LAUSR

Temporally homogenous sound textures—as produced by rain, fire or insect swarms—are thought to be represented with time-average statistics measured from early auditory representations. We explored the averaging process involved in sound texture perception using “texture steps”—stimuli whose statistics changed at some point in time. We reasoned that judgments should be biased by the stimulus history included in the averaging process. Listeners were presented with two sounds, a texture step and a probe texture (with constant statistics). Listeners were asked to compare the endpoint of the texture step to the probe and to select the sound that was most similar to a familiar reference texture. Judgments were biased by the presence of a step 1 second prior to the stimulus endpoint, but the bias was reduced when the step occurred 2.5 seconds from the endpoint. In addition, the bias was substantially larger for textures whose statistics were less homogenous (ocean waves) than textures whose statistics were more ...