Abstract
The brain computes the spatiotopic position of touch by integrating tactile and proprioceptive signals (i.e., tactile remapping). While it is often assumed that the spatiotopic touch location is mapped into extrinsic, limb-independent coordinates, an alternative view proposes that touch is remapped into intrinsic, limb-specific coordinates. To test between these hypotheses, we used electroencephalography (EEG) and a novel tactile stimulation paradigm in which participants (N = 20, 19 females) received touch on their hands positioned at various locations relative to the body. Previous findings suggest that neural activity in primate sensorimotor and parietal regions monotonically encodes limb position, with their sustained firing rates increasing or decreasing across the workspace. These amplitude gradients, detectable at the population level in somatosensory evoked potentials, can be used to test predictions from each spatiotopic coding scheme. If touch is coded extrinsically, neural gradients should reflect changes of the external stimulus location, regardless of the limb. If coded intrinsically, gradients should be tied to the position of each limb and mirror each other between hands. Both univariate and multivariate EEG analyses found no evidence for extrinsic coding. Instead, we observed neural signatures of limb-specific, intrinsic spatiotopic coding, with the earliest emerging ∼160 ms after touch in centroparietal channels, later shifting to frontotemporal and parieto-occipital channels. Furthermore, a population-based neural network model of tactile remapping successfully reproduced the observed gradient patterns. These results show that the human brain localizes touch using an intrinsic, limb-specific spatial code, challenging the dominant assumption of extrinsic encoding in tactile remapping.