#208 – 杰夫·霍金斯：智能的千脑理论

他们为什么要这么做？

And why do they do that?

我们这么做是因为你可以想象从不同角度看到的样子。

Well, we do that because you can imagine what it would look like from different angles.

你可以看看从这里看是什么样，也可以问，从车库到游泳池有多远？

You can say, Okay, look at from here, look at And you can also say, Well, how far would it get from the garage to the swimming pool?

或者类似这样的问题，对吧？

Or something like that, right?

你可以想象一下从这个角度去看。

You can imagine looking at this.

你可以说，从这个位置看会是什么样子？

And you can say, What would it be to view from this location?

所以我们建造这些实体模型，让你能够想象未来和各种行为。

So, we build these physical models to let you imagine the future and imagine behaviors.

现在，我们可以把这个相同的模型放到电脑里。

Now, we can take that same model and put it in a computer.

所以，我们现在，今天，人们都会在电脑中建立房屋模型，他们使用一组——我们稍后会回到这个术语——参考框架。

So, we now, today, they all build models of houses and a computer, they do that using a set of, we'll come back to this term in a moment, reference frames.

但最终，你会为房屋分配一个参考框架，并在不同位置为房屋赋予不同的属性。

But eventually, you assign a reference frame for the house and you assign different things for the house in different locations.

然后，计算机可以生成一幅图像，说：好的，这是从这个方向看的样子。

And then the computer can generate an image and say, Okay, this is what it looks like in this direction.

大脑正在做着与这非常相似的事情。

The brain is doing something remarkably similar to this.

令人惊讶。

Surprising.

它正在使用参考框架。

It's using reference frames.

它在构建类似于计算机模型的东西，这与建立物理模型具有同样的优势。

It's building these it's similar to a model in a computer, which has the same benefits of building a physical model.

它让我能够说：如果这个物体处于这个方向，它会是什么样子？

It allows me to say, What would this thing look like if it was in this orientation?

如果我按下这个按钮，可能会发生什么？

What would likely happen if I pushed this button?

我以前从未按过这个按钮。

I've never pushed this button before.

或者我该如何实现某件事？

Or how would I accomplish something?

我想传达我学到的一个新想法。

I want to convey a new idea I've learned.

我该怎么做呢？

How would I do that?

我可以在脑海中想象，嗯，我可以谈论它。

I can imagine in my head, well, I could talk about it.

我可以写一本书。

I could write a book.

我可以做一些播客。

I could do some podcasts.

我或许可以告诉我的邻居。

I could maybe tell my neighbor.

在我做任何这些事情之前，我就能想象出它们的所有可能结果。

And I can imagine the outcomes of all these things before I do any of them.

这就是模型能让你做到的事情。

That's what the model lets you do.

让我们规划未来，想象我们行为的后果。

Let's just plan the future and imagine the consequences of our actions.

预测，你刚才提到了预测。

Prediction, you asked about prediction.

预测并不是模型的目标。

Prediction is not the goal of the model.

预测是它的固有属性，也是模型自我修正的方式。

Prediction is an inherent property of it and it's how the model corrects itself.

所以预测是智能的基础。

So prediction is fundamental to intelligence.

这是构建模型的基础，而模型是智能的。

It's fundamental to building a model and the model's intelligent.

让我退回去，把这一点说清楚。

Let me go back and be very precise about this.

预测，你可以从两个角度来理解预测。

Prediction, you can think of prediction two ways.

一种是：如果我这么做，会发生什么？

One is like, Hey, what would happen if I did this?

这是一种预测。

That's a type of prediction.

这是智能的关键部分。

That's a key part of intelligence.

但用预测来想，比如：我拿起这个水瓶时，它会是什么感觉？

But using predictions like, Oh, what's this water bottle gonna feel like when I pick it up?

这似乎并不太智能。

And that doesn't seem very intelligent.

但有一种思考预测的方式是，它能帮助我们发现模型哪里出错了。

But one way to think about prediction is it's a way for us to learn where our model is wrong.

所以如果我拿起这个水瓶，发现它很烫，我会非常惊讶。

So if I picked up this water bottle and it felt hot, I'd be very surprised.

或者如果我拿起它，发现它很轻，我也会感到惊讶。

Or if I picked it up and it was very light, I'd be surprised.

或者如果我拧开瓶盖，却发现必须往相反方向转，我也会感到惊讶。

Or if I turned this top and I had to turn it the other way, I'd be surprised.

因此，所有这些情况都可能伴随着一种预测：好吧，我要这么做。

And so, all those might have a prediction like, Okay, I'm going do it.

我要喝点水。

I'll drink some water.

我要这么做，就是这样，我感觉到打开了。

I'm going do this, there it is, I feel opening.

如果我必须往相反方向转呢？

What if I had to turn it the other way?

或者如果它被分成了两半呢？

Or what if it's split in two?

然后我会说，天哪，我理解错了。

Then I say, Oh my gosh, misunderstood this.

我没有使用正确的模型。

I didn't have the right model.

这个东西会立刻吸引我的注意力。

This thing, my attention would be drawn to it.

我会盯着它想，这到底是怎么发生的？

I'd be looking at it going, Well, how the hell did that happen?

为什么它是这样打开的？

Why did it open up that way?

我会通过操作它、只是看着它并摆弄它来更新我的模型，然后说，这是一种新型的水瓶。

And I would update my model by doing it, just by looking at it and playing around with it, I'd update it and say, This is a new type of water bottle.

所以，你提到的是一些像水瓶这样复杂的东西，但这也适用于最基本的视觉，比如看到事物。

So, you're talking about sort of complicated things like a water bottle, but this also applies for just basic vision, just like seeing things.

这几乎就像是感知世界的一种先决条件——即预测它。

It's almost like a precondition of just perceiving the world as predicting it.

所以

So

你所看到的一切，首先都会经过

it's just everything that you see is first passed through

你的预测。

your prediction.

事实上，你所看到和感受到的一切，这是我上世纪80年代末的领悟。

Everything you see and feel in fact, this was the insight I had back in the late '80s.

不，抱歉，是80年代初。

No, excuse me, early '80s.

其他人也得出了同样的观点：你接收到的每一种感官输入，不仅仅是视觉，还包括触觉和听觉，你都会对它形成一种预期和预测。

And other people have reached the same idea, is that every sensory input you get, not just vision, but touch and hearing, you have an expectation about it and a prediction.

有时你能非常准确地预测，有时则不能。

Sometimes you can predict very accurately, sometimes you can't.

我无法预测你接下来会说出什么词，但当你开始说话时，我的预测会越来越准确。

I can't predict what next word's gonna come out of your mouth, but as you start talking, I'll get better and better predictions.

如果你谈论某些话题，我会非常惊讶。

And if you talk about some topics, I'd be very surprised.

因此，我始终在对所有感官输入进行一种背景预测。

So, I have this sort of background prediction that's going on all the time for all of my senses.

再说一遍，我认为这就是我们学习的方式。

Again, the way I think about that is this is how we learn.

这更多是关于我们如何学习。

It's more about how we learn.

这是我们理解能力的一种检验。

It's a test of our understanding.

我们的预测是在检验：这真的是一瓶水吗？

Our predictions are a test of, is this really a water bottle?

如果是，我不应该看到一只小手指从侧面伸出来。

If it is, I shouldn't see a little finger sticking out the side.

如果我看到一根小手指伸出来，我会想：天啊，这到底怎么回事？

And if I saw a little finger sticking out, I like, what the hell is going on?

这不正常。

That's not normal.

这太令人着迷了，让我再仔细想想这一点。

That's fascinating that Let me linger on this for a second.

我真的觉得，预测是根本性的，它关乎我们思维运作的方式，关乎智能。

It really honestly feels that prediction is fundamental to everything, to the way our mind operates, to intelligence.

所以，这其实是一种看待智能的不同方式，即一切皆始于预测。

So like, it's just a different way to see intelligence, which is like everything starts at prediction.

而预测需要一个模型。

And prediction requires a model.

除非你拥有对它的模型，否则你无法进行预测。

You can't predict something unless you have a model of it.

没错，但行为就是预测。

Right, but the action is prediction.

所以，模型所做的就是预测。

So like the thing the model does is prediction.

但它也可以延伸到像这样的情况：哦，如果我今天这么做会怎样？

But it also, yeah, but you can then extend it to things like, Oh, what would happen if I took this today?

我去做了这件事。

I went and did this.

那会是什么样子？

What would be like that?

你可以把预测扩展到比如：我想在工作中得到晋升。

How, you can extend prediction to like, Oh, I want to get a promotion at work.

我应该采取什么行动？

What action should I take?

你可以说，如果我这么做，我会预测可能发生什么。

And you can say, If I did this, I predict what might happen.

如果我去和某人交谈，我会预测可能发生什么。

If I spoke to someone, I predict what might happen.

所以这不仅仅是低层次的预测。

So it's not just low level predictions.

是的，这都是预测。

Yeah, it's all predictions.

这都是预测。

It's all predictions.

这就像是一个黑箱，你可以向它提出任何问题，无论是低层次还是高层次的。

It's like this black box that you can ask basically any question, low level or high level.

所以我们从这个观察开始。

So, we started off with that observation.

这是一种不间断的预测，我在书中也写到了这一点。

It's this nonstop prediction, and I write about this in the book.

然后我们提出问题：神经元究竟是如何物理上做出预测的？

And then we asked, how do neurons actually make predictions physically?

比如，神经元在做出预测时究竟做了什么？

Like, what does the neuron do when it makes prediction?

或者神经组织在做出预测时是如何运作的。

Or the neural tissue does when it makes predictions.

然后我们提出问题：我们通过什么机制构建一个能够进行预测的模型？

And then we ask what are the mechanisms by how we build a model that allows you to make predictions?

因此，我们从预测开始，将其作为某种意义上的根本研究议程，并认为：我们理解大脑是如何做出预测的。

So we started with prediction as sort of the fundamental research agenda in some sense, and say, Well, we understand how the brain makes predictions.

我们会理解它是如何构建这些模型以及如何学习的，而这正是智能的核心。

We'll understand how it builds these models and how it learns, and that's the core of intelligence.

因此，这是让我们得以切入的关键点：我们的研究议程就是理解预测。

So, it was the key that got us in the door to say, That is our research agenda, understand predictions.

那么，在整个过程中，你认为智能源自何处？

So, in this whole process, where does intelligence originate, would you say?

如果我们观察那些远不如人类聪明的生物，并从进化过程逐步构建出人类，那么这种能够进行预测的‘神奇’能力——即开始看起来更像智能的预测模型——是在哪里出现的？

So, if we look at things that are much less intelligent than humans, and you start to build up a human through the process of evolution, where's this magic thing that has a prediction model or a model that's able to predict that starts to look a lot more like intelligence?

理查德·道金斯有没有为你的书写过一篇序言？一篇极佳的序言。

Is there a place where Richard Dawkins wrote an introduction to your book, an excellent introduction.

我的意思是，这把很多事情都放在了更清晰的背景下，而且有趣的是，你这本书和达尔文的《物种起源》之间存在着诸多相似之处。

I mean, it puts a lot of things into context, and it's funny just looking at parallels for your book and Darwin's Origin of Species.

达尔文探讨了物种的起源。

So, Darwin wrote about the origin of species.

那么，智能的起源是什么？

So, what is the origin of intelligence?

是的。

Yeah.

我们对此有一个理论，但仅此而已，它只是一个理论。

Well, we have a theory about it, and it's just that, it's a theory.

这个理论如下。

The theory goes as follows.

一旦生物开始移动，它们就不再是仅仅漂浮在海洋中，也不再是像植物那样固定在某处；只要它们开始移动，以某种智能的方式移动就会带来优势。

As soon as living things started to move, they're not just floating in sea, they're not just a plant, you know, grounded someplace, as soon as they started to move, there was an advantage to moving intelligently, to moving in certain ways.

你可以做一些非常简单的事情。

And there's some very simple things you can do.

细菌或单细胞生物可以向食物浓度梯度的方向移动。

Bacteria or single cell organisms can move towards a source of gradient of food or something like that.

但一种动物可能知道自己在哪里、去过哪里，以及如何回到那个地方，或者一种动物可能会想，哦，那里曾经有食物来源。

But an animal that might know where it is and know where it's been and how to get back to that place, or an animal that might say, Oh, there was a source of food someplace.

我该怎么到达那里？

How do I get to it?

或者那里曾经有危险。

Or there was a danger.

我该怎么避开它？

How do I get to it?

或者那里曾经有配偶。

Or there was a mate.

我该怎么找到它们？

How do I get to them?

这带来了巨大的进化优势。

There was a big evolution advantage to that.

所以早期，生物面临着一种压力，需要开始理解自己的环境，比如我在哪里？我曾经去过哪些地方？那些地方发生过什么？

So early on, there was a pressure to start understanding your environment, like where am I and where have I been and what happened in those different places?

因此，我们大脑中仍然保留着这种神经机制。

So, we still have this neural mechanism in our brains.

在哺乳动物中，它位于海马体和内嗅皮层。

In the mammals, it's in the hippocampus and entorhinal cortex.

这些是大脑中较古老的区域。

These are older parts of the brain.

而且这些区域已经被研究得非常透彻。

And these are very well studied.

我们会构建出对环境的地图。

We build a map of our environment.

因此，这些大脑区域中的神经元知道我在这个房间的哪个位置，门在哪里，以及其他类似的信息。

So these neurons in these parts of the brain know where I am in this room and where the door was and things like that.

所以，许多其他哺乳动物也有这种能力

So, a lot of other mammals have this kind of

所有哺乳动物都有这种能力，对吧？

All mammals have this, right?

几乎所有能够知道自己位置并活动的动物，都必须具备某种地图系统，必须有一种方式来表示：我已掌握了我环境的地图。

And almost any animal that knows where it is and get around must have some mapping system, must have some way of saying, I've learned a map of my environment.

我后院里有蜂鸟。

I have hummingbirds in my backyard.

它们总是飞往同样的地方。

And they go to the same places all the time.

它们一定知道自己在哪里。

They must know where they are.

当它们完全清醒时，就知道自己身在何处。

They just know where they are when they're fully.

它们并不是在随机地四处飞行。

They're not just randomly flying around.

它们知道特定的花朵，并会返回那些地方。

They know particular flowers they come back to.

所以我们都有这种能力。

So, we all have this.

事实证明，让神经元做到这一点——构建环境的地图——是非常困难的。

And it turns out it's very tricky to get neurons to do this, to build a map of an environment.

因此，我们现在知道，关于位置细胞、网格细胞以及大脑较古老区域中其他类型细胞如何构建世界地图的研究非常著名，且仍在积极进行中。

And so we now know there's these famous studies that's still very active about place cells and grid cells and these other types of cells in the older parts of the brain and how they build these maps of the world.

这真的很巧妙。

And it's really clever.

显然，经过长时间的进化压力，这种能力已经变得非常出色。

It's obviously been under a lot of evolutionary pressure over a long period of time to get good at this.

所以，动物现在知道它们在哪里。

So, animals now know where they are.

我们认为发生的情况是——有许多证据支持这一点——那就是我们用来映射空间的机制被重新包装了，同样的神经元被压缩成更紧凑的形式，形成了皮质柱。

What we think has happened, and there's a lot of evidence to suggest this, is that that mechanism we learned to map like a space was repackaged, the same type of neurons was repackaged into a more compact form, and that became the cortical column.

某种程度上，它被泛化了，如果这个词可以用的话。

And it was in some sense genericized, if that's a word.

它被转化为一种非常具体的能力，从学习环境的地图扩展到学习任何事物的地图，构建任何事物的模型，而不仅仅是你的空间，还包括咖啡杯等等。

It was turned into a very specific thing about learning maps of environments to learning maps of anything, learning a model of anything, not just your space, but coffee cups and so on.

它被重新打包成更紧凑、更通用的版本，然后被复制。

And it got sort of repackaged into a more compact version, a more universal version, and then replicated.

因此，我们如此灵活的原因是，我们拥有这种映射算法的非常通用的版本，并且有十五万个副本。

So, the reason we're so flexible is we have a very generic version of this mapping algorithm and we have 150,000 copies of it.

这听起来很像深度学习的发展历程。

Sounds a lot like the progress of deep learning.

怎么说？

How so?

比如，那些在特定任务上表现良好的神经网络，将它们压缩，然后大量复制，再将它们层层堆叠。

So take neural networks that seem to work well for a specific task, compress them, and multiply it by a lot, and then you just stack them on top of it.

这就像Transformer的故事在

It's like the story of transformers in

此外，深度学习网络最终会复制某个组件，但你仍然需要整个网络才能完成任何任务。

in addition, deep learning networks, they end up, you're replicating an element, but you still need the entire network to do anything.

在这里，每个独立的元素都是一个完整的学习系统。

Here, what's going on is each individual element is a complete learning system.

这就是为什么我可以把人脑切成两半，它仍然能工作。

This is why I can take a human brain, cut it in half, and it still works.

这非常惊人。

It's pretty amazing.

它本质上是分布式的。

It's fundamentally distributed.

它本质上是分布式的、完整的建模系统。

It's fundamentally distributed, complete modeling systems.

但这是我们喜欢讲述的故事。

But that's our story we like to tell.

我猜测这个说法大体上是正确的，而且有很多证据支持这个故事，这个进化故事。

I would guess it's likely largely right, but there's a lot of evidence supporting that story, this evolutionary story.

让我产生这个想法的原因是，人类的大脑在很短的时间内迅速变大，因此很久以前就有人提出，与其创造新事物，不如只是复制某种共同的元素。

The thing which brought me to this idea is that the human brain got big very quickly, So that led to the proposal a long time ago that, well, there's this common element just instead of creating new things, it just replicated something.

我们还具有极高的灵活性。

We also are extremely flexible.

我们可以学习那些我们之前完全没有经验的东西，对吧？

We can learn things that we had no history about, right?

这表明学习算法是非常通用的。

And so that tells you that the learning algorithm is very generic.

它非常普遍，因为它不预设任何关于学习内容的先验知识。

It's very kind of universal because it doesn't assume any prior knowledge about what it's learning.

所以，把这些因素结合起来，你就会问：那么，这一切是如何产生的？

So you combine those things together and you say, Okay, well, how did that come about?

这个通用算法是从哪里来的？

Where did that universal algorithm come from?

它一定源自某种非通用的东西。

It had to come from something that wasn't universal.

它源自某种更具体的东西。

It came from something that was more specific.

所以，无论如何，这促使我们提出一个假设：在新皮层中会发现类似网格细胞和位置细胞的结构。

So anyway, this led to our hypothesis that you would find grid cells and place cell equivalents in the neocortex.

当我们首次发表关于这一理论的论文时，我们并不知道有相关证据。

And when we first published our first papers on this theory, we didn't know of evidence for that.

事实上，当时已经有一些证据了，只是我们不知道。

It turns out there was some, but we didn't know about it.

从那以后，我们开始了解到新皮层某些区域存在网格细胞的证据。

Since then, so then we became aware of evidence for grid cells in parts of the neocortex.

然后，现在又出现了新的证据。

And then now there's been new evidence coming out.

今年一月发表了一些有趣的论文。

There's some interesting papers that came out just January.

因此，我们的一个预测是：如果这个进化假设是正确的，那么我们会在新皮层的每一个柱状结构中看到类似网格细胞和位置细胞的结构，而这一点现在正逐渐得到证实。

So one of our predictions was if this evolutionary hypothesis is correct, we would see grid cell place cell equivalent cells that work like them through every column in the neocortex, and that's starting to be seen.

它们的存在意味着什么？为什么它们的存在如此重要？

What does it mean that Why is it important that they're present

因为它告诉我们，我们在探讨智能的进化起源，对吧？

Because it tells us, well, we're asking about the evolutionary origin of intelligence, right?

所以我们的理论是，皮层中的这些柱状结构遵循相同的原理，它们在建模系统。

So our theory is that these columns in the cortex are working on the same principles, They're modeling systems.

很难想象神经元是如何做到这一点的。

And it's hard to imagine how neurons do this.

所以我们说，很难想象神经元是如何学会这些事物的模型的。

And so we said, Hey, it's really hard to imagine how neurons could learn these models of things.

如果你愿意，我们可以详细讨论这一点。

We can talk about the details of that if you want.

但大脑中还有另一个部分，我们知道它能学习环境的模型。

But there's this other part of the brain we know that learns models of environments.

那么，用来学习建模这个房间的机制，是否也能用来学习建模水瓶呢？

So could that mechanism to learn to model this room be used to learn to model the water bottle?

这是同一个机制吗？

Is it the same mechanism?

所以我们认为，大脑更可能使用的是相同的机制，这种情况下，它会具备这些等效的细胞类型。

So we said it's much more likely the brain's using the same mechanism, in which case it would have these equivalent cell types.

因此，整个理论的基础是这些皮层柱具有参考框架，并且正在学习这些模型，而网格细胞则构建了这些参考框架。

So basically the whole theory is built on the idea that these columns have reference frames and they're learning these models and these grid cells create these reference frames.

所以，从某种意义上说，这个理论的主要预测是：我们将在新皮层的每个柱中发现这些等效机制，这表明它们正是在做这样的事情。

So it's basically, in some sense, the major predictive part of this theory is that we will find these equivalent mechanisms in each column in the near cortex, which tells us that that's what they're doing.

它们正在学习关于世界的感知-运动模型。

They're learning these sensory motor models of the world.

我们相当确信这会发生，但现在我们正在看到

We're pretty confident that would happen, but now we're seeing

证据。

the evidence.

因此，进化过程、自然界，会大量进行复制和粘贴，然后观察会发生什么。

So the evolutionary process, nature, does a lot of copy and paste and see what happens.

是的，是的，这个过程没有方向性，但它只是发现：嘿，如果我把这些元素复制更多，会发生什么？

Yeah, yeah, there's no direction to it, but it just found out like, Hey, if I took these elements and made more of them, what happens?

让我们把它们连接到眼睛，再连接到耳朵。

And let's hook them up to the eyes and let's hook them up to ears.

这似乎效果不错。

And that seems to work pretty well

对我们来说是这样。

for us.

再回到我们关于集体智能的讨论，你是否有时认为这也是另一种复制粘贴的体现——即在人类中复制这些大脑，大量复制，然后建立社会结构，使它们几乎像一个单一的大脑那样运作？

Again, just to take a quick step back to our conversation of collective intelligence, do you sometimes see that as just another copy and paste aspect, is copying and pasting these brains in humans and making a lot of them, and then creating social structures that then almost operate as a single brain.

我本来不会这么说，

I wouldn't have

但你说了，听起来还挺有道理的。

said it, but you said it and it sounded pretty good.

所以，在你看来，大脑是独立存在的。

So, to you, the brain is its own thing.

是的，我的目标是理解新皮层是如何工作的。

Yeah, I mean, our goal is to understand how the neocortex works.

我们可以争论这一点对于理解人脑有多重要，因为它并不是整个人脑。

We can argue how essential that is to understanding the human brain, because it's not the entire human brain.

你可以争论这一点对于理解人类智能有多重要。

You can argue how essential that is to understanding human intelligence.

你可以争论这一点对于群体智能有多重要。

You can argue how essential it is to sort of communal intelligence.

我们的目标是理解新皮层。

Our goal was to understand the neocortex.

是的，那么新皮层是什么？它在大脑的各类功能中处于什么位置？

Yeah, so what is the neocortex and where does it fit in the various aspect of what the brain does?

比如，它对你来说有多重要？

Like, how important is it to you?

当然，我前面又提到了，它约占人脑体积的70%到75%。

Well, obviously, again, I mentioned it again in the beginning, it's about 7075% of the volume of a human brain.

因此，从体积上看，它在大脑中占据主导地位。

So, it dominates our brain in terms of size.

不是从神经元的数量上，而是从体积上来说。

Not in terms of number of neurons, but in terms of size.

体积并不是一切，杰夫。

Size isn't everything, Jeff.

我知道，但情况并非如此。

I know, but it's not that.

我们知道，所有高级视觉、听觉和触觉都发生在新皮层。

We know that all high level vision, hearing and touch happens in the neocortex.

我们知道，所有语言的产生和理解都发生在新皮层，无论是口语、书面语、手语，还是数学语言、物理语言、音乐、数学，等等。

We know that all language occurs and is understood in the neocortex, whether that's spoken language, written language, sign language, whether language of mathematics, language of physics, music, math, you know.

我们知道，所有高级规划和思考都发生在新皮层。

We know that all high level planning and thinking occurs in the neurocortex.

如果我要问，你大脑的哪个部分设计了计算机、理解编程并创作音乐？

If I were to say, What part of your brain designed a computer and understands programming and creates music?

全部都是新皮层。

It's all the neurocortex.

那么，这只是一个不可否认的事实。

So then, that's just an undeniable fact.

但我们的大脑其他部分也很重要，对吧？

But then, there's other parts of our brain that are important too, right?

我们的情绪状态，调节我们的身体。

Our emotional states, regulating our body.

所以，我喜欢这样看待这个问题：你能不借助大脑的其他部分来理解新皮层吗？

So, the way I like to look at it is, can you understand the neocortex without the rest of the brain?

有些人说不能，但我认为完全可以。

And some people say you can't, and I think absolutely you can.

它们并不是不相互作用，但你仍然可以理解它们。

It's not that they're not interacting, but you can understand them.

你能不理解恐惧的情绪来理解新皮层吗？

Can you understand the neocortex without understanding the emotions of fear?

是的，你可以。

Yes, you can.

你可以理解这个系统是如何运作的。

You can understand how the system works.

它只是一个建模系统。

It's just a modeling system.

我在书中打了个比方，说它就像一张世界地图。

I make the analogy in the book that it's like a map of the world.

这张地图如何被使用，取决于使用它的人。

And how that map is used depends on who's using it.

因此，我们在新皮层中对世界的认知如何体现为人类的行为，取决于我们大脑的其他部分。

So how our map of our world in our neocortex, how manifest as a human, depends on the rest of our brain.

我们的动机是什么？

What are our motivations?

我的欲望是什么？

What are my desires?

我是个好人，还是个坏人？

Am I a nice guy or not a nice guy?

我是个骗子吗，还是不是骗子？

Am I a cheater or am I not a cheater?

生活中各种事情对我来说有多重要？

How important different things are in my life?

新皮层可以独立理解。

Neocortex can be understood on its own.

作为一名神经科学家，我知道这些互动无处不在，但我希望说我不了解它们，我们也不去思考它们。

I say that as a neuroscientist, I know there's all these interactions, I want to say I don't know them and we don't think about them.

但从普通人的角度来看，你可以说它是一个建模系统。

But from a layperson's point of view, you can say it's a modeling system.

我通常不太会思考你多次提到的智能的群体层面。

I don't generally think too much about the communal aspect of intelligence, which you've brought up a number of times already.

所以这并不是我真正关心的问题。

So that's not really been my concern.

我只是好奇，从宇宙的起源开始，是否存在着形成生命体的复杂性区域。

I just wonder if there's a continuum from the origin of the universe, like there's pockets of complexities that form living organisms.

我在想，如果我们看看人类，我们觉得自己站在顶端，但我不禁怀疑，每一种生命形式、每一个复杂性的小群体，可能都觉得自个儿是——恕我直言——最牛的。

I wonder if we're just, if you look at humans, we feel like we're at the top, but I wonder if there's like just where everybody probably, every living type pocket of complexity probably thinks they're the, pardon the French, they're the shit.

他们觉得自己站在金字塔的顶端。

They're at the top of the pyramid.

嗯，他们在思考。

Well, they're thinking.

那么，什么是思考呢？

Well, then what is thinking?

好吧，

Well,

从他们的角度来看，关键在于，在他们对世界的认知里，他们觉得自己处于顶端。

in the sense, the whole point is in their sense of the world, their sense is that they're at the top of it.

我觉得乌龟是什么，

I think What is a turtle

为了什么？

for?

但你提出了复杂性和复杂性理论的问题，这是科学中一个非常有趣的重大课题。

But you're bringing up the problems of complexity and complexity theory, it's a huge interesting problem in science.

我认为我们在理解复杂系统方面取得的进展出人意料地少。因此，圣塔菲研究所正是为了研究这一领域而成立的，甚至连那里的科学家也会说，这真的很难。

And I think we've made surprisingly little progress in understanding complex systems And in so, the Santa Fe Institute was founded to study this, and even the scientists there will say, It's really hard.

我们还未能真正弄清楚，你知道的，这门科学尚未真正成型。

We haven't really been able to figure out exactly, you know, that science hasn't really congealed yet.

我们仍在努力探索这门科学的基本要素。

We're still trying to figure out the basic elements of that science.

复杂性从何而来？它是什么？如何定义它？无论是DNA创造生物体或表型，还是个体创造社会，或是蚂蚁、市场等等。

Where does complexity come from and what is it and how you define it, whether it's DNA creating bodies or phenotypes or individuals creating societies or ants and markets and so on.

这是一件非常复杂的事情。

It's a very complex thing.

我不是一个复杂性理论专家，对吧？

I'm not a complexity theorist person, right?

我认为你应该问，大脑本身就是一个复杂系统，那么我们能理解它吗？

I think you need to ask, well, the brain itself is complex system, so can we understand that?

我认为我们在理解大脑如何工作方面已经取得了很大进展。

I think we've made a lot of progress understanding how the brain works.

但我还没有具体说明，我们在复杂性谱系上处于什么位置。

But I haven't brought it down to like, Oh, well, where are we on in the complexity spectrum?

这真是个好问题。

It's like, it's a great question.

我更倾向于认为，我们并不特殊。

I prefer for that answer to be we're not special.

如果诚实面对，我们很可能并不特殊。

It seems like, if we're honest, most likely we're not special.

所以，如果存在一个谱系，我们很可能并不处于某种重要的位置。

So, if there is a spectrum, we're probably not in some kind of significant I place to that

我认为有一件事我们可以肯定地说我们是特殊的，当然，仅限于地球，我不是说这不好，那就是，如果我们考虑知识，我们所知道的，人类大脑显然是唯一拥有某些类型知识的大脑。

think there's one thing we could say that we are special, and again, only here on Earth, I'm not saying I'm bad, is that if we think about knowledge, what we know, we clearly, human brains are the only brains that have certain types of knowledge.

我们是地球上唯一能理解地球是什么、它有多古老、宇宙整体是什么样子的大脑。

We're the only brains on this earth to understand what the earth is, how old it is, the universe is a picture as a whole.

我们是唯一理解DNA和物种起源的生物。

We're the only organism to understand DNA and the origins of species.

地球上没有任何其他物种拥有这种知识。

No other species on this planet has that knowledge.

所以，如果我们要思考，我喜欢认为人类的一项追求就是尽可能地理解宇宙。

So, if we think about, I like to think about one of the endeavors of humanity is to understand the universe as much as we can.

我认为我们的物种在这方面的进展无疑是更远的。

I think our species is further along in that undeniably.

我们的理论是对是错，我们可以争论。

Whether our theories are right or wrong, we can debate.

但至少我们有理论。

But at least we have theories.

我们知道太阳是什么，什么是核聚变，什么是黑洞。

We know what the sun is and how fusion is and what black holes are.

我们还懂得广义相对论，而其他任何动物都没有这些知识。

And we know general theory of relativity, and no other animal has any of this knowledge.

因此，从这个意义上说，我们是特殊的。

So from that sense, we're special.

我们在宇宙的复杂性层级中算是特殊的吗？

Are we special in terms of the hierarchy of complexity in the universe?

可能不是。

Probably not.

我们能看一下神经元吗？

Can we look at a neuron?

是的，你说预测发生在神经元中。

Yeah, you say that prediction happens in the neuron.

那是什么意思？

What does that mean?

所以，传统上神经元被视为

So neuron traditionally is seen as

大脑的基本单元。

the basic element of the brain.

我之前提到过，预测是我们研究的核心议题。

So I mentioned this earlier, that prediction was our research agenda.

我们提出的问题是：大脑是如何做出预测的？

We said, Okay, how does the brain make a prediction?

比如，我正要拿起这个水瓶，我的大脑正在预测我手指各个部位会感受到什么。

Like, I'm about to grab this water bottle and my brain is predicting what I'm going to feel on all my parts of my fingers.

如果我在这里的任何部位感受到异常，我就会立刻注意到。

If I felt something really odd on any part here, I'd notice it.

所以，我的大脑在抓取这个物体时，正在预测它会感受到什么。

So my brain is predicting what it's going to feel as I grab this thing.

那么，这种预测在神经组织中是如何体现的呢？

So how does that manifest itself in neural tissue, right?

我们的大脑由神经元组成，里面有化学物质、神经元、电脉冲，它们彼此连接。

Our brain's made of neurons and there's chemicals and there's neurons and there's spikes and they're connected.

预测究竟发生在哪个环节？

Where is the prediction going on?

一个可能的观点是，当我进行预测时，某个神经元必须提前放电。

And one argument could be that, well, when I'm predicting something, a neuron must be firing in advance.

也就是说，这个神经元代表了你即将感受到的东西，它正在放电，发送一个脉冲。

It's like, okay, this neuron represents what you're going to feel and it's firing, it's sending a spike.

在某种程度上，这确实会发生。

And certainly that happens to some extent.

但我们的预测无处不在，我们正在做出如此多的预测，而自己却完全 unaware，绝大多数情况下你根本不知道自己在做这些，我们试图弄清楚这怎么可能？

But our predictions are so ubiquitous that we're making so many of them which we're totally unaware of, just the vast majority of have no idea that you're doing this, We were trying to figure how could this be?

这些预测究竟发生在哪里？

Where are these happening?

除非你们坚持要听，否则我不会详细讲述整个故事，但我们逐渐意识到，你的大部分预测其实发生在单个神经元内部，尤其是最常见的神经元——锥体细胞。

And I won't walk you through the whole story unless you insist upon it, but we came to the realization that most of your predictions are occurring inside individual neurons, especially the most common neuron, the pyramidal cells.

神经元具有一种特性。

And there's a property of neurons.

每个人都知道，或者大多数人知道，神经元是一种细胞，它有一个被称为动作电位的脉冲，并传递信息。

Everyone knows or most people know that a neuron is a cell and it has this spike called an action potential and it sends information.

但我们现在知道，神经元内部存在这些脉冲。

But we now know that there's these spikes internal to the neuron.

它们被称为树突脉冲。

They're called dendritic spikes.

它们沿着神经元的分支传播，但仅限于内部，不会传出。

They travel along the branches of the neuron and they don't leave internal only.

树突脉冲的数量远多于动作电位。

There's far more dendritic spikes than there are action potentials.

多得多。

Far more.

它们一直在发生。

They're happening all the time.

我们逐渐明白，那些反复出现的树突脉冲实际上是一种预测形式。

And what we came to understand that those dendritic spikes, the ones that are recurring, are actually a form of prediction.

它们在告诉神经元：我预期自己很快就会活跃起来。

They're telling the neuron, the neuron is saying, I expect that I might become active shortly.

所以，内部的尖峰是一种暗示：你很快可能会产生外部尖峰。

So, the internal spike is a way of saying, You might be generating external spikes soon.

我预测你即将变得活跃。

I predicted you're going to become active.

我们在2016年发表了一篇论文，解释了这种现象如何在神经组织中体现，以及这一切是如何协同运作的。

And we wrote a paper in 2016 which explained how this manifests itself in neural tissue and how it is that this all works together.

但我们认为，有大量的证据支持这一观点。

But the vast, we think there's a lot of evidence supporting it.

因此，我们认为大多数预测都是内部发生的。

So, that's where we think that most of these predictions are internal.

这就是为什么你不能说它们只存在于神经元内部，你

That's why you can't be they're internal to a neuron, you

无法感知它们。

can't perceive them.

从理解单个神经元的预测机制来看，你认为这能为我们揭示大脑中更小脑区乃至整个大脑的预测能力带来深刻的洞见吗？

From understanding the prediction mechanism of a single neuron, do you think there's deep insights to be gained about the prediction capabilities of the mini brains within the bigger brain and the brain?

哦，是的，是的，是的。

Oh, yeah, yeah, yeah.

因此，单个神经元内部的预测并没有太大用处。

So having a prediction inside an individual neuron is not that useful.

那又怎样？

So what?

它在神经组织中的表现方式是，神经元会发出这些脉冲，这是一种非常单一的事件；如果神经元预测自己即将活跃，它会比平时提前几毫秒发出脉冲。

The way it manifests itself in neural tissue is that a neuron emits these spikes, so a very singular type event, if a neuron is predicting that it's going to be active, it emits its spike a little bit sooner, just a few milliseconds sooner than it would have otherwise.

就像我在书中举的比喻，这就像赛跑中站在起跑器上的短跑运动员。

It's like, I give the analogy in the book, it's like a sprinter on a starting blocks in a race.

如果有人喊‘预备，跑’，你就站起来，准备出发。

And if someone says, Get ready, set, you get up and you're ready to go.

当比赛真正开始时，你会获得一个稍早的起跑。

And then when your race starts, you get a little bit early start.

所以这个‘预备’就像预测，让神经元能更快进入待命状态。

So that ready set is like the prediction and the neuron's ready to go quicker.

当大量神经元一起工作并接收这些输入时，处于预测状态、预期激活的神经元如果真的被激活，就会更早地发生，从而抑制其他所有神经元，导致大脑中产生不同的表征。

And what happens is when you have a whole bunch of neurons together and they're all getting these inputs, the ones that are in the predictive state, the ones that are anticipating to become active, If they do become active, they happen sooner, they disable everything else, and it leads to different representations in the brain.

所以这并不仅仅局限于单个神经元。

So it's not isolated just to the neuron.

预测发生在神经元内部，但网络的行为会发生变化。

The prediction occurs within the neuron, but the network behavior changes.

在不同的预测和不同输入下，会产生不同的表征。

So what happens under different predictions, different inputs have different representations.

因此，我所预测的内容在不同情境下会有所不同。

So what I predict is going to be different under different contexts.

我的输入在不同情境下也会有所不同。

What my input will be is different under different contexts.

所以，这是整个理论的关键所在，解释了它是如何运作的。

So this is a key to the whole theory, how this works.

所以，如果要数一数‘千脑理论’中的大脑数量，你会怎么计算呢？

So the theory of the thousand brains, if you were to count the number of brains, how would you do it?

千脑理论认为，你新皮层中的每一个皮质柱都是一个完整的建模系统。

The thousand brain theory says that basically every cortical column in your neocortex is a complete modeling system.

当我问自己，我对咖啡杯的模型存在于哪里时，答案不是存在于其中一个模型中，而是存在于成千上万个模型中。

And that when I ask where do I have a model of something like a coffee cup, it's not in one of those models, it's in thousands of those models.

有成千上万个咖啡杯的模型。

There's thousands of models of coffee cups.

这就是千脑理论所阐述的内容。

That's what the Thousand Brains does.

然后还有一个投票机制。

And then there's a voting mechanism.

接下来是投票机制，这是你意识所感知到的部分，它导致了你单一的知觉。

Then there's a voting mechanism, is the thing which you're conscious of, which leads to your singular perception.

这就是你之所以能感知到某物的原因。

That's why you perceive something.

所以，这就是千脑理论。

So, that's the Thousand Brains theory.

我们是如何得出这一理论的细节非常复杂。

The details of how we got to that theory are complicated.

并不是我们某一天突然想到的。

It wasn't that we just thought of it one day.

其中一个细节是我们必须问：模型是如何做出预测的？

And one of those details is we had to ask, how does a model make predictions?

当我们谈到这些预测性神经元时，这正是这一理论的一部分。

And when we've talked about these predictive neurons, that's part of this theory.

这就像说，哦，这只是个细节，但它就像门上的一道裂缝。

It's like saying, oh, it's a detail, but it was like a crack in the door.

我们该如何弄清楚这些神经元是如何做到这一点的？

It was like, How are we going figure out how these neurons do this?

这里到底发生了什么？

What is going on here?

所以我们只是把预测看作：嗯，我们知道这是无处不在的。

So, we just looked at prediction as like, Well, we know that's ubiquitous.

我们知道皮层的每个部分都在进行预测。

We know that every part of the cortex is making predictions.

因此，无论这个预测系统是什么，它都会无处不在。

Therefore, whatever the predictive system is, it's going to be everywhere.

我们知道有无数个预测同时在发生，所以让我们试着逐步分析，提出问题：神经元是如何做出这些预测的？

We know there's a gazillion predictions happening at once, so let's see if can start teasing apart, ask questions about how could neurons be making these predictions.

这逐渐发展成了我们现在所拥有的‘千脑理论’，我可以简单地陈述它，但我们并不是一开始就想到的。

And that sort of built up to now what we have, this Thousand Brains Theory, which is I can state it simply, but we just didn't think of it.

我们必须一步一步地达到这个结论。

We had to get there step by step.

花了很多年才走到这一步。

It took years to get there.

参考框架在哪里发挥作用？

And where does reference frames fit in?

是的。

So, yeah.

好的。

Okay.

所以，再来说一下参考框架，我之前提到过一个房子的模型，我说过，如果你要在计算机里构建一个房子的模型，就需要一个参考框架。

So again, a reference frame, I mentioned earlier about a model of a house, and I said, you're going build a model of a house in a computer, they have a reference frame.

你可以把参考框架想象成笛卡尔坐标系，也就是X、Y和Z轴。

And you can think of reference frame like Cartesian coordinates, like X, Y, and Z axes.

因此，我可以说：我要设计一栋房子。

So I could say, Oh, I'm going to design a house.

我可以说：前门位于这个位置，X、Y、Z，屋顶位于这个位置，X、Y、Z，以此类推。

I can say, Well, the front door is at this location, X, Y, Z, and the roof is at this location, X, Y, Z, and so on.

这就是所谓的参考框架。

That's the type of reference frame.

结果发现，为了做出预测，我在书里引导你做了一个思想实验，当时我预测当我触摸咖啡杯时，我的手指会有什么感觉。

So it turns out for you to make a prediction, and I walk you through the thought experiment in the book where I was predicting what my finger was going to feel when I touched the coffee cup.

那是一个陶瓷咖啡杯，但这个也行。

It was a ceramic coffee cup, but this one will do.

我意识到，要预测我的手指会感受到什么，比如，我猜它会和摸这个东西的感觉不一样，如果我摸的是那个洞或者底部的这个部分，感觉会有什么不同。

And what I realized is that to make a prediction what my finger's going to feel, like, I guess it's going feel different than this, what's it feel different if I touch the hole or this thing on the bottom.

做出这个预测。

Make that prediction.

大脑皮层需要知道手指尖相对于咖啡杯的位置，确切地相对于咖啡杯的位置。

The cortex needs to know where the finger is, the tip of the finger, relative to the coffee cup, and exactly relative to the coffee cup.

要做到这一点，我必须为咖啡杯建立一个参考系。

And to do that, I have to have a reference frame for the coffee cup.

它必须能够表示我的手指相对于咖啡杯的位置。

It has to have a way of representing the location of my finger to the coffee cup.

然后我们意识到，当然，你皮肤的每一部分都必须有一个相对于接触物的参考系。

And then we realized, of course, every part of your skin has to have a reference frame relative to things that touch.

接着，我们对视觉也做了同样的分析。

Then we did the same thing with vision.

因此，无论是触摸还是视觉感知，当你在移动眼睛或手指时，要做出预测，参考系都是必需的——这是知道该预测什么的基本要求。

So, the idea that a reference frame is necessary to make a prediction when you're touching something or when you're seeing something and you're moving your eyes or you're moving your fingers, it's just a requirement to know what to predict.

如果我有一个结构，我就会做出预测。

If I have a structure, I'm going to make a prediction.

我必须知道我正在看或触摸的位置。

I have to know where it is I'm looking or touching it.

那么，我们就会问：神经元是如何建立参考框架的？

So, then we say, Well, how do neurons make reference frames?

这并不明显。

It's not obvious.

大脑中并不存在XYZ坐标。

XYZ coordinates don't exist in the brain.

事情根本不是这样运作的。

It's just not the way it works.

于是，我们转向了大脑更古老的区域——海马体和内嗅皮层，我们知道在这些区域中，存在着对房间或环境的参考框架。

So, that's when we looked at the older part of the brain, the hippocampus and the entorhinal cortex, where we knew that in that part of the brain there's a reference frame for a room or a reference frame for an environment.

还记得我之前提到过，你可以为这个房间画一张地图吗？

Remember I talked earlier about how you could make a map of this room?

所以我们说，它们在那里实现了参考框架。

So we said, Oh, they are implementing reference frames there.

所以我们知道，每个柱状区域的四分之一都需要存在参考框架，这是通过推理得出的。

So we knew that reference frames needed to exist in every quarter of a column, and so that was a deductive thing.

我们只是推断出来的。

We just deduced it.

它必须存在。

It has to exist.

因此，你利用了哺乳动物原本识别自身在特定空间中位置的能力，并将其逐步应用到更高层次上。

So you take the old mammalian ability to know where you are in a particular space and you start applying that to higher and higher levels.

是的，首先你把它应用到手指的位置上。

Yeah, first you apply it to like where your finger is.

这是我思考这个问题的方式。

So here's the way I think about it.

大脑的古老部分会问：我的身体在这个房间里的位置在哪？

The old part of the brain says, Where's my body in this room?

大脑的新部分会问：我的手指相对于这个物体在哪里？

The new part of the brain says, Where's my finger relative to this object?

在哪里

Where

我的视网膜的一小部分相对于这个物体在哪里？

is a section of my retina relative to this object?

我正在观察一小块角膜。

I'm looking at one little cornea.

它相对于我视网膜的这一区域在哪里？

Where is that relative to this patch of my retina?

然后我们把同样的原理应用到概念、数学、物理、人类学等任何领域，

And then we take the same thing and apply it to concepts, mathematics, physics, humanity, whatever you

你愿意加入清单的任何事物，你都在思考自己的死亡。

want In to the inventory, you're pondering your own mortality.

好吧，随便吧。

Well, whatever.

但重点是，当我们思考世界、拥有对世界的认知时，这些知识是如何组织的，莱克？

But the point is when we think about the world, when we have knowledge about the world, how is that knowledge organized, Lex?

它在你的大脑里处于什么位置？

Where is it in your head?

答案是，它存在于参考框架中。

The answer is it's in reference frames.

所以，我通过学习这个水瓶的结构——各个特征之间的相对关系——当我思考历史、民主或数学时，同样的基本底层结构正在发生。

So, the way I learned the structure of this water bottle where the features are relative to each other, when I think about history or democracy or mathematics, the same basic underlying structure is happening.

对于你所赋予事物的知识，存在着相应的参考框架。

There's reference frames for where the knowledge that you're assigning things to.

因此，在这本书中，我举了数学、语言和政治等例子。

So, in the book, I go through examples like mathematics and language and politics.

但神经科学中的证据非常明确。

But the evidence is very clear in the neuroscience.

我们用来模拟这个咖啡杯的机制，同样也会用来模拟高级思维，比如人类的消亡，或者你想要思考的任何事情。

The same mechanism that we use to model this coffee cup, we're going to use to model high level thoughts, the demise of humanity, whatever you wanna think about.

想想这些更高维度、更高层次概念的表征方式有多不同，它们在参考框架与空间表征上的差异，这很有趣。

It's interesting to think about how different are the representations of those higher dimensional concepts, higher level concepts, how different the representation there is in terms of reference frames versus spatial.

但有趣的是，这是一种不同的应用，却使用了完全相同的机制。

But interesting thing, it's a different application, but it's the exact same mechanism.

但更高层次的概念难道没有某种层级性吗？

But isn't there some aspect to higher level concepts that they seem to be hierarchical.

就像它们似乎只是

Like they just seem

将大量信息整合到其中。

to integrate a lot of information into them.

我们的物理对象也是如此。

So is our physical objects.

以这个水瓶为例。

So, take this water bottle.

我对这个品牌并不特别偏爱，但这是斐济水瓶，上面有个标志。

I'm not particular to this brand, but this is a Fiji water bottle, and it has a logo on it.

我在我的书里用这个例子。

I use this example in my book.

我们公司的咖啡杯上也有一个标志。

Our company's coffee cup has a logo on it.

但这个物体是分层的。

But this object is hierarchical.

它有一个圆柱体和一个盖子，然后上面还有一个标志。

It's got like a cylinder and a cap, but then it has this logo on it.

而这个标志上有一个单词。

And the logo has a word.

这个单词由字母组成，字母又具有不同的特征。

The word has letters, the letters have different features.

所以我不需要记住，也不需要去思考这些细节。

And so I don't have to remember, I don't have to think about this.

所以我只会说，这个水瓶上有个Fiji的标志。

So I say, Oh, there's a Fiji logo on this water bottle.

我不必逐一去想，Fiji标志是什么样的。

I don't have to go through and say, Oh, what is a Fiji logo?

它是由F、I、J、I组成的，还有一朵木槿花，而且上面有花蕊。

It's the F and I and a J and I, and there's a hibiscus flower, and oh, it has the stamen on it.

我不必这么做。

I don't have to do that.

我只是将所有这些内容整合成某种层次化的表示方式。

I just incorporate all of that in some sort of hierarchical representation.

我说，把这个标志放在这个水瓶上。

I say, Put this logo on this water bottle.

是的。

Yeah.

然后标志上有个文字，文字由字母组成，全部都是层次化的。

And then the logo has a word and the word has letters, all hierarchical.

所有这些东西都很庞大。

All that stuff is big.

大脑能瞬间完成这一切，真是太神奇了。

It's amazing that the brain instantly just does all that.

是的。

Yeah.

水是液体的概念，口渴时可以喝它的概念，还有品牌存在的概念。

The idea that there's water, it's liquid, and the idea that you can drink it when you're thirsty, the idea that there's brands.

是的。

Yeah.

一旦你开始处理，所有这些信息就会瞬间整合到整个系统中。

And then there's like all of that information is instantly built into the whole thing once you proceed.

所以，我想回到你关于层次结构表示的观点。

So, wanted to get back to your point about hierarchical representation.

世界本身是层次化的，对吧？

The world itself is hierarchical, right?

我可以拿起我面前的这个麦克风。

And I can take this microphone in front of me.

我知道里面会有一些电子元件。

I know inside there's going to be some electronics.

我知道会有一些电线，我知道会有一个来回振动的振膜。

I know there's going to be some wires, I know there's going to be a little diaphragm that moves back and forth.

我看不到它们，但我了解它们的存在。

I don't see that, but I know it.

所以，世界上的一切都是分层的。

So, everything in the world is hierarchical.

你走进一个房间，它是由其他组件构成的。

You just go into a room, it's composed of other components.

厨房里有一个冰箱。

The kitchen has a refrigerator.

冰箱有门，门有铰链，铰链有螺丝和销钉。

The refrigerator has a door, the door has a hinge, the hinge has screws and pin.

因此，存在于每个皮质柱中的建模系统会学习物体的分层结构。

So anyway, the modeling system that exists in every cortical column learns the hierarchical structure of objects.

所以，这个米粒大小的结构中包含着一个非常精密的建模系统。

So it's a very sophisticated modeling system in this grain of rice.

很难想象，但这个米粒大小的结构能完成极其复杂的事情。

It's hard to imagine, but this grain of rice can do really sophisticated things.

它里面包含了十万神经元。

It's got 100,000 neurons in it.

它非常精密。

It's very sophisticated.

因此，能够建模水瓶或咖啡杯的同一机制，也能建模抽象概念对象。

So, the same mechanism that can model a water bottle or a coffee cup can model conceptual objects as well.

这是维农·莫姆基斯特很多年前发现的惊人之处：我们所做的一切，背后都依赖于同一种皮层算法。

That's the beauty of this discovery that this guy, Vernon Maumkistle, made many, many years ago, which is that there's a single cortical algorithm underlying everything we're doing.

所以，常识性概念和高级概念都是以相同的方式表示的吗？

So, common sense concepts and higher level concepts are all represented in the same way?

它们都由相同的机制处理，是的。

They're set in the same mechanisms, yeah.

这有点像计算机，对吧？

It's a little bit like computers, right?

所有计算机都是通用图灵机。

All computers are universal Turing machines.

即使是我的烤面包机里那个小小的，还有运行在某个云服务器上的大型计算机，它们都基于同样的原理。

Even the little teeny one that's my toaster and the big one that's running some cloud server some They're all running on the same principle.

它们可以执行不同的任务。

They can apply different things.

所以大脑也是基于同样的原理构建的。

So the brain is all built on the same principle.

这一切都关乎通过运动和参考框架学习这些模型，它可以应用于像水瓶和咖啡杯这样简单的东西，也可以应用于思考诸如‘人类的未来会怎样？’这样的问题。

It's all about learning these models, structured models using movement and reference frames, and it can be applied to something as simple as a water bottle and a coffee cup, it can be applied to thinking like, What's the future of humanity?

你桌上为什么有个刺猬？

Why do you have a hedgehog on your desk?

我不知道。

I don't know.

没人知道。

Nobody knows.

嗯，这个

Well, the

我觉得它是一只刺猬。

I think it's a hedgehog.

没错，这是雾中的刺猬。

That's right, it's a hedgehog in the fog.

这是一个俄罗斯的典故。

It's a Russian reference.

这让你对工程化常识推理的难度有什么看法或希望吗？

Does it give you any inclination or hope about how difficult it is to engineer common sense reasoning?

那么这个过程有多复杂？

So how complicated is this whole process?

所以从大脑的角度来看，这是一种工程奇迹，还是只是一堆简单的东西堆叠在一起？

So looking at the brain, is this a marvel of engineering, or is it pretty dumb stuff stacked on top of each other over and

可以两者都是。

can be both.

两者都是？

Both?

难道不能两者都是吗？

Can't it it be both, right?

我不知道是否可以两者都是，因为如果这是一项了不起的工程成就，那就意味着进化做了大量工作。

I don't know if it can be both because if it's an incredible engineering job, that means it's So Evolution did a lot of work.

是的，但后来它只是复制了这一点。

Yeah, but then it just copied that.

所以，正如我前面所说，弄清楚如何建模像空间这样的东西非常困难。

So as I said earlier, figuring out how to model something like a space is really hard.

在进化过程中，必须经历很多技巧。

In evolution, have to go through a lot of trick.

我之前提到的这些细胞，即网格细胞和位置细胞，非常复杂。

And these cells I was talking about, these grid cells and place cells, they're really complicated.

这可不是简单的东西。

This is not simple stuff.

这种神经组织依靠的是非常出人意料、奇特的机制。

This neural tissue works on these really unexpected, weird mechanisms.

但它做到了。

But it did it.

它找到了解决方法。

It figured it out.

但现在你可以轻松复制很多份。

But now you could just make lots of copies of it.

但要找到它，是的，这是一个非常有趣的想法——它是由大量基本的微型大脑副本构成的，但问题是，找到那个可以有效复制粘贴的微型大脑究竟有多难。

But then finding, Yeah, so it's a very interesting idea that it's a lot of copies of a basic mini brain, but the question is how difficult it is to find that mini brain that you can copy and paste effectively.

如今，我们已经足够了解如何构建它了。

Well, today we know enough to build this.

我坐在这里，清楚地知道我们需要走的步骤。

I'm sitting here with, I know the steps we have to go.