Presentation
Faces of Emotion: An Investigation into the Emotional Reactions to Machine-like and Human-like Robots
SessionRobotics Summit - Session 2
DescriptionIntroduction
Robots are rapidly being integrated into the hospital environment in a variety of ways. These robots can vary in size, shape, and form depending on their task altering the physical surface-level cues that individuals see. Examples of surface-level cues are claws, wheels, arms, or legs, for example (Haring et al., 2021). Because patients, providers, and additional hospital staff have the potential to be interacting with robots with little knowledge of their capabilities, these differences in physical characteristics serve as the only cue that some individuals may have prior to interaction with a robot. For example, if you have no knowledge of what a robot’s capabilities are, you could only rely on how they look to inform you of those potential capabilities.
Based on the heterogeneous nature of the robot surface-level cues, there is potential for individuals to vary their emotional reactions to seeing and ultimately working with these robots. The emotional reactions that one has have the potential to influence behaviors and interactions with said robot during hospital-related tasks (Hwang et al., 2013; Rosenthal-von der Pütten et al., 2013). Emotional reactions can be operationalized in a multitude of ways. In the present project, we focused on the list of “basic” emotional reactions of anger, disgust, fear, sadness, and happiness (Harmon-Jones et al., 2016). As such, we performed an investigation into the relationship between surface-level cues, how the robot looked, and how it relates to participant emotional reactions to their robot teammate.
Methods
To evaluate this relationship, a between-subject study was conducted with students at a large northeastern university (N = 210). Participants had to be at least 18 years old and be current students. This study had institutional review board approval.
We utilized The Discrete Emotions Questionnaire to measure participant emotional reactions to their robot counterpart (Harmon-Jones et al., 2016). The shortened 19-item measure utilized a 7-point Likert scale from “not at all” to “an extreme amount”. The selection of a larger number on the scale indicated a stronger emotional reaction.
First, participants were told that they were to complete an interdependent task with their robot teammate. Next, participants were randomly assigned to either a human-like robot or a machine-like robot condition and received an image of their teammate. After receiving their teammate’s photo and being provided with the associated background information they were asked to complete the emotional reaction questionnaire. All of these steps were taken prior to completion of their simulated task, thus, capturing their initial reactions to their robot counterpart.
Results
An independent sample t-test was run to evaluate the relationship between emotional reactions and robot surface-level cues. All assumptions associated with this analysis were reviewed prior to conducting any analyses. Fear was found to be a significant emotional reaction based on surface-level cues of the robot teammate (t(216) = -2.24, p < 0.05). Further, there was more fear reported within groups assigned to the human-like robot (µ = 1.84) than the machine-like robot (µ = 1.52). The other emotions, anger, disgust, sadness, and happiness, did not yield significance based on the surface-level cues of their robot teammate.
Conclusion
The findings from the present work indicate the strong influence of robot physical characteristics on participant emotional reactions. More specifically, there is a greater understanding of how participants who have not interacted with a robot emotionally perceive robots within simulated environments. As it relates to the healthcare domain, these results can support decision-making efforts for robot purchasing and design. For example, it may be beneficial, for novel users, to work with a machine-like robot first before working with a human-like robot.
Future work should investigate the relationship between individual differences and their influence on emotional reactions to robots based on their surface-level cues. This study should also be replicated with providers and patients to review whether these findings vary based on task. Some of the limitations of this work are that it was conducted with university students and did not consider prior experience with other robots as it relates to participant’s emotional reactions. Despite these limitations, this work contributes valuable insights into the initial emotional response participants have to robots with varying physical characteristics.
Robots are rapidly being integrated into the hospital environment in a variety of ways. These robots can vary in size, shape, and form depending on their task altering the physical surface-level cues that individuals see. Examples of surface-level cues are claws, wheels, arms, or legs, for example (Haring et al., 2021). Because patients, providers, and additional hospital staff have the potential to be interacting with robots with little knowledge of their capabilities, these differences in physical characteristics serve as the only cue that some individuals may have prior to interaction with a robot. For example, if you have no knowledge of what a robot’s capabilities are, you could only rely on how they look to inform you of those potential capabilities.
Based on the heterogeneous nature of the robot surface-level cues, there is potential for individuals to vary their emotional reactions to seeing and ultimately working with these robots. The emotional reactions that one has have the potential to influence behaviors and interactions with said robot during hospital-related tasks (Hwang et al., 2013; Rosenthal-von der Pütten et al., 2013). Emotional reactions can be operationalized in a multitude of ways. In the present project, we focused on the list of “basic” emotional reactions of anger, disgust, fear, sadness, and happiness (Harmon-Jones et al., 2016). As such, we performed an investigation into the relationship between surface-level cues, how the robot looked, and how it relates to participant emotional reactions to their robot teammate.
Methods
To evaluate this relationship, a between-subject study was conducted with students at a large northeastern university (N = 210). Participants had to be at least 18 years old and be current students. This study had institutional review board approval.
We utilized The Discrete Emotions Questionnaire to measure participant emotional reactions to their robot counterpart (Harmon-Jones et al., 2016). The shortened 19-item measure utilized a 7-point Likert scale from “not at all” to “an extreme amount”. The selection of a larger number on the scale indicated a stronger emotional reaction.
First, participants were told that they were to complete an interdependent task with their robot teammate. Next, participants were randomly assigned to either a human-like robot or a machine-like robot condition and received an image of their teammate. After receiving their teammate’s photo and being provided with the associated background information they were asked to complete the emotional reaction questionnaire. All of these steps were taken prior to completion of their simulated task, thus, capturing their initial reactions to their robot counterpart.
Results
An independent sample t-test was run to evaluate the relationship between emotional reactions and robot surface-level cues. All assumptions associated with this analysis were reviewed prior to conducting any analyses. Fear was found to be a significant emotional reaction based on surface-level cues of the robot teammate (t(216) = -2.24, p < 0.05). Further, there was more fear reported within groups assigned to the human-like robot (µ = 1.84) than the machine-like robot (µ = 1.52). The other emotions, anger, disgust, sadness, and happiness, did not yield significance based on the surface-level cues of their robot teammate.
Conclusion
The findings from the present work indicate the strong influence of robot physical characteristics on participant emotional reactions. More specifically, there is a greater understanding of how participants who have not interacted with a robot emotionally perceive robots within simulated environments. As it relates to the healthcare domain, these results can support decision-making efforts for robot purchasing and design. For example, it may be beneficial, for novel users, to work with a machine-like robot first before working with a human-like robot.
Future work should investigate the relationship between individual differences and their influence on emotional reactions to robots based on their surface-level cues. This study should also be replicated with providers and patients to review whether these findings vary based on task. Some of the limitations of this work are that it was conducted with university students and did not consider prior experience with other robots as it relates to participant’s emotional reactions. Despite these limitations, this work contributes valuable insights into the initial emotional response participants have to robots with varying physical characteristics.
Event Type
Robotics Workshop Submission
TimeSunday, March 3011:45am - 12:15pm EDT
LocationHarbour A/B