Overview of the present study

The study consisted of four intention recognition tasks, each combining one type of target (social vs. non-social intentions) with one kind of scope (motor vs. superordinate intentions). In each of the 4 tasks, participants were instructed to watch video sequences of intentional actions performed by either a single actor in isolation (non-social tasks) or by two interacting agents (social tasks), and to identify the intention guiding the observed action. As previously mentioned, two factors were varied over the sequences: (1) the visual information conveyed by the actor’s kinematics, and (2) expectations about the most likely intention that participants had prior to seeing the action scene.

Regarding the first factor, 4 amounts of visual information have been selected (see Chambon et al., xxxx, Appendix Part A, for the selection and control of these amounts) : very high, high, moderate, and low, corresponding to varying films durations, ranging from {1880, 1640, 1560, 1480} ms after the movement onset, respectively.

Regarding the second factor, each experiment was divided into two sessions: a baseline session in which all intentions had the same probability to be achieved, and a bias session in which the probability of one particular intention (preferred intention) was increased (55%) at the expense of the others (22 % each). Each session was in turn divided into two interleaved phases:

1. a learning phase (overt blocks) in which the action scenes were shown with a very high amount of visual information, and equal (baseline) or varying (bias session) probabilities of occurrence for each intention represented.
2. A test phase (covertblocks) which allowed one to assess the strength of the current bias by varying (low, moderate, and high) amounts of visual information. In this phase, each intention had equal probability to occur (see figure 1).

Fig. 1. Experimental design: Examples of a typical experimental sequence (one overt block followed by one covert block) used in both the baseline and the bias sessions. Overt blocks: movies with a very high and constant amount of visual information. Covert blocks: movies with lower and varying amounts of visual information. In the 4 experiments, the probability of all intentions was held equal across the block, except in overt blocks of the bias session, where one particular intention had a greater probability to be accomplished than the other ones (here, the “rotate” action in red). Labels: Ov: Overt blocks; Co: Covert blocks; L: “lift” action; R: “rotate” action; T: “transport” action.

All movies were filmed using a digital camera (Sony®- HDR-SR7) and were acquired and tailored using the software Adobe Premiere®. The movies were presented on a computer monitor (IIYAMA® 19”) at a distance of 60 cm from the participant. Finally, tasks modeling intentions with the same scope or target were controlled to have close visual properties in order to allow direct comparisons between them.

Material and procedure were the same as those used in Chambon et al. (xxxx).