Neuronal correlates of auditory stream segregation - a magnetoencephalography study

Background: The ability of humans and animals to focus their attention on a single sound source among other surrounding sounds and background noises is based on the perceptual phenomenon of "fission". Using different types of stimuli (amplitude-modulated or pure tones) we investigated to which extent the percept of streaming (the perceptual segregation of auditory events into separate streams) basing on temporal cues exceeds the peripheral channeling mechanism and correlates with alterations in transient evoked brain responses. Additionally, we investigated the influence of attention on auditory stream segregation.Material and methods: Fifteen subjects (9 females) participated in the study. The stimuli were sequences of repeated triplets, either amplitude modulated (AM) or pure (PT) tone-bursts (A-B-A). Stream segregation basing on spectral cues was modeled varying the fundamental frequency of the B-tone (in PT and AM triplets). Segregation based on temporal cues was achieved varying the modulation frequency of the B-tone in AM triplets. A neuronal correlate (measured by MEG) corresponding to the change in the perception of the sequence due to changes in B-tone was assessed by P1-N1-P2 evoked complex to the B-tone in the triplets. Alterations in the component amplitudes and latencies were correlated with the perception of streaming or non-streaming.Results: We found a significant larger latency of the N1 component in the AM condition for both hemispheres in comparison to PT stimulation, whereas the P1component showed a shorter latency only in the left hemisphere. Furthermore, the amplitude of the P2 component was significantly suppressed in the AM condition compared to a PT stimulation.Discussion: These facts give an evidence for different neuronal mechanisms of auditory streaming basing on spectral and on temporal cues and on a differential specialization of both hemispheres.Conclusion: We suggest a central auditory level of streaming processing additionally to the peripheral level. This model, described as spectral vs. temporal coding of streaming could be used further for tracing the neuronal alterations in speech and hearing disorders.This project has been supported by DFG - DR 807/1-1 and FORUM - F756-201