as many as thirty-five different displays in a wide range of combinations
and sequences (Hanlon and Messenger 1996).
Traditional reasons given for the evolution of communication do
not provide particularly compelling explanations for such between-
individual signal diversity.If communication is viewed as a means of
transferring veridical information from one organism to another (see
Hauser 1996),we would expect repeated communications of the same
information by one individual or within a population to be performed in
a similar manner to avoid misinterpretation by the receiver.In the par-
ticular case of accurate species identification for mating purposes,there
should also be little variation between signals of conspecifics.If commu-
nication is seen instead as a way to manipulate the behavior of another
organism,which can include nonveridical deceit (see Krebs and Dawkins
1984),the signal used in any particular case should be the single one
found to be most effective.And if communication is considered a means
of altruistically benefiting one’s genetic relatives (Ackley and Littman
1994),we would expect convergence onto stable but possibly family-
specific ways to help one another.
What then can drive the evolution of a large variety of elaborate
communication signals? Consider the evolutionary composition tools
described in the previous section.A common problem with the auto-
mated (nonhuman) fitness functions was that they could be tricked by
musically uninteresting solutions,on which the population would then
converge because of their high fitness values.Human critics could avoid
being so tricked by changing what they were listening for in the popu-
lation,and reacting to any cheating musical behaviors.This type of
responsive fitness evaluation can also keep the population from con-
verging on a single sort of behavior and can thereby maintain a diver-
sity of musical output.As a consequence,we decided to investigate the
role that coevolution of critics and music creators could play in engen-
dering musical diversity within a population and across several genera-
tions.In particular,we wanted to test the effects of different preference
mechanisms on diversity to see if some mechanisms would lead to more
diverse populations than others.Coevolution,Sexual Selection,and Mate Choice
Coevolution can create a diversity of musical or other behavior in two
ways.First,it can produce diversity within a population at any one time.
This synchronic diversitycan be generated,for example,through the
process of sexual selection,when females choose mates based on partic-
ular traits the males bear.When both female preferences for particular
traits and male traits themselves coevolve,new species can form,374 Peter Todd