When proof is not enough: Throughout history, evidence of racism has failed to effect change

Bruner and Postman's classic psychology experiment is interesting and suggestive, the sort of thing that gets taught in psychology 101 to get students thinking beyond the naive assumption that what you see is always what is actually there. By itself it doesn't tell us how far it will generalize beyond the perception of playing cards, but other experiments have tested eyewitness testimony more directly and found that it's quite unreliable, even in situations not tinged by racism, but also in those too.

I am not sure it applies when people tell you what you're looking at and then show the evidence? That isn't about perception. It's about how much evidence you need to shift your prior beliefs.

When we're talking about generalities, the notion of "proof" doesn't quite fit the situation. Proof comes from math and logic where a single counterexample disproves a general rule. But most of our knowledge of the real world is in terms of statistical distributions - what do we consider likely? A single example can show that outliers exist, once you see it, and you can then be on the lookout for them. But outliers can also be dismissed, for example with the "a few bad apples" cliché. So it takes multiple examples to shift people's ideas about what usually happens, and scientists will hold out for statistical proof.

For example, doctors are taught "when you hear hoofprints, think horse, not zebra." Of course sometimes it really is a zebra, but seeing a rare disease once doesn't mean they expect to see it again. This has implications for people with rare diseases, where it can be quite hard to get them diagnosed correctly.

Getting people to ignore many years of life experience and think that something is common when they don't see it often is pretty hard. In an online conversation, an easier task might to try for uncertainty, to show that we are discussing something that is simply outside your experience, so you have no reason to be confident.