full genomic information being able to say what the hell are they building, we have no way to do that. But just even to guess that, wow, the the the outcome of all this activity is it's always gonna be it's always gonna build the same thing.

创造出现在这个'你'的必然性已经存在其中。所以如果从同一个胚胎开始，就会创造出非常相似的生物体。

The imperative to create the final you as you are now is there already. So you can you would so if you start from the same embryo, you would create a very similar organism.

是的。除了像异种机器人这样的例外——给它们不同环境就会发展出不同的适应方式。但总体而言，我认为进化最擅长的是创造具有稳定基础模式的硬件，即在自然状态下能稳定运作，同时具备解决问题的多种能力。当某些前提不成立时——比如细胞尺寸异常、数量错误，甚至体内被插入电极等——它仍能完成大部分必需的工作。

Yeah. Except for cases like the xenobots, you give them a different environment, they come up with a different way to be adaptive in that environment. But overall, I mean, so I think to kind of summarize it, I think what Evolution is really good at is creating hardware that has a very stable baseline mode, meaning that left to its own devices it's very good at doing the same thing, but it has a bunch of problem solving capacities such that if any assumptions don't hold, if your cells are a weird size or you get the wrong number of cells or there's a you know, somebody stuck an electrode halfway through the body, whatever, it will still get most of what it needs to do done.

你谈到了生物学的魔力与力量。如果把人类大脑放在这个语境中看，它有多特殊？你似乎在淡化大脑的重要性——我们通常认为所有特殊计算都发生在大脑，其他部分只是辅助。但你说整个系统都在进行计算。那么在这个完整的生物学背景下，人类大脑究竟有多特殊？

You've talked about the magic and the power of biology here. If we look at the human brain, how special is the brain in this context? You're kind of minimizing the importance of the brain or lessening its we think of all the special computation happens in the brain, everything else is like the help. You're kind of saying that the whole thing is the whole thing is doing computation. But nevertheless, how special is the human brain in this full context of biology?

是的。我是说，听着，我们无法否认人类大脑让我们做到了没有它就做不到的事情。

Yeah. I mean, look, there's no getting away from the fact that the human brain allows us to do things that we could not do without it.

你也可以对肝脏说同样的话。

You can say the same thing about the liver.

是的。不。这这是真的。所以，你知道，我我的目标不是...不，你说得对。

Yeah. No. The this is this is true. And so and so, you know, I I my goal is not no. You're right.

我的目标是

My goal is

不是你现在只是在给大脑说客套话。你就像个政客。听着，每个人都有自己的角色。是的。

not You're just being polite to the brain right now. Well You're being a politician. Like, listen. Everybody has Everybody has a role. Yeah.