Reconstructing neural representations of tactile space

Abstract

Psychophysical experiments have demonstrated large and highly systematic perceptual distortions of tactile space. We investigated the neural basis of tactile space by analyzing activity patterns induced by tactile stimulation of nine points on a 3 x 3 square grid on the hand dorsum using functional magnetic resonance (fMRI). We used a searchlight approach within pre-defined regions of interests (ROIs) to compute the pairwise Euclidean distances between the activity patterns elicited by tactile stimulation. Then, we used multidimensional scaling (MDS) to reconstruct tactile space at the neural level and compare it with skin space at the perceptual level. Our reconstructions of the shape of skin space in contralateral primary somatosensory (SI) and motor (M1) cortices reveal that it is distorted in a way that matches the perceptual shape of skin space. This suggests that early sensorimotor areas are critical to processing tactile space perception.\n\nSignificant StatementHere, we show that the primary somatosensory (SI) and motor (M1) cortices, rather than higher-level brain areas, are critical to estimating distances between tactile stimuli on the hand dorsum. By combining functional magnetic resonance (fMRI), Procrustes alignment, and multidimensional scaling, we reconstructed the shape of skin space in the brain. Strikingly, the shape of the skin that we reconstructed from neural data matches the distortions we found at the behavioral level, providing strong evidence that early sensorimotor areas are critical for the construction of tactile space. Our work therefore supports the view that tactile distance perception is computed at lower level in the somatosensory system than is usually supposed.