Autism is a lifelong pervasive disorder of social cognition associated in particular with difficulties in joint attention, ie sharing the focus of attention to an object of the world with another agent. Care strategies that are expected to use artificial agents to support the intensive early intervention program to rehabilitate key aspects of normal cognition in autism already focus on joint attention. Yet little is known on how early social attentional processes are impacted the use of humanoid robots in comparison to real humans. Here, we used of social version of Posner's spatial orienting task, developed to investigate how exogenous stimuli bias attention in space, to see whether human and robot upper torso posture presented centrally have similar effects on the orientation of attention to the side. In this experiment, participants respond to the appearance of a target on the left or right of the computer screen. Prior to the target, a spatial cue, in our case the human or robot torso looking left or right, is used to orient attention of the participant. Reaction time was significantly increased when a robot was used instead of a human spatial cue, suggesting it was harder to disengage attention from the robot than the human. We reproduced the congruency effect, with reduced reaction time when the spatial cue point to the side where the target later appears, with no interaction with the agent used as cue. While further work is required to clarify why our present results conflict with the existing literature with regards to human stimuli, they clearly support that a humanoid robot upper torso is at least as capable as a human stimulus to elicit the early orientation of attention investigated by the social version of Posner's task.
展开▼
