Analysis of spectral shape in the barn owl auditory system.

In a behavioral experiment, we investigated how efficiently barn owls (Tyto alba) could detect changes in the spectral profile of multi-component auditory signals with stochastic envelope patterns. Signals consisted of one or five bands of noise (bandwidth 4, 16, or 64 Hz each; center frequencies 1.02, 1.43, 2.0, 2.8, 3.92 kHz). We determined increment thresholds for the 2 kHz component for three conditions: single-band condition (only the 2 kHz component), all five noise bands with the envelope fluctuations of the bands being either correlated or uncorrelated. Noise bandwidth had no significant effect on increment detection. Increment thresholds for the different conditions, however, differed significantly. Thresholds in correlated conditions were generally the lowest of all conditions, whereas, thresholds in uncorrelated conditions mostly resulted in the highest thresholds. This can be interpreted as evidence for comodulation masking release in barn owls. If the increment in the 2 kHz component is balanced by decrementing the four flanking bands in amplitude, increment detection thresholds are not affected. The data suggest that the barn owls used information from simultaneous spectral comparison across different frequency channels to detect spectral changes in multi-component noise signals rather than sequential comparison of overall stimulus levels.