A Taxonomy of Social Errors in Human-Robot Interaction

Probably not a substantial claim: may not pass S3 (load-bearing).

People tend to trust embodied agents more than non-embodied agents.

“In terms of the appearance of a robot, it has been identified that people tend to trust embodied agents more than non-embodied agents [173].”

• cites [173] 173. Anouk van Maris, Hagen Lehmann, Lorenzo Natale, and Beata Grzyb. 2017. The influence of a robot's embodiment on trust: A longitudinal study. In Proceedings of the ACM/IEEE International Conference on Human-robot Interaction. ACM, 313-314. sourceDisputed

  The Results report no significant difference in trust between the physical robot and the virtual agent. The Conclusion likewise says the study found no significant influence of embodiment on trust. Although the Introduction describes a hypothesis that users would trust the physical robot more, the study did not confirm it. (LLM, conf=0.88)

  Checked by a reasoning model — may be wrong.

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Emotional contagion focuses on the convergence of verbal and non-verbal behaviors in interpersonal communications, also known as entrainment.

“Compared to empathy, emotional contagion focuses on the convergence of verbal and non-verbal behaviors in interpersonal communications, also known as entrainment [120].”

• cites [120] 120. Shrikanth Narayanan and Panayiotis G. Georgiou. 2013. Behavioral signal processing: Deriving human behavioral informatics from speech and language. Proc. IEEE 101, 5 (2013), 1203-1233. sourcePartly supported

  The General discussion says people tend to become more nonverbally similar to their social partners during an interaction. It does not identify this convergence as emotional contagion, extend it to verbal behavior, or contrast emotional contagion with empathy. Its reference to entrainment appears in a cited work's title, not in an explanation of the claimed relationship. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The progressive impairment can lead to reduced memory function, reduced ability in understanding others' thoughts and emotions, altered communication, depressed and unstable mood, and feelings of isolation and loneliness.

“The progressive impairment can lead to reduced memory function, reduced ability in understanding others' thoughts and emotions, altered communication, depressed and unstable mood, and feelings of isolation and loneliness [91].”

• cites [91] 91. Agnieszka Korchut, Sebastian Szklener, Carla Abdelnour, Natalia Tantinya, Joan Hernández-Farigola, Joan Carles Ribes, Urszula Skrobas, Katarzyna Grabowska-Aleksandrowicz, Dorota Szczęśniak-Stańczyk, and Konrad Rejdak. 2017. Challenges for service robots-Requirements of elderly adults with cognitive impairments. Front. Neurol. 8 (2017), 228. sourcePartly supported

  The Introduction links mild cognitive impairment and early Alzheimer disease to difficulties with activities that depend on memory, and the Discussion addresses loneliness among older people and people with dementia. The copy does not say that progressive impairment reduces the ability to understand others’ thoughts and emotions or causes depressed or unstable mood; its discussion of communication concerns reduced opportunities with age, not an altered ability to communicate. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

In psychotherapy sessions, the empathy of a therapist is a key behavioral marker for the quality and constructive outcomes of the therapy.

“For example, in psychotherapy sessions, the empathy of a therapist is a key behavioral marker for the quality and constructive outcomes of the therapy [21].”

• cites [21] 21. Brian L. Burke, Christopher W. Dunn, David C. Atkins, and Jerry S. Phelps. 2004. The emerging evidence base for motivational interviewing: A meta-analytic and qualitative inquiry. J. Cog. Psychother. 18, 4 (2004), 309-322. sourceCould not verify

  The Procedure (Toddlers’ Play with a Parent) rates parental warmth and sensitivity during toddler play, and the Results (Associations between Child and Parent Behavior Ratings and Measures of Play) relate those ratings to pretend play. Searches for psychotherapy, therapist empathy, and therapy outcomes found no passage addressing whether therapist empathy marks therapy quality or outcomes. The supplied passages have section labels but no page numbers. (LLM, conf=0.76)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Norms in smiling, laughter, and expressions of positive emotions are related to the historical heterogeneity of the population in a country or a state.

“In addition, a previous cross-cultural study found that norms in smiling, laughter, and expressions of positive emotions were related to the historical heterogeneity of the population in a country or a state [122].”

• cites [122] 122. Paula M. Niedenthal, Magdalena Rychlowska, Adrienne Wood, and Fangyun Zhao. 2018. Heterogeneity of long- history migration predicts smiling, laughter and positive emotion across the globe and within the United States. PLoS ONE 13, 8 (2018), e0197651. sourcePartly supported

  Study 1 and Study 2 Results (pages unavailable) report associations between historical heterogeneity and smiling/laughter and positive emotion at the country and U.S. state levels. But the Measures sections describe self-reported smiling or laughter and experienced enjoyment, not norms governing those behaviors or expressions of positive emotion. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing).

Emotion recognition and social context interpretation tasks have been applied to assess the degree of age-related social cognition decline, such as the Edinburgh social cognition test.

“Similar to assessments of social interaction capability used in ASD, emotion recognition and social context interpretation tasks have been applied to assess the degree of age-related social cognition decline, such as the Edinburgh social cognition test [7].”

• cites [7] 7. R. Asaad Baksh, Sharon Abrahams, Bonnie Auyeung, and Sarah E. MacPherson. 2018. The Edinburgh social cogni- tion test (ESCoT): Examining the effects of age on a new measure of theory of mind and social norm understanding. PLoS ONE 13, 4 (2018), 1-16. sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

A decline in cognitive functions occurs in elderly adults due to the loss of neurons and neuronal connections in the brain.

“A decline in cognitive functions occurs in elderly adults due to the loss of neurons and neuronal connections in the brain.”

• cites [91] 91. Agnieszka Korchut, Sebastian Szklener, Carla Abdelnour, Natalia Tantinya, Joan Hernández-Farigola, Joan Carles Ribes, Urszula Skrobas, Katarzyna Grabowska-Aleksandrowicz, Dorota Szczęśniak-Stańczyk, and Konrad Rejdak. 2017. Challenges for service robots-Requirements of elderly adults with cognitive impairments. Front. Neurol. 8 (2017), 228. sourcePartly supported

  The Introduction reports increasing cognitive impairment among older adults, and the Discussion refers to cognitive decline in the elderly. Neither section attributes that decline to the loss of neurons or neuronal connections, so the claimed cause is unsupported by this paper. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

May combine two claims.

Such a feeling leads to a decreased willingness to comply and increased willingness to punish when the desired outcome is not achieved.

“Such a feeling leads to a decreased willingness to comply and increased willingness to punish when the desired outcome is not achieved [2].”

• cites [2] 2. Jason Aimone, Sheryl Ball, and Brooks King-Casas. 2015. The betrayal aversion elicitation task: An individual level betrayal aversion measure. PLoS ONE 10, 9 (2015), e0137491. sourcePartly supported

  In “Previous Treatments of Betrayal Aversion” (page unavailable), the paper links non-reciprocated trust to decreased willingness to take a risky action or increased willingness to punish following the outcome. It does not describe the first response as decreased willingness to comply, so that part of the claim is not supported. (LLM, conf=0.86)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Social norms are external, socially defined normative standards that are the most socially appropriate action for a decision-maker in a given set of circumstances.

“Social norms are external, socially defined normative standards that are the most socially appropriate action for a decision-maker in a given set of circumstances [44].”

• cites [44] 44. Jon Elster. 1989. Social norms and economic theory. J. Econ. Perspect. 3, 4 (1989), 99-117. sourceSupported

Classified as probably not a substantial claim.

The review of human collaborative and organizational behaviors leads to additional attributes of socio-affective competence related to collaboration and prosocial behaviors, reported in Table 3.

“Our review of human collaborative and organizational behaviors allows us to zoom out from interpersonal social interactions to collaborations within and across groups.This leads to additional attributes of socio-affective competence, which we report in Table 3.”

Probably not a substantial claim: may not pass S4 (based).

How HRI errors influence users' perceptions, such as trust, is yet to be understood.

“How HRI errors influence users' perceptions, such as trust, is yet to be understood.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Previous analyses of HRI errors' impact on a user's perception have mainly focused on their influence on users' trust towards the HRI system.

“Previous analyses of HRI errors' impact on a user's perception have mainly focused on their influence on users' trust towards the HRI system.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The error types are not entirely independent from one another and can be correlated.

“Note that the error types are not entirely independent from one another and can be correlated.”

Classified as probably not a substantial claim.

Several taxonomies of HRI errors have been proposed.

“To better facilitate systematic analysis of errors in HRI, several taxonomies of HRI errors have been proposed.”

Probably not a substantial claim: may not pass S4 (based).

Other HRI error taxonomies have focused on the functional aspect of HRI, rather than users' perceptions of these errors.

“Other HRI error taxonomies have focused on the functional aspect of HRI, rather than users' perceptions of these errors.”

May combine two claims.

Most existing HRI error studies are short-term, single-session HRI experiments, and long-term, in-the-wild HRI studies are extremely scarce.

“Most existing HRI error studies are short-term, single-session HRI experiments.Long-term, in-the-wild HRI studies are extremely scarce [74].”

• cites [74] 74. Bahar Irfan, Aditi Ramachandran, Samuel Spaulding, Dylan F. Glas, Iolanda Leite, and Kheng Lee Koay. 2019. Per- sonalization in long-term human-robot interaction. In Proceedings of the 14th ACM/IEEE International Conference on Human-robot Interaction (HRI'19). IEEE, 685-686. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

Emotion is an important communicative cue that humans use to express their mind and interpret others' to facilitate collaboration.

“In Reventlov's scenario, as discussed in Section 2.1, emotion is an important communicative cue that humans use to express their mind and interpret others' to facilitate collaboration.”

Probably not a substantial claim: may not pass S4 (based).

Each category contains both instantaneous social errors that may occur in short-term HRI and social errors specific to long-term HRI.

“Each category contains both instantaneous social errors that may occur in short-term HRI and social errors specific to longterm HRI.”

Probably not a substantial claim: may not pass S2 (checkable).

The social-emotional dimension of HRI requires further analyses.

“Moreover, these existing taxonomies have focused on the functional dimension of HRI, while the social-emotional dimension of HRI requires further analyses.”

Classified as probably not a substantial claim.

Inspired by the 30 main attributes under five categories, the paper classifies social errors in HRI into five categories and 30 sub-categories in Table 4.

“In particular, inspired by the 30 main attributes under five categories that humans adopt to perceive socio-affective competence and social relationships (see Section 2), we classify social errors in HRI into five categories and 30 sub-categories in Table 4.”

Probably not a substantial claim: may not pass S4 (based).

Social skills facilitate building and maintaining human interpersonal relationships.

“Social skills facilitate building and maintaining human interpersonal relationships, as discussed in Section 2.2.”

Probably not a substantial claim: may not pass S3 (load-bearing).

An error scenario can be a mix of performance errors and social errors.

“An error scenario can be a mix of performance errors and social errors.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Understanding the influence of social norm violations may be key to advance current HRI research, especially in longitudinal scenarios where a positive human-robot relationship is desired.

“Understanding such influence may be key to advance current HRI research, especially in longitudinal scenarios where a positive human-robot relationship is desired.”

• AssumptionRests on an assumption: Error handling in HRI is not yet fully understood, especially regarding errors on the social-emotional dimension.

Classified as probably not a substantial claim.

User-centered design focuses on addressing the needs and preferences of users and the intended application scenarios of the designed systems.

“User-centered design focuses on addressing the needs and preferences of users and the intended application scenarios of the designed systems [124].”

• cites [124] 124. Donald A. Norman and Stephen W. Draper. 1986. User Centered System Design: New Perspectives on Human-computer Interaction. CRC Press. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

The social relationship between a user and a robot relies on the perceived competence of the robot on the social-emotional dimension.

“This social relationship relies on the perceived competence of a robot on the social-emotional dimension.”

Probably not a substantial claim: may not pass S3 (load-bearing).

More reliable and effective hand-eye coordination for robotic grasping can be learned through correcting errors.

“Levine et al. [100] found that more reliable and effective hand-eye coordination for robotic grasping can be learned through correcting errors.”

• cites [100] 100. Sergey Levine, Peter Pastor, Alex Krizhevsky, Julian Ibarz, and Deirdre Quillen. 2018. Learning hand-eye coordina- tion for robotic grasping with deep learning and large-scale data collection. Int. J. Robot. Res. 37, 4-5 (2018), 421-436. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

A social error may be caused by technical malfunctions, such as delayed dialogue responses, or by the imperfect design of social-emotional interaction functions of an HRI system, such as not incorporating individual variation.

“A social error may be caused by technical malfunctions, such as delayed dialogue responses, or by the imperfect design of social-emotional interaction functions of an HRI system, such as not incorporating individual variation.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

User-centered design is commonly adopted in human-computer interaction design practices and has known benefits such as improved usability and user satisfaction.

“It is commonly adopted in human-computer interaction design practices and has known benefits such as improved usability and user satisfaction [93].”

• cites [93] 93. Sari Kujala. 2003. User involvement: A review of the benefits and challenges. Behav. Inf. Technol. 22, 1 (2003), 1-16. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

In elderly care applications, HRI designs that incorporate social-emotional interactions improve health outcomes by encouraging positive moods and reducing loneliness felt by users.

“For example, in elderly care applications, HRI designs that incorporate social-emotional interactions have been shown to improve health outcomes by encouraging positive moods and reducing loneliness felt by users [58].”

• cites [58] 58. Susel Góngora Alonso, Sofiane Hamrioui, Isabel de la Torre Díez, Eduardo Motta Cruz, Miguel López-Coronado, and Manuel Franco. 2018. Social robots for people with aging and dementia: A systematic review of literature. Telemed. e-Health 25, 7 (2018), 533-540. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Empathy allows the development of shared representations, engagement, and simulation of the subjectivity of others.

“Empathy allows the development of shared representations, engagement, and simulation of the subjectivity of others.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A commonly used assessment of atypical social communicative functions is evaluating a person's ability to understand others.

“A commonly used assessment of atypical social communicative functions is evaluating a person's ability to understand others.”

Classified as probably not a substantial claim.

Failures in emotional interaction functions will result in HRI social errors on socio-affective competence dimensions.

“Failures in doing so will result in HRI social errors on these socio-affective competence dimensions, which will be discussed in more detail in Section 4.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In healthy elderly adults, impairment in social and cognitive functions may be mild and does not necessarily result in a significant decline in the quality of their personal and social life.

“In healthy elderly adults, impairment in social and cognitive functions may be mild and does not necessarily result in a significant decline in the quality of their personal and social life [157].”

• cites [157] 157. Susanne Scheibe and Laura L. Carstensen. 2010. Emotional aging: Recent findings and future trends. J. Geront.: Series B 65, 2 (2010), 135-144. sourceNot checked — no full text of the source

May combine two claims.

Elderly adults with neurodegenerative diseases typically experience more severe impairment in social cognition functions, which significantly impacts their ability to maintain independence and engage in social interactions.

“Elderly adults with neurodegenerative diseases typically experience more severe impairment in social cognition functions, which significantly impacts their ability to maintain independence and engage in social interactions.”

May combine two claims.

The same event or stimulus can evoke a spectrum of emotional responses in different people and can be perceived differently.

“The same event or stimulus can evoke a spectrum of emotional responses in different people and can be perceived differently, as demonstrated in existing studies of emotion induction and annotation tasks [136].”

• cites [136] 136. Soujanya Poria, Erik Cambria, Rajiv Bajpai, and Amir Hussain. 2017. A review of affective computing: From uni- modal analysis to multimodal fusion. Inf. Fus. 37 (2017), 98-125. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The shared anchor of different definitions of empathy is the matching relation between the emotions of two people.

“Decety and Meyer [35] also identified that the shared anchor of different definitions of empathy is the matching relation between the emotions of two people.”

• cites [35] 35. Jean Decety and Meghan Meyer. 2008. From emotion resonance to empathic understanding: A social developmental neuroscience account. Dev. Psychopathol. 20, 4 (2008), 1053-1080. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

For a robot to facilitate social interaction, it requires the ability to recognize emotions of its users, to incorporate in its decision-making process the emotions of users and artificial emotions generated by itself, and to synthesize emotional expressions according to the interaction context.

“For a robot to facilitate social interaction, it requires the ability to recognize emotions of its users, to incorporate in its decision-making process the emotions of users and artificial emotions generated by itself, and to synthesize emotional expressions according to the interaction context.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S3 (load-bearing).

The ability to exhibit empathy and entrainment is an important factor evaluated in various assessments of social skills, such as the Social Skills Inventory, the Mayer-Salovey-Caruso Emotional Intelligence Test, and the Self Descriptive Inventory.

“The ability to exhibit empathy and entrainment is an important factor evaluated in various assessments of social skills, such as the Social Skills Inventory [106], the Mayer-Salovey-Caruso Emotional Intelligence Test [111], and the Self Descriptive Inventory [143].”

• cites [106] 106. Maurice Lorr, Richard P. Youniss, and Edward C. Stefic. 1991. An inventory of social skills. J. Personal. Assess. 57, 3 (1991), 506-520. sourceNot checked — no full text of the source
• cites [111] 111. Andrew Maul. 2012. The validity of the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT) as a measure of emotional intelligence. Emot. Rev. 4, 4 (2012), 394-402. sourceNot checked — no full text of the source
• cites [143] 143. Ronald E. Riggio. 1986. Assessment of basic social skills. J. Personal. Soc. Psychol. 51, 3 (1986), 649. sourceNot checked — no full text of the source

May combine two claims.

Entrainment can be utilized in HRI design to increase the engagement of users, as well as to improve interaction outcomes, such as users' learning outcomes under an educational robot.

“These findings indicate that entrainment can be utilized in HRI design to increase the engagement of users, as well as to improve interaction outcomes, such as users' learning outcomes under an educational robot.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S3 (load-bearing).

Individuals with frontotemporal dementia have difficulties recognizing facial expressions and gaze directions.

“For example, individuals with frontotemporal dementia [145] were found to have difficulties recognizing facial expressions and gaze directions.”

• cites [145] 145. Howard J. Rosen, Katherine Pace-Savitsky, Richard J. Perry, Joel H. Kramer, Bruce L. Miller, and Robert W. Levenson. 2004. Recognition of emotion in the frontal and temporal variants of frontotemporal dementia. Dement. Geriat. Cog. Disord. 17, 4 (2004), 277-281. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

During interpersonal communications, emotions have been identified as cues that allow people to observe and perceive the latent cognitive states of their interaction partners, and to express their own internal states.

“During interpersonal communications, emotions have been identified as cues that allow people to observe and perceive the latent cognitive states of their interaction partners, and to express their own internal states [152,175].”

• cites [152] 152. Magdalena Rychlowska, Rachael E. Jack, Oliver G. B. Garrod, Philippe G. Schyns, Jared D. Martin, and Paula M. Niedenthal. 2017. Functional smiles: Tools for love, sympathy, and war. Psychol. Sci. 28, 9 (2017), 1259-1270. sourceNot checked — no full text of the source
• cites [175] 175. Alessandro Vinciarelli, Maja Pantic, Dirk Heylen, Catherine Pelachaud, Isabella Poggi, Francesca D'Errico, and Marc Schroeder. 2012. Bridging the gap between social animal and unsocial machine: A survey of social signal processing. IEEE Trans. Affect. Comput. 3, 1 (2012), 69-87. sourceNot checked — no full text of the source

May combine two claims.

An HRI system with empathic functions leads to a better understanding of the users and is required for establishing social relationships between the system and its users.

“Therefore, an HRI system with empathic functions leads to a better understanding of the users and is required for establishing social relationships between the system and its users.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S3 (load-bearing).

For individuals with limited communicative ability, an observer's evaluation can be used to assess their development level.

“For individuals with limited communicative ability, an observer's evaluation can be used to assess their development level.”

• cites [59] 59. Katherine Gotham, Susan Risi, Andrew Pickles, and Catherine Lord. 2007. The autism diagnostic observation sched- ule: Revised algorithms for improved diagnostic validity. J. Autism Dev. Disord. 37, 4 (2007), 613. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

ASD refers to a class of neurodevelopmental disorders.

“ASD refers to a class of neurodevelopmental disorders [51].”

• cites [51] 51. Uta Frith. 2003. Autism: Explaining the Enigma. Blackwell Publishing. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

HRI applications in mental health extend beyond ASD and elderly care, e.g., HRI for depression care.

“HRI applications in mental health extend beyond ASD and elderly care, e.g., HRI for depression care [15].”

• cites [15] 15. Casey C. Bennett, Selma Sabanovic, Jennifer A. Piatt, Shinichi Nagata, Lori Eldridge, and Natasha Randall. 2017. A robot a day keeps the blues away. In Proceedings of the IEEE International Conference on Healthcare Informatics (ICHI'17). IEEE, 536-540. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Current research on healthy aging has focused on identifying reliable methods to improve the emotional well-being and social functioning of individuals and preventing further decline of cognitive functions.

“Current research on healthy aging has focused on identifying reliable methods to improve the emotional well-being and social functioning of individuals and preventing further decline of cognitive functions [157].”

• cites [157] 157. Susanne Scheibe and Laura L. Carstensen. 2010. Emotional aging: Recent findings and future trends. J. Geront.: Series B 65, 2 (2010), 135-144. sourceNot checked — no full text of the source

May combine two claims.

Existing HRI studies have focused on two mental health domains: interventions for individuals with impaired cognitive and social abilities and providing companionship as well as physical and mental support to individuals living on their own.

“In particular, existing HRI studies have focused on two domains [142]: (1) interventions for individuals with impaired cognitive and social abilities and (2) providing companionship as well as physical and mental support to individuals living on their own.”

• cites [142] 142. Laurel D. Riek. 2016. Robotics technology in mental health care. In Artificial Intelligence in Behavioral and Mental Health Care. Elsevier, 185-203. sourceNot checked — no full text of the source

May combine two claims.

Emotions are fundamental to intelligence and interaction.

“Our review of the functions of human emotions demonstrates that emotions are fundamental to intelligence and interaction.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Researchers implemented artificial emotions in robots to improve their ability to generate adaptive behaviors.

“Motivated by these functions of emotion, researchers implemented artificial emotions in robots to improve their ability to generate adaptive behaviors [1,117].”

• cites [1] 1. Carole Adam and Benoit Gaudou. 2016. BDI agents in social simulations: A survey. Knowl. Eng. Rev. 31, 3 (2016), 207-238. sourceNot checked — no full text of the source
• cites [117] 117. Thomas M. Moerland, Joost Broekens, and Catholijn M. Jonker. 2018. Emotion in reinforcement learning agents and robots: A survey. Mach. Learn. 107, 2 (2018), 443-480. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Self-reported evaluations of social interaction capabilities have been used, for example, the empathy quotient questionnaire.

“Beyond emotion recognition tests, self-reported evaluations of social interaction capabilities have also been used; for example, the empathy quotient questionnaire [10].”

• cites [10] 10. Simon Baron-Cohen and Sally Wheelwright. 2004. The empathy quotient: An investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences. J. Autism Dev. Disord. 34, 2 (2004), 163-175. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

A major diagnostic cue of ASD is impairment in reciprocal social interaction and communication.

“A major diagnostic cue of ASD is impairment in reciprocal social interaction and communication [105].”

• cites [105] 105. Catherine Lord, Susan Risi, Linda Lambrecht, Edwin H. Cook, Bennett L. Leventhal, Pamela C. DiLavore, Andrew Pickles, and Michael Rutter. 2000. The autism diagnostic observation schedule-Generic: A standard measure of social and communication deficits associated with the spectrum of autism. J. Autism Dev. Disord. 30, 3 (2000), 205- 223. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Commonly used empathy assessments include the Empathy Quotient, the Empathy Scale, the Interpersonal Reactivity Index, and the Questionnaire Measure of Emotional Empathy.

“Commonly used empathy assessments include the Empathy Quotient [10], the Empathy Scale [76], the Interpersonal Reactivity Index [31], and the Questionnaire Measure of Emotional Empathy [113].”

• cites [10] 10. Simon Baron-Cohen and Sally Wheelwright. 2004. The empathy quotient: An investigation of adults with Asperger syndrome or high functioning autism, and normal sex differences. J. Autism Dev. Disord. 34, 2 (2004), 163-175. sourceNot checked — no full text of the source
• cites [76] 76. John A. Johnson, Jonathan M. Cheek, and Robert Smither. 1983. The structure of empathy. J. Personal. Soc. Psychol. 45, 6 (1983), 1299-1312. sourceNot checked — no full text of the source
• cites [31] 31. Mark H. Davis et al. 1980. A multidimensional approach to individual differences in empathy. JSAS Catalog of Selected Documents in Psychology 10 (1980), 85. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Such tests may not apply to those with limited verbal ability or across different languages or cultures.

“However, such tests may not apply to those with limited verbal ability or across different languages or cultures.”

Classified as probably not a substantial claim.

The identified socio-affective competence attributes can be used to measure users' perception of social errors and their impacts on HRI systems.

“These attributes can be used to measure users' perception of social errors and their impacts on such HRI systems.”

• AssumptionRests on an assumption: Current social-emotional interaction functions of HRI systems are far below human-level performance, and errors are inevitable during interactions.

Probably not a substantial claim: may not pass S3 (load-bearing).

Assessments with movie clips were used to capture the multimodal and context-specific nature of social interaction, such as the Movie for the Assessment of Social Cognition task.

“To capture the multimodal and contextspecific nature of social interaction, assessments with movie clips were also used, such as the Movie for the Assessment of Social Cognition task [43].”

• cites [43] 43. Isabel Dziobek, Stefan Fleck, Elke Kalbe, Kimberley Rogers, Jason Hassenstab, Matthias Brand, Josef Kessler, Jan K. Woike, Oliver T. Wolf, and Antonio Convit. 2006. Introducing MASC: A movie for the assessment of social cognition. J. Autism Dev. Disord. 36, 5 (2006), 623-636. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Variations of economic games have been applied to evaluate the perceived prosociality of robots and a user's trust towards a robot.

“Variations of these games have been applied to evaluate the perceived prosociality of robots and a user's trust towards a robot (e.g., Wu et al. [179]).”

• cites [179] 179. Jane Wu, Erin Paeng, Kari Linder, Piercarlo Valdesolo, and James C. Boerkoel. 2016. Trust and cooperation in human- robot decision making. In Proceedings of the AAAI Fall Symposium Series.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Trust has direct impacts on human decision-making and willingness to accept information or follow instructions from robots.

“Trust has direct impacts on human decision-making and willingness to accept information or follow instructions from robots [64].”

• cites [64] 64. Peter A. Hancock, Deborah R. Billings, Kristin E. Schaefer, Jessie Y. C. Chen, Ewart J. De Visser, and Raja Parasuraman. 2011. A meta-analysis of factors affecting trust in human-robot interaction. Hum. Fact. 53, 5 (2011), 517-527. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

An interesting aspect exposed by behavioral economics studies is the temporal dimension of social errors.

“An interesting aspect exposed by behavioral economics studies is the temporal dimension of social errors.”

May combine two claims.

Social norms vary across cultures and individual sensitivity to social reputation.

“Social norms vary across cultures and individual sensitivity to social reputation.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

The perceived socio-affective competence of a robot and social errors in HRI should also be investigated in longitudinal settings beyond short-term, single interaction sessions.

“This indicates that the perceived socio-affective competence of a robot and social errors in HRI should also be investigated in longitudinal settings beyond short-term, single interaction sessions.”

• AssumptionRests on an assumption: Error handling in HRI is not yet fully understood, especially regarding errors on the social-emotional dimension.

Probably not a substantial claim: may not pass S3 (load-bearing).

The attractiveness of physical appearance and voice of agents influences trust in user-agent interaction.

“The attractiveness of physical appearance and voice of agents was also shown to influence trust in user-agent interaction [183].”

• cites [183] 183. Beste F. Yuksel, Penny Collisson, and Mary Czerwinski. 2017. Brains or beauty: How to engender trust in user-agent interactions. ACM Trans. Internet Technol. 17, 1 (2017), 2. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The public goods game measures people's prosociality and investigates how awareness of social reputation motivates prosocial behaviors.

“The public goods game measures people's prosociality and investigates how awareness of social reputation motivates prosocial behaviors [45].”

• cites [45] 45. Ernst Fehr and Urs Fischbacher. 2004. Social norms and human cooperation. Trends Cog. Sci. 8, 4 (2004), 185-190. sourceNot checked — no full text of the source

The quote only partly states this claim.

Giuliani et al.'s approach of categorizing social errors as violations to two types of social norms in dialogue is limited because social interaction is more than verbal and non-verbal communication.

“Thus, their approach of categorizing social errors as violations to two types of social norms in dialogue is limited.”

May combine two claims.

An educational robot perceived as trustworthy and privacy-preserving may inspire stronger engagement and motivation in its students, resulting in better learning outcomes.

“Thus, an educational robot perceived as trustworthy and privacypreserving may inspire stronger engagement and motivation in its students, resulting in better learning outcomes.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Adherence to, or violation of, social norms has mainly been studied in the context of three economic games: the public goods game, the trust game, and the dictator game.

“Adherence to, or violation of, social norms has mainly been studied in the context of three economic games: the public goods game [96], the trust game [16], and the dictator game [80].”

• cites [96] 96. John O. Ledyard. 1994. Public goods: A survey of experimental research. Social Science Working Paper. Cali- fornia Institute of Technology, Pasadena, CA. Retrieved on January 21, 2021 from https://resolver.caltech.edu/ CaltechAUTHORS:20170823-160736011. sourceNot checked — no full text of the source
• cites [16] 16. Joyce Berg, John Dickhaut, and Kevin McCabe. 1995. Trust, reciprocity, and social history. Games Econ. Behav. 10, 1 (1995), 122-142. sourceNot checked — no full text of the source
• cites [80] 80. Daniel Kahneman, Jack L. Knetsch, and Richard Thaler. 1986. Fairness as a constraint on profit seeking: Entitlements in the market. Amer. Econ. Rev. 76, 4 (1986), 728-741. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Three main aspects of trust have been analyzed in current HRI studies: the environment perspective, the human perspective, and the robot perspective.

“Three main aspects of trust have been analyzed in current HRI studies: the environment perspective, the human perspective, and the robot perspective [156].”

• cites [156] 156. Kristin Schaefer. 2013. The Perception and Measurement of Human-Robot Trust. Ph.D. Dissertation. University of Central Florida. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The social norms questionnaire requires an individual to assess if a behavior is appropriate or not to evaluate whether the person is overly adherent to a perceived social norm or unaware of social norm violations.

“The social norms questionnaire requires an individual to assess if a behavior is appropriate or not to evaluate whether the person is overly adherent to a perceived social norm or unaware of social norm violations.”

• cites [137] 137. K. P. Rankin. 2008. Social norms questionaire. NINDS: Domain Specific Tasks of Executive Function. Retrieved January 21, 2021 from https://memory.ucsf.edu/sites/memory.ucsf.edu/files/wysiwyg/EXAMINER_UserManual_3.6r.pdf.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Adherence to social norms is tied to a person's social reputation and perceived prosociality.

“Adherence to social norms is tied to a person's social reputation and perceived prosociality [4].”

• cites [4] 4. James Andreoni and B. Douglas Bernheim. 2009. Social image and the 50-50 norm: A theoretical and experimental analysis of audience effects. Econometrica 77, 5 (2009), 1607-1636. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Erratic behaviors of robots negatively influence people's trust towards them.

“In terms of the capability of a robot, erratic behaviors of robots were shown to negatively influence people's trust towards them [176].”

• cites [176] 176. Ning Wang, David V. Pynadath, and Susan G. Hill. 2016. Trust calibration within a human-robot team: Comparing automatically generated explanations. In Proceedings of the 11th ACM/IEEE International Conference on Human-robot Interaction. IEEE Press, 109-116. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

For each attribute of socio-affective competence identified, the judgment of whether a behavior is normative or not varies across cultural, contextual, and individual factors.

“Note that for each attribute of the socio-affective competence we identified, the judgment of whether a behavior is normative or not varies across cultural, contextual, and individual factors.”

• AssumptionRests on an assumption: The judgment of whether a behavior is normative or not varies across cultural, contextual, and individual factors.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Sebo et al. [158] found that when a robot broke a promise, the participant indicated lower trust and exhibited retaliation, and apology was more effective than denial.

“They found that when a robot broke a promise previously made to a participant, the participant indicated lower trust towards the robot and exhibited retaliation in later stages of the game.They studied two types of repair strategies addressing the trust violation, namely, apology and denial, and their results suggested that apology was more effective.”

• cites [158] 158. Sarah Strohkorb Sebo, Priyanka Krishnamurthi, and Brian Scassellati. 2019. "I don't believe you": Investigating the effects of robot trust violation and repair. In Proceedings of the 14th ACM/IEEE International Conference on Human- robot Interaction (HRI'19). IEEE, 57-65. sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

The perception of socio-affective competence is flexible and context-dependent.

“However, the perception of socio-affective competence is flexible and context-dependent.”

May combine two claims.

Established trust can be violated by short instances, and the difficulty and time required for trust recovery depend on various factors, such as compensation and repair strategies.

“However, established trust can be violated by short instances, and the difficulty and time required for trust recovery depend on various factors, such as compensation and repair strategies [38].”

• cites [38] 38. Pieter T. M. Desmet, David De Cremer, and Eric van Dijk. 2011. Trust recovery following voluntary or forced finan- cial compensations in the trust game: The role of trait forgiveness. Personal. Indiv. Diff. 51, 3 (2011), 267-273. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Three standardized evaluation questionnaires are widely used for dementia patients or people at risk: the interpersonal reactivity index, the social norms questionnaire, and the social behavior observer checklist.

“In particular, three standardized evaluation questionnaires are widely used for dementia patients or people at risk, namely, the interpersonal reactivity index [30], the social norms questionnaire [137], and the social behavior observer checklist [138].”

• cites [30] 30. Mark H. Davis. 1980. Interpersonal Reactivity Index. Edwin Mellen Press.Not checked — no full text of the source
• cites [137] 137. K. P. Rankin. 2008. Social norms questionaire. NINDS: Domain Specific Tasks of Executive Function. Retrieved January 21, 2021 from https://memory.ucsf.edu/sites/memory.ucsf.edu/files/wysiwyg/EXAMINER_UserManual_3.6r.pdf.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [138] 138. K. P. Rankin, E. Baldwin, C. Pace-Savitsky, J. H. Kramer, and B. L. Miller. 2005. Self awareness and personality change in dementia. J. Neurol. Neurosurg. Psychia. 76, 5 (2005), 632-639. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The dictator game measures people's adherence to the fairness social norm.

“The dictator game measures people's adherence to the fairness social norm [40].”

• cites [40] 40. Andreas Diekmann. 2004. The power of reciprocity: Fairness, reciprocity, and stakes in variants of the dictator game. J. Conflict Resol. 48, 4 (2004), 487-505. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The impact of socio-affective competence on collaborative behaviors has been studied intensively in behavioral economics and economic games, mainly in the context of social norms.

“The impact of socio-affective competence on collaborative behaviors has been studied intensively in behavioral economics and economic games, mainly in the context of social norms [45].”

• cites [45] 45. Ernst Fehr and Urs Fischbacher. 2004. Social norms and human cooperation. Trends Cog. Sci. 8, 4 (2004), 185-190. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Social norms are constantly changing, following the shifting expectations of the majority in a social group.

“They are constantly changing, following the shifting expectations of the majority in a social group [87].”

• cites [87] 87. Erik O. Kimbrough and Alexander Vostroknutov. 2016. Norms make preferences social. J. Eur. Econ. Assoc. 14, 3 (2016), 608-638. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Trust requires time to cultivate.

“Trust requires time to cultivate.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Because of the complexity in human social interaction, a robot designed to interact with people is bound to make errors.

“Because of the complexity in human social interaction, a robot designed to interact with people is bound to make errors.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The trust game measures trust and reciprocity, which accommodated indicators of cooperativeness and adherence to social norms.

“The trust game measures trust and reciprocity [24], which accommodated indicators of cooperativeness and adherence to social norms [163].”

• cites [24] 24. David Cesarini, Christopher T. Dawes, James H. Fowler, Magnus Johannesson, Paul Lichtenstein, and Björn Wallace. 2008. Heritability of cooperative behavior in the trust game. Proc. Nat. Acad. Sci. 105, 10 (2008), 3721-3726. sourceNot checked — no full text of the source
• cites [163] 163. Dirk Sliwka. 2007. Trust as a signal of a social norm and the hidden costs of incentive schemes. Amer. Econ. Rev. 97, 3 (2007), 999-1012. sourceNot checked — no full text of the source

May combine two claims.

The mean time between technical failures for robots in the wild is often less than a few hours, and current HRI systems are not yet mature enough to effectively handle unexpected events.

“The mean time between technical failures for robots in the wild is often less than a few hours, and current HRI systems are not yet mature enough to effectively handle unexpected events [70].”

• cites [70] 70. Shanee Sarah Honig and Tal Oron-Gilad. 2018. Understanding and resolving failures in human-robot interaction: Literature review and model development. Front. Psychol. 9 (2018), 861. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The interpersonal reactivity index measures empathic concerns and ability of perspective-taking.

“The interpersonal reactivity index measures empathic concerns and ability of perspective-taking.”

• cites [30] 30. Mark H. Davis. 1980. Interpersonal Reactivity Index. Edwin Mellen Press.Not checked — no full text of the source

May combine two claims.

In both social and non-social situations, humans can feel betrayed when their trust is not reciprocated or when their expectation is unsatisfied.

“In both social and non-social situations, humans can feel betrayed when their trust is not reciprocated [81] or when their expectation is unsatisfied [55].”

• cites [81] 81. Daniel Kahneman, David Schkade, and Cass Sunstein. 1998. Shared outrage and erratic awards: The psychology of punitive damages. J. Risk Uncert. 16, 1 (1998), 49-86. sourceNot checked — no full text of the source
• cites [55] 55. Andrew D. Gershoff and Jonathan J. Koehler. 2011. Safety first? The role of emotion in safety product betrayal aversion. J. Consum. Res. 38, 1 (2011), 140-150. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The social behavior observer checklist collects a carer's observations on decreased social functions of an individual, such as showing diminished social and emotional engagement, having unstable emotional and cognitive abilities, exhibiting perseveration, or not being sensitive to others' embarrassment or privacy.

“The social behavior observer checklist collects a carer's observations on decreased social functions of an individual, such as showing diminished social and emotional engagement, having unstable emotional and cognitive abilities, exhibiting perseveration, or not being sensitive to others' embarrassment or privacy.”

• cites [138] 138. K. P. Rankin, E. Baldwin, C. Pace-Savitsky, J. H. Kramer, and B. L. Miller. 2005. Self awareness and personality change in dementia. J. Neurol. Neurosurg. Psychia. 76, 5 (2005), 632-639. sourceNot checked — no full text of the source

May combine two claims.

When a robot is unable to recognize emotions of its human teammates accurately and timely, its ability to assess the joint intention of the team can be impaired, while such ability is key to effective and successful human-robot collaboration.

“Thus, when Reventlov is unable to recognize emotions of its human teammates in an accurate and timely manner (E1.1), its ability to assess the joint intention of the team can be impaired, while such ability is key to effective and successful human-robot collaboration [13].”

• cites [13] 13. Andrea Bauer, Dirk Wollherr, and Martin Buss. 2008. Human-robot collaboration: A survey. Int. J. Human. Robot. 5, 01 (2008), 47-66. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The three scenarios are closely aligned with existing HRI studies under similar contexts.

“They are are closely aligned with existing HRI studies under similar contexts, as detailed below:”

Probably not a substantial claim: may not pass S3 (load-bearing).

In Calvin's scenario, a breach in empathic and emotional reactions is particularly harmful because of the vulnerability of its user.

“In Calvin's scenario, a breach in empathic and emotional reactions is particularly harmful because of the vulnerability of its user.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Scenario-based design is a widely used approach in human-computer interaction studies.

“Scenario-based design is a widely used approach in humancomputer interaction studies [149].”

• cites [149] 149. Mary Beth Rosson and John M. Carroll. 2009. Scenario based design. In Proceedings of the Conference on Human- computer Interaction. 145-162. sourceNot checked — no full text of the source

May combine two claims.

These social errors in HRI apply to scenarios beyond the specific areas that they originated from, and all five types of social errors may occur in any HRI scenario.

“Thus, these social errors in HRI apply to scenarios beyond the specific areas that they originated from, and all five types of social errors may occur in any HRI scenario.”

• AssumptionThe authors state this as an assumption: The 30 attributes of socio-affective competence identified from human studies are applicable to a broad class of HRI scenarios and application domains.

May combine two claims.

The induction of humor can be difficult to control, and the outcome can vary for different individuals.

“The induction of humor can be difficult to control, and the outcome can vary for different individuals.”

Probably not a substantial claim: may not pass S2 (checkable).

The proposed taxonomy illustrates the breadth and complexity of social errors in HRI.

“This proposed taxonomy illustrates the breadth and complexity of social errors in HRI.”

May combine two claims.

Such learned adaption enables a robot to respond with empathy and over time establish a companionship with its user.

“Such learned adaption enables a robot to respond with empathy [131] and over time establish a companionship with its user [58].”

• cites [131] 131. Ana Paiva, Iolanda Leite, Hana Boukricha, and Ipke Wachsmuth. 2017. Empathy in virtual agents and robots: A survey. ACM Trans. Interact. Intell. Syst. 7, 3 (2017), 11. sourceNot checked — no full text of the source
• cites [58] 58. Susel Góngora Alonso, Sofiane Hamrioui, Isabel de la Torre Díez, Eduardo Motta Cruz, Miguel López-Coronado, and Manuel Franco. 2018. Social robots for people with aging and dementia: A systematic review of literature. Telemed. e-Health 25, 7 (2018), 533-540. sourceNot checked — no full text of the source

May combine two claims.

The impact of the same social error may differ in short-term vs. long-term or repeated HRI, and researchers should take the temporal dimension of HRI into account when studying social errors.

“Thus, the impact of the same social error may differ in short-term vs. longterm or repeated HRI, and researchers should take the temporal dimension of HRI into account when studying social errors.”

Classified as probably not a substantial claim.

De Graaf et al. [32] found that the robot's lack of adaptability and sociability were the two determinants of users' abandonment.

“They found that the robot's lack of adaptability and sociability were the two determinants of users' abandonment.”

• cites [32] 32. Maartje De Graaf, Somaya Ben Allouch, and Jan Van Dijk. 2017. Why do they refuse to use my robot?: Reasons for non-use derived from a long-term home study. In Proceedings of the ACM/IEEE International Conference on Human- robot Interaction. ACM, 224-233. sourceNot checked — no full text of the source

May combine two claims.

Beane [14] found that telepresence robots were abandoned and stowed away shortly after deployment because they only served as a symbol of technological novelty and offered limited interaction and use cases.

“It was found that the robots were abandoned and stowed away shortly after deployment, as they only served as a symbol of technological novelty and offered limited interaction and use cases.”

• cites [14] 14. Matthew I. Beane. 2020. In storage, yet on display: An empirical investigation of robots' value as social signals. In Proceedings of the ACM/IEEE International Conference on Human-robot Interaction. 83-91. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Lee et al. [97] found that users with a socially oriented interaction goal responded best to apologies, while those with a utilitarian orientation responded best to compensation.

“They found that users with a socially oriented interaction goal responded best to apologies, while those with a utilitarian orientation responded best to compensation.”

• cites [97] 97. Min Kyung Lee, Sara Kiesler, Jodi Forlizzi, Siddhartha Srinivasa, and Paul Rybski. 2010. Gracefully mitigating break- downs in robotic services. In Proceedings of the 5th ACM/IEEE International Conference on Human-robot Interaction (HRI'10). IEEE, 203-210. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

A robot's failure in maintaining engagement during an interaction (E4.5) can be a consequence of it generating irrelevant or incomprehensible replies to verbal or non-verbal communications of the user (E4.2).

“For example, a robot's failure in maintaining engagement during an interaction (E4.5) can be a consequence of it generating irrelevant or incomprehensible replies to verbal or non-verbal communications of the user (E4.2).”

May combine two claims.

Through longitudinal interaction, a robot can learn to improve its ability to recognize the emotions of a specific person, and that person can better recognize the synthesized expressions of the robot, i.e., a mutual adaption.

“Through longitudinal interaction, a robot can learn to improve its ability to recognize the emotions of a specific person, and that person can better recognize the synthesized expressions of the robot, i.e., a mutual adaption.”

Classified as probably not a substantial claim. The quote only partly states this claim.

When a robot is not able to accurately recognize a user's emotional states or exhibit empathy and entrainment, providing help and mental health care in a timely fashion becomes nearly impossible.

“In individual short-term interaction sessions, when Calvin is not able to accurately recognize Frank's emotional states (E1.1) or exhibit empathy and entrainment (E1.5), providing help and mental health care in a timely fashion becomes nearly impossible.”

May combine two claims.

Kubota et al. [92] found benefits in both usability and health outcomes when target users were included in participatory design workshops for a social robot for neurorehabilitation.

“Their results demonstrated benefits in both usability and health outcomes of the robot.”

• cites [92] 92. Alyssa Kubota, Emma I. C. Peterson, Vaishali Rajendren, Hadas Kress-Gazit, and Laurel D. Riek. 2020. JESSIE: Syn- thesizing social robot behaviors for personalized neurorehabilitation and beyond. In Proceedings of the ACM/IEEE International Conference on Human-robot Interaction. 121-130. sourceNot checked — no full text of the source

May combine two claims.

Humans can experience psychological changes caused by long periods of isolation and confinement, including impaired cognitive ability, negative affect, and interpersonal tension and conflict.

“Humans can experience psychological changes caused by such a long period of isolation and confinement, including impaired cognitive ability, negative affect, and interpersonal tension and conflict [133].”

• cites [133] 133. Lawrence A. Palinkas and Peter Suedfeld. 2008. Psychological effects of polar expeditions. Lancet 371, 9607 (2008), 153-163. sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

A receptionist robot provides services to a heterogeneous population.

“A receptionist such as Olivaw's provides services to a heterogeneous population.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The paper creates three hypothetical HRI scenarios following the ecological validity guideline, i.e., these are believable situations that approximate real-life situations.

“We create three hypothetical HRI scenarios following the ecological validity guideline, i.e., these are believable situations that approximate real-life situations.”

• AssumptionRests on an assumption: The three hypothetical HRI scenarios are believable situations that approximate real-life situations.

Probably not a substantial claim: may not pass S2 (checkable).

Understanding the users' interaction goals is important when designing and evaluating an HRI system.

“This indicates the importance of understanding the users' interaction goals when designing and evaluating an HRI system.”

May combine two claims.

Violation of existing expectancy may also result in humor, and a breach in interaction norms may instead amuse a user and increase the memorability and enjoyment of an HRI session.

“However, violation of existing expectancy may also result in humor [36], and a breach in the interaction norms may instead amuse a user and increase the memorability and enjoyment of an HRI session.”

• cites [36] 36. Lambert Deckers and John Devine. 1981. Humor by violating an existing expectancy. J. Psychol. 108, 1 (1981), 107- 110. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

In short-term HRI, a breach in empathic and emotional reactions can lead to user engagement breakdowns or failures in achieving the goal of the interaction session.

“As discussed here, in short-term HRI, a breach in empathic and emotional reactions can lead to user engagement breakdowns or failures in achieving the goal of the interaction session.”

Classified as probably not a substantial claim.

If a robot carer appears unstable in its emotional states, it may worsen the user's condition.

“On the contrary, if Calvin appears unstable in its emotional states (E1.3), it may worsen Frank's condition.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Compared to other sub-categories of HRI errors, social norm violations require further analyses and more detailed classification.

“However, as shown in Figure 1, compared to other sub-categories of HRI errors, social norm violations require further analyses and more detailed classification.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

These socio-affective dimensions are connected, and together they cover the multiple facets of social-emotional interaction.

“However, these socio-affective dimensions are connected, and together they cover the multiple facets of social-emotional interaction, as discussed in Section 2.5.”

May combine two claims.

Failure in establishing mutual understanding creates obstacles for team integration and lowers the efficiency of the collaboration in the long term.

“Thus, failure in establishing mutual understanding (E1.6) creates obstacles for team integration and lowers the efficiency of the collaboration in the long term.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Scenario-based design emphasizes people and their experiences, which aligns with the user-centered perspective of the paper's taxonomy.

“Scenario-based design emphasizes people and their experiences, which aligns with the user-centered perspective of our taxonomy of HRI social errors.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

A systematic and user-centered understanding of errors on the social-emotional dimension of HRI is required.

“As discussed in Section 3, a systematic and user-centered understanding of errors on the socialemotional dimension of HRI is required.”

Probably not a substantial claim: may not pass S3 (load-bearing).

A person visiting a reception desk for the first time will be expecting a lower intimacy level in their social interaction with a robot compared to a regular visitor.

“For example, a person visiting the reception desk for the first time will be expecting a lower intimacy level in their social interaction with Olivaw compared to a member of the department who is a regular visitor.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In the three hypothetical HRI scenarios, E1 and E2 may be more relevant to Olivaw's scenario, E3 and E4 may be more relevant to Calvin's scenario, and E5 may be more relevant to Reventlov's scenario.

“Thus, in our three hypothetical HRI scenarios, E1 and E2 may be more relevant to Olivaw's scenario, E3 and E4 may be more relevant to Calvin's scenario, and E5 may be more relevant to Reventlov's scenario.”

Probably not a substantial claim: may not pass S3 (load-bearing).

In short-term interaction sessions, without emotional reaction functions, a robot can still provide basic services such as giving directions to visitors.

“Regarding Olivaw's scenario, in short-term interaction sessions, without emotional reaction functions, it can still provide basic services such as giving directions to visitors.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Human perception of a breach and its severity is flexible and context-dependent.

“Human perception of a breach and its severity is flexible and context-dependent.”

• AssumptionRests on an assumption: Human perception of a breach and its severity is flexible and context-dependent.

Classified as probably not a substantial claim. May combine two claims.

The service breakdown in Lee et al.'s study is a performance error by the paper's definition, while a robot displaying inappropriate social behaviors after a service breakdown is a social error by the paper's definition.

“The service breakdown (bringing a wrong drink) in their simulated study is a performance error by our definition, while a robot displaying inappropriate social behaviors (e.g., using an ineffective recovery strategy) after a service breakdown is a social error by our definition.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The polar expedition context puts the human-robot team in an extreme environment.

“Note that the polar expedition context puts the human-robot team in an extreme environment.”

Classified as probably not a substantial claim.

Because the goal of a robot carer is to provide longitudinal service, its socio-affective functions need to be personalized through accumulated interaction experiences.

“Because the goal of Calvin is to provide longitudinal service to Frank, its socio-affective functions need to be personalized through accumulated interaction experiences.”

Classified as probably not a substantial claim.

The paper's taxonomy of social errors in HRI focuses on how errors influence users' perception of the socio-affective competence of a robot and their perceived social relationship with it.

“To address this, our taxonomy of social errors in HRI focuses on how errors in HRI influence users' perception of the socio-affective competence of a robot and their perceived social relationship with it.”

Probably not a substantial claim: may not pass S2 (checkable).

This focus is especially important for HRI systems with socially oriented goals, i.e., social robots.

“This is especially important for HRI systems with socially oriented goals, i.e., social robots.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Honig and Oron-Gilad [70] proposed to classify HRI errors as either technical failures or interaction failures.

“In a recent effort, Honig and Oron-Gilad [70] proposed to classify HRI errors as either technical failures or interaction failures.”

• cites [70] 70. Shanee Sarah Honig and Tal Oron-Gilad. 2018. Understanding and resolving failures in human-robot interaction: Literature review and model development. Front. Psychol. 9 (2018), 861. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

If Calvin cannot understand and adjust to Frank's reception of different stages of treatment plans, it will not be able to provide effective care over time (E3.5).

“If Calvin cannot understand and adjust to Frank's reception of different stages of treatment plans, it will not be able to provide effective care over time (E3.5 incorrect assessment of a user's reception of HRI)”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

If Calvin fails to establish and maintain a social relationship with Frank over time (E2.7), it is unlikely to achieve the expected health benefits of engaging Frank in social interactions.

“if Calvin is unable to establish and maintain a social relationship with Frank over time (E2.7), it is unlikely to achieve the expected health benefits of engaging Frank in social interactions”

May combine two claims.

In longitudinal HRI, social errors can damage a user's perceived social relationship with a robot and lead to its abandonment.

“In longitudinal HRI, social errors can damage a user's perceived social relationship with a robot and lead to its abandonment”

May combine two claims.

If a robot carer fails to initiate or respond to reciprocal social communication, it may lead to the user's rejection of the robot or compromise his desire to use it.

“Thus, if Calvin fails to initiate or respond to reciprocal social communication (E2.5), it may lead to Frank's rejection of Calvin or compromise his desire to use it.”

May combine two claims.

Failures to exhibit assertiveness (E2.3) can reduce other team members' evaluation of the capability of Reventlov and their willingness to comply with the plans it proposes.

“Failures to exhibit assertiveness (E2.3) can reduce other team members' evaluation of the capability of Reventlov and their willingness to comply with the plans it proposes”

May combine two claims.

Performance errors degrade a user's impressions of a robot's competence, while social errors degrade a user's impressions of a robot's warmth and result in similar emotional and behavioral reactions to the robot as they do towards humans.

“we expect performance errors to degrade a user's impressions of a robot's competence, while social errors will degrade a user's impressions of a robot's warmth and result in similar emotional and behavioral reactions to the robot as they do towards humans”

May combine two claims.

Insufficient communicative functions may degrade a user's perception of a conversational robot's competence in achieving tasks as well as degrading her perception of the robot's socio-affective competence.

“insufficient communicative functions may degrade a user's perception of a conversational robot's competence in achieving tasks as well as degrading her perception of the robot's socio-affective competence”

May combine two claims.

Social relationship has known health benefits, and perceived social relationship with another entity represents the perceived adequacy of resources provided by the other entity.

“Social relationship has known health benefits [72], and perceived social relationship with another entity represents the perceived adequacy of resources provided by the other entity [26].”

• cites [72] 72. James S. House, Cynthia Robbins, and Helen L. Metzner. 1982. The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh Community Health Study. Amer. J. Epidem. 116, 1 (1982), 123-140. sourceNot checked — no full text of the source
• cites [26] 26. Sheldon Cohen. 2004. Social relationships and health. Amer. Psychol. 59, 8 (2004), 676. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Understanding social errors in HRI is key to design advanced error handling and failure recovery strategies, which leads to better interaction experiences and outcomes.

“Understanding social errors in HRI is key to design advanced error handling and failure recovery strategies, which leads to better interaction experiences and outcomes”