Social cognition in human-robot interaction

Research in human-robot interaction (HRI) has shown that people often interpret robot behavior in folk-psychological terms, such as beliefs, goals, and intentions.

Robot lawnmowers, for example, are often described as not wanting to run out of battery, as knowing where the charging station is, or as approaching it with the intention to recharge.

People do not necessarily think that robots actually have human-like mental states – but interpreting them this way has been shown to, for example, increase people’s trust in robots and help predict their behavior.We recently published a systematic review of 155 empirical studies of mental state attribution to robots in several scientific disciplines, including HRI, computer science, psychology, and neuroscience.

However, there are significant gaps in our scientific understanding.

Much research has addressed when people attribute mental states to robots, and why, but very little is known about the how, i.e., the mechanisms underlying such attributions.Our hypothesis is that there are three interacting mechanisms at play: projection (i.e., using one’s own human knowledge as a starting point), stereotyping (i.e., using knowledge about groups of people/robots), and interactive experience from encounters with specific robots.

We will empirically investigate the use of these mechanisms in human encounters with different types of robots and develop an integrated account of their interaction in the interpretation of robot behavior.