Are you listening? Brain activation associated with sustained nonspatial auditory attention in the presence and absence of stimulation

Abstract

Neuroimaging studies investigating the voluntary (top-down) control of attention largely agree that this process recruits several frontal and parietal brain regions. Since most studies used attention tasks requiring several higher-order cognitive functions (e.g. working memory, semantic processing, temporal integration, spatial orienting) as well as different attentional mechanisms (attention shifting, distractor filtering), it is unclear what exactly the observed frontoparietal activations reflect. The present functional magnetic resonance imaging study investigated, within the same participants, signal changes in (1) a "Simple Attention" task in which participants attended to a single melody, (2) a "Selective Attention" task in which they simultaneously ignored another melody, and (3) a "Beep Monitoring" task in which participants listened in silence for a faint beep. Compared to resting conditions with identical stimulation, all tasks produced robust activation increases in auditory cortex, cross-modal inhibition in visual and somatosensory cortex, and decreases in the default mode network, indicating that participants were indeed focusing their attention on the auditory domain. However, signal increases in frontal and parietal brain areas were only observed for tasks 1 and 2, but completely absent for task 3. These results lead to the following conclusions: under most conditions, frontoparietal activations are crucial for attention since they subserve higher-order cognitive functions inherently related to attention. However, under circumstances that minimize other demands, nonspatial auditory attention in the absence of stimulation can be maintained without concurrent frontal or parietal activations.

Keywords: attention; auditory perception; baseline shift; default mode network; fMRI.

Figure 1

Time course of a stimulation and response trial pair. Illustrated is an “attend high” Selective Attention trial. The visual cue “H” at the beginning of the trial informs the participant to count rising tone triplets in the high‐frequency melody (depicted in white) while ignoring the low‐frequency melody (black). The gray ellipses highlight the rising triplets in the to‐be‐attended stream and indicate that the correct count in this trial would be 4. Which response button is associated with this count is only revealed after the next volume acquisition, so that the BOLD images acquired after the end of the stimulation period should not reflect preparation of a specific motor response. In the present example, even numbers are mapped onto the left response button, so that the correct response would be with the left hand. The time course of stimulation trials in the Simple Attention, Beep Monitoring, and Passive Listening conditions was identical to the one depicted here. In Passive Listening blocks, one stimulation trial followed the next, without a response trial in between.

Figure 2

BOLD signal increases in mid‐STG. (A) Signal increases associated with passive listening to stimulation at a middle frequency (located between the low and high frequencies used in the attention tasks) compared to silent baseline, superimposed on an averaged anatomical image. (B) Signal increases associated with attention in silence (contrast AiS, shown in blue), attention to a single melody (contrast ATT, green), and selective attention to one of two melodies (contrast SEL, orange), compared to conditions in which participants passively listened to the same stimuli. While attending to auditory stimuli (contrasts ATT and SEL) increased the BOLD response predominantly in lateral STG, attention in silence (contrast AiS) additionally led to significant increases of the BOLD response in medial HG. Individual data (not shown) showed the same general pattern. Note: Activations outside superior temporal cortex (not contained within the white boxes) will be discussed in Figure 4. The numbers in the upper right corner of each image indicate the z‐coordinate of the depicted horizontal slice in Talairach space. All maps are thresholded at t > 3.62, P < 0.004 (the strictest single‐voxel threshold corresponding to q < 0.05 when FDR‐correction for multiple comparisons was applied to the maps).

Figure 3

BOLD signal increases in medial and lateral regions of auditory cortex. (A) Illustration of the manually drawn anatomical ROIs on mHG (thought to encompass primary auditory cortex, shown in yellow) and lateral planum temporale (latPT, thought to include nonprimary auditory belt and parabelt areas, shown in dark red). The numbers next to the image are the Talairach coordinates of the crosshairs intersection, the numbers below indicate the coordinates of the geometrical center and the size of the ROI. (B) BOLD signal changes associated with the different contrasts for the medial and lateral ROIs (left and right hemispheres combined). Identically colored bars represent conditions with identical stimulation (blue, silence; green, single melody; orange, two simultaneous melodies). In conditions AiS, att, and sel, participants performed an attention task, in conditions pass and (L+H), as well as for the silent baseline, they rested. Error bars represent ± 1 standard error (SE) around the group mean. Note that each bar, with the exception of the one for (L+H), is based on the sum of two conditions (one with stimulation/attention at low and one with stimulation/attention at high frequencies). To balance the comparison, condition (L+H) thus received a weight of 2. All conditions evoked significant BOLD signal increases compared to silent baseline in both ROIs, as did the attention‐related difference between resting and listening in silence (contrast AiS). In contrast, the difference between passive and attentive listening to auditory stimuli (contrasts ATT and SEL) only reached significance in latPT, but not mHG when applying the same threshold as used for the whole‐brain analyses (P < 0.004, corresponding to FDR‐corrected q < 0.05 or stricter). However, at an uncorrected threshold of P < 0.05, both contrasts ATT and SEL would be deemed significant in both ROIs. Thus, while attentional modulation of sound‐evoked activation in primary auditory cortex may not be completely absent, it is clearly weaker than in nonprimary, more lateral belt/parabelt regions of auditory cortex, whereas attention in silence affected both primary and nonprimary auditory cortex alike (see also Fig. 6).

Figure 4

Attention‐related BOLD signal increases outside mid‐STG, superimposed on averaged transverse and coronal views of the brain. The numbers in the upper right corner of each panel indicate the coordinates of the crosshairs intersection in Talairach space. All maps are thresholded at t > 3.62, P < 0.004 (corresponding to or stricter than FDR‐corrected q < 0.05). In addition to the effects observed in mid‐STG (shown in Fig. 2), significant BOLD response increases for attention to auditory stimulation (contrast ATT, shown in green, and contrast SEL, shown in orange) were also present in right anterior STG (A), at the left TPJ (B), right IPL (C), bilaterally in the putamen, with activation spreading into the caudate in the right hemisphere (D), in SMA and pre‐SMA (E), bilaterally on the precentral gyrus (F), along posterior IFG (BA44, see G), and at the intersection of IFG/antIns (H). Notably, none of these areas displayed an increased BOLD response for attention in silence (contrast AiS, blue). For this condition, BOLD response increases were confined to STG. The same pattern was evident in individual data (not shown).

Figure 5

Average BOLD signal changes in all ROIs outside mid‐STG (for those, see Fig. 3). Conventions are as in Figure 3. Triangles indicate those contrasts whose significance at an FDR‐corrected threshold of q < 0.05 in a whole‐brain analysis contributed voxels to the ROI (for a more detailed description on how the ROIs were identified, see section “Regions of interest”). Note that each bar, with the exception of the one for (L+H), is based on the sum of two conditions (one with stimulation/attention at low and one with stimulation/attention at high frequencies). To balance the comparison, condition (L+H) thus received a weight of 2. Beneath the name of each ROI, the Talairach coordinates of the ROI's activation peak for contrast SEL and the size of the ROI are given.

Figure 6

Direct comparison of the average BOLD signal changes associated with the three attention tasks, for all ROIs. Conventions as in Figure 5.

Figure 7

Correlations between the percentage of errors made in the selective attention task and the BOLD signal increases relative to silent baseline in the selective attention task (gray) and the simple attention task (black).