| |
Abstract:
Positron Emission Tomography (PET) was used to investigate
cortical regions associated with increasing degrees of
intelligibility in noise-vocoded speech. Noise-vocoded speech
consists of a fixed number of noise bands which fluctuate in level
in the same way as corresponding bands of energy in the original
speech (Shannon et al., Science 270: 303-304, 1995). Its
intelligibility can be readily controlled by varying the number of
noise bands, here from 1 (completely unintelligible) to 16 (highly
intelligible). Controls for acoustic complexity consisted of 3 and
16 channel noise-vocoded speech which was spectrally inverted,
hence unintelligible. A behavioural measure of intelligibility for
each condition was used as a co-variate to identify cortical
regions whose activity increased with increasing intelligibility.
Analyses revealed bilateral activation in the superior temporal
sulcus (STS). Activation on the left was greater, extending to the
posterior temporal sulcus, and anteriorly to the temporal pole.
Left posterior STS was equally activated by potentially
intelligible speech whereas left anterior STS showed activity
increasing with intelligibility. A left pre-motor region also
increased its activity with increasing intelligibility. These
results support the notion of a left-lateralised anterior stream
associated with the understanding of speech (an auditory 'what'
pathway), and a posterior stream associated with sensory-motor
representations of speech (an 'action' pathway) (Scott et al,
Brain, 2000, in press).
|
