Abstract

The purpose of the present four experiments was to examine the proposal that joint attention behaviors reflect an understanding of people as intentional agents. An object-request interactive paradigm was designed to document infant's understanding of the referential cues and of the agent (i.e., a person or a humanoid robot). The two-phase paradigm examined infant's ability to follow and to infer desires from three different referential cues: (1) gaze direction (i.e., head and eyes); (2) gesture (i.e., a grasping towards the target toy); and (3) positive vocal affect. After infants' attention was directed towards the adult, the agent focused her (its) attention towards one of two objects by displaying one of the following: a one-cue combination (gaze or gesture), a two-cue combination (i.e., gaze and gesture, gaze and positive vocal affect, gesture and positive vocal affect) or a three-cue combination (gaze, gesture, and positive vocal affect). Following cue presentation, a female experimenter requested that the infant give a toy. The first experiment examined 14-, 18-, and 24-month-olds' ability to infer desires from a human agent's prior display of object-directed behaviors. Although all infants could follow the cues to the target toy, only the 18- and 24-month-old were able to use these cues to infer a person's desire. In Experiment 2, the position of the two toys was switched following cue presentation for 18- and 24-month-olds. This switch resulted in infants' poorer performance. The findings from Experiment 3 indicated that while 14-, 18-, and 24-month-olds could follow the cues of a humanoid robot they did not attribute desires to this agent at any age. Moreover, Experiment 4 demonstrated that these latter findings were not due to the robot's inability to request the target toy on its own behalf.