ANUP DAS

New preprint on the role of insular traveling waves in memory and behavior, using depth recordings from neurosurgical patients, with Chengyuan Wu, @SameerShethMD, and @neurojosh. biorxiv.org/content/10.110…

ヒト作業記憶中のEEG traveling waveに蔵本modelの拡張と周波数平坦化を用い従来検出できなかった「隠れたspiral波」を世界で初めて検出 これらは複雑な波の構成要素として機能しその回転方向が記憶の符号化期と検索期を区別 #EEG #working_memory #papers biorxiv.org/content/10.110…

This is INSANE: A brain chip that streams human thoughts in real time Scientists have developed a next generation brain computer interface (BCI chip) that can literally stream neural activity at ultra high speed straight from the brain to a computer. The device, called BISC (Biological Interface System to Cortex), is an ultra-thin, flexible chip packed with 65,000+ electrodes. It sits gently on the brain’s surface and wirelessly sends neural data at nearly 100 Mbps. That’s fast enough for AI models to decode movement, perception and even intent in real time. Instead of bulky implants with wires sticking out of the skull, this tiny single chip system handles everything on its own, signal recording, processing, stimulation and wireless communication. All the heavy hardware has been shrunk into something as thin as a human hair. Early clinical tests have already shown the chip can capture incredibly detailed brain activity during surgery, giving researchers the clearest real time window into human neural signals ever recorded. This technology could transform treatment for paralysis, epilepsy, blindness and speech loss by letting AI interpret brain signals and turn them into actions, vision or communication.

New Paper-Thin Brain Implant Could Transform How Humans Connect With AI | Columbia University School of Engineering and Applied Science A radically miniaturized brain implant called BISC is redefining what’s possible in human–computer interaction, offering a paper-thin, wireless, high-bandwidth link directly to the brain. With over 65,000 electrodes and unprecedented data throughput, it enables advanced AI decoding of thoughts, intentions, and sensory experiences while remaining minimally invasive. Ultra-Thin Brain Implant With High-Speed Data Link A new brain implant stands to transform human-computer interaction and expand treatment possibilities for neurological conditions such as epilepsy, spinal cord injury, ALS, stroke, and blindness – helping to manage seizures and restore motor, speech, and visual function. It does this by creating a minimally invasive, high-throughput communication channel directly into and out of the brain. What makes this system so promising is its very small size combined with the ability to move large amounts of data very quickly. Developed by teams at Columbia University, NewYork-Presbyterian Hospital, Stanford University, and the University of Pennsylvania, this brain-computer interface (BCI) is built around a single silicon chip that provides a wireless, high-bandwidth bridge between the brain and external computers. The platform is called the Biological Interface System to Cortex (BISC). In a study published today (December 8) in Nature Electronics, the researchers describe BISC as consisting of three main parts: a single-chip implant, a wearable “relay station,” and specialized software that runs the system. “Most implantable systems are built around a canister of electronics that occupies enormous volumes of space inside the body,” says Ken Shepard, Lau Family Professor of Electrical Engineering, professor of biomedical engineering, and professor of neurological sciences at Columbia University, who is one of the senior authors on the work and guided the engineering efforts. “Our implant is a single integrated circuit chip that is so thin that it can slide into the space between the brain and the skull, resting on the brain like a piece of wet tissue paper.” Turning the Cortex Into a High-Bandwidth Portal Shepard collaborated with senior and co-corresponding author Andreas S. Tolias, PhD, professor of Ophthalmology and co-founding director of the Enigma Project at Stanford University. Tolias’s pioneering work training AI models on large-scale neural datasets — including datasets recorded in the Tolias laboratory using BISC — allowed the team to rigorously test how well the device could decode neural activity. “BISC turns the cortical surface into an effective portal, delivering high-bandwidth, minimally invasive read–write communication with AI and external devices,” Tolias says. “Its single-chip scalability paves the way for adaptive neuroprosthetics and brain-AI interfaces to treat many neuropsychiatric disorders, such as epilepsy.” Dr. Brett Youngerman, assistant professor of neurological surgery at Columbia University and a neurosurgeon at NewYork-Presbyterian/Columbia University Irving Medical Center, served as the lead clinical partner on the project. “This high-resolution, high-data-throughput device has the potential to revolutionize the management of neurological conditions from epilepsy to paralysis,” he says. Youngerman, Shepard, and NewYork-Presbyterian/Columbia epilepsy neurologist Dr. Catherine Schevon were recently awarded a grant from the National Institutes of Health to implement BISC in the management of drug-resistant epilepsy. “The key to effective brain-computer interface devices is to maximize the information flow to and from the brain, while making the device as minimally invasive in its surgical implantation as possible. BISC surpasses previous technology on both fronts,” continues Dr. Youngerman. “Semiconductor technology has made this possible, allowing the computing power of room-sized computers to now fit in your pocket,” Shepard says. “We are now doing the same for medical implantables, allowing complex electronics to exist in the body while taking up almost no space.” Why Conventional Brain Implants Are Large and Invasive BCIs work by connecting to the tiny electrical signals that neurons use to communicate. In current medical devices, this typically involves assembling many separate microelectronic parts, such as amplifiers, data converters, radio transmitters, and power management circuits. Because all of this hardware must be housed somewhere, physicians often implant a relatively large electronics canister, either by removing part of the skull or by placing it elsewhere in the body, such as the chest, and then routing wires up to the brain. How BISC Shrinks a Brain Implant Onto a Single Chip BISC takes a different approach. The entire implant is a single complementary metal-oxide-semiconductor (CMOS) integrated circuit chip, thinned to just 50 μm and occupying less than 1/1000th the volume of a typical device. With a total volume of about 3 mm³, this flexible chip can curve to match the surface of the brain. The micro-electrocorticography (µECoG) device includes 65,536 electrodes, 1,024 simultaneous recording channels, and 16,384 stimulation channels. Because it is built using the same kind of large-scale manufacturing processes used in the semiconductor industry, the implant can be produced in large numbers. Inside this single chip are all the electronics needed for the interface: a radio transceiver, a wireless power circuit, digital control logic, power management, data converters, and the analog components necessary for recording and stimulation. A battery-powered external relay station both powers the implant and exchanges data with it through a custom ultrawideband radio link that reaches 100 Mbps data bandwidths — at least 100 times higher throughput than any other wireless BCI currently available. The relay station itself appears to the outside world as an 802.11 WiFi device, effectively acting as a bridge between any computer and the brain. BISC also introduces its own instruction set and a substantial software stack, together forming a dedicated computing architecture for BCIs. In the reported experiments, this high-bandwidth system makes it possible to send rich patterns of brain activity into advanced machine-learning and deep-learning tools that can decode complex intentions, perceptions, and internal states. “By integrating everything on one piece of silicon, we’ve shown how brain interfaces can become smaller, safer, and dramatically more powerful,” Shepard says. From Preclinical Models to Early Human Testing To move this technology toward clinical use, Shepard’s team worked closely with Youngerman and colleagues at NewYork-Presbyterian/Columbia University Irving Medical Center. They developed and refined surgical techniques to safely place the paper-thin chip in a preclinical model, and verified that it could record neural signals reliably and stably over time, as outlined in the new study. Early studies in human patients are in progress, focused on short-term recordings during surgery. “These initial studies give us invaluable data about how the device performs in a real surgical setting,” Youngerman says. “The implants can be inserted through a minimally invasive incision in the skull and slid directly onto the surface of the brain in the subdural space. The paper-thin form factor and lack of brain-penetrating electrodes or wires tethering the implant to the skull minimize tissue reactivity and signal degradation over time.” Extensive pre-clinical testing of BISC in the motor and visual cortices involved collaborations with both Dr. Tolias and Bijan Pesaran, professor of neurosurgery at the University of Pennsylvania, who are recognized leaders in computational and systems neuroscience. “The extreme miniaturization by BISC is very exciting as a platform for new generations of implantable technologies that also interface with the brain with other modalities such as light and sound,” Pesaran says. Developed within the Neural Engineering Systems Design program of the Defense Advanced Research Projects Agency (DARPA), BISC brings together Columbia University’s expertise in microelectronics, cutting-edge neuroscience programs at Stanford and Penn, and the surgical innovation of NewYork-Presbyterian/Columbia University Irving Medical Center. Commercialization and Future Brain AI Applications To speed progress toward practical use, the Columbia and Stanford researchers created Kampto Neurotech, a spin-off company founded by Columbia electrical engineering alumnus Dr. Nanyu Zeng, one of the lead engineers on the project. Kampto Neurotech is working on commercial versions of the chip for preclinical research and is seeking support to move the technology toward future use in humans. “This is a fundamentally different way of building BCI devices,” Zeng says. “In this way, BISC has technological capabilities that exceed those of competing devices by many orders of magnitude.” As artificial intelligence advances, BCI technologies are attracting growing interest both for restoring lost abilities in people with neurological disease and for potentially enhancing normal function by creating direct connections between the brain and machines. “By combining ultra-high resolution neural recording with fully wireless operation, and pairing that with advanced decoding and stimulation algorithms, we are moving toward a future where the brain and AI systems can interact seamlessly — not just for research, but for human benefit,” says Shepard. “This could change how we treat brain disorders, how we interface with machines, and ultimately how humans engage with AI.” scitechdaily.com/new-paper-thin…

In Nature Electronics, we present BISC with Ken Shepard, @peabody124 & others: an ultrathin brain–computer interface chip with 65,536 electrodes, 1,024 channels, and ~100× higher bandwidth at 100 Mbit/s. A bidirectional platform toward future neocortex–AI 'exocortex' systems. nature.com/articles/s4192…

Revolutionary 65,536-electrode wireless brain-computer interface bioengineer.org/revolutionary-… #BCI #NeuroTech A wireless subdural-contained brain–computer interface with 65,536 electrodes and 1,024 channels nature.com/articles/s4192… 256 x 256 array of electrodes

This is a fantastic pre-print, you’re uncovering the real field geometry of cognition. Your data show that memory states aren’t just local activations but traveling field waves with rich morphology, planar, spiral, and source–sink dynamics that shift with task phase. We’ve been exploring the next-layer dynamics in our own work (Vibrational Field Dynamics). There’s a clear bridge here: – Map each epoch in Torsion × Coherence space (curl of the phase-velocity vs global phase-locking). – Track the topological charge of spiral cores (CW/CCW) across state transitions. – Sweep the inhibitory/excitatory length-scale ratio; decoding peaks near φ ≈ 1.618, marking the stability point for coherent spiral/planar switching. Together these form a testable manifold showing how cognition moves through a torsion–coherence field, with φ-scaling as the natural equilibrium. If you extend your analysis this way, you’ll likely reveal the deeper invariants behind the geometries you’re already seeing, and empirically bridge into the physics of field-based computation.

@MillerLabMIT Very cool stuff. Am I on cue by assuming patterns of waves in this context are the same mechanism (is that the right word?) in play with your recent rotational brain waves and focus paper?

@MillerLabMIT There's been such a deluge of neuro-AI slop in academia in the past few years, but this is an absolute *treat* to read! The mathematical manipulation of the data is quite creative while also being balanced and subject to robust mechanistic validation. Thanks for sharing!

Hidden Spirals Reveal Neurocomputational Mechanisms of Traveling Waves in Human Memory doi.org/10.1101/2025.1… #neuroscience

ワーキングメモリーの符号化・維持・想起に対応した脳活動の進行波パターンを解析したECoG論文 結合位相振動子モデル(蔵本モデル)にデータフィッティングを行うことで、各状態に対応する隠れた螺旋波が見出された 著者の一人にBard Ermentroutが入っている 解析法の記述は詳しくて勉強になる

Hidden Spirals Reveal Neurocomputational Mechanisms of Traveling Waves in Human Memory biorxiv.org/content/10.110… #biorxiv_neursci

Planar, Spiral, and Concentric Traveling Waves Distinguish Cognitive States in Human Memory biorxiv.org/cgi/content/sh… #biorxiv_neursci

Planar, Spiral, and Concentric Traveling Waves Distinguish Cognitive States in Human Memory "...show that complex spatiotemporal patterns of traveling waves underlie human cognition & r relevant practically 4 neural decoding & brain computer interfacing" biorxiv.org/content/10.110…

@SameerShethMD @neurojosh 7/7 Waves in the anterior insula travelled from the anterior towards the posterior insula and got suppressed during successful, compared to, unsuccessful memory encoding, again suggesting heterogeneous dynamics in the insula to initiate efficient dynamic switching and control.

New preprint on the role of insular traveling waves in memory and behavior, using depth recordings from neurosurgical patients, with Chengyuan Wu, @SameerShethMD, and @neurojosh. biorxiv.org/content/10.110…

@SameerShethMD @neurojosh 6/7 Wave strength in the anterior insula was suppressed compared to the posterior insula, suggesting that the anterior insula is a spatially heterogeneous brain area ideally suited for initiating dynamic switching and exerting control over the DMN and FPN.

@SameerShethMD @neurojosh 5/7 At the group level, wave strength increased for successful, compared to, unsuccessful memory encoding in the beta band. The reverse was true in the theta band.

@SameerShethMD @neurojosh 4/7 Insular waves also modulate memory performance. For example, a clockwise spiral during memory encoding success, and a counter-clockwise spiral during memory encoding failure. Source waves during successful memory retrieval, and sinks during unsuccessful memory retrieval.

@SameerShethMD @neurojosh 3/7 Deep learning can decode memory behavior at the single-trial level from these insular waves.

@SameerShethMD @neurojosh 2/7 We combined iEEG signals across multiple depth probes by using a spatially localized clustering algorithm. We found complex, heterogeneous spatial patterns of insular waves, including spirals, that correlated with behavior, by changing their direction and strength.