Sceptical Hens

Many animal signals are condition-dependent and hence inherently honest. But this cannot be true of vocalizations such as food calls because energetic costs are trivial: constraints on deceptive usage are likely instead to be social. Spectrogram cross-correlation analyses have revealed that male identity accounts for a substantial proportion of variation in call structure. Playback experiments using an habituation / dishabituation paradigm demonstrate vocal discrimination by hens. When a different male's call is presented in the probe trial following the habituation series, the response evoked is significantly greater than when hens hear a new call from the same male. We next asked whether hens can track differences in male reliability. This is a more challenging cognitive task because it requires construction of linking individual vocal idiosyncrasies with the probability of an experiencing a particular class of environmental event. Hens heard a series of food calls from two males, one honest (each playback was followed by food) and the other dishonest. Hens learned this relationship in three tests and had consistently different responses thereafter. They also remembered differences in male reliability for at least 24h and generalized successfully to completely novel calls. Follow-up experiments suggest that the threshold for detection of unreliable signallers is approximately 40%, a value that matches the observed frequency of deceptive calling in natural social groups. Male behaviour thus reflects the ability of hens to discriminate against dishonest callers. This is the first experimental evidence for a social constraint on production of low-cost signals.