Being oneself

This paper discusses the difference between self-identification and the self/other differentiation. Self-identification relies on the congruence of self-generated movements and their expected consequences, i.e., on the temporal correlation between the set of signals (command signals, sensory reafferences) that are related to that movement. Only the origin of the movement (i.e., whether it originates from the self or not), not its destination or its goal, is relevant for self-identification. Self/other differentiation requires both that one can understand the actions others perform, and that one can attribute these actions to them. Understanding implies that a complete description of the actions of other agents, from their origin to their end, can be available in the brain of the observer. Attributing implies that the agent can be clearly differentiated from the self. The model for the self/other differentiation proposed here can therefore be described as a "rotate/simulate" model: I first rotate myself at the location of the person I observe in order to specify her location in space. Then, I simulate the action I observe from that person in order to understand what she is doing. Assuming that the two operations are more or less synchronous, the action I simulate is automatically attributed to the person I observe, not to myself. The simulation network in the observer's brain overlaps with the execution network in the agent's brain; by contrast, the rotation network is specific to the observer's brain. (C) 2007 Elsevier Ltd. All rights reserved.