A decade in Nathan Copeland's brain: the Pitt ICMS safety paper lands

By Inside BCI · July 18, 2026

On Star Wars Day 2015, a Pittsburgh neurosurgeon opened Nathan Copeland’s skull and slid four Blackrock Utah arrays into the tissue behind his forehead. Two went into motor cortex, so the twenty-eight-year-old could move a robot arm by thinking about it. Two went into somatosensory cortex, so that when the robot arm touched something, the current running back the other way would evoke a sensation in the fingers of a hand Copeland could not otherwise feel. He had been paralysed from the upper chest down for over ten years by then, after a car accident in the rain at eighteen. Asked what the earliest evoked sensations were like, he said the majority felt like pressure or tingling; a smaller number felt warm; one category didn’t quite fit any of those words, so the team called it “spidey sense.” A year and a half later, on 13 October 2016, Copeland used the motor arrays to fist-bump President Barack Obama at the White House Frontiers Conference held on the Pitt and Carnegie Mellon campuses. Obama said it was pretty cool.

Eleven years, two months and eleven days after that Star Wars Day surgery, Robert Gaunt and Charles Greenspon published the safety data in Science Translational Medicine. Five spinal-cord-injury participants, three at Pittsburgh and two at the University of Chicago, have collectively spent twenty-seven years living with Utah arrays in the hand region of Brodmann’s Area 1. The team delivered 168 million intracortical microstimulation pulses across those twenty-seven years. Not one serious adverse event. Sixty-four per cent of the electrodes are still functional on average. Sixty per cent of the electrodes in the longest-implanted participant are still working after a decade. That participant is not named in the paper. There is no other publicly documented ten-year chronic user of Utah arrays in the Pitt program.

What the paper actually reports

The paper is the deliverable of the NIH BRAIN Initiative-funded, DARPA-seeded ICMS somatosensory programme that ran at Pittsburgh and Chicago under clinical trial NCT01894802, and it is the primary long-term-safety and efficacy report for an intracortical microstimulation approach that has been used in humans since Sharlene Flesher’s first-in-human paper in the same journal in October 2016. Charles Greenspon at the University of Chicago Department of Neurological Surgery is the lead author. Robert Gaunt at Pittsburgh’s Rehab Neural Engineering Labs is the senior author. There are twenty-two authors in total across Pittsburgh, Chicago, UT Southwestern and Northwestern. The paper’s five participants each received two Blackrock NeuroPort microelectrode arrays surgically implanted in the hand representation of somatosensory cortex. Stimulation across the five-person cohort was delivered over durations ranging from two to ten years. The primary safety finding is that no participant experienced a serious adverse event across the full 168-million-pulse dataset. About one persistent sensation was reported per twenty-three thousand stimulation trials; almost all lasted under ten seconds, none were painful, none required medical intervention. On device durability, electrode function declined but did not fall off a cliff, and the paper explicitly notes accelerated decay later in the follow-up window.

Gaunt’s on-record framing of what the data means is closer to relief than triumph. “This research plants a flag in the ground for the safety and utility of using brain-computer interfaces to deliver sensory stimulation in clinical settings and, eventually, in people’s homes,” he told the Pitt School of Medicine. Elsewhere in the same press cycle he was more direct. “If we ever want these devices to become clinical, then they have to last.” And, most bluntly: “It’s sort of obvious, but knowing if a device is going to work for ten years, it just takes ten years.” That is the piece. The trial started before the current wave of commercial BCI companies existed. It ran through the entire arc of Neuralink’s founding, Synchron’s Stentrode approvals, Paradromics’ first surgery, Precision’s Layer 7 clearance, and the Neuracle NEO regulatory approval in China this year. It survived the death of its Chicago co-principal investigator Sliman Bensmaia in August 2023. And the arrays kept working.

The absence at the top of the author list

Bensmaia is not on the paper. He died on 11 August 2023, aged forty-nine, halfway through the follow-up window. He was the University of Chicago neuroscientist who spent his career decoding how the somatosensory cortex represents touch, and the intellectual architect of the biomimetic stimulation-encoding work that the Pitt team was using to make the evoked percepts feel more like the object being handled and less like electrical shocks. His absence from the author list is a mechanical fact of academic authorship after death, not a rejection of contribution. His UChicago colleagues Nicholas Hatsopoulos, John Downey, Giacomo Valle and Ashley Van Driesche continued the trial through his death and into the paper. Whether the paper carries an explicit dedication in the acknowledgements section is not confirmed in the press coverage. What is confirmed is that the trial answered the question he had been living inside for two decades: whether the somatosensory cortex, wired to electrical current for years at a time, would remain a usable neural surface for restoring a sense of contact to people who had lost their hands to their own nervous systems.

The answer in the paper is a qualified yes, with the qualifier doing serious work. The sensations Copeland and the other four participants report are not naturalistic touch. Gaunt has been consistent about this in every interview across the paper’s press cycle. “We are generations from being able to tell the difference between silk and satin or frankly, even a block of wood and satin,” he told 90.5 WESA. What the participants can do is feel individual fingers when the corresponding electrodes are pulsed, feel changes in pressure and vibration when a prosthetic hand grips an object, and remain confident that the sensations are staying spatially loyal to the hand where they belong rather than drifting to elsewhere on the body. Copeland’s own 2016 language was that “sometimes it feels electrical and sometimes its pressure, but for the most part, I can tell most of the fingers with definite precision. It feels like my fingers are getting touched or pushed.” Eleven years later he is one of the five people whose brain generated the data in this paper.

What ten-year data does to the commercial BCI conversation

Nathan Copeland’s arrays were manufactured by Blackrock Neurotech approximately two decades ago. They were designed to be an academic research tool, not a consumer medical device. The Pitt paper reports that they have held together, held function at 60 per cent in the longest participant, and stayed safe across 168 million individual current injections, using hardware from a design generation that predates every commercial BCI company currently valued above one billion dollars. Gaunt’s own read of what this means for the newer devices in the pipeline is on record and worth quoting directly: “A twenty-year-old piece of technology can hit these milestones in terms of longevity. So it’s sort of hopeful for the newer, better technologies that are around the corner.” And: “We’re probably a couple of years away from the first generation of commercial brain-computer interfaces reaching the broader market. But I think this is coming… we’re closer now than we have ever been.”

The specific news for Neuralink, Paradromics, Precision Neuroscience and Synchron is that the durability bar they now have to meet in humans has been publicly set by a competitor from a research programme running on twenty-year-old hardware. Neuralink’s N1 array is thinner, has more channels, and uses flexible threads rather than rigid Utah spikes; the company’s stated design intent is longer functional life than the Utah generation. That is now a measurable claim rather than an aspiration, because Pitt has just published what the Utah generation actually did over a decade. Paradromics’ Connexus, which reached first-in-human at the University of Michigan and has cleared an FDA Early Feasibility Study for speech restoration, has a similarly high-count design but no chronic human duration data of its own yet. Precision’s Layer 7 is a surface electrode array and does not do intracortical stimulation; it is not in this comparison. Synchron’s Stentrode is endovascular and also does not do intracortical stimulation.

None of that includes bidirectional systems. Every current commercial BCI in human trials is motor-decode only. The Pitt paper is the durability envelope for the somatosensory half of the bidirectional loop that all four US commercial companies have publicly said they eventually want to build. If Gaunt’s read is right that the first commercial BCI will reach the broader market within a couple of years, then the sensory feedback that closes the motor loop is the next step, and the ten-year Pitt data is the safety envelope those companies will now be measured against.

Greenspon’s own summary at the launch was tighter. “This shows that this technology doesn’t just have to be a short-term solution we test in the lab; industry can start developing long-term take-home solutions for patients.”

Where this sits alongside the other 16 July paper

The Pitt paper is not the only sensory-restoration story to publish this week. The next day, a Feinstein Institutes team led by Chad Bouton published a Nature Medicine paper reporting on Keith Thomas, a quadriplegic man in whom cortical BCI implants and wearable transcutaneous stimulation patches together restored enough motor function and enough peripheral sensation for him to feed himself and drink from a cup. That paper is a different modality and a different site. The Feinstein Bouton work uses non-invasive patches on the skin plus cortical implants to route signals around the injury; the Pitt Gaunt work uses cortically implanted arrays to evoke sensations by direct microstimulation of the somatosensory cortex itself. The two papers are close cousins, both fit the sensory-restoration broadening that Inside BCI has been tracking through 2026, and neither should be conflated with the other. The Cleveland VA and Case Western Reserve peripheral-nerve-cuff programme under Dustin Tyler, which restores touch to prosthetic hands in amputees via arm-nerve stimulation, is another parallel modality with a similarly long clinical arc; Pitt’s paper is the ICMS-in-cortex durability benchmark specifically, not the last word on restoring touch generally.

What to watch

Watch what Neuralink says next about sensory feedback. Elon Musk’s team has publicly discussed adding bidirectional stim as a future N1 capability, and every day the Pitt safety envelope sits in the literature the case for Neuralink actually shipping something similar strengthens. The specific question is whether Neuralink files an IDE for a stimulating variant of the N1 within the next twelve months, and whether the FDA reviewers cite the Pitt paper’s safety envelope in their evaluation.

Watch Paradromics’ bidirectional roadmap. Connexus is currently motor-decode only in the University of Michigan cohort, but the company has repeatedly said the higher electrode count is meant for both directions. A public commitment from Paradromics to a specific ICMS clinical milestone would be a direct competitive response to the Pitt data.

Watch how quickly the ten-year cohort becomes an eleven-, twelve-, or fifteen-year cohort. Copeland is thirty-nine years old, publicly active, and running a BCI advocacy company. The Pitt team’s clinical trial is registered as active on ClinicalTrials.gov. The paper is a snapshot; the arrays are still running.

Sources

news ICMS intracortical-microstimulation somatosensory-cortex Nathan-Copeland Robert-Gaunt Charles-Greenspon Pitt University-of-Chicago UPMC Blackrock-Utah-array Science-Translational-Medicine long-term-safety Sliman-Bensmaia sensory-restoration spinal-cord-injury

What the paper actually reports

The absence at the top of the author list

What ten-year data does to the commercial BCI conversation

Where this sits alongside the other 16 July paper

What to watch

Weekly BCI Brief in your inbox