Referential Signals

CHRISTOPHER S. EVANS

Recent research on mechanisms of animal communication has been concerned largely with identifying the factors that have influenced the design of sexually-selected signals (II). Work in this area has produced exciting advances that integrate proximate and ultimate levels of analysis. It has, however, neglected a whole class of signals that encode information in addition to attributes of the sender. Birds and primates are now known to have specific calls that allow companions to predict environmental events, such as the discovery of food or the appearance of a particular type of predator (III). These signals are functionally referential.

I propose a framework for recognizing referential signals (IV), together with a strategy for the experimental analysis of such systems (V). Work on referential signalling is then reviewed, concentrating on the most obvious gaps in current models. I begin by considering the behaviour of callers, assessing production specificity (VI), the evidence for developmental plasticity in call usage (VIa) and some problems introduced by nomenclature (VIb). The factors responsible for transitions between different signal classes have been identified in some systems, but little is known about fine-grained variation in signal morphology. Conclusions about the information encoded in animal signals will be sensitive to the level of analysis selected because relatively gross changes, such as variation in call type, can reflect external events while more subtle variation in structure remains affective in nature (VII). The issue of signal design is then discussed. The factors responsible for determining the physical form of referential signals are much less well described than those that have shaped signals reflecting properties such as fighting ability or parasite load (VIII).

I suggest that there are aspects of signalling behaviour that cannot be understood by focussing only upon issues of meaning, but rather will require consideration of additional factors, including the likelihood of tonic communication and of signalling to predators. The contribution of contextual cues has probably been underestimated (IX). Social context plays an important role in determining whether signals are produced (IXa) and is thus an essential component in any model that seeks to predict calling behaviour. It is likely that contextual information also modulates the responses of receivers, both by providing additional information synchronous with the signal and by building up associations with particular signal types over the course of development (IXb).

I then assess critically some aspects of the relationship between communication and cognitive processes. Current evidence does not require us to conclude either that referential signals evoke representations of the eliciting event in the minds of receivers (X) or that they are used deceptively (XI). Finally, I consider the evolution of referential signals and suggest that ecological factors may have played an important role (XII).

 Download reprint

Return to Animal Behaviour Lab publications