Stream segregation based on interaural time differences: differential effects on transient and sustained components of the fMRI BOLD response in human auditory cortex

Auditory stream segregation is an elementary process of auditory scene analysis. Most studies of stream segregation used spectral or pitch differences to induce stream segregation, while only few studies evaluated spatial cues. Functional magnetic resonance imaging (fMRI) and magnetoencephalography studies suggest enhanced activity in auditory cortex (AC) during streaming, as a consequence of a decreased perceived tone-repetition rate. Here we investigate streaming based on interaural time differences (ITD), and in particular the effect of ITD on the blood oxygen level dependend (BOLD) response measured by fMRI in human AC. Eight listeners were presented with continuous sequences of a repetitive ABBB pattern composed of 125-ms long harmonic complex tones with a fundamental frequency of 180Hz (lowpass filtered at 5kHz, 70 dB SPL), where A and B tones were lateralized by different ITDs. A given ITD was kept constant for 32s, separated by a 34s baseline with identical ITDs for A and B tones. The A tones had a constant ITD of 687.5µs, i.e. the tones were lateralized to the left. Relative to the A tones, the B tones were lateralized to the right by 5 ITD differences: betaITD=62.5, 125, 187.5, 343.75, or 687.5µs. The betaITD=62.5µs condition was included as a control that did not generally elicit streaming. Psychophysical responses, obtained during the measurements, confirmed a low probability for perceiving streaming for the control condition (about 20%), and a high probability for all other conditions, which further increased with larger betaITD (>70% for betaITD=125 and 187.5µs; >95% for betaITD=343.75 and 687.5µs). An fMRI contrast analysis of all 5 conditions against the baseline showed distributed activation across AC, involving Heschl's gyrus, the superior temporal gyrus, and the planum temporale, with the majority of the activation in secondary areas. The reconstructed time course of the activity showed a sustained activation for the whole 32s blocks for all 4 streaming conditions, which did not increase in amplitude with larger betaITD. In contrast, transient onset responses at the transition from baseline to target ITD were much more prominent for the betaITD=343.75 and 687.5µs conditions. These results support the notion that the sustained activity reflects changes in perceived rate, while the onset peak of the BOLD response appears to be related to the salience of the corresponding perceptual change.

Supported by DFG grant GU593/3–1