The Weekly Loop - Prosthetics #1

. Diving in to the recent breakthroughs in both science and prosthetics!

Transcript

SPEAKER_00 0:01

In recent years, AI has risen to society's number one concern. Many people either don't understand AI or are fearful of it due to their own job being at stake. But on June 19th, 2026, Ohio Police Department retired their own AI, Robocop, after it heroically made zero arrests and issued zero tickets in nearly 10 months of service. So while it does seem like this big, fearful thing, AI needs to be significantly more advanced and doesn't seem as bad as it first did. Welcome back to the feedback loop. I've got five stories for you this week. One interview that rewired how I think of the word cyborg. A couple of MIT studies that go over neuroscience and can actually be implemented into prosthetics, and a quick hit of a round of general STEM news that all loops back one way or another to the limbs that we're building for people in prosthetics. Let's get into it. So I really wanted to start with an interview that Lampoon magazine published on July 15th because it's not really a news story, it's basically a worldview. The subject is Hugh Herr. At the MIT Media Lab, he runs the biomechatronics group. He co-directs the K Lisa Yang Center for Bionics. But here's the thing that actually gets me. Her isn't just studying prosthetics from the outside, he's with the wearing proof. Back in 1982, he's 17 years old, ice climbing on Mount Washington, and a blizzard traps him for days. Frostbite. He loses both his legs below the knee. And what they hand him afterward is, I mean, by his own description, basically the same tech they were using during the Civil War. Wood, foam, metal. Nothing talking to his nervous system. So he does the only thing that makes sense to him. He starts building his own legs, not just to walk again, to climb again, better than before the accident. That's the headline part of the story, but honestly, what he's building right now is more of the interesting part. And it's not really a limb, it's a conversation. Hurst team is chasing what he calls a bi-directional communication, brain to machine, and machine back to the brain. Not just think about moving the leg and it moves. Actually feel it move. He gave this example in the interview. When someone hands you a dumbbell, you don't look down at your hand to know that the grip just changed. Your body just tells you. That's proprioception. It's exactly what he wants the prosthetics to have. And they're actually getting there. Hur's team published in Nature's Medicine, this came out earlier this year, on a surgical technique where they reconnect agonist and antagonist muscles. So think biceps and triceps during the amputation itself instead of the afterward. So when someone thinks about moving a limb that isn't physically there anymore, those reconnected muscles actually move. And the brain gets real sensory feedback. Like the limb never left in the first place. They tested this on 14 people, seven with the reconnected muscles, seven without. And the difference wasn't subtle at all. Give the brain rich information and it just knows what to do. Hers exact point was that it doesn't even matter what the limb is made of, as long as the brain is getting the right signal. Then there's the sensation side of it, which, okay, I'll be honest, the mechanism almost sounds too simple to actually work. It's called the cutness mechaneural interface. Surgeons take a small patch of skin from the amputated fingertip and they wrap it around the end of the nerve that used to feel it. Hers actual words in the interview were kind of like a taco. Quote by quote, kind of like a taco. Over a few months, that nerve regenerates right into the skin, wraps some muscle around that whole setup, and now when a robotic fingertip gets touched, it fires the muscle, which presses the skin, which fires the nerve, which tells the brain, hey, there's a warm cup of coffee in your hand. That's not a metaphor for feeling something. That's the actual physical loop happening happening in real tissue. And here's why I wanted to open up with this one specifically. Every time we talk about the prosthetics industry on this show, cost, access, personalization, all of it. We're kind of assuming the ceiling is mechanical. Like better motors, better materials, better fit. Hers whole career is basically an argument that real sealing is neurological, and once you crack that open, everything else gets easier, not harder. He even makes an economic case for it. Directly in the piece. These surgeries don't really add operating time. The sensors cost a few bucks at scale. Insurance already covers the procedures. So it's cheaper long term, not more expensive, which is kind of the opposite of the story we usually tell about advanced prosthetics. And the line that actually stuck with me, he said we're moving out of what he calls the dark age of prosthetics into what he calls a dimly lit one. Not solved, just visible now. Finally. I think that's honestly the right frame for this whole episode. Sticking with MIT for a second, because there is a researcher profile the Yang Tan Collective put out on July 17th. That's basically the quiet infrastructure behind everything her just told us. Deanna Grass is a PhD candidate in the Harvard MIT program, and she works in the bioelectronics group under Paulina Anakiva. Same Yang Tan ecosystem, actually, that funds her center for bionics. Her whole research question is almost annoyingly simple to say out loud and brutally hard to actually answer. Your body is talking to itself constantly, nerves to muscles, immune system to nervous system, organ to organ, and we still don't really have the tools to listen in on that conversation in real time. Right now, medicine mostly gets snapshots, a scan there, a blood test there. Grass builds these sort of bioelectronic devices that integrate with tissue without damaging it. So you can record and stimulate neural circuits continuously instead of just catching freeze frames. But why does this matter for prosthetics? Because hers whole fingertip feeling setup only works if you can read that nerve single cleanly and continuously for years. Sensory feedback is only as good as the hardware that's actually listening to it. So Grass's work is kind of unglamorous half of the bionic limb story, not the leg, not the hand, but the sensor mesh underneath all of it that has to be soft enough, has to be stable enough, accurate enough to trust with a signal coming off a regenerative nerve. You don't get a feeling prosthetic without someone solving this part first. Okay, this next one's pure neuroscience on the surface, but stick with me because the implementation actually loops back to everything we just talked about. Researchers at MIT's and McGovernx Institute, led by Evelina Federinko, published a study on July 6th of the journal PNAS asking a genuinely old philosophical question with brand new tools. Do you actually need language to think logically? They worked with two stroke patients who had severe aphasia, meaning their language sensors were badly damaged, both understanding speech and producing it. And they gave these patients language-free logic puzzles, figure out the hidden rule, turning one number sequence into another, complete a geometric pattern. And these patients solved the puzzles just as well as a healthy control group. They just explained their answers with gestures and sketches instead of words. Then they ran brain scans on healthy adults doing similar logic tasks, and the imaging backed it up completely. The brain's language network barely lit up at all during logical reasoning. It's a totally separate system. So here's the prosthetics angle, and honestly, it's the same angle researchers themselves flagged in the piece. Every brain computer interface, every neural-controlled limb is built on the assumption that you have that you can decode intention straight from brain activity. And if logic and language are running on separate circuits, that's basically a map. It tells the engineers which part of the brain are actually worth listening to when they're trying to decode something like I want to grip this versus when parts are just narrating the decision after it already happened. You don't build a better neural interface by getting lucky. You build it by knowing exactly which wire you're tapping. Alright, rapid fire. Three stories that have nothing to do with prosthetitis on the surface and everything to do with it underneath. First, living plastic that eats itself. This one came out of American Chemical Society on July 16th, published in the ACS Applied Polymer Materials. Researchers built plastic embedded with dormant bacterial spores, engineered Bacillus subtilis that produce two enzymes working in sequence. One chops the polymer into fragments, the other breaks those fragments all the way down to their molecular building blocks, activated with warm nutrient broth, and the whole material is gone in six days, no microplastic residue left behind. They already built a wearable electrode out of it that worked completely normally, then fully degraded two weeks after activation. And think about how many prosthetic components are basically short lifespan by design. Liners, test sockets, disposable sensor patches, a material that's durable exactly as long as you need it, and then just cleanly disappears. That's genuinely useful for an industry that's constantly juggling cost, waste, and fit. Second, quantum materials where light can literally control magnetism. This one's a review out of the City College of New York, published July 16th in Nature Materials, looking at atomically thin magnetic semiconductors where light and magnetism aren't separate properties anymore. They actually influence each other directly. The practical payoff down the line is stuff like optical memory and ultrasensitive magnooptic sensors. It's early, it's foundational. Kind of a remember this name in five years story. But sensor sensitivity is a whole game for the next gen of prosthetics. The more precisely you can detect tiny electrical or magnetical magnetic signal, the finer the control you can hand back to somebody. Third, a camera that can basically see the invisible. This came out of the ETH, Zurich and EPFL, published July 17th, also in Nature Family of Journals. They built something called Platinum. Instead of the usual approach, chopping a detector into millions of tiny segments, they used a single block of light-sensitive material paired with a light field camera. Same tech behind the depth sensing cameras, plus AI to reconstruct particle paths from as few as five photons. It's built for particle physics, sure. But researchers have already filed patents applying it to PET scans, the imaging tech that tracks activity inside the body. And sharper, cheaper medical imaging has a pretty direct line to the better pre-surgical planning for the exact kind of nerve reconstruction procedures that we were just talking about, talking about in segment one. Three completely different fields, right? But it's the same underlying story every time. Better materials, better sensors, better imaging, and eventually it all lands back on the same workbench. So here's the thread running through this whole episode. If you're keeping score, Hugh Hurz proving the next leap in prosthetics isn't mechanical, it's neurological. Diana Grasse is building the soft hardware that has to exist for that neurological connection to actually hold up in the long term. The MIT logic study basically just handed engineers a clearer map of which brain signals are even worth decoding in the first place. And even the stories that feel like they belong in a totally different show. Self destructing plastic, quantum sensors, particle cameras, they're quietly solving the materials and imaging problems that this field is going to need in five years. Whether it knows it yet or not. That's the feedback loop. Science doesn't move in one direction, and neither does this show. If this one was useful, send it to someone who needs it, and I'll catch you in the next week.