On Experiential Infohazards

Content warning: Drugs, religion, abuse mention

(Epistemic status: Endorsed – lived it and seen it happen, though calling it a hazard might be overdramatic.)

Meta note: The Retrocausal Engineering Sequence isn’t over, this thought just struck me and I wanted to get it out there this week.

As I’m sure most of my readership is aware, I consider myself at least somewhat a part of the rationalist community – at the least, rationalist adjacent. This is relevant because there’s a lot of emphasis on combating biases and thinking more clearly. The question I rarely see asked (perhaps because I haven’t looked hard enough) is the one of how these biases even form, and why does thinking about them really hard seem to have such dismal results at best, and occasionally makes bias worse due to something akin to moral hazard. The answer is experience, and experiential processing is a different skill than solely the dry legible thought about patterns. The good news is experiential learning is a good way to rearrange bad patterns – the bad news is it’s also a great way to install really pernicious bad patterns. At the extremes, however, we face the possibility of experiential infohazards – the sorts of experiences that just completely blow apart your ontological frameworks and leave the experience in the vacuum left over.

Experiential infohazards are often easy to avoid at the sacrifice of some openness – drugs are the most obvious case. Some types of mental practice (meditation) are also experiential infohazards that are a little less obvious in advance without some reading – it turns out training your mind to reconceptualize suffering has some downstream effects for your value structures, which are likely to largely be built on experiences of suffering. Religious experiences can also have this quality – rituals are designed to instill a full embodied experience of the framework being engaged in, to utilize social closeness to reify sets of principles in ways that merely thinking about them will not. Experiential infohazards can also take a darker turn – abuse and trauma which tend to be less clearly signposted can lead to deeply felt experiences that alter value structure and behavior in ways that may not be endorsed.

There are multiple levels to look out for when processing experiences that are experiential infohazards – the first is obviously who has control of the experience and what do they want you to believe. Deeply ontologically violating experiences are like a much more deeply felt script breaker – you end up freezing and will take whatever lifeline you can get; the mind abhors confusion and nonsense. The next thing is what message the experience itself carries – if most people that have had a drug experience settle on universal love, then perhaps consider whether that is compatible with your long term value structure before enjoying being folded into a universal grid of souls or whatever. The third thing is, who is helping you process and what is their agenda? This one is a little less risky because in the aftermath of an experiential hazard, you’re shifting things up to the conceptual level which makes it a little easier to notice subversion. As a side note, sometimes parts of the experience will get repressed as being too far out of frame – more intense/short time frame experiential infohazards can have this quality. In this case, try to have people close to you notice behavioral distortions and connect up what you can.

In terms of generalized defenses against experiential infohazards, I don’t strictly have anything to offer. I suspect that there’s a point where you can expand the class of intense experience to which you can say “ok, had this before, it’s not that meaningful”, but I think that there will always be black swans at the edge of your experiential dataset that act as a higher class of experiential infohazard. The best I can think of from my current perspective is to develop the ability to compress experience into patterns and use this as a shield against ontologically destructive experiences that are likely to change you in unendorsed ways. Acceptance of the experience as it is can also be helpful – it makes it less of a scary, totalizing “Oh my gosh how could THAT have happened to me” thing and more into a “that was then, this is now” thing. Still, my confidence in this advice is low moderate – I haven’t gone into the experiential wilderness enough to have solid heuristics on how to maintain a value structure, or to add precision to the ways in which ontology might be altered.

Overall, experiential infohazards are usually more of an epistemic/ontological hazard than a physical “information that will kill you” hazard – though of course, there are experiences that will in fact just kill you (quite a few really!) . I suspect there are even experiential hazards that will literally kill you without causing any direct physical harm to the body – sensory data encoded in ways that turn the mind against itself. Still, occasionally going into hazardous territory can get you out of a local maxima in ways that allow you to grow – the main point is to be mindful of what you are trading for that growth, and what you are leaving behind. The you that has already had the experience will invariably consider it worth it – will the you facing the decision to have the experience think so beforehand?

Discussion questions – What are types of experiential infohazards you’ve experienced in your life? How transformative were they? What are other possible downstream effects of experiential hazards? What are some other examples?