Introduction
Imitation of a perceived, novel, action requires the action to be sufficiently “understood” in order to produce the appropriate sequence of motor commands resulting in the execution of a similar action. One way to “understand” a complex perceived motor action is to break it down into its elementary components. These components are the key: momentary positions, orientations, movements and directions, which, taken together, uniquely define the action. This seems to be the preferred way for the brain to process visual stimuli, not only in the primary visual cortex, but also in higher-level visual areas.
The literal characteristics of the observed action are not the only subject of imitation. The intentions of an agent performing an action, or the goals of the action, may also be imitated. When the intention or goal of an action is imitated, the precise movements of the imitator may differ from the model although the results of the imitated action may be the same.
The way in which a perceived action is defined depends on the perspective employed. In principle, an action can be defined from a viewer-centered, an object-centered or a goal-centered perspective (Perrett et al., 1989). A viewer-centered description defines the object or action with respect to the observer (e.g., “that person's arm moved towards me”). For object-centered descriptions (Marr & Nishihara, 1978), the principal axis of the object is taken as the basis for the reference (“that person's arm moved to a position in front of the chest”).