Octave generalization and the consolidation of melodic information

Subjects listened to a standard six-tone melody, followed by a comparison melody. The compar-ison melody was always transposed up four semitones from the standard, and this transposition was either exact or it was inexact while maintaining contour. Subjects judged whether or not the comparison melody formed an exact transposition of the standard. It was found that repetition of the standard melody, before presentation of the comparison, resulted in a significant improvement in performance compared with a condition featuring only a single presentation of the standard. A significant improvement also resulted when the repeated melody was displaced intact an octave higher or an octave lower. However, when the standard melody was repeated such that its com-ponents were displaced alternately to the higher and lower octaves, performance was significant-ly worse than when the standard melody was not repeated at all. Thus, in this paradigm, octave equivalence effects for single tones did not operate in the consolidation of melodic information. It is clear from a variety of evidence that a strong perceptual similarity exists between tones that are separated by octaves (Deutsch, 1973). Several investigators have therefore suggested that pitch be treated as a bidimen-sional attribute: the first dimension represents overall pitch level, or ‘tone height, ’ and the sec-ond defines the position of a tone within the octave, or ‘tone chroma ’ (Meyer, 1904, 1914