摘要：

Letters of the alphabet, serving as stimulus terms, were paired with either high-probability or low-probability associates, serving as response terms, in a paired-associate task. The experimental lists were presented by a tape recorder. After original learning, Ss were given an immediate or a delayed recognition task. The major experimental results indicated (a) that degree of original learning was a function of stimulus frequency and associate probability, (b) that high-probability associates were recognized significantly better than low-probability associates both at the immediate and delayed test intervals, (c) that recognition errors tended to be of a high-probability type, and (d) that there was a complex relationship among letter-frequency, associative probability, time of testing, and degree of original learning. The data were discussed in terms of the unit-sequence interference hypothesis (Underwood and Postman, 1960) and were consistent with it.

展开