Social robots designed as companions, assistants, and teachers might interact more effectively with humans if perceived as socially intelligent. We sought to determine the factors that underlie the Perceived Social Intelligence (PSI) of robots. We hypothesized four factors: social presentation, emotion, cognition, and behavior. Two hundred ninety-six MTurk workers viewed five videos of people interacting with robots and rated each robot on the PSI Scales. These scales measure overall social intelligence; the abilities to (1) recognize, (2) adapt to, and (3) predict human (a) emotions, (b) behaviors, and (c) cognitions; to identify humans, individuals, and social groups; and to present oneself as a desirable social partner. Based upon the screen test, parallel analysis, and MAP test, we extracted three factors. We examined several rotations and selected direct oblimin with delta -1 as optimal because it came closest to the ideal of simple structure. Factor 1, Theory of Mind, measures a robot’s ability to recognize, predict, and adapt to human emotions and cognitions. Factor 2, Approachable Disposition, measures the extent to which the robot was perceived as being friendly, trustworthy, caring, and helpful and not conceited, hostile, and rude. Factor 3, Reasoning about Behavior, measures the ability to recognize, predict, and adapt to human behavior. In future research, the first factor might divide into two (as originally predicted) if the selected robots were skilled with emotions but not skilled with cognitions. The number of factors is important because roboticists attempting to design better robots may want to focus on one factor.

CHS: Small: Socially-Aware Navigation, National Science Foundation PI: David Feil-Seifer, co-PI: Monica Nicolescu, Amount: $500,000, Sept. 1, 2017 - May 31, 2022

REU Site: Collaborative Human-Robot Interaction, National Science Foundation PI: David Feil-Seifer, co-PI: Shamik Sengupta, Amount: $360,000, Feb. 1, 2018 - Jan. 31, 2022