A key question in neuroscience is how cortical organisation relates to experience. Previously we showed that amputees experiencing highly vivid phantom sensations maintain cortical representation of their missing hand (Kikkert et al., 2016). Here, we examined the role of sensory hand experience on persistent hand representation by studying individuals with acquired and congenital hand loss. We used representational similarity analysis in primary somatosensory and motor cortex during missing and intact hand movements. We found that key aspects of acquired amputees’ missing hand representation persisted, despite varying vividness of phantom sensations. In contrast, missing hand representation of congenital one-handers, who do not experience phantom sensations, was significantly reduced. Across acquired amputees, individuals’ reported motor control over their phantom hand positively correlated with the extent to which their somatosensory hand representation was normally organised. We conclude that once cortical organisation is formed, it is remarkably persistent, despite long-term attenuation of peripheral signals.

Scientists traditionally use passive stimulation to examine organisational properties of the primary somatosensory cortex (SI). Recent research has, however, emphasised the close and bidirectional relationship between somatosensory and motor systems. This suggests active contributions (e.g., direct inputs from the motor system to SI) should also be considered when studying SI representations. Under such a framework, discrepant results are possible when different tasks are used to study the same underlying somatosensory representation. Here we examined whether SI hand representation is consistent when compared between active and passive tasks. Moreover, we ask whether this holds when task demands and stimulus properties are not directly matched. Using 7T fMRI, we examined three key digit representation features -spatial location, activity gradient profiles and multivariate representational structure. We show that although activity was increased overall in the active task, all key features of digit somatotopy were consistent over tasks, using both traditional univariate (activity-level) and multivariate (pattern structure) measurements. Despite overall comparability, when investigating fine-grained features of somatotopy using multivariate analysis, the active task was superior for individuating the representation across digits. We suggest that the information produced by an active task is more aligned with typical, ecologically relevant patterns of hand sensory input, resulting in more optimal SI activity patterns. Our findings validate the utilisation of active tasks for studying SI somatotopy.