Past decades of auditory research have identified several acoustic features that influence perceptual organization of sound, in particular, the frequency of tones and the rate of presentation. One class of stimuli that have been intensively studied are sequences of tones that alternate in frequency. They are typically presented in patterns of repeating triplets ABA_ABA_... with tones A and B separated in frequency by several semitones (DF) and followed by a gap of silence "_". Listeners hearing the sequence perceive either one auditory object ("stream integration") or two separate auditory objects ("stream segregation"). The initial percept is typically integration while segregation "builds up" with a time course of seconds. The psychometric build-up function, BUF (from trial-averaging), is typically monotonic and depends on DF. However, accounting for the BUF transient phase requires understanding of how the initial percept is biased to integration.
In this talk I will present a novel dynamic neural-based model for bistable auditory perception. The model is structured as a 3-layer pseudo-mechanistic network based on evidence accumulation, not competition per se. It incorporates neuronal responses from the primary auditory cortex and describes the noisy evolution of evidence against the current percept. The model accounts quantitatively for key features of streaming behavior observed in human data collected by our lab. In particular, the model matches DF-dependent mean durations, gamma-like distributions for first and subsequent durations, as well as trends of BUFs.
704 Thackeray Hall