This study consists of a series of experiments in which we try to identify core characteristics that determine whether we attribute a self to other agents. We will investigate causality, speed, equifinality, behavioural efficiency, learning and context sensitivity. We manipulate the behaviour of small robots to either suggest the presence (C+) or absence (C-) of these characteristics.
In our first experiment we tried manipulating causality showing a robot colliding with and moving a white box for C+ and stopping at the point of collision without moving the box for C-. We then asked participants to rate both robots with our own manipulation check scale, the Mind Attribution (MAS), Godspeed (GS) and Robotic Social Attribution Scales (RoSAS).
A 2x3 ANOVA with characteristic (levels are the six characteristics of interest) and robot (C+ vs. C-) as within subject factors and following pairwise t-test revealed significantly higher ratings for causality in C+, however we also found significantly higher ratings for context sensitivity in C-. A 2x3 ANOVA with subscales (levels are the subscales of MAS, GS and RoSAS) and robot (C+ vs. C-) as within subject factors and following pairwise t-tests showed significantly higher ratings for agency, experience, likeability, perceived intelligence and perceived safety ratings for the C- robot.
The data from the first experiment do not suggest that causality has an effect on mind-attribution to other entities, but there is evidence for an effect of context sensitivity.