Detecting modulated signals in modulated noise: (I) behavioural auditory thresholds in a songbird

Most signals from the auditory world have temporal patterns of amplitude modulation that either emanate from the signal source or result from environmental interference (e.g. air turbulence). To investigate mechanisms associated with the segregation and processing of amplitude-modulated signals, we trained European starlings (Sturnus vulgaris) to detect a signal noise band embedded in several flanking noise bands (FBs). We manipulated the envelope correlation between the signal and FBs, the onset synchrony between signal and FBs (0 or100 ms), signal duration (60 or 400 ms) and the spectrum level of the FBs (15 or 50 dB). The lowest signal-detection thresholds were found when the envelopes of the FBs were correlated with each other but different from the signal envelope (the 'co-uncorrelated' condition). Detection thresholds were on average 7 dB higher when both the signal and the FBs had correlated envelopes (the 'all correlated' condition). Thresholds were even higher when the envelopes of all noise bands were independent (the 'all uncorrelated' condition). The difference in detection thresholds between the co-uncorrelated and the all correlated conditions is termed 'comodulation detection difference' (CDD). Differences in signal duration and masker level had significant effects on detection threshold, but not on CDD magnitudes; differences in onset synchrony had no effects. We compare data from starlings with those from previous psychoacoustic studies of humans, and discuss possible mechanisms on which these perceptual effects may rely. Our behavioural data are the reference for a companion study investigating CDD at the neuronal level of the starling [M.A. Bee et al. (2007) Eur. J. Neurosci., 26, 1979-1994].