On the screen, each subject’s initial position in the 2 by 2 grid was displayed for each subject (red or green circle). The subjects’ task was to move their own position to the same position as the other subject (hereafter, ‘partner’) after the movement. Since they did not know the partner’s initial position with each other, they were allowed to create and send a message to the partner. Each subject was able to create his/her own message as he/she liked, by choosing two symbol marks out of five possibilities (•, ■, ♦, +, and blank, see Fig. 1 for examples of the message). The subjects tried to encode information about their initial position as well as intended destination position by their own rule. They also tried to decode information about the partner’s intention from the message. After the message exchange, the subjects were asked to make their own movement, followed by a screen that displayed both subjects’ initial positions and destinations. Successful movement to the same position was rewarded with 1 game point to both subjects. They repeated 24 trials of this movement task. More detailed information about structure and rationale of this task was reported elsewhere [3].

By using this task, we tried to examine how the subjects’ movement planning in each trial was influenced by their own and partner’s past behaviours in a course of trials during which a symbolic communication system across the two subjects was gradually developed (see Konno et al. [3] for basic profiles of the development of across-subject communication protocols during the task). For this purpose, we analysed whether the subjects’ decision of their destination position in each trial was the same as that by their own or the partner’s destination in the previous trial, as a function of numbers of repetition of the same messages they sent or received.

Specifically, for each trial from the second to the 24th trial, we compared the subjects’ destination position of that trial with their destination in the previous trial where they created and sent exactly the same message to the partner as in the current trial. We calculated percentages of trials in which they moved to the same destination as their previous destination, as a function of repetition of trials where they created and sent the same message to the partner. These data were used to index planning based on information about one’s own past behaviour (planning of a movement based on information of a past message and movement the subject made). For a comparison measure to this index, we also calculated percentages of trials in which the subjects moved to the same destination as the previous partner’s destination when they sent the same message to the partner.

Conversely, to evaluate planning based on other’s past behaviour, we compared the subjects’ destination position of each trial with the partner’s destination in the previous trial where the subjects received exactly the same message from the partner as in the current trial. We calculated percentages of trials where they moved to the same destination as the partner’s past destination, as a function of repetition of trials where the same message was received from the partner. This data indexed planning of a movement based on information of a past message and movement that the partner made. Again for a comparison measure to this index, we also calculated percentages of trials in which the subjects moved to the same destination as the previous self destination when they received the same message from the partner. By these analyses, we tried to quantify the subjects’ tendencies of planning based on the past task behaviour by themselves and that by others.