Present people with blank cards on an assembly line, and ask them to select the cards that are slightly misshapen in some specific way. About 1 in 50 cards will be misshapen. It will be some obvious thing, like a corner is cut off. Here's the rub... one in ten cards will be a different color. It will be just-noticably different. That's it...
You see that just noticeable difference? The color difference is not what the subject is scanning for. The subject is scanning for say... a punch hole in the middle of the card like this:
And again, only about 1 in 50 cards will have the hole punched in it. The subject will control the rate at which the cards appear. I think this can even be done on computer. So in the experiment there will be basically two factors: how often the "defect" occurs, and how often a different, but non-defective card appears. I want to see what combination of rates, changes the attention of the subject. That's it. I am betting that this or something similar has been done a long, long time ago. My hunch is that if one is looking for a very rare defect (a 1 in 50 defect) and it is in the midst of a common non-defect difference (the slight variation in color of ~ 1 in 10 cards), that the subject will zero in on the non-defect difference, and spend more time and attention on the non-defect difference than the "normal" cards.
Come to think of it, I think it would be better if the non-defective but "different" cards are different because of their shape rather than their color.
I think you can guess what this experiment would "prove."