Exploratory Investigation of the Effects of Perspective Taking and Awareness of Vulnerability on Impressions of Robots

41

Abstract

Objective. This exploratory study investigated whether perspective-taking and awareness of vulnerability procedures could enhance impressions of robots.
Background. A society characterized by the harmonious coexistence of humans and robots is poised for realization in the imminent future. Nevertheless, numerous challenges must be confronted for the materialization of such a societal paradigm. One among them pertains to the prevailing tendency for humans to harbor adverse perceptions of robots, the amelioration of which proves to be a complex endeavor. The present study undertakes an exploratory investigation into strategies aimed at mitigating unfavorable impressions associated with robots.
Study design. Participants were randomly assigned to one of three groups: control group, perspective perception group, and robot vulnerability awareness group, and received different instructions.
Participants. Online experiments were conducted with 360 participants who were asked to imagine and describe a day in the life of a robot, and their impressions of the robot were measured using a questionnaire.
Measurements. Upon conjecturing and articulating the robot's daily routines, participants shared their perceptions of the robot through the application of three assessment tools: the Robot Anxiety Scale, the Mind Attribution Scale, and the Familiarity Rating Scale.
Results. The manipulation checks confirmed successful manipulation, but there was no evidence that perspective-taking or awareness of vulnerability influenced impressions of the robot.
Conclusions. The efficacy of perspective-taking, a technique established as beneficial in ameliorating adverse perceptions of humans, may exhibit diminished effectiveness in the context of alleviating negative impressions associated with robots.

General Information

Keywords: social robot; perspective taking; awareness of vulnerability of the robot; empathy; human impression of the robot

Journal rubric: Empirical Research

Article type: scientific article

DOI: https://doi.org/10.17759/sps.2024150210

Funding. This work was supported by JSPS Topic-Setting Program to Advance Cutting-Edge Humanities and Social Sciences Research Grant Number JPJS00122674991

Received: 04.08.2023

Accepted:

For citation: Ikeda S. Exploratory Investigation of the Effects of Perspective Taking and Awareness of Vulnerability on Impressions of Robots. Sotsial'naya psikhologiya i obshchestvo = Social Psychology and Society, 2024. Vol. 15, no. 2, pp. 155–163. DOI: 10.17759/sps.2024150210.

Full text

Introduction

Empathy, a crucial ability for group living [23; 25], is multidimensional in nature [4] and has been defined as the capacity to recognize the emotions of others with minimal discrimination between self and others [8]. While empathy is observed in various animal species, humans are believed to possess more advanced forms of empathy as a result of evolution [6; 7; 15].
Empathy is often associated with the ability to adopt the perspective of others [5; 13]. In fact, research has suggested that empathy can be augmented through the practice of perspective-taking, where individuals strive to understand the experiences and viewpoints of others [10; 12]. For instance, taking the perspective of individuals belonging to a certain category has been shown to decrease prejudice based on age and race [9; 11].
However, recent research has highlighted that perspective-taking may not be effective in reducing prejudice towards robots [26]. Studies have reported that humans struggle to empathize with robots [3; 24], and that traditional perspective-taking procedures employed in previous studies do not improve negative impressions of robots [26]. In fact, it has been demonstrated that humans encounter difficulties in adopting the perspective of robots [27].
Given the increasing likelihood of further development in the coexistence with robots in the future, how can we recognize robots as social partners? This study investigates, in an exploratory manner, the effects of awareness of the vulnerability of robots. For instance, research has shown that humans can deliver electric shocks to a moving, talking LEGO robot at a higher intensity than to a human, upon command [1]. If humans are made aware of the fact that robots, like humans, are unable to recover from fatal injuries, they may be more likely to treat robots with a similar level of consideration as they would humans.
This study aims to investigate whether altering the perspective-taking procedure and raising awareness of the vulnerability of the robot would result in changes in impressions of the robot. Although previous research has shown that perspective-taking does not significantly impact impressions of robots [26], it is hypothesized that awareness of vulnerability may positively influence these impressions. To test this hypothesis, participants were randomly assigned to one of four experimental conditions: (1) perspective-taking group, (2) awareness of vulnerability group, (3) combined perspective-taking and awareness of vulnerability group, and (4) control group. Impressions of the robot were assessed after the experimental manipulation in each condition.
 

Method

Participants
The study was conducted using an online survey, with a total of 360 participants (136 female, 223 male, 1 non-response, Mean age = 51,01 years, SD = 10,53, range = 18-60) recruited through GMO Research (https://gmo-research.com/). The procedures for this study were approved by the Ethical Review Board of the author's institution (approval number 22H11).

 
Materials
In this study, the following three questionnaires were used to measure impressions of robots: The first is the Robot Anxiety Scale (RAS) [20]. This scale measures anxiety about robots in general with items such as "I would be nervous if I had to operate a robot in front of other people". The scale consisted of 11 items answered on a 5-point scale. The second is the Mind Attribution Scale (MAS) [14]. This scale measures the degree of mind attribution to robots through items such as "To what extent are robots able to experience joy?” The scale consisted of 18 items answered on a 7-point scale. The third is the Familiarity Rating Scale (FRS) [17]. This scale measures familiarity with the robot, and 11 adjective pairs such as "unfriendly-friendly" are presented, and the participants answer on a 7-point scale (the higher the number, the stronger their familiarity with the robot).
 
Procedure
The study was conducted on GORILLA Experiment Builder, an online experimental program. Participants accessed Gorilla from the URL provided in the GMO Research survey request and participated in the experiment. Four conditions were set for this study, and participants were randomly delivered a link that allowed them to access one of the four conditions.
The study consisted of six phases: an age and gender response phase, a free writing phase, a computational task phase, a RAS phase, a MAS phase, and an FRS phase. In each phase, participants were provided with instructions on what they were required to do, and they initiated the phase by pressing the "Start" button. There was no time limit set for each phase, unless otherwise noted. However, if the total duration of a phase exceeded 30 minutes, the phase was forcibly terminated, as there was no intention to continue beyond that point.
In the free writing phase, a picture of NAO, a bipedal humanoid robot, was presented to all conditions with different instructions for each of the four conditions. The instructions were as follows: Control group: Participants were instructed to imagine and describe a typical day in the life of the robot shown in the picture, which can talk to humans and perform simple household chores. Perspective-taking group: Participants were instructed to imagine and describe a day in the life of the robot from its point of view, as if they were the robot, which can talk to humans and perform simple household chores. Awareness of vulnerability group: Participants were instructed to imagine a day in the life of the robot and describe it in detail, while noting that the robot is fragile and cannot be restored to its original state if severely damaged. The robot shown in the picture can talk to humans and perform simple household chores. Perspective-taking x Awareness of vulnerability group: Participants were instructed to imagine a day in the life of the robot as if they were the robot, and describe it in detail from the robot's point of view. They were also instructed to note that the robot is fragile and cannot be restored to its original state if severely damaged. The robot shown in the picture can talk to humans and perform simple household chores. Participants were informed that they had 5,5 minutes to write the description, and a countdown was displayed on the screen 30 seconds before the end. In the calculation task phase, 20 one-digit addition questions were performed as fillers.
In the RAS, MAS, and FRS phases, all questions were presented on the screen in a random order. The user could not move to the next screen until all items in each phase were answered.
 
Coding
For each of the three scales, the mean of the rating values for each item was calculated for each individual and used as the score. The RAS score ranged from 1 to 5, with higher scores indicating greater anxiety toward the robot; the MAS score ranged from 1 to 7, with higher scores indicating greater mind attribution to the robot; and the FRS score ranged from 1 to 7, with higher scores indicating greater familiarity with the robot.
 

Results

Manipulation check of free writing procedures
The study examined whether there were differences in the content of the descriptions provided in the free writing phase based on the instructions given in each of the four conditions. Since the criterion of changes in the frequency of first person, often used in perspective-taking studies in Western countries, may not be applicable in Japanese language where first person is often omitted, text mining was conducted to analyze the words used in the descriptions. Correspondence analysis was performed using KH Coder version 3.Beta.07b [16], including all parts-of-speech, and the top 60 words with significant differences were used in the analysis. The results were shown in Figure, which displayed the words that were salient in each of the four conditions, arranged in the direction of the name of each condition.
The findings from Figure showed that the word "壊れる" (meaning "break" in Japanese) appeared characteristically in the group that underwent the awareness of vulnerability condition. On the other hand, the words "主人" (meaning "master" in Japanese) and "人間" (meaning "human" in Japanese) appeared characteristically in the group that underwent the perspective-taking condition. This suggests that participants in the perspective-taking group considered themselves as robots and viewed humans relatively. Additionally, the words "掃除" (meaning "cleaning" in Japanese) and "準備" (meaning "preparation" in Japanese) were common words across all groups. Given the prevalent omission of the subject in Japanese sentences, the frequent utilization of first-person pronouns, a conventional metric employed for manipulation check in Western contexts, was deemed unsuitable. Additionally, the pioneering nature of the methodology employed in this study precluded the preselection of standard words. Consequently, the operational checks implemented in this study, while not furnishing robust evidence, based on these results, it can be concluded that the manipulation of the free writing procedure was successful in eliciting different content in the descriptions based on the instructions given in each condition.
 
Fig. Results of correspondence analysis of free writing content
 
Impression Rating of the Robot
The study conducted a one-factor, four-level between-participants analysis of variance (ANOVA) to examine the effects of perspective-taking and awareness of vulnerability on impressions of the robot, using the RAS, MAS, and FRS scale scores as dependent variables (see Table). The results showed that the main effect of condition was not significant for any of the scales: RAS (F(3, 356) = 0,446, p = 0,720, ηp2 = 0,004), MAS (F(3, 356) = 0,587, p = 0,624, ηp2 = 0,005), or FRS (F(3, 356) = 0,916, p = 0,433, ηp2 = 0,008). This indicates that neither perspective-taking nor awareness of vulnerability had a significant effect on impressions of the robot, as measured by the RAS, MAS, and FRS scores.
 
Table
Average of impression ratings for robots in each condition

Variables

Control

Perspective taking

Awareness of vulnerability

Perspective x vulnerability

RAS

3,369 (0,987)

3,445 (0,820)

3,445 (0,946)

3,524 (0,846)

MAS

3,456 (1,006)

3,551 (0,874)

3,454 (0,927)

3,361 (1,031)

FRS

4,253 (1,010)

4,370 (0,891)

4,463 (0,897)

4,257 (1,171)

Note: Values in parentheses indicate standard deviations.
 

Discussion

The findings of the study suggest that perspective-taking and awareness of vulnerability, as manipulated in the study, did not have a significant effect on impressions of the robot. This is consistent with previous research that showed perspective-taking did not reduce robot anxiety [26]. However, it is noteworthy that the lack of effect of awareness of vulnerability on impressions of the robot was contrary to the predictions of the study. This finding may suggest that other factors, beyond perspective-taking and awareness of vulnerability, may play a more significant role in shaping impressions of robots. Further research may be needed to explore and understand the complex relationship between human cognition and impressions of robots.
One possible explanation for the lack of significant effects of perspective-taking and awareness of vulnerability on impressions of the robot despite successful manipulations could be related to the uncanny valley phenomenon. The uncanny valley refers to the discomfort or eeriness that humans may experience when interacting with robots or other artificial agents that closely resemble humans but are not quite identical [14; 19]. Perspective-taking and awareness of vulnerability may have made the robot feel creepy or uncanny to participants, counteracting any positive shift in impressions. Previous research [26] has shown that explicit instruction to suppress prejudice toward robots can reduce robot anxiety, but perspective-taking does not have the same effect. This suggests that direct and explicit procedures may be more effective in improving impressions of robots compared to indirect procedures like perspective-taking and awareness of vulnerability. Negative impressions of robots may be naturally held in everyday life situations [2; 18], and communicating with robots may evoke mixed responses, ranging from feeling anthropomorphic to feeling distant from humans [21; 22].
Perspective taking and vulnerability awareness, which were manipulated in the study, did not significantly affect impressions of the robot. Based on the results of this study, research should be planned to discover ways to mitigate negative impressions of robots. To promote coexistence with robots in the future, it may be necessary to consider a variety of measures, including explicit procedures to suppress prejudice, in addition to indirect approaches like perspective-taking and awareness of vulnerability. Understanding the complex and multifaceted nature of human-robot interactions and impressions of robots is important for developing effective strategies for human-robot coexistence. Further research in this area can contribute to our understanding of how to improve impressions of robots and foster positive interactions between humans and robots.
 

Conclusion

The strategic application of perspective-taking, renowned for its efficacy in ameliorating adverse perceptions of humans, may demonstrate diminished effectiveness when applied to mitigate negative impressions associated with robots. This discrepancy might be attributed to the inherent challenge of engaging in perspective-taking for robots compared to humans. In anticipation of an era characterized by the harmonious coexistence of humans and robots, the imperative arises to seek social-psychological solutions facilitating a more congruent and harmonized societal life, including the explicit attenuation of prejudice directed towards robots.
 
Declarations
Conflicts of Interest
The author declares no conflicts of interest.
Data Availability Statement
The data that support the findings of this study and materials are openly available in Open Science Framework at https://osf.io/49uns/?view_only=9cc7353e43b545d3973a6af933176ca4

References

  1. Bartneck C., Hu J. Exploring the abuse of robots. Interaction Studies. Social Behaviour and Communication in Biological and Artificial Systems, 2008. Vol. 9(3), pp. 415–433. DOI:/10.1075/is.9.3.04bar
  2. Berridge C., Zhou Y., Robillard J.M., Kaye J. Companion robots to mitigate loneliness among older adults : Perceptions of benefit and possible deception. Frontiers in Psychology, 2023, February, pp. 1–9. DOI:/10.3389/fpsyg.2023.1106633
  3. Chang W., Wang H., Yan G., Lu Z., Liu C., Hua C. EEG based functional connectivity analysis of human pain empathy towards humans and robots. Neuropsychologia, 2021. Vol. 151(November 2020), 107695. DOI:/10.1016/j.neuropsychologia.2020.107695
  4. Davis M.H. Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 1983. Vol. 44(1), pp. 113–126. DOI:/10.1037/0022-3514.44.1.113
  5. Davis M.H. Empathy: A Social Psychological Approach. Madison, Wis.: Brown & Benchmark Publishers, 1994. 260
  6. de Waal F.B.M. The antiquity of empathy. Science, 2012. Vol. 336(6083), pp. 874–876. DOI:/10.1126/science.1220999
  7. de Waal F.B.M. Putting the altruism back into altruism: The evolution of empathy. Annual Review of Psychology, 2008. 59, pp. 279–300. DOI:/10.1146/annurev.psych.59.103006.093625
  8. Decety J. The neurodevelopment of empathy in humans. Developmental Neuroscience, 2010. Vol. 32(4), pp. 257–267. DOI:/10.1159/000317771
  9. Dovidio J.F., Ten Vergert M., Stewart T.L., Gaertner S.L., Johnson J.D., Esses V.M., Riek B.M., Pearson A.R. Perspective and prejudice: Antecedents and mediating mechanisms. Personality and Social Psychology Bulletin, 2004. 30(12), pp. 1537–1549. DOI:/10.1177/0146167204271177
  10. Galinsky A.D., Ku G. The effects of perspective-taking on prejudice: The moderating role of self-evaluation. Personality and Social Psychology Bulletin, 2004. 30(5), pp. 594–604. DOI:/10.1177/0146167203262802
  11. Galinsky A.D., Moskowitz G.B. Perspective-taking: Decreasing stereotype expression, stereotype accessibility, and in-group favoritism. Journal of Personality and Social Psychology, 2000. 78(4), pp. 708–724. DOI:/10.1037/0022-3514.78.4.708
  12. Gehlbach H. Social perspective taking: A facilitating aptitude for conflict resolution, historical empathy, and social studies achievement. Theory and Research in Social Education, 2004. 32(1), pp. 39–55. DOI:/10.1080/00933104.2004.10473242
  13. Goldstein N.J., Vezich I.S., Shapiro J.R. Perceived perspective taking: When others walk in our shoes. Journal of Personality and Social Psychology, 2014. 106(6), pp. 941–960. DOI:/10.1037/a0036395
  14. Gray K., Wegner D.M. Feeling robots and human zombies: Mind perception and the uncanny valley. Cognition, 2012. 125(1), pp. 125–130. DOI:/10.1016/j.cognition.2012.06.007
  15. Herrmann E., Call J., Hernández-Lloreda M.V., Hare B., Tomasello M. Humans have evolved specialized skills of social cognition: The cultural intelligence hypothesis. Science, 2007. 317(5843), pp. 1360–1366. DOI:/10.1126/science.1146282
  16. Higuchi K.A. Two-Step Approach to Quantitative Content Analysis: KH Coder Tutorial Using Anne of Green Gables (Part I). Ritsumeikan Social Science Review, 2016. Vol. 52(3), pp. 77–91. http://hdl.handle.net/10367/8013
  17. Kanda T., Ishiguro H., Ishida T. Psychological Evaluation on Interactions between People and Robot. Journal of the Robotics Society of Japan, 2001. Vol. 19(3), pp. 362–371. DOI:/10.7210/jrsj.19.362
  18. Liu X. (Stella), Wan L.C., Yi X. (Shannon). Humanoid versus non-humanoid robots: How mortality salience shapes preference for robot services under the COVID-19 pandemic? Annals of Tourism Research, 2022. Vol. 94, 103383. DOI:/10.1016/j.annals.2022.103383
  19. MacDorman K.F. Mortality salience and the uncanny valley. Proceedings of 2005 5th IEEE-RAS International Conference on Humanoid Robots, 2005. Vol. 2005, pp. 399–405. DOI:/10.1109/ICHR.2005.1573600
  20. Nomura T., Suzuki T., Kanda T., Kato K. Measurement of anxiety toward robots. Proceedings – IEEE International Workshop on Robot and Human Interactive Communication, 2006, pp. 372–377. DOI:/10.1109/ROMAN.2006.314462
  21. Okanda M., Taniguchi K., Wang Y., Itakura S. Preschoolers’ and adults’ animism tendencies toward a humanoid robot. Computers in Human Behavior, 2021. 118, 106688. DOI:/10.1016/j.chb.2021.106688
  22. Okumura Y., Hattori T., Fujita S., Kobayashi T. A robot is watching me!: Five-year-old children care about their reputation after interaction with a social robot. Child Development, 2023, pp. 1–9. DOI:/10.1111/cdev.13903
  23. Panksepp J., Panksepp J.B. Toward a cross-species understanding of empathy. Trends in Neurosciences, 2013. 36(8), pp. 489–496. DOI:/10.1016/j.tins.2013.04.009
  24. Rosenthal-Von Der Pütten A.M., Schulte F.P., Eimler S.C., Sobieraj S., Hoffmann L., Maderwald S., Brand M., Krämer N.C. Investigations on empathy towards humans and robots using fMRI. Computers in Human Behavior, 2014. 33, pp. 201–212. DOI:/10.1016/j.chb.2014.01.004
  25. Seyfarth R.M., Cheney D.L. Affiliation, empathy, and the origins of Theory of Mind. Proceedings of the National Academy of Sciences of the United States of America, 2013. Vol. 110(SUPPL2), pp. 10349–10356. DOI:/10.1073/pnas.1301223110
  26. Wullenkord R., Eyssel F. Diversity training with robots: Perspective-taking backfires, while sterotype-suppression decreases negative attitudes towards robots. Frontiers in Robotics and AI, 2022. Vol. 9(March), pp. 1–14. DOI:/10.3389/frobt.2022.728923
  27. Xiao C., Xu L., Sui Y., Zhou R. Do People Regard Robots as Human-Like Social Partners? Evidence From Perspective-Taking in Spatial Descriptions. Frontiers in Psychology, 2021. Vol. 11(February), pp. 1–11. DOI:/10.3389/fpsyg.2020.578244

Information About the Authors

Shinnosuke Ikeda, PhD in Education, Associate Professor, Human and Social Administration Department, Kanazawa University Kakuma-machi, Ishikawa, Japan, ORCID: https://orcid.org/0000-0001-9128-5627, e-mail: odenshin@gmail.com

Metrics

Views

Total: 113
Previous month: 13
Current month: 7

Downloads

Total: 41
Previous month: 6
Current month: 1