Effect of frequency variation and covariation on auditory streaming of tone sequences

Abstract

Recent results from our lab show the masking of one tone sequence by another to be strongly related to the information divergence of sequences, a measure of statistical separation of signals [Gilbertson et al., POMA 19, 050028 (2013)]. The present study was undertaken to determine if the same relation holds for the auditory streaming of tone sequences. An adaptive procedure was used to measure thresholds for streaming of ABAABA tone sequences wherein the frequencies of the A and B tones varied independently of one another (r = 0) or covaried within the sequence (r = 1). The procedure adapted on the difference Δ in the mean frequencies of A and B tones (normally distributed in cents) with the mean frequency of A tones fixed at 1000 Hz. For most listeners, Δ increased monotonically with increases in the variance of the tone frequencies (σ = 0-800 cents), but did not differ significantly for r = 0 and r = 1. For other listeners, Δ was a nonmonotonic function of variance and differed for r = 0 and r = 1. The results fail to support a strong relation between auditory streaming and the information divergence of tone sequences.