Pervasive Memory: The future of long-term social HRI lies in the past

The technologies underlying long-term social Human-Robot Interaction (sHRI) continually advance, with impressive results. However, applications typically remain on relatively short time-scales, or if long-term, are more focussed on the human perspective than that of the robot. What are still missing are the broader theories of temporal organisation of robot social behaviour for long-term interaction, and more particularly, how experience informs the adaptivity of behaviour. Based on this necessity, we propose in this position paper that memory should play a more central role in accounts of sHRI. Specifically, we introduce the concept of "Pervasive Memory", i.e. a broader notion of memory than is typically used: memory as underlying and involved in all aspects of social behaviour, beyond the mere passive storage of (symbolic) semantic information. The working hypothesis is that by committing to such a perspective, an integrated and coherent approach to long-term sHRI can be formulated. A number of examples are described in this position paper, including studies currently under way, to support this approach. We suggest that, in this sense, the future development of social robotics really does lie in the past.