Crossposted from Telescopic Turnip.

As I am writing, there is a tiny angel-version of myself suspended in mid-air over my left shoulder. His comforting voice is telling me to live a normal life and pursue happiness in normal hobbies like play-dough, warhammer or ironing. Floating over my right shoulder, a little devil-myself is telling me: “hey, you know what would be a great idea? Go on the darknet and buy some roofies!

The thing with experimental psychology is that each participant can only be tested once. For a compelling example, look at Greenstein (2020). They wanted to know if being angry makes you more gullible to misinformation. Being scientists, they performed the Scientific Method: they split their sample of participants in half – one half got a nice and professional experimenter, the other half got an absolute dickhead annoying them on purpose (“the experimenter was disorganized, dismissive, insulting, lost documents, provided only vague instructions, created unnecessary work, and interrupted the participant”). Then, they could confirm that the angry group was indeed more sensitive to fake news.

But this statistical significance comes with a bitter price: they had to deliberately exasperate ~40 innocent undergraduates. And that’s a small study. If they really wanted solid results, they would have to recruit many more participants, and taken together this would amount to a lot of total suffering. Wouldn’t it be great if we could do one arm of the experiment on one person, obliviate them and re-use the same person for the other arm?

Fortunately for us scientists who care about ethics, there may be a way. According to urban legends I’ve heard, some drugs can induce anterograde amnesia, the loss of memory of recent events (typically, what happened after you took the drug). These are reportedly used by bad people to commit all kinds of mischiefs. But we are good people – can we use these drugs for fun science instead?

For experimental purposes, the perfect drug would create strong and reproducible anterograde amnesia, with minimal side effects. If such a drug were available to the average reckless citizen scientist, there would certainly be a lot of opportunities for exploration, especially combined with good self-blinding techniques. You could test how much determinism there is in your thought process. Look for exotic Schelling Points known only to yourself – can you come up twice with the exact same haiku? What about anti-cooperation games – playing the prisoner’s dilemma or the game of chicken against yourself? Are you bad enough for some acausal trade, or evil enough to betray yourself as an act of self-trolling?

Then there is all the literature on unconscious bias and discrimination, like when researchers send fake rental applications to landlords and see if the name/race/sex of the applicant affects how many replies they get. Except here, with good anterograde amnesia, you could test your own biases. Does the sex of a politician affect whether you vote for them? Does the sex of a scientist affect how you evaluate their research? What about their start-up pitches? Take it to the meta-level: how do you evaluate a study’s methodology, depending on whether the results match your favourite ideology?

(For this kind of things, we can’t rely on self-blinding, since the participant must not know what hypothesis is being tested. Maybe ask random Internet people to submit pairs of survey questions, then go through all of them over a few sessions of anterograde amnesia. The more questions, the better, since it makes it harder to remember the specifics in case the drug is not completely effective.)

I don’t expect the perfect drug to exist. But we can look for imperfect alternatives. Most benzodiazepines (including the stricto sensu roofies) seem to be highly incapacitative at the doses required for amnesia, so they are probably not a good choice. That being said, midazolam has been used in several studies on memory, and it looks like patients were able to take part in the experiments about fine[1]. GHB could also be a candidate: apparently, it was used as an alternative to MDMA in the 90s – if you’re conscious enough to party, you’re conscious enough to take a damn survey. Regarding side-effects, Wikipedia lists nausea, dizziness, drowsiness, agitation, visual disturbances, depressed breathing, amnesia, unconsciousness and death. Not perfect, but nothing beyond what a mad scientist with iron resolve can do. Here are my questions:

  • Have psychologists ever tried to use drug-induced amnesia for experiments? If not, why?
  • Are there any known drugs that would work for that?
  • What experiments would you do if you had access to such a drug?
  • How dangerous are such drugs for self-experimentation?
  • If afriend of mine” was to order such drugs on the darknet for self-experimentation, what are the odds of being visited by the FBI shortly after? Do FBI agents even enjoy fun science?

Counting on your erudition.


  1. ^

    Here is an horror story about midazolam used as an anaesthetic, in combination with a myorelaxant. What if sedated patients feel pain during surgery, but just can’t remember it afterwards? (I’m not competent to verify these claims.)

New to LessWrong?

New Comment
5 comments, sorted by Click to highlight new comments since: Today at 12:47 AM

Just because a drug produces amnesia for episodic memory doesn't mean that it shuts down all other types of memory as well. Without a drug whose amnesic properties are well studied in a variety of contexts, you don't know to what extent the drug actually shuts down the memory mechanisms that are relevant to your experiment and to what extent.

GHB could also be a candidate: apparently, it was used as an alternative to MDMA in the 90s – if you’re conscious enough to party, you’re conscious enough to take a damn survey. 

Just because you are conscious enough to take the study, doesn't mean that your answer won't be severely affected by the drug. 

In general, I don't believe that blinding helps substantially with seeing for personal experiments. If you want to know whether you actually understand phenomena, making predictions around the phenomena and then checking your credence is more promising than fetishizing blinding. 

I think the easiest way to do these kinds of self-experiments is just to wait a long time, say 6 months, or maybe a year, between doing each branch of the experiment. If you do a lots of these self-experiments, then you really won't remember at all the minutia of what you did the first time. I think taking roofies regularly in order to learn these little things about yourself has massively negative expected payoff, though I guess taking them once to see if the concept itself works wouldn't be too destructive.

There are really no drugs that work the way you describe. Rohypnol or any other benzo will just knock you out, GHB gets you high, etc.. 

As long as you get the dosage right, there is very little risk in consuming drugs. You can also order testing kits that will verify that the substance you expect is in the mixture you've ordered. However, the economics of online drug marketplaces strongly disincentivize and harshly punish anyone "stomping" on the product, so they tend to be very safe.

It is also safe to order from them. In the best case, your drugs are shipped, the dealer deletes all information that could link you to them, and you're done. In the absolute worst case, the whole market is a honeytrap, but as far as I'm aware this has never happened, even though several LEAs have seized working markets that they could have used for this. In the average worst case, the dealer does not delete your information, but it's extremely unlikely that they could use this for any bargaining, and it's extremely unlikely that you would get any attention from it.

This assumes you're in the US, where you have a constitutional right to be secure against the search and seizure of your packages without a warrant signed by a judge. YMMV if you are in a less free country like the UK or Australia, but you will probably still be fine. Generally, the demand-side of the drug industry is not prosecuted, unless there's some other aggravating factor like getting into a car accident or beating someone up.

[+][comment deleted]2y4