From Being Wrong: Adventures in the Margin of Error by Kathryn Schultz:
When it comes to observing imperceptibly slow natural processes—flowers blooming, weather systems forming, stars moving across the sky—we rely on time-lapse photography. If we wanted to isolate the wrongness implicit in our own gradual changes, we would need a kind of internal equivalent to that—which, as it happens, we have. Unfortunately, it is called memory, and as we have seen, it is notoriously unreliable. Moreover, it is most unreliable precisely with respect to accurately recalling past beliefs. This effect is widely documented. For instance, in 1973, the psychologist Greg Markus asked over 3,000 people to rate their stances (along one of those seven-point "strongly disagree / strongly agree" scales) on a range of social issues, including affirmative action, the legalization of marijuana, and equal rights for women. A decade later, he asked these same people to assess their positions again—and also to recall how they had felt about the issues a decade earlier. Across the board, these "what I used to think" ratings far more closely reflected the subjects' current beliefs than those they had actually held in 1973. Here, it wasn't just the wrongness that disappeared from the process of belief change. It was the change itself.
