What Neuralink Is Really Working on | Dr. Matt MacDougall & Dr. Andrew Huberman

00:00:00.000 | We could talk a little bit about what I consider really the holy grail of the nervous system,

00:00:07.900 | which is neuroplasticity.

00:00:08.900 | It's an incredible capacity of the nervous system to change its wiring, strengthen connections,

00:00:15.140 | weaken connections, maybe new neurons, but probably more strengthening and weakening

00:00:18.500 | of connections.

00:00:19.500 | Nowadays, we hear a lot of excitement about so-called classical psychedelics like LSD

00:00:25.280 | and psilocybin, which do seem to "open plasticity."

00:00:28.920 | They do a bunch of other things too, but through the release of neuromodulators like serotonin

00:00:33.800 | and so forth.

00:00:35.840 | How do you think about neuroplasticity?

00:00:38.420 | And more specifically, what do you think the potential for neuroplasticity is in the adult,

00:00:44.400 | so let's say older than 25-year-old brain, with or without machines being involved?

00:00:52.120 | Because in your role at Neuralink and as a neurosurgeon in other clinical settings, surely

00:00:58.360 | you are using machines and surely you've seen plasticity in the positive and negative direction.

00:01:06.080 | What do you think about plasticity?

00:01:07.360 | What's possible there without machines?

00:01:09.920 | What's possible with machines?

00:01:12.400 | So as you mentioned or alluded to, the plasticity definitely goes down in older brains.

00:01:21.020 | It is harder for older people to learn new things, to make radical changes in their behavior,

00:01:26.480 | to kick habits that they've had for years.

00:01:33.420 | Machines aren't the obvious answer, so implanted electrodes and computers aren't the obvious

00:01:37.700 | answer to increase plasticity necessarily, compared to drugs.

00:01:42.760 | We already know that there are pharmacologics, some of the ones you mentioned, psychedelics,

00:01:47.960 | that have a broad impact on plasticity.

00:01:51.320 | It's hard to know which area of the brain would be most potent as a stimulation target

00:01:55.740 | for an electrode to broadly juice plasticity compared to pharmacologic agents that we already

00:02:04.800 | know about.

00:02:07.320 | I think with plasticity, in general, you're talking about the entire brain.

00:02:11.320 | You're talking about altering a trillion synapses all in a similar way in their tendency to

00:02:20.040 | be rewirable, their tendency to be up or down weighted.

00:02:26.760 | An electrical stimulation target in the brain necessarily has to be focused.

00:02:32.920 | With a device like potentially Neuralinks, there might be a more broad ability to steer

00:02:37.980 | current to multiple targets with some degree of control, but you're never going to get

00:02:44.120 | that broad targetability with any electrodes that I can see coming in our lifetimes, say

00:02:53.920 | that would be coating the entire surface and depth of the brain the way that a drug can.

00:02:59.740 | I think plasticity research will bear the most fruit when it focuses on pharmacologic

00:03:06.160 | agents.

00:03:07.160 | I wasn't expecting that answer, given that you're at Neuralink.

00:03:11.200 | Then again, I think that all of us, me included, need to take a step back and realize that

00:03:17.440 | while we may think we know what is going on at Neuralink in terms of the specific goals

00:03:23.120 | and the general goals, and I certainly have in mind, I think most people have in mind

00:03:28.000 | a chip implanted in the brain or maybe even the peripheral nervous system that can give

00:03:33.120 | people super memories or some other augmented capacity.

00:03:38.180 | We really don't know what you all are doing there.

00:03:40.960 | For all we know, you guys are taking or administering psilocybin and combining that with stimulation.

00:03:46.600 | We really don't know.

00:03:47.600 | I say this with a tone of excitement because I think that one of the things that's so exciting

00:03:54.620 | about the different endeavors that Elon has really spearheaded, SpaceX, Tesla, et cetera,

00:04:01.480 | is that early on, there's a lot of mystique.

00:04:05.720 | Mystique is a quality that is not often talked about, but it's I think a very exciting time

00:04:13.720 | in which engineers are starting to toss up big problems and go for it.

00:04:20.060 | Obviously, Elon is certainly among the best, if not the best, in terms of going really

00:04:25.000 | big.

00:04:26.000 | I mean, Mars seems pretty far to me.

00:04:27.920 | Electric cars all over the road nowadays are very different than the picture a few years

00:04:31.400 | ago when you didn't see so many of them, rockets and so forth, and now the brain.

00:04:38.920 | To the extent that you are allowed, could you share with us what your vision for the

00:04:46.040 | missions at Neuralink are and what the general scope of missions are?

00:04:50.880 | Then if possible, share with us some of the more specific goals.

00:04:55.360 | I can imagine basic goals of trying to understand the brain and augment the brain.

00:04:58.880 | I could imagine clinical goals of trying to repair things in humans that are suffering

00:05:03.320 | in some way, or animals for that matter.

00:05:05.800 | Yeah.

00:05:06.800 | It's funny what you mentioned.

00:05:09.840 | Neuralink, and I think Tesla and SpaceX before it, end up being these blank canvases that

00:05:16.440 | people project their hopes and fears onto.

00:05:20.600 | We experience a lot of upside in this.

00:05:23.120 | People assume that we have superpowers in our ability to alter the way brains work,

00:05:29.160 | and people have terrifying fears of the horrible things we're going to do.

00:05:33.560 | For the most part, those extremes are not true.

00:05:37.720 | We are making a neural implant.

00:05:40.420 | We have a robotic insertion device that helps place tiny electrodes, smaller than the size

00:05:48.360 | of a human hair, all throughout a small region of the brain.

00:05:54.160 | In the first indication that we're aiming at, we are hoping to implant a series of these

00:06:00.720 | electrodes into the brains of people that have had a bad spinal cord injury.

00:06:06.400 | People that are essentially quadriplegic, they have perfect brains, but they can't use

00:06:11.200 | them to move their body.

00:06:12.400 | They can't move their arms or legs.

00:06:14.720 | Because of some high level spinal cord damage.

00:06:17.040 | Exactly right.

00:06:18.040 | And so this pristine motor cortex up in their brain is completely capable of operating a

00:06:24.000 | human body.

00:06:25.000 | It's just not wired properly any longer to a human's arms or legs.

00:06:29.920 | And so our goal is to place this implant into a motor cortex and have that person be able

00:06:38.120 | to then control a computer.

00:06:41.440 | So a mouse and a keyboard, as if they had their hands on a mouse and a keyboard, even

00:06:46.720 | though they aren't moving their hands, their motor intentions are coming directly out of

00:06:51.320 | the brain into the device.

00:06:54.600 | And so they're able to regain their digital freedom and connect with the world through

00:06:59.640 | the internet.

00:07:01.260 | Why use robotics to insert these chips?

00:07:04.740 | And the reason I ask that is that sure, I can imagine that a robot could be more precise

00:07:10.940 | or less precise, but in theory, more precise than the human hand.

00:07:15.220 | No tremor, for instance.

00:07:20.020 | More precision in terms of maybe even a little micro-detection device on the tip of the blade

00:07:28.860 | or something that could detect a capillary that you would want to avoid and swerve around

00:07:33.020 | that the human eye couldn't detect.

00:07:35.260 | And you and I both know, however, that no two brains nor are the two sides of the same

00:07:41.540 | brain identical.

00:07:44.060 | So navigating through the brain is perhaps best carried out by a human.

00:07:50.420 | However, and here I'm going to interrupt myself again and say, 10 years ago, face recognition

00:07:58.340 | was very clearly performed better by humans than machines.

00:08:04.140 | And I think now machines do it better.

00:08:07.340 | So is this the idea that eventually, or maybe even now, robots are better surgeons than

00:08:12.860 | humans are?

00:08:14.740 | In this limited case, yes.

00:08:17.260 | These electrodes are so tiny and the blood vessels on the surface of the brain so numerous

00:08:21.980 | and so densely packed that a human physically can't do this.

00:08:26.980 | A human hand is not steady enough to grab this couple micron width loop at the end of

00:08:33.940 | our electrode thread and place it accurately, blindly, by the way, into the cortical surface

00:08:41.460 | accurately enough at the right depth to get through all the cortical layers that we want

00:08:45.300 | to reach.

00:08:47.540 | And I would love if human surgeons were essential to this process, but very soon humans run

00:08:57.900 | out of motor skills sufficient to do this job.

00:09:02.120 | And so we are required in this case to lean on robots to do this incredibly precise, incredibly

00:09:09.580 | fast, incredibly numerous placement of electrodes into the right area of the brain.

00:09:15.640 | So in some ways, Neuralink is pioneering the development of robotic surgeons as much as

00:09:20.420 | it's pioneering the exploration and augmentation and treatment of human brain conditions.

00:09:26.100 | Right.

00:09:27.100 | So as the device exists currently, as we're submitting it to the FDA, it is only for the

00:09:33.340 | placement of the electrodes.

00:09:35.440 | The robot is part of the surgery.

00:09:37.680 | I or another neurosurgeon still needs to do the, you know, the more crude part of opening

00:09:42.380 | the skin and skull and presenting the robot a pristine brain surface to sew electrodes

00:09:48.600 | into.

00:09:49.600 | Well, surely getting quadriplegics to be able to move again, or maybe even to walk again

00:09:55.640 | is a heroic goal and one that I think everyone would agree would be wonderful to accomplish.

00:10:03.900 | Is that the first goal because it's hard, but doable, or is that the first goal because

00:10:12.320 | you and Elon and other folks at Neuralink have a passion for getting paralyzed people

00:10:18.860 | to move again?

00:10:19.860 | You know, broadly speaking, you know, the mission of Neuralink is to reduce human suffering,

00:10:24.840 | at least in the near term.

00:10:26.940 | You know, there's hope that eventually there's a use here that makes sense for a brain interface

00:10:33.380 | to bring AI as a tool embedded in the brain that a human can use to augment their capabilities.

00:10:41.900 | I think that's pretty far down the road for us, but definitely on a desired roadmap.

00:10:48.540 | In the near term, we really are focused on people with terrible medical problems that

00:10:52.800 | have no options right now.

00:10:56.680 | With regard to motor control, you know, our mutual friend recently departed, Krishna Shenoy,

00:11:04.920 | was a giant in this field of motor prosthesis.

00:11:09.720 | It just so happens that his work was foundational for a lot of people that work in this area,

00:11:15.080 | including us.

00:11:16.080 | And he was an advisor to Neuralink.

00:11:19.560 | That work was farther along than most other work for addressing any function that lives

00:11:25.560 | on the surface of the brain.

00:11:27.720 | The physical constraints of our approach require us currently to focus on only surface features

00:11:33.320 | on the brain.

00:11:34.500 | So we can't say, go to the really very compelling surface, deep depth functions that happen

00:11:43.000 | in the brain, like, you know, mood, appetite, addiction, pain, sleep.

00:11:51.480 | We'd love to get to that place eventually, but in the immediate future, our first indication

00:11:56.640 | or two or three will probably be brain surface functions like motor control.

00:12:01.680 | [MUSIC PLAYING]

What Neuralink Is Really Working on | Dr. Matt MacDougall & Dr. Andrew Huberman

Chapters