Handbook of Psychology, Volume 4: Experimental Psychology

Sound Source Segregation 141

amplitude modulated (unmodulated masker) is added simul-
taneously to the amplitude-modulated probe, there is only a
small change in the threshold for discriminating a change in
modulation depth. This is consistent with the critical-band
view of auditory processing in that the unmodulated masking
tone’s frequency is not near the carrier frequency of the probe.
If the masker is now amplitude modulated with the same am-
plitude modulation pattern (same rate of modulation) of the
probe, then threshold is increased to around 20% (a 15–16 dB
increase). The increase in modulation-depth threshold due to
the common pattern of modulation between the probe and
masker is referred to as modulation detection interference,
MDI, and typical results are shown in Figure 5.16.
One argument for why MDI occurs is that the probe and
masker are grouped together as a single sound source based on
the common pattern of amplitude modulation. Because the
common pattern of amplitude modulation is the basis of

the grouping, the auditory system has difficulty detecting

changes in modulation unless it affects the pattern of modula-

tion. Because changes in amplitude modulation depth of one

tone would have a small affect on the modulation pattern of the

mixture of the two tones, it is difficult for the auditory system

to detect changes in the depth of amplitude modulation for the

probe. One test of this argument is to make the pattern (rate) of

masker modulation different from that of the probe. In this

case, the masker and probe would not be grouped as a single

sound source, and MDI would be less or disappear. The data

shown in Figure 5.16 are consistent with this argument.

Models or Theories of Sound Source Segregation

One key aspect of accounting for sound source segregation is

the recognition that such processing requires the auditory

system to process sound across a wide spectral range and

Figure 5.15 Both the basic CMR task and results are shown. At the bottom the time-domain waveforms

for the narrow-band maskers (Target and Cue Bands) and the amplitude spectra for the maskers and the

signal are shown in a schematic form. The dotted line above each time-domain waveform depicts the am-

plitude envelope of the narrow-band noises. The listener is asked to detect a signal (S) which is always

added to the target band. In the target-band alone condition, the signal is difficult to detect. When a cue

band is added to the target band such that it is located in a different frequency region than the target band

and has an amplitude envelope that is different (not comodulated with) from the target band, there is little

change in threshold from the target-band alone condition. However, when the target and cue bands are co-

modulated, the threshold is lowered by approximately 12 dB, indicating that the comodulated condition

makes it easier for the listener to detect the signal. The waveforms are not drawn to scale. Source:From

Yost (2000), based on data from Hall et al. (1984), with permission.