Invest Like the Best with Patrick O'Shaughnessy
00:31:09 · Andrew Huberman
A guy that came up, not mainly through my lab, but he spent some time in my lab, he's now the head neurosurgeon at Neuralink, Matt McDougall, amazing guy, has a little receiver implanted under his skin of his hand between his thumb and his index finger. And his wife does too, and it opens their door to their home, to keys. Like, I was like, what are you doing? He's like, it's a sort of biohacking Bay Area ish kind of stuff. He's like, yeah, like, I'm just playing with the idea of what is it like to have a small device under your skin that can do stuff in the world for you. That's very neural link ish, and obviously they have much bigger aspirations than that. But I think within twelve months, 20% of the entrepreneur real world will be asking, if it's only as invasive as getting a piercing, yeah, maybe I want a little vagal stimulator. And that's coming next. People are thinking like peptides, what are we gonna do? I'm not worried about peptides. I'm thinking implantable devices is where it's gonna go next. The commercially available vagal stimulators, some have shown some promise. They're not super impressive. To my neurosurgeon colleagues who like to get it under the skin and under the skull. But I will say there's some interesting results there, but we're not quite there. But think about ten years ago, if I told you, listen, you're going be wearing a ring or a band on your wrist that measures your sleep and HRV, that's pretty out there.
00:32:25 · Patrick O'Shaughnessy, Andrew Huberman
[00:32:25] S0: I'm so fascinated by this. So there's reading the underlying activity of the brain at different levels of fidelity. There's writing. You talked about ultrasound stimulation, I think there's electrical, I think there's magnetic stimulation potentially of these areas of the brain. One central part that I don't really know much about is the model of the brain itself. [00:32:43] S1: How [00:32:44] S0: far along in our understanding of that we even are. I know there's some things that we can say activation of this area produces this thing. But like, it feels like we need those three things, like good input modalities, good reading methods, and a model. [00:32:55] S1: Yeah. I'm glad you're asking. Yeah. I mean, these are my favorite topics. It's just that usually people are like, what peptide should I take? [00:33:01] S0: Which peptide should we take by the way?
00:33:03 · Andrew Huberman
[00:33:03] S1: I mean, it depends on what your pain points are. So at the periphery, the retina, the cochlea, the tongue, the olfactory system, the skin, neuromuscular control, which isn't at the periphery, but I think we know pretty darn well how the nervous system works. And we know how different wavelengths of light are transformed into visual images at a, not even a first approximation, but we understand that in significant detail, how motion is processed. We understand a fair amount about memories, although I want to put that on the shelf for a moment because there's an important finding in the memory field that will make you feel like it's in kind of twilight zone, perhaps in neuroscience. It worries me all the time. The hypothalamus. So circuits that are deep in the brain that control hunger, anger, sexual appetite, lack thereof hormone secretion. Those are switches. So do we understand those? Yes. Do we understand everything? No, but we understand them. I mean, one of the reasons the GLPs are so effective and one of the reasons they got so much velocity is because we understand where they bind in the brain and they bind to these hypothalamic and other areas, of course, that activate neurons that are involved in the suppression of appetite. And so it's a very simple equation, really. Ramp these GLP levels up a thousand fold, people aren't as hungry. But it turns out they're not as driven for other things too if you get beyond a certain level. I think Sam Altman talked about his experience with that at one point, and why he decided to either back off the dosage or stop taking them. So as you move in from the periphery of the sensory organs, right? And as you move up from the base of the brain, the deeper limbic structures and hypothalamic structures to the prefrontal cortex, we can say, yeah, like your prefrontal cortex, know based on lesion studies in humans, activation studies is really important for contextual learning, for strategy setting in different environments, for suppression of impulse, what we call top down suppression. So do we understand what the prefrontal cortex is doing? Yeah. The place where it starts to get murky is at the level of memory. We'll talk about that in a moment. And at the level of thoughts, you won't find a chapter in a neuroscience textbook that says, this is how thoughts are made.
00:35:10 · Andrew Huberman
And we can talk about what's starting to emerge there. These dynamic attractor states, we'll get to those in a moment, but here's the eerie results from the study of the hippocampus and memory that freak me out and a lot of neuroscientists don't want to talk about this because it throws all the read write stuff that we were talking about earlier, all the technologies and pharmacology into a complicated place because it doesn't allow for clean matching of conceptual understanding of the brain from experiments to levers that actually allow us some control. The experiments are like this. Mark Mayford's lab has done this, so down at Scripps, Susumu Tonogawa, who won a Nobel for his work on immunoglobulins, but then became a very successful neuroscientist. They've done these experiments where they have an animal do some sort of learning task. Could either learn that they get shocked over here and not there, or they get a reward over here and not there, or they learn a maze, or they learn a lever press combination that gets them a reward, you know, they can learn pretty sophisticated things for a mouse. Monkeys can do this too. Humans can do this too. Because ways to read from the brain, including the hippocampus, a deep ish brain structure, they've gotten pretty good. You can say, okay, these are the patterns of signals that happened as the animal learned. And here are the patterns of signals that occur every time the animal expresses this thing. We imagine there's a correlate where every time somebody serves a tennis ball the way they perfected it, that the same cluster of neurons is activated in the brain, They're communicating with each other, the same circuit rather, and in a particular sequence, A leads to B leads to C, and so on, and you get the tennis serve. The data tell us that if you tag the neurons that were involved in this behavior and then reactivate them, you expect the behavior, right? And indeed, that's what you find. Hypothetically, you serve the tennis ball. I'm able to measure, let's just say with the highest degree of fidelity that the patterns of activation from start to finish that occur is you toss the ball in the air and you serve. Great. Now I stamp those neurons with a chemical tag. This has been done. Then I come back later and I activate those neurons in the same sequence and you do the serve perfectly. I say, great.
00:37:18 · Andrew Huberman
But then they did the important control experiment. What if I activate those neurons in the opposite sequence? What if I activate all those neurons at once, like banging on the keys of the piano? It's not the same as playing the song. If I just say, what are all the keys involved in this motif of music? And instead I just bang on them all once you get a very different sound coming out of the piano. It turns out you get the exact same behavior. This is problematic for the field of neuroscience because what it says is that this notion of these circuits firing in a particular sequence, at least for memory, it's probably not true. So now if I want to write to your nervous system, maybe I'm better off because all I have to do is activate these same neurons. And it doesn't matter what sequence I activate them in. That's kind of cool, right? Like you want someone to walk, maybe you just need to activate the brain areas involved in walking, and the body and the other brain circuits will figure it out. But here's the problem. First of all, it's very hard to activate these select sets of neurons with high spatial and temporal fidelity to begin with. But the other problem is these neurons, I didn't tell you, are all also involved in lots of other perceptions and behaviors. So it's not like you have a circuit for serving a tennis racket that isn't also used for some aspect of thinking and remembering about something from your childhood. I mean, that's one of the beauties of the nervous system is it repurposes different neurons to participate in different circuits, depending on the perceptions and behaviors that you're experiencing, that you're engaging in. Our model of the brain works at the level of the periphery, at the deeper brain structures, but once you start getting into memory and thinking, it's complicated because we don't yet have a good model of whether or not the temporal ordering of firing of neurons matters. And every textbook says it does. And the flow from like eyes to the thalamus up to the cortex to create a visual perception, we know that direction of flow is critical. If you cut that off at the level of this deeper brain structure of the thalamus, there's no perception. We know this from human patients, but memory thinking brain plasticity, these things all rest on the assumption that the temporal order matters. And apparently it doesn't.
00:39:27 · Andrew Huberman
And that's one of the unspoken like within our field And people are trying to understand that and what it means, but it's going to present a serious problem for the neurotechnologies. Again, it's going to alleviate some problems because maybe you don't need as much spatial fidelity. But in my mind, this is a huge gap in understanding. And then the other piece is that we don't really understand what a thought is. And so if you're gonna get in there and start stimulating neurons in even just motor cortex to generate movement without the acknowledgement that these neurons are extensively interconnected with structures involved in thinking and planning, you could send the system into haywire. That said, if there's a goal in mind, like drink the glass of water, the brain generally figures it out. And that's what's incredible. I mean, there are people working on brain damaged patients that are finding that if you just go to the areas of the brain where the signals are generated to inform the action output ends of the brain, then you can do pretty well. That's how my friend Eddie Chen gets people with locked in syndrome to speak through a computer. He's not measuring from the speech production areas. He's carefully characterized for many years, decades, the signals coming out of the speech planning areas. And because the signals can't get down to the pharynx and larynx, because speech is just the pharynx and larynx controlling exhales, what he's doing is just transforming those signals into electrical signals, so that the person can talk through a computer. So there's advantages and disadvantages to this non specificity. I think that's the big one. That's the really big one.