The effects of decorrelation on the discriminability of noise samples
Susan M. Fallon, Donald E. Robinson · The Journal of the Acoustical Society of America · 1987
In a previous study, it was demonstrated, using a same-different paradigm, that the discriminability of samples of reproducible noise is a function of the correlation between members of a pair of samples to be discriminated [S. M. Fallon and D. E. Robinson, J. Acoust. Soc. Am. Suppl. 1 78, S46 (1985)]. On “different” trials, the second sample of a pair was generated by repeating a temporal segment of the sample presented in the first interval, and combining it with a new segment of noise. The data from that experiment indicate that discriminability at “threshold” is independent of the total duration of the sample, but is highly dependent on the interpair correlation (proportion of common elements). Discriminability was also affected by the temporal position of the new segment of noise. In the work reported here, the correlation (proportion of common variance) between a pair of identical samples was varied by adding a new, independent sample of noise to the sample presented during the first observation interval. The duration and position of the added sample were also varied. The similarities and differences between the two methods used to decorrelate the pairs of samples will be discussed. [Work supported by AFOSR and NMRDC.]