"On the Perception of Incongruity: A Paradigm," is a brilliant psychological study on the nature of perception that was done in 1949 by Jerome S. Bruner and Leo J. Postman. It was first published in the Journal of Personality, No. 18, September 1949. In the experiment, 28 test subjects were repeatedly shown playing cards in a tachistoscope for increasing lengths of time, gradually increasing from 10 milliseconds to a full second, with three exposures at each speed. Unbeknownst to the subjects, half of the cards they looked at were doctored — specifically, their color and suit were mismatched. For example, there was a black three of hearts and a red six of spades. After each exposure, they asked the subject "What did you see?" If the subject answered correctly twice in a row, it was considered a correct identification, and the time exposure was recorded.
Bruner and Postman found that the average threshold recognition time for the normal cards was 28 milliseconds, whereas the average for the doctored cards was a whopping 114 milliseconds. Not only that, but all of the normal cards were recognized by 350 milliseconds, whereas 10% of the doctored cards were still not recognized at the longest exposure used, 1000 milliseconds (i.e. one full second). They also found that subjects who had already identified one trick card were somewhat faster at correctly identifying another trick card, and those that had identified two or more trick cards were much faster at identifying further trick cards. Furthermore, they found no statistically significant difference between trick card identification times with subjects that had not used the tachistoscope yet and those that had already identified one normal card, ruling out the possibility that practice with the tachistoscope was the primary cause of the time difference. (In fact, they found that subjects who had correctly identified a normal card took slightly longer to identify a trick card than did those who had not used the tachistoscope before at all.)
What truly makes this study extraordinary, however, is that the results do not stop here. Bruner and Postman found that they could qualitatively classify the different types of reactions that they saw the test subjects have to the trick cards. The reactions fell into four groups: dominance reactions, compromise reactions, disruption, and recognition of incongruity.
Dominance reactions occur when the test subject simply sees the trick card as a perfectly normal card. For example, a black ace of diamonds might be seen as a regular ace of spades or a regular ace of diamonds, with no evidence of uncertainty on the part of the subject. 96% of the test subjects showed dominance reactions to the trick cards in their records, some much more so than others. In a few cases, dominance reactions prevented recognition of the trick cards entirely; for example, one subject identified the black three of hearts as a three of spades 44 times in a row and never saw otherwise.
Compromise reactions occur when a test subject describes a trick card as being somewhere between a normal card and its actual appearance. For example, red spades were described by some subjects as black with red edges, black lit with red light, black on a reddish card, blackish brown, black but not quite, etc.. Approximately half of the subjects showed such reactions to trick cards at some point in the experiment.
Disruption is "a gross failure of the subject to organize the perceptual field." Basically, it's when the subject expresses that they don't know what they saw with a response such as "I'll be damned if I know now whether it's red or what!" Disruption responses did not occur on many exposures, but 57% of the test subjects went through disruption at some point in the experiment. Typically, they would occur in the following manner: the subject would have a dominance reaction at first, and start to erroneously report with confidence the identity of the card somewhere around 50ms (close to average for a normal card). This would go on for a series of exposures up to some time exposure well over the usual recognition threshold for normal cards, say 100 or 200 ms. Then the subject would suddenly start to lose confidence and eventually would be disrupted, with an exclamation such as: "I can't make the suit out, whatever it is. It didn't even look like a card that time. I don't know what color it is now or whether it's a spade or a heart. I'm not even sure now what a spade looks like! My God!"
Recognition of incongruity is exactly what it sounds like: correct recognition of the trick cards. This did not always happen, and when it did it was frequently accompanied by exclamations such as "Good Lord, what have I been saying? That's a red six of spades." — what Bruner and Postman informally termed the "My God!" reaction.
Bruner and Postman theorize that these reactions are all caused by the mind trying to match perception to expectation. As they put it, "perceptual hypotheses [tend] to fixate after receiving a minimum of confirmation." The subjects were expecting to see a normal playing card, and once they had even the slightest evidence that that was what they were seeing, they ran with it — a dominance reaction. If they didn't get overpowered by the dominance reaction, then perhaps they realized that there was something wrong, but unwilling to make the full leap to recognition, they had a compromise reaction, or became disrupted. In most cases, the correct identification was made eventually, but it was almost always preceded by one of these reactions.
This whole thing turns everyday ideas about how perception works upside down. Traditionally, perception is supposed to work something like this:
1. Data from the external world reaches sense organs.
2. Sense organs send data to brain.
3. Brain looks at data and sees what is in external world.
This experiment would indicate that the process is more like:
1. Data from external world reaches sense organs
2. Sense organs send data to proper sensory center of brain.
3. Sensory center molds data to fit known patterns and sends altered data to higher-level decision-making areas of brain.
If you want to take it one step further, the experiment can be viewed as proof that what we see is not primarily determined by the nature of the external world, but rather by what we expect to see. More accurately, it's evidence that the basic perceptual machinery of the human mind is made to look for patterns, and more importantly, when reality does not fit a deeply-ingrained pattern, our perception of reality may be warped by our own mind to fit with the pattern anyhow. And if you're really extreme, you can take it even further and say that it's empirical evidence that reality is, to some degree at least, subjective. (This last idea is a little bit circular, but arguing for or against it is another node's job.) At any rate, it's certainly a cool idea, and it's been heavily debated in the field and continues to influence a great deal of research into perception (as do Bruner and Postman personally, both of whom are still alive and very much kicking). The study has also been widely cited in numerous places outside of psychology, most notably in Thomas S. Kuhn's 1963 classic, The Structure of Scientific Revolutions.
The study itself is available here, if you're interested:
http://psychclassics.yorku.ca/Bruner/Cards/