Subject: Abstract - humpback whale bioacoustics (fwd)

Mike Williamson (pita@www1.wheelock.edu)
Sun, 26 Jul 1998 12:52:22 -0400 (EDT)

Subject: dissertation abstract - humpback whale bioacoustics (fwd)


Mercado, E., III (1998). Humpback whale bioacoustics: From form to
    function. Unpublished doctoral dissertation, University of Hawaii,
    Honolulu.

This thesis investigates how humpback whales produce, perceive, and use
sounds from a comparative and computational perspective.  Biomimetic
models are developed within a systems-theoretic framework and then used to
analyze the properties of humpback whale sounds. First, sound transmission
is considered in terms of possible production mechanisms and the
propagation characteristics of shallow water environments frequented by
humpback whales.  A standard source-filter model (used to describe human
sound production) is shown to be well suited for characterizing sound
production by humpback whales.  Simulations of sound propagation based on
normal mode theory reveal that optimal frequencies for long range
propagation are higher than the frequencies used most often by humpbacks,
and that sounds may contain spectral information indicating how far they
have propagated. Next, sound reception is discussed.  A model of human
auditory processing is modified to emulate humpback whale auditory
processing as suggested by cochlear anatomical dimensions.  This auditory
model is used to generate visual representations of humpback whale sounds
that more clearly reveal what features are likely to be salient to
listening whales.  Additionally, the possibility that an unusual sensory
organ (the tubercle) plays a role in acoustic processing is assessed.
Spatial distributions of tubercles are described that suggest tubercles
may be useful for localizing sound sources.  Finally, these models are
integrated with self-organizing feature maps to create a biomimetic sound
classification system, and a detailed analysis of individual sounds and
sound patterns in humpback whale ``songs'' is performed.  This analysis
provides evidence that song sounds and sound patterns vary substantially
in terms of detectability and propagation potential, suggesting that they
do not all serve the same function.  New quantitative techniques are also
presented that allow for more objective characterizations of the long term
acoustic features of songs.
        The quantitative framework developed in this thesis provides a
basis for theoretical consideration of how humpback whales (and other
cetaceans) might use sound. Evidence is presented suggesting that
vocalizing humpbacks could use sounds not only to convey information to
other whales, but also to collect information about other whales.  In
particular, it is suggested that some sounds currently believed to be
primarily used as communicative signals, might be primarily used as sonar
signals. This theoretical framework is shown to be generalizable to other
baleen whales and to toothed whales.


I hope to make a postscript version of this disseration electronically
available within the next few months.


                                                Eduardo Mercado III
                                                mercado@hawaii.edu