Principles of brain function can be disclosed by studying their limits during performance. Tactile stimuli with near-threshold intensities have been used to assess features of somatosensory processing. When stimulating fingers of one hand using near-threshold intensities, localization errors are observed that deviate significantly from responses obtained by guessing – incorrectly located stimuli are attributed more often to fingers neighbouring the stimulated one than to more distant fingers. Two hypotheses to explain the findings are proposed. The ‘central hypothesis’ posits that the degree of overlap of cortical tactile representations depends on stimulus intensity, with representations less separated for near-threshold stimuli than for suprathreshold stimuli. The ‘peripheral hypothesis’ assumes that systematic mislocalizations are due to activation of different sets of skin receptors with specific thresholds. The present experiments were designed to decide between the two hypotheses. Taking advantage of the frequency tuning of somatosensory receptors, their contribution to systematic misclocalizations was studied. In the first experiment, mislocalization profiles were investigated using vibratory stimuli with frequencies of 10, 20 and 100 Hz. Unambiguous mislocalization effects were only obtained for the 10-Hz stimulation, precluding the involvement of Pacinian corpuscles in systematic mislocalization. In the second experiment, Pacinian corpuscles were functionally eliminated by applying a constant 100-Hz vibratory masking stimulus together with near-threshold pulses. Despite masking, systematic mislocation patterns were observed rendering the involvement of Pacinian corpuscles unlikely. The results of both experiments are in favor of the ‘central hypothesis’ assuming that the extent of overlap in somatosensory representations is modulated by stimulus intensity.