Adopting the intentional stance interferes with social attention when interacting with a social robot

Joint attention is a pivotal mechanism of human social cognition. Traditionally, joint attention has been experimentally investigated by means of the gaze cueing task in screen-based settings. Recently, the gaze cueing task has been implemented in more naturalistic and interactive settings with the embodied iCub robot, allowing for the investigation of concurrent factors influencing joint attention. Previous studies found that a communicative gaze before the gaze cue (i.e., mutual or avoiding gaze with the participants) influenced the subsequent joint attention by eliciting the gaze-cueing effect only in the mutual gaze condition. In the present work, we first manipulated the attribution of the intentional stance toward the robot by letting participants share a familiar context. Subsequent to this shared experience, we integrated the gaze-cueing paradigm with the iCub robot, such that the robot cued participants’ attention after establishing either mutual or avoiding gaze with them. In contrast to previous studies where the adoption of the intentional stance was not manipulated, our current results show that when the robot was perceived as an intentional agent, a gaze-cueing effect was present for both mutual and avoiding gaze, thereby suggesting that participants might have interpreted both gaze conditions as intentional (and, thus as communicative signals).