It has been pointed out that my previous article about the famous “the talk: non-black version” seemed to imply that I do not understand the full weight of the “only when you have no better information” qualifier.
The problem is that the qualifier doesn’t work, not because the qualifier’s not strong enough, but because humans are not rational enough. To put it simply, humans can’t actually make rules like that work. There are a few issues.
Issue #1: Lazy evaluation
The human brain inclines strongly towards not doing unneeded work. What this means is that if you have a heuristic to hand, your brain will tend to use it and present the result as if it were knowledge or well-supported reasoning. You can override this, most of the time, but doing so requires extra effort, and humans cannot maintain that kind of extra effort consistently over time. (For a much more detailed discussion, see Kahneman’s Thinking, Fast and Slow. Give particular attention to the statistics on judicial review and how outcomes vary over the course of a day’s work.)
So even if you have a rule which is duly qualified to be used only when no better data are available, the qualification will be silently disregarded. You might catch this sometimes. You won’t catch it all the time.
Issue #2: Theory-laden data
So assume you correctly apply the qualifier. Does it work? The answer, perhaps surprisingly is still “no.” The problem is that we often forget that we don’t really have access to unfiltered experience; we have to make sense of things in terms of our understanding of how the world works. What this means is that the data from which we reason are theory-laden, meaning that they already reflect our presuppositions.
When dealing with social interactions, we are well outside the realm of things we can evaluate without assumptions. The key is that assumptions about relationships and attitudes can radically transform the experiences we have.
Two of my friends, who have been close friends for many years (close enough that their parents kept assuring them that, if they were a couple, they could just say so and no one would reject them), habitually insult each other. People who don’t know them hear their interactions as hostile and abusive. And this leads to feedback; someone who hears the hostile interactions first will tend to interpret their friendly interactions as a temporary ceasefire with resentment bubbling underneath, and thus continue to evaluate them as much less friendly than they really are.
By contrast, people with social anxiety disorders will massively misinterpret social cues as more hostile or judgemental than they are. One of my friends once complained about how mean a mutual friend had been, making fun of her teeth. I was confused, so she explained; he’d made some comment, then looked at her mouth and made eye contact with me, which clearly indicated that he’d been talking about her teeth. (She felt she had bad teeth. I was totally unaware of this. So was the friend.)
So if you start with a set of rules you use “only when you don’t have specific knowledge”, the specific knowledge you get will be biased to at least some degree by those presuppositions. In social interactions, especially when forming first impressions, that can create wildly divergent sets of interpretations. Even apart from the theoretical problems, people recognize things they are prepped for more readily, even when this is inaccurate; shown neutral facial expressions, people are more likely to “recognize” expressions corresponding to a mood they have reason to anticipate.
Issue #3: Game theory.
Social interactions with strangers are very much subject to the classic failure modes of things like the prisoner’s dilemma. Imagine that two people meet, and that they are able to tell in advance that each is in a group which has some history of hostility or bigotry towards the other. Doesn’t really matter which groups you use; it could be a black guy and a white guy, or it could be a leather fairy and a priest. Each will start at least alert to the possibility that the other will be hostile. The thing is, guarded-and-defensive and hostile look like each other at least a little. So if they both start out defensive, and prepped for hostile interactions, they are likely to end up being mutually hostile, with each believing that the other initiated it.
You can somewhat overcome this simply by overtly rejecting hostility and demonstrating friendly intent. But to do this, you have to be willing to genuinely commit to it (or be a much better actor than most people will ever be). A set of rules that encourages you to be defensive and not-open rules that out.
So if both parties follow a set of rules similar to those proposed when dealing with perfect strangers, they will be very likely to have consistently negative (or at the very least non-positive) experiences. This then reinforces the stereotypes and presuppositions.
So in conclusion…
While in general it’s true that you should favor accuracy over desireability in evaluating the world, doing so with social rules can be massively counterproductive, because your presuppositions can influence outcomes quite dramatically. Starting out prepared to deal with hostility makes you more likely to perceive hostility whether or not it is present, and more likely to act in ways that provoke hostile responses. It does this no matter how smart you are, and no matter how committed you are to avoiding it; human brains simply cannot be that rational.