One way to characterize the special relation that one has to one's own body is to say that only one's body appears to one from the inside. Although widely accepted, the nature of this specific experiential mode of presentation of the body is rarely spelled out. Most definitions amount to little more than lists of the various body !"#$%&"'()*+()++ + .+ basis of the sense of bodily posture, or proprioception, is well known. It includes muscle spindles, which are sensitive to muscle stretch, Golgi tendon organs, which are sensitive to tendon tension, and joint receptors, which are sensitive to joint position. The question then arises: how does one obtain reliable proprioceptive information about posture and movement on the basis of raw signals about muscle stretch, tendon tension, and joint angle? But this is not the only puzzle. One also needs to derive full-fledged bodily experiences from proprioceptive information. The body senses then suffer from two main problems: the limits of their reliability and the limits of their informational scope. When you lie in bed at night without moving at all, you may sometimes feel unsure of the exact location of your limbs. Proprioceptive signals are weak when one is not moving (Rossetti et al., 1994; van Beers et al., 1998, 1999; Helms Tillery et al., 1991). Proprioception is then only of limited accuracy; even a few minutes of complete stillness can make you lose your body, so to speak. In order to be the most reliable about the location of one's limbs, one must be moving. But even then the accuracy of proprioception is limited. More precisely, it decreases with the number of joint angles that must be computed. The more distal the body part, the more complex the computations are. For example, information about the location of one's fingertip in space requires taking into account information from many receptors on many joints, muscles and tendons, each of them sending noisy signals, the noise increasing in proportion to the degree to which the body part is distal. The second problem with the body senses is that they do not directly carry information about the shape of the various parts of the body, their size, and their spatial configuration. For example, the facts that we have two arms, that they are cylinder-shaped, that they are of a certain length, and that they are connected to the !"#$%&"'()*+()++ + /+ torso on one end and to the hands on the other end, cannot be easily derived from the body senses. It is true that one receives somatosensory feedback when acting, which can carry information about the size of the limbs. But this can give only a rough estimate. In particular, to know how far one can reach with one's hand does not indicate the respective size of one's fingers, palm, forearm and upper arm. Active exploration of each body part by haptic touch seems to fare better and to be more specific. However, this involves complex tactile-proprioceptive processing, and that in turn requires taking into account the size of the exploratory body parts (e.g., fingers). Hence, ...