DETECTING MISTUNING IN THE PRESENCE OF SYNCHRONOUS AND ASYNCHRONOUS INTERFERING SOUNDS

Listeners were presented with the first seven harmonics of a 500-Hz fundamental (the ''test complex''), and were required to detect a mistuning imposed on its fourth harmonic (the ''target''), which had a nominal frequency of 2000 Hz. All components were frequency modulated by a 5-Hz sinusoid, and, in the signal interval of each 2IFC trial, a 180 degrees phase delay was imposed on the modulation applied to the target. This led to a mistuning proportional to the depth of FM, which was varied adaptively to obtain thresholds. When an unmodulated 2000-Hz ''interfering'' sinusoid was gated on and off together with the test complex, thresholds were largely unaffected, provided the level of the interferer was more than 3 dB below that of the target. However, when the interferer started before and ended after the test complex there was a substantial elevation in threshold, even at low interferer levels. The elevation was mainly due to the portion of the interferer preceding the test complex, but the part following the complex also had some effect. When the offset asynchrony between the interferer and the test complex was held constant at 100 ms and the onset asynchrony varied, the highest thresholds occurred at an onset asynchrony of 200 ms. A threshold elevation could also be achieved by replacing the interferer with a narrow-band noise centered on 2000 Hz, or by preceding the test complex with an interferer consisting of all components except 2000 Hz. The elevation could be reduced by presenting the leading portion of the 2000-Hz interferer as part of a harmonic complex with an F0 of 333 Hz, or even by presenting it as part of a complex with which it did not form a harmonic series. It is concluded that the effect of the asynchronous interferer was to pull the target component into a separate perceptual stream from the rest of the test complex, and that this prevented listeners from detecting mistuning by comparing its frequency to that of the remaining components. Implications of these findings for theories of perceptual streaming and of mistuning detection are discussed.