Using this technique,we compared the rate of change of population
modal songs over time for our six conditions.Surprise scoring yielded
greater change than either global or local transition scoring.Local
scoring,in fact,made the population converge rather rapidly to locally
preferred transitions so that male songs often degenerated to repetition
of a single note or alternation between two notes.(This also gave these
runs very low within-generation synchronic diversity scores,so we did
not analyze this type of preference further.) Furthermore,coevolution
led to faster change than fixed female preferences,primarily when sur-
prise scoring was used.But the parameter with the biggest effect was
choir size:listening to only two males yielded much faster evolutionary
change than choosing from twenty.This effect could occur because with
bigger samples,traits could match preferences much more closely,and
so little movement of either would be necessitated over time.
We can easily visualize the difference between rate of change in the
fastest case and its parametric “opposite”(i.e.,changing all parameters),380 Peter Todd
Figure 20.1
Change in modal song from current generation G (left to right) to all previous generations
G¢(from G-1 at top to G-999 at bottom).Here a coevolving surprise-preference sample-
size-two population shows continuous rapid change over time.Fig.20.1