The functional specialisations of cortical sensory areas were traditionally viewed as being tied to specific modalities. A radically different emerging view is that the brain is organized by task rather than sensory modality, but it has not yet been shown that this applies to primary sensory cortices. Here we report such evidence by showing that primary ‘visual’ cortex can be adapted to map spatial locations of sound in blind humans who regularly perceive space through sound echoes. Specifically, we objectively quantify the similarity between measured stimulus maps for sound eccentricity and predicted stimulus maps for visual eccentricity in primary ‘visual’ cortex (using a probabilistic atlas based on cortical anatomy) to find that stimulus maps for sound in expert echolocators are directly comparable to those for vision in sighted people. Furthermore, the degree of this similarity is positively related with echolocation ability. We also rule out explanations based on top-down modulation of brain activity – e.g. through imagery. This result is clear evidence that task-specific organization can extend even to primary sensory cortices, and in this way is pivotal in our reinterpretation of the functional organisation of the human brain.
Data summary
Summary excel spreadsheet that contains the data that was entered into the multiple linear regression reported in the manuscript
retinotopic mapping summary data.xls
Matlab functions for data analysis
Custom/modified Matlab functions that were used in the cross-correlation analysis pipeline. These functions are designed to be used in conjunction with the mrTools analysis suite (Gardner Lab, Stanford University, USA). These functions are:
a. corAnal – designed to replace the function of the same name in the mrTools suite. This function includes a call to the following function:
b. computeCoranal –designed to replace the function of the same name in the mrTools suite. This function includes a call to the following function:
c. myXcorr – this is an entirely custom built function that carries out phase-encoded mapping on sparse sampling mri data in a way that is compatible with the mrTools analysis pipeline.
Matlab functions.zip
Subject mapping data
Neural mapping data and fitted probabilistic retinotopic maps for each individual subject. Each data folder corresponds to a single subject (e.g. EE1, SC2) – details on these subjects can be found in the manuscript and supplemental materials. Within each of these folders are the following data files:
• aseg.hdr and aseg.img
o header and image files
• ret_areas.hdr and ret_areas.img
o The probabilistic atlas for cortical visual areas fitted to this subject’s cortical anatomy.
o The .img file can be loaded into matlab using the function mlrImageReadNifti (Gardner Lab, Stanford University, USA)
o The file includes a 3-dimensional data matrix, in which each cell corresponds to a single voxel. Each of these cells includes a single value to denote which visual area (if any) that voxel is mostly likely to be part of. Please see https://hub.docker.com/r/nben/occipital_atlas/ for further details.
• ret_eccen.hdr and ret_eccen.img
o The probabilistic atlas for retinotopic eccentricity fitted to this subject’s cortical anatomy.
o The .img file can be loaded into matlab using the function mlrImageReadNifti (mrTools analysis suite, Justin Gardner)
o The file includes a 3-dimensional data matrix, in which each cell corresponds to a single voxel. Each of these cells includes a single value to denote which point in visual space (eccentricity) that voxel is mostly likely to represent. Values range from 0 (central) to +90 (most eccentric) in each hemisphere. Please see https://hub.docker.com/r/nben/occipital_atlas/ for further details.
• corAnal.mat
o Results of the cross-correlation analysis (described in the manuscript). This is a standard matlab .mat file, which was created using the mrTools analysis suite (Gardner Lab, Stanford University, USA). For proper exploration of these data we recommend installing the mrTools analysis package for Matlab.
o Within the corAnal data variable, field “overlays” is a 1 x 3 structure array. These 3 subfields correspond to the following data from the cross-correlation analysis
Coherence values
Amplitude values (unused in our analysis)
Phase values
o Within each of these subfields there are further subfields corresponding to the data for each of the stimulus conditions that the subject took part in (in the order echo, source, vision)
o Each of these data variables is a 3-dimensional data matrix, where each cell corresponds to a single voxel. The positions of these voxels correspond to those in the ret_areas, ret_eccen and aseg data files.