Playback experiments involve the broadcast of natural or synthetic sound stimuli and provide a powerful tool for studying acoustic communication in birds. Playback is a valuable technique for exploring vocal duetting behavior because it allows investigators to test predictions of the various hypotheses for duet function. Here, we adopt a methodological perspective by considering various challenges specific to studying duetting behavior, and highlighting the utility of different playback designs for testing duet function. Single‐speaker playback experiments allow investigators to determine how duetting birds react to different stimuli, but do not simulate duets in a spatially realistic manner. Multi‐speaker playback experiments are superior to single‐speaker designs because duet stimuli are broadcast with spatial realism and unique and additional predictions can be generated for testing duet function. In particular, multi‐speaker playback allows investigators to evaluate how birds respond to male versus female duet contributions separately, based on reactions to the different loudspeakers. Interactive playback allows investigators to ask questions about the time‐ and pattern‐specific singing behavior of birds, and to understand how singing strategies correspond to physical behavior during vocal interactions. Although logistically challenging, interactive playback provides a powerful tool for examining specific elements of duets (such as the degree of coordination) and may permit greater insight into their functions from an operational perspective. Interactive playback designs where the investigator simulates half of a duet may be used to describe and investigate the function of pair‐specific and population‐wide duet codes. Regardless of experimental design, all playback experiments should be based on a sound understanding of the natural duetting behavior of the species of interest, and should aim to produce realistic and carefully controlled duet simulations. Future studies that couple playback techniques with other experimental procedures, such as Acoustic Location System recordings for monitoring the position of birds in dense vegetation or multimodal techniques that combine acoustic with visual stimuli, are expected to provide an even better understanding of these highly complex vocal displays.

No abstract