You talkin’ to me? Interactive playback is a powerful yet underused tool in animal communication research

Abstract

Over the years, playback experiments have helped further our understanding of the wonderful world of animal communication. They have provided fundamental insights into animal behaviour and the function of communicative signals in numerous taxa. As important as these experiments are, however, there is strong evidence to suggest that the information conveyed in a signal may only have value when presented interactively. By their very nature, signalling exchanges are interactive and therefore, an interactive playback design is a powerful tool for examining the function of such exchanges. While researchers working on frog and songbird vocal interactions have long championed interactive playback, it remains surprisingly underused across other taxa. The interactive playback approach is not limited to studies of acoustic signalling, but can be applied to other sensory modalities, including visual, chemical and electrical communication. Here, I discuss interactive playback as a potent yet underused technique in the field of animal behaviour. I present a concise review of studies that have used interactive playback thus far, describe how it can be applied, and discuss its limitations and challenges. My hope is that this review will result in more scientists applying this innovative technique to their own study subjects, as a means of furthering our understanding of the function of signalling interactions in animal communication systems.

1. Introduction

It is well known that in the context of animal communication, signalling facilitates the exchange of information between individuals [1]. One signal produced by an individual has the power to induce a response in a receiver that, in turn, can alter the behaviour of the signalling animal [2]. Thus, communication between individuals is definitively interactive.

Researchers have long explored the function of signals in animal communication systems with the use of playback experiments. Traditionally, however, such experiments have been non-interactive, because the stimuli played back do not change with respect to the varying vocal or physical behaviour of the study subjects [3]. Yet, there is some evidence that the information conveyed in a signal may only have value when presented interactively [4,5]. While it can be considered challenging to conduct, a flexible playback design allows the researcher to manipulate the signals played back as a function of the study subject's currently changing behaviour, thus uncovering signal function within an interactive setting [6]. An interactive playback design therefore incorporates realism into an experimental framework, allowing previously inaccessible questions to be addressed. As such, it offers a more powerful tool for examining animal communication than traditional non-interactive approaches, allowing us to explore the interactive features of communicative strategies that include signal exchanges and the role of graded signalling systems. Indeed, the dynamic nature of aggressive signalling has meant that interactive playback has provided unparalleled insights into the role of signal interactions in male–male competition [2,7]. It is consequently highly surprising that this influential experimental technique remains so underused.

2. A brief history of interactive playback

Interactive playback is not a new technology and has been used by a small subset of researchers since the early 1980s. In 1982, Narins used an interactive playback design to assess the importance of call timing in male neotropical tree frogs (Eleutherodactylus coqui and Hyla ebraccata) [8]. By using a ‘call-triggered playback system’ to vary the temporal patterning between the playback stimulus and the focal subject's vocalizations, he was able to identify call latencies that were salient to each species for avoiding acoustic interference in noisy environments [8]. This early work was followed by a number of notable studies by Schwartz and others, who used interactive playback experiments to investigate male–male acoustic interactions in frogs [7,9–15]. These experiments enabled researchers to show that male anurans avoid overlapping calls to preserve signal integrity in the spring peeper (Pseudacris crucifer), grey treefrog (Hyla versicolor), small-headed treefrog (Hyla microcephala now Dendropsophus microcephalus) [10] and túngara frogs (Physalaemus pustulosus) [13]; that aggressive calling is a graded signalling system in the spring peeper [9]; and that unison bout singing is likely an adaptation to conserve energy in chorusing males in the small-headed treefrog [11]. Interactive playback was pivotal in uncovering the function of call alternation in male anurans [7,14]. For example, one study showed that male túngara frogs varied their call rate in response to the calling strategy of the simulated male, which matched observed patterns of natural calling behaviour, yet this pattern was not observable when traditional playback approaches were used [16].

At a similar time, researchers examining male–male vocal interactions in songbirds were also using interactive playback. In the mid 1980s, Dabelsteen and Pedersen developed the digital sound emitter, a computerized interactive playback system that was used to explore vocal interactions in songbirds [17–19]. These studies supported previous predictions that strangled song in male blackbirds (Turdus merula) represented the highest level of aggressive arousal [17]. A subsequent study by McGregor et al. with great tits (Parus major) showed that the nature of the playback (interactive versus non-interactive) significantly affected the singing response of the focal subject [2], indicating that vocal interaction played an important role in the signalling system of great tits, and likely other songbirds. Subsequently, in order to examine the prediction that matching in songbirds was a reliable signal of a close approach between two individuals in a territorial dispute [20], researchers began to use a more interactive approach in their playback design. One such approach involved using custom-written software to view the focal male's song output in real time, with the experimenter then quickly identifying the song type and choosing an appropriate song, from a selection of pre-prepared stimuli, to respond with [21]. A number of studies with song sparrows (Melospiza melodia) and black-capped chickadees (Poecile atricapilla) used such an interactive playback design to show that song-type matching reliably elicits a stronger approach by subjects that are matched [21–26]. It was only with the use of interactive playback that researchers have been able to show that song-type matching in songbirds is a graded signal of aggressive intent. Researchers continue to use this approach in studies of songbird vocal interactions. For example, one study used interactive playback to nicely demonstrate that personality traits play a role in territorial behaviour and resource defence in male great tits [27].

Studies with songbirds have also highlighted the importance of interactive playback designs in elucidating the functions of eavesdropping by third party receivers. Eavesdropping is defined as the extracting of information from an interaction between individuals that could not be gained from one signal alone [28]. Studies have used interactive playback to engage males in counter-singing interactions with a simulated intruder [29] (as per matching studies described above) or have used it to simulate two males counter-singing with each other [30] and then observed the response of individuals nearby. These studies have very successfully shown that eavesdropping on male song contests directly influences female reproductive decisions in black-capped chickadees [29], acts as song-tutoring events for juvenile song sparrows [30], and allows male great tits to indirectly assess rivals [31]. This work has effectively shown that animals extract information from the interaction between two signallers that they could not obtain from one animal signalling alone.

Although interactive playback has long been used as an experimental approach by researchers studying frog and songbird vocal interactions, surprisingly few studies have applied this approach to other taxonomic groups. It remains particularly underused by those researchers studying mammalian communication, though two studies that have used an interactive playback approach to explore mammalian vocal interactions are both worth mentioning. The first used automated interactive playback software with pre-set time delays to show that antiphonal call timing is behaviourally significant in common marmosets (Callithrix jacchu) [32]. Subjects were more likely to continue with antiphonal exchanges with conspecifics when call latency delays were short. In fact, it has been shown that interactive playbacks are significantly more effective at eliciting antiphonal calling in common marmosets than traditional playback techniques [33]. The second study used experimenter-operated interactive playback software to show that the immediate matching of vocal signals is a friendly means of addressing in bottlenose dolphins (Tursiops truncatus) [34]. It identified a critical time interval during which animals must exchange signals in order to successfully address one another. This study also used an interactive playback design to elicit calling in the focal animal, enabling the experimenter to then choose which stimuli to play back from a selection of computer-synthesized dolphin whistles with the use of computer hotkeys. Both these studies, along with [8], nicely demonstrate how interactive playback can be used to examine which call latencies are important in mediating vocal interactions [32,34].

3. Interactive playback: design and limitations

In order to carry out an interactive playback, it is required that an animal's behaviour is monitored in real time by the experimenter. There are, however, a number of different ways a playback can be considered interactive. I will discuss this in the context of interactive playback experiments with acoustic signals. However, the four categories of interactive playback I present can, in principal, be applied to other sensory modalities.

The first is that the experimenter views the animal's vocalizations in real-time and waits for the animal to produce a specific vocalization or set of vocalizations before beginning the playback. This allows the sound playback to occur in the same context each time. Although at the simpler end of an interactive design, this approach may prove to be particularly important for the collection of behavioural response data on animals that are frequently out of sight, e.g. marine mammals. It ensures a standardized approach within an experimental framework and makes results between playbacks more comparable. The second is that playbacks are used to elicit calling in a focal animal. This can be particularly useful if the subject needs to be engaged in order for a playback experiment to be conducted. Experimental stimuli that are salient to the species can therefore be used to initiate calling in the focal animal [5,34,35]. The third is perhaps considered truly interactive in the sense that the experimenter or an automated routine continually makes decisions on the type and timing of calls to play back based on the animal's currently changing vocal behaviour [2,3,14]. This allows the experimenter to realistically simulate calling conspecifics. This is important because the interaction itself may convey the function, such as type-matching in songbirds as a form of threat escalation [21]. This approach facilitates the rapid interaction between the researcher and the study subject, with the focal animal essentially driving the interaction with a simulated individual. It allows researchers to investigate how both the type and timing of signal production correspond with the vocal and physical behaviour of the study subject [3]. Finally, interactive playback can be used to simulate two or more counter-calling animals, which can then be used to explore the function of eavesdropping [30] or to simulate the dynamics of a natural chorus to observe how an individual may adjust its behaviour based on the changing dynamics of its communication network. For example, one interactive playback design with frogs was used to simulate call interruptions by a number of different males. Using this approach, the study determined that the spatial arrangement of calling males directly influences the temporal patterning of calls as a means of preserving call integrity [12].

Although interactive playback can provide a powerful experimental design, we should acknowledge its potential limitations. Many of these limitations are shared with traditional non-interactive playbacks in that a good understanding is needed of the behavioural or vocal repertoire of the focal species, the context in which these calls or signals are used (i.e. mate attraction, territory defence) and the animal's motivation for communicating [3]. This knowledge will ensure that the interactive playback experiments are well designed and that any likely effects of stimuli on the focal subject are known prior to the playback. As with any experimental design, interactive playback needs careful forethought and planning. It is true that the interactive playback protocol may be thought of as more challenging than those traditionally used in non-interactive playback experiments, as the researcher usually only has a short amount of time in which to select the playback stimulus based on the focal animal's current behaviour [4]. However, given the technological advancements of recent years, if automated computer algorithms are used then response times can be rapid. The challenge with interactive playbacks is in fact to choose the right software or to develop new hardware that is suited to the interactive questions the researcher wishes to address. One final limitation that may be specific to an interactive design is that variation in the stimuli played back to individual subjects is typically much greater in interactive compared with non-interactive playbacks. This is because the type and number of stimuli used are driven by the individual subject's response and this potential disparity in stimuli presentation between individuals can provide some challenges during the analysis stage [6].

4. Technological advances

Technological advancements have meant that interactive playback designs are more accessible than ever. Early studies on anurans used analogue synthesizers: a system triggered by a call from the focal animal, which then presented the playback stimulus after a pre-set time delay. This approach allowed the experimenters to respond rapidly to changes in calling behaviour of focal animals and incorporate realism in their experimental design through the simulation of neighbouring conspecifics [8–10]. Researchers soon began to develop their own types of custom-written software, which were made freely available. These types of software have been written for frogs [16], birds [36–38] and insects [39], among others. Many of these user-friendly software packages allow large numbers of sounds to be stored and rapidly played back in the field. Available technology is such that computer programs can be written to automatically recognize different call types produced by the focal animal and then choose a stimulus to play back based on a pre-programmed calling strategy for the simulated animal [7,16]. Other approaches include the experimenter using a key press to inform the program of the focal subject's behaviour, i.e. aggressive or affiliative signalling, and the computer algorithm then automatically selecting which type of signal to broadcast in response [40]. This has meant that increasing flexibility is afforded to researchers in the experimental design of their interactive studies. Thus, rather than the hardware restricting the playback design, as Schwartz eloquently surmises, playback design is now only ‘limited by the skill and imagination of the programmer’ [7, p. 617].

5. Broader applications

So far, I have largely discussed interactive playback designs in the context of animal acoustic communication. However, although examples are limited, interactive playback can be applied to other signalling modalities in animal communication, such as visual, electrical and chemical. For example, inventive approaches have been used in applying interactive video playback in studies of visual communication in male jacky dragons (Amphibolurus muricatus) [40,41]. These studies either used a computer-controlled library of digitized sequences of a previously recorded male [40] or used a three-dimensional computer-modelled artificial lizard to simulate a digital individual that precisely matched the display behaviour of its live counterparts [41]. These studies allowed researchers to explore how variation in signalling strategy influences an opponent's behaviour in male–male interactions. Analogous to interactive playback studies with acoustic signals, these studies also demonstrated that the ability to interact with a digital video sequence altered the response of the focal subject in comparison with non-interactive video playback. Thus, this research highlights the importance of interaction in an opponent assessment context [40,41]. Interactive video playback and computer-animated stimuli have also been nicely used to explore courtship behaviour and mate choice in swordtail fish (Xiphophorus birchmann) [42]. An exciting avenue of future work would be to combine sensory modalities within an interactive framework: this could be used to investigate the interaction between visual and chemical cues in certain types of social interactions in those species that, for example, rely on both sensory modalities.

6. Conclusion

There are many benefits of traditional non-interactive playbacks, which have significantly contributed to our understanding of animal communication [43]. But conformity starves innovation, and as well as using traditional techniques, we should look towards creative and innovative approaches as a means of furthering our knowledge of interactions in animal communication systems. While there are challenges that come with incorporating interactive playback designs into an experimental study, they are by no means insurmountable. By their very nature, vocal or behavioural exchanges are interactive. As discussed, studies in anurans, songbirds, lizards and non-human primates have all shown that the nature of the playback (interactive versus non-interactive) significantly affects the resulting behaviour of the study subjects [2,16,32,40]. Interactive playback has also shown itself to be incredibly useful for investigating time-specific responses that are salient to the animal during vocal exchanges. Accordingly, interactive playback experiments can offer far greater insights into the signal value of interactive features of communication [6] than standard playback approaches and so uncover the function of signalling strategies that were previously unknown.

A number of researchers have championed interactive playback over the years [4,6,14,19], and I heartily recommend the reader to explore these texts, but this approach still remains widely underused. Given the technological advancements, exciting avenues are now open to animal communication researchers. For example, one could develop an automated network of artificial individuals to simulate some of the real-life complexity of dynamic communication systems or one could develop interactive playbacks that encompass multi-modal communication, with some combination of visual, acoustic, chemical or electrical stimuli. My hope is that, rather than conforming to standard approaches, more scientists will take up this innovative playback design and apply it creatively to their own study subjects.

Competing interests

I report no competing interests.

Funding

I received no funding for this study.