Page 43 - 2018Fall

maintaining attention on one source in a multitalker envi- ronment. However, are there other areas of the brain recruit- ed to help us listen for one talker or to coordinate switching of attention from one talker to another? Does it depend on what features we are attending to, such as listening for the speaker on my left side or listening for the speaker with the highest pitch?
In the visual attention literature, there is a prevalent dual- path theory for spatial and nonspatial feature processing. The dorsal stream (or the “where” pathway) is involved with processing spatial information, whereas the ventral stream (or the “what” pathway) is involved with object identifica- tion and recognition (Desimone and Duncan, 1995; Figure 4a). A similar dual pathway has also been proposed for audi- tory information processing (Rauschecker and Scott, 2009), with areas of the primary auditory cortex and surrounding areas organized spatially to feed into this dual-pathway con- figuration (Rauschecker, 1998). Even before a sound begins, we can direct our attention volitionally to “listen out” for a specific feature of sound. In a recent M/EEG study us- ing inverse imaging constrained by MRI information, our laboratory asked subjects to report back one of two spo- ken digits (Lee et al., 2013b). These two spoken digits were monotonized (one with a higher pitch and the other with a lower pitch) and convolved with two different head-related transfer functions, one simulating the source coming from the left and the other from the right. We cued the subjects using a visual arrow (viz., left, right, up, or down), indicating which feature they should prepare to attend to even before the onset of the stimulus. We found that the neural response in the left frontal eye fields, part of the dorsal “where” path- way, was enhanced in the preparation and during a spatial attention task when subjects were prompted to pay attention to a location by a left or right arrow. When, instead, subjects were prompted to attend to a higher or lower pitch by an up or down arrow, the left posterior superior temporal sul- cus, an area previously implicated for pitch categorization and possibly in the auditory “what” pathway, was greater in preparation for a pitch attention task. These findings are in line with other neuroimaging studies (e.g., fMRI, EEG; Diaconescu et al., 2011), showing that the deployment of auditory spatial attention engages similar cortical regions associated with visual attention, but selectively attending to nonspatial features invokes cortical areas that are specific to the attended sensory modality. These observations generally agree with the notion that the frontoparietal attention sys- tem comproed of the precentral sulcus/inferior frontal sul-
cus, dorsolateral prefrontal cortex/middle frontal gyrus, and intraparietal sulcus/inferior parietal lobule regions (Figure 4b) is engaged when we want to listen out for a talker in a crowded environment (Eckert et al., 2016).
To carry on social conversations, we also need to switch at- tention from one talker to another. Converging evidence suggests that the frontoparietal attention system also par- ticipates in attention switching in both the visual and audi- tory modalities. When listeners switched attention between a male and female stream, elevated activation was found in the posterior parietal cortex (Shomstein and Yantis, 2006). A similar region in this cortical area was also more active when listeners had to switch attention from one ear to the other. Combined with other converging evidence found in the vi- sual attention literature, this suggests that the posterior pari- etal cortex participates in both spatial and nonspatial atten- tion in auditory and visual modalities (Serences et al., 2004). However, one recent M/EEG study suggests that switching based on different features might recruit different behavioral strategies, leading to recruitment of different cortical areas (Larson and Lee, 2014). In this study, listeners were cued ini- tially to attend to one of two talkers. These two talkers were either separated in auditory space (to the left or right) or monotonized to have different pitches. Subjects were cued to listen to one of the two talkers in the beginning of the trial and were asked to either maintain or switch attention during an inserted silent gap in this two-speaker mixture. The result shows that the right temporoparietal junction (TPJ; Figure 4a) was more activated when listeners had to switch atten- tion based on the available spatial cues only and that the ac- tivity in right TPJ during the attention-switching gap period correlates significantly with how well the listeners perform this task. However, when listeners were told to switch atten- tion and the only cue available was the pitch difference be- tween talkers, the left inferior parietal supramarginal area was more engaged throughout the trial in the maintain tri- als compared with the switch trials. This suggests that the listeners might use different behavioral strategies to accom- plish this nonspatial switching task, thereby recruiting a dif- ferent cortical area with a different activation time course.
What if we make our listening condition even more chal- lenging than switching attention between speakers, such as reading the conference abstract while attending to the speaker? In a recent meta-analysis study of 10 speech-rec- ognition experiments using fMRI, it was found that the cortical regions in the cingulo-opercular executive-control system consistently showed elevated activity during the
Fall 2017 | Acoustics Today | 41