Effect of sensory modality and attention on layer-specific activations in sensory cortices

Introduction: Accumulating evidence suggests that audiovisual interactions are not deferred to higher order association areas, but start already at the primary cortical level. Moreover, visual stimuli have been shown to induce fMRI activations in the visual system, with deactivations in primary auditory areas, and vice versa for auditory stimuli. It is unknown whether these crossmodal deactivations are caused by top-down effects from association areas reflecting the withdrawal of attentional resources from the non-stimulated modality, or via direct connections between sensory areas. Previous neurophysiological studies in animals have demonstrated that visual and auditory stimuli induce activations in auditory cortices with distinct layer-dependent profiles (feedforward/granular vs. backward/infra+supragranular). This study used 7T fMRI to resolve BOLD responses at different cortical depths and characterize the effects of sensory stimulation and modality-specific attention on the cortical layer-specific activation profiles. Methods: Four participants took part in this fMRI experiment at 7T (MAGNETOM 7T). In a 3 (stimulation modality) X 2 (attention) design, they were presented with visual (V, concentric looming white circles on a black background), auditory (A, looming frequency-modulated pure tones) or audio-visual stimuli (AV). They attended and responded to targets either in the visual or auditory modality. Using a 24 channel phased array head coil (Nova Medical Inc, Wilmington MA, USA), we acquired a whole-brain T1 map (MP2RAGE (Marques et al. 2010); spatial resolution: (0.7 mm)3; TR=5000 ms; TE=2.45 ms; TI1/2=900 ms/2750 ms; FA1/2 = 5°/3°; iPAT=2) and 46 axial EPI slices with an axial coverage of 3.6 cm that included the entire primary visual and auditory cortices and much of the posterior superior temporal gyri (spatial resolution: (0.75 mm)3; TR=3000 ms; TE=25 ms; FA=90; iPAT=4). Primary and secondary visual cortices were identified by retinotopic mapping. Four regions of interest were defined including left and right Heschl's gyri, the planum temporale, the left/right superior temporal gyri and the left/right primary auditory cortices (i.e. TE 1.0 based on cytoarchitectonic probability maps). For cortical depth-dependent analysis, the cortex was segmented, upsampled to a (0.4 mm)3 resolution and automatically contoured into 6 laminae using the CBS Tools and MIPAV (Waehnert et al. 2013). Using SPM, the functional data were realigned, unwarped, coregistered to the up-sampled whole-brain T1 map and regridded to (0.4 mm)3 resolution. The subject-specific GLM analysis included 7 regressors, modeling the 6 conditions of the 3 X 2 design, target onsets and subject responses with a canonical HRF and its temporal derivative. Contrast images were computed for each of the 3 X 2 conditions. Parameter estimates for each contrast were averaged for each layer of each ROI and then across subjects. Results: Overall mean accuracy was above 90, assuring participant compliance with the attention modulation instructions. Preliminary fMRI analyses revealed: 1. Cross-modal deactivations: auditory stimuli induced deactivations in primary visual areas, as did visual stimuli in primary auditory areas. 2. Attention amplified the sensory evoked activations and deactivations in primary sensory areas. 3. Audiovisual interactions (i.e. AV ≠ A + V) were sub-additive. 4. The cortical profile showed an activation gradient, maximal at the cortical surface. Conclusions: The study replicated cross-modal deactivations operating from vision to audition and vice versa. Moreover, we demonstrate that these cross-modal deactivations are amplified when attention is directed to the sensory input. Further analyses will be needed to delineate how much attention and sensory modality modulate the layer-specific activation profiles