create note transition expectations,rather than evolving them,will allow
the population to change its tune even more rapidly than the cases
described here,because expectations will be able to shift just as rapidly as
the songs themselves—learning operates faster than selection.
Furthermore,we could allow learning in females to occur at an even
faster time scale so that instead of habituating to songs heard too many
times last week,for example,each female could habituate to notes and
phrases heard too many times within the current male’s song.In this case,
females would seek novelty and expectation-violation within each song
they hear.To sing preferred songs,males would have to balance the
amount of repetition and novelty in their song.We expect that this type
of real-time preference learning will lead to increased complexity of the
internal structure of the songs themselves,not just of the population of
songs.
Being able to witness and analyze interactions of musical behavior,
learning,culture,and evolution in these new ways is the best argument
for using evolutionary computer simulations to study the origins of
music.By harnessing the power of computers to mimic these adaptive
processes from nature,we gain a new way of listening in on the nascent
songs of bygone worlds.385 Simulating the Evolution of Musical Behavior
Figure 20.5
Diversity of songs in a coevolving surprise-preference sample-size-twenty population,
showing evolution of two tightly clustered song species between generations 600 and 800.Fig.20.5