9

respond to food calls by rapidly approaching the male, but we cannot exclude the possibility that this response is mediated by other aspects of the male's behaviour, such as visual displays, or by motivational factors quite unrelated to food, such as sexual receptivity. Interpretation of the data currently available on food-associated calls in chimpanzees (Hauser and Wrangham, 1987; Hauser et al., 1993; but see Clark and Wrangham, 1993,1994), golden lion tamarins (Benz et al., 1992; Benz 1993), toque macaques (Dittus, 1984) and rhesus macaques (Hauser and Marler 1993a) is similarly constrained by the lack of playback experiments. The idea of functional reference has the merit of making a clear theoretical prediction about the behaviour of hens, which is that presentation of food calls in the absence of other cues will be sufficient to elicit anticipatory feeding movements. Recently-completed experiments have confirmed this prediction. Hens approach a loudspeaker playing food calls, moving their heads repeatedly downward to fixate frontally on the substrate and occasionally pecking at the floor. These responses are significantly more probable during playback of food calls than during playback of either ground alarm calls (which are structurally-similar) or contact calls (which are produced under similar social circumstances) (Evans, in prep). The data from studies of production and perception thus suggest that chickens have a total of at least three functionally referential calls, which encode information about the discovery of food and about the appearance of predators (Fig. 1).

VI. SPECIFICITY

Analyses of referential signalling have always been concerned with the issue of specificity, that is, with the size of the stimulus set (Fig. 1a) corresponding to a particular call type (Fig. 1c). Analogies between the natural communication of animals and properties of human language (e.g., Cheney and Seyfarth, 1990; Evans and Marler, 1995) will be most compelling when animal signals are shown to be evoked by a small class of environmental events. Unfortunately, we have data on so few referential systems that it is not yet possible to evaluate sensibly patterns of specificity. It is, however, clear that in some species the external stimuli evoking calls are quite narrowly-defined. For example, the 'eagle alarms' produced by adult vervet monkeys are elicited principally by martial eagles (Seyfarth and Cheney, 1980). Comparisons between the morphology of martial eagles and those of the other raptors to which vervets give eagle alarms suggest that the monkeys may be responding not only to the overall form of potential predators, but also to quite subtle details, such as melanic markings beneath the head and under the wings, and perhaps also to dynamic cues such as characteristic differences in flight pattern.

Studies of the natural vocal behaviour of chickens provide a clear contrast to the vervet system. Chicken aerial alarm calls are associated with a relatively large set of airborne objects (Gyger et al., 1987). Laboratory experiments, in which birds were presented with computer-generated animations, have permitted the systematic manipulation of stimulus characteristics. The results demonstrate that aerial alarm calls are dependent upon attributes such as size and apparent speed (Evans et al., 1993b). Spatial location and shape are also important (Evans and Marler, in prep). These studies have provided a precise mapping of the stimulus parameters corresponding to call production and, together with measurements of non-vocal behaviour, they allow us to infer something about the way in which visual stimuli are recognized and categorized (Fig. 1b). Work of this kind is thus logically complementary to studies of visual categorization in animals using instrumental (e.g., Herrnstein et al., 1976; Herrnstein, 1984, 1991; Lea, 1984) or neurophysiological techniques (Kendrick and Baldwin, 1987; Perrett et al.,1987).

VIa.Developmental Plasticity
The specificity of calls produced by adult animals is likely to be a consequence not only of selection pressures operating over evolutionary time, but also of developmental factors.