On intuition

I was talking to another parent at my kids’ school, and we were talking about science. You have to understand — it’s a Steiner School, so the people there are much more likely to be new-age types and homoeopathy believers (among other things) than at other schools, with all the anti-science baggage that goes with it. (One parent once told me, “I like that science can’t explain why homoeopathy works.” My answer: “It can — gullibility.”)

Anyhow, this parent said something like “Science is all very well, but there’s a higher form of knowledge, and that’s intuition.”

Is intuition any good?

We have amazing brains. We notice things without even noticing that we’re noticing. Our brains are great at taking mountains of data and coming up with patterns. Not that we do it optimally, of course, but very well in most normal cases. So we could see intuition as our brains trying to come up with conclusions, in such a subtle way that we aren’t able to articulate how we arrived at these conclusions.

And yet… one has to watch out for confirmation bias. When we find out that an intuition was correct, we take it as a ‘hit,’ but we seldom notice intuitions that turn out wrong or that go unfulfilled.

There’s another possibility. I remember when a prominent BYU professor came out and said he was gay. Afterward, I heard a few students say something like “Oh, yeah, I had a feeling about him,” after the fact. I never heard anyone say that beforehand! Did they really have an intuition? Or did they retroactively rewrite the story of what intuitions they had in order to fit what actually happened? I suspect the latter. It’s easy to do. It makes you feel more knowing.

So if intuitions are so malleable and interpretable, what good are they?

In my work in Natural Language Processing, I get intuitions all the time. After all, I’ve been speaking a language for quite a while, so I should be able to have some shadowy ideas about what kinds of features would work best on a certain task, or what kinds of patterns we should expect in dialogue. But I never trust my intuitions unreservedly. I make every attempt to verify them. They don’t always hold up, but if they do, I keep them.

It’s all about keeping everything in its place. Intuitions, though useful, are down there with hunches and guesses. They’re certainly not above facts, as that Steiner parent would argue. That’s just upside-down. You have to know how seriously you can take an idea.