Cognitohazards
The brain as an attack surface
It’s December 16th, 1997, Tuesday. Young Misaki came back home after a long day of school and is finishing up dinner with her family. It’s finally 6:30 PM, just in time to watch her favourite anime: Pokémon. It only airs weekly, if you miss it, you gotta wait until next week. Excitedly, she huddles in front of the big screen TV. In tonight’s episode, Ash and his friends will be going inside a computer to fight Team Rocket with a Pokémon called Porygon, an artificial creature made completely out of programming code. Exciting. 20 minutes in, Pikachu is using thunderbolt to block a couple missiles heading their way, and since they’re in a computer, the explosion is a rapid electric of red and blue. Misaki feels something’s terribly wro-
White. Ceiling. Cold bright lights stare at her.
Misaki blinks. Metal taste in her mouth, tongue thick and swollen. Taste of blood.
She tries to move, but everything aches. Her neck especially. She seems to be in a bed? Her parents come into view, kissing and hugging her intensely, thanking the heavens. What happened? Her parents look at each other, not sure what to say. A man in white coat comes into view, addressing her. He tells her she’s fine now, and that she experienced involuntary muscle contractions and was unconscious for around an hour. She bit her tongue during the spasms and hurt herself really bad. Misaki doesn’t understand, she doesn’t remember anything, only that she was watching an episode of Pokémon, the details of which she cannot recall, and now she’s here.
The doctor tells Misaki that she had an epileptic seizure. She was one of the 685 hospitalized children to be watching the episode Dennō Senshi Porygon and be given a seizure as a consequence of it.
The human brain is a computational device. And like any other computational device, it can be attacked. It can be done bluntly, with a bullet or tumour. But there’s something extremely fascinating to me about the idea of “hacking the brain” in order to conduct a subtle attack.
You show a 3 second video to a person, and they break. Knocked out, unconscious. This exists. It’s estimated that this works for 2 million people right now (clinically significant effect). You have a 1 in 4000 chance in inducing this to someone. Not only that, but we have EEG data showing that for flashing light stimulation, in children aged 1-16, the average abnormal brain reaction (photoparoxysmal responses) percentage is 7.6%, 1 in 13 children. That’s 154 million children. And it somehow disappears as children become adults, with 50% of cases being resolved?
Insane. Insane. We can already hack the brain, it’s already here.
How far can it go? Sound induced? Smell induced?
Thought induced? Feeling induced?
In rodent models, simple sound seizures are actually very common. For humans? Basically nonexistent, unless we’re counting startle epilepsy: seizures induced by loud noises or sudden surprises. But music induced epilepsy does exist (1 in 10 million people have it), and can be very specific. It can be a piece of classical music, it can be church bells, it can even happen by just thinking about the song.
There’s epilepsy from calculation, solving arithmetic problems, playing chess or cards, solving a Rubik’s cube, abstract reasoning.
There’s a version of epilepsy induced by reading.
I’ve read qntm’s There is No Antimemetics Division. I’ve read Nick Bostrom’s Infohazard PDF. I’ve read BLIT / “Basilisk” by David Langford.
We already have some of these hacks and it works for at least 2 million people.
For the abnormal brain reaction, that’s around 200 million people worldwide.
If you were extremely malicious and evil, how far could you extend it?