This diversity could actually be increased if female song preferences
could change faster,altering within any given female’s lifetime rather
than just between mother and daughter.This is exactly the role that
learning can play,enabling adaptations faster than evolution can accom-
plish (Todd and Miller 1991b).By combining (co)evolution and learn-
ing,we may be able to explore further questions about musical diversity
relating to culture and individual song complexity.
One obvious place we could add learning to our system is in the cre-
ation of female musical expectations:where should their transition tables
come from? In our current setup,females inherit transition tables from
their mother and father.Because of this,“surprising”note transitions can
be surprising only relative to a particular female’s inherited expectations.
But certainly for humans,and for other animals as well,expectations
are built up through experience and learning within one’s lifetime (see
Bharucha and Todd 1989).Instead we can let a female learn expectations
about note transitions based on a set of songs from her current genera-
tion,or from the previous generation,as if she has heard those songs and
picked up knowledge of her culture from them.Then she will be surprised
when she hears something new that toys with these learned expectations,
building them up,and violating them.We expect that using learning to
384 Peter Todd
Figure 20.4
Diversity of songs in a noncoevolving global-preference sample-size-twenty population,
showing loss of diversity over time.
Fig.20.4