As AI-enabled robots enter the realm of healthcare and caregiving, it is important to consider how they will address the dimensions of care and how they will interact not just with the direct receivers of assistance, but also with those who provide it (e.g., caregivers, healthcare providers etc.). Caregiving in its best form addresses challenges in a multitude of dimensions of a person's life: from physical, to social-emotional and sometimes even existential dimensions (such as issues surrounding life and death). In this study we use semi-structured qualitative interviews administered to healthcare professions with multidisciplinary backgrounds (physicians, public health professionals, social workers, and chaplains) to understand their expectations regarding the possible roles robots may play in the healthcare ecosystem in the future. We found that participants drew inspiration in their mental models of robots from both works of science fiction but also from existing commercial robots. Participants envisioned roles for robots in the full spectrum of care, from physical to social-emotional and even existential-spiritual dimensions, but also pointed out numerous limitations that robots have in being able to provide comprehensive humanistic care. While no dimension of care was deemed as exclusively the realm of humans, participants stressed the importance of caregiving humans as the primary providers of comprehensive care, with robots assisting with more narrowly focused tasks. Throughout the paper we point out the encouraging confluence of ideas between the expectations of healthcare providers and research trends in the human-robot interaction (HRI) literature.

Trust and capability transfer between tasks and environments is common in human-human interactions. For human-robot interactions it is unclear how a robot's performance of a task in one environment affects humans predictions about the robot's performance of another related or unrelated task in a different environment. When making assessments about a robot's task capabilities, three main sources of information are pertinent: the human's "default mental model" of robots, the robot's appearance, and the robot's performance. We hypothesized that past task performance would be the most salient information source, and that participants who saw the robot perform tasks in one environment would transfer their assumptions about the robot's capability to a new environment with new tasks. However, the results of our first study did not support this hypothesis. We then performed a second study to exclude the possibility that because the robot worked well in the first environment, it did not supply any salient, different information from the participants' default mental model of robots (that robots are functional, etc.). If this hypothesis was correct, a faulty robot in the first environment would be rated significantly lower at the tasks in the second environment. However, the results did not support the second hypothesis either. We then conducted a third study investigating whether the tasks themselves or the environment had a stronger effect on trust assessments. The results showed that because individual judgments varied dramatically no systematic trust and task transfer result can be obtained. The upshot for HRI is that trust and task transfer are solely dependent on the individual's background and judgment rather than on task or environmental properties.

Potential applications of robots in private and public human spaces have prompted the design of so-called “social robots” that can interact with humans in social settings and potentially cause humans to attach to the robots. The focus of this article is an analysis of possible benefits and challenges arising from such human-robot attachment as reported in the HRI literature, followed by guidelines for the use and the design of robots that might elicit attachment bonds. We start by analyzing the potential benefits for humans becoming attached to robots, which might include increased natural interaction, effectiveness and acceptance of the robot, social companionship, and well-being for the human. Turning to the potential risks associated with human-robot attachment, we discuss the possibly suboptimal use of the robot in the most benign cases, but also the potential formation of unidirectional emotional bonds, and the potential for deception and subconscious influence of the robot on the person in more severe cases. The upshot of the analysis then is a recommendation to reconceptualize relationships with social robots in an attempt to retain potential benefits of human-robot attachment, while mitigating (to the extent possible) its downsides.

We present a novel integration between a computational framework for modeling attention-driven perception and cognition (ARCADIA) with a cognitive robotic architecture (DIARC), demonstrating how this integration can be used to drive the gaze behavior of a robotic platform. Although some previous approaches to controlling gaze behavior in robots during human-robot interactions have relied either on models of human visual attention or human cognition, ARCADIA provides a novel framework with an attentional mechanism that bridges both lower-level visual and higher-level cognitive processes. We demonstrate how this approach can produce more natural and human-like robot gaze behavior. In particular, we focus on how our approach can control gaze during an interactive object learning task. We present results from a pilot crowdsourced evaluation that investigates whether the gaze behavior produced during this task increases confidence that the robot has correctly learned each object.

Much research effort in HRI has focused on how to enable robots to learn new skills from observations, demonstrations, and instructions. Less work, however, has focused on how skills can be corrected if they were learned incorrectly, adapted to changing circumstances, or generalized/specialized to different contexts. In this paper, a skill modification framework is introduced that allows users to modify a robot's stored skills quickly through instructions to (1) reduce inefficiencies, (2) fix errors, and (3) enable generalizations, all in a way for modified skills to be immediately available for task performance. A thorough evaluation of the implemented framework shows the operation of the algorithms integrated in a cognitive robotic architecture on different fully autonomous robots in various HRI case studies. An additional online HRI user study verifies that subjects prefer to quickly modify robot knowledge in the way we proposed in the framework.

Attachment theory is a research area in psychology that has enjoyed decades of successful study, and has subsequently become explored in realms beyond that of the original infant-caregiver bonds. Now, attachment is studied in relation to pets, symbols (such as deities), objects, technologies, and notably for our purposes, robots. When we discuss attachment in Human-Robot Interaction (HRI), is “attachment” to a robot the same as being attached to a pet? Or does it more closely resemble attachment to a technology device such as a smartphone? Through untangling the concept of attachment in HRI, we summarize a breadth of the existing attachment literature in a unified spectrum. We present a notion of weak attachment, and strong attachment before setting both as distinct ends of a spectrum of attachment. We motivate this spectrum by teasing out the underlying theoretical basis for strong attachment, and how capabilities of the attachment figure could lead to stronger or weaker attachment. This more nuanced, multi-dimensional representation of attachment allows us to present a clarified categorization of where various human-robot bonds explored in HRI studies fit on the spectrum, where robots in general could place, and how a clearer definition of human-robot attachment can benefit future HRI studies.

Trust in human-robot interactions (HRI) is measured in two main ways: through subjective questionnaires and through behavioral tasks. To optimize measurements of trust through questionnaires, the field of HRI faces two challenges: the development of standardized measures that apply to a variety of robots with different capabilities, and the exploration of social and relational dimensions of trust in robots (e.g., benevolence). In this paper we look at how different trust questionnaires (Lyons & Guznov, 2019; Schaefer, 2016; Ullman & Malle, 2018) fare given these challenges that pull in different directions (being general vs. being exploratory) by studying whether people think the items in these questionnaires are applicable to different kinds of robots and interactions. In Study 1 we show that after being presented with a robot (non-humanoid) and an interaction scenario (fire evacuation), participants rated multiple questionnaire items such as "This robot is principled" as "Non-applicable to robots in general" or "Non-applicable to this robot." In Study 2 we show that the frequency of these ratings change (indeed, even for items rated as N/A to robots in general) when a new scenario is presented (game playing with a humanoid robot). Finally, while overall trust scores remained robust to N/A ratings, our results revealed potential fallacies in the way these scores are commonly interpreted. We conclude with recommendations for the development, use and results-reporting of trust questionnaires for future studies, as well as theoretical implications for the field of HRI.

Robots are increasingly embedded in human societies where they encounter human collaborators, potential adversaries, and even uninvolved by-standers. Such robots must plan to accomplish joint goals with teammates while avoiding interference from competitors, possibly utilizing bystanders to advance the robot's goals. We propose a planning framework for robot task and action planners that can cope with collaborative, competitive, and non-involved human agents at the same time by using mental models of human agents. By querying these models, the robot can plan for the effects of future human actions and can plan robot actions to influence what the human will do, even when influencing them through explicit communication is not possible. We implement the framework in a planner that does not assume that human agents share goals with, or will cooperate with, the robot. Instead, it can handle the diverse relations that can emerge from interactions between the robot's goals and capacities, the task environment, and the human behavior predicted by the planner's models. We report results from an evaluation where a teleoperated robot executes a planner-generated policy to influence the behavior of human participants. Since the robot is not capable of performing some of the actions necessary to achieve its goal, the robot instead tries to cause the human to perform those actions.

As robots begin to enter roles in which they work closely with human teammates or peers, it is critical to understand how people trust them based on how they interpret the robot’s behavior. In this paper we investigated the interplay between trust in a robot and people’s perceptions of the robot’s emotional intelligence. We used a vignette-based method to explore the following questions: (1) Do subjects perceive differences in robot EI, and is their trust in the robot influenced by differences in the robot’s reliability and capability? (2) Does a robot’s EI influence how much it is trusted and conversely does a robot’s capability and reliability influence how emotionally intelligent it is perceived to be? (3) Do people trust male and female robots differently when the robots exhibit different levels of EI or different levels of capability and reliability, and do gender stereotypical expectations related to EI transfer to trust?; (4) Does focusing on the robot’s EI increase one’s trust in the robot? (5) Is the interplay between trust, EI and gender the same for different levels of evoked social presence and human-likeness (i.e., when the interaction is presented in different modalities, text or spoken dialogue when the robot’s voice is actually heard)? We found that trust in the robot was influenced by the level of the robot’s EI (p < .001) and that gender stereotypical expectations related to EI were transferred to trust (p  $$=$$  .006), but gender effects on trust disappeared when only capability and reliability (robot’s trustworthiness) were manipulated but not the robot’s EI (p = .103). Surprisingly, we found that people trusted the robot more when the interaction was presented in text format (p  $$=$$  .024), going against our hypothesis that spoken dialogue would evoke more social presence and thus bolster EI perception and instill more trust. We suggest that this effect might be due to people’s expectations of a more expressive and human-like voice. Finally, we also found that people’s trust ratings in the robot were higher when they were made to notice and think about the robot’s EI, by answering EI questionnaires prior to trust questionnaires (p = .022). We discuss the implications of our findings for robot design and HRI research.

There is a close connection between health and the quality of one's social life. Strong social bonds are essential for health and wellbeing, but often health conditions can detrimentally affect a person's ability to interact with others. This can become a vicious cycle resulting in further decline in health. For this reason, the social management of health is an important aspect of healthcare. We propose that socially assistive robots (SARs) could help people with health conditions maintain positive social lives by supporting them in social interactions. This paper makes three contributions, as detailed below. We develop a framework of social mediation functions that robots could perform, motivated by the special social needs that people with health conditions have. In this framework we identify five types of functions that SARs could perform: a) changing how the person is perceived, b) enhancing the social behavior of the person, c) modifying the social behavior of others, d) providing structure for interactions, and e) changing how the person feels. We thematically organize and review the existing literature on robots supporting human-human interactions, in both clinical and non-clinical settings, and explain how the findings and design ideas from these studies can be applied to the functions identified in the framework. Finally, we point out and discuss challenges in designing SARs for supporting social interactions, and highlight opportunities for future robot design and HRI research on the mediator role of robots.

We describe an approach to generating explanations about why robot actions fail, focusing on the considerations of robots that are run by cognitive robotic architectures. We define a set of Failure Types and Explanation Templates, motivating them by the needs and constraints of cognitive architectures that use action scripts and interpretable belief states, and describe content realization and surface realization in this context. We then describe an evaluation that can be extended to further study the effects of varying the explanation templates.

Emotions are crucial for human social interactions and thus people communicate emotions through a variety of modalities: kinesthetic (through facial expressions, body posture and gestures), auditory (the acoustic features of speech) and semantic (the content of what they say). Sometimes however, communication channels for certain modalities can be unavailable (e.g., in the case of texting), and sometimes they can be compromised, due to a disorder such as Parkinson's disease (PD) that may affect facial, gestural and speech expressions of emotions. To address this, we developed a prototype for an emoting robot that can detect emotions in one modality, specifically in the content of speech, and then express them in another modality, specifically through gestures. The system consists of two components: detection and expression of emotions. In this paper we present the development of the expression component of the emoting system. We focus on its dynamical properties that use a spring model for smooth transitions between emotion expressions over time. This novel method compensates for varying utterance frequency and prediction errors coming from the emotion recognition component. We also describe the input the dynamical expression component receives from the emotion detection component, the development and validation of the output comprising of the gestures instantiated in the robot, and the implementation of the system. We present results from a human validation study that shows people perceive the robot gestures, generated by the system, as expressing the emotions in the speech content. Also, we show that people's perceptions of the accuracy of emotion expression is significantly higher for a mass-spring dynamical system than a system without a mass-spring when specific detection errors are present. We discuss and suggest future developments of the system and further validation experiments. This paper is part of a larger project to develop a prototype for a socially assistive robot for PD persons. The goal is to present the technical implementation of one robot capability: emotion expression.

Individuals with Parkinson's disease (PD) often exhibit facial masking (hypomimia), which causes reduced facial expressiveness. This can make it difficult for those who interact with the person to correctly read their emotional state and can lead to problematic social and therapeutic interactions. In this article, we develop a probabilistic model for an assistive device, which can automatically infer the emotional state of a person with PD using the topics that arise during the course of a conversation. We envision that the model can be situated in a device that could monitor the emotional content of the interaction between the caregiver and a person living with PD, providing feedback to the caregiver in order to correct their immediate and perhaps incorrect impressions arising from a reliance on facial expressions. We compare and contrast two approaches: using the Latent Dirichlet Allocation (LDA) generative model as the basis for an unsupervised learning tool, and using a human-crafted sentiment analysis tool, the Linguistic Inquiry and Word Count (LIWC). We evaluated both approaches using standard machine learning performance metrics such as precision, recall, and F1scores. Our performance analysis of the two approaches suggests that LDA is a suitable classifier when the word count in a document is approximately that of the average sentence, i.e., 13 words. In that case, the LDA model correctly predicts the interview category 86% of the time and LIWC correctly predicts it 29% of the time. On the other hand, when tested with interviews with an average word count of 303 words, the LDA model correctly predicts the interview category 56% of the time and LIWC, 74% of the time. Advantages and disadvantages of the two approaches are discussed.

Robots are machines and as such do not have gender. However, many of the gender-related perceptions and expectations formed in human-human interactions may be inadvertently and unreasonably transferred to interactions with social robots. In this paper, we investigate how gender effects in people's perception of robots and humans depend on their emotional intelligence (EI), a crucial component of successful human social interactions. Our results show that participants perceive different levels of EI in robots just as they do in humans. Also, their EI perceptions are affected by gender-related expectations both when judging humans and when judging robots with minimal gender markers, such as voice or even just a name. We discuss the implications for human-robot interactions (HRI) and propose further explorations of EI for future HRI studies.

In typical human interactions emotional states are communicated via a variety of modalities such as auditory (through speech), visual (through facial expressions) and kinesthetic (through gestures). However, one or more modalities might be compromised in some situations, as in the case of facial masking in Parkinson's disease (PD). In these cases, we need to focus the communication and detection of emotions on the reliable modalities, by inferring emotions from what is being said, and compensate for the modalities that are problematic, by having another agent (e.g., a robot) provide the missing facial expressions. We describe the initial development stage of a robot companion that can assist the communication and detection of emotions in interactions where some modalities are totally or partially compromised. Such is the case for people living with Parkinson's disease. Our approach is based on a Latent Dirichlet Allocation topic model as a principled way to extract features from speech based on a trained classifier that can be linked to measures of emotion. The trained model is integrated into a robotic cognitive architecture to perform real-time, continuous speech detection of positive, negative, or neutral emotional valence that is expressed through the facial features of a humanoid robot. To evaluate the integrated system, we conducted a human-robot interaction experiment in which the robot credibly detected and displayed emotions as it listened to utterances spoken by a confederate. The utterance were directly extracted from interviews with people with Parkinson's Disease. The encouraging results will form the basis for further developments of finer prediction models to be employed in a companion robot for persons with PD.

BACKGROUND:As robots are increasingly designed for health management applications, it is critical to not only consider the effects robots will have on patients but also consider a patient's wider social network, including the patient's caregivers and health care providers, among others. OBJECTIVE:In this paper we investigated how people evaluate robots that provide care and how they form impressions of the patient the robot cares for, based on how the robot represents the patient. METHODS:We have used a vignette-based study, showing participants hypothetical scenarios describing behaviors of assistive robots (patient-centered or task-centered) and measured their influence on people's evaluations of the robot itself (emotional intelligence [EI], trustworthiness, and acceptability) as well as people's perceptions of the patient for whom the robot provides care. RESULTS:We found that for scenarios describing a robot that acts in a patient-centered manner, the robot will not only be perceived as having higher EI (P=.003) but will also cause people to form more positive impressions of the patient that the robot cares for (P<.001). We replicated and expanded these results to other domains such as dieting, learning, and job training. CONCLUSIONS:These results imply that robots could be used to enhance human-human relationships in the health care context and beyond.

We present an approach to generating natural language justifications of decisions derived from norm-based reasoning. Assuming an agent which maximally satisfies a set of rules specified in an object-oriented temporal logic, the user can ask factual questions (about the agent's rules, actions, and the extent to which the agent violated the rules) as well as "why" questions that require the agent comparing actual behavior to counterfactual trajectories with respect to these rules. To produce natural-sounding explanations, we focus on the subproblem of producing natural language clauses from statements in a fragment of temporal logic, and then describe how to embed these clauses into explanatory sentences. We use a human judgment evaluation on a testbed task to compare our approach to variants in terms of intelligibility, mental model and perceived trust.

Despite the importance of social interactions for infant brain development, little research has assessed functional neural activation while infants socially interact. Electroencephalography (EEG) power is an advantageous technique to assess infant functional neural activation. However, many studies record infant EEG only during one baseline condition. This protocol describes a paradigm that is designed to comprehensively assess infant EEG activity in both social and nonsocial contexts as well as tease apart how different types of social inputs differentially relate to infant EEG. The within-subjects paradigm includes four controlled conditions. In the nonsocial condition, infants view objects on computer screens. The joint attention condition involves an experimenter directing the infant's attention to pictures. The joint attention condition includes three types of social input: language, face-to-face interaction, and the presence of joint attention. Differences in infant EEG between the nonsocial and joint attention conditions could be due to any of these three types of input. Therefore, two additional conditions (one with language input while the experimenter is hidden behind a screen and one with face-to-face interaction) were included to assess the driving contextual factors in patterns of infant neural activation. Representative results demonstrate that infant EEG power varied by condition, both overall and differentially by brain region, supporting the functional nature of infant EEG power. This technique is advantageous in that it includes conditions that are clearly social or nonsocial and allows for examination of how specific types of social input relate to EEG power. This paradigm can be used to assess how individual differences in age, affect, socioeconomic status, and parent-infant interaction quality relate to the development of the social brain. Based on the demonstrated functional nature of infant EEG power, future studies should consider the role of EEG recording context and design conditions that are clearly social or nonsocial.

Determining whether social attention is reduced in Autism Spectrum Disorder (ASD) and what factors influence social attention is important to our theoretical understanding of developmental trajectories of ASD and to designing targeted interventions for ASD. This meta-analysis examines data from 38 articles that used eye-tracking methods to compare individuals with ASD and TD controls. In this paper, the impact of eight factors on the size of the effect for the difference in social attention between these two groups are evaluated: age, non-verbal IQ matching, verbal IQ matching, motion, social content, ecological validity, audio input and attention bids. Results show that individuals with ASD spend less time attending to social stimuli than typically developing (TD) controls, with a mean effect size of 0.55. Social attention in ASD was most impacted when stimuli had a high social content (showed more than one person). This meta-analysis provides an opportunity to survey the eye-tracking research on social attention in ASD and to outline potential future research directions, more specifically research of social attention in the context of stimuli with high social content.