Pod Insights From First Principles Can AI Help Wake Coma Patients? The Science of Consciousness (EP 35) About this episode Hosted by Lester Nare and Krishna Choudhary, this episode is a deep dive into one of the hardest questions in neuroscience: what breaks in the brain during a coma, and can we figure out how to turn consciousness back on? We unpack a new paper from Daniel Toker et al. that uses an interpretable AI framework — not a generic black box chatbot model — to reverse engineer the biological mechanisms of prolonged unconsciousness, recover known features of coma, predict new ones, and propose a possible new target for deep brain stimulation.
Summary
Why diagnosis is so hard — disorders of consciousness are not just about whether a patient is awake, but whether awareness is still present even when motor output is gone.
The mesocircuit hypothesis — the episode explains how the cortex, thalamus, and basal ganglia may work together like an electrical grid to support consciousness.
Interpretable AI, not black-box hype — Daniel Toker’s team built a biophysically grounded model that rediscovered known coma features and predicted two new biological mechanisms.
A possible stimulation target — the subthalamic nucleus emerged as a standout candidate for deep brain stimulation, suggesting a new path toward restoring wakefulness.
Support the show
Donate: FFPod.com/donate
Follow: @FFPod on X / Instagram / TikTok / Facebook
Show Notes
Daniel Toker et al. — Adversarial AI reveals mechanisms and treatments for disorders of consciousness
Nicholas Schiff et al. — deep brain stimulation in a minimally conscious patient
Adrian Owen et al. — fMRI evidence of covert awareness in a patient diagnosed as vegetative
Transcript 12,060 words · 2 speakers
Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Krishna Choudhary? There's a question that haunts them every time, which is, is my loved one still there? And this question is one of the hardest problems in all of neuroscience, right? What is consciousness? He used AI to reverse engineer the biological mechanism of unconsciousness in coma patients. AI has rediscovered known features of unconsciousness. And then it's also predicted two entirely new biological mechanisms.
0:27 Lester Nare? Hello, internet. This is your captain speaking, Lester Nare, joined as always by my co-host and our resident PhD, Krishna Choudhary. We are back for another deep dive episode where we are going to talk about a paper today that's trying to tackle one of the hardest questions in neuroscience about comas and how AI is helping us better understand the state of deep, prolonged unconsciousness. This was a paper published in Nature Neuroscience by friend of the pod, Daniel Toker, at all on March 26th. And as always, we are going to learn about the science from the ground up because this is from First Principles.
1:24 Krishna Choudhary? Every year, tens of thousands of people fall into comas. And it's either from traumatic brain injury or a stroke, cardiac arrest, drug overdoses. And whenever this happens, some patients recover and some patients don't. We don't really know why some recover and why some don't. The other thing is for families that are sitting in that intensive care unit, there's a question that haunts them every time, which is, is my loved one still there? And this question is one of the hardest problems in all of neuroscience, right? What is consciousness? It's not just philosophically hard, which that's above my pay grade and we're not going to get into, but even like mechanistically and clinically, this is a very hard problem, right?
2:08 Krishna Choudhary? What is consciousness even physically speaking? Obviously, it has something to do with the brain.
2:15 Krishna Choudhary? The human brain has eighty-six billion neurons. It's a muscle of fat and neural tissue. And it's quite incredible that that like sack of fat and tissue is where we get this profound phenomenon of consciousness, right? Our ability to perceive, to feel, to think, and to have this subjective experience of ourselves and the world around us. And it's all just from like this sack of like blood and fat in our skull, right? And one can ask, so it's somewhere in the brain, but what are the specific circuits that generate consciousness?
2:55 Krishna Choudhary? And if a part of that circuit breaks, how does it affect consciousness? And then can I actually figure out how to repair it, right? Now, doctors can observe that a patient is conscious or unconscious, right? Because they're not responsive. And they can classify the depth of unconsciousness on behavioral scales. But all we can do really is like wait for them to wake up. There are some new therapies right now that we're going to get into that are maybe getting at trying to wake them up. But at the end of the day, we still don't have a way to peer inside the machinery of awareness and start asking like, what is the thing that is broken? And how do we fix it? Okay, or how do we replace it?
3:40 Lester Nare? There, there's still this delta between sort of what we have as a scientific understanding currently and the actual mechanism that generates that it's kind of a little bit of a
3:51 Krishna Choudhary? Yeah.
3:52 Lester Nare? Dark cave.
3:52 Krishna Choudhary? Yeah, very much a black box in that sense, right? And that's why I really like this new paper. It's a new study that I think represents a genuine step change in that field. Okay, so it's a team that was led by Daniel Toker at UCLA, who is, as you said, a friend of the pod. He went to college with us at Princeton undergrad. Then I think he went to Berkeley for a PhD and then came to UCLA and now has like a gig with UCLA and USC as a postdoc. He's also the brain scientist on Instagram and TikTok. He's got a bunch of followers, really into science communication. You guys should check out his page. It's really, really cool. So.
4:33 Krishna Choudhary? The cool thing about this paper, the cool thing that Daniel did was he used AI to reverse engineer the biological mechanism of unconsciousness in coma patients. And the way that he used AI is very different from, I think, when people think about using AI to do science. They think like, oh, black box LLM model, I like trained it on a bunch of data and then it's going to spit out like a classifier, consciousness or not consciousness. There's part of that here, but what really I think is cool about it is there's, the AI has rediscovered known features of unconsciousness. And then it's also predicted two entirely new biological mechanisms.
5:16 Krishna Choudhary? Because the way that he has engineered this AI is it's not a black box. There's actual mechanistic understanding about a model of the human brain that he's hardcoded in. And that's what the AI is sort of acting on. That's what the evolution of the AI is acting on. And we're going to get into that, but I think it's, it's very, very cool. And the implications are in every direction.
5:40 Lester Nare? I just want to make a quick note here. We use the term AI in a very general sense. When we discuss on the pod, it is a very expansive area of research with a lot of different subcategories and subboxes. So a lot of times we will just say AI for shorthand because we have a lot to get to. But we don't mean AI equals LLM when we do so. It is sort of talking about the entire field of research at a high level. And we then get into the weeds of what we're talking about specifically. But it's just as a small caveat note, that's important because it is a vast, vast space that is beyond just chatbots.
6:19 Krishna Choudhary? Exactly. And this is, this is very much not a chatbot. It's very, very cool. We're going to get into it, right? And the implications, I think, radiate in all directions, right? Something like 5,500 Americans currently are in a persistent vegetative state. And there's a larger number that's in a minimally conscious state. And this research offers a genuine scientific pathway towards treatment. From the philosophy of mind perspective, it provides like a very mechanistic, specific picture of what makes a brain conscious. And then from, for an AI perspective, it's also very cool to think about how one can use.
6:56 Krishna Choudhary? Underlying principles of AI architectures to understand something as complicated as consciousness, right? So it's a deeply personal question that I think this is answering. Like consciousness is the one thing I think that we can all be sure exists because at least we experience it. It's the Descartes, I think, therefore I am. Clearly, consciousness is a thing. Maybe the world isn't.
7:18 Lester Nare? Right.
7:18 Krishna Choudhary? And quantum fields aren't, but consciousness like 100% like I am me, right? Sitting here talking in this microphone and you are you and everybody agrees that like that is a thing. So it's, it's a deeply personal question. And it's still a phenomenon that we, I think, understand the least out of everything. So any, any light that is shed on this mystery, I think is certainly welcome. All right. So with that in mind, let's get into some of the history behind this type of research.
7:49 Lester Nare? Yeah.
7:49 Krishna Choudhary? And we'll start all the way with the ancient Greeks.
7:52 Lester Nare? Always.
7:53 Krishna Choudhary? Okay. Because the ancient Greeks, they had some crazy ideas, like Aristotle thought that the cool, the brain was just a cooling organ for the blood and consciousness actually lived in the heart. We're gonna, we're gonna let that slide because other people like Hippocrates, the guy from the Hippocratic oath, right?
8:13 Lester Nare? Yes.
8:13 Krishna Choudhary? He noted that patients who fell into deep unresponsive sleep after a head trauma rarely ever recovered. Okay. And the word coma comes from deep sleep in Greek. So they had already sort of classified normal sleep versus this pathological deep sleep coma.
8:33 Lester Nare? Yes.
8:33 Krishna Choudhary? They had already kind of figured that out. But again, it was just sort of an observation, right? Then in the Renaissance, there came the anatomical revolution with Viselius. He published this massive manuscript called De Humani Corporis Fabrica in 1543. He was basically like really into dissecting human bodies and gave us incredible drawings of human anatomy. So from there, now we've got an understanding of brain structure, right? But it's kind of like the machine is getting catalogued, but we don't really know what any of the parts are doing.
9:08 Lester Nare? So already by the early 1600s, we both had like a philosophical framework or concepts developing about classifying what these different parts of the body are doing and how that. Like how that translates behaviorally.
9:21 Krishna Choudhary? Yeah.
9:21 Lester Nare? As well as sort of a very detailed cataloguing of the physical structures.
9:27 Krishna Choudhary? Yeah.
9:27 Lester Nare? At least as it related to the brain and other key parts of the human body. You know, four, 500 years ago.
9:33 Krishna Choudhary? Yes, yeah, exactly. It was a catalogue. We didn't know what each thing did.
9:37 Lester Nare? Right.
9:37 Krishna Choudhary? But there was a pretty good catalogue already, right? And then in the 1800s, we had some pretty revolutionary ideas. Paul Broca in 1861, he discovered this idea of cortical localisation. There's an area of the brain now called Broca's area. And now we know that that damage to that area causes language loss. We also know that electrical stimulation in the motor cortex causes muscle movements. So now we're starting to get into the brain, it turns out, is a machine that has specialised parts. And each of these parts is doing something, right? The other invention in the 1800s was electrophysiology. In the 1870s, Richard Caton placed electrodes on exposed skulls of rabbits and found that there was electrical activity.
10:22 Krishna Choudhary? And then we see Hans Berger in 1929. He develops the EEG, the electroencephalogram, which is you put leads on the skull and you can actually read electrical activity. And he found that the brain electrical activity dramatically changes with the state of consciousness. Okay, so different patterns mean waking, sleep, anaesthesia. So already now it's like, okay, the brain is doing different things during unconsciousness. So perhaps the brain is really the seat of, you know, consciousness.
10:53 Lester Nare? Yes.
10:54 Krishna Choudhary? That kind of makes sense.
10:55 Lester Nare? Yes, yeah, yeah, yeah, yeah.
10:57 Krishna Choudhary? Finally, 1949, Giuseppe Morozzi and Horace Magouin, they published this landmark paper. And this demonstrates something called the reticular activating system. It's a diffuse system that runs in the brainstem. The brainstem is where your brain meets the rest of the body with the spinal cord. So it's kind of like the connecting part of the brain. And they found that that part of the brain, the part that connects the brain to the spinal cord, that is essential for maintaining wakefulness. So now we've honed in on this is the sort of general area. Where consciousness is probably happening.
11:35 Lester Nare? So to make a crude analogy, we figured out the brain is the factory. And we now know that different parts of the factory have different workers.
11:44 Krishna Choudhary? Yeah.
11:44 Lester Nare? The engineers are downstairs, the sort of manufacturing line is upstairs. The finance department's over here.
11:51 Krishna Choudhary? Yeah.
11:52 Lester Nare? And, and, but, but we didn't really know where the boss's office was.
11:55 Krishna Choudhary? Yeah, yeah.
11:56 Lester Nare? Where is the boss's office?
11:57 Krishna Choudhary? Yeah, yeah. It seems like that's where it's the brain stem.
12:00 Lester Nare? Right.
12:00 Krishna Choudhary? It's that part, right? And over the following decades, we've mapped out that brain stem in more and more detail. So we figured out there's something called the locus coeruleus. There's something called a basal forebrain. There's something that's part of the brain stem. There's the hypothalamus, there's the raphinuclei. And so all of these like sort of parts, you see where the spinal cord is going in. That junction is where all of this is happening.
12:27 Lester Nare? Which it's, it's almost like even just from a crude structural, like just looking at it. You would imagine if the things that are most sensitive have the most protection, just locationally.
12:36 Krishna Choudhary? Yes. Locationally, that makes sense from one perspective of evolution. And the other perspective of evolution, I think, is that like the brain sort of grew out from that part, right? So it's like the oldest part of the brain is the part that connects it to the rest of the body. Because, I mean, let's face it, like the brain evolved to control movement, right? And so the brain controlling the rest of the spinal cord is sort of probably how it started. And then like higher and higher cognition formed around that, right? Like, oh, okay, I need a sense of place. I need a sense of play, a sense of time.
13:09 Lester Nare? Yep.
13:09 Krishna Choudhary? I need a sense of problem solving. All of that starts forming around this like nucleus.
13:15 Lester Nare? And then the prefrontal cortex being the most advanced.
13:18 Krishna Choudhary? That's the one that's like, yeah.
13:19 Lester Nare? Right in the front, which.
13:21 Krishna Choudhary? You know, it's, it's very cool to think about it that way. Just like in terms of stages of development over the millions of years of evolution.
13:27 Lester Nare? Okay.
13:27 Krishna Choudhary? Okay. So now we know that it's like kind of the brain stem, but we still don't have really an idea of like. Okay, this thing connects to this, this thing connects to this, so this is powering that, that, that, that, and so on and so forth.
13:40 Krishna Choudhary? 2007, there's a pivotal moment in consciousness research because Nicholas Schiff and his colleagues published this paper in Nature, and it showed that continuous deep brain stimulation, this is the idea of you, we take an electrode and we pulse it with electricity, that is going to excite the neurons near that electrode. It's just you're pumping electricity deep into the brain, that's why it's deep brain stimulation. And deep brain stimulation of the central thalamus helped a minimally conscious patient regain the ability to communicate and feed himself. Okay, this was a very big deal, because this was an ability that he had lost about six years earlier after a traumatic brain injury. It's a preliminary result, single patient, notoriously in human consciousness studies
14:25 Krishna Choudhary? and things like that, the N is like very low. Single digits is you're already getting a lot, because human, I mean, you're like experimenting on human beings, right? There's tons of ethical concerns, there's not a lot of like, you know, people who are willing to do this. And so, even though the data is low, the fact that it worked means that there's something going on. So that sort of rejuvenated this research. And in 2010, that same author, Nicholas Schiff, he came up with something called the mesocircuit hypothesis. Okay? Effectively, what he's saying is the following: the consciousness relies on a network of interactions between the cortex and the central thalamus. Okay. And that interaction is regulated by a part of the brain called the basal ganglia.
15:09 Krishna Choudhary? We're going to focus on effectively three regions of the brain, okay? And I'm gonna use an analogy of an electrical grid. Okay. Okay? A city's electrical grid. Yes. So, the cerebral cortex, this is the part that's on the outside, the most sort of recently developed in humans, this is the, when we think about the brain and all the wrinkles, that's the cerebral cortex that we're seeing on the outside. And that outer layer, it's responsible for higher thought, memory, sensory processing. And in this analogy of a city's electrical grid... This is your network of homes, podcast studios, you know, businesses that consume the electricity, right? Now, the central thalamus is the deep brain relay station.
15:54 Krishna Choudhary? That's the part that's sort of on the right over there in the pink.
15:58 Lester Nare? Just above the brain stem.
16:00 Krishna Choudhary? Just above the brain stem, okay? So that's getting input from the cortex and it's also giving input to the cortex. Okay, so it's a two-way street. It's kind of like the primary power substation.
16:09 Lester Nare? Yes.
16:09 Krishna Choudhary? It's pumping excitatory input to the cortex, but it's also getting excitatory input back from the cortex. So it's relaying a bunch of electricity one way or the other.
16:19 Lester Nare? So the one of the examples being if you have solar on your house, you're sending electricity back to the grid. It's not just you're receiving it as a.
16:27 Krishna Choudhary? That's very good. Yeah, yeah, yeah. No, that's exactly right. And this feedback loop kind of helps both run.
16:33 Lester Nare? Right.
16:33 Krishna Choudhary? Okay, in some sense. And then finally, there's the basal ganglia and that's the striatum and the globus pallidus. That is a kind of regulatory system and it's kind of like a voltage controller that manages the flow of power in this entire network.
16:49 Lester Nare? So in the summer, when it gets hot, they have to decide where to send energy, where to not send energy in order to not overflow. Like there has to be, there's somebody decides, okay, we're going to shut down this neighborhood.
16:58 Krishna Choudhary? Yeah.
16:59 Lester Nare? Because we need to power the nucleus.
17:01 Krishna Choudhary? Yeah, yeah. Yeah, exactly. As a crude analogy, it's kind of routing signals and it's controlling both the power stations and the homes to make sure like nothing's. Yeah, there's balance and nothing's, nothing's going out of the way.
17:13 Lester Nare? Yep.
17:13 Krishna Choudhary? Okay, so that's how sort of the consciousness mechanism is working in a healthy patient. What happens during coma? Well, according to the mesocircuit hypothesis, here's what happens. When there's widespread brain injury, right? Like from a traumatic brain injury or a stroke, so a bunch of neurons die because they don't get oxygen and things like that. Your power lines are going to be down on the cortex, right? So you're going to get localized areas of power cuts.
17:37 Lester Nare? Yep.
17:37 Krishna Choudhary? Like we do in LA sometimes when somebody steals our copper.
17:40 Lester Nare? Yes.
17:41 Krishna Choudhary? From underground.
17:42 Lester Nare? Yes.
17:42 Krishna Choudhary? Okay.
17:42 Lester Nare? Yes.
17:43 Krishna Choudhary? But anyways, the cortex is now going to reduce its excitatory input to the thalamus. That connection is broken, as you can see on the left. The thalamus and the cortex, that's now a dotted. Line instead of on the left, it's a solid line.
17:55 Lester Nare? Yep.
17:56 Krishna Choudhary? Okay, the other thing is, um, there are these special neurons in the striatum,
18:03 Krishna Choudhary? called medium spiny neurons, and what they do, they're super prone to metabolic stress death, meaning you like, remove a little bit of oxygen and they'll, they'll just die. Okay, they're not very resilient, they're very finicky. And if those guys die, then the striatum is going to not inhibit the part that is inhibiting the thalamus. It's a double negative now.
18:25 Lester Nare? Okay.
18:25 Krishna Choudhary? Okay, so because of that, so what it's, what it's doing is usually the striatum is like inhibiting the part that is putting brakes on the thalamus. Okay, but now that that brake is gone, the, the first brake is gone, right? The second negative is gonna go even more negative. Okay, and it's gonna start inhibiting the thalamus even more. That's why that red line just got fatter.
18:50 Lester Nare? That makes sense.
18:51 Krishna Choudhary? Right, so the thalamus is one, not getting any input from the cortex because the power lines are down.
18:55 Lester Nare? Yes.
18:55 Krishna Choudhary? And second, the part that was putting brakes on the thalamus has now gone hyperactive and putting even more brakes on the thalamus. So the thalamus is getting inhibited. It's getting shut down more and more in a coma because of these very tiny defects that have happened in exactly the wrong areas.
19:13 Lester Nare? Yeah, so this is, this is very interesting and what we're sort of saying here is the, the traumatic, the traumatic brain injuries are disrupting the normal flow between these three fundamental component parts.
19:26 Krishna Choudhary? Yeah.
19:28 Lester Nare? The cortex, which is like the end destination, the Holmes example, the thalamus, which is sort of sending and receiving, and then the striatum and palladium that are kind of controlling the amount of what is or is not being sent.
19:40 Krishna Choudhary? Yeah.
19:40 Lester Nare? Where it is or is not being sent.
19:42 Krishna Choudhary? Exactly, and, and, and it's exactly at the right part where the thalamus, which is the relay station,
19:47 Lester Nare? Yes.
19:47 Krishna Choudhary? is just getting signaled to shut down.
19:49 Lester Nare? Yeah, yeah, yeah.
19:49 Krishna Choudhary? Okay, and when that shuts down, now this entire circuit of consciousness is getting shut down.
19:55 Lester Nare? It's, it's almost like the failsafe system that happens at a substation. If there's some overload, just gets stuck in the locked position.
20:00 Krishna Choudhary? Exactly, exactly, and there's, and there's some evidence that this mesocircuit hypothesis is probably correct.
20:07 Lester Nare? Okay.
20:07 Krishna Choudhary? There's certain stimulant drugs, for example, for example, um, amantadine, which enhances the inhibition to that brake, like the part that's inhibiting the thalamus.
20:18 Krishna Choudhary? Amantadine inhibits that part, so it's no longer going to inhibit the thalamus. Kind of works and it kind of wakes up patients. It's turning it to overclock. Yes, yes, exactly. So, so there, there's some mechanism where these drugs are kind of working with the mesocircuit hypothesis. But at the end of the day, still a hypothesis because it's not very granular, right? It's just like, oh, giant brain area. You know, like there's, but the giant brain area has millions of neurons. Right. That are all different types. What type? Right. What type is doing what? Right? And also you can't really test it directly because the bottleneck is standard animal models are really bad at mimicking human comas. This prolonged unconsciousness.
21:01 Krishna Choudhary? And that just has to do with the fact that animal models like rodents, the brain is structurally a very different thing. Right. Right? Even though it's still a mammal, one of the things that we think happens during traumatic brain injury that causes this kind of coma is the idea of a diffuse axonal injury. Here's the idea. You've got the cerebral cortex on the outside. There, there's gray matter, which is like sort of the cell bodies. And then there's white matter. You might've heard of that gray matter and white matter. Gray matter is kind of like the cell bodies. The white matter is the axons. These are these like cables, like, you know, the high voltage cables that come from, you know, Hoover Dam to LA. The axons are kind of like that. They're the high voltage cables. Now, if we go to a photo 14, what you'll see is
21:46 Krishna Choudhary? the white matter is kind of in the middle and it's like connecting these areas of gray matter. But if you have a traumatic brain injury, let's say you got in a car crash or like, boom, like your brain had severe G-forces on it. Those axons, which are the cables, those are going to rupture first. Yeah. Right? Yeah. Now with a small rodent, the brain is not big enough. To feel that kind of G-force stress, right? Ours is big enough, we're like, and we're like doing crazy things as humans. We're like, we're driving around at 70 miles an hour. We're like, we're now subject to G-forces if something goes wrong, right? And so the, our brain has just not evolved. When we were hunter-gatherers in Africa, like evolving.
22:31 Lester Nare? Right.
22:31 Krishna Choudhary? The evolution didn't think, hey, I should probably make this brain, um, you know, withstand getting hit by.
22:39 Lester Nare? Another 350-pound man in an NFL game.
22:41 Krishna Choudhary? Yeah, exactly. Okay, so, so, so there's a mismatch between like what the brain is designed, like the, the, the environment that the brain is designed in, and then what the brain was capable of creating, right? Which is like, which is like, human beings in cars and all this other crazy stuff, skiing and then hitting a tree.
22:59 Lester Nare? Right.
22:59 Krishna Choudhary? And things like that, right?
23:01 Lester Nare? Right.
23:01 Krishna Choudhary? So, and we simply can't test in rodent models.
23:06 Lester Nare? Right.
23:07 Krishna Choudhary? Okay?
23:07 Lester Nare? And this is something we talked about on a recent previous episode too, about just, you know, uh, there are a lot of opportunities to use animal models, which, the purpose of which, yes, for folks who are, you know, PETA-adjacent, et cetera, the, the, the ethical concerns about testing directly on humans are so prevalent that it's perceived as a lower ethically bad option.
23:32 Krishna Choudhary? 100%.
23:32 Lester Nare? Um, and for an outcome that's going to save millions of lives at some point in the future.
23:36 Krishna Choudhary? Yeah, but here, like.
23:38 Lester Nare? It doesn't.
23:39 Krishna Choudhary? It doesn't, the ethics like is even more problematic because well, what, you're not going to get anything out of it.
23:44 Lester Nare? Right.
23:44 Krishna Choudhary? Right?
23:44 Lester Nare? Right, and so you're just.
23:46 Krishna Choudhary? Yeah, you're just like wasting animals at that point, right?
23:50 Lester Nare? Right.
23:50 Krishna Choudhary? And, and that's not something that we want to do in science research.
23:53 Lester Nare? Right.
23:53 Krishna Choudhary? So, that's one bottleneck.
23:55 Lester Nare? Okay.
23:55 Krishna Choudhary? Which is just, animal models are not a thing. The second bottleneck is the diagnostic crisis.
24:01 Lester Nare? Okay.
24:01 Krishna Choudhary? That this is in general with consciousness in medicine, okay? Consciousness has two fundamental components. There's wakefulness, which is like arousal, like when you're awake. And then there's also awareness. That's the qualitative experience of like, me being me and seeing you and everyone else, right? This is some qualitative subjective experience that only I have about what it's like to be me, right? So. Disorders of consciousness that comes from physical trauma and oxygen deprivation. There's several different types. So in a coma, you have neither arousal nor awareness. Right? Because you can't be awoken.
24:35 Lester Nare? Yes.
24:35 Krishna Choudhary? There's no sleep-wake cycle.
24:37 Lester Nare? Yes.
24:38 Krishna Choudhary? And they fail to respond to stimuli.
24:40 Lester Nare? The lights are off.
24:41 Krishna Choudhary? Yeah, lights are off and maybe nobody's home.
24:43 Lester Nare? Right.
24:43 Krishna Choudhary? Okay. Vegetative state means it's also called unresponsive wakefulness syndrome.
24:49 Lester Nare? Okay.
24:49 Krishna Choudhary? You have wakefulness.
24:52 Krishna Choudhary? And you have a sleep-wake cycle, but perhaps you have no awareness. So the lights are on, but maybe nobody's home.
24:58 Lester Nare? Okay.
24:59 Krishna Choudhary? But now it gets kinda weird. Like, what if they are aware?
25:03 Lester Nare? Yeah.
25:04 Krishna Choudhary? But they just can't have any agency over their motor controls.
25:09 Lester Nare? Mm-hmm.
25:10 Krishna Choudhary? Right? So how can you tell?
25:12 Lester Nare? So someone's home, but they can't answer the door when you ring the doorbell.
25:15 Krishna Choudhary? Exactly.
25:16 Lester Nare? The lights are on.
25:16 Krishna Choudhary? Lights are on.
25:17 Lester Nare? And someone is home.
25:18 Krishna Choudhary? Yeah.
25:18 Lester Nare? You ring the doorbell, but they can't answer.
25:20 Krishna Choudhary? That's very good. Yeah, exactly. How do you, how can you tell? Right? This is a
25:24 Lester Nare? Yeah, that's subtle.
25:25 Krishna Choudhary? That's very subtle and for the longest time wasn't really considered until science paper 2006 by Adrian Owen. It's a landmark paper. It's quite incredible. This was a wake-up call, so to speak. Okay? He used functional MRI to ask a patient that was diagnosed in a vegetative state to imagine playing tennis.
25:45 Lester Nare? Mmm.
25:46 Krishna Choudhary? And imagine walking around her home.
25:48 Lester Nare? I love scientists so much.
25:49 Krishna Choudhary? Okay? And the brain activity from that FMRI study
25:53 Lester Nare? Yes.
25:53 Krishna Choudhary? was identical to someone who was fully aware.
25:56 Lester Nare? Mm-hmm.
25:57 Krishna Choudhary? So she, like, she's, she's there in this vegetative state. There's no way that she can express
26:03 Lester Nare? Mm-hmm.
26:03 Krishna Choudhary? her awareness.
26:04 Lester Nare? Mm-hmm.
26:05 Krishna Choudhary? But if we look into the brain MRI, it's the same as someone who's aware.
26:08 Lester Nare? Mm-hmm, mm-hmm. The, the idea being they may not be able to move their hands to touch a button.
26:15 Krishna Choudhary? Mm-hmm.
26:15 Lester Nare? Or verbalize as the
26:17 Krishna Choudhary? Or even move their eyes.
26:18 Lester Nare? Move their eyes, which is techniques that have been used before
26:21 Krishna Choudhary? Yeah.
26:21 Lester Nare? to communicate. Now we're just saying it's because the awareness is in the brain.
26:26 Krishna Choudhary? Yeah.
26:27 Lester Nare? And so you shouldn't actually need any of these secondary Uh, meth avenues.
26:32 Krishna Choudhary? Yes.
26:32 Lester Nare? To communicate. And the only way to make the differentiation between the, uh, is someone home versus is someone not home? Reference point, you have to go directly to the source.
26:44 Krishna Choudhary? Yeah.
26:44 Lester Nare? If all of their physical capacities.
26:46 Krishna Choudhary? Yeah.
26:47 Lester Nare? Are, are, are not available to them, which does not necessarily mean.
26:50 Krishna Choudhary? Yeah.
26:50 Lester Nare? That no one is home.
26:51 Krishna Choudhary? Exactly. It's like, no one's answering the door, but here it's kind of like, we did an X-ray of the home.
26:55 Lester Nare? Yeah.
26:55 Krishna Choudhary? And the person is moving around.
26:57 Lester Nare? Right, right. We have the thermal and you can see.
26:58 Krishna Choudhary? Exactly, you know?
26:59 Lester Nare? Okay.
27:00 Krishna Choudhary? It's, I thought that was a very cool paper.
27:01 Lester Nare? That is very clever.
27:02 Krishna Choudhary? It's very clever and now, now it's, it's a real crisis, right? Because now there's a third, it's called minimally conscious state. Okay? And it turns out that 15 to 20% of patients that are classified as vegetative. Actually might have covert conscious awareness. That's a lot.
27:21 Lester Nare? That's gotta be terrifying.
27:22 Krishna Choudhary? And that's gotta be so terrifying, right?
27:24 Lester Nare? Cause you can hear and see everything happening around you and you cannot engage in that environment. That's just unbelievable.
27:29 Krishna Choudhary? And this was only 20 years ago, 2006, right?
27:32 Lester Nare? After this 500 year cycle we've just talked about.
27:34 Krishna Choudhary? Yeah. So, I mean, this field is just one of those where you can really sit back and be like, wow, we really don't know much.
27:41 Lester Nare? Right.
27:42 Krishna Choudhary? About the brain.
27:42 Lester Nare? Right.
27:43 Krishna Choudhary? Like.
27:44 Lester Nare? Because this is like, kind of, you know, people's, this goes to the materialist, uh.
27:49 Krishna Choudhary? Yes.
27:49 Lester Nare? Non-materialist debate. You know, is it that consciousness arises out of matter? Or is consciousness primary? Which gets a little, that's, again, that was above our pay grade and not the focus of this podcast. But it is, it is one of those, it's the first question.
28:07 Krishna Choudhary? Yes.
28:07 Lester Nare? Because once you.
28:08 Krishna Choudhary? Yeah, yeah, yeah.
28:09 Lester Nare? You know what I mean?
28:09 Krishna Choudhary? Yeah.
28:09 Lester Nare? You, you can't, everything else downstream is really impacted by which of the two.
28:14 Krishna Choudhary? Which of those two, right? What are we talking about?
28:16 Lester Nare? Right.
28:17 Krishna Choudhary? Who am I? Who are you? And what the hell is going on? That's, that's, that's how I like to say it.
28:25 Krishna Choudhary? You know, and so with with these kinds of studies, it's it's really like underscoring how much we don't know.
28:32 Lester Nare? Right.
28:32 Krishna Choudhary? About this, right?
28:33 Lester Nare? Right.
28:33 Krishna Choudhary? And now fifteen to twenty percent of patients, we're saying, are classified incorrectly.
28:39 Lester Nare? Which is fascinating, because I mean, I'm sure after that study, a variety of because the FMRI's are not hugely
28:45 Krishna Choudhary? Yeah.
28:46 Lester Nare? A negative for a patient.
28:48 Krishna Choudhary? No.
28:48 Lester Nare? So at least as a way of especially for families or anybody to try.
28:52 Krishna Choudhary? Mm hmm.
28:53 Lester Nare? But in any event.
28:55 Krishna Choudhary? Exactly. So now we finally get to Daniel's paper. Daniel Toker in Nature Neuroscience, adversarial AI reveals mechanisms and treatments for disorders of consciousness. It's a really innovative way of using machine learning and the vast amount of data that's now available on comatose patients, vegetative patients. He's made a in silico model of coma, so a computer model of the brain that circumvents circumvents the lack of mouse models.
29:26 Lester Nare? Yes.
29:26 Krishna Choudhary? So now I have a model in the brain that I can play with.
29:29 Lester Nare? Yes.
29:30 Krishna Choudhary? Of a model of the brain in my computer that I'm fairly confident is quite good, that I can play around with.
29:35 Lester Nare? Because the idea is it's based off of this real world data of actual pay the anonymized.
29:40 Krishna Choudhary? Yeah.
29:41 Lester Nare? Data.
29:42 Krishna Choudhary? Exactly. And then I can figure out a kind of mechanistic model about how consciousness works, what gets disrupted to make a coma. And then it becomes a kind of discovery engine. From that, we can mess around with it and propose new treatments. And he is doing all of these.
29:56 Lester Nare? This is this.
29:57 Krishna Choudhary? Very well done, Daniel, by the way.
29:59 Lester Nare? Nicely.
29:59 Krishna Choudhary? Quite good.
30:00 Lester Nare? Nicely done.
30:00 Krishna Choudhary? Quite good. Again, the brain scientist on Instagram. And before we deep dive into it.
30:06 Lester Nare? Yes, yes.
30:06 Krishna Choudhary? Let's do a little bit of housekeeping.
30:08 Lester Nare? Yes. So as always, we are so grateful for all of you to join us for these research deep dives. It's just an incredible, not only an incredible time to be alive, but the opportunity to really understand not only what we're learning today, but how it works and what the history of how we got here. You're not going to get it anywhere else. And this is why we love doing this show at From First Principles. And it is hugely helpful to us to continue to bring you the best and greatest breaking science research. As we, as I always say, fight the billionaire algorithms, a like, a share, a comment, a DM, sending it to a friend, bringing it into Journal Club is hugely, hugely helpful for us,
30:51 Lester Nare? where it's again, still just the two of us running the show, to be able to give this to you multiple times every week. If you would like to become a patron, you can donate to the pod at ffpod.com backslash donate. Any and all support monetarily or non-monetarily is hugely, hugely valuable and we really greatly appreciate it. I have no other show notes this week and I'm very excited to dive back into this paper.
31:16 Krishna Choudhary? Let's do it. Okay. So, first thing I want to do, right, I want to use, I want to use artificial intelligence to make a discovery engine for my brain, figure out why comas happen, and then propose treatment. That's the end goal of this paper, okay? First thing I want to do is train an AI detective that is capable of diagnosing consciousness from raw brainwaves. Okay? Just can I identify consciousness if you give me the, you know, sort of brain activity, the electrical activity of a brain? So he made something called a deep convolutional neural network. These are just our normal, you know, CNNs that we hear about, giant network of artificial neurons that is trained through back propagation,
32:00 Krishna Choudhary? you know, supervised learning, which is just, we've got about 680,000 10-second electrophysiological recordings. So 680,000 recordings of brains, electrical activity. These recordings come from humans, monkeys, rats, and bats, under anesthesia, in a coma, and wakeful. So, a swath all across mammals. So that means it's not going to really hone in on some human feature.
32:27 Lester Nare? Right.
32:27 Krishna Choudhary? It's gonna, it's gonna really start thinking about what is the nature of consciousness itself, right? And he made three discriminators. So he made a cortical neural network that was gonna get trained on the cortical data. There's a thalamic, and then there's a pallidial. That's the three sort of big brain regions that we were talking about in that mezzocircuit model, right? The cortex is your grid with all the homes and the podcast studios. The thalamus is the relay station, and the pallidial station is kind of like governing what's going on between these two, right?
33:05 Lester Nare? Yes.
33:06 Krishna Choudhary? Okay. Once he trains it, he validates it. It's pretty good. The AI score for consciousness, which is plotted on the Y axis, is correlated with the ground truth score of consciousness that it was given to each sample in the test sample. Okay, which means that like the AI, this deep convolutional neural network, is given a sample that it's never seen before, and it's told to guess how conscious it is based on a number between zero and one. That number is correlated with the ground truth. Okay, so that means it's working.
33:39 Lester Nare? Yes.
33:40 Krishna Choudhary? Okay?
33:40 Lester Nare? Yes.
33:40 Krishna Choudhary? The other big one that I quite liked is from figure 1J, if you see over there. What that shows is that the AI was able to classify fully paralyzed ALS patients as conscious. Remember in ALS, we did the story where it's a motor neuron disease, right? All of your motor neurons go away. So you can still be conscious, but you won't be able to fully express that. Here, they looked at EPHIS recordings from ALS patients, and there's no significant difference between them and healthy cortical activity.
34:14 Lester Nare? This is actually a fascinating follow up to our ALS story because we did a great deep dive on understanding what is, what it actually is mechanistically.
34:23 Krishna Choudhary? Yes.
34:24 Lester Nare? And it dovetails with what we said earlier, which is just because your motor functions go away, ALS is a perfect example of what we were just talking about. Doesn't necessarily mean that you're, there's no one home.
34:35 Krishna Choudhary? Yeah, yeah.
34:36 Lester Nare? That's very.
34:36 Krishna Choudhary? Exactly, and this was able to find that, right? Without motor output, you can now classify.
34:41 Lester Nare? That's.
34:41 Krishna Choudhary? So already, big win, figure one.
34:43 Lester Nare? Yes, that is huge.
34:45 Krishna Choudhary? Right? That's good.
34:46 Lester Nare? That is huge.
34:46 Krishna Choudhary? Okay, now, next, we've, we've made a detector. We've made a detector of consciousness. Now that we have a consciousness detector, let's make a brain simulator that can make my own brain signals.
34:58 Lester Nare? Okay.
34:59 Krishna Choudhary? Okay? And this is where we get into the real why there's AI in the title.
35:04 Lester Nare? Okay.
35:04 Krishna Choudhary? Okay? Because just making a deep convolutional neural network that will classify, that's just an autonomous classifier. That's like 10 years old. What are you doing, right? You're not gonna get into nature neuroscience like that, and Daniel knows that.
35:17 Lester Nare? And this is for the social clips. If we end up clipping the first half without the second half.
35:22 Krishna Choudhary? Yeah.
35:23 Lester Nare? Please watch the pod so you get the full story.
35:25 Krishna Choudhary? So they're not like, that's not even, I can't read that's a nature narrative.
35:29 Lester Nare? Yes, we know. We can't fit an hour-long thing onto Instagram.
35:33 Krishna Choudhary? Yeah. Okay, so now what he's gonna do is make a brain simulator. He's going to combine that discriminator that he got, the AI, the consciousness detector, with something called a generative adversarial network. These are GANs. Traditionally with GANs, I mean, they're kind of used to make like generative AI, like photographs and things like that. Here's how it works. So this is very different from like a diffusion model, which is usually kind of used in the, in the zeitgeist. GANs kind of didn't have popular, they had popularity for a while.
36:02 Lester Nare? Yes.
36:02 Krishna Choudhary? And then, and then diffusion kind of went up.
36:04 Lester Nare? Yes.
36:04 Krishna Choudhary? And became the generative sort of paradigm. But then now GANs are kind of back up. We don't really know which architecture is gonna win in the end. But here's the idea. Okay, so you've got something called a forger, which is trying to forge real-life data. For example, in this case, what we're trying to do is create real faces. So we're gonna have a dataset of real faces. And then we're gonna have a forger, that's the generator over there. And that's gonna generate faces.
36:33 Lester Nare? Yes.
36:33 Krishna Choudhary? Okay. And then we're gonna have a discriminator on the other side that is gonna take the real data and gonna take the forged data and try to figure out which one is forged and which one is real. Now, in the beginning, it's gonna be obvious.
36:46 Lester Nare? In the Benning game.
36:47 Krishna Choudhary? Yeah. In the beginning, because the forger doesn't have any training. He's just gonna be making up random nonsense.
36:52 Lester Nare? Right.
36:52 Krishna Choudhary? So the discriminator is gonna be able to look and be like, oh yeah, this is real, this is fake. But as the forger gets better and better, because that output of, hey, this is fake. You keep making fake stuff. I want the real stuff. That output is getting back, back propagated through to the forger. And the forger is thinking, okay, how do I get better? How do I get better?
37:10 Lester Nare? Yes.
37:10 Krishna Choudhary? Pretty soon, after multiple rounds of training, the forger is going to get good at making the real data, such that it fools your discriminator.
37:20 Lester Nare? Yes.
37:20 Krishna Choudhary? Okay, that's the generative part. And the adversarial part is you've got two networks that are adversaries. There's a detective that's trying to detect the fakes. And then there's a forger that's trying to create even better fakes. That's the adversarial part, right?
37:33 Lester Nare? For anyone who's ever used image generation, particularly mid-journey, this structure is exactly the reason why you'll see a blurry, weird thing.
37:44 Lester Nare? When you first initiate the image generation, and then it's going through the cycle, you just talked about it. The fidelity gets closer and closer and closer over and over, until it gives you the final, like, produced, like, image. This is like the whole stable diffusion kind of came in.
37:59 Krishna Choudhary? Yeah, and that one's, that one's
38:01 Lester Nare? Slightly different.
38:02 Krishna Choudhary? Slightly different because, so what you're talking about is very, very close to generative adversarial networks. Because at the end of the day, how the forger creates the fake stuff is it starts with some kind of random noise, which is what you see there. And then, and then it starts manipulating that noise to create the real thing, right? So what you're seeing there is the act of the forger,
38:21 Lester Nare? Yes.
38:21 Krishna Choudhary? in some sense, going from noise to the real thing.
38:25 Lester Nare? Yeah.
38:25 Krishna Choudhary? But the training part is something that's already been done by Mid Journey on the backend.
38:29 Lester Nare? Hundred percent.
38:29 Krishna Choudhary? Right?
38:29 Lester Nare? Hundred percent.
38:30 Krishna Choudhary? And so, that's the generative part.
38:34 Lester Nare? Yes.
38:34 Krishna Choudhary? Okay, great.
38:34 Lester Nare? Yes.
38:35 Krishna Choudhary? So now, um, great. That's how you make a fake Picasso and a real Picasso and things like that. How are we gonna use that to figure out coma?
38:43 Lester Nare? Yes.
38:44 Krishna Choudhary? All right? This is, this is where Dan comes in hot with an interpretable, biophysically grounded mean field model.
38:53 Lester Nare? Okay.
38:54 Krishna Choudhary? And that is what is cool. So on the upper left, that's A, figure A.
38:57 Lester Nare? Yes.
38:58 Krishna Choudhary? Okay. On the left-hand side, we've got our D, CNN, the convolutional neural network. That's our detector.
39:03 Lester Nare? Yes.
39:04 Krishna Choudhary? That we had trained previously.
39:05 Lester Nare? Yes.
39:05 Krishna Choudhary? In figure one, right? There's other detectors in here that'll actually like, um, figure out, is this real or fake data? That one is saying, is this unconscious or conscious data? There's other ones that are not in this backend. You have to go to the supplement to see it. Where it's like, is this real data or fake data? Because first, we want to just create real-looking data. Then we'll worry about, okay, am I creating conscious data or unconscious data, right? So there's multiple steps to this process. But the key is that the forger is not a black box. Okay? The forger is actually a three-component, which, within which, there's multiple components.
39:41 Lester Nare? Yes.
39:42 Krishna Choudhary? But you know that mesocircuit hypothesis of like, there's a cortex, there's the basal ganglia, there's the thalamus, and these guys are talking to one another?
39:49 Lester Nare? Yes.
39:49 Krishna Choudhary? Each of those boxes is its own little neural network, right? And it's a biophysically realistic neural network in the sense that it's not just a bunch of artificial neurons that start at noise. What it's doing is It's saying, okay, how does a biophysical neural network work? Well, some input comes in, then there's gonna be like channels, like, you know, some neurotransmitter is gonna have some characteristic time scale. Other neurotransmitters are gonna have other characteristic time scales. There's gonna be excitatory neurons which do positive feedback. There's gonna be negative inhibitory neurons that do negative feedback. These I can hook up mathematically using differential equations. And I can create a biophysically realistic model.
40:34 Krishna Choudhary? It is no longer a black box. And instead of using backpropagation to change the weights between each neuron, what I'm gonna do is use something called a genetic algorithm to change the parameters of my biophysically realistic model. Meaning, how many excitatory neurons are there? How many inhibitory neurons are there? How is the connectivity, right? What is the time scale of the interaction? Things like that, things that are actually relevant when we think about what is a brain doing?
41:06 Lester Nare? And is in some sense, the reason why it's like, that algorithm is necessarily bounded by what the realistic value ranges are.
41:16 Krishna Choudhary? Exactly.
41:17 Lester Nare? And that's why it's not a black box. Because you know that, oh, it's either gonna be excitatory neurons or they're gonna go between X and Y.
41:23 Krishna Choudhary? Yeah.
41:23 Lester Nare? And that's a realistic normalized range.
41:25 Krishna Choudhary? Exactly, that's one of them. And the other one is, we can literally point to components of the model and be like, oh, it's the excitation in the thalamus.
41:32 Lester Nare? Ah, yeah, yeah, yeah.
41:33 Krishna Choudhary? Right?
41:34 Lester Nare? Yeah, yeah, yeah, yeah, yeah.
41:35 Krishna Choudhary? I can literally be like, oh, it's this part.
41:37 Lester Nare? Yeah.
41:37 Krishna Choudhary? With again, it's like with normal neural networks, I don't have no idea.
41:42 Lester Nare? Yeah.
41:42 Krishna Choudhary? Like, you know, I type in, make me a cat on a horse. I have no idea in the trillions of parameters where it decided cat.
41:51 Lester Nare? Yes.
41:52 Krishna Choudhary? Where it decided horse, how it figured out to put the cat on top of the horse. Here, literally, I could just be, I could look into the parameters. I can see how they evolve with training and figure out what is actually going on.
42:03 Lester Nare? That's very.
42:03 Krishna Choudhary? And then make that extrapolation from there to the brain.
42:06 Lester Nare? Very good.
42:07 Krishna Choudhary? That's very good, I think.
42:08 Lester Nare? That's very good.
42:09 Krishna Choudhary? Right?
42:09 Lester Nare? Yeah.
42:09 Krishna Choudhary? And, and that's one of the things I like a lot.
42:12 Lester Nare? Yeah.
42:12 Krishna Choudhary? Right? This loss function, he had a loss function that sort of like figures out, right? How to, how to change the parameters? And this loss function incorporates the input, outputs from the networks that train to, that's trained to classify real versus synthetic data. It also has outputs from the consciousness detector that we had done earlier. It's also got a way to identify seizures. Um, and as you said, it's got these empirical constraints on the firing rates of these neurons. As you said, like, right?
42:39 Lester Nare? Right.
42:39 Krishna Choudhary? Excitation can't be that high.
42:41 Lester Nare? Right.
42:41 Krishna Choudhary? Inhibitory neurons are usually higher firing rate than excitation, things like that, right? And the interregional communication patterns.
42:47 Lester Nare? Right.
42:48 Krishna Choudhary? Like, excitation is long range, inhibition is usually short range. This is stuff you can bake into the model.
42:54 Lester Nare? This is quite nice.
42:55 Krishna Choudhary? Right?
42:55 Lester Nare? This is, this is quite nice.
42:56 Krishna Choudhary? I love the fact that it's not black box, because that's one of the things that I hate about.
43:00 Lester Nare? Right.
43:00 Krishna Choudhary? Like, just large neural networks.
43:02 Lester Nare? Yeah.
43:02 Krishna Choudhary? It's just, I have no idea where anything is happening.
43:05 Lester Nare? Yeah, and it's also so different from the culture of software.
43:07 Krishna Choudhary? Yes.
43:08 Lester Nare? Since the, it's beginning, which has been that everything is like explicit. And you can literally point to where exactly, like the, the trace callback of where something's.
43:15 Krishna Choudhary? Exactly.
43:16 Lester Nare? Coming from.
43:16 Krishna Choudhary? This is the going back to that type.
43:18 Lester Nare? Yeah.
43:18 Krishna Choudhary? You know?
43:19 Lester Nare? Yeah, which is great.
43:19 Krishna Choudhary? Which is great.
43:20 Lester Nare? This is like, yeah, fantastic.
43:22 Krishna Choudhary? Yeah. And so, once this genetic algorithm has trained my, um, you know, brain maker. It's an artificial brain. Now we can ask, okay, um, how about we reprogram the objective now?
43:35 Lester Nare? Okay.
43:35 Krishna Choudhary? It was just making real world data.
43:37 Lester Nare? Yeah, yeah, yeah.
43:37 Krishna Choudhary? Now I reprogram the objective to make comatose data. So I can use that first neural network to be like, give me coma.
43:45 Lester Nare? Right.
43:46 Krishna Choudhary? Data.
43:46 Lester Nare? Yes.
43:47 Krishna Choudhary? Right? And then now that genetic algorithm is going to tweak the parameters such that I get coma type data.
43:52 Lester Nare? Yes. And, and just, I just want to re-bring this back in because we have done the training and done the validation based off of a ground truth, real source of actual information.
44:02 Krishna Choudhary? Yeah, we have all of those 10 second long clips.
44:05 Lester Nare? Right, it, that, that's. Like I just, it's not making this from whole cloth.
44:11 Krishna Choudhary? No.
44:11 Lester Nare? Like I just, that's like a really important concept.
44:13 Krishna Choudhary? Yeah, and there's multiple stages of training. It's like, he's, he's figured out a way to, you know, you can't just tell a biophysical model, make me coma.
44:22 Lester Nare? Right, right.
44:23 Krishna Choudhary? You gotta first be like, no, no, no, let's make just a, a normal data. Give me data that can fool even me to thinking, I don't know if that's from a real patient or from my brain simulation. Right? I'm sure he did like checks just on his own.
44:36 Lester Nare? Yes.
44:37 Krishna Choudhary? Where like after the whole thing, I mean, you look at it, it's like, okay, that looks pretty good.
44:40 Lester Nare? Yes.
44:40 Krishna Choudhary? Right? And then you go, okay, now give me a coma. And he was successfully able to recreate known phenotypes, like on the right-hand side. We've got high voltage delta oscillations. This is something that happens in coma, like the, the slow sort of oscillations that are super high voltage. We've got burst suppression that's on the bottom, bottom side. You got these big bursts and then, and then just seconds of inactivity. Again, something that's there in coma patients. What's cool is there's no explicit programming of prior neuroanatomical knowledge on what is the difference between a coma?
45:13 Lester Nare? Yeah, yeah, yeah.
45:14 Krishna Choudhary? Versus a normal brain.
45:15 Lester Nare? Yeah, yeah, yeah.
45:16 Krishna Choudhary? Okay?
45:16 Lester Nare? Yeah.
45:16 Krishna Choudhary? Just the genetic algorithm.
45:18 Lester Nare? Yes.
45:18 Krishna Choudhary? And that detector that I had of coma versus conscious.
45:22 Lester Nare? Yes.
45:23 Krishna Choudhary? Just that and the genetic algorithm has now independently discovered some of the core tenants of that mesocircuit hypothesis, right? What did the mesocircuit hypothesis said? It said that if, if I weaken cortical drive coming in from the cortex.
45:37 Lester Nare? Yeah.
45:38 Krishna Choudhary? That's gonna cause a bunch of random crap.
45:40 Lester Nare? Yep.
45:41 Krishna Choudhary? That's what the, that's what.
45:42 Lester Nare? Yeah.
45:42 Krishna Choudhary? My, um, mo- model is doing. So as a first pass, it's already kind of on the right track. You can smell it in the water, right? Like there's, I'm on the right track because.
45:53 Lester Nare? Right.
45:54 Krishna Choudhary? The normal stuff that I kind of know about the mesocircuit that I hadn't baked in explicitly, it's already figuring that out.
46:01 Lester Nare? It is in, it is being, it is able to intuit.
46:04 Krishna Choudhary? Mm-hmm.
46:05 Lester Nare? Uh, what we've already done over this last five, 600 years that we talked about. Um, out of the box without, this is sort of what people say about benchmarks and AI models. It's like, well, if you train it on the. benchmarks that it's gonna do good on the benchmarks.
46:18 Krishna Choudhary? Yeah.
46:19 Lester Nare? And so the point here, it's not, the, the answers to the test were not provided as a part of training. And so it's getting answers that are not tainted by sort of giving them the cheat sheet beforehand.
46:33 Krishna Choudhary? Yeah, yeah, yeah. It's just looking at how do I make a coma and already it's recreating parts of the mesocircuit hypothesis that we know are probably true.
46:41 Lester Nare? Right.
46:42 Krishna Choudhary? Okay.
46:42 Lester Nare? Right.
46:42 Krishna Choudhary? So that's already impressive. The next part is the impressive part, which is what is the new stuff?
46:47 Lester Nare? I just want to say, this is all already very impressive. And just as another caveat, we're talking about Daniel specifically. This was obviously a team and this is incredible work by everybody on the team.
46:56 Krishna Choudhary? Yeah, yeah, yeah.
46:57 Lester Nare? Everybody on the team.
46:57 Krishna Choudhary? Daniel's our friend.
46:58 Lester Nare? So that's why we're talking about Daniel.
46:59 Krishna Choudhary? Yeah.
46:59 Lester Nare? All right, all right, so.
47:01 Krishna Choudhary? But clearly, yeah, this is, this is very, I mean, it takes a lot of data, it takes a lot of like, it takes a lot of brains to.
47:08 Lester Nare? It's always a team effort, guys. Every time we're talking, it's always a team effort.
47:11 Krishna Choudhary? Yeah. So let's get back to Daniel.
47:17 Krishna Choudhary? Okay, um, here's, here's what he's gonna do next, okay?
47:20 Lester Nare? Okay.
47:20 Krishna Choudhary? So the, the one part of the mesocircuit hypothesis, which is like cortical drive goes down, so the power station, like, the normal power grid is down. Or the, um, there's elevated firing in the palladian population, so that's gonna stop the, that's gonna inhibit the thalamus even more and things like that. That stuff is already corroborated. Stuff we already knew.
47:40 Lester Nare? Yes.
47:41 Krishna Choudhary? This thing is kind of confirming.
47:42 Lester Nare? Okay.
47:42 Krishna Choudhary? Okay. Now, does it do anything new? Well, yes. There's something called the selective disruption of the indirect pathway. This is a new prediction from the model. Here's what's happening. So it turns out, the prediction is that the coma is actually driven by the degradation of certain medium spiny neurons. Remember those neurons that I was telling you about that like, they just die if there's like any stress whatsoever?
48:10 Lester Nare? Or like, like, little bit of oxygen deprivation immediately.
48:14 Krishna Choudhary? Immediately. There's like, okay, no, I'm, I'm done.
48:17 Krishna Choudhary? There's selective degradation of a certain type of medium spiny neuron in the striatum that is projecting to the part that's putting on brakes to the thalamus.
48:27 Lester Nare? Okay.
48:27 Krishna Choudhary? Okay, and this very specific pathway is new. Okay, the classical mesocircuit hypothesis is just like, oh, there's like stridal dysfunction. Right, it's like the the whole brain region.
48:42 Lester Nare? Yeah, yeah.
48:42 Krishna Choudhary? Right, this thing is saying that the the indirect pathway, which is specifically this D2 receptor expressing medium spiny neuron, this very specific subpopulation in the striatum, that's the part that is getting weakened. And when that gets weakened, that's going to ultimately suppress the thalamus at the end of the day, like that thing that I was telling you about. But now it's identified a little part.
49:06 Lester Nare? It's not just the all of LA County. We've got it down to a street block.
49:10 Krishna Choudhary? Yeah, and a specific type of house.
49:11 Lester Nare? Right.
49:11 Krishna Choudhary? Type thing.
49:12 Lester Nare? Right.
49:12 Krishna Choudhary? Right, okay, so how do you find that's your prediction? Predictions are only as good as the data that corroborate.
49:20 Lester Nare? Right.
49:20 Krishna Choudhary? So how do we corroborate it? We use something called diffusion tensor imaging. So this is a advanced form of like MRI.
49:26 Lester Nare? Oh, this is beautiful.
49:27 Krishna Choudhary? Right, what we can look at is trace, this is very cool. What you can do with MRI is MRI is magnetic resonance imaging, right? Where you look at like hydrogen molecule, hydrogen atoms in whatever biological tissue, and you can actually see that using this magnetic resonance imaging technique. I'm not going to get into how MRI works, but effectively we can trace hydrogen atoms.
49:50 Lester Nare? Yeah.
49:50 Krishna Choudhary? Okay, water has a bunch of hydrogen atoms. Okay, so we can measure the directional diffusion rate of water molecules along the axons, along those fibers of those neurons, right? And we can measure in white matter, the water is going to move faster along the length of the axon than across. And so we have a directional idea of like how diffusion of water is going. And diffusion of water is a proxy for kind of how neurons talk to one another.
50:19 Lester Nare? Where is it? How's the traffic and flowing?
50:20 Krishna Choudhary? Right, yeah, exactly. So what they did. Was use this DTI diffusion tensor imaging of 51 patients with disorders of consciousness. And you can show that there's significantly lower striatum to GPE streamlines. That's the break on the thalamus.
50:38 Lester Nare? Yeah.
50:39 Krishna Choudhary? That prediction.
50:39 Lester Nare? Yeah.
50:40 Krishna Choudhary? Is lower in vegetative state versus minimal consciousness.
50:45 Lester Nare? No way.
50:46 Krishna Choudhary? Yeah.
50:47 Lester Nare? So just to take a step back, right? The, from this model, it made a prediction about where specifically in this region of the brain that is controlling the flow between the thalamus and the cortex.
50:59 Krishna Choudhary? Mm hmm.
51:00 Lester Nare? Is, is the point where the degradation, like the degradation of this specific area is what's actually driving the problem.
51:07 Krishna Choudhary? And this connection.
51:08 Lester Nare? This connection.
51:09 Krishna Choudhary? Is where we would find it.
51:10 Lester Nare? And so it said, this is, this is the road and the house where the problems like arise out of.
51:17 Krishna Choudhary? Yeah.
51:17 Lester Nare? And then we said, okay, well, we have all this actual MRI data from real patients. So can we look at this road, this house in real patients that the model predicted you would see degradation in and in an outsized ways compared to surrounding area?
51:35 Krishna Choudhary? Yeah. Yeah. I mean, no, they compared vegetative versus minimally cognit.
51:39 Lester Nare? Right, right.
51:40 Krishna Choudhary? Cause we've got those two sets.
51:41 Lester Nare? Yes.
51:41 Krishna Choudhary? And if there's a difference, there should be some statistically.
51:44 Lester Nare? Some delta between those two. The vegetative will have more of the degradation than the minimally.
51:49 Krishna Choudhary? Oh my god. Yeah, yeah. And so, and so on the, on the, on the left hand side, we see, um, we see that significance. There is a significant difference. On the right hand side, he did sneak this in. The P value is only 0.07. So it's not significant. But it's there, right? And perhaps if you were to pull the two data, it would become even more significant. But it's the, there's, there's at least something, right? And the N is low.
52:14 Lester Nare? Yeah.
52:14 Krishna Choudhary? Right? So obviously like significance is not, we're not going to get like Higgs boson 10 to the minus five significance here, right? Where literally to do that, they had billions of particles colliding with each other, right? So you'll never get the, the significance level of traditional like physics. But the effect it seems is there.
52:35 Lester Nare? Yep.
52:36 Krishna Choudhary? Right? And with more and more data, perhaps it's going to be an even bigger effect.
52:39 Lester Nare? Makes sense.
52:39 Krishna Choudhary? Right? It was, it was really kind of a proof of like minimally viable proof.
52:44 Lester Nare? Yeah, yeah, yeah.
52:44 Krishna Choudhary? That this hypothesis of this little.
52:49 Krishna Choudhary? Prediction that I have, it's probably true.
52:51 Lester Nare? It was good enough for nature neuroscience.
52:53 Krishna Choudhary? Yes.
52:53 Lester Nare? Not bad.
52:53 Krishna Choudhary? Exactly. The second thing that they did was predict this, this whole AI architecture predicted that there's increased synaptic coupling between inhibitory interneurons, meaning those negative feedback neurons that I was telling you about, these are sometimes called fast spiking PV+ neurons in the cerebral cortex. Those negative feedback neurons in our cerebral cortex have a lot of coupling in between them. So it's like the negative feedback is coupled to the negative feedback, which is coupled to the negative feedback, right? That is what the AI model is predicting. So to test this, you go to transcriptomics. Transcriptomics means
53:33 Krishna Choudhary? I'm going to now read the mRNA that is in my cerebral cortex. Okay. And I'm going to see what genes are being expressed.
53:43 Lester Nare? Yes.
53:43 Krishna Choudhary? In patients that have vegetative state.
53:46 Lester Nare? Yes.
53:46 Krishna Choudhary? And patients that don't.
53:47 Lester Nare? Yes.
53:47 Krishna Choudhary? Okay. Healthy versus those in coma.
53:49 Lester Nare? Yes.
53:50 Krishna Choudhary? What they find is there's an upregulation of two genes, the VGF and the SCG2. These are genes that when expressed in these interneurons, they drive synaptogenesis, meaning they drive this feedback mechanism, this coupling.
54:08 Lester Nare? Yeah, yeah, that.
54:09 Krishna Choudhary? Right?
54:10 Lester Nare? This is.
54:10 Krishna Choudhary? This one's, this one's pretty significant.
54:12 Lester Nare? Yeah.
54:13 Krishna Choudhary? This one's, this one's way more than, in both those genes.
54:16 Lester Nare? Yeah, yeah, yeah.
54:16 Krishna Choudhary? Right? And this is again, data that's out there.
54:20 Lester Nare? Right, right, right.
54:21 Krishna Choudhary? Right?
54:21 Lester Nare? Exactly. And I think the, this is kind of, again, in terms of like a frame of reference for, I, I've seen the conversation around quote unquote AI be very different within some corners of the science community versus the general public, because a lot of researchers view it in this way, where it's like, okay, this can be a sort of intermediary layer where I can rapidly prototype and generate predictions and rapidly be able to test those, test that against real data in a way where I don't have to abuse my postdocs.
54:56 Krishna Choudhary? Mm hmm. Yeah, yeah, yeah. Just, just abuse the underbreds.
55:03 Krishna Choudhary? Here's a thousand images labeled.
55:07 Lester Nare? Right, no, right.
55:08 Krishna Choudhary? Sorry, undergrads are always gonna.
55:09 Lester Nare? That's, that's tough.
55:10 Krishna Choudhary? Yeah, it's a tough life. We've all been there. We all want those letters of recommendation.
55:16 Krishna Choudhary? You gotta, you gotta work the sweat.
55:17 Lester Nare? This is really interesting though, because now there's, the predictions are also on two very different planes.
55:25 Krishna Choudhary? Yes, that's, that's a very good point.
55:26 Lester Nare? You know what I mean? Like in terms of what's?
55:28 Krishna Choudhary? Yeah, because when we think about those three, those three components of our city, right? The surreal court cerebral cortex is the homes and the businesses. That PV plus interneuron thing has to do with that part.
55:41 Lester Nare? Right.
55:42 Krishna Choudhary? And then, and then the, the, the brain imaging part.
55:46 Lester Nare? Yes.
55:46 Krishna Choudhary? That one had to do with the thalamus and the relay station.
55:50 Lester Nare? Correct.
55:50 Krishna Choudhary? You know, so all three parts of the mesocircuit hypothesis are working together. And this thing is giving predictions, as you said, on all of them.
55:57 Lester Nare? Right.
55:57 Krishna Choudhary? Right?
55:58 Lester Nare? Right. In a very interesting ways. And again, this, this continues to go back to the fact that we have all these existing tools like transcript, transcriptomics.
56:10 Lester Nare? Enables the ability to know what genes are being expressed.
56:14 Krishna Choudhary? Yeah, and one thing I just want to say is like to the team that, that made this paper happen, right? I admire the resourcefulness. Right? Because it's one thing to have an AI model and make predictions.
56:26 Lester Nare? Right.
56:27 Krishna Choudhary? And it's another thing to think, okay, here are the predictions. How can I make the argument that this is real?
56:34 Lester Nare? Right.
56:35 Krishna Choudhary? Right? How can I check? It's always about checking.
56:37 Lester Nare? Yes.
56:37 Krishna Choudhary? And they were so resourceful that they found these kinds of data sets.
56:41 Lester Nare? Yep.
56:41 Krishna Choudhary? You know, maybe they talked, I don't, I don't know exactly the details. I don't know if these are open or not, but you, you go talk to somebody who has that data set.
56:47 Lester Nare? Right.
56:47 Krishna Choudhary? You meet someone at a conference and they're like, hey, actually you can look into mine. You know, put me as an author.
56:54 Krishna Choudhary? You know, so like.
56:55 Lester Nare? Very, yeah, it's very clever.
56:57 Krishna Choudhary? Yeah, I, I think that part is nice.
56:59 Lester Nare? The whole from the, from the ideation of like, let's give it a try. To then, how are we gonna be able to prove to reviewer two that this is not all nonsense?
57:10 Krishna Choudhary? Yeah, yeah, exactly. I got, I got lunch with Daniel about three weeks ago. And he was telling me about this paper. And he, he told me like, and you know, the thing predicted this. And the first thing I asked, well, yeah, but like, you know. How do you?
57:22 Lester Nare? How do you know?
57:24 Krishna Choudhary? The model isn't just bullshitting. And then he told me about these two techniques. I was like, okay, that's actually pretty dope.
57:29 Lester Nare? Yeah.
57:29 Krishna Choudhary? All right, well, as soon as it's out.
57:30 Lester Nare? We'll cover it.
57:31 Krishna Choudhary? Yeah. You know, um, so, yeah, this was, this was pretty cool. Okay, the final thing we're going to talk about.
57:38 Lester Nare? Yeah.
57:38 Krishna Choudhary? Is a proposed strategy for awakening patients.
57:42 Lester Nare? Okay.
57:42 Krishna Choudhary? With coma, right?
57:43 Lester Nare? Right.
57:43 Krishna Choudhary? Cause at the end of the day, that's what really matters clinically.
57:46 Lester Nare? That's how we started.
57:47 Krishna Choudhary? That's how, that's how it started and that's how we're going to end it. So the intervention that they looked for and how to test was deep brain stimulation. This is again, the idea of you, you test, um, you, you put in an electrode deep into the brain and then you give it electrical activity and you try to wake up the neurons. Okay. Now, the researchers here, they, what they did was test a bunch of targets in their biophysical model and say, what if I provide stimulation here? What if I provide stimulation here? What's the best target in my three component model? Right? Cause there's a bunch, there's cortex, there's thalamus, there's subthalamic nuclei, there's the palladium and each of those has their own little subpopulations.
58:32 Krishna Choudhary? So we can get really granular, granular now, right? We can get really granular and try to think, what is the best target?
58:39 Lester Nare? This is so good.
58:40 Krishna Choudhary? They settled on the subthalamic nucleus. It had overwhelmingly standout result for consciousness recovery in their model.
58:49 Lester Nare? Okay.
58:49 Krishna Choudhary? Okay. Then again, they tested this with humans. They tested 130 hertz stimulation in the subthalamic nucleus in six awake human patients. And, um, the cortical CNN, the, the cortical consciousness detector that they had previously trained, detected a significant shift towards optimal consciousness.
59:14 Lester Nare? This is crazy.
59:15 Krishna Choudhary? With only stimulation in the STN. If they went anywhere else, only the subthalamic nucleus showed this effect. If they did it in the cortex, they did it in the palladium, didn't show any significant increase in consciousness.
59:27 Lester Nare? Yes.
59:28 Krishna Choudhary? And that's what the model predicted.
59:29 Lester Nare? So are you saying we found the boss's office?
59:32 Krishna Choudhary? Yeah, it seems we found.
59:34 Lester Nare? We may have found the boss's office.
59:35 Krishna Choudhary? Where if you turn the lights on and off, the boss will wake up.
59:38 Lester Nare? Yeah.
59:39 Krishna Choudhary? Right?
59:41 Lester Nare? This is, I'm so mad because it's so clever. Each of the three layers, also in how we went through it, of like detection, characterization and evaluation almost is like, is, is so good.
59:54 Krishna Choudhary? No.
59:55 Lester Nare? Because I think the, the, the part that's getting me about this piece is because you've set up like the genomic algorithm and the, the model, the brain simulator, the way they thought to set it up that way means that you almost have this like alpha fold adjacent kind of thing where you can in silica like test.
1:00:18 Krishna Choudhary? Yeah.
1:00:18 Lester Nare? Stuff to try to narrow the sandbox of where to look.
1:00:24 Krishna Choudhary? Exactly.
1:00:24 Lester Nare? Right? And like, that's so valuable.
1:00:27 Krishna Choudhary? Yeah, yeah, exactly.
1:00:28 Lester Nare? It doesn't necessarily have to be the exact answer.
1:00:30 Krishna Choudhary? No.
1:00:30 Lester Nare? But if it's even somewhat directionally correct, that's unbelievably valuable.
1:00:35 Krishna Choudhary? Yeah, dude, it's great. And one thing that, one thing that's kind of funny is, so Daniel, he, he is himself a science communicator, right? So he's got an Instagram and a TikTok. And he did like a short three minute video being like, hey, my paper came out and he kind of explained it at a very high level. I mean, here we've gone really in depth. So at that high level, he was describing his generative AI framework. And there was someone in the comment who was like, that's not really AI, right? Because you're using a biophysical model. And it's like, dude, that's the point. That's the point. It's like without a biophysical model, I wouldn't be able to make these predictions of, oh, it's the PV plus neurons in the cortex. It's this highway in between the striatum and the palladium.
1:01:17 Krishna Choudhary? It's the subthalamic nucleus that we need to probe, right? It's just going to be some random neuron in some random layer of a trillion parameter model. What good is that?
1:01:27 Lester Nare? The, the, a hundred percent, I still come back to one of the, there's so many interesting insights in this entire paper's architecture, like the experimental design architecture. But I think a key piece was, uh, you, you can get traceability and then subsequently reproducibility because of that brain simulator.
1:01:50 Krishna Choudhary? Yeah.
1:01:52 Lester Nare? Like, and the way in which it is explicit, it is, um, what's the, there's the terminology, and this may not be correct, there's this terminology where probabilistic versus deterministic, and it's more on the deterministic side and less on the probabilistic side, which in science, it's a good thing.
1:02:12 Krishna Choudhary? That's a good thing. Yeah, it's very much a good thing. And, right, and then finally, we now have an idea that it should be the subdylemic nucleus that we target, which is this tiny, like, lentil-sized structure, right? It's actually also quite incredible, if you think about it, like, that is the boss's office, right? For at least stimulation of the brain. Consciousness could be a distributive thing, but just stimulating that tiny structure of maybe, let's say, hundreds of thousands of neurons at the max is going to wake up a brain that is 86 billion neurons, right? Consciousness is a crazy thing, and we're getting closer.
1:02:52 Lester Nare? So with actually a correct analysis, because there's going to be probably some thoughts about, oh, like, consciousness is, like you discussed at the beginning, a very expansive thing, and really, all we're identifying in this last piece is, I keep using the grill lighter example, this is, might just be the igniter, the little button you press on the grill, to turn consciousness on.
1:03:16 Krishna Choudhary? On, yeah.
1:03:16 Lester Nare? But that's, that's, that's all we're saying.
1:03:19 Krishna Choudhary? I don't know what the grill being on means.
1:03:21 Lester Nare? Right, we don't know how big the grill is, is it a black, black stone? Is it, you know, whatever.
1:03:27 Krishna Choudhary? Exactly.
1:03:27 Lester Nare? We're just saying, we may have found where we can wake
1:03:30 Krishna Choudhary? Yes.
1:03:31 Lester Nare? An unconscious, from this vegetative or unconscious state.
1:03:35 Krishna Choudhary? Yes.
1:03:36 Lester Nare? And the interesting dovetail of how the 2007 paper we talked about, that identified this idea of minimally waking consciousness, was actually a key step in order for this to actually also work? Because that was where we started to zoom in and be able to validate the mesocentric model, and it's just like, everything... Has to.
1:04:03 Krishna Choudhary? Yeah, one after the other.
1:04:04 Lester Nare? It has to stack.
1:04:05 Krishna Choudhary? Right, yeah.
1:04:06 Lester Nare? In order for this to even be possible.
1:04:08 Krishna Choudhary? Yeah.
1:04:08 Lester Nare? This is really quite nice.
1:04:09 Krishna Choudhary? It was a good paper. He's, Daniel does a lot of really cool work with coma. Um, he's actually also in an organoid lab at UCLA. So one of these days, you know, we'll, we'll come by UCLA and see your lab, Daniel.
1:04:22 Lester Nare? Yes.
1:04:22 Krishna Choudhary? But until then, nicely done.
1:04:24 Lester Nare? Very, very nicely done. Obviously, the, um, this is early.
1:04:30 Krishna Choudhary? Yes.
1:04:30 Lester Nare? But there's a huge, the, the, the unlocks that are there, you know, you could talk about it forever, just in terms of not only the clinical or medical context, but even for the philosophy of mind kind of stuff. Having some mechanistic details.
1:04:46 Krishna Choudhary? Yes, that's, that's big. And then, and I think, I think the part that is going to be most influential for this paper is actually the,
1:04:58 Krishna Choudhary? the, the way that he went about using this biophysical model.
1:05:02 Lester Nare? Yes, I totally agree.
1:05:03 Krishna Choudhary? And tweaking, right? That's a, that's a very interesting way to do things.
1:05:07 Lester Nare? I totally agree.
1:05:07 Krishna Choudhary? And I can already imagine it being used for all sorts of stuff, material science.
1:05:15 Lester Nare? Yes, yes.
1:05:16 Krishna Choudhary? Like physical things, you know?
1:05:17 Lester Nare? Yes.
1:05:17 Krishna Choudhary? Like, we've got really good models about how atoms work with each other.
1:05:21 Lester Nare? Exactly.
1:05:22 Krishna Choudhary? Make a GAN, genetic network type thing, you know, like that is the part that is really,
1:05:27 Lester Nare? Yep.
1:05:28 Krishna Choudhary? Like, I think mechanistically very interesting to me. The methods is very nice. I mean, obviously, the results are incredible, but to me, the, the one I'm most impressed about is the way that he used this biophysical model.
1:05:40 Lester Nare? As someone who's coming more from the technology and software world, it is naturally what I'm able to better engage in at a, at, at a deep level.
1:05:51 Krishna Choudhary? Mm-hmm.
1:05:51 Lester Nare? Um, but I agree with you. I think they've basically created a conceptual framework around how to design.
1:06:03 Lester Nare? ML and AI architecture as a, as an end to end system and process that is subject matter agnostic.
1:06:15 Krishna Choudhary? Mm hmm. Yeah, yeah.
1:06:16 Lester Nare? Like, it could work.
1:06:17 Krishna Choudhary? Yeah, yeah, yeah. That thing can work.
1:06:19 Lester Nare? And the whole variety of, the brain was just the first target. But the exact same conceptual framework could work in a variety of different areas.
1:06:27 Krishna Choudhary? Yeah.
1:06:27 Lester Nare? Particularly the data feedback loop and validation process, et cetera. Really, really, I just, this was great.
1:06:33 Krishna Choudhary? Yeah, good stuff.
1:06:34 Lester Nare? Well done, Daniel Toker et al. Again, this was in Nature Neuroscience on March 24th. Fresh, hot off the presses. Again, there is nowhere else on the internet. Not even on the direct pages of the authors themselves where you're going to get this level of excitement, passion, and deep dive. They need a break. Daniel needs a break anyway, so he doesn't need to, like, it's our job to do this and step in.
1:06:58 Krishna Choudhary? Yeah, we'll see what he thinks about it.
1:07:01 Lester Nare? You actually guys did pretty well, did pretty well. No, just another fantastic paper. And we, we just really want to thank those of you who have stayed this long and listened to the episode for staying with us and enjoying this journey of curiosity and discovery. I will do a brief pause to ask if we would like to do a comment.
1:07:23 Krishna Choudhary? For the audience. Yeah.
1:07:26 Krishna Choudhary? GAN.
1:07:27 Lester Nare? GAN.
1:07:27 Krishna Choudhary? Oh, yeah, yeah. This has become our shtick.
1:07:29 Lester Nare? That's, if we can't think of anything, yeah.
1:07:31 Krishna Choudhary? What, what's, what's another alternate?
1:07:35 Krishna Choudhary? Full form of GAN.
1:07:37 Lester Nare? Yeah.
1:07:37 Krishna Choudhary? G A N.
1:07:38 Lester Nare? For, for.
1:07:39 Krishna Choudhary? Generative adversarial networks. Come on, you guys can think of something better.
1:07:42 Lester Nare? Something great. We really appreciate you all. My name is Lester Nare. Joined as always by my co-host and our resident PhD Krishna Choudhary. This is the last reminder I'll make on this. We are moving to single story episodes multiple times a week. We have gotten confirmation from our deep listeners that you do appreciate it and you like it. And there was a good idea that we will think through. I don't know if, actually, I don't even know if you saw this.
1:08:08 Krishna Choudhary? Yeah, yeah, yeah.
1:08:09 Lester Nare? The, the rundown. When we do the rundown at the end of the week, we don't really go in depth. And there's a lot of folks who would like us to go in depth on any number of those particular stories. So. We will look into having basically a patron system to vote on what rundown story should we follow up on for a deep dive. It's a great idea. Excuse me for not remembering the commenter who said it, but you can comment again. We do see it. Thank you so much. We will see you all later this week.
Previous AI Cancer Vaccines, Strange Fish, Ketamine, and Ancient Life (EP. 34) Next Artemis II, Claude Code Leak, iPhone Spyware & Project Hail Mary (EP 36) Page updated October 30, 2026 · transcript and summary are machine-generated