I used to be a high school biology teacher, and I noticed that students often found it easier to learn irrelevant information than the information I was trying to teach. When learning the steps involved in the process of cell division, the students simply memorized the pictures in their textbook. If I tested them using a different set of diagrams depicting the same process, they had difficulty completing the task. They had memorized a set of pictures, instead of the concept of mitosis.
We can learn things even when we're not consciously trying to. For example, if we try to learn a set of shapes, some of which are relevant to the task, and some of which are not, we'll often end up learning the irrelevant ones as well. Take a look at the following image:
If people are trained to recognize blocks with stars in them, they will also learn that stars are usually paired with circles, even though they were never explicitly trained on shape pairings.
How does this accidental learning (usually called "statistical learning") occur? Is there a pattern to the "extra" information we notice? Chris Baker, Carl Olson, and Marlene Behrman (of the Mellon Institute, the University of Pittsburgh, and Carnegie Mellon) developed a series of experiments to test some of the factors involved ("Role of Attention and Perceptual Grouping in Visual Statistical Learning," Psychological Science, 2004).
Baker and his colleagues trained volunteers to recognize 8 shapes out of a set of 16. Shapes were displayed in pairs, one above the other, and connected with a bar. Each shape could be paired with 4 of the other shapes, but two of the pairings occurred four times as often as the other two pairings. Participants were more accurate and responded quicker for the frequent pairings. They also reported that the frequent pairings were more "familiar" than the infrequent pairs, and both types of pairings were more familiar than the trained objects paired with new shapes. In a new experiment, when the pairs of shapes were not connected with a bar, the results were the same.
Next they repeated the experiment, but this time, the shape the volunteers were asked to recall always appeared below the paired shape. Now they reacted faster, and still reacted faster for the more frequent pairings, but the difference in reaction time between frequent and less frequent pairings was smaller.
Finally, they conducted the experiment with no bars connecting the pairs, and again with the shapes the participants were learning always appearing in the same place. The irrelevant paired shape was close by, but there was no physical connection to the trained shape, and there was never a reason for participants to look in the spot where it appeared. In this condition, all differences between frequent and infrequent pairings disappeared.
So when participants had to pay attention to both locations where a shape could occur, the results were the same whether or not the shapes were connected. But when they only had to pay attention to one location, then the paired shape only influenced results when it was connected to the studied shape. Participants don't accidentally learn irrelevant things all the time; there must be some reason for them to connect the irrelevant object with the relevant one.
In other words, if I had taught my students using many different diagrams of mitosis, they probably would have attended only to the actual phases of the process, and not to the particular stylistic differences between the drawings. Let's hope I didn't squander the careers of too many future biologists with my teaching methods!
It's very intesting! I'll also contribute to your website when I've similar experience to share.