UC San Diego
The role of pitch in the recognition of vocalizations in songbirds and humans
- Author(s): Bregman, Micah Rendler
- et al.
Recognizing other individuals by sound is a primary function of many vertebrate communication systems. Both speech and birdsong provide an auditory "face," allowing individuals to recognize each other without visual contact. Humans and songbirds have brain mechanisms that have evolved to support these behaviors. In this thesis, I present results from three studies investigating the recognition of complex sounds. Sounds are typically described in terms of four perceptual features: pitch, timbre, loudness and duration. Here we focus on the role of pitch and timbre in auditory recognition. We trained European starlings, a species of songbird, to recognize excerpts of conspecific song and human melodies. Starlings were more flexibly able to recognize song excerpts that had been shifted than shifted melodies, although they were able to learn to recognize both well. This indicates that the use of auditory cues may be stimulus dependent. We then describe a study investigating the interaction of pitch and timbre cues in recognition. Our hypothesis was that adding timbre cues to tone sequences would enable continued recognition even when pitch was altered. However, the results of this study suggest that songbirds are maintaining acoustic memories closely tied to specific spectral features and are unable to recognize tone sequences based on pitch cues alone. In contrast, human listeners performed well when asked to recognize pitch- altered tone sequences. We then investigated the role of individual differences in humans learning to recognize unfamiliar voices. Individuals who learned a language early in life were able to recognize talkers in that language better than those who learned the language later. Extensive music training also plays a role--listeners with more music training learned to recognize voices in an unfamiliar language more quickly than those with less training. Together, these studies contribute a new comparative perspective on the recognition of complex sounds. A description of natural vocalizations consisting of pitch, timbre, loudness and duration is convenient but limiting. The cognitive processes underlying sound recognition are complex, relying on interactions between spectrotemporal features that depend not only on the features of the signal itself, but also on the listener's auditory experience