Just four months ago, Noland Arbaugh had a circle of bone removed from his skull and hair-thin sensor tentacles slipped into his brain. A computer about the size of a small stack of quarters was placed on top and the hole was sealed.

Paralyzed below the neck, Mr. Arbaugh is the first patient to take part in the clinical trial of humans testing Elon Musk’s Neuralink device, and his early progress was greeted with excitement.

Working with engineers, Mr. Arbaugh, 30, trained computer programs to translate the firing of neurons in his brain into the act of moving a cursor up, down and around. His command of the cursor was soon so agile that he could challenge his stepfather at Mario Kart and play an empire-building video game late into the night.

But as weeks passed, about 85 percent of the device’s tendrils slipped out of his brain. Neuralink’s staff had to retool the system to allow him to regain command of the cursor. Though he needed to learn a new method to click on something, he can still skate the cursor across the screen.

Neuralink advised him against a surgery to replace the threads, he said, adding that the situation had stabilized.

The setback became public earlier this month. And although the diminished activity was initially difficult and disappointing, Mr. Arbaugh said it had been worth it for Neuralink to move forward in a tech-medical field aimed at helping people regain their speech, sight or movement.

“I just want to bring everyone along this journey with me,” he said. “I want to show everyone how amazing this is. And it’s just been so rewarding. So I’m really excited to keep going.”

From a small desert town in Arizona, Mr. Arbaugh has emerged as an enthusiastic spokesman for Neuralink, one of at least five companies leveraging decades of academic research to engineer a device that can help restore function in people with disabilities or degenerative diseases.

While Mr. Musk’s pitches have centered on sci-fi ambitions like telepathy for high-tech consumers, Mr. Arbaugh’s experience shows the potential for advancement in one medical realm where federal authorities will allow such risky research.

Neuralink announced this week in news reports that it had received permission from the Food and Drug Administration to continue testing implants in additional patients. The company has not offered much detail on the unexpected flaw, and did not respond to requests for comment.

Mr. Arbaugh has been paralyzed since a swimming accident in the lush hills of Northeastern Pennsylvania, where he worked after college as a camp counselor. Lunging into waist-deep water in a lake with a group of friends, he sank to the bottom.

“I was facedown in the water and I just thought, well, I can’t move. So what do I do? I guess nothing,” Mr. Arbaugh said. “So I took a big drink and passed out.”

Mr. Arbaugh became paralyzed from the fourth vertebra in his neck down.

Adjusting to life as a person with quadriplegia was far from the future he had envisioned. As a young person growing up in Yuma, Ariz., Mr. Arbaugh had pursued all his community had to offer. If he wasn’t playing soccer, football or golf, he was at an Academic Decathlon competition or a chess tournament. Among the first in his family to go to college, he enrolled at Texas A&M University, where he admittedly smoked too much marijuana, spent a slightly aimless semester in Australia and skipped too many classes to graduate as a senior.

In the years after the accident, he tried to adjust to an array of devices meant to help people who are paralyzed. Most failed to work effectively for long periods, though Siri on his iPad emerged as his most reliable assistant, enabling him to call and text his friends.

Last year, a friend, Greg Bain, told him about Neuralink and urged him to apply for the company’s first trial in humans.

Mr. Arbaugh said he did not have strong feelings about Mr. Musk, but felt that he drove progress and that “things he touched turned to gold.”

After the implant was embedded in late January, he began working long days with Neuralink staff to link the neuronal patterns picked up in his brain to the actions he intended to take. He found the work tedious and repetitive, but rewarding.

Once the training was complete, engineers gave him control of the cursor on a computer. “I was like, once you guys take these restraints off me, I’m just gonna fly,” Mr. Arbaugh recalled.

On his first day flying solo, Mr. Arbaugh beat a 2017 world record in the field for speed and precision in cursor control. “It was very, very cool,” Mr. Arbaugh said.

The long days of training computer models with Neuralink staff at his side have now been reduced to remote work in four-hour time blocks, Mr. Arbaugh said. The team continues to work on tasks like spelling words, as he envisions making sign language letters or writing on a chalk board.

But the Neuralink device continued to lose its connection, the tendrils gradually sliding out of the tissue of his brain and presumably resting in the fluid that surrounds it.

When only about 15 percent of the threads remained in place, Mr. Arbaugh lost command of the cursor altogether. Engineers recalibrated the computer programs to perform most tasks he had been able to do before. Because he can no longer get the system to do mouse clicks, he’s using a new tool that allows him to click by hovering a cursor over the item he intends to select.

The flawed implant underscores the concerns of some experts in the brain-computer-interface field. The small, round device implanted in the skull is supposed to keep the thin tendrils of electrodes in place. But like a finger in a pie that’s wobbling about, the threads can pull out.

Mr. Arbaugh said his brain moved more than engineers had expected, and have revised the surgical plan to implant the threads deeper in the next patient’s brain.

Neuralink is vetting applications from others interested in taking part in trials. Their expenses, such as travel, are covered by the company, according to Neuralink.

This first Neuralink experiment also highlights how complicated the mechanics of the connection between the brain and a device are.

Lee Miller, a professor of neuroscience and rehabilitative medicine at Northwestern University, described the difficulties of working with the brain. It is bathed in salt water, moves around as the head swivels and bobs, and it is equipped with immune defenses meant to wall off invaders. Researchers have observed the brain forming scar tissue around sensors and even rejecting an entire sensing unit that used a grid of tiny needles.

Cristin Welle, a University of Colorado neurophysiologist who started the neural interfaces program at the Food and Drug Administration, which approves medical devices like the implants, said the first Neuralink case suggested that the company still faced hurdles in developing a durable device.

If the threads were implanted deeper, they could still ease out and leave fibers rubbing on the surface of the brain, possibly increasing the amount of scarring — and signal loss — in the area, she said.

“It’s hard to know if that would work,” Dr. Welle said. “It may be the case that a fully flexible device may not be a long-term solution.”

Mr. Arbaugh said his team had expected his brain to form scar tissue around the threads at the base of the brain — which they believed would help hold them in place. He said he will have the option to leave the study after a year, but expected to keep working with the company longer. Neuralink has said the initial study will take about six years to complete.

Other leading commercial companies have taken different approaches.

Synchron, based in Brooklyn, has avoided the delicate tissue of the brain by going through a vessel to implant a tiny metal tube near the motor cortex of the brain. Yet the device does not pick up as much subtle neural activity as others that penetrate brain tissue, according to researchers in the field. It registers louder signals, so to speak, like the intent to select an option from an on-screen menu. The company has human trials underway.

Precision Neuroscience, based in Manhattan has implanted a flexible strip equipped with sensors on the surface of people’s brains and is reviewing the data it is gleaning from patients with the strip placed temporarily, Michael Mager, the company’s chief executive, said.

Researchers have been studying brain-computer devices for decades. The standard had been a grid of 96 pins, called the Utah Array, that rests on the top of the brain and picks up activity up to 1.5 millimeters below the surface. It tends to be linked through a wire in the skull to a small box mounted on the head during continuing human trials. The hole in the skull that lets the wire through is prone to infection, though, and Blackrock Neurotech in Salt Lake City is working on a fully implantable upgrade.

Paradromics, which also uses a device with a grid based on the Utah Array, is testing its implantable device in sheep and expects to test it in humans in about a year, according to Matt Angle, the company’s chief executive.

All of the work is closely regulated by the F.D.A., which weighs the risks and the benefits for procedures and is expected to first consider use of these devices in people with major disabilities or degenerative diseases. (The agency would not comment specifically on Neuralink, but said it requires routine reports on expected and unexpected events in such trials.)

Beyond that, researchers are divided over the prospect of widespread use by people with no disability, who might want an implant to communicate without speech or to download a language, as Mr. Musk has mused. Some researchers predict availability for general consumers in decades. Others argue that will never be authorized for activities like web surfing in the shower, given the infection risk of repeated brain surgeries over a lifetime.

Mr. Angle, of Paradromics, said he could envision a progression from usage in people with no ability to speak or walk to those with serious mental health problems that have resisted treatment.

From there, he said, consumer use could be a decade away. After all, he said, the idea of Botox went from preposterous to mainstream in about as much time.

“If 100 years ago you said, ‘Hey, rich people are going to inject botulism in their face,’ that sounds totally nuts,” he said. “Once you understand the risks, and people can make informed decisions, then it becomes reasonable.”

Mr. Arbaugh hopes to see the technology deployed first to restore function in those who lost it.

“And then it can go to enabling people to enhance their capabilities,” he said, adding, “As long as we don’t give up our humanity along the way.”