Many models of animal signal evolution fail to incorporate an explicit stra
tegy for receivers prior to the evolution of signals. When reasonable assum
ptions are made for such strategies, we have shown that there is a minimal,
accuracy of signal coding that is required before receivers should attend
to signals (Bradbury & Vehrencamp 1998, Principles of Animal Communication)
. Depending upon the relative payoffs of correct and incorrect decisions by
receivers, this minimal accuracy can be quite high. Here we use this resul
t to explain why so many signals appear to be traits that provided useful i
nformation to receivers before becoming ritualized into signals. Our model
also supports one prediction of sensory drive models: that latent preferenc
es may selectively favour some signal precursors over others. However, it i
mposes a serious constraint on sensory drive by requiring that there be suf
ficient benefits to a receiver to compensate for the costs of disrupting th
e optimal receiver strategy used before exploitation. Finally, we discuss t
he overlap between signal honesty and accuracy and show how senders that co
mpletely disagree with receivers about appropriate receiver decisions may s
till-benefit by providing moderately honest and accurate signals. (C) 2000
The Association for the Study of Animal Behaviour.