ascending auditory pathway to at least one of the auditory cortices. The rare condition of
cortical deafness occurs following lesions to both superior temporal lobes (see Ref. 19 for
review). The perception of pitch, as opposed to the detection of a pure tone at a given fre-
quency, has been investigated using complex stimuli where the pitch is not immediately
determined by the spectral characteristics of the sound. Zatorre^20 used a stimulus where
the pitch is determined by the ‘missing fundamental’of the spectral representation. The
stimulus in this experiment, considered in the frequency domain, is a harmonic series
where the lower harmonics including the fundamental stimulus are removed. The pitch of
this stimulus is determined by the missing fundamental, against the idea that for a pitch to
be heard there needs to be a representation of that pitch in the stimulus spectrum. The
determination of pitch by any frequency domain model requires a form of template fit-
ting;^21 the spectrum still contains harmonics that are spacedby an amount equal to the
fundamental frequency. Alternatively, the pitch of this stimulus might be explained by tem-
poral domain models; the waveform of this stimulus has a periodicity equal to the recip-
rocal of the fundamental, both before and after removal of the lower harmonics. Using the
missing fundamental stimulus Zatorre investigated the perception of this complex pitch in
subjects with well-characterized lesions affecting the superior temporal lobes. Subjects
with right lesions involving the primary auditory cortex had deficits in perception of this
pitch, but not subjects with right superior temporal lobe lesions sparing the primary aud-
itory cortex, or subjects with lesions of the left primary auditory cortex. This suggests a par-
ticular involvement of the right primary auditory cortex in complex pitch perception.
Patterson, Yost, and others have developed stimuli containing regular intervals where the
pitch properties are most parsimoniously explained by the fine temporal rather than
the spectral characteristics of the sound.22,23Of course, both characteristics are interchange-
able at the stimulus level, but these stimuli are manipulated in such a way that spectral
template models are difficult to apply because of the filtering properties of the auditory sys-
tem. Iterated rippled noise (IRN) is one such type of regular interval sound produced by
taking bandpass noise and adding it to itself in a delay-and-add cycle. With increasing delay-
and-add cycles (or gain in each cycle) the stimulus becomes more regular, as shown by the
increasing first peak in the autocorrelation spectrum. In parallel with this temporal change,
a pitch emerges from the noise with a signal-to-noise ratio that parallels the degree of tem-
poral regularity of the auditory signal. In contrast, the spectrum of the auditory signal
retains its bandpass properties. It is therefore easier to infer in the case of this signal (than
for the missing fundamental) that changes in perceived properties are due to the fine tem-
poral structure rather than the spectral structure in the auditory signal (see Ref. 2 for more
detailed discussion). Using this stimulus I investigated one of Peretz’s patients, I.R., with
apperceptive amusia following bilateral temporal lobe lesions^24 (see Chapter 13 by Peretz,
this volume, for further discussion of amusia). This patient has normal appreciation of pitch
but is unable to perceive musical sequences. She has an intact pathway up to and including
the right primary auditory cortex, but effective ‘isolation’of that cortex from other areas of
cortex that are lesioned. Testing this patient with IRN showed that she had a normal thresh-
old for the detection of this stimulus presented to both ears. Testing each ear separately with
contralateral masking noise showed that her normal performance was entirely accounted for
by the performance at the left ear, the predominant input to the intact right auditory cortex.
171