Proprioceptive Feedback Plays a Key Role in Self-Other Differentiation

How do humans know whether the hand in front of their sight belongs to themselves? The question concerning the development of self-other differentiation remains one of the fundamental problems before we can truly understand and simulate the cognitive process of human social behaviors. Opposing to the traditional associative sequence learning models, our proposed model adds a closed loop of the proprioceptive perception of an agent, which conceptually simulates the imaginary body scheme. During a learning phase, this simulated body representation is corrected by the feedback of the actual sensation of the agent. Therefore, after learning, the agent becomes to be able to visually distinguish self-produced actions from others' even without proprioceptive information. This paper presents how the utilization of predicted proprioceptive feedback enables the agent to better differentiate the self from others.