132 Audition
Figure 5.11 Equal loudness contours showing the level of a comparison
tone required to match the perceived loudness of a 1,000-Hz standard tone
presented at different levels (20, 40, 60, 80, and 100 dB SPL). Each curve is
an equal loudness contour. Source: Based on international standards.
approximation and over a considerable range of durations, if
the energy of the sound remains constant, detection thresh-
olds also remain constant (equal-energy rule). The duration at
which thresholds no longer change with increases in duration
(i.e., at 300 ms) is referred to as the integration time for detec-
tion (Viemeister & Plack, 1993).
Masking
When the threshold for detecting sound A is increased in the
presence of another sound, sound B, sound B is said to be a
masker for the signal, sound A. The amount of masking is the
amount of the increase in signal detection threshold due to
the presence of the masker. Figure 5.12 shows the thresholds
for detecting a signal tone of one frequency as a function of
tonal maskers of different frequencies. For these data, listen-
ers were asked to detect the presence of a short-duration tonal
signal presented just a few decibels above its threshold of
hearing. The level of the tonal masker that yielded threshold
performance for detecting the signal was determined for each
masker frequency. Both the similarity of the shape of the data
curves in Figure 5.12 to those in Figure 5.7 and the method-
ological similarities result in these psychophysical data being
referred to as psychophysical tuning curves. It is assumed
that the frequency selectivity suggested by psychophysical
tuning curves results from the frequency selectivity measured
in the auditory periphery (Moore, 1997).
The observation from Figure 5.12, that masking is greatest
when the frequency of the masker is near that of the signal,
was studied extensively by Harvey Fletcher in the 1940s. He
formed the concept of the critical band (Fletcher, 1953), stat-
ing that only a band of frequencies near that of the signal was
critical for masking. He further theorized that the amount of
masking of a tonal signal was proportional to the power of
the critical masking band. These observations have been con-
firmed by many experiments since the 1940s.
The tonal psychophysical curves are one method used to
measure this critical band. However, several interactions can
occur between a tonal signal and a tonal masker that can com-
plicate interpretation of some tone-on-tone masking results
(Wegel & Lane, 1924). If the signal and masker frequencies
differ by 20 or fewer Hz, then the tones interact to produce
slow fluctuations in overall intensity that result in the percep-
tion of beats (alteration in loudness), which can be used as a
cue for detecting the presence of the tonal signal. In addition,
the nonlinear properties of auditory transduction can produce
aural distortion products that can also provide a detection
cue. The tonal masker can produce aural harmonics, which
are frequencies at the harmonics of the masker frequency
caused by the nonlinear process. The nonlinear properties of
transduction can produce difference tones, which are fre-
quencies equal to differences between the frequencies of the
masker and signal. The psychophysical tuning curve method
reduces, but does not always eliminate, the effect of many of
these stimulus interactions as possible detection cues.
The preferred method for measuring the critical band is
the band-reject noise paradigm as shown in Figure 5.13. A
band-reject noise has a frequency region filtered out of the
noise, which for masking is a frequency region surrounding
the signal frequency. This band-reject, noise-masking proce-
dure (Moore, 1986) reduces or eliminates all of the interactiveFigure 5.12 Three psychophysical tuning curves for simultaneous mask-
ing are shown. The different curves are for conditions in which the signal fre-
quency was 300, 1000, and 3000 Hz. Source: From Yost (2000), adapted
from data of Wightman, McGee, and Kramer (1977), with permission.