2. Methods and results

The study consisted of four experiments, each combining one type of target (social vs. non-social) with one kind of scope (motor vs. superordinate).All four experiments had the same design with two distinct sessions (baseline and bias session) that differed in whether participants were, or were not, biased towards recognising one particular intention. In the baseline session, no bias was assigned – all intentions thus had equal probabilities –, whereas, in the bias session, one particular intention (preferred intention) was more likely to be achieved than the other ones (non-preferred intentions). Both sessions consisted of interleaved blocks (overt and covert blocks) that differed in the amount of visual information that was given to the subjects. Overt blocks were characterized by a very high amount of visual information that allowed participants to easily recognize the different intentions and therefore to integrate the effective probability distribution of the different intentions associated with each session. In the covert blocks, however, action sequences carried varying and lower amounts of visual information. This design allowed us to assess how the bias (i.e. the participants’ prior expectations) interacts with the amount of visual information during an intentional inference task.

Fig. 1. Examples of stimuli for the motor non-social intention experiment (A); the superordinate non-social intention experiment (B); the motor social intention experiment (C) and the superordinate social intention experiment (D).