Current Issues – Perspectives and Reviews
Functionally Referential Communication in Mammals: The Past, Present and the Future
Article first published online: 24 SEP 2012
© 2012 Blackwell Verlag GmbH
Volume 119, Issue 1, pages 1–11, January 2013
How to Cite
Townsend, S. W., Manser, M. B. (2013), Functionally Referential Communication in Mammals: The Past, Present and the Future. Ethology, 119: 1–11. doi: 10.1111/eth.12015
- Issue published online: 10 DEC 2012
- Article first published online: 24 SEP 2012
- Manuscript Accepted: 17 AUG 2012
- Manuscript Revised: 6 JUN 2012
- Manuscript Received: 31 MAY 2012
- Top of page
- Functional Reference – the Framework
- Problems Classifying Calls as Functionally Referential
- Literature Cited
Whether animal vocalizations have the potential to communicate information regarding ongoing external events or objects has received considerable attention over the last four decades. Such ‘functionally referential signals’ (Macedonia & Evans 1993) have been shown to occur in a range of mammals and bird species and as a consequence have helped us understand the complexities that underlie animal communication, particularly how animals process and perceive their socio-ecological worlds. Here, we review the existing evidence for functionally referential signals in mammals according to the framework put forward in the seminal Macedonia and Evans review paper. Furthermore, we elucidate the ambiguities regarding the functionally referential framework that have become obvious over the last years. Finally, we highlight new potential areas for investigation within referential signalling. We conclude the functionally referential framework is still informative when interpreting the meaning of animal vocalizations but, based on emerging research, requires further integration with other approaches investigating animal vocal complexity to broaden its applicability.
Functional Reference – the Framework
- Top of page
- Functional Reference – the Framework
- Problems Classifying Calls as Functionally Referential
- Literature Cited
…I took a stuffed and coiled-up snake into the monkey-house at the Zoological Gardens, and the excitement thus caused was one of the most curious spectacles which I ever beheld. Three species of Cercopithecus were the most alarmed; they dashed about their cages, and uttered sharp signal cries of danger, which were understood by the other monkeys. (Darwin 1871)
Although unbeknown to Darwin at the time, this presentation of a snake to a group of captive monkeys represented perhaps the first experimental investigation into one of evolutionary biology's most fascinating, yet inherently challenging, conundrums: what do the vocalizations of animals ‘mean’? Even in human language, the quest to understand the relationship between meaning and signs is an inherently arduous task that has intrigued linguists and philosophers for centuries (see Hurford 2008). Hence, attempting to address similar problems in the communication systems of animals, and unpack the meaning of their vocalizations, represents a non-trivial feat (Smith 1977).
Since Darwin's initial observations, Tom Struhsaker and later Robert Seyfarth and Dorothy Cheney revisited this question under more natural settings and with more rigorous empirical approaches. By recording the vocalizations of wild vervet monkeys (Chlorocebus pygerythrus), three distinct alarm call variants produced systematically in accordance with the type of predator encountered were identified (Struhsaker 1967; Seyfarth et al. 1980a). In response to martial eagles (Polemaetus bellicosus), vervets would produce a ‘cough-like’ vocalization, when spotting a leopard (Panthera pardus) a harsh ‘bark’ would be emitted and when coming across a snake a ‘chutter’ would result (Struhsaker 1967; Seyfarth et al. 1980a). Upon hearing these calls, conspecific monkeys within the same group were observed to respond in a similar way as if they had detected the predator themselves, looking up into the sky or running for cover when hearing ‘eagle’ alarms, running into trees following ‘leopard’ alarm calls and finally adopting a raised posture and scanning the surrounding ground area after hearing ‘snake’ alarms. These qualitatively different behavioural responses and their similarity to the natural anti-predator behaviour prompted the question, what exactly the calls referred to. To answer this, Cheney, Seyfarth and Marler implemented playback experiments whereby the alarm calls were played to monkeys through hidden speakers, but in the absence of the eliciting predator. As predicted, responses of listening monkeys to played-back alarm calls mimicked those elicited when encountering the predator directly, leading the researchers to postulate that vervet monkey alarm calls are intimately linked, and refer to, the eliciting external event (Seyfarth et al. 1980b).
In human language, this linguistic capacity of reference is generally known as semantics (Bickerton 1990). However, because the exact cognitive mechanisms underlying signal production are still unclear, calls that allow animals to accurately predict environmental events are termed ‘functionally referential’, as the call functions to provide potential information to the receiver with regard to an external event or object (Marler et al. 1992; Macedonia & Evans 1993; Evans 1997; Zuberbühler 2000). In parallel with Cheney, Seyfarth and Marler's pioneering experimental work demonstrating striking similarities between vervet vocalizations and the external reference of human words, Chris Evans and his colleagues were investigating similar meaning-based questions in the vocalizations of chickens (Gallus gallus). Such findings demonstrating the acoustic signalling and perception of external events in birds, even rivalling those of primates, prompted Chris Evans in collaboration with Joseph Macedonia to outline a theoretical framework guiding the classification of calls as functionally referential (Macedonia & Evans 1993; Evans & Marler 1994; Evans 1997). Such a progressive conceptual approach undoubtedly contributed to establishing the fields of animal vocal communication and language evolution as we know them today.
The framework proposed by Macedonia and Evans dictates that only when specific production and perception-based criteria are fulfilled can a call be accepted as functionally referential. Firstly, there must be a tight association between the call type and its eliciting context such that it could be used as a proxy of an object or event in the external world. Secondly, the call should cause the same behaviour in receivers as would occur if naturally encountering the stimulus, but in the absence of the putative referent (Evans et al. 1993; Macedonia & Evans 1993). With these criteria for functional reference logically conceptualized, the stage was set for comparative investigations into similar communicative complexity in primate, non-primate mammals and birds. Focusing here on mammals, the last four decades has demonstrated that an array of species do produce and use functionally referential vocalizations (Manser 2009; Clay et al. 2012), although the contexts in which they occur seem to be relatively limited.
What is the Referent in ‘Referential’?
Functionally referential alarm calls
In general, functionally referential calls in mammals fall within three main categories: predatory contexts, discovery of food and social interactions (see Table 1). To date, most evidence for functional reference has accumulated within the predatory context with a range of species producing different alarm calls in the presence of aerial and terrestrial predators and receivers responding to playback of such calls as if they had encountered the predator themselves. For example, within the primates, aside from vervet monkeys, functionally referential alarm calls have also been documented in Diana monkeys (Cercopithecus diana) (Zuberbühler et al. 1997), Campbell's monkeys (Cercopithecus campbelli) (Zuberbühler 2002) and ring-tailed lemurs (Lemur catta) (Pereira & Macedonia 1991). More recent findings have indicated that not only the acoustic structure alone, but also the sequence of primate alarm calls can encode referential information with regard to predator types. Black-and-white colobus monkeys (Colobus guereza), for example, produce two acoustically distinct alarm call variants, snorts and roar phrases, when exposed to aerial and terrestrial predators, but neither is given exclusively to one of the predator classes. Instead, a stronger relationship exists between predator class and the number of ‘roar phrases’ present in the calling sequence, with terrestrial predators eliciting snorts followed by a few phrases, whilst aerial predators are responded to with sequences consisting purely of roar phrases (Schel et al. 2009, 2010). These findings suggest considerable flexibility in the referential system, as not only the acoustic structure, but also the composition of discrete calls can potentially provide reliable information regarding the presence of an external event or object.
|External context species||Call type or sequence||Source|
|Predation||Vervet monkeys (Chlorocebus pygerythrus)||Call type||Struhsaker (1967), Seyfarth et al. (1980a,b)|
|Diana monkeys (Cercopithecus diana)||Call type||Zuberbühler et al. (1997)|
|Campbell's monkeys (Cercopithecus campbelli)||Call type||Zuberbühler (2002)|
|Ring tailed lemurs (Lemur catta)||Call type||Pereira & Macedonia (1991)|
|Black and white colobus monkeys (Colobus guereza)||Call Sequence||Schel et al. (2009, 2010)|
|Gunnison's prairie dogs (Cynomys gunnisoni)||Call type||Kiriazis & Slobodchikoff (2006)|
|Meerkats (Suricata suricatta)||Call type||Manser (2001), Manser et al. (2001)|
|Food||Marmosets (Callithrix geoffroyi)||Call type||Kitzmann & Caine (2009)|
|Chimpanzees (Pan troglodytes)||Call type||Slocombe & Zuberbühler (2005, 2006)|
|Bonobos (Pan paniscus)||Call Sequence||Clay & Zuberbühler (2009, 2011)|
|Social||Rhesus macaques (Macaca mulatta)||Call type||Gouzoules et al. (1984)|
|Baboons (Papio cynocephalus ursinus)||Call type||Owren et al. (1997), Rendall et al. (1999)|
|Dog (Canis familiaris)||Call type||Faragó et al. (2010)|
Outside of the primates, Gunnison's prairie dogs (Cynomys gunnisoni) and meerkats (Suricata suricatta) have been shown to produce functionally referential alarm calls. Prairie dogs have four alarm call variants that vary consistently with different predator types, including hawks, humans (Homo sapiens), coyotes (Canis latrans) and dogs (Canis familiaris). Playback of these alarm calls in the absence of the actual predators elicited similar escape responses to those displayed when naturally encountering the predator types (Kiriazis & Slobodchikoff 2006). More recent observational data have additionally indicated that the referential specificity of prairie dog alarm calls may be considerably greater than previously thought, with alarm calls also varying based on specific physical attributes of the predator (Slobodchikoff et al. 2009). Unfortunately, the relevance of these results remains difficult to interpret given the absence of playback experiments to test receiver perception of this information.
Cooperatively breeding meerkats living in the Kalahari Desert also provide an intriguing example of functionally referential communication, as, despite their non-primate mammal status, their alarm call system probably represents a more sophisticated example than so far described for any primate species (Manser et al. 2002). Meerkats are exposed to a range of aerial and terrestrial predatory threats and various snakes (Manser 2001). Due, in part, to the openness of their semi-desert habitat and their foraging technique, which directly compromise their visual system (Manser 1999; Townsend et al. 2011), meerkats are highly vulnerable to predation (Clutton-Brock et al. 1998). Given that it is not possible to scan for predators and forage efficiently at the same time (Manser 1999), and their need to coordinate escape responses as a group (Furrer & Manser 2009), meerkats have evolved a referential alarm call system that not only indicates predator type, and hence which escape response to employ, but also the relative urgency of the imminent attack (Manser 2001). Traditionally, alarm calls were considered to lie along a continuum from purely emotional (only the internal state of the caller is attributed to the call, such as the urgency experienced) to referential (external events or objects are attributed to the call) (Marler et al. 1992; Blumstein & Armitage 1997; Blumstein 1999). However, detailed acoustic analysis of meerkat alarm calls indicate that both ‘ends’ of the motivational–referential continuum can be encoded in unison and should not necessarily be seen as juxtaposed alternatives (Manser et al. 2002; Manser 2009; Fischer 2011). Moreover, playback experiments have demonstrated that this multidimensional acoustic variation described is used by receivers, inducing qualitatively different escape responses to aerial and terrestrial calls and more extreme responses to calls encoding higher urgency (Manser et al. 2001; Manser 2009). This dual encoding of referential and emotional information allows meerkats to make informed decisions regarding not only how to respond to approaching predators, but critically, also when to respond and ultimately avoids compromising an individual's foraging success (Manser 2001; Manser et al. 2001; Amsler 2009).
Functionally referential food calls
In a similar way to detecting a nearby predator, the discovery of food has been shown to elicit acoustically distinct vocalizations in a number of species. Whilst this may point towards a similar communicative phenomenon as functionally referential alarm calls, the picture is complicated by two issues. Firstly, in some of the instances where ‘food calls’ are produced, the same vocalization is also produced in non-food contexts, hence violating the ‘context specificity’ assumption within the empirical framework used to deduce referentiality (see Clay et al. 2012). Toque macaques (Macaca sinica), for example, produce specific ‘hum’ vocalizations when encountering a large quantity of food in their natural environment, but the same call is also elicited when the first rains arrive after a long dry period (Dittus 1984; Clay et al. 2012). Secondly, for some species where strong context specificity exists between calls, food presence, food type or food quality, such as in golden lion tamarins (Leontopithecus rosalia, Benz et al. 1992) and cotton-top tamarins (Saguinus oedipus, Elowson et al. 1991), playback experiments demonstrating receiver perception are currently lacking (Clay et al. 2012). Exactly how these examples therefore contribute to understanding food calls as referential signals remains unclear.
As for species whose food calls do fulfil the functionally referential criteria, it seems there exists only a handful (see Table 1). Marmosets (Callithrix geoffroyi), chimpanzees (Pan troglodytes) and bonobos (Pan paniscus) have been shown to produce either acoustically distinct vocalizations or sequences of vocalizations in the presence of food. In these instances, such vocal behaviour is intimately tied to the eliciting stimulus whether it be the discovery of food, as in marmosets (Kitzmann & Caine 2009), or the food quality and/or type, such as in chimpanzees (Slocombe & Zuberbühler 2005, 2006) and bonobos (Clay & Zuberbühler 2009). Playback experiments of these food calls, in comparison with control calls, have demonstrated that receivers use this information and modify their behaviour accordingly. For example, in marmosets, playback of food-associated calls reliably elicited higher rates of ‘food-related behaviours', than did either of the control playbacks, including both social cohesion ‘twitter’ calls and ambient noise (Kitzmann & Caine 2009). Similarly in chimpanzees, playback of food grunts elicited during discovery of high quality, or preferred food, resulted in more searching behaviour for the food type associated with the calls by receivers, than when exposed to grunts produced in the presence of low-quality, less-desirable food (Slocombe & Zuberbühler 2005, 2006). In bonobos, however, a stronger relationship has been shown to exist between the arrangement of distinct calls in sequences and food quality, rather than the actual call structure (Clay & Zuberbühler 2009). Playback of food sequences produced when finding preferred food induced significantly more searching effort in receivers, than when hearing sequences of calls produced in the presence of less-desired food (Clay & Zuberbühler 2011). These results, demonstrating context specificity at the sequence, rather than the call level, mirror the referential alarm call sequences in black-and-white colobus monkeys (Schel et al. 2009) and emphasize that compositionality may play a more important role in referential signalling than previously thought.
In sum, it appears that food-related calls in mammals do qualify as referential signals (Clay et al. 2012), but only in a very limited number of species and solely within the primates. In non-primate mammals, evidence for context-specific food calling appears to be absent. Greater spear-nosed bats (Phyllostomus hastatus) and bottlenose dolphins (Tursiops truncatus) have both been observed to produce discrete call types when uncovering food; however similarly to some primate species, the calls are not tightly linked to the ‘food-specific’ context, but are also produced in other more general behavioural contexts, such as social recruitment (Wilkinson & Boughman 1998; Janik 2000). Why food calling is absent in non-primate mammals is currently not clear. One possibility is that non-primate mammals often forage as a single cohesive group encountering patchy or distributed food items that cannot easily be divisible amongst multiple group members. Hence, selection pressure to announce the presence of a common shareable food source may not be sufficient to drive the evolution of food-associated calls. Future comparative work assessing socio-ecological and sensory-physiological similarities and differences between primates and bird species also possessing functionally referential food calls may help to shed light on this matter.
Functionally referential social vocalizations
Whether vocalizations also produced in social contexts can function in a truly referential manner has been subject to some contention, given that it is often not clear whether social vocalizations reference the affective state of the signaller or the conspecific as the external stimulus eliciting the call (Smith 1981; Fedurek & Slocombe 2011). Nevertheless, there is some evidence suggesting certain social calls can qualify as referential signals. Rhesus macaques (Macaca mulatta) have been shown to reliably produce five acoustically variable scream vocalizations depending not only on the intensity of aggression received, but also on the rank of the aggressing monkey (Gouzoules et al. 1984). Playback experiments of offspring screams, varying in their acoustic structure, to mothers, resulted in differential behavioural responses. Mothers responded most strongly to screams indicating their offspring was being attacked severely or by lower-ranking individuals than those produced when offspring were just chased or attacked by higher-ranked individuals. A heightened response to more severe aggression is predictable when considering that such interactions can often end in injury. Why screams given when being attacked by low-ranking individuals should also elicit a stronger response, however, is less obvious unless the rhesus macaque social system is considered. Given the matrilineal nature of rhesus macaque societies, receiving aggression from lower-ranking conspecifics can ultimately lead to dominance usurping, which has serious impacts on future reproductive output as a benefit to the higher-ranked and a loss to the lower-ranked individuals (Gouzoules et al. 1984). Being able to keep track of aggression severity and particularly from whom it is received therefore allows females to make informed decisions about whether to interfere and mobilize an attack (Gouzoules et al. 1984). This context specificity, in combination with the variable response to playback of scream types, prompted Gouzoules et al. (1984) to posit that the screams function in a referential way, transferring external information specifically with regard to opponent rank.
The harmonically rich grunts of baboons (Papio cynocephalus ursinus) have also been highlighted as candidate vocalizations that have the potential to refer to external social events. Observational data and acoustic analyses have demonstrated that structurally different grunts are produced in qualitatively different behavioural contexts, for example, when approaching a mother and infant dyad (‘infant grunt’) or when travel is initiated across open areas (‘move grunts’) (Owren et al. 1997). Playback experiments have shown that the two grunt variants elicit functionally distinct behavioural responses similar to how baboons respond to these calls under natural conditions, although perception also seems to be mediated by the current context (Rendall et al. 1999). Given that at least one of the behavioural contexts eliciting the call can be considered as external to the signaller, such as the onset of group movement, and resulted in appropriate, reliable receiver responses during playbacks, baboon travel grunts have been purported to have a rudimentary referential function (Rendall et al. 1999).
Outside of the primates, recent research has shown that dogs reliably produce different growl variants depending on the external social event, particularly during threatening situations [when approached by an unknown human individual (‘stranger growl’), or if guarding a food item (‘food guard growl’)] and in positive social situations, such as playing with a human (‘play growl’) (Faragó et al. 2010). With playbacks, Faragó et al. (2010) also showed that the variation in growls is meaningful to receivers. When presented with a food item, subjects spent less time in contact with the food after being played back food guarding growl variants simulating the food belonged to another dog, than when played back either the play growls or threatening stranger growls. Because dogs were as likely to approach the food source when played threatening stranger growls in comparison with play growls, the authors argue it can be ruled out that subjects were just responding purely to the emotional information in the call and instead that the calls provide specific information to receivers regarding the ongoing social context (Faragó et al. 2010).
From reference to representations
A central issue concerning the semantic properties of referential signals is whether they have their effects by evoking representations of the eliciting event in the mind of the receiver. This is also important as such a language-like attribute of a natural animal signal can help shed light on the evolutionary route via which language emerged (Evans 2002). Although animals do respond ‘as if’ they understand the semantic content of the vocalization, one common criticism endured by proponents of a vocal route to language evolution is it is virtually impossible to empirically prove this (Rendall et al. 2009). A more parsimonious explanation for the receiver's behaviour is simply that animals are responding to the underlying acoustic structure of the vocalizations, which infiltrate evolutionarily primitive subcortical brain structures and induce involuntary, reflexive behavioural responses (Smith 1991; Owren & Rendall 2001; Rendall et al. 2009). Whilst it is extremely challenging to identify the true nature of a monkey's understanding when it hears an aerial alarm call, there are observations and field experiments which directly contradict such affect and conditioning explanations and even point towards a representational-like perception of external events or objects (see also Evans & Evans 2007). Firstly, from a structural point of view, although an intimate relationship exists between acoustic features and function – noisy and harsh calls, for example, commonly elicit avoidance responses (Morton 1977) – there are numerous exceptions to the rule (Fischer 2011). Secondly, to assess if animals respond to the acoustic structure alone requires finding vocalizations that match in their broad external ‘meaning’, but differ in their acoustic properties. If animals rely on physical call characteristics, animals should always respond to calls irrespective of the call's information content. Alternatively, if animals respond to the actual external information in the call, response intensity should be mediated by similarity in meaning (Cheney & Seyfarth 1990).
Like vervets, the alarm calls of Diana monkeys function referentially (Zuberbühler et al. 1997), but the specific cognitive processes underlying alarm call perception remained elusive. To address this issue, Zuberbühler et al. (1999) designed a prime–probe-based experiment to be able to differentiate between ‘perceptual semanticity’ and ‘conceptual semanticity’. In perceptual semanticity, the acoustic features of the call induces the behavioural response, whilst in ‘conceptual semanticity’, some form of representation or intervening mental concept drives subsequent behaviour (Zuberbühler et al. 1999), as is the case in human language (Yates & Tule 1979; Gyger et al. 1987). The experiment consisted of various discrete playbacks, each time involving a prime (predator call, such as an eagle shriek or leopard growl or a Diana monkey alarm call) and a probe stimulus (a predator call). In the baseline condition, eagle shrieks (prime) were followed again by eagle shrieks (probe). In this instance, both acoustic and semantic features are congruent with one another, and hence, Diana monkeys showed a reduced response to the probe in comparison with the initial prime call. In the test condition, male Diana monkey alarm calls for an eagle or leopard were played back and, then again the corresponding predator vocalization (shriek or growl) acted as the probe. The logic here was, if Diana monkeys purely use the acoustic-perceptual features of vocalizations to classify calls, there should be a strong response also to the probe given that eagle alarm calls and eagle shrieks are acoustically different. However, subjects showed a similar response as in the baseline condition, with a weaker response to the probe predator vocalization in comparison with the predator alarm call prime. All necessary combinations and controls verified that Diana monkeys were not responding to the acoustic features alone, but also engage a common associate to help make informed decisions regarding the best behavioural response. Whether common associates are engaged through the generation of a mental representation or through associative learning processes is still to be determined.
It therefore seems that the suggested gulf separating functionally referential calls and words in human language may not be as wide as often assumed in as far as the processing of vocalizations is not purely affect-based, or perceptual (Rendall et al. 2009), but also potentially conceptual. Although ‘encouching’ animal vocalizations within a linguistic framework has received recent criticism (Rendall et al. 2009), without taking such a comparative approach, it would be impossible to understand where the differences between animal vocal systems and human language lie (Fischer 2011) and ultimately what physiological, social and ecological factors are responsible for this.
Problems Classifying Calls as Functionally Referential
- Top of page
- Functional Reference – the Framework
- Problems Classifying Calls as Functionally Referential
- Literature Cited
How do Animals Partition Their Worlds?
Analysing calls within the functionally referential framework was designed to identify the selective conditions driving the evolution of communication related to external stimuli and events, in comparison with calls elicited by behavioural contexts animals experience. Such an approach was vital in also identifying what type of information receivers might be exposed to and providing some insight into the cognitive processes leading to the specific response. However, as with any theoretical ‘construct’ used to interpret empirical data, several obvious problems in classifying calls according to this framework have been identified over the last years. One issue arising from both the production and perception perspective is that we simply do not know enough regarding how animals categorize their environment. For example, we are often limited to identify the function of all call types of a species resulting in difficulties when explaining the variation underlying call production. Furthermore, ambiguous behavioural responses to playbacks complicate our understanding of perception specificity.
Although certain vocalizations seem to fit the criteria necessary for functioning referentially, other calls that seem to be elicited by external stimuli are produced in more than just one context. For example, in the redfronted lemur (Eulemur fulvus rufus) and white sifakas (Propitheus verreauxi verreauxi), a specific call type is given to aerial predators, but the alarm call given to terrestrial predators is also given in other contexts, suggesting that production specificity for the different call types can vary (Fichtel & Kappeler 2002). However, it may just be that the terrestrial call is elicited by a category of stimulus that occurs in several contexts and from the animals' perspective is as specific as the aerial call. Along the same lines, Rhesus macaques produce specific vocalizations when food of differing quality is encountered (Hauser & Marler 1993). A habituation–dishabituation experiment demonstrated that individuals habituated across calls with similar food-quality referents, and neuro-imaging of the ventral prefrontal cortex indicates similar levels of neuronal activity (firing rate) to the acoustically distinct calls, suggesting receivers classify the calls on the basis of their concept rather than their perceptual features (Hauser 1998; Gifford et al. 2005). Together, this work suggests rhesus macaque food calls fit perfectly into the referential framework, until one considers that the same calls are also produced in contexts unrelated to food, such as group movement (Hauser & Marler 1993).
Another aspect complicating the categorization of calls as functionally referential is that detailed acoustic and sequence analyses have demonstrated not only the encoding of multiple external and internal information sets within the same call type (Manser et al. 2002), but also the use of multiple call types to encode a single external event (Clay & Zuberbühler 2009). For example, the acoustic structure of meerkat alarm calls differ depending on the predator type (raptor flying in the air, and terrestrial predator on the ground) and the distance of the predator (Manser 2001), and receivers seem to adjust their response accordingly to this variation (Manser et al. 2001). Similar abilities have previously been described in dwarf mongoose (Helogale parvula) (Beynon & Rasa 1989), prairie dogs (Slobodchikoff et al. 2009) and more recently in titi monkeys (Callicebus nigrifrons) (Caesar et al. 2012). However, for these studies, the playback experiments required to confirm the biological relevance to the receiver are missing. Emerging research also suggests that, in various food or alarm realms, external events are not referred to via single discrete calls but instead through the combination of different calls (Schel et al. 2009). By restricting functionally referential calls to only those that demonstrate production specificity of a discrete call to a very specific context, we are likely failing to describe and understand the potential complexity that underlies animal vocalizations and their wealth of meaning to receivers.
Uncertainty regarding the referential value of calls has also emerged because playbacks of certain calls can elicit subdued and subtle reactions in receivers (Donaldson et al. 2007), implying a lack of perception specificity. This is seen most obviously in socially referential calls. For example, when rhesus macaques' perception of the information in screams was tested, mothers responded by simply looking more in the direction of the speaker (Gouzoules et al. 1984). This differs considerably from when vervet monkey or meerkats are exposed to an alarm call and exhibit the same escape behaviour, as when they would detect the predator themselves. One explanation is that, from a structural point of view, alarm call variants are quantitatively different and are produced in qualitatively different predation contexts. Screams on the other hand show more graded acoustic variation, which corresponds with the subtly different aggressive contexts eliciting their production. Receivers may therefore perceive this variation as representing ‘shades of meaning’ (Cheney & Seyfarth 1990), and hence, differences in response intensity may also fall along a graded continuum.
Another potential, though often overlooked, explanation could be that the presence of a valuable food or a potential predator are, in general, more evolutionary urgent events where the benefits of responding far outweigh the costs. Agonistic social events on the other hand, though sometimes beneficial to respond to, also represent a certain additional degree of danger, given that aggression has the potential to escalate and can often be re-directed (Wittig & Boesch 2003). Clear behavioural responses therefore may only occur in more extreme situations when the benefit accrued sufficiently outweighs the cost of intervening. Currently, it seems that opinion is still divided regarding functionally referential social calls (Donaldson et al. 2007; Fedurek & Slocombe 2011). Future experimental work potentially outside of agonistic contexts, where social calls co-vary with external entities or occurrences and costs and benefits can be more easily quantified, may help to shed further light on this contentious issue.
Where do functionally referential communication studies head now?
Given the ambiguities that have emerged when attempting to apply the functionally referential framework to animal vocalizations, it may be time to reflect and ask whether sticking rigidly to this theoretical construct may be creating more problems for the field of animal communication than we are solving? A growing school of thought proposes we focus our attentions away from evolutionary urgent calls which are tightly linked to an eliciting stimulus and hence potentially have a very fixed hypothesis space (Fitch 2010; Meise et al. 2011) and concentrate more on those calls produced across multiple contexts (Fischer 2011). Such contextually variable call types are particularly interesting because they can potentially tell us more about the flexibility of information processing in animals (Fischer 2011). Whilst this may be the case, such a qualitative switch in unpacking animal vocalizations would suggest that, firstly, we have exhaustively covered the occurrence of animal's abilities to refer to external events and, secondly, that identifying calls as referential in the traditional sense brings nothing additional to the field. Neither of which, we argue, is correct.
Given the evolutionary importance of efficiently signalling predation events, it is no surprise that alarm calls represent the clearest examples most faithful to the referential criteria. However, there still exist additional avenues, particularly within the social world of animals, worthy of investigation if we are to better understand the pervasiveness of referential signals. One particularly fruitful area, we believe, may be within the socio-reproductive realm of mammal daily life. Signalling ongoing reproductive events and keeping track of changes in the reproductive behaviour of social group members would clearly be beneficial in guiding an individual's future mating decisions, and indeed, it has been demonstrated that calls of certain primate species in reproductive contexts do encode external information sets. Yellow baboons (Papio cynocephalus), for example, frequently emit loud vocalizations when copulating (Semple 2001), and subsequent acoustic analysis has attributed not only variation in calls to the identity of the caller, but also the rank of the male mating partner (Semple et al. 2002). Unfortunately, playback experiments testing the perception of this information have yet to be conducted, and hence, how ‘meaningful’ this variation is to receivers remains unclear (Semple et al. 2002). Nevertheless, given the immediate fitness consequences associated with advertising and monitoring available mating opportunities, one might predict to see similar specificity, in both the context eliciting the call and the distinctness of the categorical behavioural responses as is seen with mammal alarm calls.
As to whether such a framework can actually help us progress in the field of animal vocal complexity, a growing interest in animal meaningful syntax would argue in favour of this (Arnold & Zuberbühler 2006; Ouattara et al. 2009; Hurford 2011). To reliably interpret the potential information content of combinatorial or indeed compositional sequences of vocalizations, it is critical to first elucidate the meanings of the independent call types (Fedurek & Slocombe 2011). As it stands, the functionally referential framework proposed by Macedonia and Evans lends itself perfectly to this emerging field providing the necessary tools to do this. Moreover, without implementing this framework in the first place, the fact that there exists considerable variation in the context specificity of vocalizations may not have been stumbled upon. We therefore argue that functionally referential calls still represent a crucial starting point from where we can compare and interpret the variation in information content and the mechanisms underlying this, across the entire spectrum of animal vocalizations.
Instead of abandoning the referential scaffold completely, one possible compromise would be to revisit the specific criteria laid out by Macedonia and Evans and, based on recent findings within the field of animal communication, amend some of the underlying assumptions. In particular, the requirement for context specificity between signal and event should be given more consideration so as not to exclude calls produced in multiple contexts from functioning referentially (Fischer 2011). Furthermore, given the common aim shared by most studies of animal vocal communication is to better understand how such information is perceived and processed, we also encourage the employment of additional playback methodologies to actively disentangle perceptual from conceptual processing mechanisms (Zuberbühler et al. 1999). Whether it be through prime–probe, habituation–dishabituation or violation of expectation (Proops et al. 2009), if applied in a truly comparative way, this cognitive dimension to the functionally referential framework could begin to unravel the similarities and differences between how animals see and accordingly categorize the world and the evolutionary pressures responsible for this.
We conclude that, although the functionally referential framework by itself only explains a small part of animal vocalizations, as a conceptual guideline it forces us to explain the following: (1) Why do we find variation in the specificity of call production and perception? (2) How can we distinguish calls induced by external referents rather than being the expression of the emotional and motivational state of an animal experiencing a given behavioural context? (3) What are the differences in cognitive processes with regard to calls fulfilling the functional referential framework versus more context-based calls? To avoid underestimating the distribution of functionally referential communication throughout the mammals and beyond, a greater discussion and agreement between researchers as to what could potentially be referenced by animals and subsequently operationalizing definitions of external ‘objects’ or ‘events’ would be beneficial. This could then lead to the design of experiments testing specific hypotheses based on a broader approach to understand what cognitive processes are involved and what this means for the representational aspect of animal vocal communication.
- Top of page
- Functional Reference – the Framework
- Problems Classifying Calls as Functionally Referential
- Literature Cited
Thanks to Mark Hauber for the invitation to write this review, Julia Fischer, Katie Slocombe, Dan Blumstein, Robert Seyfarth, Zanna Clay and Roman Furrer for comments on previous versions of the manuscript and Brandon Wheeler for discussions. Funding was provided by the University of Zurich.
- Top of page
- Functional Reference – the Framework
- Problems Classifying Calls as Functionally Referential
- Literature Cited
- 2009: How Urgency Levels in Alarm Calls Influence the Forager's Response in Meerkats (Suricata Suricatta). MSc thesis, Univ. of Zurich, Zurich. .
- 2006: Language evolution: Semantic combinations in primate calls. Nature 441, 303. &
- 1992: Relation between food preference and food-elicited vocalizations in golden lion tamarins (Leontopithecus rosalia). J. Comp. Psychol. 106, 142—149. , &
- 1989: Do dwarf mongooses have a language?: warning vocalisations transmit complex information. S. Afr. J. Sci. 85, 447—450. &
- 1990: Language and Species. Univ. of Chicago Press, Chicago.
- 1999: The evolution of functionally referential alarm communication: multiple adaptations; multiple constraints. Evol. Comm. 3, 135—147.
- 1997: Alarm calling in yellow-bellied marmots: I. The meaning of situationally variable alarm calls. Anim. Behav. 53, 143—171. &
- 2012: The alarm call system of wild black-fronted titi monkeys, Callicebus nigrifrons. Behav. Ecol. Sociobiol. 66, 653—667. , , &
- 1990: Attending to behaviour versus attending to knowledge: examining monkey's attribution of mental states. Anim. Behav. 40, 742—753. &
- 2009: Food-associated calling sequences in bonobos. Anim. Behav. 77, 1387—1396. &
- 2011: Bonobos Extract Meaning from Call Sequences. PLoS ONE 6, 4. &
- 2012: Food-associated vocalizations in mammals and birds: what do these calls really mean? Anim. Behav. 83, 323—330. , &
- 1998: Costs of cooperative behaviour in suricates (Suricata suricatta). Proc. Roy. Soc. London, B 265, 185—190. , , , , , , , , &
- 1871: The Decent of man and Selection in Relation to sex. D. Appleton and Company, New York.
- 1984: Toque macaque food calls: semantic communication concerning food distribution in the environment. Anim. Behav. 32, 470—477.
- 2007: The evolution of functionally referential meaning in a structured world. J. Theor. Biol. 246, 225—233. , &
- 1991: Food-associated calls correlate with food preferences in cotton-top tamarins. Anim. Behav. 42, 931—937. , &
- 1997: Referential signals. In Perspectives in Ethology, (Owings, D. H., Beecher, M. D. & Thompson, N. S., eds). Plenum, New York, pp. 99—143.
- 2002: Cracking the Code: Communication and Cognition in Birds. In: The Cognitive Animal (Bekoff, M., Allen, C. & Burghardt, G. M., eds). MIT Press, pp. 315—322. .
- 2007: Representational signalling in birds. Biol. Lett. 3, 8—11. &
- 1994: Food calling and audience effect in male chickens, Gallus gallus: their relationships to food availability, courtship and social facilitation. Anim. Behav. 47, 1159—1170. &
- 1993: On the meaning of alarm calls: functional reference in an avian vocal system. Anim. Behav. 46, 23—38. , &
- 2010: ‘The bone is mine’: affective and referential aspects of dog growls. Anim. Behav. 79, 917—924. , , , &
- 2011: Primate vocal communication: a useful tool for understanding human speech and language evolution? Hum. Biol. 83, 153—173. & .
- 2002: Anti-predator behavior of group-living Malagasy primates: mixed evidence for a referential alarm call system. Behav. Ecol. Sociobiol. 51, 262—275. &
- 2011: Where is the Information in Animal Communication? In Animal Thinking: Contemporary Issues in Comparative Cognition, (Menzel, R. & Fischer, J., ed.). MIT Press, Cambridge, pp. 151—161.
- 2010: The Evolution of Language. Cambridge Univ. press, Cambridge.
- 2009: The evolution of urgency-based and functionally referential alarm calls in ground-dwelling species. Am. Nat. 173, 400—410. &
- 2005: The neurophysiology of functionally meaningful categories: Macaque ventrolateral prefrontal cortex plays a critical role in spontaneous categorization of species-specific vocalizations. J. Cognit. Neurosci. 17, 1471—1482. , , &
- 1984: Rhesus monkey (Macaca mulatta) screams: representational signaling in the recruitment of agonistic aid. Anim. Behav. 32, 182—193. , &
- 1987: Semantics of an avian alarm call system: The male domestic fowl, Gullus domesticus. Behaviour 102, 15—40. , &
- 1998: Functional referents and acoustic similarity field playback experiments with rhesus monkeys. Anim. Behav. 55, 1647—1658.
- 1993: Food-associated calls in rhesus macaques (Macaca mulatta): I. Socioecological factors. Behav. Ecol. 4, 194—205. &
- 2008: The Origins of Meaning. Oxford Univ. Press, Oxford, U. K.
- 2011: The Origins of Grammar. Oxford Univ. Press, Oxford, U. K.
- 2000: Food-related bray calls in wild bottlenose dolphins (Tursiops truncatus). Proc. Roy. Soc. B 267, 923—927.
- 2006: Perceptual specificity in the alarm calls of Gunnison's prairie dogs. Behav. Process. 73, 29—35. &
- 2009: Marmoset (Callithrix geoffroyi) food-associated calls are functionally referential. Ethology 115, 439—448. &
- 1993: Variation among mammalian alarm call systems and the problem of meaning in animal signals. Ethology 93, 177—197. &
- 1999: Response of foraging group members to sentinel calls in suricates, Suricata suricatta. Proc. Roy. Soc. Series B 266, 1013—1019.
- 2001: The acoustic structure of suricates' alarm calls varies with predator type and the level of response urgency. Proc. Roy. Soc. Series B 268, 2315—2324.
- 2009: What do functionally referential alarm calls refer to? In: Cognitive Ecology II, (Dukas, R. & Ratcliffe eds, J.). University of Chicago Press, Chicago, USA, pp. 229—248. .
- 2001: The information that receivers extract from alarm calls in suricates. Proc. Roy. Soc. Series B 268, 2485—2491. , &
- 2002: Suricate alarm calls signal predator class and urgency. Trends Cognit. Sci. 6, 55—57. , &
- 1992: Animal signals: reference, motivation, or both? In: Nonverbal Vocal Communication: Comparative and Developmental. Approaches (Papoucek, H., Jürgens, U. & Papoucek, M. eds). Cambridge Univ. Press, Cambridge, pp. 66—86. , & .
- 2011: Sources of acoustic variation: implications for production specificity and call categorization in chacma baboon (Papio ursinus) grunts. J. Acoust. Soc. Am. 129, 1631—1641. , , &
- 1977: On the occurrence and significanceof motivation-structural rules in some bird and mammal sounds. Am. Nat. 111, 855—869.
- 2009: Campbell's monkeys concatenate vocalizations into context-specific call sequences. Proc. Nat. Acad. Sci. U.S.A. 106: 22026—22031. , & .
- 2001: Sound on the rebound: Bringing form and function back to the forefront in understanding nonhuman primate vocal signaling. Evol. Anthropol. 10, 58—71. &
- 1997: The acoustic features of vowel-like grunt calls in chacma baboons (Papio cynocephalus ursinus). J. Acoust. Soc. Am. 101, 2951—2963. , &
- 1991: Ringtailed lemur anti-predator calls denote predator class, not response urgency. Anim. Behav. 26, 760—777. &
- 2009: Cross-modal individual recognition in domestic horses (Equus caballus). Proc. Nat. Acad. Sci. 106, 947—951. , &
- 1999: The meaning and function of grunt variants in baboons. Anim. Behav. 57, 583—592. , , &
- 2009: What do animal signals mean? Anim. Behav. 78, 233—240. , &
- 2009: The alarm call system of two species of black-and-white Colobus Monkeys (Colobus polykomos and Colobus guereza). J. Comp. Psychol. 123, 136—150. , &
- 2010: Predator-deterring alarm call sequences in Guereza colobus monkeys are meaningful to conspecifics. Anim. Behav. 80, 799—808. , &
- 2001: Individuality and male discrimination of female copulation calls in the yellow baboon. Anim. Behav. 61, 1023—1028.
- 2002: Information content of female copulation calls in yellow baboons. Am. J. Primatol. 56, 43—56. , , &
- 1980a: Vervet monkey alarm calls: Semantic communication in a free-ranging primate. Anim. Behav. 28, 1070—1094. , &
- 1980b: Monkey responses to three different alarm calls: Evidence for predator classification and semantic communication. Science 210, 801—803. , &
- 2009: Prairie dog alarm calls encode labels about predator colors. Anim. Cog. 12, 435—439. , &
- 2005: Functionally referential communication in a chimpanzee. Curr. Biol. 15, 1779—1784. &
- 2006: Food-associated calls in chimpanzees: responses to food types or relative food value? Anim. Behav. 72, 989—999. &
- 1977: The Behavior of Communicating: An Ethological Approach. Harvard Univ. Press, Cambridge, MA.
- 1981: Referents of animal communication. Anim. Behav. 29, 1273—1274.
- 1991: Animal communication and the study of cognition. In Cognitive Ethology: The Minds of Other Animals. (Ristau, C. A., ed.). Lawrence Erlbaum Association, Hillsdale, NJ, pp. 209—230.
- 1967: Auditory Communication among Vervet Monkeys (Cercopithecus aethiops). In Social Communication Among Primates, (Altmann, S. A. ed.). Univ. of Chicago Press, Chicago, USA, pp. 281—324. .
- 2011: All clear? Meerkats attend to contextual information in close calls to coordinate vigilance. Behav. Ecol. Sociobiol. 65, 1927—1934. , &
- 1998: Social calls coordinate foraging in greater spear-nosed bats. Anim. Behav. 55, 337—350. &
- 2003: The choice of post-conflict interactions in wild chimpanzees (Pan troglodytes). Behaviour 140, 1527—1559. &
- 1979: Perceiving surprising words in an unattended auditory channel. Quart. J. Exp. Psychol. 31, 281—286. &
- 2002: A syntactic rule in forest monkey communication. Anim. Behav. 63, 293—299.
- 2000: Referential labelling in wild Diana monkeys. Anim. Behav. 59, 917—927.
- 1997: Diana monkey long-distance calls: messages for conspecifics and predators. Anim. Behav. 53, 589—604. , &
- 1999: Conceptual semantics in a nonhuman primate. J. Comp. Psychol. 113, 33—42. , &