Introduction. Have you ever had the experience of waving at someone or saying "hi" to a person, only to realize moments later that the person is not whom you thought she was? Have you ever been driving and misread a billboard or sign? Have you ever confused a 25 cent piece with a Susan B. Anthony dollar? Have you ever been cheated when getting change back from a purchase, because the clerk thought that the 20 dollar bill you gave him was a 10 dollar bill? All these situations are cases of confusion between similar patterns. Two faces can look similar, two letters or words can look similar, two coins can look similar, two paper designs can look similar.
Not everything looks equally similar to everything else, however. Presumably this is because the mind carves up patterns in particular ways, accentuating some aspects and deemphasizing other aspects. A good theory of pattern recognition should be able to account for which patterns look similar and which patterns look dissimilar.
Try it yourself. You will see a series of letter pairs, and you will give a rating of how similar the two letters are. After making 40 ratings, you will take a look at your data and decide whether a template matching model fits your data.
Do not print this page or the rating cases. After the ratings, a message will tell you what to print.