Putting the science in fiction - Dan Koboldt, Chuck Wendig 2018

What’s possible with cyborgs and cybernetics
Things to know for when Skynet takes over

By Benjamin C. Kinney

First things first: There are already cyborgs among us. Early twenty-first-century medical science has plenty of ways to surgically integrate a device into your body: cochlear implants, hip replacements, deep brain stimulation, pacemakers, and so forth. Right now, the first few human beings have chips implanted in their brains that allow them to control a robotic limb directly with their thoughts. For example, the BrainGate clinical trials allowed its first tetraplegic patient to control a computer cursor in 2006, gave another patient control over a robot arm in 2012, and used muscle-stimulation techniques to restore control of the patient’s paralyzed limb in 2017.

From here on out, we’ll focus on cybernetics suited for space-opera life: mechanical or other devices controlled directly by the human brain.

The hardest part is neural decoding

If you want to control your cyber-arm just as smoothly and naturally as your real arm, the key technology is neural decoding. That means understanding the activity in your nervous system so well that you can convert neural signals into a machine-readable format: in other words, the neural-to-technological interface. (I say “machine” for simplicity, but this would hold true for any technology.)

Neural decoding requires a lot of information. The system doesn’t just need to know your goal (e.g., “pick up coffee cup”), it also needs to know where you want to move your arm (kinematics, e.g., “reach to the left of my water bottle”) and how you would accomplish that with a pattern of muscles and joints in motion over time (dynamics, generally below our conscious control).

Imagine how it would feel to have a system that didn’t use your own kinematics: You’d tell it what to get done, but not how to do it. That kind of system would feel less like controlling a limb and more like pressing a “bring me coffee” button. A big step up for a paralyzed person, but it’s a far cry from making the cybernetic arm a natural part of your life.

Brain function is distributed and hard to measure

Your brain doesn’t have single cells that do specific things, outside of the most basic sensory perception. Information is shared across the activity and connections of hundreds or thousands of brain cells (mostly neurons, but also other cell types called glia). When your brain controls a movement, no single cell carries all the relevant information. To decode the neural signals for arm movement, you need to measure the neural activity in most of those hundreds or thousands of cells.

This is a big technological sticking point: how to record so much activity from the brain? Current brain-computer interface (BCI) methods involve implanting a chip onto the surface, with dozens of tiny microelectrodes sticking down into the brain tissue below.

Figure 35.1: “BCI chip circa 2012. (A) Pedestal connector, grounding wires, and chip. Only the chip (B) makes contact with the brain. From Hochberg et al. 2006, Nature.”

These electrodes don’t penetrate neurons, but they get within a few hundred microns, which is close enough to measure electrical activity. Stick a hundred microelectrodes into the motor-control part of the brain, and you should be able to pick up activity in 50 to 150 random neurons.

It’s like listening to a symphony orchestra by plucking out ten random musicians. You’ll probably get the gist of the piece, but wouldn’t it be better if you could choose which ten musicians? (On the upside, if you’ve randomly picked musicians, it doesn’t matter if you lose track of a musician and replace it with a new one, which can happen as the electrodes move a little.)

Space-opera quality cybernetics will require one of three solutions: either a much bigger sample of neurons, a drastic improvement in our ability to extrapolate from a few neurons, or some way to identify exactly the right neurons to measure. Everyone’s brain is unique, so if you want either of the biological solutions, they must be custom-tailored to each individual.

Peripheral nerves provide an easier, but limited, option

You might be able to minimize the above two problems by recording from a peripheral nerve: For instance, to control your cyber-arm, we might try to decode signals in the nerves that would normally control your boring old human arm. Down there in your arm, you have far fewer neurons, and most of those carry the information you want: movement and sensation. But like all things in life, there’s a tradeoff.

First, peripheral-nerve implants are no use at all for patients with spinal cord injuries or other nerve damage. You can’t solve the problem at the arm level if the arm has lost its connection to the brain. Second, the arm nerve will only carry information about dynamics (joints and muscles) and maybe kinematics (movements), but not higher-level goals. Without also knowing the movement goal, your system has a lot more room for error compared with a brain-based system with access to all three kinds of information.

Action without sensation is possible, with major limitations

Can your fictional cyber-arm sense touch and motion like a human arm? If not, you’ll encounter some problems, similar to a person with nerve damage. If you can’t feel touch, you won’t be able to grasp and manipulate objects. If you can’t feel your body position (a sense called proprioception), you will lose track of your arm’s location in space, and your limbs will start wandering all over the place as soon as you stop paying attention.

But there is a partial workaround, buried in that idea of “paying attention.” There’s another way to find out where your hands are and what they’re doing, albeit one that takes work and vigilance to pull off. If you can’t feel your hands, you can learn to use them effectively as long as you’re looking at them.

The learning process is remarkably easy

If you can record from the neurons in the motor-control areas of your brain, you’ll have no problem learning the basic control of a prosthesis. Matt Nagle, the first human to receive a modern microelectrode array neural implant, was able to control a computer cursor immediately after a few seconds of software calibration because neural decoding interprets the brain’s natural movement-control signals.

Hand transplant patients illustrate another trick: Most of the tendons that control hand movement (other than drawing the thumb toward or away from the hand) are controlled in the forearm, so a replacement hand could use intact nerves.

Transhuman prostheses might be impossible

Can the human brain control truly inhuman things: an extra pair of arms, a pair of wings, a starship, Doctor Octopus? Goal-level control is certainly possible; Doc Ock thinks about his coffee cup and his robot arms will go get it. But it may be impossible to move your starship’s steering flanges as naturally as you control your own arm. This is because the human brain is an evolved system that developed in parallel with our bodies, toward the goal of producing actions that improve our survival or reproduction.

Nearly everything in our brains is rooted in our motor and sensory capabilities, and vice versa. If you ask people to make yes/no answers by pressing a button, difficult questions (e.g., controversial moral judgments) lead to different movements: When you hesitate over a difficult call, even your arm hesitates to commit. Action is not like software uploaded into the brain; it is the brain. Radically different movement control would require a radically different brain.

The brain is shaped by the body

Inhuman movement control would require an inhuman brain, but that’s not nearly as impossible as it sounds. In the last section we saw what the brain has evolved to do, but we can also frame it as a brain evolved to change. We’re born not with a detailed genetic program of our bodies encoded in the brain, but with the flexibility to mature and learn in the context of living in that body. This mutability is a phenomenon called neuroplasticity, which includes everyday learning, but also encompasses any persistent change in the brain. (Contrast “plasticity” with “elasticity,” which would be a change that doesn’t persist.)

Our brains’ perfect fit to the human body is a consequence of life in that body, with countless moments of action, perception, and feedback. When scientists modify an animal’s body right after birth, those changes to the body produce drastic changes in the brain. If the animal never uses its eyes, the visual part of its brain instead develops to control touch, hearing, or something else. However, these drastic changes require a young brain. Adult brains have less plasticity than young brains, and that difference runs deep in the motor system: Humans who lose a hand in accidents when they’re adults have different patterns of movement and brain activity compared with people born without a hand. Maybe controlling a second pair of arms is more like learning a language: If you start young, the human brain can accomplish almost anything.

This is all conjecture, of course. A young human brain might be able to wire up appropriately for a nontraditional body, or it might fail. For example, the rest of the human brain might not function well without the patterns of movement and sensation that arise from its co-evolved body.

Changing the brain could have catastrophic consequences

Here, at last, we may find a limit to human possibility. For the brain to control any kind of body, human or otherwise, it needs to start young. We can dream about rejuvenating the brain back to youthful levels of neuroplasticity, but that will always have a major, terrifying cost. After all, our brains stabilize for a reason. To destabilize the brain is to open all of it to change: our memories, personality, identity, and everything else. To return a brain to childhood is to wipe out its memories and personality.

We can’t avoid this by targeting our rejuvenation at movement-control parts of the brain. As we learned earlier, our entire brains are rooted in motor and sensory capabilities, precisely the first areas we want to update for cybernetics. It might be technically possible to rejuvenate the human brain, with advanced drugs to reawaken the neuroplastic potential latent in every cell’s genome. However, because that process would involve wiping away the brain’s growth and experiences, few people would accept it voluntarily. This “treatment” is so harmful, it’s the kind of thing a totalitarian space empire might force on conscripts and prisoners.

To add insult to injury, rejuvenating the brain could also cause brain damage. Animal research suggests that the brain’s maturation mechanisms protect cells from long-term damage. If a brain spends an unnatural amount of time in a state of youthful plasticity, it would be at increased risk for psychiatric problems such as schizophrenia. Not something you want to cause, even in your space-empire soldiers.

We know a great deal about the possibilities for human cybernetics, but authors still have plenty of room to make choices. What will we achieve, and when? What are the as-yet-untested limits of the human brain? And most important of all, how far will our society go to find those limits?