Well, as it happens I found a copy of Norman's book, and discovered that it is the wrong one. The perception model that I had in mind is called "template matching" and is mentioned only in passing in the book I mentioned.
I haven't been able to find a better reference, but the essence of the model is that the human senses (vision, in particular) can deliver more information to the brain than can be easily processed in real time. If something in the current "scene" comes close enough to matching a previously-experienced image, the brain essentially decides that the two correspond, and ceases to evaluate further input from the current scene. In effect, the remembered "template" is substituted for the present scene. Not all cognitive psychologists agree with this theory, but it does handily account for a lot of phenomena, such as the notoriously unreliable eyewitness testimony, the effectiveness of camouflage, and many magician's tricks.
What this has to do with reproducing "vintage" pictures is that the initial impression has to be strong enough for it to be identified as something that it is not (i.e., an actual vintage image). If this can be accomplished, it will be somewhat difficult for the average untrained observer to concentrate on whatever discrepancies are actually present. (The class of "trained observers" in this context includes laboratory scientists, crime scene investigators, some artists, and........photographers!)
If I can find the reference that I had in mind, I'll post it; if anyone else knows of a good one, I'd be interested in hearing of it.