Abstract:
When listeners are asked to discriminate between trials on which a sample of noise is presented twice and trials on which two nonidentical samples are presented, discriminations are easier if samples differ near the end rather than near the beginning. The effect of temporal position [S. F. Coble and D. E. Robinson, J. Acoust. Soc. Am. 92, 2630--2635 (1992)] is robust over a wide range of stimulus parameters including duration, level, and correlation. In previously reported work [M. E. Rickert and D. E. Robinson, J. Acoust. Soc. Am. 98, 2906(A) (1995)], discriminability was measured with wideband (100--3000 Hz) and narrow-band (455--655 Hz) noise. The size of the effect of temporal position is smaller with the narrow-band noise. However, one could argue that these effects are restricted to the lower region of the frequency spectrum because center frequency was held constant ((approximately equal to) 545 Hz). In the experiment reported here, both center frequency (545 and 2000 Hz) and bandwidth (200 and 1000 Hz) were investigated. Data were also obtained with wideband (100--3000 Hz) noise. Although the effect of temporal position is consistent across stimulus conditions, the size of the effect is reduced for narrower bandwidths. There is no evidence of an effect of center frequency. Results are discussed in terms of a leaky integrator model [D. E. Robinson and M. E. Rickert, J. Acoust. Soc. Am. 98, 2906(A) (1995)]. [Work supported by NIH and AFOSR.]