WEBVTT 1 00:00:00.200 --> 00:00:04.320 Welcome to Neuralink Live Update. We're gonna tell you about 2 00:00:05.280 --> 00:00:09.919 the progress of the first patient with the neuralink olland 3 00:00:10.000 --> 00:00:15.359 and sort of your recap of the progress there, then 4 00:00:15.400 --> 00:00:19.679 talk about what changes we were making for the second patient, 5 00:00:19.719 --> 00:00:23.239 which we're hoping to do an implant in the next 6 00:00:23.239 --> 00:00:26.280 week or so. And this is for our first product, 7 00:00:26.280 --> 00:00:29.199 which is called Levathy, which enables you to control a 8 00:00:29.239 --> 00:00:33.320 computer or a phone just by thinking. So let's in fact, 9 00:00:33.359 --> 00:00:36.719 so we'll start off with just some introductions. Do you start? Hi? 10 00:00:36.719 --> 00:00:39.600 Everyone, My name is DJ sull I'm an electrical engineer 11 00:00:39.640 --> 00:00:43.119 and a chip designer by training. I led the design 12 00:00:43.240 --> 00:00:48.600 of first several generations of the Neuralink implant. Currently I 13 00:00:48.640 --> 00:00:50.560 was on the founding team and currently a president. 14 00:00:51.799 --> 00:00:52.799 I'm Matthew McDougal. 15 00:00:52.840 --> 00:00:56.359 I'm a practicing neurosurgeon and a head of neurosurgery at Neuralink. 16 00:00:57.759 --> 00:01:03.119 Yeah go ahead, yeah, head of burning the first as applications. 17 00:01:02.560 --> 00:01:04.599 And I'm plussed some software engineer at Nerlink trying to 18 00:01:04.599 --> 00:01:07.079 figure out how to turn brain activity into the stuff 19 00:01:07.079 --> 00:01:07.439 in the work. 20 00:01:08.280 --> 00:01:10.040 All right, thank you, Well, so let's see, so we'll 21 00:01:10.079 --> 00:01:16.000 just get going into the presentation. So our first product 22 00:01:16.200 --> 00:01:19.400 is something like I said, we call telepathy, which enables 23 00:01:20.400 --> 00:01:24.040 the person with the neural link implant to control their 24 00:01:24.040 --> 00:01:26.799 phone or computer just by thinking. And once you can 25 00:01:26.799 --> 00:01:29.319 control your phone and computer, you can essentially control almost 26 00:01:29.319 --> 00:01:32.439 anything just and literally just by thinking. So there's no 27 00:01:33.040 --> 00:01:37.680 eye tracking or anything. It is purely purely your thoughts. 28 00:01:38.280 --> 00:01:44.159 So this is really quite quite a profound device that 29 00:01:44.280 --> 00:01:46.640 can help a lot of people who have lost the 30 00:01:46.680 --> 00:01:51.200 connection between their brain and body. So imagine people like 31 00:01:52.560 --> 00:01:55.799 Stephen Hawking who you know. You imagine if if he 32 00:01:55.840 --> 00:01:58.359 could communicate at the same speed as someone who had 33 00:01:58.400 --> 00:02:01.079 still had the connection to their brain and body. So 34 00:02:01.239 --> 00:02:04.840 it's it's really uh something that can help millions of 35 00:02:04.840 --> 00:02:07.599 people around the world, and it's a it's part of 36 00:02:07.640 --> 00:02:13.039 our overall goal of enabling a very high bandwidth connection 37 00:02:13.159 --> 00:02:16.919 between the brain and and your and and the rest 38 00:02:16.919 --> 00:02:20.240 of the world and your computers. The long term goal, 39 00:02:20.360 --> 00:02:23.599 which sounds a little esoteric, is to mitigate the the 40 00:02:23.719 --> 00:02:28.680 risk of of the civil civilizational risk of AI by 41 00:02:29.280 --> 00:02:34.439 having a sort of closer symbiosis between human intelligence and 42 00:02:35.080 --> 00:02:39.240 digital intelligence. But but that that that will take many years. Uh. 43 00:02:39.800 --> 00:02:42.759 Along the way, We're we're going to help solve a 44 00:02:42.759 --> 00:02:47.479 lot of brain injury or spinal injury issues. Uh so 45 00:02:48.439 --> 00:02:52.599 uh and with without first product sypathy, that's that's going 46 00:02:52.639 --> 00:02:56.000 to be really quite profound. Uh that there is also 47 00:02:56.120 --> 00:03:00.960 potential long term for bridging the gaps. So if there 48 00:03:01.000 --> 00:03:05.759 are damaged or severed neurons, being able to span the 49 00:03:05.759 --> 00:03:08.360 gap between the brain's motor cortex to the spine to 50 00:03:08.439 --> 00:03:11.759 enable someone to use their body again, I think that 51 00:03:11.800 --> 00:03:16.400 would be very exciting, and it's you know that that 52 00:03:16.520 --> 00:03:21.520 is something that is possible in the long term and 53 00:03:21.520 --> 00:03:26.039 they're not. Our second product, which we've demonstrated to work 54 00:03:26.199 --> 00:03:30.159 with monkeys, is blind sight, which would enable someone who 55 00:03:30.240 --> 00:03:34.479 is completely blind, lost both eyes or completely lost that 56 00:03:34.560 --> 00:03:38.319 are optic nerve to be able to see. So that's 57 00:03:39.560 --> 00:03:41.719 that's something that we hope to demonstrate in the future. 58 00:03:47.479 --> 00:03:49.039 So this just gives you a sense of what the 59 00:03:49.080 --> 00:03:52.280 device is. A way to think about the neuralink device 60 00:03:52.520 --> 00:03:54.520 is kind of like a fitbit or an Apple watch 61 00:03:54.919 --> 00:04:00.800 with tiny wires or electrodes. Those those tiny wires are 62 00:04:00.800 --> 00:04:04.800 implanted in the in the brain and they are read 63 00:04:04.800 --> 00:04:09.159 and write electrical signals. So a lot of people think 64 00:04:09.199 --> 00:04:12.479 the brain is this incredibly mysterious thing. It's it is 65 00:04:12.520 --> 00:04:14.479 mysterious a lot of ways, but but it is. Actually 66 00:04:15.319 --> 00:04:17.560 it does operate with like electrical signals, so if you 67 00:04:17.560 --> 00:04:20.680 can read and write those electrical signals, you can interface 68 00:04:20.720 --> 00:04:25.959 with the brain. And the device is sized so that 69 00:04:26.120 --> 00:04:29.560 it is the same size as the as the piece 70 00:04:29.560 --> 00:04:32.720 of skull that is removed. So if it's like a 71 00:04:32.720 --> 00:04:36.199 few centimeters diameter of skull skull that's removed, we replace 72 00:04:36.279 --> 00:04:40.639 that with the device after implanting the tiny wires with 73 00:04:40.639 --> 00:04:45.639 the surgical robot, and that enables read write capability to 74 00:04:45.839 --> 00:04:50.800 the neurons completely wirelessly. Yeah, yes, exactly, it's completely wirelessly. 75 00:04:50.920 --> 00:04:54.480 So like I could I could have a neuralank right now. 76 00:04:54.519 --> 00:04:58.959 You wouldn't know. And it charges inductively so you could 77 00:04:58.959 --> 00:05:04.240 just uh it's they have electromagnetic ad that that you 78 00:05:04.360 --> 00:05:07.600 charge the device with. So yeah, it's like an Apple 79 00:05:07.600 --> 00:05:08.240 Watch exactly. 80 00:05:08.279 --> 00:05:12.199 So uh, except that it's actually a much harder technical 81 00:05:12.680 --> 00:05:15.759 challenge to solve that there's limit as to how much 82 00:05:16.000 --> 00:05:20.160 get the print issue or whereas in for phones and. 83 00:05:19.800 --> 00:05:21.560 Really care if you're sitting on the table. 84 00:05:23.160 --> 00:05:25.040 Yeah. Also it's got to go through skin and stuff 85 00:05:25.079 --> 00:05:28.439 as well. And okay, so it is a tougher challenge 86 00:05:28.480 --> 00:05:32.959 to to charge and to have vivemand with communications, given 87 00:05:32.959 --> 00:05:35.240 that has got to go through skin and hair and 88 00:05:35.279 --> 00:05:41.519 stuff we have we have solved. So yeah. 89 00:05:41.600 --> 00:05:44.759 Yeah, So our first step with the telepathy is basically 90 00:05:44.800 --> 00:05:50.720 to unlock digital independence for people with policies and to 91 00:05:50.879 --> 00:05:53.600 allow them to control the computer just with their mind 92 00:05:53.639 --> 00:05:57.600 without moving their body. And our goal is to provide 93 00:05:57.639 --> 00:06:01.879 them the same level of control, functionality, and reliability that 94 00:06:01.959 --> 00:06:04.199 I have when I'm using a computer, even better than 95 00:06:04.800 --> 00:06:05.839 the level of control I have. 96 00:06:09.360 --> 00:06:11.160 Just to be clear, this guy he's controlling us with 97 00:06:11.199 --> 00:06:13.120 his brain. So he's not like you can't see his 98 00:06:13.120 --> 00:06:15.199 hands in this video, but he's not using a mouse 99 00:06:15.240 --> 00:06:18.079 and keyboard, just thinking about how to move the cursor 100 00:06:18.160 --> 00:06:20.319 and playing civilizations. 101 00:06:19.519 --> 00:06:27.560 No eye trackers from meat, just thinking. 102 00:06:26.680 --> 00:06:28.279 Just a couple of days a cursor move here. 103 00:06:28.839 --> 00:06:29.319 Yeah. 104 00:06:29.399 --> 00:06:31.000 Yeah, it's like the last standard two nights ago. 105 00:06:31.439 --> 00:06:33.240 Yes, yeah, I think I think the way he also 106 00:06:33.240 --> 00:06:35.720 described that is he's using them. 107 00:06:35.800 --> 00:06:40.240 Yeah, there's many more videos on his on the platform. 108 00:06:40.560 --> 00:06:43.959 Definitely checked them out. Yeah, so he can he's string 109 00:06:43.959 --> 00:06:45.800 with that live and also can talk and like move 110 00:06:45.839 --> 00:06:47.240 his head with that problem multi time. 111 00:06:48.199 --> 00:06:49.879 Yeah, you guys, like, if you join his livestream, you 112 00:06:49.879 --> 00:06:52.000 can ask him questions. You'll he'll tell you all about 113 00:06:52.319 --> 00:06:53.399 what it's like to move. 114 00:06:54.199 --> 00:06:58.040 Also, I think I haven't played Civilization myself, but I 115 00:06:58.079 --> 00:06:59.079 think this is actually not easy. 116 00:06:59.079 --> 00:07:01.120 About this is expert Emperor mode. 117 00:07:01.519 --> 00:07:03.519 Emperor if you have played stiff empro mote is like 118 00:07:03.560 --> 00:07:05.639 the highest difficulty level. Just the point is that this 119 00:07:05.680 --> 00:07:07.839 is a kind of demanding task. And while wife streaming 120 00:07:07.839 --> 00:07:10.800 playing the hardest motive team and he's able to do 121 00:07:10.839 --> 00:07:11.360 that while. 122 00:07:11.160 --> 00:07:14.519 Moving on talking engaging with you know, the audience. 123 00:07:14.959 --> 00:07:17.720 Playing One of the other games he likes to play 124 00:07:17.759 --> 00:07:20.480 a lot is chess. I think it gets lost sometimes 125 00:07:20.519 --> 00:07:24.879 that he's actually playing speed chess against me, which requires 126 00:07:25.040 --> 00:07:29.720 an incredibly high fidelity degree of control and speed of 127 00:07:29.759 --> 00:07:31.199 control in order to be able to win. 128 00:07:35.199 --> 00:07:38.279 So also, another cool stuff about about our device is 129 00:07:38.279 --> 00:07:41.839 that is it anywhere anytime, also on a plane during 130 00:07:41.839 --> 00:07:46.199 a flight, while sating really cool memes of care. And 131 00:07:47.000 --> 00:07:53.120 also our device unlocks things that previously were impossible for participants. 132 00:07:53.399 --> 00:07:57.839 For example, we're able to connect him to his living consoles, 133 00:07:57.879 --> 00:08:00.879 which they marcott with friends and family, and it was 134 00:08:01.120 --> 00:08:04.920 lovely to see them playing together after years that he 135 00:08:04.959 --> 00:08:06.720 couldn't do it since his injury. 136 00:08:07.279 --> 00:08:10.399 I know, if you're sitting over from this guy playing 137 00:08:11.079 --> 00:08:14.959 look over, he's going a cap me love, no hands, 138 00:08:15.040 --> 00:08:15.759 no movement. 139 00:08:17.519 --> 00:08:18.279 Even away over. 140 00:08:18.600 --> 00:08:20.160 Yeah, it's strange, strange time. 141 00:08:21.839 --> 00:08:22.079 Yeah. 142 00:08:22.120 --> 00:08:25.720 And he loves using the device and using independently daily 143 00:08:25.839 --> 00:08:32.039 to watch videos and read play games using the baser. 144 00:08:32.720 --> 00:08:35.200 And the key metrics that week are is to make 145 00:08:35.200 --> 00:08:38.519 sure our device is actually useful to basically the amount 146 00:08:38.519 --> 00:08:41.039 of hours and use the device daily and weekly, and 147 00:08:41.120 --> 00:08:44.879 we track it weekly since the since the surgery, and 148 00:08:45.039 --> 00:08:47.799 on weeks that he's not too busy and not traveling, 149 00:08:48.159 --> 00:08:51.200 he can even reach seventy hours of using the. 150 00:08:51.240 --> 00:08:52.360 Device a week. 151 00:08:52.639 --> 00:08:55.399 This is amazing and he would of course love to 152 00:08:55.480 --> 00:08:59.639 use it more, but need to run result sessions, he 153 00:08:59.720 --> 00:09:02.080 needs to sleep sometimes and also of course to charge 154 00:09:02.080 --> 00:09:04.639 the device once in a while hopefully will improve that 155 00:09:04.720 --> 00:09:05.159 over time. 156 00:09:06.320 --> 00:09:08.240 Maybe not obvious to people who are watching this like 157 00:09:08.279 --> 00:09:10.279 it's a normal math book. He's controlling. This isn't like 158 00:09:10.320 --> 00:09:13.279 some limited addition thing. But there's only a few options, 159 00:09:13.279 --> 00:09:14.440 Like you can just do anything that you can do 160 00:09:14.480 --> 00:09:16.519 on a mapbook pro that same one I have on 161 00:09:16.559 --> 00:09:20.039 my desk. Actually it's the exact same one. 162 00:09:19.960 --> 00:09:24.919 Yeah, And another interesting point is that on the first 163 00:09:25.000 --> 00:09:29.759 day he used bciurn Control, was able to break the 164 00:09:29.879 --> 00:09:34.159 previous world record for curser control just by using the 165 00:09:34.200 --> 00:09:37.559 brain and recently he even doubled it and was able 166 00:09:37.639 --> 00:09:41.960 to outperform about ten percent of our engineer at New Link. 167 00:09:42.080 --> 00:09:43.960 And you can be sure that we are very good 168 00:09:43.960 --> 00:09:46.039 in this game and very quick. And if you want 169 00:09:46.080 --> 00:09:48.279 to check out how well how well you can do it, 170 00:09:48.320 --> 00:09:51.039 you can do it on our website. And it's very 171 00:09:51.320 --> 00:09:54.440 addictive games. Yeah, it's a very simple game. You just 172 00:09:54.480 --> 00:09:55.279 have to click on the square. 173 00:09:55.399 --> 00:10:01.600 But it's actually even though it sounds silly, it's it's 174 00:10:01.639 --> 00:10:04.519 quite a Yeah, it can be quite a It sounds 175 00:10:04.519 --> 00:10:07.080 like it could be quite addictive, and it's especially if 176 00:10:07.080 --> 00:10:08.519 you get a low score and you think there's no 177 00:10:08.600 --> 00:10:11.679 way I got it. So, I mean, anyone who wants 178 00:10:11.720 --> 00:10:13.679 to try this, I recommend going to the neuralink dot 179 00:10:13.679 --> 00:10:16.159 com its website and seeing seeing if you can beat 180 00:10:16.159 --> 00:10:19.000 Nolan's record, and it's that you will find that's actually 181 00:10:19.039 --> 00:10:22.559 quite difficult to do so. And this is really with 182 00:10:23.080 --> 00:10:25.559 version one of the device and with only a small 183 00:10:25.600 --> 00:10:29.519 percentage of the electrodes that are working. So this is 184 00:10:30.840 --> 00:10:32.879 this is really just the beginning, but even the beginning 185 00:10:33.240 --> 00:10:36.279 is twice as good as the world record. This is 186 00:10:36.320 --> 00:10:40.480 important to emphasize the you know, the media has a 187 00:10:40.519 --> 00:10:44.679 habit of saying that the glasses ten percent empty, but 188 00:10:45.039 --> 00:10:48.759 actually it's ninety percent full. So I think it's really 189 00:10:48.799 --> 00:10:52.960 quite accomplishment of the Neuralink team to have achieved with 190 00:10:52.960 --> 00:10:57.960 with the first patient, first device, twice the world record 191 00:10:58.120 --> 00:11:05.759 for the range computer madwidth. That's really an astonishing, an 192 00:11:05.799 --> 00:11:09.320 amazingly great outcome, and it's only going to get better 193 00:11:09.320 --> 00:11:12.679 from here. So the potential is to ultimately get I 194 00:11:12.720 --> 00:11:18.600 think to megabit level. So that's that's probably the long 195 00:11:18.679 --> 00:11:22.200 term goal of improving the bandwidth of the brain computer interface. 196 00:11:22.639 --> 00:11:26.240 If you think about like how low the bandwidth normally 197 00:11:26.320 --> 00:11:30.039 is between a human and advice, it's the average bandwidth 198 00:11:30.120 --> 00:11:33.759 is extremely low. It's it's i say, less than one 199 00:11:33.799 --> 00:11:36.159 bit per second over the course of a day. If 200 00:11:36.159 --> 00:11:38.080 there are eighty six thousand, four hundred seconds in a day, 201 00:11:38.559 --> 00:11:42.559 you're outputting less than that number of bets to any 202 00:11:42.559 --> 00:11:50.240 any given device except perhaps very rare circumstances. So this 203 00:11:50.320 --> 00:11:57.080 is it's actually quite important for for AI. You know, 204 00:11:57.080 --> 00:12:03.960 basic for human AI sebiosis is just being able to 205 00:12:03.960 --> 00:12:09.679 communicate at a speed. Yeah, I can follow. 206 00:12:11.600 --> 00:12:15.879 So yeah, just to emphasize again, he's performing at this 207 00:12:16.000 --> 00:12:20.039 extremely high level with about fifteen percent of his channels functional, 208 00:12:22.120 --> 00:12:25.039 and so we want to mitigate any of the problems 209 00:12:25.039 --> 00:12:28.639 that led to that situation. So, you know, the brain 210 00:12:28.720 --> 00:12:31.879 is a fascinating organ. I'll share with you some of 211 00:12:31.879 --> 00:12:35.399 the secrets about the brain. During any typical brain surgery, 212 00:12:36.200 --> 00:12:38.799 a small amount of air is introduced into the skull. 213 00:12:40.000 --> 00:12:42.480 That's because neurosurgeons like to have as much room as 214 00:12:42.519 --> 00:12:46.919 possible around the brain, and so there's this little known 215 00:12:46.960 --> 00:12:53.759 control mechanism of allowing the CO two concentration in the 216 00:12:53.799 --> 00:12:57.519 blood to rise a bit, which allows the brain to 217 00:12:57.679 --> 00:13:00.919 either expand or contract, depending on where you you target 218 00:13:00.960 --> 00:13:04.080 that CO two. But typically neurosurgeons will have the brain 219 00:13:04.120 --> 00:13:07.799 shrink by lowering CO two. What we're going to do 220 00:13:07.879 --> 00:13:11.960 in our future surgeries is keep the CO two concentration 221 00:13:12.080 --> 00:13:15.720 actually quite normal, maybe even slightly elevated, and I'll allow 222 00:13:15.759 --> 00:13:18.639 the brain to stay its normal size and shape during surgery. 223 00:13:19.000 --> 00:13:21.200 That should eliminate this air pocket that we saw in 224 00:13:21.200 --> 00:13:24.720 the first participant that air pocket we think may have 225 00:13:24.799 --> 00:13:29.080 contributed to eating up some of the thread slack as 226 00:13:29.120 --> 00:13:32.240 the air bubble migrated to be under the implant push 227 00:13:32.279 --> 00:13:35.799 the brain away from the implant, and so that's easy 228 00:13:35.879 --> 00:13:41.639 enough to fix. Another consideration that we want to focus 229 00:13:41.679 --> 00:13:47.159 on for our upcoming participants is that the brain. Think 230 00:13:47.200 --> 00:13:51.159 of it like a really complex folded onion. It's layer 231 00:13:51.240 --> 00:13:54.879 upon layer of sheets of neurons all over the surface 232 00:13:54.919 --> 00:14:00.759 of the brain, folded into this odd looking shape. 233 00:14:00.960 --> 00:14:01.639 The folds of. 234 00:14:01.600 --> 00:14:05.159 The brain traveled down deep into the brain, and along 235 00:14:05.159 --> 00:14:08.759 with it go those onion layers of neurons. And if 236 00:14:08.840 --> 00:14:11.879 we insert very close to one of the folds where 237 00:14:12.240 --> 00:14:15.759 there may be very useful information encoded in neurons, we 238 00:14:15.799 --> 00:14:19.799 may end up traveling with our threads parallel to some 239 00:14:19.840 --> 00:14:22.000 of the layers of neurons that were most interested in 240 00:14:22.360 --> 00:14:25.919 avoiding them Entirely. To avoid that possibility, we're going to 241 00:14:25.960 --> 00:14:30.440 insert in our future participants more close to the middle 242 00:14:30.519 --> 00:14:34.720 of the apex of the folds, ensuring that we're crossing 243 00:14:35.120 --> 00:14:37.519 the layers of interest layer five of the cortex. 244 00:14:37.919 --> 00:14:42.200 I also think that it's important to highlight here those 245 00:14:42.240 --> 00:14:44.159 tiny wires that Elon mentioned. 246 00:14:44.559 --> 00:14:46.039 They're fraction of a human hair. 247 00:14:46.120 --> 00:14:50.679 They're very flexible intentionally so because brains constantly moving and 248 00:14:50.720 --> 00:14:53.039 you want the electros to be moving with the brain 249 00:14:53.399 --> 00:14:59.000 causing less of the scarring. And it's actually impossible for 250 00:14:59.159 --> 00:15:03.080 human eurosurgeon, however talented matthew Is to actually maneuver them 251 00:15:03.080 --> 00:15:05.799 by so we have a surgical robot that we built 252 00:15:05.960 --> 00:15:08.799 that can actually precisely target them in any three dimensional 253 00:15:08.840 --> 00:15:11.320 space x, y as well as z with micron level 254 00:15:11.360 --> 00:15:15.840 precision while avoiding bascoature so that you don't disrupt and 255 00:15:16.240 --> 00:15:19.360 cause immune response from happening. So we actually have the 256 00:15:19.440 --> 00:15:21.879 technology to be able to place them exactly where we 257 00:15:21.919 --> 00:15:23.360 want them in three. 258 00:15:23.679 --> 00:15:25.919 Yeah, it's truly amazing to see the surface of the 259 00:15:25.919 --> 00:15:29.320 brain after the robot had inserted all the electrodes on 260 00:15:29.399 --> 00:15:32.120 the first participant, without a drop of blood in sight. 261 00:15:33.320 --> 00:15:35.080 It's really quite an achievement. 262 00:15:36.200 --> 00:15:39.440 Yes, something that probably most people don't realize is that 263 00:15:39.480 --> 00:15:42.799 the brain appears to be sort of somewhat undifferentiated. So 264 00:15:42.799 --> 00:15:45.000 if you look at the cortex, it looks like a 265 00:15:45.000 --> 00:15:48.720 whole bunch of folds that were you know, maybe like that. 266 00:15:49.159 --> 00:15:51.559 It's not obviously just looking at to take a picture 267 00:15:51.559 --> 00:15:55.480 of the brain. That that it's the brain is highly differentiated. 268 00:15:55.519 --> 00:16:00.200 That there's you pretty much know exactly where the part 269 00:16:00.200 --> 00:16:03.039 of the brain is that controls your right hand and 270 00:16:03.080 --> 00:16:05.919 your left hand and your leg and that kind of 271 00:16:05.919 --> 00:16:11.799 thing or vision. It's it's actually quite precisely located. It's 272 00:16:11.840 --> 00:16:15.919 not some people like might might think look the brain like, oh, 273 00:16:15.960 --> 00:16:19.399 could be it could be anywhere, but actually we it 274 00:16:19.399 --> 00:16:23.000 it's it's your brain is highly differentiated, even though it 275 00:16:23.000 --> 00:16:25.399 doesn't look it. It's. 276 00:16:25.679 --> 00:16:28.360 Yeah, do you want to describe how we actually were, 277 00:16:28.639 --> 00:16:31.080 like how we identify word to drill the quen So. 278 00:16:31.440 --> 00:16:34.559 We can we can put a patient that is considering 279 00:16:34.600 --> 00:16:39.279 this implant into an f MRI so a functional magnetic 280 00:16:39.279 --> 00:16:43.240 resonance imaging machine and ask them to imagine hand movements 281 00:16:43.279 --> 00:16:45.879 that you know, because of the spinal cord injury don't happen, 282 00:16:46.360 --> 00:16:49.440 but just imagining those hand movements causes these areas of 283 00:16:49.440 --> 00:16:52.799 the brain to light up in the fMRI scanner. And 284 00:16:52.879 --> 00:16:56.200 so we have a pretty good idea based in in fact, 285 00:16:57.039 --> 00:17:01.039 for each individual participant which part their brain is going to, 286 00:17:02.720 --> 00:17:06.200 you know, respond to imagined movements of the hand. And 287 00:17:06.240 --> 00:17:08.759 so we can map those imagined movements, much as we 288 00:17:08.799 --> 00:17:12.480 all do when moving a mouse to controlling a cursor 289 00:17:12.519 --> 00:17:14.400 on a screen, even without the use of a mouse. 290 00:17:17.039 --> 00:17:19.400 Yeah, but I think it's kind of an important point that, like, 291 00:17:19.440 --> 00:17:21.839 it's not like your the part of your brain that 292 00:17:21.839 --> 00:17:24.440 controls your hand might be anywhere in the cortex. It's 293 00:17:24.440 --> 00:17:25.640 this is not the case. It's going to be in 294 00:17:25.680 --> 00:17:28.799 a very specific region, and it's going to be extremely 295 00:17:28.839 --> 00:17:31.039 common across people. Precision is key too. 296 00:17:31.640 --> 00:17:37.279 Yeah, the left handed right handed in my mind too, Like, 297 00:17:37.319 --> 00:17:39.200 if you're right handed, you want. 298 00:17:39.039 --> 00:17:40.559 The fight on the left side. 299 00:17:41.039 --> 00:17:42.920 Let the lotto side to the hand gets your commet. 300 00:17:43.400 --> 00:17:45.240 Yeah, the left side of your brain controls right right. 301 00:17:45.319 --> 00:17:46.079 Yeah, everything's crop. 302 00:17:47.839 --> 00:17:50.119 Another of the risk mitigations we're looking at in the 303 00:17:50.160 --> 00:17:54.680 future is that, you know, the implant has a certain size. 304 00:17:54.920 --> 00:17:57.759 The depth of the bottom of the implant. It's actually 305 00:17:58.240 --> 00:18:02.440 dinner than the average human skull. And so what we 306 00:18:02.480 --> 00:18:05.559 want to be able to do is control the size 307 00:18:05.559 --> 00:18:09.079 of the gap under the implant, give the threads that 308 00:18:09.720 --> 00:18:12.960 travel from the implant into the brain as much slack 309 00:18:13.000 --> 00:18:17.880 as possible. We didn't do this in the first participant 310 00:18:17.960 --> 00:18:21.160 because we didn't want to manipulate any of their tissue 311 00:18:21.160 --> 00:18:24.359 that we didn't absolutely have to in upcoming implants. 312 00:18:24.920 --> 00:18:26.640 Our plan is to sort. 313 00:18:26.519 --> 00:18:30.960 Of sculpt the surface of the skull very intentionally to 314 00:18:31.000 --> 00:18:34.079 minimize the gap under the implant, such that the bottom 315 00:18:34.119 --> 00:18:38.680 of the implant travels perfectly flush with the normal contour 316 00:18:38.720 --> 00:18:41.160 of the inner side of the skull. That will put 317 00:18:41.160 --> 00:18:44.359 the implant closer to the brain. It will eliminate some 318 00:18:44.440 --> 00:18:46.640 of the tension on the threads, and we think it 319 00:18:46.680 --> 00:18:49.680 will reduce some of the tendency of threads to retract. 320 00:18:50.920 --> 00:18:52.640 And we actually built the tool to do. 321 00:18:53.759 --> 00:18:56.359 Yeah, this is actually this is a very important detail. 322 00:18:57.920 --> 00:19:02.039 You really want the inner contraur of the skull to 323 00:19:02.079 --> 00:19:06.279 be blushed so that the implant there's there's no the 324 00:19:06.480 --> 00:19:08.720 brain doesn't want to pucker up into the into the gap. 325 00:19:08.920 --> 00:19:14.839 That's really quite a big deal. So, like like minimizing 326 00:19:14.880 --> 00:19:18.599 the air procket and the implant being flushed with the 327 00:19:19.240 --> 00:19:23.440 inside contour of the skull are two very important improvements. 328 00:19:23.640 --> 00:19:26.240 The additional benefit here is that, uh, you know, you 329 00:19:26.279 --> 00:19:28.839 do see some amount of stick up what we call 330 00:19:28.880 --> 00:19:31.119 stick ups, so you might have bumped in the head, 331 00:19:31.160 --> 00:19:33.079 but this actually eliminates it even further. 332 00:19:33.200 --> 00:19:36.440 Yeah, yeah, I mean it's like, really our goal is 333 00:19:36.440 --> 00:19:40.039 that that if you run your hand over the top 334 00:19:40.039 --> 00:19:41.839 of the skull, you don't feel any any momp, you 335 00:19:41.839 --> 00:19:45.519 don't feel any any device, and that even if someone 336 00:19:45.599 --> 00:19:48.599 was bold, you wouln't really even notice it. And uh 337 00:19:48.759 --> 00:19:51.440 and then from the inner inner contraur of the skull 338 00:19:51.440 --> 00:19:54.680 that the brain from physical standpoint, doesn't really notice that 339 00:19:54.720 --> 00:20:06.599 there's a divot in the skull because there's no dibit. Okay. 340 00:20:06.640 --> 00:20:11.160 Another aspect of the human brain that you know, obviously 341 00:20:11.200 --> 00:20:13.519 differs from any of the animals that we tested in 342 00:20:14.440 --> 00:20:16.200 is that the human brain is a lot bigger. And 343 00:20:16.279 --> 00:20:19.319 so you may not realize that that means the human 344 00:20:19.359 --> 00:20:24.440 brain moves quite a bit more than any of these 345 00:20:24.480 --> 00:20:28.160 other smaller brained creatures. And so when we open the skull, 346 00:20:29.519 --> 00:20:33.079 we see the brain travel toward and away from the 347 00:20:33.160 --> 00:20:35.799 robot about three millimeters in total, as the heart beats 348 00:20:35.799 --> 00:20:41.319 and the breathing takes place. And so that movement, you know, 349 00:20:41.599 --> 00:20:45.200 it adds a small challenge for the robot in precisely 350 00:20:45.680 --> 00:20:48.200 choosing a depth to insert each thread. It's not an 351 00:20:48.319 --> 00:20:52.160 enormous challenge, and we've already upgraded the robot's capabilities to 352 00:20:52.160 --> 00:20:57.079 be able to even more precisely target depth in even 353 00:20:57.119 --> 00:21:01.519 a very rapidly moving brain with a high amplitude of movement. 354 00:21:05.079 --> 00:21:07.839 You may think the most obvious mitigation for threads that 355 00:21:07.880 --> 00:21:09.720 pulled out of the brain is to insert them deeper. 356 00:21:10.000 --> 00:21:14.480 We think so too, and so we're going to broaden 357 00:21:14.559 --> 00:21:18.160 the range of depths at which we insert threads. So 358 00:21:19.119 --> 00:21:22.000 for the very first participant, we had an enormous amount 359 00:21:22.000 --> 00:21:25.640 of data from our animal work, and we had very 360 00:21:25.680 --> 00:21:31.000 highly optimized our insertion depth to maximize the crossing of 361 00:21:31.160 --> 00:21:34.240 the layers of interest in the cortex with the electrodes 362 00:21:34.240 --> 00:21:38.680 that we're recording from. Now that we know retraction is 363 00:21:38.720 --> 00:21:43.039 a possibility, we're going to insert at a variety of 364 00:21:43.039 --> 00:21:46.680 depths that even in several cases of different amounts of 365 00:21:46.720 --> 00:21:50.319 retracting threads, we're going to have electrodes at the proper 366 00:21:50.359 --> 00:21:54.359 depth and with the deepest threads be able to track 367 00:21:55.000 --> 00:21:57.799 how much retraction has occurred across the surface of the 368 00:21:57.839 --> 00:22:01.720 brain from each thread. And so we're going to you know, 369 00:22:01.799 --> 00:22:04.880 both have more threads in the right layer and have 370 00:22:05.000 --> 00:22:07.240 better data on how much retraction has occurred. 371 00:22:07.920 --> 00:22:09.559 If you're a pc I NERD. You might know that 372 00:22:09.759 --> 00:22:12.720 being able to control individual z depth per thread it 373 00:22:12.799 --> 00:22:16.720 is not something that most neural interface devices offer. Most 374 00:22:16.759 --> 00:22:19.640 neural interface devices are kind of a static, fixed rigid 375 00:22:19.720 --> 00:22:21.519 array that you push in and all the lectures are 376 00:22:21.559 --> 00:22:22.000 on depth. 377 00:22:22.200 --> 00:22:22.440 Right. 378 00:22:22.720 --> 00:22:24.799 To be able to do this is actually pretty pretty 379 00:22:24.839 --> 00:22:25.920 novel part of the robots. 380 00:22:26.920 --> 00:22:30.039 Yeah, the historical approach is to actually pound in a 381 00:22:30.119 --> 00:22:32.079 sort of bed of nails with an air hammer into. 382 00:22:31.920 --> 00:22:35.160 The Look it looks crazy that that is, Yeah, just 383 00:22:36.480 --> 00:22:39.680 with a with a aumuntic hammer. That's the that's what 384 00:22:39.839 --> 00:22:41.839 this is. It sounds come somewhat by braic. This is 385 00:22:41.880 --> 00:22:44.480 not what we do, but this is the what's been 386 00:22:44.480 --> 00:22:47.799 done before. Is that's literally just hammering and what looks 387 00:22:47.839 --> 00:22:50.319 like a better nails with the brain, which actually works. 388 00:22:50.359 --> 00:22:51.200 It's astonishing that. 389 00:22:53.279 --> 00:22:55.319 Some people like man like DBS props you're just sticking 390 00:22:55.400 --> 00:22:56.519 in by hand. 391 00:22:58.160 --> 00:23:01.559 But those are several several or is a magnitude more 392 00:23:01.640 --> 00:23:03.720 volume of brain tissue that you're destroying. 393 00:23:03.359 --> 00:23:04.400 Compared to what we're doing. 394 00:23:04.559 --> 00:23:06.880 But that deep brain simulation stuff does actually work and 395 00:23:06.920 --> 00:23:10.720 it actually helps people a lot. Yeah, Yeah, that's a 396 00:23:10.720 --> 00:23:13.960 great product. Yeah, I mean I think we'll be able 397 00:23:14.000 --> 00:23:16.480 to do a much more finessed version of that down 398 00:23:16.519 --> 00:23:22.680 the road. So I mean it's really difficult. Like the 399 00:23:23.039 --> 00:23:26.359 neuraling device is something that really absolutely minimizes damage to 400 00:23:26.400 --> 00:23:30.960 the brain, absolutely minimizes the load on the patient, and 401 00:23:31.200 --> 00:23:32.960