It's a cognitive bias, since we remember the events that match our expectations and don't keep track of experiment over time.

This is a rather large assumption. I have had plenty of times when I thought I had noticed a trend of some sort and turned out to be mistaken, and so stopped relying on the heuristic. Insisting that everyone is biased (as opposed to observing that anyone can be) is a good way to filter out unexpected and perhaps unwelcome observations.