Orienting auditory spatial attention engages frontal eye fields and medial occipital cortex in congenitally blind humans

Abstract

What happens in vision-related cortical areas when congenitally blind (CB) individuals orient attention to spatial locations? Previous neuroimaging of sighted individuals has found overlapping activation in a network of frontoparietal areas including frontal eye fields (FEF), during both overt (with eye movement) and covert (without eye movement) shifts of spatial attention. Since voluntary eye movement planning seems irrelevant in CB, their FEF neurons should be recruited for alternative functions if their attentional role in sighted individuals is only due to eye movement planning. Recent neuroimaging of the blind has also reported activation in medial occipital areas, normally associated with visual processing, during a diverse set of non-visual tasks, but their response to attentional shifts remains poorly understood. Here, we used event-related fMRI to explore FEF and medial occipital areas in CB individuals and sighted controls with eyes closed (SC) performing a covert attention orienting task with endogenous verbal cues and spatialized auditory targets. We found robust stimulus-locked FEF activation of all CB subjects, similar to and stronger than in SC, suggesting that FEF plays a role in endogenous orienting of covert spatial attention even in individuals in whom voluntary eye movements are irrelevant. We also found robust activation in bilateral medial occipital cortex in CB but not in SC subjects. The response decreased below baseline following endogenous verbal cues but increased following auditory targets, suggesting that the medial occipital area in CB does not directly engage during cued orienting of attention but may be recruited for processing of spatialized auditory targets.

Figure 1

(a) During the covert attention switching task, oral cue words and virtually spatialized auditory targets (see Methods for details) were delivered via headphones in the scanner. The scan was clustered with a 600 ms silent gap during which all the auditory stimuli were presented. (b) Response times and (c) Accuracy (% correct responses) for different conditions (C: centre cue and target; VL: valid left cue, left target; VR: valid right cue, right target; IVL: invalid left cue, right target; IVR: invalid right cue, left target). Responses to valid targets were significantly faster and more accurate than to invalid ones. Error bars indicate ± SEM.

Figure 2

(a) Group activation map of the blind subjects during covert attention switching trials superimposed on a 3D surface; white arrows indicate FEFs. (b) Coronal (y = −11) and axial (z= 49) views of FEF ROI (green areas in the left column), sighted (middle column), and blind (right column) group maps. Voxels activated by the covert attention switching task are shown in red/yellow with the green FEF areas superimposed to show overlap. All maps were created with whole brain random effects GLM and used cues > baseline contrast at FDR < 0.05 and a cluster minimum of 50 voxels; the legend bar shows the corresponding t(64) values for the CB case, p < 0.0039; for the SC group the corresponding t(72) ranged from 2.67 to 6.00, p < 0.0095).

Figure 3

Individual axial (z= 53) activation maps in and around the FEF region of all subjects performing the covert attention switching task. All (a) blind and (b) sighted subjects showed consistent activation around the intersection of superior frontal and precentral sulci bilaterally. (c) Axial slices detail of one blind subject (08JS_CB) shows activation aligned closely with precentral sulcus as it descends from its top (z = 57) to bottom (z = 47). All maps were generated using a cue > baseline contrast at FDR < 0.05 and a cluster minimum of 50 voxels; the legend color bar shows the corresponding t-values.

Figure 4

Event related response of the FEF region aligned to cue delivery, showing percent change over baseline in BOLD signal averaged across all voxels in: (a) left FEF in blind, (b) right FEF in blind, (c) left FEF in sighted, (d) right FEF in sighted. The cue was delivered at scan 0; target followed after 2 scans (only trials where targets arrived after 2 TRs or 4.4 seconds are shown here for visual clarity; targets after 1TR resulted in similar patterns - see Fig. 5 for the combined analysis of all targets). The baseline was computed by averaging the three scans before and including 0 separately for every cue epoch (both valid and invalid targets were combined). Responses to “centre” cue (and centre targets) shown with black lines and circles, to “left” cue shown with blue lines and triangles, and to “right” cue shown with red lines with squares. Responses to left and right cues are similar in both groups, but the responses to central cues (black lines with crosses) tend to differ between SC and CB. Error bars indicate ± SEM.

Figure 5

Event related response of the FEF region aligned to target delivery, showing percent change over baseline in signal averaged across all voxels for responses aligned to targets classified according to whether they followed central/valid/invalid cues and whether they were go/nogo; (a) left FEF in blind, (b) right FEF in blind, (c) left FEF in sighted, (d) right FEF in sighted. The target was delivered at scan labelled 0. The baseline was computed by averaging the three scans previous to and including scan 0 for every target epoch (the baseline thus incorporates the first part of the response to delivery of the preceding cue). Blue lines with triangles show “invalid go”, green lines with squares show “valid go”, red lines with filled circles show “valid nogo”, pink lines with diamonds show “invalid nogo” targets, and black lines with unfilled circles show responses to “centre” targets. Responses to most targets are similar in both groups, with response to invalid-go targets being the strongest; the exception is the response to central targets for which the blind response tended to be stronger than sighted. Error bars indicate ± SEM.

Figure 6

Group contrast maps showing Targets > Baseline (p(FDR) < 0.05, minimum cluster size > 50) along the medial occipital wall and calcarine sulcus of a) congenitally blind and sighted (b) subjects during covert attention switching trials with sagittal view (at x = 2) shown on the left and coronal view (at y = −81) on the right side. Voxels activate during the covert attention switching task are shown in red/yellow with the occipital ROI shown with a dotted outline. The white cross hairs are centred on the peak voxel (x = 2; y = −81; z = 5) on the lower bank of the calcarine sulcus.

Figure 7

Average event-related response from the medial occipital ROI (see Fig. 6) showing percent change over baseline. (a) The occipital responses aligned to cues for CB group are shown separately for targets that followed the cue by a delay of 1 TR (blue lines) or 2 TRs (green lines). Cue delivery is at scan 0. (b) The average response of the occipital region to targets after 2TRs, (green line), is compared to the corresponding response in the FEF region, shown (red line). The occipital response shows an initial decrease followed by a sharp rise after target delivery. The cue delivery is at scan 0 and the target at scan 2. (c–d) Occipital responses to targets are shown for the CB group in (c) and SC group in (d). The event-related response under different target conditions. Blue lines with triangles show “invalid go”, green lines with squares show “valid go”, red lines with filled circles show “valid no-go”, pink lines with diamonds show “invalid no-go” targets, and black lines with unfilled circles show responses to “centre” targets. The target was delivered at scan labelled 0. CB responses to both cues and targets were significantly stronger than SC responses and clearly differ for different kind of targets. Error bars indicate ± SEM.