2.2.2. Procedure

A. Motor experiment: Each trial started with the onset of an incomplete movie showing an actor performing a basic action on one of the three available cubes. As soon as the sequence ended, participants were instructed to infer the intention underlying the actor’s action by pressing, as quickly and as accurately as possible, one of the three corresponding keyboard buttons (T for ‘transport’, L for ‘lift’, or R for ‘rotate’). In the bias session, the type of motor intention the probability of which was increased (i.e. preferred intention) was counterbalanced across participants. Finally, each action was performed equally often with all objects (cube 1, 2, or 3).

Fig. 2. Examples of stimuli for the motor non-social intention experiment (A); the super-ordinate non-social intention experiment (B); the motor social intention experiment (C) and the superordinate social intention experiment (D).

B. Superordinate experiment: Each trial started with the onset of an incomplete movie showing an actor performing a series of three successive actions on the three cubes available to him. Each individual sequence of actions (i.e. each trial) led to building of a final shape composed of those three cubes, and situated in the centre of the scene. Participants were asked to identify the action performed on the last cube, given the final shape being currently arranged. Commutative sequences were used so that each shape could be constructed from distinct sequences of actions (e.g. the shape S1 could be obtained from the sequences ‘lift-lift-rotate’ and ‘lift-rotate-lift’). Identifying the last action of the sequence therefore required having previously inferred which superordinate intention (i.e. which final shape) this action contributes to achieve. As previously, participants were instructed to indicate what the not-yet completed last action of the sequence was, by pressing as quickly and as accurately as possible,one of the three corresponding keyboard buttons (T, L, or R).

In the bias session, the probability of a particular shape (S1, S2, or S3) was increased at the expense of the two others, while keeping equal the probability of each basic action contributing to building of that shape. The type of the biased shape was counterbalanced across participants.