The above image is of what is known in Russian folklore as a “Solar Plexus Clown Glider”. As the legend goes, the appearance of these rare humanoid creatures sets off a reaction in the brain’s ‘solar plexus’ region, an ancient part of the brain stem associated with bodily function. Looking at the creature for too long can cause headaches, which then rapidly evolve into seizures, and ends with the entire body ceasing to function. The entire process takes only about an hour after viewing the creature.

Another important thing I neglected to mention is that you don’t have to see a real Solar Plexus Clown Glider in order to cause such an attack. Studies show that if one views an image of the creature for only a minute or longer, the viewer will still suffer the same terrible fate…

Of course, everything I just said was a fucking lie. The Solar Plexus Clown Glider was created as part of an annual April Fool’s prank done by 4chan’s /x/ board, and is just an edited photo of a guy’s friend wearing a Burger King crown. The “solar plexus” isn’t even a real part of the brain.

And yet, the concept of the SPCG bears a striking resemblance to something that has been debated for centuries: the concept of a cognitohazard. A cognitohazard, in basic terms, is an idea which can kill you. It is a concept, belief, story, abstraction, or — yes — image, which can cause physical harm to a viewer.

Debating the existence of cognitohazards is a complicated concept. Some would argue that they are real, but the definitions of them differ tremendously. In this post, we’re going to dive into the history, starting with more fantastical and creative modern interpretations, then moving into more real and philosophical examples.

I understand that cognitohazards are something that can scare people, especially nowadays where its become popular in horror communities. As a spoiler, I’m going to come to the conclusion that cognitohazards do not exist, and thus it’s my prognosis that you’ll be safe reading this. Yet, if you really are afraid, you might want to look away now. We’re going to run through, like, a lot of these.

This is your last warning.

Modern Inventions of the Cognitohazard

‘Cognitohazard’ is a new term used to describe a very old concept. These ideas first started being seriously considered during the grand age of alchemy, where its sister concept, ideas of great power — uh, cognitobenefit? — were written about in texts relating to things such as the Akashic Records and the Aleph. It naturally came to people’s minds that, if there are ideas that can really help you, there must be ideas that can really hurt you. Yet, just like these so-called cognitobenefits, cognitohazards only existed purely in theory.

It was an idea that was given serious thought by both serious scholars (Umberto Eco, Jorge Luis Borges) and wicked crackpots (Aleister Crowley, Julius Evola) alike. But in today’s age, the mantle has been taken up by an unlikely faction: creepypasta writers on the internet.

The Solar Plexus Clown Gliders are a relatively popular example of a modern-day, creepypasta-based cognitohazard. Other good examples include SCP-2718 (which use a lot of clever warnings to make you believe you’re doing something very wrong by looking for this information) or Erratas (a rather cheesy horror story that relies on using a relatively common word to make it appear as if mentions of it are deleted off search engines — this however has stopped working after the story gained massive popularity). These cognitohazards are cute, harmless, and often creative. They rely on things such as unsettling imagery and psychological tricks to make you believe the information you have now come into contact with is ultimately dangerous to you, thus giving the illusion of the cognitohazard.

And yet, they are not cognitohazards.

These are stories about cognitohazards, but they do not get us any closer to discovering what those brazen alchemists were looking for (or rather, trying to avoid) back in ancient times. I, being the reckless one, decided to look and see if there were any real cognitohazards out there in the world. What follows is my theories.

Ideas That Can (Really) Kill You

If I show an epileptic a strobing image, they go into a seizure. This, by all accounts, is directly analogous to the way a normal person would theoretically see a Solar Plexus Clown Glider, or read the description of SCP-2718. You come into contact with the imagery, it hurts your brain, and potentially kills you. By this account, strobing imagery to an epileptic must be a cognitohazard.

This is where my mention of conflicting definitions come back into play. Most people would not consider epilepsy a “cognitohazard”, just a medical issue. Even though it follows all the same rules and exists in real life, there is something perhaps unsatisfying with the result.

The problem is that a strobe effect is not an idea. When an epileptic thinks about a strobe, they do not go into a seizure — they need to see the strobe directly. Of course, in our SPCG example, the idea doesn’t hurt you either — once again, it depends on how wide or narrow your definition of cognitohazard is.

Here’s another interesting thought experiment. Say you are a person living in a dictatorship. You hold the idea of free speech, of open beliefs, and of criticizing the leader. Then, the leaders men come in and shoot you. Is this a cognitohazard?

Once again, it depends. Some people say no because in order to get injured by the belief, you have to act on it — in other terms, other people have to know you believe such a thing. If you go throughout your entire life hating the dictator and loving democracy, yet no one ever knows, then you could probably make it through the dictatorship safe and sound. However, one could argue that beliefs as strong as free speech or individualism or democracy are rather intrusive: you would constantly live in fear of slipping up and saying the wrong thing, thus getting put on the chopping block. Seems pretty hazardous to me.

Speaking of which, most good cognitohazards are religious in nature. Take Pascal’s wager for instance: if you believe in God, there’s a chance you go to heaven, but if you don’t believe in God, you’ll always go to hell [1]. Therefore, the unbelief (?) in God would be a cognitohazard.

And yet, some would still argue that religious issues don’t actually have a basis in real life, as there is no real way to prove God exists. So if God doesn’t exist, and therefore hell doesn’t exist, then the belief was never really a cognitohazard — we just thought it might be one. Well alright then, Mr. Atheist… this last example might be more up your alley.

The King Of All Cognitohazards (And Why It Sucks)

The most famous cognitohazard that is both modern and real is that of Roko’s Basilisk.

Roko, a user of LessWrong (we’ll get to what that is), once made the following thought experiment: if a sentient AI were to one day take control over the world, there would be some incentive for it to “silence” all those who went against its existence. In order to do that, it could read the historical data of individuals and use their behavioral patterns to determine whether they were ‘Anti-AI’. Interestingly enough, Roko posits that this also includes anyone who did not actively help bring up the AI’s existence. Therefore, if you have read this story of Roko’s Basilisk, and then proceed to not dedicate your life to helping the AI, you are on the chopping block…

There were two reasons why this cognitohazard gained so much ground. The first I already mentioned — this theory comes from LessWrong, which is considered to be one of the more legit sites on the web. Actively used by people like Mark Zuckerberg and Elon Musk, it is essentially a congregation of software engineers who talk about deeper, more philosophical issues in regards to technology — particularly to AI. The second is that the owner of LessWrong, Eliezer Yudkowsky, replied directly to this thread expressing his clear concern about the possibility of it being a cognitohazard, and immediately locked the thread [2].

Certainly this is as legit of a cognitohazard as you can get… right?

The truth is, this entire story is riddled with holes, and I feel as though it doesn’t take too long to look at in order to figure out what those holes are. Roko’s entire premise revolves around two insane concepts: 1), that a superintelligent AI would want to torture and murder people who are “against it”, and 2) that “against it” includes people who read a forum post twenty years ago! Both of these premises conflict with the idea of a superintelligent AI — they are ridiculous, barbaric, and proven to be incompatible with what we know about smart leadership or strategic tactics. A superintelligent AI would be concerned with its detractors, certainly, but it would read the data that generals like Alexander the Great, Napoleon, and Genghis Khan had well-known strategies of “forgive the innocent and eradicate the malevolent”. In simple terms, this strategy involves only destroying those who pose an immediate, direct threat, while treating kindly your other subjects in order to assimilate them faster. Perhaps Roko had just never played Crusader Kings before… who knows.

And then there’s that thing about Yudkowsky. If you read about the CEO of “niche intellectual forum the earth’s richest people actively read” being afraid and censoring a cognitohazard, you’d put a lot of stock into what he thinks. And, to be clear, Yudkowsky is a really smart guy. But he is also… eccentric. Severely eccentric. And perhaps severely autistic too but I’m not a doctor so you don’t have to take that one as seriously. Basically, if you do read LessWrong, and already know Eliezer’s beliefs and activity on the forum, you’d know to treat this as a whatever statement.

POSTSCRIPT – What Philosophers Like To Say

Occasionally when I write these posts I discover blind-spots I woefully ran into while writing. Sometimes, the blind-spots are discovered so deep into the writing process that there is no way I can rewrite or append it, so I must put it into a postscript.

Fortunately, this one does little to negate what we’ve already talked about — its just an interesting addendum. I spent so much time talking about alchemists and mystics and HorrorTubers, that I had completely neglected to mention another very important group who actively discusses cognitohazards: philosophers! In their terms, a cognitohazard is instead an “information hazard”, and operates mostly the same way — the only difference is that in the philosophical definition it can also include information that does not directly hurt the host but can still do damage to the wider world, such as knowing the nuclear launch codes. You can read more about information hazards by checking out its Wikipedia entry.

[1] – Now that I reread this it has literally nothing to do with Pascal’s wager. But it is important for the example I’m giving.

[2] – I’m copying his exact response here because it’s pretty damn insane:

I don’t usually talk like this, but I’m going to make an exception for this case.

Listen to me very closely, you idiot.

YOU DO NOT THINK IN SUFFICIENT DETAIL ABOUT SUPERINTELLIGENCES CONSIDERING WHETHER OR NOT TO BLACKMAIL YOU. THAT IS THE ONLY POSSIBLE THING WHICH GIVES THEM A MOTIVE TO FOLLOW THROUGH ON THE BLACKMAIL. […]

You have to be really clever to come up with a genuinely dangerous thought. I am disheartened that people can be clever enough to do that and not clever enough to do the obvious thing and KEEP THEIR IDIOT MOUTHS SHUT about it, because it is much more important to sound intelligent when talking to your friends.

This post was STUPID.