About ten years ago the discovery of mirror neuron systems (MN systems) in both human and nonhuman primates (Rizzolatti et al. 1996; Gallese et al. 1996) initiated a revolution in our philosophical views of the relations between perception and action. The particularity of MNs is that they are activated during the performance of transitive actions (such as grasping, holding, bringing something to the mouth) as well as during the observation of other individuals performing the same or similar actions. Actions performed and perceived are said to be ‘broadly’ or ‘strictly’ congruent depending on the similarity of goals, objects, movements (Rizzolatti and Sinigaglia 2006). What this revolution seems to highlight is the centrality of action for social understanding. However, one of the most debated problems in the philosophy of action is whether MNs have simply a triggering and preparatory function towards action, or if they also play a role in the recognition and understanding of the actions and emotions of other individuals.

According to Gallese (2005) mirror neurons help an observer to understand the actions of other individuals by allowing her to form a motor representation that “simulates” the action. Much like the imitation ability in human infants, the simulation capacity is for the adult the basis for social understanding and possibly for cooperative joint action (Pacherie and Dokic 2006). This form of agent-neutral representation is generally interpreted as a simulation of the action (even a potential action) on the part of the observer. Potential actions were already known to scientists since 1988 (Rizzolatti and Sinigaglia 2006); but these kind of potential actions are evoked in the ordinary neurons of the visuo-motor area by the sight of targets and objects. What mirror neurons add to the picture is the eliciting of potential actions by action performed by others.

The aim of the paper is to provide a philosophical account of what is required if MNs are to play a clear cognitive role in social understanding. To bridge the gap between motor competence (based on MNs and other motor neurons) and action understanding (whether this is seen as triggering, simulating, or even mimicking) we need a theory of the “minimal units” or “constituent building blocks” of our agentive competence both as agents and observers. I propose that the notion of an individual’s “action repertoire” constituted of these “minimal units” can provide an answer to many problems in the MN research program.

The “action repertoire” of an individual is the system of all and only actions (types) the individual is able to perform and can be said to be competent on. An individual can recognize that a particular action is being performed (by an individual other than him) only if the action is of a kind which belongs in her “action repertoire”. The action repertoire is therefore the framework within which all understanding and expectation can take place. A system is not a bare set of items: it is what shapes our experience by determining the possible contents of our expectations. The system is thus constituted of “action units” that represents the types of action the individual is able to perform and recognize as (being) performed by other individuals. In order to underpin the notion of the ultimate constituents of an individual’s agentive competence I offer an argument against reducing genuine actions into sub-parts such as internal tryings or peripheral movements.

We know that mirror neurons are also “phase-sensitive” in the sense that they are differently activated during the various temporal phases of the action performed. This phase-sensitivity, however, in no way undermines the unity represented by the action, say, of grasping a cup as a whole. Action phases, in other words, are never taken to be actions or sub-actions in their right. They cannot be segregated without ceasing to make a “sense” to the observer and the agent as well. Therefore phases of actions are not actions by any point of view. What they lack is precisely the unity which goals confer on genuine actions. But the phase-sensitivity conspires with this “repertoire of actions” (the ultimate constituents) to produce expectations that are sufficiently determinate and generally correct.