Previous studies have shown that the sensory modality used to identify regions of the body hidden from sight, but frequently viewed, influences the type of the body representation employed for reaching them with the finger. The question then arises as to whether this observation also applies to body regions which are rarely, if ever, viewed. We used an established technique for pinpointing the type of body representation used for the spatial encoding of targets which consisted of assessing the effect of peripheral gaze fixation on the pointing accuracy. More precisely, an exteroceptive, visually dependent, body representation is thought to be used if gaze deviation induces a deviation of the pointing movement. Three light-emitting diodes (LEDs) were positioned at the participants’ eye level at -25 deg, 0 deg and +25 deg. Without moving the head, the participant fixated the lit LED before the experimenter indicated one of the three target head positions: topmost point of the head (vertex) and two other points located at the front and back of the head. These targets were either verbal-cued or tactile-cued and the participants had to reach them with their index finger. We analysed the accuracy of the movements directed to the topmost point of the head, which is a well-defined, yet out of view anatomical point. Based on the possibility of the brain to create visual representations of the body areas that remain out of view, we hypothesized that the position of the vertex is encoded using an exteroceptive body representation, both when verbally or tactile-cued. Results revealed that the pointing errors were biased in the opposite direction of gaze fixation for both verbal-cued and tactile-cued targets, suggesting the use of a vision-dependent exteroceptive body representation. The enhancement of the visual body representations by sensorimotor processes was suggested by the greater pointing accuracy when the vertex was identified by tactile stimulation compared to verbal instruction. Moreover, a control condition showed that participants were more accurate in indicating the position of their own vertex than the vertex of other people. Together, our results suggest that the position of rarely viewed body parts are spatially encoded by an exteroceptive body representation and that non-visual sensorimotor processes are involved in the constructing of this representation.