Versace, E., Endress, A.D. & Hauser, M.D. 2008. Pattern recognition mediates flexible

timing of vocalization in nonhuman primates: experiments with cottontop

tamarins. Animal Behaviour, 76, 1885-1892.

The authors test the prediction that selection would favor the ability of a sender to recognize patterns within discontinuous noise and adjust the timing of his call appropriately in order to prevent masking. In this study, eight adult cottontop tamarins were isolated and exposed to a series of playbacks that consisted of cyclic alternations of white noise and silence. The moments of silence were either long enough for the tamarin to perform full calls (3s) or too short (1.5s). In trials where the silence was long enough, the white noise was either 4s or 6 s long and the silence was always preceded by a speech syllable ‘ti’. In trials where the silence was too short to call, the white noise was either 5.5s or 7.5 s long and there was no speech syllable preceding the silence. The authors found that for most individuals and at the group level, tamarins called more often in the first part of the long silence than in the first part of the short silence. This suggests, that tamarins are able to use either I) the cue ‘ti’ or II) the duration of white noise to predict when long gaps of silence are about to occur.

To distinguish between these two possible mechanisms, the authors then performed another experiment similar to the first. In trials where the period of silence was long enough to call, this gap was preceded by either the syllable ‘ti’ or ‘lu’. In trials where the gap was too short, the silence was preceded by the syllable ‘ja’. (This time, the durations of preceding white noise were the same in both trials). Again, they found that the tamarins called more often during the first part of long gaps of silence. This suggests that they can use the specific speech syllable as a cue to predict long silences.

This study relates to what we’ve discussed in lecture when we consider why this flexibility in call timing is adaptive. Assumingly, there is some energetic cost of calling since, as we’ve learned, sound is produced due to a change in pressure of the medium caused by a vibration from the sender. Therefore, it would be maladaptive to waste energy calling if the intended receiver cannot detect the call. Furthermore, we learned that sound waves could interfere with each other both positively or negatively. In a noisy environment, the call of the tamarin would likely be degraded due to interference caused by the background noise. Thus, signal optimization predicts that the tamarin should maximize the probability that his call reaches the intended receiver. We learned that this could be done in several ways such as increasing call amplitude (Lombard effect) or altering call frequency so that the signal occupies an open channel. In this case, the white noise in the playback is much louder (25 dB) than an average tamarin call and has a bandwidth that covers most of the tamarin’s possible frequency range. Consequently, these constraints cause the tamarin to optimize his call by altering its timing (instead of amplitude or frequency) in order to take advantage of gaps in interfering sound. In this way, it is easy to see that the theories about the production and propagation of sound allow us to predict both the characteristics of vocal signals and the behaviors of the signal senders.