BCI: Politically-Incorrect Game Theory

Human beings love games. Human beings love competition, especially when that competition allows the comparison between two entrants who previously could not share an arena. Deep Blue over Garry Kasparov in 1997 seems almost quaint in retrospect, but its attention shows the allure of juxtaposing two wildly different agents and then receiving the answer to a burning question. In this case, “can a machine beat a human at chess?”

Of course, most of us aren’t world-champion grandmasters. And yet, we’ve all ridden along for this journey. First, board games required human beings seated together, using their limbs to move around some pieces of elegantly carved wood (or cheap plastic, depending on the relative affluence of grandparents). Next, technology allowed a human being to compete against a machine, when a human being was unavailable. Soon after, the internet allowed two human beings separated by oceans to compete against each other or in cooperation against a machine. In fact, this might offer a narrowly defined Turing test.1 I certainly couldn’t tell if the agent kicking my butt all over a chessboard was Magnus Carlsen or an algorithm! This still required human beings to look at a screen and push some buttons. No one thought much of this constraint.

But what if your limbs and digits are paralyzed? Enter brain computer interface technology (BCI). Suddenly, a human lacking use of his limbs regains agency, playing mahjong with his mind. Imagine if suddenly, we could play chess against a distant opponent (flesh or machine) simply with thoughts. Would we even need to represent the board on a screen, or could we communicate its state to the brain directly? This is sufficiently intriguing, but your diligent blogger has other thoughts.

A few years back, John McEnroe found himself embroiled in a controversy2 by comparing Serena Williams with male professional tennis players unfavorably. Periodically, some asinine pundit asks some athlete what might unfold if Serena played Roger Federer or some other current star, the star has the audacity to answer the question accurately (the guy would win handily3), and the pundits rage for a couple weeks about chauvinism, we discuss Billy Jean King and Bobby Riggs4 again, etc.

But what if Roger Federer could compete against Serena Williams? What if in lieu of some rage-inducing, flame war about upper-body strength and lateral agility we could facilitate one virtuoso’s skill competing against another’s? Strategy, stroke-production, positioning, and the brilliant electrical impulses in both genius’s neural networks. BCI may never provide the answer to settle all past and future debates, but it might facilitate a spectacle more than 50 million would watch (I would!). What if those incessant debates about Lebron vs. Michael Jordan on some astral plane’s blacktop could be settled?

Better still, what if you could compete against a childhood idol by simulating the totality of the experience? What if electrical impulses, delivered to the proper locations of your neural architecture placed you upon your own field of dreams? What if that same simulation allowed for your physical abilities to become irrelevant5–your brain being all that is required?

Ah, but before we settle all athletic disputes with technology (admittedly, I’m sure there are fewer bar brawls over some piece of arcane statistical sports trivia since the invention of the smartphone), matters aren’t so simple. A dualist might argue that the brain can be disembodied. However, the current prevailing view characterizes brains as embodied agents, linking mind and body via feedback from each. In other words, Michael Jordan’s brain in 2021 is a byproduct of his 2021 body.6 More concretely, if I were to amputate my arm, the areas of my brain responsible for that arm’s movement would immediately begin ceding ground to other functions. My arm movement repertoire doesn’t disappear entirely, which is why paralyzed individuals can control robotic arms with their minds. Still, the plasticity of the brain that allows extraordinary human achievements ensures that it will be forever intertwined with the body it inhabits. So Lebron still wins and my Chicago-born wife is still angry about the discussion.

The technology for such competitions seems impossible. Of course, the idea of a human losing a chess match to a pile of circuits and microchips seemed equally unthinkable not long before. And if that seemed unthinkable, even when a supercomputer ultimately did best a human, how would the idea of a fair competition with a device the size of your iPhone have sounded? Landing on the moon seemed unthinkable when Charles Lindbergh was attempting to fly across the Atlantic.7 It always seems impossible, and then, before we can lift our heads to look over the horizon, it is upon us.

Ah, but even this is not always so. Computers rapidly overtake human beings in the types of well-defined games tailored to the advantages a computer possesses like memory, speed, precision, etc. Computers overtaking humans in Go required an additional two decades beyond Deep Blue. Transfer learning8 is still an arena where computers cannot even begin to compete with average human beings, let alone the most skilled professionals of the species. A human being with a driver's license can sit behind the wheel of almost any car, locate the gas and brake, and navigate a physical world of traffic lights and jughandles.9 The models of the world one must calibrate internally to manage such a feat is miraculous, and whether an in silico solution can exist is hardly settled. Language models improve rapidly, but do they understand human intellect or just deploy increasing computational power?

So how might we traverse the intellectual landscape between chess and simulations of athletic competitions? Perhaps we begin with simple inputs (e.g. a game controller rather than a human limb’s degrees of freedom). Next, we would attempt to decode complex human motion, then provide stimulation of neural regions to simulate the sensory experiences associated with that motion and the responses it might yield. Could a computer process the trajectory of a tennis ball faster than a human mind, feed the result to the human mind, and facilitate a response even faster than our intuitions and experiences might allow? Who knows, but it begins with understanding the latent patterns of human neurology. (And we’re excelling at that!).

Today, folks play mahjong with their minds. What’s next?

1

Turing tests have academic definitions, but the basic idea is the inability to tell the difference between Arthur C. Clarke’s HAL and Hal Johnson sitting on a bench, sipping a Starbucks latte.

2

McEnroe? Outspoken? “You cannot be serious!”

3

At the risk of wading into the very mire I decry, let’s settle this. I can agree that Muhammad Ali (heavyweight) would have obliterated Sugar Ray Leonard or Marvin Hagler (middleweights) without asserting the inferiority of middleweight boxing or the middleweight boxers themselves. Serena herself has acknowledged that she could not compete with top male players. And she’s still among the greatest athletes of the last century regardless. Shelley-Ann Fraser-Pryce is an impossibly gifted sprinter, and cannot post times comparable to Usain Bolt. Got it? Good.

4

Again, not a good line of argument for either party. Riggs was 55 when the famous match occurred, and had beaten the current women’s #1 (Margaret Court, age 30 at the time) earlier in the same year. That fact is blithely omitted by one side while the other seems to forget that the 50 million who watched helped raise the profile of women’s tennis for decades to come, to the benefit of basically every fan of the game.

5

When I was a child, I dreamed of being an adult and dunking a basketball. Now I’m an adult…and I’m still waiting. I feel like those teachers who told me I could do anything deserve frank conversation with gravity.

6

And possibly the psychological trauma endured mismanaging the Charlotte Bobcats/Hornets.

7

Just over 42 years before Apollo 11.

8

Again, there’s an academic definition, but the idea is that a model designed for one task can help with another.

9

At least if they’re from New Jersey. They seem to survive despite the aggression and the jughandles.

No one works with an agency just because they have a clever blog. To work with my colleagues, who spend their days developing software that turns your MVP into an IPO, rather than writing blog posts, click here (Then you can spend your time reading our content from your yacht / pied-a-terre). If you can’t afford to build an app, you can always learn how to succeed in tech by reading other essays.