"Yes," said Dumbledore, as he descended to the bottom of the dark stone stairs. "Let us all go home, indeed." His blue eyes were locked on Harry, as hard as sapphires.

It suddenly occurs to me that Dumbledore has seen two interactions between Harry and a Dementor. In the first one, it almost destroys him. In the second, he casts a Patronus that destroys it. Neither would seem to provide the kind of evidence that you would need to confidently assume that other Dementors would run away from you if you said "Boo" to them.

So, is this enough evidence for Dumbledore to decide that he's wrong about who broke Bellatrix Black out of Azkaban?

Also here:

I can't end without mentioning that there has been some empirical work done on investigating which cognitive features make people libertarians. The main example that comes to mind is Philip Tetlock's investigation of taboo tradeoffs. Roughly, if you present subjects with a dilemma about a hospital administrator who has to choose whether to spend a million dollars on buying a six-year-old child a kidney, or spend the same million dollars on hospital equipment, doctor salaries, et cetera, what you discover is that most subjects, liberal or conservative, want to punish an administrator who even thinks about the question. People who identify as libertarian don't get angry at the administrator for thinking about it. And the first obvious interpretation of an experimental result isn't always the correct one, but sometimes, you know, it is.

One thing you've left out of the analysis is the cost of the signal. An ivy league degree may or may not be mostly signal, but either way, very few low status people will be able to get one.

So, one way that signals can lose their meaning is if the cost drops. Take the example of buying a drink. If drinks are expensive, then buying drinks for women you don't know is a costly signal. (Note that there's an equilibrium here; the more successful the tactic becomes, the less costly the signal becomes.) As we all become wealthier and the relative price of drinks goes down, the statement being made about disposable income goes away, and it becomes just a ritual.

I would say, then, that the primary driver for a signal to expire is not for it to be common knowledge that it's a signal, but for it to be common knowledge that it's a cheap signal. Knowing which signals to pay attention to reduces to the question of accurately estimating how costly it is.

Unless you're going to cloister your children until the state forces you to release them into the world, they're going to encounter plenty of irrationality without your needing to deliberately lie to them. I would try leading by example, rather than hoping for an epiphany that may never come.