"If I check again, I'll be sure"

This is the belief the habit rests on, and it's the one that turns out to be exactly wrong. It says that doubt is a shortage of evidence, and another look supplies evidence, so another look reduces doubt. It seems obvious. It's the opposite of what happens.

Marcel van den Hout and Merel Kindt, in Maastricht, set out to explain something puzzling about people who check compulsively. They check a great deal and remain unsure. If checking worked as evidence, the most thorough checkers should be the most certain. They're the least. The researchers built a computer animation of a gas stove and had healthy volunteers, with no diagnosis, check it. One group checked the stove over and over. A comparison group checked light bulbs over and over instead, and the stove only at the start and end. Then everyone was tested on what they remembered about their last stove check.

Across three experiments, with about a hundred people in all, repeated checking of the stove did nothing to accuracy. People who'd checked many times remembered the state of the stove as correctly as people who hadn't. What changed was the quality of the memory. The recollections of the repeated checkers were less vivid and less detailed, and their confidence in their memory was lower. The light bulb group showed none of this. Checking the stove again and again hadn't made anyone wrong. It had made them unsure of being right.

Adam Radomsky and colleagues in Montreal then did the same thing with a real stove. Fifty students turned a kitchen stove on and off and checked it, in a set routine, nineteen times, while a comparison group did the routine on a tap. After the repeated stove checking, memory confidence, vividness and detail all fell. After the tap checking they didn't. The paper's title is "Repeated checking really does cause memory distrust", and the "really" is there because the first finding had seemed too strange to believe.

Look at what this means for the walk back to the door. The first check gave you a clear picture. You went back and pushed the handle, and the picture was now two pictures, and less clear. If you'd gone back a third time it would have been less clear still. By the fourth you'd have had a general, fuzzy sense of having checked, with no particular check in it. That fuzziness is the doubt. You produced it by checking. The lock was never in question.

I'd put it as plainly as I can. Doubt after checking isn't a report on the door. It's a report on what checking has done to the picture. Taking it as a reason to check again is taking the symptom as an instruction to repeat the cause.