Some figural properties of auditory patterns.

Listeners identified one of six permutations of three frequencies, presented as brief three-note melodies. Identification performance remained high in spite of transposition of the original three frequencies throughout a two-octave range, so long as the musical intervals or frequency ratios between the adjacent pairs of frequencies remained constant. Even when those intervals were compressed or expanded, while remaining about equal to each other, identification was quite good for the range between the lowest and highest frequency of no more than approximately 1/3 octave. Performance decreased sharply when the span was much wider. Unequal intervals, where the low and middle frequencies were closer together or farther apart than the middle and high frequencies, did not retain good identification performance. When the three-tone patterns were embedded in longer sequences of seven or eight tones, the identification performance was best when the pattern occurred at the beginning or the end of the sequence, and when the range of frequencies from which the irrelevant background tones were chosen lay outside the range of pattern frequencies. Under conditions where the background frequencies were fixed and the pattern frequencies were moved, thus combining the manipulation of embedding with that of transposition of the pattern, overlap of pattern and background frequencies was still the principal cause of deterioration in performance. The findings are related to some analogies to the perceptual rules of Gestalt theory, as well as to certain aspects of musical practice.