The neural basis of stream segregation in the primary auditory cortex.<p>

Christophe Micheyl (1), Biao Tian (2), Robert P. Carlyon (3), and
Joseph P. Rauschecker (2)<p>

1- UMR CNRS 5020, Lyon, France <br>
2- Georgetown University, Washington DC, USA <br>
3- MRC - Cognition & Brain Sciences Unit, Cambridge, UK <br>
<p>
This study was funded in part by an Engineering and Physical Sciences
Research Council Life Sciences Network grant GR/M90146 to
Prof. C.J. Darwin, Exptl. Psychol., University of Sussex.
<p>
The sounds reaching our ears often contain components arising from
multiple acoustic sources. The task of our brain is to organize this
complex raw sensory material into meaningful perceptual entities. One
of the fundamental perceptual processes of auditory scene analysis is
known as "stream segregation". Auditory stream segregation refers to
the perceptual teasing apart, and linking over time of frequency
components in sound sequences, which results in the formation of
auditory "streams". In everyday life, a stream may be the voice of a
speaker in a cocktail-party background, or the sound of a violin among
other instruments in an orchestral symphony. In the laboratory, a now
classic demonstration of the stream segregation phenomenon involves
the presentation of tones alternating in frequency in an ABAB pattern:
when the tempo is slow and/or the frequency difference between the A
and B tones is small, listeners report hearing a single stream of
tones alternating in frequency; when the tempo is fast and/or the
frequency difference is large, listeners report hearing two distinct,
monotonic streams. Although stream segregation has been largely
studied over the past fifty years, the brain mechanisms behind it
still remain unclear to date. In the present study, we investigated
further the hypothesis suggested by earlier results from Fishman et
al. (2000), that perceptual auditory stream segregation is reflected
in the primary auditory cortex by the differential suppression of
neural responses to off-BF (best frequency) tones by preceding on-BF
tones. We recorded the responses of single units in the primary
auditory cortex (A1) of an awake monkey to 10-sec ABA_ sequences of
tones - where A and B represent 125-ms tone pips of a different
frequency, and _ stands for a 125-ms silent gap. The A-tone frequency
was always close to the unit's BF, whilst the B-tone frequency was set
to 1, 3, 6, or 9 semitones (STs) above A. The results revealed that
for A-B separations of 3, 6, and 9 ST, but not 1 ST, the amplitude of
the neural response to the B-tones decreased over part or all of the
sequence duration. Conversely, a trend for A tone responses to
increase over time was observed. As a result, the ratio of B-tone to
A-tone responses increased over time. These electrophysiological
results are consistent with psychoacoustical data in humans, which
indicate that with the type of sequences used here, auditory stream
segregation "builds up" over roughly the first 5-10 secs of
stimulation. They reveal that the differential suppression of
responses to off-BF tones by preceding on-BF tones, which Fishman et
al. (2000) have proposed to be the neural basis of stream segregation
at the cortical level, is a dynamic phenomenon. The decrease in
amplitude of the B-tone responses was not observed when these tones
were presented alone - i.e. with the surrounding A tones removed -,
suggesting that the effect was not simply due to long-term adaptation
of off-BF responses.