Extensions of population-interval models for pitch:
pitch multiplicity, harmonic resolvability and pitch masking
Peter Cariani
Eaton Peabody Laboratory of Auditory Physiology
Massachusetts Eye & Ear Infirmary
Harvard Medical School
Boston, MA
ABSTRACT
Population-wide all-order interspike interval distributions at the
level the auditory nerve have been used to successfully account for
the low pitches of complex tones. We will present results from recent
auditory nerve simulations that illustrate how population-interval
models might be extended to handle pure tone pitch discrimination,
pitch multiplicity, harmonic resolvability, and pitch masking. Such
models implicitly incorporate both cochlear excitation patterns (in
the competition for cochlear territory) and the interspike interval
distributions that they produce. Pitch masking patterns predicted from
simulated population-interval distributions reflect basic features of
pure tone masking audiograms that depend on cochlear excitation
patterns (e.g. upward spread of masking). Predicted pitch saliences of
AM tones were found to depend upon both carrier frequency and harmonic
number (higher carriers and harmonic numbers produce weaker
pitches). For a harmonic complex (F0=200 Hz; n=1-12), harmonics 1-4
were resolvable. We will discuss the pros and cons of such
representations as neural substrates for explaining differences
between resolved vs. unresolved harmonics and between low- and high
frequency hearing.