[Paper Review] A Robot by Any Other Frame: Framing and Behaviour Influence Mind Perception in Virtual but not Real-World Environments

Sebastian Wallkötter, Rebecca Stower+2 more|arXiv (Cornell University)|Apr 16, 2020

Social Robot Interaction and HRI57 references4 citations

TL;DR

This study investigates how framing (how a robot is introduced) and social behavior influence mind perception and moral concern in human-robot interaction. While online experiments showed that both framing and behavior significantly affect mind perception, these effects failed to replicate in a real-world experiment, suggesting that real-world interactions involve more confounding factors, making mind perception harder to manipulate than in isolated virtual settings.

ABSTRACT

Mind perception in robots has been an understudied construct in human-robot interaction (HRI) compared to similar concepts such as anthropomorphism and the intentional stance. In a series of three experiments, we identify two factors that could potentially influence mind perception and moral concern in robots: how the robot is introduced (framing), and how the robot acts (social behaviour). In the first two online experiments, we show that both framing and behaviour independently influence participants' mind perception. However, when we combined both variables in the following real-world experiment, these effects failed to replicate. We hence identify a third factor post-hoc: the online versus real-world nature of the interactions. After analysing potential confounds, we tentatively suggest that mind perception is harder to influence in real-world experiments, as manipulations are harder to isolate compared to virtual experiments, which only provide a slice of the interaction.

Motivation & Objective

• To investigate how framing (introduction context) and social behavior influence mind perception and moral concern in robots.
• To compare the effects of these factors in virtual versus real-world human-robot interaction settings.
• To explore whether mind perception mediates the relationship between anthropomorphism and moral concern toward robots.
• To identify why experimental effects observed in virtual settings may fail to replicate in physical interactions.
• To contribute to ethical HRI by understanding the conditions under which robots are perceived as having agency and moral standing.

Proposed method

• Conducted two online experiments using Amazon Mechanical Turk (AMT) to test the effects of framing (high vs. low mind perception) and social behavior (social vs. non-social robot videos).
• Performed a real-world experiment with a NAO robot in a physical setting, manipulating framing and behavior in a controlled interaction (e.g., Tower of Hanoi task).
• Used standardized scales to measure mind perception, anthropomorphism, and moral concern toward robots.
• Applied one-way and two-way ANOVAs to analyze the effects of framing, behavior, and experimental condition (virtual vs. physical) on mind perception and moral concern.
• Conducted post-hoc analyses to explore potential confounds and interactions, particularly focusing on the discrepancy between virtual and real-world results.
• Used effect size and power analysis to assess the robustness of findings, especially in the real-world condition.

Experimental results

Research questions

• RQ1 Does framing the robot with high mind perception increase participants’ perception of the robot’s mental states?
• RQ2 Does observing a robot displaying social behaviors (e.g., speech, gestures) increase mind perception compared to non-social behavior?
• RQ3 Do the effects of framing and behavior on mind perception replicate in a real-world interaction setting?
• RQ4 Is there a significant correlation between mind perception, anthropomorphism, and moral concern toward robots?
• RQ5 What factors explain the failure of framing and behavior effects to replicate in real-world settings compared to virtual ones?

Key findings

• In online Experiment 1, framing significantly influenced mind perception (F(3,173) = 4.691, p = .004), with high-mind framing increasing perceived agency.
• In online Experiment 2, social behavior significantly increased mind perception (F(3,157) = 5.421, p = .001), with socially behaving robots perceived as more mentally capable.
• In the real-world experiment, neither framing nor behavior showed significant main effects on mind perception, and the interaction was non-significant (p = .069).
• The main effect of experimental condition (virtual vs. physical) was significant (p = .012), suggesting that physical embodiment may lead to higher baseline mind perception.
• The effect size for framing and behavior was notably smaller in the real-world setting, indicating reduced manipulability of these factors in physical interactions.
• Post-hoc analysis suggests that real-world interactions involve more uncontrolled confounds (e.g., intergroup dynamics), reducing the detectability of isolated manipulations compared to virtual settings.