Significance of frequency- and amplitude-based temporal cues in Japanese macaque coo vocalizations.

Abstract

The smooth early (SEH) and smooth late (SLH) high coo calls produced by Japanese macaques differ in the temporal location of a frequency inflection from rising to falling and are generally used in different social situations. Coo calls also contain temporal variation in relative harmonic level, which may or may not have communicative significance. Macaque monkeys that had not received explicit training to attend to specific features were trained to discriminate among SEH and SLH coos. We then used multidimensional scaling (MDS) to analyze discrimination latencies and determine the perceptual similarity of these coos. Macaque subjects perceived stimuli as similar based on patterns of variation in the relative levels of the harmonics. When relative level manipulations were conducted on synthetic coo replicas, the MDS perceptual similarity maps were reorganized. Because the temporal position of the frequency inflection has been considered the primary communicative cue within Japanese macaque coo calls, we used MDS to closely examine the salience of this acoustic feature. We conducted temporal reversals of the frequency contours of several coo calls. Original and frequency-reversed calls could not be discriminated by Japanese macaques if the frequency peak was near the temporal midpoint of the coo, but could be discriminated when the peak was near a call endpoint. However, the perceptual similarity of reversed calls was high unless the call contained substantial frequency modulation. Reversals of amplitude contours were also conducted. Although macaques are quite sensitive to amplitude increments, reversal of the relatively flat amplitude-contours of these calls did not affect discrimination responses. In a final experiment, we added amplitude increments to the standard level of coo call harmonics and determined threshold sensitivity using psychophysical methods. Variation in relative harmonic level was detectable, although thresholds were elevated when compared to those obtained for vowel-like stimuli. This elevation was probably a function of natural amplitude modulation in the coo stimuli. Taken together, these data suggest relative harmonic level variation is present in coo calls, and can be detected. The functional significance of such variation has not been determined, however.