(A) Rubber hand illusion. Participants sit in front of a table with their left hand hidden under the table and a fake limb (white detached arm in figure) placed on the table in front of them. If the fake limb is visibly stroked (schematic red brush in figure) together with the real limb (not visible), participants experience the illusion that the touch is referred to the fake limb (illusion) and that their real limb posture shits toward the fake limb (proprioceptive drift). The size of such effects is greater if the touches on the real and fake limb are delivered synchronously (right panel, light blue columns) than asynchronous (right panel, green columns) (Redrawn from the original data of: Botvinick and Cohen, 1998). (B) Mirror box illusion. Participant execute right-hand movements while the left arm is hidden from view and kept still inside a box, the right wall of which is replaced by a mirror (Ramachandran et al., 1995). As compared with a no-mirror condition (left panel, green columns), the mirror reflection of the right hand mimics the movements of the left hand inside the mirror box, biasing the participants feeling (assessed through a questionnaire) of ownership (Question 1: “The reflection in the mirror looks like the hand behind the mirror”, left panel, left light blue column) and inducing the illusion of apparent movement (or a true, involuntary, unconscious movement) of the hand inside the box (Question 2: It seems as though the hand behind the mirror is moving; left panel right light blue column) (Romano et al., 2013). (C) Crossmodal effects induced by robotic hand training. Prolonged use of an electromyography-driven, detached robot hand (drawn in gray on the right side of the table) providing sensory feedback referred to the participant’s arm (white circles), increased the interference from visual distracter leds located near the robot hand fingers (reddish shadowed circles) tested with the crossmodal congruency paradigm (right panel, light blue columns), as compared to the pre-training assessment (right panel, green columns). This pattern of results suggests a training-dependent expansion of crossmodal integration properties, typical of the peripersonal space near the body, to the space surrounding the robot hand (Marini et al., 2014).