Authors:Alexandra Bendixen, Tamás M. Bőhm, Orsolya Szalárdy, Robert Mill, Susan L. Denham, and István Winkler
Sound sources often emit trains of discrete sounds, such as a series of footsteps. Previously, two different principles have been suggested for how the human auditory system binds discrete sounds together into perceptual units. The feature similarity principle is based on linking sounds with similar characteristics over time. The predictability principle is based on linking sounds that follow each other in a predictable manner. The present study compared the effects of these two principles. Participants were presented with tone sequences and instructed to continuously indicate whether they perceived a single coherent sequence or two concurrent streams of sound. We investigated the influence of separate manipulations of similarity and predictability on these perceptual reports. Both grouping principles affected perception of the tone sequences, albeit with different characteristics. In particular, results suggest that whereas predictability is only analyzed for the currently perceived sound organization, feature similarity is also analyzed for alternative groupings of sound. Moreover, changing similarity or predictability within an ongoing sound sequence led to markedly different dynamic effects. Taken together, these results provide evidence for different roles of similarity and predictability in auditory scene analysis, suggesting that forming auditory stream representations and competition between alternatives rely on partly different processes.