136 Audition
location in space. When such HRTF filters are used, listeners
are much more likely to localize the sounds in space at a
location appropriate for the specific HRTF used than they are
to lateralize the sound inside the head. Thus, HRTF-filtered
sound presentations over headphones can create a virtual au-
ditory environment simulating sound localization in the real
world. Under the proper conditions, sounds delivered over
headphones are perceived as nearly indistinguishable from
the sound delivered from actual sources (Gilkey & Andersen,
1997; Wightman & Kisltler, 1989b).
The Effects of Precedence
Although reflections from surfaces may aid distance judge-
ments, they could also offer a confusing auditory scene for
sound localization, because each reflection could be misinter-
preted as a possible sound source location. In most real-world
spaces, reflections do not have a significant effect on either
the location or on the fidelity of the sound from the originat-
ing source. The sound from the source will reach a listener
before that of any reflection due to the longer path any reflec-
tion must travel. Hence, it is as if the sound from the source
takes perceptual precedence over that from reflections
(Litovsky, Colburn, Yost, & Guzman, 1999).
The effects of precedence (Litovsky et al., 1999) include
that fact that the reflections are rarely perceived as separate
echoes (fusion), the perceived location of a sound source in a
reflective environment is dominated by the location of the
source and not by the location of reflections (location domi-
nance), and information about reflections is suppressed rela-
tive to that about the source (discrimination suppression).
Evidence also suggests that the effects of precedence may be
influenced by a listener’s prior listening experience in an
acoustic environment. A common paradigm (Litovsky et al.,
1999) for studying the effects of precedence involves the pre-
sentation of a transient from one loudspeaker (the lead or
source sound), followed a few milliseconds later by an identi-
cal transient presented from a different loudspeaker (the lag or
reflected sound). In most cases in this lead-lag paradigm, a sin-
gle transient is perceived (fusion), at the location of the lead
loudspeaker (localization dominance), and the spatial acuity
of the lag is reduced relative to conditions when the lag was
presented in isolation of the lead (discrimination suppression).
SOUND SOURCE SEGREGATION
Any animal’s auditory experience probably involves process-
ing several simultaneously or nearly simultaneously occurring
sound sources. Several stimulus cues have been suggested as
possibilities for segregating sound sources in the complex
acoustic world: spectral separation, temporal separation, spa-
tial separation, pitch and timbre (harmonicity and temporal
regularity), spectral profiles, common onsets and offsets, and
common modulation (Yost & Sheft, 1993; Yost, Popper, &
Faye, 1993).
Recall from the description of the auditory periphery that
the auditory nerve codes for the spectral-temporal properties
of sound. Sounds from every sound source in an acoustic en-
vironment are combined into a single complex sound field
that stimulates the inner ear. The auditory periphery codes
for the spectral-temporal structure of this complex sound
field. The spectral-temporal code must be analyzed to deter-
mine the potential sound sources. That is, the spectral-
temporal neural properties must be deconvolved into subsets
of spectral-temporal patterns representing the sound origi-
nating from each individual sound source. This form of
analysis is presumably performed by the central auditory
nervous system. Note that in order for this type of analysis
to take place, computations must be made across frequency
and over time (Bregman, 1990).Spectral SeparationIf two sound sources had very different and nonoverlapping
spectral structures, the frequency-resolving ability of the au-
ditory system might segregate the two sound sources very
nicely into two patterns. Thus, in some cases the frequency-
resolving abilities of the auditory system can aid in sound
source segregation, but not in all cases. The difficulty arises
when the spectra of sounds from different sources overlap in
frequency and time.Temporal SeparationClearly, if the sound from two sources occurs at different
times, and there is little, if any, temporal masking, then sound
source segregation is possible. In many real-world situations,
the sound from one source is intermittent and may overlap
in time with sounds from other sources that are also intermit-
tent. In addition to the question of segregation of one sound
source from other sound sources, this stimulus situation also
addresses the question of how an intermittent sound from a
source continues to be identified as originating from that
source, especially if other sounds occur at or about the same
time. A series of studies referred to as auditory stream analy-
sis investigates this type of stimulus condition (Bregman,
1990).
An early context for the study of auditory stream process-
ing involved the presentation of two tones of different