Autonomy and robotonomy: The role of gender

Abstract

This study examines how gender affects people's perception of autonomy in social robots. It investigates whether an agent's name (masculine/feminine/neutral/ technical/no-name) and type (human/robot) can impact agency, personification, competency, and gender evaluations and whether these evaluations are at an unconscious level. The study consists of a pre-study and a main study with 150 participants. Participants watched 4-second 18 videos in the main study and answered autonomy-related questions. They also completed three implicit association tasks, adapted according to the study's themes, autonomy, agents, and gender. ANOVA analysis was conducted to analyze the agent, naming, and gender effects for the explicit part. The d-score was calculated for all participants in the implicit part, and an ANOVA analysis was conducted. Regression analysis was conducted to determine gender attribution in the pre-study and main studies. Correlation analyses were also conducted to determine if explicit-implicit parts were correlated. Lastly, a thematic analysis was conducted for qualitative inputs in the explicit part and categorized into nine themes. The study found the main effect of action and agent in agency-level attribution. In competency and gender attribution, agents had no main effect in none of the name conditions. In the implicit part, women and men participants differed in men-independent/women-dependent association IAT-1. The other two IATs, women and men participants, responded similarly. The study suggests that name manipulation does not affect people's autonomy perception, but rather agent types and actions characteristics affect them. Furthermore, people's implicit and explicit answers do not predict each other.