ABSTRACT
Real-world perspective-taking problems frequently involve interactions among individuals, suggesting a potential social element to this seemingly spatial problem. Previous studies have suggested that the agency of the target in a perspective-taking task might influence reasoning. This hypothesis is tested directly by manipulating whether one takes the perspective of a potential agent or an object. The results were striking: Even though no overall differences in performance were observed with and without agents, performance was differentially associated with social skills. In particular, participants with better social skills were more accurate than less social peers when the target was a potential agent, whereas no such relationship was observed when the target was an object. These results suggest that bringing domain-specific investigations to bear on real-world problems requires understanding how that domain exists in the broader context of interacting skills and biases.
