从脑电信号到突破性进展：神经科技、人工智能与金融科技——Gerwin Schalk博士专访

尽管从技术上讲仍有许多粗糙之处，但你可以以这种非常亲密的方式与大脑互动。

And you can really work with the brain in this very intimate kind of way, even though it's technically still crude in many ways.

这确实是我们唯一能直接接触大脑的情况。

It's really the only circumstance where we have this direct access to the brain.

所以从多种角度来看，这都是非常令人兴奋的经历。

So that was very exciting from, yeah, really lots of different perspectives, very exciting experiences.

是的，非常感谢。

Yeah, thank you so much.

显然这些经历激励你进一步近距离研究人脑，并与全球各地的科学家开展不同项目和合作。

Obviously all those experiences inspired you to further study the human brain up close and personal, and you develop different projects and collaborations with the various scientists across the globe.

有没有哪项研究是你特别引以为豪的？

What is maybe one study that you are particularly proud of?

通过与这些患者合作，利用这些电极及其信号，我们清楚地发现它们不仅可用于脑机接口实验和刚才描述的功能映射技术，还能促进对大脑的深入理解和神经科学的进步。

By working with these patients and by using these electrodes and the signals from these electrodes, it became very clear that they can be useful for not only for brain computer interfacing experiments, not only for this functional mapping technique that we just described, but also to achieve a better understanding of the brain and a betterment of neuroscience.

回想当时，当我们刚开始时，实际上只有三种常用的成像技术。

Again, when you think about at that time, there really were only three imaging techniques that were in common use when we started.

那是已经存在很长时间的头皮脑电图记录技术，它便宜、简单且安全，但当然无法告诉你大脑中事件发生的位置。

That was scalp recorded EEG that's been around for a long time, that again is cheap, easy, and safe, but of course it cannot tell you where things happen in the brain.

它只能告诉你大脑中事件发生的时间，而且连这一点也做得不太好。

It can only tell you when things happen in the brain, and even that not so great.

但当然它非常容易操作。

But of course it's very easy.

或者记录大脑内部单个神经元的活动，现在你虽然能记录单个神经元，但你确实不知道整个大脑发生了什么。

Or recordings from individual neurons inside the brain, now you're recording from individual neurons, but you really don't know what happens across the brain.

因此，现在我们首次拥有了一种技术，能够在空间和时间上提供极高的分辨率。

So now for the first time we actually have a technique that can give you both in space and in time, very good resolution.

它实际上可以展示大脑信号如何从一个区域传递到另一个区域。

It can actually show you how the brain signals progress from one area to another.

这确实是其他技术无法以这种方式实现的。

This is really something that no other technique can do in this way.

通过研究这些信号，我们清楚地发现它们具备这种能力，最初是通过复制三十年前在猴子身上进行的简单研究开始的，比如人们移动操纵杆或在空间中移动手部、说出不同词语、聆听不同音调，我们能够观察到大脑中某个位置正在发生什么，并从中了解到这些信号所揭示的惊人功能细节。

And it became clear by working with these signals that they had this capability, which first started by replicating simple studies that have been done in monkeys thirty years ago, people moving a joystick or moving their hand around in space, saying different words, listening to different tones, and we could see, oh, there's something happening in the brain at this location or that location, and we can learn really amazing detail of function from these signals.

我们可以识别出他们正在听什么声音。

We can learn what sounds they are hearing.

我们可以知道他们手部移动的位置，他们想象中手部移动的位置，以及他们意图移动或实际移动的位置。

We can learn where they move their hand, where they imagine to move their hand, where they intend to move or actually move.

我们可以判断他们是否看着中心点却将注意力放在这边或那边的物体上。

We can learn whether they look in the center but pay attention to this object over here or versus paying attention to this object over here.

我们能够从脑信号中观察到所有这些现象。

We can see all these things from brain signals.

我们给许多患者播放了平克·弗洛伊德的《De Walt》这首歌，实际上成功通过直接从大脑表面记录的信号重建了这首歌曲的原声。

We played Pink Floyd, the song De Walt, to many, many patients and actually were able to reconstruct the song, the actual song from recordings directly from the surface of the brain.

我们与一位杰出的合作者——医生共同完成的研究在不到一年前刚刚发表。

There was actually a study that we just got published less than a year ago with a great collaborator, Doctor.

伯克利的鲍勃·奈特博士，以及另一位我们合作多年的优秀同事。

Bob Knight at Berkeley, and just another terrific colleague who we've been working with for many years.

从大脑表面记录的这些脑信号中，我们确实能提取出极其精妙的细节。

So really terrific, terrific detail that we can extract from these brain signals recorded from the surface of the brain.

我认为最令人兴奋的或许是，我们不仅能解码、理解和观察大脑功能的这些不同方面，还能实际看到大脑各区域之间是如何相互沟通的。

What I found perhaps the most exciting was the combination of not only the fact that we can decode and understand and see all these different aspects of function in the brain, but we can actually see how the brain communicates itself from one area to another.

而且，如果你愿意的话，我们可以实时追踪单次试验中的脑信号。

And we can follow, if you so want, the brain signal in real time, in single trials.

举个例子，想象一个人看到闪光灯，他们被告知当看到闪光灯这种视觉刺激时，需要用右手按下按钮。

So for example, imagine a person sees a flashing light and they're being told when you see the flashing light, when you see this visual stimulus, I want you to push the button with the right hand.

根据基础神经生理学和解剖学知识，我们知道当你看到东西时，光线进入视网膜，通过视神经传递到后脑勺的枕叶皮层或视觉皮层区域，那里是大量视觉处理发生的地方。

Now we know from just basic neurophysiology and basic anatomy that when you see something, the light goes into the retina, goes through the optic nerve, goes back to the back of the head in an area called the occipital cortex or visual cortex where a lot of the visual processing is happening.

这些我们都知道。

So we know that.

我们还知道，当你移动手部时——比如按下按钮——大脑对侧（即与按键手相反的一侧）的特定区域，运动皮层，正是控制手部动作的区域。

We also know that when you move a hand, just like when you push a button, the area on the other side, so not the same side as the hand that pushes the button on the other side, the contralateral side, there's a certain area in the brain, the area motor cortex that makes the hand move.

所以当我看到光并通过按键作出反应时，视觉皮层必然先被激活，接着运动皮层也会被激活。

So somehow of course when I see a light and I respond by pushing a button, there has to be some activation in the visual cortex and then some activation in motor cortex.

但当然，过去没有人能真正看到这些区域是如何沟通的，因为我们缺乏能实现这一观察的成像技术。

But of course nobody could really see how these areas would communicate because we didn't have an imaging technique that would allow us to do that.

所以脑电图基本上不够精确，单神经元记录只能在单一位置进行。

So the EEG is basically not accurate enough, the single neuron recordings, they are only in one place.

是的，我们可以把它们放在这里或那里，或者前额，但我们无法覆盖整个大脑。

So yeah, we can put them back here or back here or up in the front, but we cannot cover the whole brain.

我们也可以做功能性磁共振成像，但那又是基于血流，需要很多秒才能形成图像。

We could also do fMRI, functional magnetic resonance imaging, but that again is based on blood flow and that takes many, many seconds to develop.

而这个过程相当快，就像你能理解的，看到光后可能在300毫秒左右就有反应，这就是反应时间。

So whereas this process is rather fast, it happens of course as you can appreciate, you may respond in three hundred milliseconds or something when you see a light, right, the reaction time.

所以功能性磁共振成像无法捕捉到这种信号。

So fMRI cannot capture that.

因此理论上，皮层脑电图是唯一能同时捕捉这些激活的空间细节和时间进程的技术。

So electrocorticography is really the only technique that can, in theory, capture both the spatial detail and the temporal progression of these activations.

我们开始研究这种大脑内部通讯，并开发了几种新算法，使我们能够以前所未有的细节和方式观察这一进程。

And we started to study this kind of brain internal communication and develop several different new algorithms that would allow us to see this progression in a detail and in a way that just wasn't possible before.

有很多惊人的例子，你可以直观地看到神经元如何在视觉皮层兴奋起来，就像点亮圣诞树一样，你能清晰地看到大脑是如何完成这一过程的。

And there's many striking examples where you can literally see how neurons get excited in visual cortex and then it's almost like you turn on the Christmas tree, the light goes, you know, and you can literally see how the brain does this.

信号先传递到运动皮层，按下按钮，然后随着按钮按压的感官反馈返回，又传递到感觉皮层。

It goes, motor cortex, button press, and then it goes to sensory cortex as you get the sensory feedback back from the button press.

你可以看到大脑中这一过程是如何发生的，而且能在单次试验中就观察到。

And you can see how this happens in the brain, and you can see it in individual trials.

你只需进行一次试验，就能追踪整个过程的动态。

You do that one time, and you can follow how this all happens.

这简直不可思议。

I mean, it's incredible.

这种能直观看到大脑运作方式的能力，实在令人着迷。

This ability to see how the brain does something is really, really so fascinating.

当时我们基本上是世界上首批以这种方式观察大脑此类活动的人。

And we were at a place where we were basically the first people in the world that had seen the brain do things like this in this way.

这确实非常令人振奋。

So that was really very exciting.

再次强调，当你完全学到新知识并开始理解和欣赏大脑的沟通方式时，就会自然而然地产生许多疑问。

And again, totally learning something new and you start to appreciate and understand how the brain communicates, then you can start to ask yourself a lot of questions.

那么，为什么当你执行相同的任务时，有时反应时间是200毫秒，有时却是400毫秒呢？

Well, why is it then that when you do the same task, that one time you respond in two hundred milliseconds and one time you respond in four hundred milliseconds?

是因为大脑信号有时这样传输——非常直接，而有时却那样传输吗？

Is it because the brain signal one time goes like this, they're very directly, and one time it goes like this?

不，事实并非如此。

No, it's not the case.

无论你反应快还是慢，激活的都是相同的脑网络通路。

They go the same way, the same brain networks when you respond quickly and when you respond slowly.

那这又是为什么呢？

Well, but then why is it?

因为动作电位传导速度——也就是大脑中神经元实际传递信号的速度——是非常接近的。

Because action potential conduction velocity, for example, that's the velocity of action potentials, actual neuronal communication in the brain, that's very, very similar.

我的意思是，这也许能解释不到1毫秒的差异，但绝不可能达到200毫秒的程度。

So I mean, that could account for variations of less than a millisecond maybe, but not like two hundred millisecond.

那么，这究竟是怎么回事呢？

Well, why is it then?

嗯，这个问题的答案可能需要另一期播客来讨论，但我们现在可以理解其中的原因了，举例来说。

Well, the answer would probably take another podcast to talk about, but we can see and we can now understand why that is, for example.

而这个答案实际上与许多不同现象密切相关。

And the answer is actually very closely related to many different phenomena.

例如，当你听到处于感知阈值的声音时——即你只有50%的时间能做出反应的声音，它非常微弱，有时你能听见，有时听不见。

Like, for example, when you hear a sound at perceptual threshold, like meaning a sound that you respond to only 50% of the time, it's so silent, so quiet that sometimes you can hear it, sometimes you can't.

根据定义，感知阈值是指你只有50%时间能做出反应的声音响度。

By definition, perceptual threshold is defined as the loudness of a sound when you respond only 50% of the time.

那么，为什么你50%的时间会做出反应，而另外50%的时间不会呢？

Well, why is it then that 50% of the time you respond and 50% of the time you don't?

其实这个答案与之前的问题密切相关。

Well, the answer is actually very closely related.

同样，这需要很长时间来解释，但它与大脑反应时间时快时慢的原因密切相关。

Again, it would take a long time to explain, but it's very closely related to why it is that the reaction time in the brain, so to speak, sometimes is fast and sometimes is slow.

所以你会发现许多令人兴奋的新现象，这完全是一个全新的研究领域，对人类大脑理解的新领域，绝对是一个激动人心的时代。

So there's really exciting things that you can start seeing and really totally a new area of research, a new area of understanding of the human brain, just an absolutely exciting, exciting, exciting time.

是的，完全同意。

Yes, absolutely.

这让我想起多年前我刚到佛罗里达医院工作时组织的那场会议，当时正筹建脑机接口与功能映射项目。

And this reminded me about the conference that I organized many years ago when I just came to work at Florida Hospital and to establish Brain Computer Interface and Functional Mapping program.

您作为特邀嘉宾出席，我至今仍对此心怀感激。您开场引用了一句话，谈到大脑活动就像一门我们需要学习理解的语言。

You came as our guest speaker, so I'm still grateful to you for doing that, and you started your lecture giving a quote and talking about the activity of the brain as a language that we need to learn to understand.

那么您通过ECOG技术开展的所有工作，如何帮助您更好地理解大脑的语言？

So all this work that you did with ECOG, how did it help you to understand the language of the brain better?

在开始使用ECOG之前，有哪些认知是您无法获得的？

What did you learn that wasn't possible before you started working with ECOC?

这又是一个非常棒的问题。

This is another really great question.

正如许多人常说的那句话：神经科学是数据丰富但理论贫乏的领域。

It goes to the saying that many people have said and it's very common, they said, Neuroscience is data rich but theory poor.

意思是虽然收集了大量数据，但在解释大脑运作机制时，现有理论只能在非常狭窄的领域内成立。

And what they mean is while you collect a lot of data and then when you are trying to explain how the brain does something, you have a lot of theories, but they really only work in very narrow domains.

例如，我们可以开始理解决策任务。

So for example, we can start to understand decision making tasks, for example.

比如，如果我需要选择用这只手或那只手按下按钮，在需要决定用哪只手的任务中，就有关于大脑如何处理这种特定任务的理论。

Well, if I have a choice between pushing the button with this hand or with this hand and there's some kind of task where I have to decide whether it's one or the other, then there's some kind of theory about how the brain does this particular task.

你可以让任务变得更复杂，然后我们就能理解大脑会发生什么变化以及发生在哪个区域。

You can make the task more difficult and then we would have an understanding of what would happen in the brain and where it would happen in the brain.

但这些理论都非常局限，而且这类小型理论有很多。

But it's very, very narrow and there's many of these small theories.

但我们缺乏的是关于大脑整体运作方式的可靠观点。

But what we didn't have is many good ideas about how does the brain overall work?

大脑是如何协调我们的行动的？

How does the brain coordinate our actions?

当然，我并不是说我现在已经完全理解了，即使是我认为已经充分理解的内容，要完全发展它们可能也需要我毕生的时间。

And of course, I'm not claiming that I have a complete understanding now, even the things that I think I now understand fully developing them would basically take my whole life.

但我确实相信，通过研究这些ECOG信号，我开始对大脑的运作方式有了更深刻的认识。

But I do believe that by studying these ECOG signals, I started to get a much, much deeper appreciation of how the brain operates.

这实际上是我们目前正在收尾的一项研究。

And so this is actually a study that we're finalizing right now.

这项研究已开发多年，我们现在称之为‘皮层模拟器’的研究项目正在完善中。

It's been in development for about many years, but a study that we're developing now and finalizing now that we call the cortical simulator.

基于我提到的许多其他研究结果，我们开始总结出大脑皮层运作的简单规则，并用这些规则开发了一个计算机程序，用于模拟感觉皮层和运动皮层，或不同的感觉皮层，进而模拟各种任务。

So we actually started to, as sort of the results from a lot of the other studies I talked about, develop simple rules of how the cortex operates and then use these rules to make a computer program that would simulate sensory cortex and motor cortex or different sensory cortices and then simulate different tasks.

如果你在视觉刺激下需要按键反应会发生什么？

Well what happens if you have visual stimulation and you respond with a button press?

现在你在感知阈值下接受伴有响亮刺激的视觉刺激。

Well now you have visual stimulation with a loud stimulus at perceptual threshold.

那么反应时间会有什么变化？

Well then what happens to reaction time?

行为会有什么变化？

What happens to behavior?

如果给予多感官刺激会发生什么？

What happens if you give multisensory stimulation?

你同时给予听觉和视觉刺激，或者在不同时间给予，诸如此类。

You give auditory and visual stimulation at the same time, at different times, and so forth.

我们能够用非常简单的规则，解释之前许多其他研究中展示的各种行为。

And we were able to, with very simple rules, be able to explain all different kinds of behaviors that have been shown in many other studies before.

这让我明白，当然，我确信没有哪个理论、没有哪个简单的理论能永远正确且完美。

So what that tells me is that of course, I'm sure there's no theory, no simple theory will always be right and will be perfect.

这实际上是首个开始让我们对大脑功能的普遍原则有更全面理解的概念之一。

This is really one of the first concepts that actually starts to get towards having a more holistic understanding of general principles of brain function.

这也是我一直感到非常非常兴奋的事情。

That's really also something I've been very, very excited about.

我认为从长远来看，这类概念将产生重大影响。

And I think in the long term those kind of concepts will have a lot of impact.

绝对如此，我们期待这项新研究发表并阅读它。

Absolutely, and we are looking forward to this new study to be published and to read it.

非常感谢。

Thank you so much.

你当然没有止步于此，我想你从未停止前进，但我正要谈到下一步。

You didn't stop there, of course, I don't think you ever stopped, but I'm just getting to the next step.

你还开发了一个自适应神经技术项目，该项目也在教导他人如何进入神经技术领域，提供了非常重要的信息和实践培训，我也参加了，非常高兴能参与其中。

You also developed an adaptive neurotechnology program, which also was teaching other people how to get into the field of neurotechnologies, was giving very important information and practical training, which I also attended, I'm so happy I did.

你能详细说说什么是自适应神经技术吗？

Can you tell more what is adaptive neurotechnology?

我认为这是一个非常独特的项目，可能是世界上首个此类项目。

And I think it's such a unique program, probably the the first in the world program of this kind.

所以这就是为什么我非常希望你至少能简单谈谈它。

So that's why I really would like you to speak at least a little bit about it.

是的。

Yeah.

我们很幸运地从美国国立卫生研究院为我们在奥尔巴尼的研究所争取到了一笔大额资助。

We were fortunate enough to attract a big grant from the NIH to our institute, in Albany.

这是一个为期五年的大型中心资助项目，可以续期。

This is a big center grant that's five years, can be renewed.

目前已经续签，这是第二次续签了。

It's now has been renewed, now for the second time.

该中心的主要组成部分或目标之一是开发自适应神经技术、脑机接口及相关领域的新技术，并加以应用。

One of the major components or the purpose of the center is to develop new technologies in the area of adaptive neurotechnologies, brain computer interfacing, and related areas, and also apply them.

其中很重要的一部分是开设培训课程，向处于职业特定阶段的人员教授所有必要的工程学、生理学、医学和神经学概念。

And a big component of all of this is a training course that would teach all the necessary engineering, physiology, medical, neurological concepts to people that are at a certain stage in their careers.

正如你所指出的，没有人能一次性掌握所有这些概念。

Because as you pointed out, I mean, nobody learns about all these concepts at once.

我的意思是，有些人成为工程师，有些人成为医生，有些人成为心理学家。

I mean, some people are engineers, some people are medical doctors, some people are psychologists.

虽然现在确实没有统一的教育项目——尽管目前各大学开设的神经技术专业课程在逐渐增加，质量参差不齐——但要全面掌握所有这些不同领域的专业知识和综合技能仍然相对罕见。

While there really isn't any educational program, there's some now an increasing number, some are very good and some are not so good neurotechnology sort of specializations at certain universities, but it's still relatively rare that you get this comprehensive expertise and comprehensive knowledge in all these different areas.

因此，基于这个原因，同时也因为中心拨款要求我们必须开设教授这些概念的课程。

So for both that reason and also because it was mandated by this center grant that we had to form a course that would basically teach these concepts.

这就是我们所做的。

And that's what we did.

在你参加的课程中，我们共有24名来自美国各地的学员，他们需要满足多项标准：必须处于职业发展的特定阶段，工作领域需确保他们学以致用的可能性，而非仅因兴趣参与后便搁置不用。

So in the course that you attended, we had, I believe, 24 participants from many places in The United States, and they had to satisfy quite a few criteria so that they had to be at a certain stage in their career, they had to work in an area where they can actually then there's a reasonable likelihood that they would actually then use the things that they are learning in some way and not just go because they're interested and then they leave.

这就是我们所做的。

So that's what we did.

我们进行了大量的内部培训。

There was a lot of in house training.

我们邀请了来自全国不同地区、不同大学、不同学科领域的教师，他们会在现场进行为期三周的教学。

We had faculty from other areas in different disciplines, from different universities and different places in the country, and they would teach on-site for three weeks.

正如你指出的，这确实是世界上首批（如果不是首个）此类培训课程之一。

As you pointed out, this is really one of the first, if not the first training course of its kind in the world.

我相信它取得了巨大成功，并且仍在持续进行中。

And I believe it's been a great success, and it's still ongoing.

是的，完全同意。

Yes, absolutely.

我深受启发，因此根据所学内容开设了一门课程，当然还融入了我的全部专业知识，在维尔纽斯大学为研究生神经生物学学生教授类似的课程。

I was so inspired by it that I offered a course based on what I learned, and of course adding all my expertise, the course like this at Vilnius University, where I'm still teaching graduate neurobiology students.

因此我们正在扩展课程内容，原本只有理论部分，现在又增加了实践环节。

And so we're expanding because we had just theoretical part and now we added also a practical component.

我们购置了一些设备，学生们已经开始动手做实验并分析数据，这门课程对我个人以及一代代学生都产生了巨大影响。

So we've got some equipment and our students are already doing hands on experiments and analyzing the data, so that course really created a huge impact on me and now on generations of students.

所以非常感谢。

So thank you.

再见。

Bye.

我们都很感激。

We all appreciate it.

好的。

Okay.

那很好。

That's good.

谢谢你详细解释这些。

So thank you for for explaining that.

然后就是你现在正在进行的这个语音识别新项目。

And then there is this new project that you are working on with the voice recognition.

所以识别，语音的意图，它有多种语音调制方式，但我更想了解预测方面的内容。

So recognition, the intent of the voice, he has different modulations of the voice, but I would like to learn more and predictions.

你是怎么开始这个项目的？能多讲讲吗？

So how did you come to this project and can you tell more about it?

好的。

Yeah.

这其实源于我们之前对ECOG信号的大量研究，开始理解大脑信号中的语言和情感。

This really started with a lot of the research we have done with the ECOG signals starting to understand language and emotions in brain signals.

我们发现，仅通过监听大脑活动就能获取大量语言细节和人类情感信息，这当然非常鼓舞人心。

So we learned that it's possible to learn a lot of details about language and a lot of details about human emotion just by listening to brain And that's of course that's very inspiring.

当然，其中涉及的基础神经科学非常有趣，但目前还没有实际应用。

Of course there's a lot of basic neuroscience that is very interesting, but of course this has no application.

我的意思是，如果你想了解这些，总不能直接给人植入电极吧。

I mean if you want to know something about them, you cannot just implant them with some electrons.

这种情况当然只会在非常有限的条件下发生。

This really happens only of course in very limited circumstances.

所以我一直在思考，如何能将这种理解转化为实际有用的东西？

So I've been thinking for a long time, well how could I actually take some of that understanding and translate it into something that would actually be useful?

我意识到，声音的语调在许多方面与脑电信号并没有太大不同。

And I realized that the tone of the voice is actually in many ways not too different from brain signals.

事实上，它看起来几乎就像脑电图或皮层脑电图信号。

In fact, it actually sort of looks almost like EEG or ECOG signals.

而且分析方式也非常相似。

And it also the ways in which it's analyzed is very similar.

我了解到，虽然声音的许多方面已被研究很久，比如你所说的词语。

I understood that while many aspects of voice have been studied for a long time, like the words that you're speaking, for example, they've been studied for a long time.

这是一个被称为自然语言处理或NLP的完整领域。

It's a whole field called natural language processing or NLP.

这个领域已经存在很长时间了。

It's been around for a long, long time.

它真正起源于计算机科学领域，大概在——我不确定具体时间——至少是70或80年代左右，并逐渐形成了一种认知：你确实可以分析人们说话句子中的词汇，从中提取有用信息。

And it really started in computer science in probably the, I don't know, at least the 70s or 80s or something, and producing understanding that you can in fact analyze the words that are in sentences that people speak and can derive useful information from this.

这种技术从计算机科学领域逐渐扩展到其他领域，比如金融市场，并最终催生出了能提供实用信息的产品。

And that would go from computer science to then different type of domains like the financial markets, for example, and would then lead to products that would provide useful information.

举个著名的例子：大约十年前开发的一套词典在金融市场非常知名，叫做Lachron McDonald词典。该词典指出，如果企业高管使用某些特定词汇，在金融市场语境下往往具有积极含义。

So for example, a dictionary that was developed about ten years ago, it's very famous in the field of financial markets, called the Lachron McDonald Dictionary, and that dictionary says that well, if an executive says these words, that actually indicates or has some positive meaning in the context of financial markets.

而如果他们使用另一些词汇，则意味着消极信号。

If they say these words, well then it means that it's negative.

当公司发生负面事件时，高管们往往会使用这类词汇。

So if there's something bad happening about the company, then they tend to use these words.

当出现正面事件时，他们则倾向于使用另一类词汇。

If something good happens, they tend to use these words.

这自然是投资者和其他利益相关者都渴望掌握的信息。

So that's of course something that investors are interested in and other stakeholders are interested in learning.

我们虽然理解这种规律，但一直缺乏系统性研究的，其实是情感交流中更为核心的要素——那就是声音的语调。

And we understood that that's the case, but what hasn't been available for systematic study is something that is actually maybe even more integral to emotional communication, that is the tone of the voice.

这可以追溯到社会心理学文献和神经科学领域。

And that goes back to both social psychology literature and it goes back to neuroscience.

神经科学方面的解释是，我们所说的话语主要由大脑左半球产生，至少对大多数人来说是这样。

Like the neuroscience aspect is the words that we're speaking, they're produced predominantly by the left hemisphere in the brain, at least in most people.

所以如果大脑左侧特定区域中风，你就无法再说话了。

So if you have a stroke on the left side in certain areas in the brain, you can no longer speak.

然而，声音的语调——比如你说的韵律，你知道的，像是你的音调变化、声音的抑扬顿挫——这些主要由大脑右半球控制。

However, the tone of the voice, you would call that prosody, for example, that, you know, like how your inflection, how you modulate the tone of your voice, That is governed mostly by the right side in your brain.

这就是为什么如果你的右脑中风，你仍然可以说话，但可能听起来像个机器人。

So that's why if you have a stroke on the right side of your brain, you can still speak, but you may sound like a robot.

所以我们从最初就明白，即使在大脑内部，你如何说话和你说什么也是完全分离的。

So we understand that from the very, very beginning, even inside the brain, how you speak the words and what you say is completely separated.

然后我们从社会心理学文献中得知，确实有篇著名论文指出，信息中约40%的内容蕴含在语音语调中。

We then know from the social psychology literature that indeed there's famous paper saying that about 40% of the information in the message is in the tone of the voice.

当然人们对这40%的比例和具体含义存在争议，但基本概念是：说话内容与表达方式并非同一回事。

And of course people debate about the 40% and what it is, but the basic concept is what you say and how you say is not the same.

从基本的社交互动中

From basic social interactions.

我们知道，比如丈夫忘记倒垃圾，妻子看着他生气，他说：'嘿，怎么了？'

We know that if, you know, like the man forgets to take out the trash and the wife looks at him and is mad and he says, Hey, what's wrong?

而她说：'没什么'

And she says, Nothing.

'我很好'

I am fine.

但你心里明白情况并不好

Well, you know it's not fine.

对吧？

Okay?

知道情况并不好

Know it's not fine.

话语是积极的，但语气却不是

The words are positive, the tone is not.

明白吗？

Okay?

我和同事肖恩·奥斯汀开始意识到，我们的沟通中有一个极其重要的部分，目前我们作为人类可以某种程度上理解，但无法系统性地掌握。

My colleague, Sean Austin, and I, we started to understand that there's a completely important part of our communication that currently we can sort of as humans understand, but cannot understand on a systematic basis.

这实际上构成了这个想法的基础：虽然我们可能没有使用脑信号，但我们运用声音信号，采用我们现在从神经科学中了解到的同类技术，并运用这些技术来理解声音的语调。

And that's really the basis for this idea that while maybe we are not using brain signals, but we're using voice signals to understand with the same kind of techniques that we now know from neuroscience, and with using those kind of techniques to understand the tone of the voice.

正是那时我们基本上启动了一个名为'赫利俄斯生命企业'的新项目和事业，其目标是将这种新信息——或者说我们人类已经理解的信息——系统化地呈现出来，并在某种程度上教会计算机更多关于人类情感的知识。

That's when we basically started a new project and a new venture that's called Helios Life Enterprises that sets its goal to make this new information or this information that we understand as humans anyway, but make that available systematically and in a way teach computers a little more about human emotions.

我们正在金融市场的背景下进行这项工作。

And we're doing that in the context of financial markets.

有多种方式可以应用这项技术，例如，它可以应用于医疗领域进行疾病诊断，比如帕金森病或其他类型的疾病。

There's different ways that one could apply this, For example, it could be applied in the medical domain for diagnosis of disease, Parkinson's, for example, or other types of disorders.

它也可以应用于呼叫中心，例如当人们感到不满时，让机器能够理解人们的情绪。

It could be applied in call centers, for example, when people get upset and then have machines understand that people get upset.

出于多种原因，我们决定将其应用于金融市场的概念中，投资者可能对了解一位高管是否言行不一感兴趣——他说了很多积极的话，但实际上听起来相当消极。

We decided for different reasons to apply it in the concept of financial markets where, again, investors may be interested in understanding whether an executive well, he says a lot of positive things, but actually he sounds pretty negative.

嗯，这可能意味着些什么。

Well, that probably means something.

实际上，在这个背景下，我认为有趣的是，无论是面部表情、肢体语言还是你使用的词语，人类很多时候都会主动利用这些来引导沟通，甚至可以说是操控他人。

Actually, this in this context, I think it's just interesting to know that both facial expressions, body language, and the words that you're using are actually used by humans a lot of times actively to steer the communication or you could even say to manipulate others.

比如，扑克脸，对吧？

So, you know, poker face, for example, right?

你摆出一副扑克脸，可能是在虚张声势，也可能不是，诸如此类。

You show a poker face, maybe you're bluffing, maybe you're not or something.

人类总是利用这些渠道来操控他人。

Humans use these channels all the time to manipulate others.

但你确实无法用声音的语调做到这一点。

Well, you really can't do that with the tone of the voice.

如果你刚遭遇了可怕的事情，你不可能听起来兴高采烈。

If you just had something horrible happen, you cannot sound exuberant.

这非常难以做到。

That's very difficult to do.

当你对某件事非常兴奋时，你无法假装出沮丧的声音。

Or you can't sound depressed when you're actually very excited about something.

这真的很难做到。

It's really hard to do.

或许可以通过训练达到，但即便能做到，也会非常困难。

Maybe you can be trained to do that, but if you can, it will be very difficult.

因此，如果我们想更深入了解人们及其情感状态，语音语调确实是一个重要的切入点。

So it's really something that if we want to learn more about people and more about their emotional context, the tone of the voice is a place to go.

这就是多年工程工作和神经科学研究积累的巅峰成果——虽然这些研究本身令人兴奋，但往往难以转化为改善人们生活的实际应用，当然我们之前讨论的功能映射是个例外。

So that's really the culmination, a lot of the engineering work and a lot of the neuroscience I've done over the years that, of course, is very exciting, often cannot lend itself to applications that make people's lives better, except of course we just talked about this functional mapping before.

但即便如此，我的意思是，尽管这项研究令人振奋，它也只适用于神经外科领域非常狭窄的范围。

But even that, I mean, that's in a lot of ways, as exciting as it is, it's useful for neurosurgery in a very, very narrow domain.

对吧？

Right?

它只适用于这个非常特定的应用场景。

It's useful for this very specific application.

这项能力——如果我们能教会计算机更好地理解人类，将能在众多不同领域得到广泛应用。正因如此，我们感到非常振奋，这已成为我多年来参与并推动的一个重要项目。

This capability, if we can teach computers how to understand humans better, that could have a lot of applications in really many, many different areas For all these reasons, that that got us very excited, so that's now an important project that I'm have been involved with and have been driving for quite a few years now.

是的。

Yes.

这确实非常令人兴奋，看到这种发展进程很有趣——通过掌握信号分析技术，你实现了对脑信号的惊人解析。

That that is very exciting, and it's very interesting to see that progression, that knowing how to analyze the signal, you were able to really do amazing analysis of the brain signal.

而你积累的所有脑信号分析经验，现在又被用于处理从人们那里收集的声音数据，这种知识体系的相互关联与促进真是妙不可言。

And then all that experience that you accumulated analyzing the brain signal, now you are using already for processing of the sound, yes, of the sound that you collect from people, so how everything is really connected and helping each other.

太美妙了。

That's beautiful.

关于这个项目，我在观看你的YouTube视频后产生了一个小疑问。

In regards to this project, just a quick question which I became very curious about after watching some of your YouTube videos.

你在某个视频中提到过一份名单，上面列出了人们声音中可识别语调变化的程度排序。

In one of the videos you mentioned that you have a list of people, at least people listed in terms of the amount of that recognizable inflection in their voice.

如果我没理解错的话，你说排名第一的是埃隆·马斯克。

So if I understood correctly, and number one, you said that it was Elon Musk.

实际上，根据他的声音变化和你获取的数据，你做出了一个非常有趣的预测，结果证明是正确的。

And actually based on on his voice inflection and the data that you acquired, you made a very interesting prediction which turned to be true.

你能稍微谈谈这个吗？

Can you a little bit maybe talk about this?

是的。

Yeah.

那确实太棒了。

That was really terrific.

正如你指出的，通过理解声音的语调，我们现在可以开始理解声音的语调与公司未来发生的事情之间的关系。

So the as you pointed out, by understanding the tone of the voice, we can now start to understand how the tone of the voice is related to things that happen to the firm in the future.

比如，你可以用非常简单的术语来理解这一点。也许高管被问及阿拉巴马州刚发现的有毒废物倾倒场，他们会说‘哦，他们知道这是个重大问题，因为他们刚和律师谈过这事’。

Like, for example, and you can understand that in very simple terms, So maybe the executive is being asked about the toxic waste dump in Alabama that was just found, and they say, Oh, is they know it's a big problem because they've just talked to their attorneys about it, so they know it's a big problem.

但当然，他们被指导要给出非常政治正确的回答，对吧？

But of course they are being coached that they will, of course, give a very politically correct answer, right?

所以他们会说‘哦，没问题’之类的话，但他们的声音却暴露了问题的存在。

So they will say, Oh, it's no problem or something, but their voice gives away that there is an issue.

如果这种情况发生并且你意识到了，你可能会知道，也许一个季度后的下一次财报电话会议上，有毒废物倾倒问题现在被公认为是一个大问题，股价会因此变动或者公司业绩受到某种负面影响。

Well, if it does that and you know that, you would know that maybe a quarter later at the next quarterly earnings call or something, maybe the toxic waste dump is now known to be a big problem and now the stock price changes or the company's performance is affected in some negative way.

所以关键在于，在金融市场的背景下，声音的语调有助于了解公司未来可能的表现或发展趋势。

So the bottom line is the tone of the voice is useful in the context of financial markets to learn something about how the firm is going to do or maybe doing in the future.

当然，这并不是一个水晶球，它不会总是准确的。但最终，有些高管是毫无保留的，比如埃隆·马斯克就是一个完全透明的人。

And of course, it's not a crystal ball, it won't be correct all the time, But the end, of course, there will be executives that are an open book, and Elon Musk is a complete open book.

如果他生气了，听起来就是生气的样子。

If he is pissed, he will sound pissed.

如果他开心，听起来就会很开心。

If he is happy, he will sound happy.

你跟蒂姆·库克交谈时，他的语气总是如出一辙。

Well, you talk to Tim Cook, Tim Cook always sounds the same.

所以他声音语调中蕴含的信息量远不如埃隆·马斯克丰富。

So the the amount of information in his tone of the voice is much less in Elon Musk's voice.

我们内部——虽然从未公开或发表过——但确实有一份全国高管榜单，按他们声音语调中泄露的信息量排名，正如你所说，埃隆是第一名。

So we internally and we've never publicized this or published it or anything, but we had an internal listing of the, as you point out, of the top executives in the country ranked by how much information they give away in the tone of their voice, and Elon was number one.

事情是这样的，这其实发生得相当早，那是几年前在伦敦的一个会议上，我们基于对埃隆声音的分析做出了一个非常大胆的预测。

So what happened was this was actually relatively early on, this is several years ago, there was a conference in London, and we made a very bold prediction based on analysis of Elon's voice.

我们在他的声音中发现了某些特征，这些特征基本上是非常积极的信号指示器，用简单的话来说。

And we found particular characteristics in his voice that were very indicative of basically very positive signal, so want in in simplified terms.

于是我们在伦敦的舞台上做了一个可变预测，一个前瞻性预测：特斯拉股价将突破历史新高。

So we made a variable prediction on stage in London, a forward looking prediction that Tesla is going to break its all time high.

现在我不记得当时特斯拉的具体股价了，可能是200美元左右。

Now I don't remember exactly what Tesla was at this moment, maybe 200 or something.

所以这个预测非常大胆，而且我们预测这将在下一个季度内实现。

So that was very bold, you know, and we said it would happen within the next quarter.

几天后，埃隆在我们分析完他们的季度财报电话会议后，发布了季度财报电话会议。

So a few days later, Elon introduced after this earnings call, their quarterly earnings call that we analyzed, we then had the quarterly earnings call.

之后我们在伦敦召开了会议并做出这个预测。

We then had the conference in London where we made this prediction.

几天后，埃隆发布了Cybertruck，那是一场灾难性的发布会。

A few days later, Elon introduces the Cybertruck, and that was a disastrous presentation.

股价下跌了。

The stock price went down.

我们简直要抓狂了，因为这太尴尬了。

We are ripping our hair out because we're like, this is so embarrassing.

我们完全错了。

We're so completely wrong.

然后股价开始飙升，很快就涨到了800美元左右。

And then the stock price was just like skyrocketing, and it went to like $800 or something very quickly.

所以我们的预测是正确的。

So our prediction was correct.

这非常大胆。

It was very bold.

预测非常精准。

It was spot on.

当然，这是在数百人面前做出的预测。

And of course, this was in front of hundreds of people.

所以这并不是事后诸葛亮之类的情况。

So this was not like after the fact or something.

现场有数百人见证。

This was hundreds of people.

那确实相当令人兴奋。

So that was quite exciting.

基于这个故事，几个月后另一次会议上，我在纽约另一个舞台上现场讲述了这个故事，随后我们参加了约30个项目的路演比赛，结果我们以绝对优势赢得了第一名。

And then actually based on that story, there was another conference just a few months later and I told that story live on stage at a different conference in New York, and then we had a competition of about 30 different pitches if you so want, and we won with by far, we were number one out of these 30.

这个特斯拉的故事非常激动人心，整个故事都基于埃隆·马斯克的语气基调。

So that was very exciting with this Tesla story, very exciting story that was based on the tone of the voice of Elon Musk.

确实。

It is.

非常有意思。

It is very interesting.

我...我想我看过那些演讲视频。

I I think this is what I watched, those speeches.

你需要在三分钟内展示你的项目。

It was like three minutes during which you need to present your project.

是的。

Yes.

所以你成功了。

So you it was successful.

非常好。

Very good.

非常好。

Very good.

太棒了。

Excellent.

这让我想到你现在在上海的工作。

So it brings me now to your current work in Shanghai.

是的。

Yes.

一步一步来，我相信中间还有很多很多其他项目，但至少我们回顾了主要的几个。

So step by step, I'm sure there were many, many other projects in between, but, you know, at least we went through the main ones.

那么你现在在上海做什么工作呢？

And now what do you do in Shanghai?

能谈谈你的工作以及未来的规划吗？

Can you tell about your work and also, you know, your future steps?

你计划从那里往哪个方向发展？

Where are you planning to go from there?

我刚才在商业项目部分提到，我们真正想做的是能改变人们生活的创新。

I just talked about for this commercial venture about this desire to actually do something that really changes people's lives.

所以我们正在通过之前描述的语音语调分析技术，在商业领域实现这个目标。

So we're doing that in this commercial domain with this voice tone analysis I just described.

另一种实现方式是：将这些已在动物或人类身上经过验证、成功应用数百次的神经技术，转化为能改善大众生活的解决方案。

Another way to do that would actually be to take all these neurotechnologies that have been developed and tested either in animals or in humans and have been really applied and shown success with in hundreds and hundreds of demonstrations, take them and make them into solutions that improve the lives of many people, improve the lives of mass populations.

这种动力其实源于一种深刻的矛盾——既对这些神经技术的前景感到振奋，又对现状感到沮丧。

That's really motivated by a very deep frustration in a way that comes from the exciting promise on one hand that these neurotechnologies have.

虽然我们现在可以与大脑交互，能够理解大脑如何运作，并利用它来诊断疾病、治疗障碍，我们可以做到所有这些不同的事情。

While we can now interface with the brain, we can understand how the brain works, we could use it to diagnose disease, how to treat disorders, we can do all these different things.

而且我们获得了许多生物学上的洞见。

And we have a lot of biological insights.

我们对大脑有了更深入的理解。

We understand the brain better.

我刚才已经提到了一些。

I just talked about some before.

我们取得了许多技术突破。

We have a lot of technical breakthroughs.

我提到过ECOG脑电图记录技术。

I talked about ECOG recordings.

有新算法、新设备，有许多有趣的技术突破，这些加深的理解和突破都促成了技术演示。

There's new algorithms, there's new devices, there's a lot of interesting technical breakthroughs, and both the improved understanding and the breakthroughs, they lead to technical demonstrations.

比如现在就有关于如何治疗失语症患者的技术演示。

Well, there's technical demonstrations how we can now treat people that have aphasia.

我们可以通过大脑信号解码人们的笔迹。

We can decode people's handwriting from their brain signals.

我们能够通过记录大脑信号来恢复语言功能。

We can restore speech by recording brain signals.

我们可以实现所有这些激动人心的突破，但归根结底，目前仍缺乏真正改变患者生活的解决方案。

We can do all these exciting things except, of course, at the end of the day, we don't have any solutions that actually are changing people's lives.

这确实是个普遍存在的遗憾。

So that's a very general frustration.

尽管我们拥有诸多技术展示，但真正能改变生活的解决方案依然空缺。

And really this issue that even though we have all these demonstrations, we really don't have solutions that change people's lives.

其实这个困境早在2002年就被领域内所认知——当时我们团队在纽约州北部举办了第二届国际脑机接口会议。

That's something the field has actually kind of been aware of since 2002, when I think it was the second, right, the second international BCI meeting that was held by our group in 2002 in Upstate New York.

那次会议的名称恰如其分地定为《超越技术展示的脑机接口》。

And the name was aptly named Brain Computer Interfaces Moving Beyond Demonstrations.

即便在当时，我们的目标就很明确：要推动技术落地实现实际应用。

So even back then the idea was, well, we want to take this out and actually want to do something.

如今二十年过去了，我们在转化应用方面进展甚微。

Well now we're twenty years later and we haven't done a lot of translation.

这在很多方面都令人非常沮丧。

So that's very frustrating in many ways.

虽然我之前提到过功能映射方面的一个成功案例，但总体而言，这个领域确实未能将诸多成功案例和演示转化为能改变大众生活的解决方案。

And I told you about a success before with dysfunctional mapping, but in general the field has really not been able to take a lot of the successes and the demonstrations and make it into solutions that change the lives of many people.

我们没有足够时间深入探讨具体原因，但我们在前沿应用神经技术实验室（天桥和陈前沿研究所的研发部门）正致力于——这个由天桥和陈氏夫妇创立的慈善机构专注于支持脑科学研究与发展。

We don't have enough time to go into the details for why that all is, but what we're trying to do at the Chien Frontier Lab for Applied Neurotechnologies, is the really R and D component of the Tianqiao and Chrissy Chien Institute, which is a philanthropic organization dedicated to supporting brain research and development.

我们正在努力开发能够实现这一跨越的技术。

And what we're trying to do is to develop technologies that are going to make that jump.

我们仔细分析了为何先前的演示未能成功转化，以及我们需要采取哪些措施来实现突破。

And there's a careful analysis of why it is that previous demonstrations haven't made it and what do we have to do to actually make it.

事实上，我与几位合著者正在撰写一篇评论文章，即将投稿至《自然·生物工程评论》，该文将详细探讨这个问题及其潜在解决方案。

There's actually a review article that I'm now working on with several co authors that very soon will go to Nature Reviews Bioengineering that actually talks about this problem in potential solutions.

我们正尝试将这些解决方案融入前沿实验室的工作中——该实验室的使命正是通过神经技术改善人类生活。

And the solutions we're trying to incorporate or we are now incorporating into the work that we're doing at the Frontier Lab whose mission it is to improve people's lives with neurotechnologies.

作为这项工作的一部分，我们正在开发一套全新的多模态脑电图记录系统。

And as part of this, we are developing a completely new multimodal EEG recording system.

它被称为智能脑电图设备。

It's called Smart EEG.

你可以将它佩戴在前额。

You can put it on your forehead.

它具备双通道脑电信号采集功能。

It has two channels of EEG.

配有运动传感器。

It has a motion sensor.

内置麦克风。

It has a microphone.

支持血氧饱和度测量（PPG和SpO2技术），可监测心率。

It has SpO2 measurements or PPG and SpO2 so it can measure heart rate.

还能检测血氧水平。

It can measure oxygenation.

因此它能实时完成所有这些功能。

So it can do all that in real time.

它与我们的BCI 2000软件平台完全集成。

It's fully integrated with our BCI 2,000 software platform.

所以这本质上是一个非常强大的软硬件平台，极具实用性。

So it's essentially a very powerful hardware software platform that's extremely practical.

这也成为了巨大挫折的根源——因为即使是人们描述的那些转化神经技术，包括研究中描述的许多技术，实际上很多都非常成功。

And that's been a source of a tremendous frustration because a lot of the even the translational neurotechnologies that people have described and are described in studies, many of them are actually very successful.

非侵入式的技术中有很多都非常成功。

The non invasive ones, but there's many non invasive ones that are very successful.

我提到过失语症训练，同时你还能诊断抑郁症患者。

I spoke about aphasia training while you can diagnose people that have depression.

这样的例子有很多。

There's many examples of that.

你可以对患有轻度认知障碍或痴呆症的人进行认知评估。

You can have cognitive evaluations of people that either have mild cognitive impairment or dementia.

你可以区分它们。

You can tell them apart.

你可以将正常人、轻度认知障碍患者和痴呆患者区分开来。

You can tell the normal people apart from the people with mild cognitive impairment and the people with dementia.

如果我们能拥有这种能力而不必去诊所，那不是很好吗？

Well, wouldn't that be great if we could just have that capability and you didn't have to go to a clinic?

这个领域的许多研究使用64通道脑电图，需要博士后参与，博士后必须设置研究、收集数据、进行大量数据分析并非常谨慎地处理，然后可能才能展示一些成果。

A lot of the studies in this area, they would use sixty four channel EEG and they would use a postdoc, and the postdoc has to set up the study and then collect the data and do a lot of data analysis and be very careful with data analysis, and then maybe you can show something.

你能展示或人们能展示这些固然很好，但这并不能使其成为一个真正可应用的实用系统。

Well, that's great that you can show that or that people can show that, but what it doesn't do, it doesn't make it a practical system that actually can be applied.

而这正是我们现在正在努力的方向。

And this is now what we're working on.

我刚才提到的硬件基本上只是一种硬件接口，如果你愿意这么称呼的话。

And the hardware I just talked about is basically just the sort of the hardware interface, if you so want.

它基本上是一个柔性贴片，可以像这样弯曲，大概这么大。

It's basically a flexible patch, so it actually can flex like this, it's about this big.

你将标准电极嵌入在背面的粘性贴片上

And you put standard electrodes on the back that are embedded in sticky patch.

整个贴片都是粘性的，内部嵌有电极，这些电极与全球通用的心电图电极类似

So the patch, the whole patch is sticky, it has electrodes embedded in them, they are similar to EKG electrodes as they are used everywhere in the world.

只需将设备夹好，贴在额头上，然后运行某种计算机程序和与之交互的系统，执行特定协议

You clip them in and you simply take the device, you put it on the forehead, and then you can run some kind of computer program and some kind of system that interacts with it and executes a certain protocol.

因此它是实时闭环能力的多模态数据收集与强大协议相结合的产物，这种能力在全球范围内都是独一无二的

So it's in each combination of both the multimodal data collection of the real time closed loop ability and powerful protocols associated with it, it's really a capability that's completely unique worldwide.

当然，我们也有用于研究用途的脑电图头盔

So we have EEG helmets, of course, for research purposes.

现在我们有了面向消费者的脑电图设备，但它们可能仍难以佩戴、价格昂贵或仅能检测脑电波信号

Now we have EEG devices that are for consumer use, but they still are maybe hard to put on or maybe they are expensive or maybe they only do EEG and not other signals.

所以这些设备都存在各种局限性

So they have all kinds of limitations.

我们的设备是多模态的，自主生产且成本极低，计划将提供给成千上万的研究人员，让他们能用自己的方式开展研究，这完全是种全新的范式

Our device is multimodal, we produce it ourselves, it's very cheap to produce, and the plan is that we will provide it to thousands and thousands of people, to many researchers that actually that can use the system for their own studies and do it basically in a way that's a completely different paradigm.

这将把应用场景从需要依靠资助的博士后手中转移出来——他们可能要花费数月时间进行设备调试——转变为普通技术人员甚至个人在家就能使用，比如可以佩戴该设备睡觉，从而让设备能够分析使用者的睡眠阶段。

It would take the application away from the postdoc who is funded on a grant by, you know, and is spending months and months setting things up into a situation where a normal technician can maybe use it or maybe even the person at their own home can use it and for example sleep with the device and so that the device can provide an understanding of the person's sleep stages, for example.

这正是我们在应用神经科技前沿实验室开展初期工作的一部分，确实是非常激动人心的研究。

And so that's really part of the initial work we've begun at the Change Frontier Lab for Applied Neurotechnologies and really a very exciting work.

再次强调，我们的目标是改善人们的生活，这实际上建立在大量基础研究、基础工程和技术开发的成果之上，现在正将这些整合起来，开始创造真正能改善全球许多人生活的解决方案。

Again, trying to make people's lives better, and that really sits on the sort of tail end of a lot of the basic research and basic engineering and developing techniques, really putting all this together and now starting to make solutions that actually will improve the lives of many people worldwide.

嗯。

Yeah.

这太棒了。

That's beautiful.

当然，我衷心祝愿这项研发工作一切顺利。

And I I wish, of course, all the best with this development.

现在我们来回顾一下你至今完成的所有工作。

Now we'll look at everything you did so far.

你认为什么是你拥有的可迁移核心能力，能帮助你成功完成一个又一个项目？

What do you think is one main skill that you have that is transferable, that helped you to go from one project to another and be able to complete all of them so successfully?

这肯定不是技术技能。

It's certainly not a technical skill.

肯定不是某种计算机程序之类的东西。

It's certainly not, you know, oh, you know, this computer program or something.

我认为好奇心可能是第一位的。

I think curiosity is probably number one.

真正对事物保持兴奋和好奇。

Really being excited by and curious about things.

不仅仅是对已知事物感到兴奋，而是对未知事物更感兴趣，然后去学习其他东西。

Not just excited about, like, the kind of things you know, but actually more excited about the things you don't know then you learn about the other things.

如果你总是停留在已知领域，比如掌握某种编程语言就只应用它，就永远不会和医生交流，也不会和神经科学家对话。

If you always stay with the things you know, you'll know a certain programming language, you will apply that programming language, you'll never talk to a doctor, you'll never talk to a neuroscientist.

如果你是神经科学家，就永远不会和临床医生交谈。

If you're a neuroscientist, you'd never talk to a clinician.

事实上，这正是我经常看到的最大问题之一。

And in fact, this is actually one of the biggest problems I see all the time.

工程师或神经科学家能与临床医生高效合作的概率非常低，因为通常其中一方或双方对对方的工作领域并不真正感兴趣。

The chance that an engineer or neuroscientist can work together productively with a clinician is very low because typically one or both of them just aren't really interested in what the other person can do.

他们可能会说：'你其实对我的病人了解不多，但我却需要你大力协助。'

They may just say, Well, you really don't know very much about my patients, but I want you to help me a lot.

工程师当然会认为：'你根本不懂我正在做的精妙工程，而你对待病人的方式非常愚蠢，因为你在用30年前的老方法。'

You know, the engineer of course thinks, you really don't know anything about the amazing engineering that I'm doing, and what you're doing with your patients is very stupid because you're doing something that's 30 years old.

这种局面显然不会产生什么好结果，而这种情况却屡见不鲜。

And so that kind of situation is clearly not going to lead to a whole lot, and that happens all the time.

因此要保持开放心态，保持热情，始终对新事物充满兴趣。

So being open, being excited, being interested in new things and being like that all the time.

当然还要学会倾听那些否定意见，同时仍能勇往直前地付诸实践。

And then of course having to listen to people that tell you how it's not going to work and still forging ahead and actually doing something.

记得我刚提出功能映射这个概念时，曾咨询过美国顶尖的电刺激映射专家——一位极受尊敬的神经外科权威。

I remember when I first came up with this idea for functional mapping, I talked to one of the premier, most respected neurosurgeons in The United States who specialize on electrical stimulation mapping.

他叫乔治·奥格曼，来自神经外科世家，是位德高望重的资深前辈。

His name is George Orgemann, a very, very senior, very highly respected neurosurgeon coming from a whole dynasty of neurosurgeons.

我去找他时大概是2006年左右，当时我有了第一段关于这种不使用电刺激的被动功能定位的视频。

I went to him and this must have been 2,006 or so and I had my first video of this passive functional mapping that does not use electrical stimulation.

这本来就应该让我有所迟疑。

Now that already should have given me pause.

而我当时正在和一个毕生研究都基于电刺激的人交谈。

Now I'm talking to a person whose whole life was based on electrical stimulation.

我去找他说：'嘿乔治，你觉得怎么样？'

And I went to him and I said, Hey George, so what do you think?

我有个新技术。

I have this new technique.

它不需要用电刺激。

It does not use electrical stimulation.

你觉得如何？

What do you think?

当然，你猜得到他会说什么。

Well of course, I mean what do you think he said.

他花了半小时训斥我，说我刚才告诉他的想法有多愚蠢，列举了各种理由。

Well he gave me a half an hour lecture about how stupid it was, what I just told him, because all these different reasons.

当然现在回想起来，我当时很沮丧——其实也不算真的沮丧，但确实感到挫败，不过还是硬着头皮继续推进了。

And of course now, you know, I was depressed, well not really, but you know, I was frustrated, but of course pushed ahead anyway.

所以我认为，这本质上是一种改变事物的驱动力，一种内在的激情。

And so it's, I think, really a drive to change things, a drive, an inner passion.

我不认为这和智力有多大关系。

I I don't think it has very much to do with intelligence.

当然，你需要理解某些基本概念。

Of course, you know, understand certain things.

必须具备一定的基础智力水平才能有所作为。

There has to be a certain base level of intelligence so that you can do something.

但我认为最重要的能力其实是好奇心、驱动力、毅力，以及真正拥有远见。

But I think the biggest skills are really the curiosity, the drive, the perseverance, and really having this vision.

哇，如果我们能做到，那将会非常了不起。

Wow, if we do this, then this is really amazing.

我是说，你知道吗，就在我们和ECOG以及平克·弗洛伊德乐队合作这项研究时，我在想——让人们听平克·弗洛伊德的音乐，然后尝试从他们脑电信号中解码出音乐。

I mean, you know, I was just thinking when we were doing this study with ECOG and Pink Floyd, having people listen to Pink Floyd and trying to decode the music from the person's brain signal.

我当时就想，哇，如果这能实现，你就能听到一个人大脑里的声音了。

I was like, wow, if you can do that, you can listen to a person's brain.

这难道不神奇吗？

Isn't that amazing?

你明白吗？

You know?

而且你真的能听到。

And you can listen.

这就像你知道最早的演唱会录音是刻在蜡筒上的那种东西，现在你拿着设备，把唱针放上去录音然后就能回放？

It's almost like you know how they have like the first recordings of some concert or something that's imprinted on a wax drum or something, and now you have the device and you put the needle in and you record and you can play it back?

虽然这些录音可能有250年历史了，但你依然能听出是音乐，能辨认出是什么曲子，当然噪音很大也不完美。

You can kind of be it's two fifty years old or however old these things are, you can kind of hear it's kind of this music, but, you know, you know it's music, you kind of know what it is, but of course it's very noisy and not perfect.

但你会惊叹：这可是200多年前的东西啊，这可是人类最早的录音。

But you can see, wow, this is 200 something years old, you know, and this is the very first recording.

我当时就想，哇，如果我们能听到一个人的大脑活动，就能知道这个人正在听什么。

And this is the way I thought, wow, if we can listen to the person's brain and we can hear, you know, what the person is hearing.

这简直不可思议。

Now that is incredible.

其实这背后并没有什么科学假设，或者说没有很好的假设，纯粹是被这种可能性激发了灵感。

Now there's really no scientific hypothesis behind this or not a very good one anyway, but just being inspired by this possibility.

哇，如果能看见神经元如何兴奋起来，动作电位如何传递，那该多神奇啊。

Or wow, wouldn't it be amazing if you could see how neurons kind of like get excited and then this one, and then you can almost see the action potentials.

不是真的看见，但你可以想象动作电位从这里传递到那里，然后观察整个过程如何运作。

Not really see, but you can almost imagine like here and then the action potentials go over here and then and you can see how that works.

这才叫激动人心。

Now that is exciting.

说实话，要做这类研究并为之兴奋，可能需要某种特定性格的人。

And so doing something like this and being excited by it, it probably requires a certain personality, to be honest with you.

我不知道这在多大程度上可以后天培养，但这确实是部分必备条件。

I don't know how to what extent that can be taught or something, but that's certainly in part required.

如果这类事情不令人兴奋，如果因为没人做过或觉得太可怕而令人畏惧，那么如果你被吓到了，你就不会去做。

If things like this aren't exciting, if they are too scary because nobody's done it or something and it's intimidating, well, you're not going to do it if you're intimidated by this.

如果你感到害怕，而且你非常内向，害怕与知名临床医生交谈，那你根本就不会去和他说话。

If you're intimidated by and you're so introverted and intimidated by talking to a famous clinician, well, you're not going be talking to him.

我的意思是，你永远也学不会这个。

Mean, you're never going to learn this.

所以在某种程度上，这可能需要特定类型的人。

And so it probably requires a certain kind of person in a way.

好奇、外向、充满激情的人，而且可能还得勤奋努力。

Curious, extroverted, you know, passionate person being, you know, and probably hardworking.

我是说，这种事情可不是每天七小时就能完成的，明白吗？

I mean, this is not something that happens in seven hours a day, you know?

这需要大量工作。

It's a lot of work.

所有这些事情都需要付出大量努力。

And all of these things have been a lot of work.

现在说起来很容易，比如我有100多篇论文，所有这些记录、业务和新发明什么的。

And it's now easy to talk about, Oh, you know, like I have 100 and something papers and all this record and this business and this new technique or this new device or something.

说起来容易，但实际需要付出大量努力。

It's all easy to say, but it's a lot of work.

当然，你不会展示那些不成功的东西，对吧？

And of course, you don't show the things that don't work, you know?

总会有一些你本以为会很激动人心但最终未能实现的事情。

There's, of course, always things that don't work that you think are going to be exciting, but they don't.

它们没有成功。

They don't pan out.

所以虽然需要大量工作，但这段经历非常精彩。

So it's a lot of work, but it's been quite the ride.

我绝不会用这段经历交换任何东西。

I would not trade it for anything.

是的，这确实是一段探索不同领域的毕生难忘之旅。

It's been, yeah, really the ride of a lifetime learning about all these different areas.

一旦你理解了这些不同领域如何协同作用，你确实可以——你刚才提到了可迁移技能。

And once you understand how these different areas all play together, you can really and you talked about transferable skills.

你可以把这些应用到任何地方。

You can apply this anywhere.

比如我现在就创办了一家涉及金融市场并从事语音分析的企业。

You can well, so now I created a business in that's in financial markets and does voice analysis.

虽然这与ECOG脑电信号没有直接关联，但以不同方式运用了许多相同技能。

Well, it has nothing to do with ECOG brain signals direct I mean, directly, but it uses a lot of the skills in a different way.

沟通能力——既要能与临床医生对话，又能与神经科学家交流，理解他们的兴奋点和需求。

Communication, just to be able to talk with a clinician and with a neuroscientist and understand what gets them excited and what they want.

这真是项关键技能。

It's such a critical skill.

沟通能力可能是所有技能中最重要的。

Like communication is probably one of the most important skills in all of this.

如果你无法与不同领域的人沟通，事情就办不成。

If you cannot communicate with different people in different fields, it's not going to work.

毕竟，我们必须理解所有这些事物，而最令人兴奋、真正改变生活、重新定义类别并带来变革的，往往发生在不同学科的交叉界面上。

Because after all, all of these things we have to understand, and probably most things that happen that are exciting and really life changing and change category defining and changing are things that happen at the interfaces of different disciplines.

如今我们正处在ChatGPT和新型生成式AI的风口浪尖，这些主要是人工智能领域内极其重要的新进展。

Now we're, you know, on the verge of chatty PT and new generative AI, and those are primarily very important new developments and they happen within the domain of artificial intelligence.

它们可能是人工智能领域五十年来最重要的突破，但这类突破每五十年或一百年才会出现一次。

They're probably the most important developments in AI in fifty years, but those things happen every fifty years or every one hundred years.

大多数时候，新创新都来自不同学科的交叉点，因此我们需要能够真正连接这些不同学科的人才。

Most of the time new innovations come from the interface of different disciplines, and that's why you need people that can actually bridge these different disciplines.

这就是为什么需要一个让人们接触不同学科的环境，而不仅仅是教授算法的工程学院，然后你偶尔去找精神科医生获取数据，回来做数据分析。

That's why you need an environment that exposes people to these different disciplines, not just an engineering school that teaches the algorithms, and then you go over to some psychiatrist one time and get some data and go back and do some data analysis.

实际情况并非如此运作。

This is not how it works.

你需要经常过去，需要全程参与，需要持续与精神科医生交流，了解他们的工作方式，掌握他们的诊断方法，清楚他们当前的能力范围——这样你就不会重复他们已经做得更好、更轻松或更经济的工作。

You need to go over, you need to be there, you need to talk to the psychiatrist all the time, you need to understand what they do, you need to understand how they make their diagnosis, you need to understand what they can currently do, so you don't do something that they're already doing much better or very, very easily and cheaply.

你必须专注于他们无法做到，或者虽然能做但成本极高、耗时极长或问题重重的领域，对吧？

You have to do the things that they can't do or that they can only do and it's very expensive or time consuming or problematic, right?

所以就是这类事情，这类技能。

So it's those kinds of things, those kinds of skills.

它们与掌握多种算法、特定编程方式之类的事情毫无关系。

And they have nothing to do with like knowing many different algorithms or writing computer code a certain way or something.

主要是沟通能力、好奇心和毅力。

It's mostly communication and curiosity, perseverance.

因此，如果要我给任何人建议的话，我会说：这就是你需要的，若你真心想成功， 这些就是你需要培养的技能。

And so if I would give advice to anybody, I would say, well, this is what you need, and if you really want to succeed, then that's the kind of skills that you need.

确实如此。

Absolutely.

这也与博士的观点一致。

And it's also in line with what Doctor.

尼古拉斯·奥皮在我们本周即将发布的播客中谈到了坚持不懈的重要性，因为无论发生什么，都需要时间和极大的努力与决心才能前进。当然，他也提到了不同学科融合的关键性——他们开发的支架电极（stentrode）既是工程学的产物，又源于介入放射学技术，现在被重新设计用于脑机接口领域。

Nicholas Oepi was talking in the podcast that we recorded, it's coming up this week, he was talking about perseverance, because it takes time, it takes a lot of effort and determination to move forward, whatever is happening, and also he mentioned, of course, this merge between different disciplines, how important it is, because the stentrode that they develop, yes it's engineering, and it's interventional radiology, yes something that was used by interventional radiology, but engineered now for brain computer interfaces.

非常感谢你提到这一点。

So thank you so much for mentioning that.

你已经经营着自己的公司，有这么多研究生与你共事，你在寻找什么样的潜在员工和学生来与你合作？

And you already run your own company, you had so many graduate students working with you, What are you looking for in your potential employees, in students that you invite to work with you?

至少在我自己的工作中，我发现从具备扎实工程基础的人开始合作最为成功。

At least in my own work, I've been most successful starting with people that have a very strong engineering basis.

当然，每个人都有适合的位置。

Of course, there's a place for everyone.

如果是一个大项目，你知道的，总会为其他人也留有一席之地。

And if it's a big project and, you know, there's always a place for for other people too.

但工程能力——比如编写计算机代码、理解数学——这些是你无法事后补上的技能。

But the engineering is the like writing computer code, understanding mathematics, that's the thing you're not going to pick up afterwards.

这需要多年的学习积累。

It's something that requires many years of study.

它需要投入大量时间。

It requires a lot of time.

神经科学当然也需要学习，但理解起来并不困难。

The neuroscience, of course you have to learn something, but it's not hard to understand.

是的，你需要阅读资料。

Yes, you have to read things.

当然，你需要记住一些东西。

Yes, of course, you have to remember things.

你需要理解所读内容，但这本身并不难。

You have to understand what you're reading or something, but it's not hard per se.

医学也不难。

Medicine is also not hard.

当然，你需要理解一些概念。

You have to, of course, understand some concepts.

而且，如果你上医学院，那当然会很辛苦，因为要记忆大量内容。

And, of course, if you go through medical school, of course, it's very hard because you have to memorize a lot of things.

我可能都做不到，因为我的记忆力没那么好。

And I probably couldn't even do that because my memory is not that great.

但理解起来并不困难。

But it's not hard to understand.

例如，理解大脑可能存在白质畸形并不困难。

It's not hard to understand, for example, that you may have a white matter malformation in the brain.

所以大脑会有这些迁移障碍，比如大脑发育时，它大致是这样生长的。

So the brain has these migration disorders, for example, as the brain develops, it sort of grows like this.

可能会出现大脑中某部分本该到达皮层、位于大脑外部，却滞留在白质中的情况。

And it can happen that there's a part in the brain that's destined to go to the cortex, to be on the outside of the brain, but it actually gets stuck in white matter.

现在这基本上是一块位于你大脑内部的组织，但它没有到达应该去的位置。

Now it's basically this is a piece of tissue that's inside your brain, and it's not at the place where it's supposed to go.

而且显然它的连接也不正确，因为它本不该在那里。

And it's obviously not wired correctly because it's not supposed to be there.

这就是引发癫痫发作的常见原因之一。

That's one of the common reasons for having epileptic seizures, for example.

嗯，这部分并不难理解。

Well, that part is not hard to understand.

那么，你会怎么做呢？

Well, what would you do?

好吧，做个核磁共振。

Well, do an MRI.

做完核磁共振后，你会看到白质中有个灰色的东西。

You do an MRI and then you see some gray thing that's in the white matter.

你看到了本不该在那里的东西。

You see something that's not supposed to be there.

那你会怎么做？

Well what do you do?

那就把它切除掉。

Well you cut it out.

所以这并不复杂。

So it's not complicated.

因此做核磁共振的原因并不难理解。

So it's not why you do the MRI is not difficult to understand.

我的意思是，是的，你必须知道这些情况确实存在。

I mean, yes, you have to know that these things exist.

你必须理解并了解这一点，但这并不难理解。

You have to understand and know about this, but it's not hard to understand.

如果你和一个有医学学位的人讨论傅里叶变换，那就要困难得多。

If you talk about a Fourier transform to somebody with a medical degree, well that's much harder.

你没办法走捷径。

You can't there's no shortcut.

你不能简单地说‘只学快速傅里叶变换’就行。

You cannot just say just learn the fast Fourier transform.

你还得学习数学中的许多其他内容。

You have to learn many other things in mathematics too.

在这个特定案例中，我只需要了解我刚才描述的这个特定现象。

I only have to know about like in this particular case, I just have to know about this particular phenomenon I just described.

我不需要研究整个医学领域。

I don't have to study all of medicine.

我只需要理解这一点就够了。

I can just understand that.

因此，我认为从扎实的工程基础开始入门是非常好的途径。

So starting with a solid foundation in engineering, I think, is great way to get in.

这并不是说没有持有医学学位的优秀人士取得成功，当然这样的例子很多。

That's not to say that there aren't great people that have, you know, medical degrees that have been successful, of course, and, you know, there's many.

但我认为对于想进入这个领域的年轻人来说，拥有扎实的工程基础是非常非常有用的。

But I think from a young person's perspective that wants to get into the field, I think having a solid basis in engineering is I think very, very useful.

当然还包括许多可迁移技能——如果你会编程、懂数学并能正确应用数学和统计学，你就能胜任很多不同的工作。

And also, of course, a lot of transferable skills that go if you know how to program and you know how to do mathematics and apply mathematics and statistics correctly, you can do a lot of different things.

所以我认为这是非常非常重要的技能。

So that I think is a very, very important skill.

如果有人面临学习方向的选择，那就去学最难的数学、最多的数学、最多的统计学和最难的编程——这些技能将为你奠定终身受用的坚实基础，支撑你掌握我刚才描述的所有其他能力。

If somebody had a choice of what to learn and how to get in, well, learn the hardest math and the most math, the most statistics and the hardest programming that you can because that will literally set you up for a lifetime, a very strong foundation upon which you can be with all the other things I've just described.

非常感谢，希望我们的听众——特别是学生们——能听取这个建议。

Thank you so much, I hope our listeners are listening to this advice, our students.

我们的播客节目叫做《神经科医生挑战不可能》。

Our podcast is called Neurocarers Doing the Impossible.

在你的职业生涯和工作中，有哪些曾被你认为不可能实现的事情最终被证明是可能的？你是如何实现它的？

What is something that you may be thought of being impossible that actually proved to be possible in your career, in the work that you did, and how did you make it possible?

确实，这样的例子可能有很多，但我们能从大脑中学习到的功能细节层次是极其深刻的。

Really, again, and there's probably several examples, but the level of detail of function that we can learn from the brain is really profound.

当我刚开始时，我们基本上只能判断一个人是否在动手。

When I started, we could basically tell, oh, is somebody moving the hand or not moving?

我们完全不知道手是朝这个方向还是那个方向移动，也不知道手是在做这个动作还是那个动作，甚至不清楚手具体在做什么。

We had no idea whether the hand is moving this way or that way, whether the hand does this or this or what the hand does.

我们当时完全没想到。

We had no idea.

五年、十年后，我们已经能解码手的运动了。

Well, five years, ten years later, we can decode the movement of the hand.

我们不仅能解码是哪一根手指在弯曲（五根中的哪一根），还能解码每根手指的具体运动轨迹。

We can decode not only which finger is flexing, which of the five fingers, we can decode the specific trajectory of each of the five fingers.

这简直太不可思议了。

I mean, it's ridiculous.

这完全令人难以置信。

It's completely unbelievable.

我们不仅能识别某人是在说话还是想象词语，还能解码单词中的元音和辅音，甚至能完整解码实际说出或想象的句子中的完整单词和句子。

We can learn not only whether somebody is speaking a word or imagining a word, we can decode the vowels and the consonants in the word, we can decode the full words, the full sentences in actually spoken or imagined sentences.

这太疯狂了。

This is crazy.

我是说，这完全超乎想象。

I mean, this is completely unbelievable.

十年前，我根本无法想象我们能实现这样的技术。

Ten years earlier, I would have never imagined that we'll be able to do things like this.

就像我说的，这还远未结束，我们还能发现其他各种类型的信息。

And like I said, it doesn't just stop with that, it goes on to all these other different types of information that we can find.

所以我的意思是，通过这些电极从大脑解码的信息量是极其庞大的。

So I mean, amount of information with these electrodes that we can decode from the brain is absolutely mobile.

问题在于我们无法在普通人身上植入电极——如果我们能通过不依赖这些复杂电极的方式获取这类信息，社会将发生翻天覆地的变化。

It's just we cannot put electrodes in normal people because if we could, if we could just get access to this kind of information in a way that doesn't require these complicated electrodes, society would change would would change dramatically.

我们将拥有一种与大脑的接口，让人们真正开始与互联网融合。

We would have an interface to the brain where people literally would start fusing with the Internet.

我们可以实现从大脑直接到互联网的即时通讯。

We could make direct communication from our brain directly to the Internet.

现在我们知道这在理论上是可行的。

Now we know it's theoretically possible.

我们知道理论上我们可以做到这些事情。

We know that we can in theory do these things.

但实际上我们无法实践，因为这需要脑部手术以及所有相关的并发症、风险和成本等等。

But we really can't do it in practice because it requires brain surgery and all these complications and risks and costs and so forth that are associated with this.

显然，由于这些原因，这不会在很多人身上实施。

So obviously this is not going to be done in a lot of people for these reasons.

但真正看到什么是可能的，就像我说的，从能看到某人想象移动右手到解码完整句子，你知道，这在十年间就实现了。

But actually seeing what's possible and going from, like I said, yeah, you can see that somebody's imagining to move the right hand to decoding full sentences, you know, and that in the time span of, ten years.

这真的令人难以置信。

I mean, this is really unbelievable.