Selective attention to human voice enhances brain activity bilaterally in the superior temporal sulcus.

Regional cerebral blood flow was measured with positron emission tomography (PET) in 10 healthy male volunteers. They heard two binaurally delivered concurrent stories, one spoken by a male voice and the other by a female voice. A third story was presented at the same time as a text running on a screen. The subjects were instructed to attend silently to one of the stories at a time. In an additional resting condition, no stories were delivered. PET data showed that in comparison with the reading condition, the brain activity in the speech-listening conditions was enhanced bilaterally in the anterior superior temporal sulcus including cortical areas that have been reported to be specifically sensitive to human voice. Previous studies on attention to non-linguistic sounds and visual objects, in turn, showed prefrontal activations that are presumably related to attentional control functions. However, comparisons of the present speech-listening and reading conditions with each other or with the resting condition indicated no prefrontal activity, except for an activation in the inferior frontal cortex that was presumably associated with semantic and syntactic processing of the attended story. Thus, speech listening, as well as reading, even in a distracting environment appears to depend less on the prefrontal control functions than do other types of attention-demanding tasks, probably because selective attention to speech and written text are over-learned actions rehearsed daily.

Linguistic processing in visual and modality-nonspecific brain areas: PET recordings during selective attention.

Positron emission tomography (PET) was used to investigate the neural basis of selective processing of linguistic material during concurrent presentation of multiple stimulus streams ("cocktail-party effect"). Fifteen healthy right-handed adult males were to attend to one of three simultaneously presented messages: one presented visually, one to the left ear, and one to the right ear. During the control condition, subjects attended to visually presented consonant letter strings and ignored auditory messages. This paper reports the modality-nonspecific language processing and visual word-form processing, whereas the auditory attention effects have been reported elsewhere [Cogn. Brain Res. 17 (2003) 201]. The left-hemisphere areas activated by both the selective processing of text and speech were as follows: the inferior prefrontal (Brodmann's area, BA 45, 47), anterior temporal (BA 38), posterior insular (BA 13), inferior (BA 20) and middle temporal (BA 21), occipital (BA 18/30) cortices, the caudate nucleus, and the amygdala. In addition, bilateral activations were observed in the medial occipito-temporal cortex and the cerebellum. Decreases of activation during both text and speech processing were found in the parietal (BA 7, 40), frontal (BA 6, 8, 44) and occipito-temporal (BA 37) regions of the right hemisphere. Furthermore, the present data suggest that the left occipito-temporal cortex (BA 18, 20, 37, 21) can be subdivided into three functionally distinct regions in the posterior-anterior direction on the basis of their activation during attentive processing of sublexical orthography, visual word form, and supramodal higher-level aspects of language.

Hemispheric lateralization of cerebral blood-flow changes during selective listening to dichotically presented continuous speech.

Regional cerebral blood flow (rCBF) was measured with positron emission tomography (PET) while subjects were selectively listening to continuous speech delivered to one ear and ignoring concurrent speech delivered to the opposite ear, as well as concurrent text or letter strings running on a screen. rCBF patterns associated with selective listening either to the left-ear or right-ear speech message were compared with each other and with rCBF patterns in two visual-attention conditions in which the subjects ignored both speech messages and either read the text or discriminated the meaningless letter strings moving on the screen. Attention to either speech message was associated with enhanced activity in the superior temporal cortex of the language-dominant left hemisphere, as well as in the superior and middle temporal cortex of the right hemisphere suggesting enhanced processing of prosodic features in the attended speech. Moreover, enhanced activity during attention to either speech message was observed in the right parietal areas known to have an important role in directing spatial attention. Evidence was also found for attentional tuning of the left and right auditory cortices to select information from the contralateral auditory hemispace.