This past weekend, I was sitting on a tarmac, my plane queued up in a long line, waiting for a runway to open, and I was staring absently at the lights embedded in the asphalt. On some of the straightaways, I could see a long line of green lights, but there were a couple spots along the way where the lights alternated orange-green-orange-etc.

I started automatically generating hypotheses. I couldn't see any straight line paths with orange lights, but maybe it was a question of the angle I was seeing them at. The nearest set of orange lights I could see was on a curve; maybe unless you were looking at them dead-on, they were orange, reminding the planes to stay on track. I craned my neck and looked around once my plane started trundling forward, to see if any individual light flicked over once I was at a different angle. No dice.

Ok, on to the next idea. Maybe the orange-green-orange pattern marked curves, so they were neatly differentiated from straightaways. This one held a little longer, until I spotted some short, straight stretches of alternating lights, just the width of a runway, where one trail crossed another. That led to my last hypothesis, that the orange lights marked any place where a plane could be crossing someone else's path (after all, most of the turns were merges). But my much delayed plane finally took off before I had the chance to try to falsify that last idea.

While I was trying to puzzle out the runway lights, every time a hypothesis failed I had a quick flair of frustration. I'd actually learned more, but it felt a little like I'd screwed up, and I had an impulse to add epicycles and twist the facts to fit the theory. Ruling out hypotheses meant I had more data, and a slightly richer model of the signal, even if I didn't yet know the rules. The flicker of annoyance faded out when I reminded myself of this fact.

But I was playing on an easy setting. I had no skin in the game, and I didn't even have a seatmate to embarrass myself in front of since I was speculating privately. It's nice to encounter the rationalization reflex in a low-stakes environment, where it's easier to notice and overcome. Generating and discarding hypotheses is a way to be engaged with my surroundings and to drill the habit of being excited to discover a theory doesn't work.

The next time I start to flinch away from noticing a hypothesis doesn't hold up, I can remember that I know not to trust that feeling. Or, more accurately, that it's a signal that I know more than I did a moment ago, just like the pain in my arms when I do push-ups is a sign I'm getting stronger.