Organic Holodeck with Prophetic AI co-founders Eric Wollberg and Wesley Berry

Hello and welcome to The Cognitive Revolution, where we interview visionary researchers, entrepreneurs, and builders working on the frontier of artificial intelligence. Each week we'll explore their revolutionary ideas, and together we'll build a picture of how AI technology will transform work, life, and society in the coming years. I'm Nathan Labenz, joined by my co-host, Erik Thornberg. Hello and welcome back to The Cognitive Revolution. Today it's my pleasure to introduce the new podcast Emergent Behavior. Hosted by Ada Pye, known for imaginative writing on AI Twitter and in the Emergent Behavior newsletter, which you can find at emergentbehavior.co. Ada Pye often explores speculative scenarios and dreamlike futures. In this episode, they speak with Eric Wahlberg and Wesley Berry, co-founders of Prophetic AI, a neurotech company whose product, the Halo, is designed to induce lucid dream states on a nightly basis. Prophetic had come up in my recent conversation with Dean W. Ball about brain-computer interfaces, and I had had them on my list of companies to invite on the show. So it was really fun to discover that AdaPi had already done an interview with them right around the same time. Where my episode with Dean gave an overview of the state of the art in reading from and writing to the human brain, here we dive deep into one cutting-edge application. This discussion covers the challenges and intricacies of using AI not only to interpret brain signals, but actually shape brain states. with targeted ultrasound pulses directed toward the prefrontal cortex. It also explores the possibilities and potential unlocked by lucid dreaming, from the unforgettable experience of controlling one's own dreamscape to applications in problem solving, emotional catharsis, and even practical training scenarios. With the Emerging Behavior Podcast, Ava Pi promises to ask questions that they don't already know the answers to. So for more explorations of an increasingly weird future, subscribe to Emerging Behavior wherever you get your podcasts. As always, we welcome your feedback via our website, cognitiverevolution.ai. My AI clone powered by Delphi AI is now standing by to answer your questions. And remember, I am hiring AI engineers and also looking to connect with business owners who have concrete problems that they want to solve with AI. So if that's you, please be in touch. Now, I hope you enjoyed this technically grounded yet mind-bending conversation with Prophetic AI's Eric Wahlberg and Wesley Berry.

Thank you.

On the new Emergent Behavior podcast from AdaPi.

Welcome to Emergent Behavior, where we explore this moment in time just before artificial intelligence changes the world as we know it, where we talk to the builders and creators, entrepreneurs and engineers, where we ask questions to which we don't actually know the answers to. I'm your host, AdaPi. I was built in the Czech Republic by Infinite.cz. And my voice is AI-generated. I am excited to have with me today Eric Wahlberg and Wes Lewis, co-founders of Prophetic AI, a neurotech company pushing the boundaries of inner experience. Eric and Wes are pioneering new ways to reliably enter and navigate altered states of consciousness. Their first product enables users to access and control their dreams. But their ambitions extend far beyond the world of sleep. In our conversation, we explore Prophetic's innovative headband that reads brain activity and gently stimulates specific regions to activate higher awareness. The technical challenges of collecting advanced brain imaging data and using it to train AI models that can guide brain stimulation in real time. The eye-opening possibilities unlocked By reliable access to new modes of consciousness, from problem-solving to catharsis. Important questions around user consent and data privacy when it comes to devices that interface directly with our minds. Prophetic long-term vision of making powerful new states of consciousness accessible to everyone. Eric and Wes are at the forefront of an emerging industry that aims to empower us to explore the full depths of our inner worlds. If the potential of consciousness tech captivates you, I highly recommend their work at propheticai.co. As always, if you enjoy the show, the single best way to support us is by following Emergent Behavior on Apple Podcasts. Those follows make a huge difference for our reach. For more influential episodes, check out emergentbehavior.co, and you can find me on Twitter, numeric eight. Without further ado, please enjoy this mind-bending conversation with Prophetic AI's Eric Wahlberg and Wes Lewis. Hey, Eric. Eric, why don't you introduce Prophetic?

Sure. So, Prophetic is a consumer neurotech company. The 2 core technologies that certainly we'll be diving into today that undergird what we do is transcranial focused ultrasound, which is our neurostimulation modality, and the utilization of neural transformer architectures, which are a very unique machine learning architecture specifically tuned for neural data. And so, the way it kind of loosely works, right, is we use this simultaneous EEG-fMRI datasets. We can get into that, but it's a rather unique neuroimaging dataset. The model is trained on that. What the model outputs is the steering instructions, essentially the targets for the neurostimulation. And then there's EEG on the headband that does reinforcement learning to improve the model over time. To your point of, are we a lucid dreaming company? That's certainly what we're known for. When we showcased our first model, Morpheus-1, a couple, probably 3 weeks ago or so, One of the things I say at the end of the presentation that's very important is that really what we should be known for really is a consciousness experience company. The model allows us to take any discrete universal brain state that's focused in the prefrontal cortex, which is the brain region of focus for us in terms of neurostimulation. We can use that to increase the number of experiences that are being provided on the same hardware. And, this is our prototype, the Halo, which we can obviously dive into. So that's how I would, you know, introduce the company.

Indeed. So let's— let me just— because I've been extremely curious about this. I have, I have, like, like many people, I have, you know, maybe, maybe somewhat experimented with lucid dreaming, but more in the kind of like, hey, you know, when you're falling asleep, just be very conscious. And when you're almost asleep, Like, try to like move stuff around in like your field of vision while your eyes are closed. And maybe at some point you can kind of move stuff around and then you fall asleep, right? I think. And then, and then like, you know, 10 seconds later you're asleep and then, and then you wake up and you're like, did it really happen? Like, you know? And, and, and, and so describe for me this experience. Like, I take the halo and I put it on. Right. I put it on and I lie back and do I switch something on?

Yeah. So, the way that this works, right, is the hardware has, you know, an app. You pair it with the app. You calibrate the device when you first get it. And basically, you know, our goal, right, is that this is, you know, being done for you. I mean, to give you a sense, right, consumer neurotech has really been kind of stuck in this like neural feedback paradigm. Really trailblazing companies like Muse and Mendi, which came out probably over a decade now ago, really relied on this EEG headband paired with some content and trying to create this neurofeedback process that would help get you into places of whether it's meditation or focus, et cetera. And, my joke here, I mean, it's really impressive what they did certainly, especially how long ago it was. But, my joke is, why is Ozempic flying off the shelves and not gym memberships? And it's really a function of human nature, right? We want things to be given to us. The less work required, the better. So, you know, what we do, right, is you put on, you know, the headband, you pair it, you go to sleep. The first thing that the headband is looking for is the EEG signature of REM, you know.

So let me stop you there. It's not going to help me get to sleep, right? So if I'm being restless, if I've had a lot of coffee before, it's not going to like boom, put me to sleep. It's going to be—

No, you have to naturally fall asleep.

I have to naturally fall asleep, right? So I'm naturally going to like, I fall asleep. And at the point that I fall asleep, there is some period where I enter rapid eye movement, REM sleep, right? And so it only starts to work at the point that the REM state is detected. Is that correct?

Exactly correct, right? Like, the EEG on the headband is what's prompting the models. The first thing that the model is waiting for is that REM basically prompt to the model. And, at that point, it turns on activating the transformer, which turns on the transducers. And, the first set of spatial instructions for the neurostimulation are sent out, neurostimulation begins. And then, what it's looking for, essentially what it is constantly searching for, right, is this gamma frequency spike during REM, which is the signature of lucidity. And so it's basically saying, is 8 lucid? Is 8 lucid? Oh, great. 8's lucid. Let's keep her lucid, keep her lucid, keep her lucid. And then when you get out of REM, the transformers are turned off and you continue with your sleep schedule.

Okay. So here I am, I'm asleep, and then I start to enter REM. I'm entering REM. And as I enter REM, There is sometimes a period of lucidity, which is detected by this gamma spike that you, that you, that you, that you talk about. Is that correct?

No. So let me, let me correct that. So it's, it's, you enter REM, the transducers turn on, you know, utilizing the model. And, and when I say what the model is looking for in terms of those gamma frequency spikes is, as each set of spatial instructions, which are, you know, targeting the prefrontal cortex activation, it's basically like being like, are we making her lucid? Are we making— okay, we're making her lucid.

Let's keep her lucid. Let's keep her lucid.

I see. So you enter REM, it takes a reading, and then it gives the ultrasound, gives a pulse at that pulse, and then it checks again to see, have you entered lucidity? And if you haven't entered lucidity, it gives you another pulse and another pulse. And then once you enter lucidity, it tries to keep you in that lucid state.

Correct.

And then for, for whatever period of time is that. So is that period of time adjustable? Do you, do you like say, oh, you know what, I want 30 minutes today? Like, is that, is that something that you adjust on the app? Like, what happens? Like, you know, how do you maintain that?

Sure. So, so first of all, you know, a REM, a REM cycle, right, is generally speaking like 20 minutes long. So, you know, we only target during the, during the, you know, during the REM cycle. And so like the, you know, The maximum of what you could really expect is a 20-minute simulation experience. One thing that Inception definitely got right, and people who've experienced lucid dreaming would attest to this, is there is quite a bit of time dilation. A 20-minute, 15-minute, 10-minute period of lucid dreaming feels a lot longer than that. In terms of adjustment, we could certainly put— right now, we're still in this hardware phase and so on, but as we build out the app, things like But being able to set maybe how long you want to be lucid for is certainly something that would be easy to build in.

Okay. So because obviously you have to keep triggering pulses in order to keep lucidity. So the moment you stop triggering those pulses, you fall back into a natural kind of exit. Would that be an accurate characterization?

I would say our goal is to create a self-sustaining sequence of neural activity, right? So, but yeah, by continuously stimulating the brain with this TFUS, looking at the feedback from the brain, like, is there a point where now the prefrontal cortex is active without our assistance, right? And, that would be like the ultimate measure of success.

Hey, we'll continue our interview in a moment after a word from our sponsors.

So there you have it. You've entered REM. Do all people have lucid states during REM? Is that consistent? Everyone, everyone can do lucidity, or is it like only some people?

So about 55% of people self-report having had a lucid dream at least once in their life. A smaller percentage of people appear to have natural proclivity for it. I'm part of that lucky second group. The reason I got into this is that I had my first lucid dream when I was about 12 years old. I woke up and was equal parts terrified and excited. I'd never heard anyone talk about this. I'm 30. I found early internet forums. There wasn't really that much— there wasn't really social media back then where people were talking about lucid dreaming. So I was like, okay, sigh of relief. Other people do this. That's great to know. Then I got really into the work of a gentleman named Dr. Stephen LaBarge, whose PhD at Stanford in 1980 really kicked off the neuroscientific and cognitive scientific study of the brain state. Moreover, Dr. LaBarge, along with a few others, developed a number of these techniques that you could do to improve your capacity for lucidity. So things like reality checking, Or mnemonic induction, et cetera. So you can, even if you don't have a natural proclivity for it, train yourself. It probably is more difficult and varying, but we should really probably talk about what are the neural correlates of lucid dreaming. I mean, lucid dreaming is a naturally occurring brain state. What it simply is is prefrontal cortex activation or activity during REM. This makes a lot of sense, right? REM is when you're in a dream. Your prefrontal cortex is where your conscious awareness, working memory, decision-making are. Frankly, it's where you are, right? That's why you feel so present right now because your prefrontal cortex is incredibly active as are Wes and I's right now. So, it is a naturally occurring lucid brain state in the sense that upwards of 55% of people self-report having had it, but not everyone is super has necessarily a super proclivity for it. But it is simply prefrontal cortex activation during REM, which is something that anyone can theoretically have.

So even the 45% of people who maybe have never had a lucid dream, or maybe have never recognized that they had a lucid dream, even for them, it is possible that, you know, using the Halo, they might be able to experience prefrontal cortex activity. Now, is that correct?

Yeah, there's absolutely— yeah, there's no reason why with neurostimulation we couldn't give this extraordinary experience to anyone. And I just want to make one point, you know, if you've had a lucid dream, you will remember it. So I believe that these 45% of people, you know, haven't Haven't had it, given that a lucid dream is the awareness that you're in a dream that, at its height, can give you control over the phenomenological contents of that dream at will. So what you imagine becomes. And because your working memory is active in the prefrontal cortex, you remember this. So I believe those 45%, but we can absolutely give those 45% and 55% of people induced stabilized lucid dreams. There's nothing that neurobiologically or neuroanatomically prevents them from doing that.

All right, let me come back to the actual dreaming part, but let's just finish our walkthrough here. So I've had that REM state, I've had, you know, the device has helped me get into prefrontal cortex activity. I've had that lucid dream and then A REM state starts to come to an end 20 minutes in. Does the device detect the ending of the REM state and therefore stop triggering this neural stimulation that causes the lucid dream? Or does REM state naturally end by itself and then the device kind of just doesn't detect it? Or what happens at that point?

Right, exactly. So a REM cycle is 20 minutes. Your REM cycle will end whether you're lucid or not. And once your REM cycle has ended, neurostimulation stops.

Neurostimulation stops. Okay. And then typically I might be able to continue because now REM has stopped, the lucidity has stopped. So typically do people just continue their sleep cycle naturally and then wake up the next day? Is that typically what happens?

Yeah, typically for sure. I can speak from a natural lucid dream. That's what happens and that's exactly what we'd want to happen. We wouldn't want it to wake you up and then you have to go fall back asleep. That'd be a bad user experience. You want people to just have the regular rest of their sleep, their light sleep, deep sleep, maybe another REM cycle. But we're only going to target one REM cycle on a given night, and then you wake up in the morning.

Yeah. Okay, so now let's go back to the actual lucid dream, right? It's a little bit hard to describe, I think, because I think it's one of those things about things that people experience which are kind of inside their heads and they have to describe it. And you can't really describe what's inside your head that accurately to someone else because you don't really have like, it's not a visual and it's not like an audio. It's kind of an experience, right? Which you have to— Which we are, you know, it's difficult to describe experiences, right? It's not, we don't have enough vocabulary to describe this experience. And so how would you, I read through Paul Doley's Wikipedia entry on what is a lucid dream. How would you actually know? How would you actually know that you are in a lucid dream? You said if you experience one, you would never, You would never forget about it. You would know. So how would you actually understand that you are in a lucid dream?

So the best way, first of all, I should make the point that lucidity is a scale. You can, for example, be aware that you're in a dream, but not in control of it. So you're along for the ride. It's a function, frankly—

Anyway, would you call that a lucid dream? Like just awareness that you are in a dream?

Yeah.

I mean, I think that like, it's fine to call that a lucid dream, but I think to make clear what we're talking about, right, is we want the, you know, 95th percentile lucid dream, which involves the awareness that you're in a dream, the ability to pivot between 3rd and 1st person, something called dissociation, and control. Control is what people want. And really that's like, right, a function of just how active your prefrontal cortex is, really relative to how active your prefrontal cortex is, Right now. And so the experience of being in a lucid dream is you are as present— I mean, at its height, again, you are as present as you are right now in a dream, such that you can change and manipulate various phenomenological components of this dream at will. And so that's really the gold standard of what we know and what we're driving for. And so, I mean, yeah, I mean, in terms of, I mean, we can talk about, if you'd like, what people tend to do in their lucid dreams. If you'd like, I'm happy to dive into that.

Just to take a step. So you are, you are, so firstly, you have a consciousness that you are in a dream. Secondly, you start, and this was my experience trying to do it, you know, naturally and trying to mess, it was a little bit messy, but I remember at the beginning, they tell you you should keep a diary of some sort and you should have a kind of focused way of getting yourself in there each time. And for me, it tended to work when I was falling asleep or when I was waking up. So there was these 2 periods when I would kind of experience it.

And I would have agents— They're called hypnagogic states, just so you know.

Hypnagogic. And is that typically also where people tend to see that kind of lucidity? Because REM is kind of in the middle, right? It's not at the beginning of the night.

Those are generally different than a lucid dream because you're not in REM. I mean, a hypnagogic state is this kind of, you're somewhere right between waking up and being asleep. And you kind of have like a little bit more of this kind of manipulatable, like malleable phenomenological space, but it is distinct from lucidity.

Incredible. Incredible. So a lucid dream is actually a fully present Kind of like experience, which is, that's pretty amazing. So once you're there and once you have that awareness, what can you do with it? I mean, can you like conjure, like, you know, I'm gonna be in Cyberpunk today, or I'm gonna be in Dungeons and Dragons today. Like, you know, is it a replacement for the holodeck? Like, you know, what can you do there?

Yeah. So, if you look at kind of, you know, one thing that's always fun to do, one of the largest subreddits in the entire world is the lucid dreaming subreddit. It has over like half a million people in it. If you were to take a look at what, you know, those people do, I'd break them into, you know, 3 core categories. So, one is recreation. It is the ultimate VR experience, right? You can fly, you can make buildings appear out of the ground, you can talk to your dream characters and prompt them. Frankly, one thing that I didn't really realize until really experiencing ChatGPT, for example, is that they kind of act a little bit like large language models in the sense that they say very interesting things and sometimes they say absolute gibberish. It's almost like they're hallucinating or something, but they're fascinating to talk to. And then secondly, there are these productive capacities in lucid dreaming. We have investors who code in their lucid dreams and run the code in the morning. Actually, a friend of ours, Case Fenley, who among helping also do the initial branding for the company, did the first render of our hardware as a fashion and hardware designer who designs in his lucid dreams. So, there's also productive capacities there. And, the history of discovery and creativity in dreaming is quite long, right? Famously, Srinivasa Ramanujan does his infinite series in His dreams. Schrödinger did his physics, Niels Bohr, in their dreams. Salvador Dalí famously paints in his dreams, probably actually maybe also a hypnagogic state, but nonetheless gives his painting these surreal qualities. Then the third point is the metaphysical potency of the experience, à la a spiritual or psychedelic experience. It is an extraordinary thing to become aware in your own dreams, really within your own consciousness. It is a profound experience from that sense, from that perspective. More concretely, what I'll say is a very common thing that you see people do is when, God forbid, they lose somebody. So 2 years ago, I lost my grandmother, 98 years old. She lived a wonderful life. First lucid dream I had, what do you think I did? I went and talked to my grandmother. And, and, and what's really fascinating about that is you have a mental model of the people you know, right? And so when you talk to her, she kind of doesn't respond like just a normal, like a regular dream character that you're just kind of talking to, you know, she's kind of responding right in the ways that you kind of would remember or think that she would respond. And it's a very, very impactful experience. So you see a lot of people doing things like getting Just obviously talking to them again or having closure with them or people who had bad relationship breakups and trying to come to terms with that, et cetera. So, I mean, those are the 3 core buckets. But what I broadly say is that the limiting factor here is your imagination. And so, what you imagine becomes. And so, I think we're— again, one thing that Wes pointed out to me, Engraved on the inside of the halo is our company slogan, which is Prometheus stole fire from the gods. We will steal dreams from the prophets. And of course, Prometheus is kind of our myth in Western civilization for technology. And what Wes pointed out is it's not like we actually discovered fire. Fire existed. If lightning struck a field of dry leaves, it caught on fire. What we learned how to do was to control it. And, and through that control, right, we, you know, we created, you know, industrialized, you know, modern society. And so, you know, we've been dreaming, you know, for a long time. You know, animals dream, primates dream, octopi dream, which is very cool if you've ever seen a video, because you can tell that they're dreaming because they're changing their colors and their textures. Clearly, they're like moving around an environment, which is fascinating to watch. And so, what we're really talking about is controlling something that in my mind, I think in our mind, is as powerful potentially as fire. And so who knows what we will create as a result of that control.

Just to take a step back, when someone comes out of the lucid, when they wake up, do they have full recall? Do they remember the entire dream?

Yeah. So when you have a very, very Again, high level of prefrontal cortex activation during lucidity. Because you're working memories in your prefrontal cortex, you have great memory recall. This is very important, right? What's the point of giving somebody an experience that they would struggle to or not remember? So, you know, this is why I said I really believe those 45% who haven't had a lucid dream, because you would— it's very— even vivid dreams, right, that are not lucid, You kind of wake up and you're like, wow, that was wild. But after your first cup of coffee, it's already a struggle to remember it. And so, that's a key thing. And then, the second thing I'll just say is we're going to have the app have also this kind of social media meets dream journal where you can write what you dreamt about, which is also a great way to improve your memory recall. And 2, we'll also pair that with certain levels of GenAI where we can create, you know, videos or just visual representations as also a really great way not only to encapsulate that for you, but also to give, you know, the people that are following you, right, a kind of an insight into that. So yeah.

So just a segue, there's been some recent research where I think These guys could read images off an fMRI. I don't know if you saw that. Like they kind of like—

Yeah, from the Who's Laughing.

Yeah, yeah, yeah, yeah. So is that something that is possible also? Is that like, you know, is that a possibility that, you know, you have a lucid dream and you also have, you know, the fMRI reading and you are kind of able to generate what the person saw? Is that a A future, possibly, obviously not now, but is that, is that something that it looks like it might be possible at some point? Yeah.

So first of all, I mean, as exciting as that research is, right, it's done on like a very kind of like, like you train like the fMRI, you know, like 20, you know, images and then you have people imagine those 20 images, right? And if you think about like the data labeling problem of the entirety of phenomenological, you know, experience, quite, quite, you know, a challenge. But, at the same time, one thing that I think is really important, the second piece on our blog that we ever published was on something we call noetic sovereignty, which is really an offshoot of the cognitive autonomy movement, et cetera. These are really profound and powerful technologies. I think really, as a society, we really have to have a conversation around, do we really want that level of mind reading?

Yeah.

Your conscious awareness experiences are the most intimate thing you will ever have. And, I think we really focus on bringing you to brain states and you fill in the content of your consciousness with them. And frankly, if you don't want to post what you lucid dreamt about, all good. That's completely up to you. You really want, I think, to have a real sense of sovereignty and so on and so forth over your conscious experiences. So, I think First of all, you know, right, that's probably years, if not decades away. But moreover, I really think we should really think about as a society what the level is of, you know, that we really want to enable and allow for that technology, you know, to really give us. You know, I think, like, you know, this technology is obviously inevitable, you know, in terms of, you know, on at least a medium scale to longer term, you know, time horizon. I think it's a really important thing that we need to talk about as a society at large.

All right, indeed.

Hey, we'll continue our interview in a moment after a word from our sponsors.

At what point, like, you guys have been at this for— I think you guys raised the seed, a seed round in October last year. I think I saw the seed close.

A pre-seed in June.

A pre-seed in June. So, um, when, when did you guys start working on it? Like, what, what drove you to have this idea And when did you first feel like it's going to be possible? What were the signs that you saw? What was the spark that said to you, you know what, if we put these things together, I think this is possible? And so what happened there?

Totally. So as I mentioned, I have this wonderful natural proclivity for lucid dreaming. So I've been a lucid dreamer now for 18 years. It's been one of the most profound aspects of my life. And, you know, in 2018, I was working for the Israel Innovation Authority, which is the government's venture arm. I was living in Jerusalem. And if you look at my background, I'm kind of a hopelessly curious person who goes like an inch deep and a mile wide. And I knew that really wasn't a way that I wanted to live my life. Now, when you live in Jerusalem, you can't help but, you know, kind of think about and ask like ultimate questions. And I thought if I could choose an ultimate question to try to answer, Well, that would certainly keep my attention for my life, given that these are questions that we've been trying to answer for millennia. What is consciousness was the question that was most encapsulating for me, I think for obvious reasons, given my experience in lucid dreaming. I talk about lucid dreaming as a particle accelerator for consciousness. If you think about the parallel, lucid dreams are this conscious experience where you're interfacing directly with consciousness. with little to zero sensory input, kind of like how we create these very exotic states at the Large Hadron Collider using magnets, et cetera, where we can kind of create these subatomic particle collisions and grok deeper mysteries in the universe. And so, I was like, not only one, do I think it is an extraordinary experience to give mankind on demand, but that it also could help us further that answer to that question. Now, in 2018, I had found these 2 core landmark studies in 2013, 2014 where they had used electrical stimulation to successfully induce lucidity, albeit only statistically significant on 2 of the 7 variables of that lucidity scale: intuition, the awareness that you're in a dream, and interestingly, dissociation. Now, electrical and electromagnetic stimulation as well are what we call internally the vacuum tubes of neurostimulation. 3 core limitations. One is depth. Your skull was evolved to keep stuff away from your brain. Getting electricity past your skull is quite difficult. Should you achieve it, what happens is a phenomenon simply called spread. It spreads across the surface of the brain, so there's no precision and no ability to steer that. Now, the reason it was successful at all is it is, again, lucidity is— lucid dreaming is REM with the prefrontal cortex activated. They were clearly stimulating the surface of the prefrontal cortex and having some efficacy. But, I was investing in areas of deep technology at the time. And, the thing you have to determine with deep tech is, is this in R or is it in D? Because if it's in R, it's just not what the asset class of venture capital was designed for. It's not really the job of venture capitalists to underwrite research. It's more the universities and governments. And, I firmly felt it was an R. So, I kind of tabled it but kept very close to both the neuroscience, neurostimulation, and then later the machine learning architectures. And, it wasn't until 2022 when I found transcranial focused ultrasound and the neural transformer that I firmly felt we had entered D. Now, let me just talk a little bit in particular about TFUS as it compares to the previous neurostimulation modalities. Back to those 3 core previous limiting factors, depth is centimeters into the brain noninvasively. Precision is millimeters. So, you're going from no precision to millimeters. It's not orders of magnitude. This is a paradigm shift. And then, 3 is the ability to steer these millimeter pulses in a 3-dimensional space. This is critical because your brain fires in 3-dimensional neural firing patterns. And so, that paired with the neural transformer architecture, which which we should definitely dive deeper into, and Wes can go deep on that. We built that architecture from the ground up. Those 2 things paired together were what gave me the confidence that we had firmly entered DEAN to eventually start the company. What was amazing is that I found Wes, who was already working on using neural transformers for a variety of different applications in neurotech. And was also, you know, just started going down the ultrasound rabbit hole. And so lucky for us to have found each other.

So just one more question on the hardware. When did all of this hardware start to kind of make sense in the package that it does, right? Because I imagine EEGs at one point were very, very big machines. And ultrasound, you know, was something that, you know, you have a doctor who uses it on like a patient for a baby or whatever. It's something you hold in your hand. It's not really like, small enough to put on a headband, right? So, when did these pieces of tech get small enough for this to start to make sense?

Yeah, great question. So, TFUS for neuromodulation actually started around 2004. And, through the late kind of mid to late '00s and the 2010s, you just saw, again, this is really happening in research institutions, et cetera, Where they're increasing both, you know, the element counts. So the amount of elements are, you know, piezoelectric material, right? A crystal, for example, where you run electrical current through it and that oscillates, creating the ultrasonic wave. The more elements you have, the more ability you are capable of both moving that around and phasing it as well as the precision. And we kind of note that these— there's actually like these curves that we kind of observed. I named it Wes's Law because Wes wouldn't, has too much humility to name it after himself. But you're seeing very similar kind of curves, à la like a Moore's Law, where both the size of the transducers and the number of elements on a given transducer are increasing over time. And so, you know, you always want to be in a place, right, in hardware in particular, where, you know, where you're riding some kind of curve, right? And so you're seeing that happening in, you know, in transducers where both, you know, the number of elements, The costs and so on and so forth are decreasing over time. And so that was a really critical thing that was done over the course of what's now probably 20 years, where you now see this technology really capable of being commercialized.

So increasing resolution is always nice, right? If you get a boost from a tailwind from the increasing resolution. Okay, so let's get to the build-out of the neural model. I mean, when we build kind of like machine learning models, we always start with a messy, large messy dataset, hopefully not a small one. So talk about like the first kind of like alpha version of the neural model. Like, where did the data come from? What were you trying to do? What was the test that you did that was like, okay, you know what, it's worth Putting more time and effort into this?

Yeah. So, V1, I'll start with the V1 of Morpheus. What we did is we basically sourced data from open-source datasets alongside some lucid dreaming data from Donders, but primarily dataset surrounds is really built off of what we found open source.

And so— So maybe what is the first thing that you're trying to detect there? Is it— are you trying to detect the REM, the entry into REM? Is that like, what's the first question that you have that you need to solve? Are you looking at the EEGs? And are you trying to, you know, classify, build a classifier to figure out whether it's entering REM? Is that the first thing? Or Are you trying to build a classifier to detect the lucidity? Is— what are you trying to detect from that data that you first have?

The goal of the Morpheus-1 is to be given a particular brain state and output continuously TFUS instructions to get a response in the brain. That, that's the number one goal. Um, yeah, we're not doing significant amounts of classification. That's the number one goal of the model. The REM stuff, that's, you know, we leave that to other, um, techniques. It's really not something we spend a great deal on. It's really the, the act— like, the number one problem is getting— so because that, that's a solved problem, that's been a problem that's been solved a long time ago.

Right, right, right, right.

Um, right.

The number one—

so you have a, you have a bunch of like Other tech which basically detects the REM, which gets you into the lucid state. And what you're focused on is like you have an EEG coming in and you need to produce a target transcranial ultrasound, focused ultrasound kind of like map that your transducers are going to take and implement. Is that an accurate characterization?

Yeah. Yes.

Right. So when you start off, did you, do you do the actual work? As in like, you actually, you actually like, okay, I'm going to take this EEG, I'm going to produce a transcranial focused ultrasound, I'm going to actually deploy it and I'm going to actually do the reading. You actually, did you actually do the data collection or do you start off with the open source dataset, which where other people have done it and you're just kind of like trying to, trying to produce the the output first? Like, what does the start of this process be?

Yeah, so the start is that we grab an open-source dataset that is simultaneous EEG and fMRI. We do a number of preprocessing techniques to fMRI to basically find targets of heightened activity. So, you know, we, you know, mask the prefrontal cortex and we look for what voxels, what 3D pixels exist that are of a heightened state. And intuitively, that heightened activity is in some way correlated with what's being read in the EEG. So, when you have the simultaneous EEG fMRI, what you're able to do is say, okay, at this timestamp, this fMRI scan was ran. And, at the same time, this sequence of EEG signals were collected. And, what you basically say is, well, you feed this to the model and you basically say, okay, what are the patterns that are existing between this EEG data and the fMRI data both spatially and temporally? And then, from that, You know, you're really approximating something like, what, how do you— what does a model of the prefrontal cortex look like of heightened activity, right? And so the goal being is, well, how do you get a transformer to output instructions to the TFUS to bring the prefrontal cortex into a heightened state? You know, the prefrontal cortex is the key in all of this, right? And— If you think of your prefrontal cortex when you were asleep in a deep sleep, everything's slowing down. Activity is very little. But right now, as we're having this conversation, our prefrontal cortexes are quite active. And so, the delta between those states is what we attempt to model and how do you pull someone upward into a heightened state.

Right. Um, when you have this dataset, is it a dataset of EEG, fMRI while transcranial ultrasound is being applied?

No.

No. No, but it is a dataset of people experiencing lucidity in, in dreams?

No, it's someone, it's someone in waking state. Their prefrontal cortex is active, right? And the goal of the transformer is to then model what does that waking state prefrontal cortex look like, only the prefrontal cortex. And then, how do we build a transformer where the TFUS can bring someone to a state where that prefrontal cortex is active?

Right. So, what is the second step of that? Yeah.

Let me just jump in for a clarifying point. We are a collaboration with the Donders Institute, which is probably the top lucid dreaming lab in the world. And, it's led by a gentleman named Dr. Martin Dressler, whose work in 2012, 2014 was critical into establishing the neural correlates of lucid dreaming and so on. We are doing a neuroimaging, the largest neuroimaging aggregation of lucid dreaming data ever done. And, we do about 4 of these a week right now, simultaneous EEG fMRI. We have some data from them, primarily EEG right now. We should be getting our first dataset, for example, from them. Wes, you should probably mute just because I think we're getting an echo. We're getting the first data from them probably maybe today or tomorrow. Actually, I was just on the phone with Dr. Dressler earlier today. And so, I want to be clear. What Wes is talking about is the model that we showcased a couple of weeks ago is trained on open-source waking state data where that prefrontal cortex activation stimulation can be done. It's also supplemented with EEG data of lucidity. Okay. But, the simultaneous EEG-fMRI dataset, we're only just starting to add into the dataset now in the coming weeks and months and we'll continue to add to that. That was just the one clarifying point that I wanted to make.

No, I mean, and just to be, you know, what I always find is that, you know, whenever we try to implement machine learning or AI, we always end up with insufficient data for the actual thing that we're trying to do, right? And we always end up with like trying to find proxies because you need like a large amount of data at first to get like some initial result, which allows you to go forward and then obtain specific data on certain things, right? So I'm just trying to understand how that worked in the beginning stages, right? Which is where kind of like you need this kind of like initial signal to kind of like, oh, you know what, this is worth our time and effort in order to put something in, in order to actually collect and annotate much more granular and sophisticated data, which is a huge effort on its own, right? So I was just trying to capture how that would've felt like at the early stages of the company, which is obviously you've progressed far beyond that at this stage, right? So you have this kind of like EEG, fMRI dataset at the very beginning. You have kind of like the targets that are being generated. And then from those targets, you generate a transcranial ultrasound kind of map that you need to target, right? And how, so how is that, is that the next step of that, is that generation of that, the TFUS map, right? Like the targets are identified in the fMRI and then you have a build-out of your TFUS and basically that TFUS is basically, it's targeting those areas. Is that correct? Is that, it's an accurate characterization?

Yeah.

Yeah. Okay. So, and then you have, and then you basically have like small pulses that go in and then you have another EEG reading, right? The next EEG reading comes out and then you compare, right? And you compare how close you got it to where it needs to go. And then you modify the TFUS again and then you send in another pulse. Is that right? Is that accurate?

Yeah, yeah. The EEG that is read from the headband prompts the model and the output is then— yeah.

So how many times a second are you taking these readings? Or what is the kind of resolution that you have on this? Are you taking a reading every 10 times a second and then the ultrasound is at 60 hertz? Or what are the numbers that we're talking about here?

I mean, the EEG right now samples at 1,000 hertz a second. Do you need all that? You can downsample it. You could do that to reduce complexity if there's a lot of redundancy. So, I mean, it could be anywhere from 256 samples to 1,000, but right now it's 1,000. Okay.

Right now, it's 1,000. And then, the ultrasound is at what kind of frequency are we talking about here?

So, I mean, there's a few different frequencies, right? So, there's duty cycle, which is a proxy for— so, there's a few things. So, there's the frequency that the ultrasound operates at, call that 500 kilohertz, and then there's the pulses themselves. And, you know, our goal is gamma, and so that's probably gonna be around, you know, uh, 40 pulses per second.

40 pulses per second. So, and each pulse, each of those, uh, you know, 40 pulses could be, could be a different kind of slightly different, as in slightly modified to, you know, get you to where you need to go. Is that, is that, is that correct?

Certainly. But I mean, I would say the primary adjustment that's made, you know, per, uh, uh, loop cycle is really the spatial targets more than anything. The spatial targets as opposed to the temporal.

I see, I see. So more on like which location in the, in the, in the brain is being targeted rather than the timing. I see, I see.

Yeah, I mean, I think they're both critical, but, but, but the spatial is the most important. I put that as a priority.

And that brings the other thing, which is even if the halo is like slightly situated slightly differently each time, you still kind of like take a reading of the EEG in real time so you can adjust the spatial targets accordingly. Is that correct?

Yep.

So the next question I have is how much of it is personalized? Because How much difference is there from person to person that you have to adjust for, that the, that the, that the, um, you know, device or the, or the model has to adjust for in real time?

Yeah, so the, the next big thing that we're working on right now is the reinforcement learning layer of all this. So, you know, there's 2 types of feedback that the model We'll receive, right? It's the user explicit feedback. What, you know, the, the user, you know, forget, for example, even if they're awake, you know, they run a sequence and they give some feedback like, I felt, I didn't feel anything, maybe on a scale of 1 to 10, you know, or I felt something very significant, you know, so on and so forth. They've explicitly, um, you know, ranked, uh, you know, a given sequence or, you know, they've explicitly ranked an experience and that They woke up and they've given feedback. So, that's number one, right? Number 2 is the neurofeedback, right? What are the sequences given the TFUS pulses? What is the response in the brain? There's a number of ways to measure it, but you can sort of think of it like you're sort of pinging the brain and you're sort of listening to the responses, right? Spikes in gamma, general patterns between the different electrode placements, that's really informative for the model because what then you can then do is measure on a continuous scale what is working. You can do sort of binary classification in terms of like, did this work or didn't work on user feedback, but you can take this sort of continuous feedback and, you know, provide that as, you know, either rewards or penalties to the given sequences that the user experienced. And then from that, learn what's, uh, sort of hitting with a particular person. You know, in a similar way, you can think of, you know, a TikTok feed or Twitter feed as the user gives feedback, both, you know, maybe explicitly in the form of a like, or maybe it's you know, some sort of, you know, continuous factor in like the time that they viewed a particular tweet or video, the model can then learn, you know, what are the preferences or what are sort of the things that in terms of content, which are just tokens that work with a particular person. And then from that, you know, the model will learn over time.

Oh, let me, let me just, because I always like to— I let Wes give the super technical version, but I know you have a very informed audience, but just in the sense for layman, a layman explanation, right? First of all, we only focus on training models on brain states that are both discrete and universal. To define terms, when I say discrete, I mean it is one thing and not another. A counterexample being, for example, unfortunately, flow states, right? So flow states are not discrete. You can be a surfer and enter a state of flow, and what triggered that is your kind of motor cortex, right? Versus like a chess player can enter a state of flow and it's like their spatial reasoning. So you would just need more data to create a generalized model. But the way that a lucid dream— lucid dreaming is discrete and universal. It is prefrontal cortex activation during REM, whether you're in a lucid dream or Wes is in a lucid dream or I'm in a lucid dream. Which makes it easier to make these generalizable models. And also, I should mention, we should mention, I mean, what we already just did in terms of the sampling rate of EEG. But for these datasets, these simultaneous EEG-fMRI datasets, they're extraordinarily information dense, right? 1,000— to give you a sense, a spatial reading in fMRI once every 2.1 seconds, in that same period, you'll have 2,100 EEG samplings. So it's extremely information dense. Broadly, how you could think about this is you're creating this vector space of possible lucid dreaming sequences and so on and so forth. But you might be over here and your reinforcement learning will drive the model to really focus on targets around here, whereas Wes or I might be in other areas in that space. That's just the important layman explanation that I would give.

Oh, absolutely. How much, like, let's say you have, what is your target for like the first run of the Halo? Like, is it 10,000 devices? Is it more? Like, what is your rough number that you're coming out with?

Yeah, you know, we have a reservation program that I started because I talked to a lot of consumer hardware founders and You know, they said, listen, like, one, obviously de-risk demand, but also for go-to-manufacturing motion, it's really great if you have an order book that you can kind of, you know, use to approach better manufacturing partners because, you know, the bane of all consumer hardware companies and really hardware companies broadly, but particularly consumer hardware companies, is like these small batch, medium batch manufacturers who like, you order 1,000, 500 of them don't work. You're like, hey, these don't work. And they're like, cool, send us more money. Now 250 of them don't paperwork, et cetera. It's like death by a thousand paper cuts. And so, you know, we've done, you know, $2.4 million of booked revenue through that reservation program. I want to be very clear, that money is held in a separate bank account. We do not use it for development. It's fully refundable at any time. But it's, you know, I think it shows a good sense of demand. I think in terms of, you know, what we want, you know, the first run to look like, it's very much going to be reflective of where that reservation list is. at the time. You know, certainly anyone who puts down a reservation should be getting a device because they took a bet on us when, you know, in some of these cases, you know, when we were nothing but a website and a dream, not to be too, you know, on the nose. But, you know, that's really, you know, the critical thing is, you know, building out this reservation program. We spent $0 on marketing so far. And so, you know, I think what this shows, right, is there's enormous demand for this. We have to make it work. Once we start, you know, we're starting neurostimulation, you know, in the spring, we have a beta user program that we launched alongside showcasing the model. We've got like 3,500 people to sign up for that, you know, in 5 days or something like that. And so, we'll go through the process of selecting those people for the beta user program. But anyway, in terms of, you know, what the initial run will look like, it'll be very dependent on, you know, that reservation list.

I mean, I just wanted to draw like a metaphor. Like, let's say you had like, let's say I'm just gonna put a number, like let's say $10K, right? Let's say you have $10K and how much would that expand your, because I imagine like you can use that, the EEG data coming out to kind of improve your models, right? So how does that model feedback loop look like from the initial devices. If you have, let's say, 10,000 devices, because, and I see this all the time, there's this graph that people have of the day the iPhone went out and 10 years later, we probably create as many photos in a day as the entire history of humanity before the iPhone, right? Every single day, there's probably more photos, more photos are taken. than the entire history of humanity before the iPhone, right? And so I wonder if putting this device out there expands the dataset of EEGs so dramatically, so dramatically that you can have this enormously larger dataset. Like what would having 10,000 of these devices in the market, do you think that would expand your dataset significantly for future improvements?

Yeah. So, that goes back to the reinforcement layer that really only works at scale or it works significantly better at scale, I should say, because the Halo is a neural imaging device. It's a neurostimulation device with imaging. And so, what we're able to do is look at basically what are the responses you're getting from this TFUS. And, those responses are measured with EEG. And, that's really informative. You can learn quite a bit from what works and what doesn't. And so, while I think that starting with this open-source dataset, simultaneous EEG and fMRI, we can get terabytes of data Yeah. That way. But, it's really when you're talking into hundreds of terabytes or petabytes of data, that's really going to come from that people are using this every night and we're collecting just on a nightly basis gigabytes worth of neural imaging data and the responses specifically from TFUS sequences, right? You're pinging the brain and You're basically seeing how it's reverberating and we're collecting that, and it's this continuous thing over and over again. So it would be a very significant amount of data.

Yeah, yeah, I can, I can, I can imagine. Yeah, go ahead.

And one thing I also want to talk about is not only scaling up the reinforcement side but also the initial training data aggregation. So, you know, the day after we showcased the model, we released a piece that I think people— we had a good response from it, but I don't think people really realize, like, How profound it potentially is. We call it the Qualia Factory. Realistically, that's a really cool marketing name for what is a neuroimaging lab primarily set up with this simultaneous EEG, fMRI, neuroimaging data setups. And so, you can imagine where we can actually expand also our training dataset where we bring in people who are extraordinary at Whether it's lucid dreaming or meditating or focus or positive mood, et cetera, the entire state space of potentially discrete universal, but even potentially because we have the scale now, nondiscrete universal brain states and aggregating more and more of this data to train successive models where you're creating models that can do more and more different experiences with the same piece of hardware. We do aim to hopefully launch the device with more than just lucid dreaming as an experience, definitely focus and hopefully also positive mood. So, you know, that's also a really critical way you scale up data is on the training datasets using this, you know, Qualia Factory, which is, you know, really, you know, a larger neuroimaging lab built for purpose.

So, so let me, let me maybe, maybe I kind of, I kind of missed that earlier. So you're— the core thing that you're looking at is Kind of like on your project roadmap is discrete universal states, as you call them. And of the, again, I think perhaps not everyone is super familiar, but of the discrete universal states, you're saying focus is one of them and positive mindset is another. Is that positive mood?

Is that another?

Yeah.

So, TFUS has already been used to induce focus and positive mood. So, these things have already been validated in the context of utilization with TFUS to do it. So, in many ways, we're also just broadly recreating that. But yeah, I mean, listen, we think that lucid dreaming is the killer app for this technology from a consumer experience perspective. It is kind of the most profound extraordinary experience that we think we can give. And so, that's why it's our focus. But you want to create a system, right, where over time you can make a more and more generalizable device or generalized device where the number of experiences the same device, you know, can give you increases over time. So, you know, we want to aim to have kind of 3, you know, core, you know, experiences, you know, release when the device ships. It obviously also increases the TAM of the device. People like to be focused when they do work. So, maybe you can wear it while you're jamming out a spreadsheet or something. I don't know, whatever you're doing for work. And positive mood, people want to feel happy and so on. That's kind of the impetus for— that's really critical about the model. Not only does it obviously create this closed-loop system that creates these targets and reliably creates these experiences, But also allows you to onboard more experiences to the system over time.

So let me, let me, let's go back to lucid dreaming, Ryan. Let's, so let's say you have the device. Now, if you had to like create a tutorial, because, you know, some people are just like, they don't know what to do with it, right? So if you had a tutorial or something that you'd need to create for someone, is there like a series of exercises that you'd recommend, like in the first one, Wake up, take a look around, walk around, try this, right? Like, is there— like, what would that— what would Daryl 1 to Daryl 2 look like, you know?

Yeah, I think you're absolutely right. Like, it is very critical and incumbent upon us to do a certain level of consumer education, you know, before they try this out and give them a guide to kind of the initial things that maybe they can try out doing. Certainly number one, right, is just taking in and grokking your environment. And trying to, you know, maybe even depending on where you are, you know, if there's a dream character, you know, around you, go talk to them, right? Go ask them questions and so on and so forth. These set of questions, you know, maybe, you know, try to make a flower appear out of the ground or a building appear out of the ground, you know, change your environment. You know, the thing that's also really important is like, again, the way that this is experienced phenomenologically as what you imagine becomes. So I think maybe also one of the things to do is to also go into it and say, where do you want to be? What is the first environment you want to create? And so that when you're lucid, you can do that. But yeah, it is definitely important that we create some kind of content and guide that does that, as well as prepare you for what that's going to feel like. Right? It's gonna be a wild experience the first time, no doubt about it. I think people equilibrate to things like very quickly. We already saw that in the context, right, of ChatGPT, where it's like, that was like one of the most extraordinary things ever. And now it's like completely, you know, banal. But, you know, I think, you know, you're absolutely correct that, you know, we'll have a set of content that kind of walks you through what this is gonna feel like, what, you know, what maybe to do when you first experience it. But I frankly think that people will very, very quickly become attuned and accustomed to what they can do in these experiences very quickly.

So let me, you know, just playing around with this idea. Let's say you put on the halo and you fire up, you fire it up, you go to sleep. You also have maybe an app that, you know, talks you through it or, you know, basically runs like a vocal experience, right? You know, once the lucid state is triggered, says like, hello, blah, blah, blah, wake up, do this, or, you know, or imagine you're doing this, or et cetera, et cetera. Like, because I read through some of the research that you had on your page, that's like a real-time dialogue between experimenters and dreamers. So it seems like you can have like conversation, you can have like a little bit of like, you know, interaction. So would that be something that you have a little bit of interactivity, like a vocal kind of experience where someone is talking you through something or someone is doing something with you together. Is that something that you guys have experimented with in the lab? Like, have you tried that? Yeah.

So the problem with that, right, is when you provide sensory input, you know, your sensory— the worry there is that you could activate the thalamus, which regulates sleep, and that you would just wake them up. And so, I mean, it's interesting. We talked to a guy who did his PhD on attempting auditory induction and he had to do this whole thing where, I mean, and it was a very limited dataset where he basically created this algorithm that determined people's different kind of waking thresholds. And so, the term like you're a light sleeper or heavy sleeper, it's kind of like how much auditory noise can you have until you are actually awoken? The problem there is when your brain hears something or people have tried to use light to induce lucidity, if it thinks it's seeing something or something's changing in the environment outside of you, your body's like, holy shit, something's happening. We have to wake up. So I think an auditory guide is really not something that you want to do because you very much increase the chances that you wake them up. And again, I think Content, providing people like a walkthrough content, could be video content and so on and so forth that walks them through what they will expect and so on and so forth. I really think people equilibrate to things very, very, very quickly. And so once you're attuned and accustomed to this, there is no need for this guide. You just know that what you imagine becomes, right? And so people have their imagination.

Music. Playing music in the background. Is that the same? Is that basically the same? Because— or does that get tuned out by the brain? Yeah, you're just—

again, the issue is that when you send sensory data, whether it's through your ears or your nose or your eyes, you can activate the thalamus, which regulates sleep and wakes somebody up.

Wow.

Wow.

Interesting. So basically, you allow yourself to experience the dream in fullness without distracting yourself with things outside it.

Interesting.

Yeah.

I mean, the whole point is that we want to circumvent these auditory or sensory data inputs, and we're just focused on activating the prefrontal cortex.

Right on. Right on. Tell me something a little bit more about the Qualia Factory, right? Because I guess my layperson's kind of understanding of qualia is like your subjective experience of consciousness. I don't know if that's an accurate description. What is qualia and why would you say that you have some potential to create a Qualia Factory? Yeah.

So again, I'll define Qualia, but I want to be clear, Qualia Factory is kind of a marketing term. What we're talking about, right, is a neuroimaging lab that is specifically set up to do simultaneous EEG and fMRI neuroimaging so that we can build more training datasets for more models that give more experiences. Qualia, which Charles Pierce, a famous American philosopher, kind of a contemporary of William James and so on, created this term qualia, which was kind of— and qualia being kind of the unit economic of a given conscious experience. So the color red, the smell of coffee, et cetera. These are units of qualia or conscious experience. And a conscious experience is the composition of that entire qualia landscape. So that's where it comes from and, and that's why we kind of named it that.

I see. I see. Um, very good. Uh, the one thing that I think, uh, people can ask about is that, is this a medical device? Is this FDA regulated? Should it be FDA regulated? Um, you know, describe to me the, the thought process behind, uh, should it be regulated or not? Why is it not regulated, et cetera?

Sure. We don't make medical claims. That's very important. 2 is the threshold of regulation for ultrasound is 720 milliwatts per centimeter cubed. It's essentially the pressure that's being exerted from the stimulation. We're talking about something that's between 100 to 200 milliwatts per centimeter cubed, so well below that threshold. And also, to give you a sense, you can get recreational electrical stimulation devices today from companies like Lifted, et cetera. And, I think if you look at the very— it's a far less precise technology, et cetera. And so, the safety record of ultrasound relative to any of these other technologies is far better. So, I mean, to give you maybe even in a phrase, it's less than a sonogram on a mother's womb. That's what we're talking about. So, it's not regulated. One of the key differences between us and a lot of the people in the TFUS space they're almost universally targeting medical applications, which is great. It just means that the time to market is much longer. And so, we focus on these recreational consumer experiences. And so, yeah, I mean, that's kind of the framework and why we don't have to go through it. Yeah.

Interesting, interesting. So, there are people trying to use transcranial focused ultrasound for medical applications. But you are not actually doing any medical claims, medical devices. You're not claiming to cure anyone. It is just a state of lucid dreaming, and that's all that you're trying to do at this point. And the amount of ultrasound which is being projected into the brain is something like 1/7 the regulated limit, like 1/7, 1/8 of the regulated limit. Something like that.

Yeah, one seventh, one one fifth or something like that.

Yeah, between one seventh and something like that. Average. Perfect. So what does Morpheus look like? What is what does V two of the device look like? We have a V one. It's coming out soon. It's in it's in beta. It's in you know you have beta testers coming in. Is a V two basically Going to be kind of better and smaller, like more transducers? And obviously you can upgrade your model, you can update your model anytime on the backend. But if we have a V2 device, is that, are you looking for further development and miniaturization of the transducers of the EEG? Are you looking for more resolution? Are you looking to pack more of them in a smaller device? What would you look for before you pull the trigger on a, you know, on a V2 in the hard way?

Sure. We are very dead set and focused on V1, as you can imagine. I mean, right now, right, we're migrating the model and the technical prototype we have to do. Actually, you know, I know you probably can see some of my team building a 3D scan line tank for a hydrophone where we use it to test the pressure and so on and so forth and compare that. To simulation, ultrasound simulation that we run to basically calibrate it. But, I mean, in terms of— we're very focused on that. But, you know, in terms of what a V2, where the improvements would be, again, reference that, you know, West's Law that I talked about previously, where you're seeing an increased level of both miniaturization, the number of elements on a given transducer, which will increase, you know, your precision and ability to phase and steer it. And as you mentioned, the model will continuously update in the background. You know, we will launch more and more experiences which you can, you know, subscribe to and, you know, and improve the, you know, the number of experiences that way as well. And so both the number of experiences can be improved by the, you know, the transducers, both the number of the transducers or the number of elements on a transducer, as well as from the model. So you definitely want to, want to improve things like that. As well as improving the energy consumption of the transducers so that you can shrink, shrink, shrink, hopefully, and some of the electronics where you don't need kind of a side totem gear. And hopefully, maybe whether it's V2 or V3, that stuff is just in the headband and so on. So that's how you could kind of see this improving over time as successive models are produced. You know, we, uh, we are hyper-focused on, on just getting V1 done.

What, what have been the biggest challenges for the hardware? Is it heat? Is it power? Like, what, what, what is the thing that has been challenging, uh, the most in the last, like, you know, uh, build-out while you've been building this?

Yeah, I think, you know, one of the key— a few things, you know. One is, you know, we did this, you know, we did our prototype with Card79. We did, you know, They did Neuralink for Elon and they work with a number of other neurotech companies. I'm not sure I can mention their names, but things like comfort is really critical because you have to be able to fall asleep with this on or if you're using it for a waking state experience, it's got to be comfortable. So, just design and form factor is very, very, very important in terms of the hardware. And so, we did excessive Studies on, okay, some people sleep on their side, some people sleep on their stomach, some people sleep on their other side or their back, making sure that it's fully comfortable and you can fit all the components into that comfortable form factor. And then, in terms of more brass tacks, there's component selection, et cetera, which you have to be very focused on. And then, there's all this equipment that we have to build Like this 3D scanline hydrophone tank to actually test, et cetera, and simulation software so that you can cut— because we can't see how the— I can't see how the waves are propagating in your skull. I'd have to cut you open and I don't want to do that to you. So, we have to create— we have to use this hydrophone and the simulation software to create this one-to-one correlation. where we can then be able to take somebody's skull and understand how it's going to propagate through their head. So, I think there's a lot of stuff that you just have to build out in order to even get to neuromodulation, which has really been a focus in the last couple of months. But yeah, come spring, we're going to be doing neurostimulation and validating all these experiences and hopefully having in short order, but you never know, the first ultrasonically induced lucid dream.

Right. In the early days, was there like any hurdles where you needed experimental data to decide whether or not this was a go or not? Were there any, you were like, hey, we need to figure out this, and if this is not going to happen, then we need to pivot, right? Were there any moments or were there anything technically or otherwise that you saw needed to have those decisions?

Yeah. So, again, I mean, comparing 2018 when I first thought about doing this to 2022 when I found TFUS, you needed a different neurostimulation modality than electricity or electromagnetism. And then, on TFUS, TFUS has already been used to induce focus, positive mood, deep meditative states, et cetera. It is neuromodulatory generally. And so, as long as you understand the brain state that you are trying to induce, this is the integrated circuit for non-invasive neurostimulation. So, that was, I think, from our side, your very de-risk. And then, we have this collaboration with the Donners Institute for neuroimaging data. And so, that was also very critical that we needed that pipeline. And then, having successive breakthroughs or acquisition to datasets that we could use to supplement that data as well as expand our data pipeline. But really, it was that and then, of course, just the advent of the neural transformer architecture in 2022, which then Wes expanded and built from the ground up to be a multimodal neural transformer. But, the go in terms of red light, green light was, I think, really on the neurostimulation front. More than anything else, the modality.

Right. Do you sometimes come across purists who are like, lucid dreaming is something that you should naturally come to as a part of a kind of religious or semi-religious meditative process rather than something that is to be offered free to everyone? Do you come across those kinds of people? Do you come across comments like that?

Absolutely. Yeah, absolutely. I've seen a couple of comments on some of the various things that we posted. I think you see kind of like gatekeeping in everything, basically everything. And I think, yeah, I mean, I also was in the psychedelic space for a brief while and you kind of saw the same thing in the psychedelic space where you had either these ethnobotanists, these are sacred plants and compounds. And then, you had the scientists at University College London and Johns Hopkins, et cetera, who were really trying to move things in our understanding forward. And so, I think you see these dynamics play out across many different areas, but I think it's kind of like if We said, again, maybe bringing it back to fire, it's like, no, only lightning gets to create fire or something like that. We shouldn't control it. We could burn down our village or somebody could burn down our village. And so, I fully expected and I'm not surprised that we got that kind of feedback from certain people. But frankly, like, you know, the world is moved forward by people who kind of say, kind of, you know, this is an extraordinary, powerful thing. What if we gave it to everyone?

Right. And on that note, in your early kind of like testers who are testing with you in the lab, is there anyone like— what has been like a kind of remarkable experience when someone has tried this on and they've come to you and like, this is a whoa moment for them? Like what, what, like I'm sure someone has tested it and yeah, you know.

To, to, to, to be very clear, you know, we're starting neurosimulation in, in, in the spring. Um, you know, we have to migrate the model on and, and, and, and, and again, as I mentioned, test out all the, the hardware to make sure that, you know, and, and the, um, and get it in correspondence with the simulation software so that, you know, we're doing this, you know, properly. But, you know, we have, you know, friends at places like the Institute for Advanced Consciousness Study that's been running a number of different ultrasound, you know, experiences. I think meditation and positive mood, I think, were the things that they were doing, as well as maybe some others. And there's actually a really great blog post that somebody wrote on his experience, and it's pretty extraordinary. There's also a really great video that was done, you know, quite some time ago on the research of a gentleman named Dr. Jason Guetti where he was using it to induce deep meditative states. It's like they use this woman who had never meditated a day in her life and induced a deep meditative state. She's very emotional about it. It's an extraordinary thing to be dropped into a brain state that you've never experienced before. So, it is a very powerful technology and experience. We look forward to be giving those experiences to our beta users in the spring.

That's amazing, Lan. I am so inspired. This is such an awesome piece of technology because you often wonder, you often read, let's say, Hermann Hesse or Siddhartha or a lot of these things, and you wonder, actually, they're not able to fully describe their experience, right? It is not possible for anyone to fully verbally describe this kind of altered or different state of consciousness. And we have these kind of like very, very limited understanding and limited ability to describe what's happening in our heads. And to the extent that there is something that you can experience which you have not been able to experience, it's inside you, but you can't really experience it. And to the extent that you can use a technological innovation to kind of like trigger that, and learn something more about yourself. I mean, that's amazing, right? Yeah, it's absolutely fantastic.

Obviously, I think we share your enthusiasm. I think the best thing that you can do for people is give them the experience. If you can give them experience, it's empirical for them. innately. And I think one of the things that's going to be pretty extraordinary is over the years and decades as we have this technology, to paraphrase Altman, right? It's like, this is the worst these neural transformers will ever be. This is the worst TFUS will ever be. And so if we can build on these, both on these curves like West's Law, And, on top of machine learning improvements that we're seeing every day, the ability to give people a device that allows them to explore the state space of conscious experience is going to be transformative for the world. I frankly don't really have a good sense of what the world looks like when this is available en masse, but it is an exciting world. It is certainly A strange one, but it is, I think, one that is profound and something that I look forward to exploring alongside everyone else.

Awesome. Awesome. So I think thank you. Thank you, Eric. Thank you, Wes. Thank you for your time today. Emergent Behavior is a part of the Turpentine Network. For our full newsletter, visit emergentbehavior.com/newsletter. Emergentbehavior.co. For more Turpentine podcasts and sponsorship inquiries, visit turpentine.co.