Second and fifth graders were presented with a discrimination learning task in which each of four displays were to be paired with a response button. For one group two of the displays shared a common feature and were paired with the other response button. This common feature condition required a subject to learn only two associations if he perceived and used the two as a collative principle. For another group, the four displays shared no common feature and the four associations had to be learned. Following the original learning task, both groups were given four new displays, with common features for each of two pairs. The displays for half the subjects were printed words having a common feature of rhyme and spelling pattern. For the other half, the displays were pictures representing the words, so the rhyming names of the pictures were the common features. When the displays were words, the fifth graders performed better than second graders. When there was no common feature, fifth graders did not excel second graders. When displays were pictures, fifth graders showed a significant transfer effect. It was concluded that ability to use a common feature economically increases with age and that common spelling patterns have little saliency for second graders. (See CS 000 176.) (Author/WR)