202: Chapter 202 Feasibility Study of Brain-Computer Interface

Chen Youhua went from his dormitory to the laboratory in the morning, and from the laboratory back to his dormitory at night.

In between, he seemed nailed to his chair, staring at the dense literature and data on the screen, occasionally taking a sip of tea that had already gone cold, occasionally standing up to pace the room twice, and then sitting back down.

He was thinking about a problem.

A problem he had been thinking about for a long time, but had never dared to truly delve into.

The prosthetic limb project had reached the threshold of clinical trials. The connection success rate between artificial neurons and biological nerves had reached 99.97%; monkeys could use their thoughts to control a robotic arm to grab apples, peel bananas, and even mimic the strength of their grip, holding a grape without crushing it.

From an engineering perspective, this was already a remarkable achievement.

But from a scientific perspective, it was far from enough.

Current technology was essentially building a bridge.

Building a bridge between the two ends of a severed nerve to allow signals to flow again—whether this bridge was one-way or two-way, it ultimately only restored the original pathway.

It was like a bombed-out road; you build a temporary makeshift bridge so cars can pass, but the bridge surface is potholed, its load capacity is limited, and it cannot handle heavy trucks or high-speed traffic.

What Chen Youhua had truly wanted all along was not a prosthetic limb.

What he wanted was an information channel that could connect the brain directly to a machine, with sufficient bandwidth, low enough latency, and high enough capacity.

He wanted to create a brain-computer interface.

This was the reason why he wanted to push forward with the prosthetic limb project.

His individual intelligence was too limited, and there was also a conclusion buried in his heart that had always unsettled him.

He did not believe humans could control enlightenment.

A low-intelligence creature controlling a high-intelligence creature was like asking a Border Collie that couldn't even speak to direct humans in their work.

Humans would even find it very difficult to understand the language of a Border Collie.

Then, did enlightenment view humans in the same way?

...

On the screen was a review article published in Nature Neuroscience, the title of which, translated, was "Clinical Progress and Challenges of Implantable Brain-Computer Interfaces."

Chen Youhua had already read this article four times.

Every time he read it, he could discern something new, but every time it also reminded him of the same fact: the current state of this field was much cruder than he had imagined.

The mainstream solution for existing implantable brain-computer interfaces was the so-called "Utah Array."

A 4mm by 4mm silicon-based chip with 100 microneedles thinner than a human hair standing on it.

By inserting it into the cerebral cortex, each needle could record the electrical signal of one neuron; one hundred needles meant one hundred neurons.

It sounded like a lot.

But the human brain has 86 billion neurons.

One hundred versus eighty-six billion was like scooping a ladle of water from the Pacific Ocean and then saying, "I understand this ocean."

Even worse, after these microneedles were inserted, the human immune system would immediately treat them as invaders.

Glial cells would surround them, forming a layer of scar tissue on the surface of the microneedles, and the signal quality would deteriorate day by day.

After a few months, the signals that were originally clear would turn into blurred noise.

Therefore, these devices had a common name: temporary solutions.

Chen Youhua leaned back in his chair, eyes closed, thoughts churning over these data points in his mind.

He recalled the experiment Li Wei had conducted. A small segment of artificial neurons was connected to a Bullfrog's leg; whenever an electrical signal arrived, the hind leg would kick. That was in vitro, where the environment was controllable, with no immune system causing trouble and no scar tissue interference.

Now, putting the same thing into the brain—an organ ten thousand times more complex than peripheral nerves—was not an additive increase in difficulty, but a multiplicative one.

He began to write on the whiteboard.

This was his habit when thinking through problems. There were too many things on the screen, it was too cluttered, and not as good as the whiteboard. The whiteboard was linear; with every stroke, his train of thought followed his hand.

He wrote one word at the very top of the whiteboard: Signal.

Then he drew two lines below it.

On the left was "Acquisition," reading signals out of the brain. On the right was "Writing," writing signals into the brain.

On the acquisition side, he listed several keywords: resolution, bandwidth, invasiveness, and long-term stability.

On the writing side, the keywords were different: precision, safety, reversibility, and adaptability.

After writing, he stepped back two paces and stared at the whiteboard for a long time.

Any one of the problems on either side, taken individually, was a world-class challenge.

And what he had to do now was solve both sides simultaneously, while also integrating them into a single system.

Chen Youhua leaned back in his chair, eyes closed, thoughts churning in his mind.

He suddenly laughed.

It wasn't because he felt relaxed, but because he felt it was absurd.

He recalled his freshman year, when he proposed the concept of an artificial cytoskeleton at Professor Zhang Zhenhua's lecture; he had been full of such passion back then, his mind filled with wild, imaginative ideas, completely unaware of the immensity of heaven and earth.

But things were different now.

Back then, he didn't understand anything, so he dared to imagine anything.

Now that he knew enough, he instead knew how many pitfalls lay behind every statement, and how many walls stood before every step.

...

His phone vibrated once.

He picked it up to look; it was a message from Li Wei: "You haven't left yet? The cafeteria is about to close."

Chen Youhua paused, then glanced at the time in the bottom right corner of the computer screen: 8:40.

He had only eaten a sandwich for lunch, and his stomach was empty, but he didn't feel hungry at all.

He replied: "Leaving soon."

After sending it, he didn't move. The phone vibrated again.

"Are you struggling with some difficult problem again?"

Chen Youhua looked at the message, hesitated, typed a few words, deleted them, typed a few more, and deleted them again. In the end, he only replied with one word:

"Yeah."

Li Wei's reply came quickly: "You really are... Do you want me to bring you something to eat?"

"No need."

"Alright then. Get back early, don't stay up too late."

"Okay."

He placed his phone face down on the table, not wanting to look at it anymore.

The brain-computer interface was still in the feasibility study stage; there was no need to tell them. He would notify them once he could determine whether this could be done, and whether it should be done.

He turned back to the whiteboard, looking at the scribbled handwriting on it.

Signal, Acquisition, Writing, Resolution, Bandwidth, Invasiveness, Long-term Stability, Precision, Safety, Reversibility, Adaptability.

Behind every word was a mountain.

He picked up the pen and drew a line under "Invasiveness."

This was the biggest point of contention at the moment.

To acquire high-quality neural signals, the sensor must be close enough to the neurons.

The closer, the better; ideally, inserted directly into the cortex.

But the deeper it was inserted, the greater the trauma, the stronger the immune response, and the worse the long-term stability.

This was a classic "Impossible Trinity."

High signal quality, low invasiveness, and long-term stability—it was impossible to have all three.

Was there any way to break this triangle?

Instead of inserting it, what if it was placed on the surface, like an EEG, with electrodes attached to the scalp?

But the spatial resolution of an EEG was too poor; it could show that a brain region was active, but it couldn't tell which neuron was firing.

What if it was placed even closer? Not on the scalp, but on the dura mater? On the arachnoid membrane? Or even on the surface of the cerebral cortex?

This raised another problem: even if you placed it close enough, the electrodes themselves had physical limits.

One electrode could only record the signals of a small circle of neurons around it; to cover a large area, many electrodes were needed. But the more electrodes, the larger the volume of the implant, the greater the trauma, and the greater the risk.

He rubbed his temples, feeling as though his brain was a tangled mess of hemp, where every thread was connected to another; pulling one out brought ten others with it.