Abstract:
Alternating tones and noises are rated as segregating into different perceptual streams [Dannenbring and Bregman, J. Exp. Psychol. 2, 544--555 (1976)]. The current study examined segregation by timbre using an interleaved melody task [Dowling, Cognit. Psychol. 5, 322--337 (1973)] in an adaptive, 2IFC procedure. The discrimination threshold for a melody interleaved with distractors was measured. The melodies were five pure tones, randomly chosen from a two-thirds octave range. In experiment 1, the distractors were either: (1) pure tones from the same frequency range; (2) narrow-band noises from the same frequency range; or (3) pure tones from a frequency range 1 oct higher (a comparison condition expected to induce streaming). The timbral difference between the melody and distractors in (2) improved performance, though it was less effective than the frequency difference in (3). In experiment 2, narrow-band noise distractors (aperiodic) were compared with two-component complex tone distractors (periodic), chosen to evoke a similar excitation pattern. Performance improved equally when pure tone distractors were replaced either by narrow-band noises or two-tone complexes. This indicates that differences in bandwidth influence streaming, but provides no evidence for an effect of differences in periodicity. [Work supported by EPSRC, UK.]