Two conditions were employed: (1) an autonomous ‘high
immediacy’ robot tutor [4], and (2) a human tutor (Fig. 1).
The robot tutor was designed to regularly gesture, look at the
child, make small body movements to appear ‘relaxed’, and
lean forwards. The human was given a word-by-word script to
match the lesson content of the robot, but was not constrained
in terms of social behaviour. Due to the script providing precise
lesson content (and the study focus on social behaviour and
embodiment differences) an expert tutor was not required. A
total of 22 children took part: 11 in the robot condition and
11 in the human (age M=8.8, SD=0.4; 12F, 10M). Interactions
lasted for M=14m05 (SD=3m16) in the robot condition, and
M=13m10 (SD=3m39) in the human condition.