But a number of players have entered the field of neurotech with a different goal: cognitive enhancement of healthy humans, made possible by the formal, physical union of computers and our brains.

The largest contributor to neurotech has been Darpa, the research arm of the Department of Defense, which has a long record of pushing scientists and engineers to achieve the seemingly impossible. Darpa birthed the Arpanet, which eventually became the internet, and was also behind GPS and the world’s first successful self-driving cars. Its medical-research division has funded studies that use implants similar to the one created by Neuropace to treat everything from traumatic brain injury to psychiatric disorders.

Darpa has also committed $60 million to create what’s known as a “direct cortical interface,” a brain-computer connection unlike any that exists today. Neural implants like Emily Borghard’s can stimulate and record from just a handful of neurons. Darpa hopes to create a neural interface that can connect to as many as one million neurons.

A fully functional brain-computer interface on this scale would, in theory, turn a person into a programmable, debuggable machine—just like a computer. What used to be accomplished through drugs, training, education and psychotherapy could someday be achieved by more direct means. The goal is something akin to the scene in “The Matrix” where Keanu Reeves’s character learns kung fu from a quick download. “It’s almost like the design-build-test cycle” for designing new hardware and software, says Justin Sanchez, director of the biological technologies office at Darpa. This cycle of building a prototype, measuring its performance in the real world, and tweaking accordingly is how humans refine new technologies. Thanks to neurotech, we could someday use the same process to refine our brains.

A direct cortical interface is the goal of Neuralink, the latest moonshot startup from the serial entrepreneur Elon Musk. Neuralink is pursuing what Musk has described as “neural lace” technology—billions of tiny brain electrodes that may one day allow us to upload and download thoughts. Musk has said that his goal is “uncompressed” communication of thoughts between people—the ability to communicate concepts without first expressing them in language. To hear Musk tell it, he’s trying to replace texting with telepathy. “Your output level is so low, particularly on a phone, your two thumbs just tapping away,” he said last June. Neuralink declined to offer any compressed communication in the form of a comment for this article.

Bryan Johnson, who sold his payments startup to PayPal for $800 million in 2013, is currently founder and CEO of Kernel, a neurotech company he launched last August with $100 million of his own capital. “I consider this to be the most important thing we could be working on in the human race,” Johnson says. He’s convinced that cognitive enhancement from neurotech will unlock radical progress in every conceivable field. “The brain is the master tool,” he says. “Everything else is downstream: health, climate science, governance, education, love—everything.”

It isn’t clear what, exactly, startups such as Kernel and Neuralace are doing at the moment besides staffing up. “There have been a lot of defections from academia to industry in this space,” says Michael Kahana, the head of the Computational Memory Lab at the University of Pennsylvania. Professors from universities like Stanford and Caltech helped turn Silicon Valley into a hub of innovation, and it’s exactly these sorts of defections that have enabled far-out technologies to be rapidly commercialized.

The first step to installing self-contained computers into our brains is deciding where to put the hardware without distorting the skull. Implant accommodation is the goal of Longeviti Neuro Solutions, a startup founded by Chad Gordon, a plastic surgeon at Johns Hopkins, and Jesse Christopher, a veteran of medical-device companies. Dr. Gordon was developing prosthetics for patients who had lost portions of their skull to surgery and cancer when he realized that his replacement skull panels could be fitted with functional parts, from drug-administering ports to electronics.

On a recent morning in a conference room at Johns Hopkins Hospital, Dr. Gordon showed me one of Longeviti’s skull panels, a smooth, transparent, oblong piece of plastic. Each panel is custom-machined by a robotic arm fitted with a laser—custom-cut to fit a patient. Today, Longeviti carves indentations into the panels to house hardware, but the goal is to encapsulate implants and the batteries that power them inside the plastic. While we’re talking, Henry Brem, a Longeviti collaborator and head of neurosurgery at Johns Hopkins, picks up my iPhone and notes that it’s roughly the same size as the piece of artificial skull in his other hand. “Why couldn’t we put the electronics from this,” he says, indicating the phone and then the prosthetic, “into this?”

But Dr. Brem is quick to point out that the biggest obstacle to advanced neurotech implants isn’t the development of new hardware or the surgery required to implant it. “Opening and closing the skull isn’t the challenge for us,” Dr. Brem says. “Conceptually it sounds very frightening, but the technology to do that is very simple. The rate-limiting step is what you do when you get there.” According to the chief of neurosurgery at what is arguably the best neurosurgery department in the world, the limiting factor is our limited understanding of how the brain works. And Dr. Brem isn’t the only expert concerned that neurotech startups are putting the “tech” before the “neuro.”

“What most people fail to realize is there’s a whole different aspect of the problem, and that’s the basic science,” says Andy Schwartz, a professor of neurobiology at the University of Pittsburgh and a pioneer of brain-computer interfaces. “I would say almost everybody is skipping over that. I call that engineering arrogance.”

To see how far we have to go, you need only look at attempts to use a wireless implant to reconnect a monkey’s brain and limb after the animal’s spinal cord has been severed. In an experiment conducted at the Swiss Federal Institute of Technology, scientists used an implant to create a wireless connection between a monkey’s brain and a battery-powered stimulator in its paralyzed leg, allowing the monkey to walk again. It’s astonishing research that could someday make an end run around the problem of healing the spinal cord in patients with paralysis. But while monkeys in the experiment regained mobility, their marionette-like movements didn’t begin to approach the functionality they lost. And how our motor cortex controls our limbs is relatively well understood compared with more complex cognitive processes like how we represent ideas and language in our brains.

This lack of basic understanding is unlikely to deter Elon Musk, who has attempted to disrupt everything from automobiles to high-speed rail travel to rocket ships. But whether the “move fast and break things” ethos of Silicon Valley can be effectively applied to a delicate and mysterious organ remains to be seen.

Kernel founder Bryan Johnson hosts regular dinner parties of intellectuals, scientists and entrepreneurs where, he says, the conversation often turns to his work. His guests accept that brain-computer interfaces are coming but are troubled by the ethics. Once life-changing cognitive enhancement is realized, who will have access to it? Will governments give it to their citizens? Will businesses subsidize it for employees? Will companies someday require brain-computer interfaces? Won’t these complex and expensive systems exacerbate inequality? Should we be doing this at all?

According to Johnson, these are the wrong questions to ask. “People assume that somehow we have this permission structure that determines which technology is going to be developed in our society,” Johnson says. James Giordano, chief of the neuroethics studies program at Georgetown University, agrees. “Like it or not,” he says, “neurotechnology is happening. The cat is already out of the bag. It’s foolish if not naive to think you can stop it.”