Social offloading has been proposed as a process by which individuals engaged in joint action are able to selectively de-prioritise co-actor task information so as to enhance cognitive efficiency (Tufft and Richardson, 2020). Previous work using the picture-word interference paradigm has shown how distractor interference is removed when participants believed they were the targets of a co-acting partner. We build on these findings using an adapted version of the retroactive cuing (RAC) paradigm.

In the RAC task, participants are presented with two polygon shapes, one on the left and one on the right of a visual display. After 250ms, a central left or right pointing arrow cue is presented followed by a target polygon on the left or right side. Participants must judge if this target is the same or different as the preceding polygon at the same location. Improved performance is found when targets are presented at validly cued locations (same side). In our adapted version, central arrows were replaced by exogenous cues (red dots) presented on the left or right side. Participants were told these represented the gaze of a co-acting partner who they believed were working with them on the task. Remarkably, rather than improvement, we found reduced sensitivity to validly cued locations but only when participants believed their partner was competent or working collaboratively as team. We interpret these results as a further example of social offloading in which locations deemed to be gazed at by an appropriate partner are de-prioritised, supporting a division of labour.