When you go to sleep each night, how do you know there isn’t a monster under the bed? You behave as if there isn’t, but how do you know?

Technically, you don’t. There’s a nonzero chance that there is a monster under there. But a monster under the bed would be instant proof of a creature that you’ve never heard of, probably a brand new order of animals, and new branches of physics that explain how a monster materialized under your bed from thin air.

Occam’s razor says go to sleep.

But suppose you’re five years old. Here are a few things that your 5.5 years may have taught you about reality:

These are all things you know to be true. They are the beliefs that best match both your observations and what others (including adults) say are their observations.

So let’s revisit the initial question, this time as a five-year-old: when you go to sleep each night, how do you know there isn’t a monster under the bed? As an adult, a monster would violate everything you know about the world. But as a five-year-old, it would violate none of your priors. It’s just a new creature, one of dozens you’ve learned about this year. It sneaks in like the Tooth Fairy. One more of Earth’s unknown-unknown dangers.

Interestingly, the consequences of a monster under the bed don’t change as you get older. No matter your age, the result is death. (Presumably. I don’t know if they eat you or if it’s a territorial attack. Maybe self-defense? God help you if it’s a mother monster with cubs.)

The only reason we discount monster danger as we age is that the probability-weighted consequences change. The five-year-old believes that on 0.27% of nights (one in 365), a magical reindeer farmer flies to their house from his snowbound panopticon to dole out justice. A monster under the bed seems at least that likely. So suppose they intuit a lower bound of 0.27% risk of monster attack — and each night they go to sleep, they’re buying one more ticket in that lottery. That’s unacceptably dangerous, which is why they’re so worried about it.

An adult, on the other hand, assesses the risk of monster attack at something like one in many quadrillion. Not worth worrying about, even when the penalty for being wrong is death.

So we have a child who believes they are at risk for monster attack. Our framework says that based on their experience, this is the most rational belief — it would be deeply closed-minded for a kid to believe that a monster under the bed is a priori impossible. The kid who’s not afraid of monsters is, given their information, taking on suicidal levels of risk.

In adults, first-principles reasoning is horrible at persuading people to change their beliefs (ex: all of politics). Entrenched beliefs only change through emotions and experience. So what would that look like here?

One strategy is to test the monster hypothesis experimentally, from within a child’s own belief system. For example, ask them if a monster sliding under the bed would disturb packing peanuts. If so, put some packing peanuts under the bed for a few nights. When they stay undisturbed, that’s some evidence (like Santa’s cookies) that no monsters showed up. Remember, the kid believed in monsters because that was the most rational thing to believe. Rack up a string of solid anti-monster data points, and the most rational belief will start to change.

The second strategy is home defense. As long as a child believes in a nonzero risk of monster attack, they’ll put a positive expected value on reasonable countermeasures. Learning how to do a bedtime defensive sweep of the room is simple. Many kids arm themselves with a squirt gun loaded with anti-monster spray. Some monster species die instantly when exposed to a night light. And so on.

This is a multi-purpose set of tools. And once you know them, you can use them by yourself, to hunt whatever monsters you do believe in.