Wet vs Dry Electrodes: Interactive BCI & EEG Simulation

Okay, we'll be reviewing this new application. You can try it yourself on bionykills.com/electrodes. It's covering different electrode interface modalities. In this particular simulation, you can control the cognitive state from deep sleep relaxed to active. Can simulate the EMG noise. And choose different types of electrodes. And we'll go for this overview by Notebook LM. Well, it's actually doing a demo for the application and what it can do. >> Imagine uh trying to eavesdrop on a whisper, right? But this whisper is happening inside a completely soundproof bank vault. >> Oh, wow. >> And you are standing outside the vault just trying to listen in. But also, you're standing right next to an active, incredibly loud jackhammer. >> That is a that's quite the acoustic nightmare. >> Right, it's an absolute nightmare. But that is basically what engineers are dealing with every single day when they attempt to, you know, read human thoughts from the outside of the skull. >> Exactly, it's a huge problem. >> exactly what we're going into today. So, welcome to this deep dive. You aren't just listening to us talk through a stack of dry research papers today. >> Definitely not. >> No, today we are acting as your personal technical guides. Because right now, you are looking at a live, highly interactive neuroengineering dashboard right there on your screen. >> Yeah, the uh the Bionyk Chaos laboratory interface. Bionic Chaos, I mean, what you have in front of you is a real-time biophysical simulation of an EEG system. >> system, right? >> Exactly. We are looking at a fully functional brain-computer interface simulator. And by the end of this deep dive, as we, you know, walk you through this dashboard, you are going to understand the actual physics and the biology of how different sensors and placements, and even the subject's state of mind dramatically alter those electrical signals we extract from the human brain. >> So, let's get you oriented because at first glance, I mean, this dashboard looks like the cockpit of a fighter jet. >> It's a lot. There is a lot going on. >> There's so much going on. >> Mhm. >> So, let's just start by directing your eyes over the left side of your screen. You've got this top-down kind of wireframe model of a human head. >> Right. >> And it's covered in this crazy web of glowing dots. >> Yeah. So, those glowing nodes, they actually map out the international 10-20 system, which in clinical neurology and neuroscience, I mean, this is the absolute gold standard for where you place electrodes on a human scalp. >> Okay, because I'm looking at the labels on these nodes and they have names like FES and O1. >> Yeah. >> But, I'd imagine mapping the human head is just notoriously difficult. >> Well, incredibly difficult. >> Because you've got a toddler's head, an adult's head, entirely different bone structures. I mean, you can't just hand a doctor a tape measure and say, "Hey, put the sensor exactly 3 in above the left ear." >> Right, because that 3 in means something completely different depending on the actual patient. >> Exactly. >> Which is why the system doesn't use absolute measurements at all. The the 10 and 20 in the name actually refer to percentages. >> Oh, percentages, okay. >> Yeah, the whole grid is based on proportional distances between fixed anatomical landmarks on the skull. So, for instance, a technician will take a tape measure and find the total distance from the >> Which is the bridge of the nose, right? >> Right, the bridge of the nose. And they measure all the way back to the inion, which is that little bony bump at the very base of the back of your skull. >> Okay, I see. So, if you place a sensor at exactly like 10% or 20% of that specific patient's total measurement, the grid just dynamically scales. >> It stretches to fit the map perfectly, yes. >> That's brilliant. It's like a dynamic GPS grid. That means FES is always going to sit dead center over the frontal lobe and O1 is always going to anchor right over the occipital lobe in the back. >> Precisely. >> Whether the patient is, you know, a massive NFL linebacker or a premature infant in the ICU. >> Exactly. It creates this universal coordinate system. So, if a researcher in Tokyo publishes a paper saying, "Hey, we detected a specific anomaly at electrode Fes." A surgeon in Toronto knows the precise square centimeter of the cerebral cortex they're talking about. >> Wow. Okay, that makes perfect sense. >> But, to understand what that anomaly actually looks like, let's pull your attention over to the right side of the canvas now. You should see an oscilloscope. >> Yeah, I see it. It's the the black box with all the green traces jumping around. Looks a bit like a hospital heart monitor, but I mean the lines are way more erratic, really jagged. >> Yeah, they are. And what you are witnessing on that oscilloscope is not just abstract data. Those specific voltage fluctuations are the synchronized electrical firing of literally millions of pyramidal neurons in the cerebral cortex. >> Wait. Rendering in real time? >> Rendering in real time, right in front of you. >> Okay, pyramidal neurons. So, these are the specific brain cells actually doing the heavy lifting of our conscious thought. >> Yeah, they are the primary excitatory neurons in the cortex and they're named that because their cell bodies literally look like tiny pyramids. >> Okay, that's a cool visual. >> But, you know, their shape isn't actually the most important part for our dashboard here. It's their orientation. >> Their orientation? >> Yeah, evolution basically arranged these particular neurons so they are lined up completely perpendicular to the surface of the brain. So, they point straight up toward the skull, kind of like trees growing in a forest. >> Wait, does the orientation actually matter for the electrical signal? Like if they were just a tangled random web of cells, wouldn't they still produce electricity? >> They would, absolutely, but we would never ever be able to read it from the outside. >> Why not? >> Well, think about it. If the neurons were facing all random directions, firing off their microscopic electrical charges, those highly localized electromagnetic fields would just cancel each other out. >> Oh, because they're opposing each other. >> Right, they'd cancel out before they ever even left the tissue. But because millions of these pyramidal trees are all standing parallel to one another, pointing the exact same way, their tiny individual charges physically summate. >> Meaning they add up. >> Exactly. They stack together into a macro field that is just large enough to push through the layers of tissue and bone so that our surface sensors can catch it. >> That is just wild. It's kind of like um like a massive stadium full of people holding a flashlight. >> Okay. Yeah. >> If they're all just pointing their flashlights at each other in completely random directions, from far away, it's just this diffuse blurry glow. But if everyone points their flashlights straight up into the night sky at the exact same second, >> then from a satellite in space, you see one giant unified beam of light. >> Exactly. >> That is a fantastic way to visualize it. And the rhythm of those flashes, like the frequency and amplitude of that unified beam, tells us an incredible amount about the cognitive state of the brain in that specific moment. >> Which means we finally get to the really fun part of this dashboard. We know where the signals are coming from now, so let's actually manipulate them. >> Let's do it. >> Okay, look down at the bottom right corner of the UI, everyone. You'll see a control panel labeled the neural dynamics engine. We are going to change what our digital subject is currently experiencing. Let's shift the simulation into deep sleep. >> Okay, keep your eyes on the green lines on the oscilloscope. And engaged. >> Whoa. Look at that shift. The traces on the screen just completely, fundamentally altered. The jagged spikes just smoothed out into these massive, slow, rhythmic waves rolling across the entire grid. >> Yeah, it's a dramatic change. >> like deep ocean swells, like really tall hills and very wide valleys. >> You are looking at delta waves. They operate at a very low frequency, typically between like 1 and 4 hertz. >> Meaning that wave only cycles 1 to 4 times per second, right? >> Exactly. And biologically, this is the absolute hallmark of stage three deep restorative dreamless sleep. What is happening in the cortex right now is that the brain has largely disconnected from processing external sensory input. >> So the neurons just aren't reacting to sights or sounds in the room. >> Right. They are basically offline from the outside world. And because they aren't busy calculating or reacting to anything, millions of neurons just fall into this shared rhythm dictated by deeper brain structures. They're resting together, then firing together, then resting together. >> Oh, I see. >> It's an intense, completely unified synchronization, which is why the physical wave on your screen is so tall and powerful. >> Okay. Well, let's see what happens when we disturb that peaceful rest. I'm going to use the engine here to wake our subject up instantly, and we are going to assign them a complex multi-step calculus problem. >> this should be good. >> So watch those big slow ocean waves right now, and switch. >> There it goes. The signal transforms immediately. >> Okay, wait. Hold on a second. The guy just woke up and started doing intense calculus. And the signal just flatlined into tiny static. >> Yep. >> Did the simulation just crash? Because it looks completely broken. I mean, when the subject was asleep, the waves were huge and powerful. Now that the subject is awake, hyper-focused, and doing heavy cognitive lifting, the big waves are just gone, replaced by this chaotic low amplitude buzz. >> It is a bit shocking to see at first. >> I feel like this is backwards. Shouldn't a brain doing complex math produce a much larger, more powerful signal than a brain doing absolutely nothing? >> It seems totally counterintuitive, I know. But you have to go back to your stadium analogy. When the brain is in deep sleep, it's like the entire stadium of neurons is doing the wave together. Everyone stands up and sits down at the exact same moment. >> Right. And that mass synchronization creates a giant visible signal from space. >> Exactly. But, what happens when you wake the subject up and ask them to solve calculus? >> Well, I guess the stadium stops doing the wave, cuz, you know, people have to actually get to work. >> Precisely. The visual cortex starts rapidly decoding the numbers on the paper. The prefrontal cortex starts holding all those variables in working memory. The parietal lobe starts processing the spatial relationship of the complex equation. >> Now, I see where this is going. >> The unified stadium basically fractures into thousands of highly specialized groups, all firing incredibly fast, producing what we call beta and gamma waves. >> But, they're all firing at completely different rhythms now. >> Right. And because these different cortical networks are now totally out of sync with one another, their electrical fields are no longer stacking perfectly. >> Gotcha. >> In fact, on a macro level, they start canceling each other out. That is called desynchronization. So, that chaotic, tiny, noisy trace on your oscilloscope right now. >> Yeah. >> That isn't a broken signal at all. That is the beautiful, messy visual representation of active human thought. >> That is amazing. Okay, so that actually brings us to a massive engineering wall, though. If an active thinking brain produces a signal that is microscopic, totally chaotic, and actively desynchronizing itself, how do we ever capture it reliably enough to control a machine? >> It's the million-dollar question. >> Because, frankly, if the signal is this tiny, why isn't the whole world wearing brain-reading hats right now to, I don't know, type on our phones? >> Well, because reading these microvolt-level signals from outside the head is just a physics nightmare, and the primary culprit for that is evolutionary biology. >> skull. >> The human skull. I I from a biological standpoint, the skull is an absolute marvel. It is this dense, calcium-rich armor perfectly designed to protect our most vital organ from blunt force trauma. >> Right, it keeps us alive. >> But from an electrical engineering perspective, calcium phosphate is a highly effective electrical insulator. It completely scatters, absorbs, and blocks the weak electrical fields generated by the brain. >> So, you have these microscopic whispering neurons trying to shout their complex math problems through a literal brick wall of bone. And on top of that, dead skin. >> A very thick brick wall. >> To get the clean signal you are currently seeing on the oscilloscope, we clearly have to optimize the point of contact. So, direct your eyes to the panel on the screen labeled placement and invasiveness. Right now, the simulation is running the medical clinical standard, which is wet silver silver chloride electrodes. >> Yeah, and the operative word there being wet. If you have ever had a clinical EEG in a hospital setting, you are deeply familiar with this really messy reality. >> Oh, yeah. >> A technician literally takes an abrasive paste and physically scrubs your scalp to strip away the stratum corneum, which is the outermost layer of dead skin cells and oils. >> Which uh sounds terrible. Just aggressively scrubbing the scalp. >> Yeah. >> But I assume dead skin is also a terrible conductor of electricity. >> It is highly resistive, yes. So, once the skin is scrubbed raw and red, they apply a highly conductive electrolytic gel right between the silver electrode and your scalp. >> I always think of this like building a custom computer. >> Oh, really? How so? >> Well, if you are putting a metal heat sink onto a really hot computer chip to cool it down, you don't just slap the dry metal directly onto the dry chip. >> Right. >> Even if they look perfectly flat to the human eye, there are microscopic imperfections, right? Like tiny air gaps between the metals. And air is an insulator, it traps the heat. So, you have to squeeze thermal paste between them to fill those microscopic gaps and and a perfect bridge. >> That is >> That wet electrogel is basically doing the exact same thing, but for electricity instead of heat. It chemically bridges the microscopic air gaps caused by hair and skin texture. >> The thermal paste analogy is incredibly spot-on. By chemically bridging those gaps, you drop the skin impedance, the actual electrical resistance, down to about 5 kΩ. >> Which is very low. And if you glance over the telemetry dash on the UI, the result just speaks for itself. The signal is gorgeous. >> It really is. >> You get this beautifully clean high signal-to-noise ratio. The brain waves are distinct, the background static is almost zero. >> Yeah, it's the clinical gold standard for a reason. But here is the major problem. Bringing this back to the real world of consumer brain-computer interfaces, the wet gel presents an absolutely insurmountable friction point. >> Right. Because let's be honest, I am not going to aggressively scrub my scalp with abrasive paste and slather chemical conductive gel into my hair every single evening just because I wanted like play a virtual reality game or turn off my smart lights with my mind. Nobody is doing that. >> Nobody. The daily friction is just way too high, which is why the entire neurotech industry has basically spent the last decade aggressively pursuing dry consumer headsets. You know, devices you can just slip on like a pair of headphones. >> Well, let's test that out right now. I'm going to guide you back to the interface panel. Keep your eyes locked on the oscilloscope as I switch our hardware from the wet clinical standard to the dry modality. >> This represents modern consumer tech using dry pin-like electrodes that just kind of comb through the hair to rest directly on the scalp. >> Okay, ready? 3 2 1 switch. >> And there goes the signal. What you are seeing on the dashboard right now is the catastrophic reality of dry electrodes. Without that wet electrolytic gel bridging the microscopic air gaps we talked about, the electrical resistance just shoots through the roof. >> impedance on our telemetry dash just spiked from a highly conductive 5 kiloohms all the way past 250 kiloohms. >> It's a massive jump. >> The green lines just shrank to a whisper. The raw neural signals fighting so much resistance it's barely even registering on the screen anymore. >> And when your primary brain signal is that incredibly weak, something else really problematic happens. You open the door to environmental bleed. >> What does that mean? >> Your dry electrodes essentially functioning as raw antennas now, and they just start picking up whatever stronger electromagnetic signals are floating around the physical room. >> Oh, because the human body is basically just a giant bag of salt water, right? We are essentially walking antennas. >> Very much so. >> So, if the sensor can't hear the brain properly through the skull, it starts listening to the environment vibrating through us. Like, what specifically? What are we picking up? >> Look at the heavy thick oscillation that just took over the trace on your screen. In North America, the electrical power grid runs at 60 hertz. Because the impedance of the scalp is so terribly high that ambient 60 hertz electromagnetic noise, which is literally emanating from the wiring inside your walls, your desk lamp, your computer monitor, >> it's bleeding right into the sensor. >> Exactly. The brain waves are completely swamped by the electricity powering the building itself. >> You're trying to read a human mind, and instead you're literally reading the wall outlet. >> It's a massive headache for engineers. >> But wait, it actually gets worse than just the room noise. Everyone, find the button on your dashboard labeled artifact. I'm going to trigger it right now. This simulates our digital subject doing something entirely mundane, like just slightly shifting their weight in their chair, or you know, just clenching their jaw. Here we go. >> And the screen just blew out. >> Whoa. The green lines just shot into these massive jagged cliffs way off the top and bottom of the oscilloscope. The brain waves data is completely obliterated by these giant spikes. What just happened? >> You just triggered a motion artifact, specifically an electromyographic or EMG response. >> Okay. >> This highlights the absolute brutal trade-off of non-invasive headsets. You are locked in a constant battle for signal fidelity, not just against the environment, but against the subject's own body. >> How so? >> When you clench your jaw or swallow or even just blink your eyes, you are contracting muscles. And muscles operate using electrical potentials, just like the brain does, but muscle fibers are massive compared to neurons, and their electrical charges operate in millivolts, not the tiny microvolts of the brain. >> So, a jaw clench is basically a a megaphone screaming right next to the whispering neurons. >> Exactly. The muscle electricity drowns everything out. Plus, with dry electrodes, any slight movement of the subject's head creates tiny physical shifts of the sensor against the dry skin, which causes static electricity spikes. >> Uh that makes sense. >> So, just imagine a consumer playing an intense video game using a dry EEG headset. They get stressed during a boss fight, they grit their teeth, and instantly the system goes completely blind. The game character freezes because the computer can only see the jaw muscle. >> Okay, this brings us to a really critical logical roadblock then. >> Mhm. >> If non-invasive tech is this vulnerable, I mean, the skull blocks it, the power line swamp it, a simple jaw clench completely blows it out, how do we ever achieve perfect signal? >> It's tough. >> Because we've all seen the incredible videos on the news, right? Of paralyzed patients controlling robotic prosthetic arms with their minds, and they're able to slowly take a drink of water. They clearly aren't doing that through static and jaw clenches. >> No, they certainly aren't. To achieve that level of instantaneous high-precision control over a complex external machine, there's really only one engineering solution left on the table. We have to bypass the skull completely. >> to go under the bone. >> We move from non-invasive wearables to invasive neural implants. >> Okay, I want you to watch the traces on your screen one last time. We are going to change the simulation setting from the consumer dry modality directly to an invasive SEEG implant. That stands for stereo-electroencephalography. Get ready to see the absolute payload of modern neuroengineering. Switching now. >> It is genuinely like night and day, isn't it? >> It's stunning. The traces on the oscilloscope are pristine. The 60-Hz hum from the walls is entirely gone. The artifact blowouts are gone. We are just seeing incredibly clean, sharp, rapid data. >> The simulation just modeled the surgical placement of microscopic electrodes directly into the physical tissue of the cerebral cortex. By doing this, we've entirely bypassed the insulating properties of the dead skin, the scalp, and that thick bone. >> And now your telemetry dash, the impedances have dropped down to near zero. >> Yep. >> We took down the brick wall. We aren't standing outside the bank vault listening to the whisper anymore. We have stepped completely inside the vault. >> And because we are inside, something else truly amazing happens. The spatial blurring disappears. >> The spatial blurring? >> Yeah. When an electrical signal travels through the skull, the bone acts as a diffuse medium. It smears the signal out, making it incredibly difficult to pinpoint exactly which tiny millimeter of cortex generated the thought. >> Oh, I see. >> But with an invasive implant, that blur is gone. We know precisely which localized cluster of neurons is firing. >> And if you look closely at these new pristine traces, the actual frequency of the wave is incredibly fast. We aren't looking at those slow delta ocean swells anymore. We are looking at high gamma frequencies. >> Right. High gamma activity oscillates anywhere from 70 up to 150 Hz or even more. And from a biophysicist standpoint, high frequency waves have very short wavelength. They don't travel through dense physical barriers well at all. >> It's kind of like hearing a car pulling up outside with the stereo blasting, right? The high-pitched vocals get completely blocked by the walls of your house, but the long, slow, low-frequency bass notes easily rattle right through the glass. >> That is the exact same physics principle. The skull easily blocks the high-pitched treble of the high gamma waves, so we almost never see them with surface headsets. But once you surgically implant a sensor inside the brain, high gamma is everywhere. >> And why is capturing high gamma the holy grail for these engineers? >> Because high gamma frequencies are the fundamental neural code for precise motor intent. If you want to move your right index finger to push a button, a highly localized rapid burst of high gamma activity occurs in a very specific millimeter of your motor cortex. >> Oh, wow. >> By surgically unlocking access to those specific frequencies, a computer algorithm can read your exact physical intention in real time with zero lag, and translate it perfectly to a robotic finger. >> And there it is. That is the grand compromise as the core lesson of the entire dashboard you're looking at today. The choice of how we interface with the human brain is always a massive unforgiving engineering compromise. >> It really is. >> You constantly have to balance the pristine zero latency signal quality of a surgical implant against the daily friction of a wet clinical cap, against the absolute messiness of a consumer dry headset, and against the severe medical risks of opening up a healthy human skull. >> It is a delicate, extremely high-stakes balancing act. And right now, every researcher, neurologist, and consumer tech company on Earth is fighting tooth and nail to find the sweet spot on that triangle. >> Well, on that note, we are going to formally conclude the guided tour portion of this deep dive. I am officially yielding the controls of the Bionik EOS gas board back to you. I highly encourage you to stay and enjoy the sandbox. Click around. >> Yeah, definitely play around with it. >> Interact directly with that 10-20 head map. Experiment with different electrode materials. Maybe see what happens when you mix wet electrodes with the deep sleep state. Just watch how the biophysics of the brain adapt in real time. It really is mesmerizing once you understand the mechanics behind the screen. >> It is. But, you know, before we completely sign off, I want to leave you with one final thought to ponder while you play with the settings. >> All right, what is it? >> We just established that surgically unlocking high gamma frequencies gives us the fundamental neural code for perfect, instantaneous machine control. And right now, that technology is rightfully used to help paralyzed patients operate prosthetic limbs. But, imagine a future where these invasive implants become completely frictionless, safe, and mainstream for everyday consumers. >> A world where anyone can just walk into a clinic and get one. >> Right. When the latency drops to absolute zero and the signal between your mind and a computer is completely perfect and bidirectional, will we even be able to tell the difference between our own organic thoughts and the digital feedback from the machine? >> Oh, wow. >> If you can just query a massive artificial intelligence by simply thinking a question and the answer appears in your consciousness instantly, whose thought was that really? When the interface is perfect, where exactly does the human brain end and the computer begin? >> That is a very heavy, very real question. But, it's exactly the kind of muddy waters we end up in when we start tweaking the dials on the human mind. Keep an eye on that oscilloscope and keep questioning what you see. >> You can try it yourself on brainwaves.com/electrodes. There's also a description at the bottom of the page. And check uh other simulations as well. Like this epilepsy discharges overview. Or this uh sleep cycle simulator. All these tools are available on brainwaves.com. Go check them out and provide your feedback.