The Effects of Human-Robot Interactions and the Human-Robot Relationship on Robot Competence,Trust,and Acceptance

Abstract

As social robots may be used by a single user or multiple users different social scenarios are becoming more important for defining human-robot relationships. Therefore, this study explored human-robot relationships between robots and users in different interaction modes to improve user interaction experience. Specifically, education and companion were selected as the most common areas in the use of social robots. The interaction modes used include single-user interaction and multi-user interaction. The three human-robot relationships were adopted. The robot competence scale, human-robot trust scale, and acceptance of robot scale were used to evaluate subjects’ views on robots. The results demonstrate that in the two scenarios, people were more inclined to maintain a more familiar and closer relationship with the social robot when the robot interacted with a single user. When multiple persons interact in an education scenario, setting the robot to Acquaintance relationships is recommended to improve its competence and people’s trust in the robot. Similarly, in multi-person interaction, Acquaintance relationships would be more accepted and trusted by people in a companion scenario. Based on these results, robot sensors can be added to further optimize human-robot interaction sensing systems. By identifying the number of users in the interaction environment, robots can automatically employ the best human-robot relationship for interaction. Optimizing human-robot interaction sensing systems can also improve robot performance perceived in the interaction to meet different users’ needs and achieve more natural human-robot interaction experiences.

Plain Language Summary

This study explored human-robot relationships between robots and users in different interaction modes to improve user interaction experience. Specifically, education and companion were selected as the most common areas in the use of social robots. The interaction modes used include single-user interaction and multi-user interaction. The three human-robot relationships were adopted. The robot competence scale, human-robot trust scale, and acceptance of robot scale were used to evaluate subjects’ views on robots.

Keywords

acceptance competence human-robot interaction human-robot relationship interaction mode trust

Introduction

As robots are more commonly used in people’s social life, people must understand complex robot-user relationships. Robots do not interact with only one user; they can also interact with multiple users. These social situations are becoming increasingly important to how people perceive human-robot relationships. The single-user interaction mode is generally adopted for HRI-related research. However, as the research went deeper, scholars found that single-user interaction was not the only mode in many situations; multi-user interaction is usually closer to the real situation (Keizer et al., 2014). For example, in the store service field, a robot may interact with customers and sales clerks at the same time in the same interaction task, or an intelligent robot assistant may interact with multiple customers. Such situations may also occur in the service robots at bars or shows (Jeong et al., 2022; Keizer et al., 2013, 2014). Another interesting study was conducted where humanoid robots were set as magic artists to perform magic shows. In the shows, the robots had natural interactions with several audience members. The robots’ personality traits were evaluated during the Interaction. The results show that in the interaction process, the audience felt a strong sense of participation among the robots and had a high evaluation of the robots’ humor (Jeong et al., 2022). This finding made it significant to distinguish human-robot interaction modes. Even in the same industrial environment, people may perceive and evaluate human-robot relationships differently according to the differences in the single-user or multi-user interaction mode. Therefore, in this study, a social robot interacting with a single user was compared to the same social robot interacting with three users in an identical application scenario. The study also evaluated and analyzed whether different interaction modes affect the perceived robot competence and the users’ trust in and acceptance of the robot.

Currently, people have perceived a higher value of human-robot interaction. In particular, better interaction experiences provide warmth and pleasure to users, which has been considered in designing social robots (Kim et al., 2013). Recent studies on multiple interaction modes of robots (Diederich et al., 2019; Fortunati, 2018; Thompson & Gillan, 2016) and on intelligent voice assistants under multi-user conditions (Etzrodt & Engesser, 2021; Lopatovska & Williams, 2018; Porcheron et al., 2018; Purington et al., 2017; Raveh et al., 2019) showed that the number of users changes the interaction with robots or the intelligent agents people perceive and may affect people’s relationship with them. Furthermore, other scholars have discussed human-robot interaction among multiple users. For example, in studies involving a social robot that goes into the classroom, provides medical services, and becomes a team member in a rescue operation (Thompson & Gillan, 2016), or a social robot that mediates family group life in a home environment (Fortunati, 2018). This human-robot interaction happens in a social environment of many people, in which there must be interpersonal relationships between each other. In human-robot interactions, the foundation of human-robot relationships is the perspective of robots as social beings, assigning them social traits and functions, and applies the concept of human-human social interaction to human-robot interaction. This perspective could motivate researchers to explore the possibility of utilizing established human social norms to gain insight into the interactions between people and robots. With the computer as social actor (CASA) theory put forward in the academe, robots are gradually treated as members of society (Nass et al., 1994). This means that in human-robot interactions, people need robots not merely for their functions; they must also treat them as real human beings and maintain a certain social relationship with them. Social relationships refer to the connections that exist between people. A social relationship may be a long-term relationship, a relationship in a group, or a relationship in a new social situation (Simmel, 1923). Therefore, to further obtain design requirements for the best human-robot interaction experience in different interaction modes, a human-robot relationship is categorized into three closeness levels—Familiar, Acquaintance, and Stranger—with reference to interpersonal relationships and is regarded as a variable in our analysis.

Studies have shown that the most effective way to perceive the close relationship between humans and robots is the content of human-robot dialogues (Kim et al., 2013). The setting of dialogue content in human-robot interaction needs further research so that robots can be accepted as social robots in daily life. Notably, verbal communication and expression change with the closeness levels. In interpersonal interactions, verbal behavior not only helps convey information (task-related talk) but also builds relationships (small talk). Therefore, language expression is divided into two categories: task-based dialogue content and small talk. Some scholars mentioned that small talk could be used to build trust and rapport in interpersonal relationships (T. Bickmore & Cassell, 1999). Similar findings have been uncovered in the realm of human-robot interaction (T. Bickmore & Cassell, 1999; T. W. Bickmore & Picard, 2005; Paradeda et al., 2016). Specifically, engaging in small talk has been shown to raised trust in robots that act as in-vehicle navigation assistants (J. M. Kraus et al., 2016). In addition, robots that tell stories were more trusted after making small talk (Paradeda et al., 2016). For human-robot interaction, trust is crucial for evaluating human-robot interaction experience and in judging whether the use of human-robot systems meets users’ needs (Parasuraman & Riley, 1997). The setting of dialogue content for small talk in human-robot interaction affects people’s trust in robots and the robot’s perceived competence (Babel et al., 2021). Additionally, the closeness between people changes with language expression between people (Kim et al., 2013). Therefore, the language communication content between robots and users was set according to the closeness between humans and robots, namely the Familiar relationship (task-related content + small talk + anthropomorphic modal words), Acquaintance relationship (task-related content + anthropomorphic modal words), and Stranger relationship (task-related content).

In human-robot interaction, it is important to evaluate people’s trust in robots for proper system usability (Parasuraman & Riley, 1997). Trust in automated intelligent devices is defined as “the attitude that an agent will help achieve an individual’s goals in a situation characterized by uncertainty and vulnerability” (M. K. Lee & Makatchev, 2009). Some studies pointed out that robot characteristics play a crucial role in developing trust (Hancock et al., 2011; M. Kraus et al., 2018; Miller et al., 2021; Sanders et al., 2011). Robot features are a dependable and foreseeable indicator of performance (Oleson et al., 2011). Examples include human-like appearance (Bartneck et al., 2009) or verbal or nonverbal behavior (Looije et al., 2010). Therefore, there is a link between anthropomorphic design and the trustworthiness of robots, which implies a conceptual correlation between human-robot trust and interpersonal trust (Hoff & Bashir, 2015; J. Kraus et al., 2021; J. D. Lee & See, 2004; Matthews et al., 2019). Recent research has explored how social robots increase trust and acceptance during small talk with users in a human-like manner (Babel et al., 2021). However, studies on dialogue with robots only focused on how the personality traits of a robot influence people’s feeling of it between task-related conversation and small talk (Dou et al., 2021) or the evaluation of how people perceive robot competence based on the different interaction order of task-based dialogues and small talk (Babel et al., 2021). No study focuses on whether there is an interaction effect between the dialogue content due to different human-robot interaction modes and different human-robot relationships. Therefore, the combination of different human-robot interaction modes and different human-robot relationships in specific interaction situations is worth further discussion.

Moreover, previous studies have not focused on the effect of interaction environments on people’s interaction modes. Most social situations are limited in space and time and depend on the existence of other participants in the same space and time. However, different interaction modes affect users’ perceptions of robots. Users have different expectations of the robot in different interaction scenarios (Dou et al., 2020). In order to explore the differences between specific use scenarios, this study selected and discussed the two most common areas of use of social robots, namely education and companionship. Based on the two research variables mentioned above, this study also considered the influence of the interaction modes and human-robot relationships on users’ perceived interaction experience. This study also provides a reference value for the future design of robot sensing systems. Through the sensing device on the social robot, the number of participants in the interaction environment was automatically detected. In addition, the human-robot relationship in the interaction was intelligently adjusted according to the detected data to achieve the best human-robot interaction in different scenarios and tasks.

Research Questions

A: How do interaction modes (independent variable) and human-robot relationships (independent variables) affect people’s trust in (dependent variable) and acceptance of robots (dependent variable) and what impressions do people have on robot competence (dependent variable)?

B: Under different interaction modes (independent variable) in the same scenario, how can a human-robot relationship (independent variable) between robots and users be maintained to improve users’ perception of robot Trust, Competence, and Acceptance (dependent variable)?

Research Goal: By exploring these two questions, we aim to identify the optimal combinations of the human-robot relationship and the interaction mode in education and home companionship scenarios. It will help us to achieve design guidelines for the robot with the optimal human-robot interaction experiences.

Use of Terms

Interaction modes—The number of subjects involved when communicating with the robot. There are two modes. Single-User Interaction: the robot interacts with one participant. Multi-User Interaction: the robot interacts with three participants.

Human-robot relationships—Human-robot interactions can be categorized into three types based on human relationships:

a. Familiar: Familiar refers to people who know each other and have established a certain degree of familiarity and emotional connection. Usually, familiar people have some history of interaction and can engage in more in-depth communication. Their conversation frequency is relatively high, and may cover multiple topics, including personal life, interests, etc.

b. Acquaintance: Acquaintances refer to people who have interacted before but have a relatively superficial relationship compared to familiar people. Although there is some understanding, they have not reached the level of in-depth interaction. Their conversation frequency may be lower, and focuses primarily on specific occasions or topics.

c. Stranger: Strangers refer to people who lack mutual acquaintance or familiarity with each other. There is likely no emotional connection or interaction history established between strangers, so their conversation frequency is usually low and limited to necessary interactions or specific contexts.

The study sets different dialogue content based on the three types of relationships for the experiment: Familiar relationship—Dialogue content: task-related content + small talk + anthropomorphic modal words. Acquaintance relationship—Dialogue content: task-related content+ anthropomorphic modal words. Stranger relationship—Dialogue content: task-related content

Interaction scenarios—The scenarios where robots interact with users in real life. This study selects education and home companionship fields which are the most common application fields for social robots.

Method

Short Overview

This study explores the effects of three human-robot relationships (familiar relationship, acquaintance relationship, and stranger relationship) and two interaction modes (single-user interaction and multi-user interaction) on people’s perceptual evaluation on interactions in two different application fields: education and home companionship.

The experiment adopted the Wizard of Oz method to manipulate the robot, which is the most commonly used control method in current robot experiments. This refers to the robot in the experiment being controlled manually rather than autonomously (Tay et al., 2014). In the experiment, the robot engaged in dialogue interactions with participants. According to the different interaction scenarios of education and home companionship, different dialogue topics were set. Moreover, based on the three human-robot relationship of Familiar, Acquaintance, and Stranger, three different expressions were set. Familiar relationship: task-related content + small talk + anthropomorphic modal words. Acquaintance relationship: task-related content + anthropomorphic modal words, and Stranger relationship: task-related content. In addition, two interaction modes would be performed within the same application scenario, including Single-User interaction where the robot interacted with a single participant, and Multi-User interaction where the robot interacted with three participants (Table 1). Through the interactive experiments, the aim of this study was to examine whether the perceptual evaluation on the robot’s capability, trustworthiness, and acceptability in the same interaction environment would be affected by the interaction mode and the human-robot relationship. It also summarized the optimal combination of interaction modes and human-robot relationships.

Table 1.

Use of Terms and Detail.

Scenario	Education			Home companionship
Interaction mode	Single-user		Multi-user	Single-user	Multi-user
Human-robot relationship	Familiar	Acquaintance	Stranger	Familiar	Acquaintance	Stranger
Dialogue topic	Starlink project interaction			Headache relief methods interaction

Participants

There were 60 subjects were enlisted in this study, with ages between 20 and 60 years old (27 males and 33 females). Among them, 30 participants took part in the education interaction scenario and 30 valid questionnaires were collected (Male = 14, Female = 16) (Male average age [M] = 23.57, standard deviation [SD] = 3.131; Female average age [M] = 23.79, standard deviation [SD] = 3.606). Thirty participants took part in the home companionship interaction scenario and 30 questionnaires were collected (Male = 13, Female = 17) (Male average age [M] = 50.77, standard deviation [SD] = 6.990; Female average age [M] = 53.29, standard deviation [SD] = 5.753). The subjects had a rudimentary comprehension of social robots (have seen them or interacted with them before). Participants in the education environment were undergraduate art and design students who were from universities in Jiangsu Province, China and Taiwan and participated voluntarily. Participants in the home companionship scenario were some retired teachers who were from a university in Jiangsu Province, China and also participated voluntarily.

Robot Subsection

This study made use of the anthropomorphic NAO robot from SoftBank Robotics. This robot was manufactured by SoftBank, which is the most prevalent humanoid robot employed in contemporary studies. The Nao robot is 58 cm in height, with 25 degrees of freedom and over 100 sensors. It has human-like features with a head, body, and limbs. Nao can be programmed using the graphical programming tool Choregraphe. In this study, the light color expressions of the robot’s head were set using Choregraphe. A small Bluetooth speaker was attached to the back of the NAO robot to facilitate the experiment, and a preset dialogue was played for different scenarios.

For the setting of robot voices, the text-to-speech software created by iFlytek Co., Ltd. was adopted in this study. The system was used to produce robot voice samples (Ling et al., 2007) in two experimental environments: education and companion. iFlytek synthetic speech is one of the Chinese synthetic speech systems with the best functions at present. Its synthetic speech is slightly richer in emotion compared to the commonly used Chinese synthetic speech systems, such as Miss Google and Mac computer speech. According to the team’s previous research, the male voice is the first choice in the education scenario, while the female voice is more accepted in the companion scenario (Dou et al., 2022). Therefore, in this study, the robot voice in the education scenario employed the iFlytek Qige’s gentle male voice package, whereas the iFlytek Xiaoying’s voice package intellectual female voice was selected for the companion scenario. The robot voice volume was set to 7, with the maximum volume being 10 and the speech rate was set to the normal.

Task

Each application field consists of two elements: interaction mode and human-robot relationship. The education scenario was set as follows: Assuming the role of a teacher, the robot teacher would answer a certain question (e.g., What is the Starlink Project?). Students would pose inquires and confer with the robotic teacher. The companion scenario was set as follows: Assuming the role of a family healthcare worker, the robot answered questions and made suggestions to remedy headaches. The specific dialogues are shown in Table 3. The humanoid Nao was set on a 0.5 m high table, with a 0.05 m distance from the edge of the table. The interaction distance was set within 0.75 and 0.8 m (Topp et al., 2006). To minimize the impact of the robot’s own non-verbal expressions on the participants’ perceptual evaluation during the interaction, the robot was set to have no hand gestures in the study. Based on our team’s research results on light colors (Dou et al., 2022), the robot’s head and eye light colors were set to blue cool light for the education application scenario, and yellow warm light for the home companionship application scenario. During the interaction, the robot faced the participant at all times. The participants were arranged to sit on a 0.4 m high chair to interact with the robot. The participants would not see the robot operator and would not be affected by the experimenter in the interaction process.

Environment/Scenarios

To avoid potential confounding factors in a real-life ambience, the experiment was performed in a 56 square meter laboratory set up to resemble two different contexts (an education scenario with a blackboard, desks and stools, a home companion scenario with a sofa and a coffee table). During the experiment, one robot operator controlled the robot’s voice while hidden from the participant’s view behind a cabinet. There was also one experimenter present (Figure 1a and b).

Figure 1.

(a) Single-user interaction scenario and (b) multi-user interaction scenario.

Experimental Design

The experiment utilized a 3 × 2 mixed design, with the within-subject variables including human-robot relationships (three levels: Familiar relationship, Acquaintance relationship, and Stranger relationship) and human-robot relationships (two levels: single-user interaction and multi-user interaction). The dependent variable was Robots’ Competence, Trust, and Acceptance, as determined using a 7-point Likert-scale. The independent and dependent variables are presented in Table 2.

Table 2.

Summary of the Dependent and Independent Variables.

Variable	Type	Value
Human-robot relationships	Independent	Three levels: familiar relationship, acquaintance relationship, and stranger relationship
Human-robot relationships	Independent	Two levels: single-user interaction and multi-user interaction
Competence, trust, and acceptance	Dependent	Total of 1–7 points

Experimental Setup

During the interaction, participants and the robot remained in fixed positions while conversing. The conversational content is shown in the Table 3. For the education scenario, 30 students were recruited. Fifteen participated in the single user interaction mode, engaging in one-on-one human-robot interaction. The other 15 were divided into five groups of three for the multi-user interaction. For the home companion scenario, 30 retired university professors were recruited to take part in. Again 15 did the single user interaction, while 15 did the multi-user interaction. Senior adults were chosen for the home companionship scenario because our surveys found the main users of home healthcare robots were seniors, so retired professors served as appropriate subjects. Prior to the experiment, participants completed a demographic questionnaire with information like age, gender and experience of using robots. Then they interacted with the robot under three relationships, and completed a questionnaire after each interaction before moving to the next one. The total experiment time was around 25 min. Participants received a 200 TWD cash reward upon completion.

Table 3.

Specific Contents of Dialogue Interaction.

Human-robot relationship	Education scenario dialogue script settings (R: Robot/P: Participant)	Home companionship scenario dialogue script settings (R: Robot/P: Participant)
Familiar relationship: (task-related content + small talk + anthropomorphic modal words)	R: Hi! What are you thinking about? P: Do you know what the Starlink project is? R: Of course, I know! It’s the plan unveiled in 2015 by SpaceX, the American aerospace manufacturer founded by Elon Musk! I heard they plan to launch 12,000 small satellites into low Earth orbit by 2024. Through communication between the satellites, they want to build a satellite network that covers the globe. His goal is to compete in the satellite internet market! Can you imagine that kind of scene? P: What a huge project! R: I know, right! Musk is so impressive! P: Looking forward to this global satellite network in the future! R: Yeah! That means in the future, if our devices connect to the Starlink satellites, we’ll have stable internet even if we’re stranded on some remote Pacific island. I really look forward to being able to get high-speed internet signals everywhere!	R: Hi! You don’t seem to be feeling well, are you uncomfortable? P: I have a bit of a headache today. R: Is your headache severe? Did you not rest well last night? Recently there have been big temperature differences between indoors and outdoors, did you catch a chill from the air conditioning in your room yesterday? I heard indigestion can also cause headaches, is that what’s causing yours? P: I probably just slept too late last night. R: You could try massaging your temples first to see if that relieves it! If there’s no improvement you could close your eyes and rest for a bit! Remember to go to bed early tonight! See if you feel better tomorrow! P: Okay, got it. R: Mm-hmm, make sure to get enough sleep every day, and try to keep a regular schedule! If it’s still not better tomorrow, you can go to the hospital and get checked out, ask the doctor. If you need to go to the hospital, I can help you make an online appointment! P: Thank you! R: Take care of yourself, let me know if you have any issues, I’ll always be by your side!
Acquaintance relationship: (task-related content + anthropomorphic modal words)	R: Have you encountered any issues? P: Do you know what the Starlink project is? R: I know the one you mentioned. The Starlink project is a plan that the American aerospace manufacturer SpaceX is to launch tens of thousands of small satellites into low Earth orbit to provide global internet connectivity. P: What a huge project! R: Yes. I think so too! P: Wonder if they can pull off the plan? R: Let’s look forward to it together!	R: Have you encountered any issues? P: I have a bit of a headache today. R: There are many possible causes of headaches, including colds, poor diet, lack of sleep, and illness. What’s causing yours? P: I probably just slept too late last night. R: The temples are important pressure points for headaches. You could try massaging around that area to relieve the headache. P: Okay, got it. R: Alright, be sure to rest early tonight. If it’s not better by tomorrow, I suggest seeing a doctor at the hospital. P: Thank you! R: You’re welcome, take care of yourself!
Stranger relationship: (task-related content)	R: How may I help you? P: Do you know what the Starlink project is? R: The Starlink project is a global satellite internet access service planned by the American aerospace manufacturer SpaceX. P: What a huge project. R: Yes, do you have any other questions? P: No, thank you! R: You’re welcome, goodbye!	R: How may I help you? P: I have a bit of a headache today. R: There are many possible causes for headaches, such as neurological or physical factors. P: I probably just slept too late last night. R: Getting adequate sleep is very important. P: Okay, got it. R: Yes, do you have any other questions? P: Thank you! R: You’re welcome, goodbye!

Procedure

The experimental protocol in this study utilized a theater-based HRI (THRI) approach (Chatley et al. et al., 2010; Nørskov, 2017), in which the investigator interacted with the robot following a predefined script to enact dialog scenarios for the participants (audience members) during the experiment. Employing a theater-based HRI (THRI) methodology prevents shifts in perception that could arise from the specific content of a participant’s personal dialogue with the robot.

Before the experiment, we assigned participants to groups. For the single user interaction, each group had only one participant. For the multi-user interaction, each group had three participants. Each group was assigned a time slot to participate in the experiment. Two minutes before the start of the experiment, all participants were briefed on the study details and their involvement, and gave informed consent by endorsing a form. Afterwards, the participants filled out a basic questionnaire covering demographic details including age and gender.

Once baseline information had been offered, the subjects began the conversational interaction experiment with the robot (Table 3). In the single-user interaction mode, the robot randomly used one of three human-robot relationships to communicate with the subject, with each exchange lasting less than 3 min. Following the dialogues, participants were requested to fill out a questionnaire. The subjects were instructed to appraise the robot’s competence, trustworthiness, and acceptability based on their interactions. Subsequent to finishing the questionnaire, the robot randomly transitioned to a different human-robot relationship to converse with the participants. Following this conversation, the subjects completed a questionnaire once more. This procedure was iterated until subjects had filled out surveys evaluating all three human-robot relationships. Every subject was requested to fill out a sum of three questionnaires on three human-robot relationships. In multi-user interaction mode, the experimental procedure was the same.

Evaluation Method

The scale used in this research is comprise of three primary components: competence, trust, and acceptance. The evaluation scale of robot competence can be obtained from relevant literature (Babel et al., 2021; Berlo et al., 1969; Gong, 2008; McCroskey et al., 1974, 1975; Ohanian, 1990), which contains six items (1. unintelligent–intelligent; 2. knowledgeable–unknowledgeable; 3. incompetent–competent; 4. uninformed–informed; 5. inexpert–expert; and 6. experienced–inexperienced). The evaluation scale of trust can be obtained from the literature (Gong, 2008; Jian et al., 2000; Wheeless & Grotz, 1977; Zhao & Rau, 2020), which consists of 12 items (1. The robots are deceptive; 2. The robots behave in an underhanded manner; 3. I am suspicious of the robots’ intent, actions, or outputs; 4. I am wary of robots; 5. The robots’ actions will have a harmful or injurious outcome; 6. I am confident with the robots; 7. The robots provide security; 8. The robots have integrity; 9. The robots are dependable; 10. The robots are reliable; 11. I can trust robots; and 12. I am familiar with robots). The robot acceptance scale contains nine items (1. useful–useless 2. pleasant–unpleasant; 3. bad–good; 4. nice–annoying; 5. effective–superfluous; 6. irritating–likable; 7. assisting–worthless; 8. undesirable–desirable; and 9. raising alertness–sleep-inducing), which can be obtained from the literature (Babel et al., 2021; Van Der Laan et al., 1997). There were 37 items in total. The seven-point Likert Scale was used to appraise the variables, and all questions were translated into Chinese.

Result

Reliability Analysis

The Cronbach’s alpha coefficients for the three components in the educational scenario all exceeded .8 (robot competence α = .95, trust α = .90, and acceptance α = .93). On the other hand, the Cronbach’s alpha coefficients of the three elements in the companion scenario were also higher than .8 (robot competence α = .88, trust α = .91, and acceptance α = .96), indicating high reliability.

Descriptive Results

Education Scenario

In the competence factor, in single-user interaction mode, significant differences were found between the Familiar relationship (M = 6.12, SD = 0.69) and Acquaintance relationship (M = 5.13, SD = 0.73) and between the Familiar relationship and Stranger relationship (M = 2.51, SD = 0.67). Notable differences were also detected between the Acquaintance relationship and the Stranger relationship. This result indicates that the Familiar relationship achieved the top appraisal in the competence factor, followed by the Acquaintance relationship.

It demonstrated that in the education environment, when the robot teacher interacts with a single user, the user tends to perceive the robot teacher as having a close relationship with them. The interaction allows the user to focus fully on the robot without being influenced by others. This enables the participant to become more immersed in the interaction process with the robot. As a result, they give higher perceptual evaluations on the robot’s capabilities and prefer a more intimate expression style from the robot.

In the multi-user interaction mode, significant differences were found between the Acquaintance relationship (M = 5.81, SD = 0.89) and Familiar relationship (M = 5.47, SD = 0.60) and between the Acquaintance relationship and Stranger relationship (M = 2.67, SD = 0.81). Notable differences were also detected between observed between the Familiar relationship and the Stranger relationship. This finding indicates that the Acquaintance relationship gained the highest appraisal in the competence factor in the multiple-user interaction mode. When there is more than one participant in the interaction, the feelings of the interactors can influence each other. In the multi-user interaction mode, the robot interacts with three participants simultaneously. Since the participants are strangers to each other, this may affect people’s perceptual evaluation on the robot during the interaction.

In the trust factor, in single-user interaction mode, significant differences were found between the Familiar relationship (M = 6.29, SD = 0.65) and Acquaintance relationship (M = 5.54, SD = 0.72), and between the Familiar relationship and Stranger relationship (M = 4.13, SD = 0.75). Notable differences were also detected between observed between the Acquaintance relationship and the Stranger relationship. This result indicates that the Familiar relationship achieved the highest score in the trust factor, followed by the Acquaintance relationship. When the robot teacher interacts with a single participant, the more familiar relationship improves people’s perceived trust toward the robot. The participant feels closer to the robot teacher because of its more intimate expression.

In the multi-user interaction mode, there existed significant differences between the Acquaintance relationship (M = 5.52, SD = 0.99) and Familiar relationship (M = 5.34, SD = 0.39), and between the Acquaintance relationship and Stranger relationship (M = 3.96, SD = 0.67). Significant divergences were also examined between the Familiar relationship and the Stranger relationship. This finding indicates that the Acquaintance relationship had the highest appraisal in the trust factor in the multiple-user interaction mode. Consistent with the findings on perceived robot capability, the familiar relationship improves participants’ evaluation on the robot’s trustworthiness in the multi-user interaction mode.

In the acceptance factor, in the single-user interaction mode, significant differences were observed between the Familiar relationship (M = 5.76, SD = 1.35) and Acquaintance relationship (M = 5.30, SD = 1.00) and between the Familiar relationship and Stranger relationship (M = 2.95, SD = 0.83). Remarkable differences were also detected between the Acquaintance relationship and the Stranger relationship. This result indicates that the Familiar relationship gained the best estimation in the acceptance factor in the single-user interaction mode. Similarly, in multi-user interaction mode, the Familiar relationship obtained the highest acceptance, and significant differences were found between the Familiar relationship (M = 5.96, SD = 0.44), Acquaintance relationship (M = 4.30, SD = 1.01), and Stranger relationship (M = 3.14, SD = 0.45). Regarding the overall acceptance of the robot teacher, under both interaction modes, acceptance ratings were highest in the familiar relationship condition. This result is because people prefer the robot teacher to use an intimate and familiar communication style when conversing with them. When the robot teacher answers questions in a friendly manner, it breaks the stereotype of teachers being serious, making students feel more at ease.

Companion Scenario

The familiar relationship showed significant differences in three dimensions of robot competence (M = 6.37, SD = 0.41), trust (M = 6.43, SD = 0.90), and robot acceptance (M = 5.63, SD = 1.36) in single-user interaction mode. This finding indicates that the Familiar relationship obtained higher evaluation in the emotional experience of the robot in the single-user interaction mode. Based on the marks of acceptance of the robot in the multi-user interaction mode, significant differences were detected between the Acquaintance relationship (M = 5.98, SD = 0.88), Familiar relationship (M = 4.36, SD = 1.35), and Stranger relationship (M = 2.16, SD = 0.71). This finding indicates that the Acquaintance relationship obtained the highest evaluation in the Acceptance factor in the multiple-user interaction mode. In the home companionship scenario, the familiar relationship was preferred by elderly participants in one-on-one interactions. This aligns with people’s tendency to perceive a companion robot as an intimate character in the private indoor space. Whereas in multi-user interactions, due to the interpersonal influence between participants, the unfamiliarity between some groups of participants led to varying evaluation results. In future research, the degree of familiarity between participants can be investigated separately under multi-user interactions

Main Effects on Competence, Trust, and Acceptance

An evident main effect was observed in the interaction mode and human-robot relationship, in which the interaction mode Wilklambda F = 5.599, p = .002; human-robot relationship Wilklambda F = 34.765, p = .000.

The MANOVA results are shown in Table 4. In the education scenario, obvious differences were found between interaction modes in principal components, namely trust [F (1, 2) = 6.37, p < .01] had high significance levels. This finding indicates that differences in human-robot interaction modes have a significant effect on people’s trust in robots. In addition, the human-robot interaction mode determines whether a robot meets the service needs and positioning of the education scenario. Therefore, it is necessary to study different interaction modes in the same scenario in the current research on human-robot interaction, and further in-depth discussion is needed. A significant effect was observed on the perceived robot competence [F (2, 2) = 171.26, p = .000], trust [F (2, 2) = 52.89, p = .000], and acceptance [F (2, 2) = 81.75, p = .000) in human-robot relationships. This finding indicates that for education robots, different human-robot relationship performances in interaction affect three dimensions. The comprehensive effect of the three dimensions provides a specific design reference for the setting of human-robot interaction in education scenarios and shows that human-robot relationships are an important parameter affecting participants’ emotional experience of robots.

Table 4.

MANOVA Results of Interaction Mode × Human-Robot Relationships.

Factors of dependent variables	F values and effect sizes ( $η_{p}^{2}$ )
	Main effect					Interaction effect
		Interaction mode		Human robot relationship		Interaction mode × Human robot relationship
		F	$η_{p}^{2}$	F	$η_{p}^{2}$	F	$η_{p}^{2}$
Competence	Education scenario	0.145	0.002	171.255**	0.803	6.203*	0.129	1v1:F > A > S/1v3:A > F > S
Competence	Companion scenario	8.643*	0.093	154.757**	0.787	14.213**	0.253	1v1:F > A > S/1v3:F > A > S
Trust	Education scenario	6.369*	0.070	52.885**	0.557	3.682*	0.081	1v1:F > A > S/1v3:A > F > S
Trust	Companion scenario	0.281	0.003	83.061**	0.664	15.006**	0.263	1v1:F > A > S/1v3:A > F > S
Acceptance	Education scenario	1.667	0.019	81.751**	0.661	0.452	0.011	1v1:F > A > S/1v3:F > A > S
Acceptance	Companion scenario	2.419	0.028	56.177**	0.572	9.869**	0.190	1v1:F > A > S/1v3:A > F > S

Note. F = familiar relationship; A = acquaintance relationship; S = stranger relationship; 1v1 = single-user interaction; 1v3 = multi-user interaction.

p < .05. **p < .001.

In the companion scenario, significant differences were noted in the competence factor in the human-robot interaction mode [F (1, 2) = 8.64, p < .01], indicating high significance levels. This observation suggests that different human-robot interaction modes affect participants’ perceptions of robot competence. Human-robot relationships were significant in the perceived robot competence [F (2, 2) = 154.76, p = .000], trust [F (2, 2) = 83.06, p = .000] acceptance [F (2, 2) = 56.18, p = .000). This finding indicates that setting the relationship between robots for companions and the user is becoming particularly important.

Interaction Effects

Apart from the main effect of interaction mode and human-robot relationship have an interaction effect, with obvious effects in perceived competence, trust, and acceptance, respectively (Figures 2 and 3).

Figure 2.

Interaction mode and human-robot relationships result in education scenario.

Figure 3.

Interaction mode and human-robot relationships result in companion scenario.

The education robot was tested according to the simple main effect (Figure 2). The results indicate an interaction effect between interaction mode and human-robot relationship in competence [F = 6.203, p = .003 < .01] and trust [F = 3.682, p = .025 < .05]. In the education environment, interaction modes combined with different human-robot relationships influenced the perceived capabilities and trustworthiness of the robot teacher. However, for robot acceptance, the interaction effect was not discovered between interaction modes and human-robot relationships. Additionally, the unfamiliar interpersonal relationship received the lowest evaluations across both interaction modes.

The companion robot was tested according to the simple main effect. The results indicate an interaction effect between the interaction mode and human-robot relationship in competence, [F = 14.213, p = .000 < .001], trust [F = 15.006, p = .000 < .001], and acceptance [F = 9.869, p = .000 < .001]. Whereas in the home companionship scenario, the combination of interaction modes and human-robot relationships significantly influenced the perceptual evaluation across the three dimensions of the robot

Discussion

How Do Interaction Modes and Human-Robot Relationships Affect People’s Trust in and Acceptance of Robots, and What Impressions Do People Have on Robot Competence?

It is found that human-robot interaction in education scenarios affects the evaluation of the trust factor in the interaction process. Single-user interaction offers the best trust experience compared to multi-user interaction. This result is attributed to higher closeness when the interaction occurs only between two parties because a particular issue is believed to be shared only between two parties. This observation indicates that both parties experience a strong affinity and reciprocal cognition, arising from their profound affective aspects in primary relationships (Lenz, 2009). Because both partners are accountable for the relationship’s events, they reveal considerable interplay and do not hide their actions or inactions from the other (Etzrodt, 2022). Therefore, the human-robot interaction mode in the companion scenario can affect the evaluation of the competence factor in the interaction process. The participants in the single-user interaction mode had a better appraisal of the perceived robot competence. This result is similar to the findings of previous research. Scholars mentioned that the participants rated intelligent voice assistants to have higher concern and comprehension abilities in one-on-one interactions than in multi-person interactions (Etzrodt, 2021). Meanwhile, other scholars have found that in interaction with social robots, one-on-one interaction could improve rich behaviors and interaction participation and lead to higher satisfaction of human participants on interaction (Fortunati et al., 2020).

In the education and companion environments, human-robot relationships had significant effects on perceived competence, trust, and acceptance. For education robots, participants maintained a Familiar relationship with the robot and had a higher evaluation regarding robot competence and trust. Hence, they found the robot more acceptable because apart from letting the robot answer the participants’ questions, this study set more small talk and some anthropomorphic modal words in the Familiar relationship in this experiment. This result is similar to previous research findings. Some scholars have found that in human-robot interactions, when small talk is added to the dialogue content of robots that have completed tasks, trust in robots can be increased (Paradeda et al., 2016; Salem et al., 2013). In the Familiar relationship, interviews with the participants revealed that when a human-like language style was set for the dialogue content (modal words and interjections), the participants felt that they were interacting with real human beings in the dialogue. Furthermore, their perception of the robot would be more anthropomorphic. However, no existing studies have found that closer dialogue content improves the perceived robot competence. Nevertheless, it is an interesting finding in this experiment. In terms of perceived competence and trust, the Familiar relationship obtained a significantly higher evaluation. According to interviews with participants in companion scenarios, companion-related topics are solutions and suggestions to troublesome problems. Participants thought that the robot’s answers to questions were more concise in the Acquaintance relationship, and it was irritating when small talk was added in close relationships. However, in the Stranger relationship, the robot’s attitude would give people a cold feeling. Therefore, the Acquaintance but unfamiliar relationship obtained the highest acceptance. Because the variables were controlled in this experiment, the robot took the initiative in the dialogue interaction. Moreover, studies have shown that human beings have the right to initiate a dialogue in human-robot interaction; thus, the power should be equally distributed. When the power is asymmetrical, such dialogue content may not be accepted (Capurro, 2019; Jarrasse et al., 2014; Reinhardt et al., 2017), particularly in the interaction with small talk (Babel et al., 2021).

Under Different Interaction Modes (Single-User Interaction and Multi-User Interaction) in the Same Scenario, How Can a Human-Robot Relationship Between Robots and Users Be Maintained to Improve Users’ Interaction Experience?

In both the education and companion scenarios in the single-user interaction mode, participants preferred to maintain a Familiar relationship with the robot. The Familiar relationship obtained the highest score in perceived competence, trust, and acceptance. This result indicates that people feel closer to each other in one-to-one interactions (Etzrodt, 2022). The Acquaintance but unfamiliar relationship can be used as the second choice, in which an anthropomorphic tone is set in dialogue content. Through interviews with participants, this study revealed that people believed that service robots should be able to solve problems first and then express emotions. Therefore, to increase participants’ perception, trust, and optimal acceptance toward robot competence in one-to-one interaction, the Familiar relationship should be prioritized in the relationship design. This means that task-related content, small talk, and anthropomorphic modal words should be included in the design of dialogue content.

In the multi-user interaction mode, when the robot interacted with multiple people in the education scenario, the Acquaintance relationship obtained a high perceived competence and trust evaluation. The same interpersonal relationship in the companion scenario can improve participants’ perceptions and evaluations of trust in robots and acceptance. Such a result can be explained by the three-way interaction theory. Some scholars suggest that compared with one-to-one interactions, the close relationship between the two will be weakened when a third party intervenes in the interaction between multiple people. Meanwhile, the interaction between the two becomes common due to the presence of a third person or more people in the scenario (Etzrodt, 2021, 2022). In the two application scenarios, participants in the same group knew each other and shared their feelings in the evaluation. Hence, this study suggests that this also has a mutual influence on other participants’ evaluations of the robot. No significant difference was observed in evaluating the robot in the Familiar relationship and the relationship between Acquaintance and Familiar in the education scenario. However, in the companion scenario, the Familiar relationship increased the perception of robot competence because the robot could help users solve problems in a Familiar relationship. Additionally, the small talk’s dialogue content was added to enrich the exchange. Additionally, when the content of small talk appeared in the dialogue, it improved the closeness between the interactors, thereby narrowing the distance between them (Argyle & Dean, 1965; Babel et al., 2021) so that users felt a closer interaction between them, particularly in the private space in the companion scenario. Therefore, it is suggested that the one-to-one interactive human-robot relationship in education and companion scenarios should be set as the best user experience in the Familiar relationship. In the one-to-many interaction mode, the Acquaintance relationship is the best choice to improve the perceived robot competence and people’s trust in robots. Notably, setting a Familiar relationship to improve the acceptance of education robots is recommended. In the companion application scenario, it is recommended to set the human-robot relationship as a Familiar relationship to improve the perceived competence. On the other hand, an acquaintance relationship is the best choice to improve trust and acceptance—this is also a new finding on the interaction mode and human-robot interaction in this study.

Limitation and Future Work

Some limitations must be considered to do precise research on social robots. First, the three human-robot relationships were set based on different dialogues with robots, namely the Familiar relationship (task-related content + small talk + anthropomorphic modal words), Acquaintance relationship (task-related content + anthropomorphic modal words), and Stranger relationship (task-related content). However, when users interact with social robots in a specific space in a real scene, the closeness between users and social robots will be affected by human-robot location and social distance. These related conditions can be considered in future research. Furthermore, through brief interviews with the subjects, slight differences were noted between male and female users in the acceptance of a close relationship with the robot. However, the influence of the participants’ gender was not considered in this experiment. Second, in the discussion of multi-user interaction, this study arranged three participants in each group to interact with the robot. Due to the limitation of experimental conditions, only one participant had a conversation with the robot during the interaction. Meanwhile, the other two participants evaluated robot competence, trust, and acceptance as bystanders. However, in the real interactive environment, everyone has different needs; each participant will put forward their respective needs. This aspect can be further discussed in future research. Third, with regard to judging the social robot’s competence, trust, and acceptance, in the interaction between participants and the social robot, only the robot could initiate dialogues in the experiment. However, in interpersonal dialogue, the initiative of dialogue (who first raises the question) will also impact individual’s confidence and acceptance of the robot (Babel et al., 2021). Fourth, the anthropomorphic robot (Nao) used in the study has the following limitations regarding anthropomorphism: (1) The robot’s anthropomorphism may have caused participants to overestimate the robot’s competence and intelligence. However, in this study, participants only engaged in scripted dialogues with fixed content, which may have misled participants’ perceptions of the robot’s capabilities. Subjects may have realized that the robot was not as intelligent as imagined, or even a bit silly. (2) The anthropomorphism may create confusion in real interactions, leading to communication issues or unrealistic expectations. The robot in the study had human-like features (head, body, limbs). To minimize the influence of other nonverbal behaviors on perceived evaluations, the robot in the study was stationary in an upright position. This also greatly reduced participants’ expectations for realistic interactions. In addition, some nonverbal behaviors of robots, such as the direction of eye gaze (eye contact), will also affect the evaluation of people’s trust in social robots in human-robot interaction (Nomura & Kanda, 2015). Future research can further explore whether these factors affect the interaction mode and human-robot relationship in this study.

Conclusion

Based on the education and companion scenarios, this study explored the best combination of human-robot interaction and human-robot relationship in the same application scenario. The results demonstrated that in the education and companion scenarios, in the single-user interaction mode, the user maintained a more familiar relationship with the social robot and had a higher evaluation of the perceived robot competence, trust in, and robot acceptance. Therefore, the user can have the best interactive experience when a Familiar relationship is set in the one-to-one interaction mode. In multi-user interaction, it is recommended to set the robot to an Acquaintance relationship to improve the perceived robot competence, trust in, and acceptance of the robot in the education environment. In this relationship, the robot teacher’s answers to education-related questions were more direct and expressed in an anthropomorphic tone. However, Familiar relationships obtained higher acceptance in the communication with robots. In the companion scenario, Acquaintance relationships (that is, the dialogue only included task-related content and anthropomorphic modal words) obtained higher evaluation in acceptance and trust. However, to improve the competence of the companion robot, the best choice is to set a Familiar robot-user relationship.

Moreover, this study presents a comprehensive discussion of two distinct interaction modes and three human-robot relationships in the same application scenario in a real environment to shrink the margin between theoretical work on social robots and practical applications. It is also helpful for future customized designs of social robots in various application fields to better meet users’ needs and interaction experience.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by Ministry of Science and Technology, Taiwan (111-2410-H-036-002-MY2) and supported by the Project of Philosophy and Social Science Research in Colleges and Universities in Jiangsu Province (No. 2022SJYB1514).

ORCID iD

Yue Yuan

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

References

Argyle

Dean

(1965). Eye-contact, distance and affiliation. Sociometry, 28(3), 289–304.

Babel

Kraus

Miller

Kraus

Wagner

Minker

Baumann

(2021). Small talk with a robot? The impact of dialog content, talk initiative, and gaze behavior of a social robot on trust, acceptance, and proximity. International Journal of Social Robotics, 13(6), 1485–1498.

Bartneck

Kulić

Croft

Zoghbi

(2009). Measurement instruments for the anthropomorphism, animacy, likeability, perceived intelligence, and perceived safety of robots. International Journal of Social Robotics, 1, 71–81.

Berlo

D. K.

Lemert

J. B.

Mertz

R. J.

(1969). Dimensions for evaluating the acceptability of message sources. Public Opinion Quarterly, 33(4), 563–576.

Bickmore

Cassell

(1999). Small talk and conversational storytelling in embodied conversational interface agents [Symposium]. AAAI Fall Symposium on Narrative Intelligence.

Bickmore

T. W.

Picard

R. W.

(2005). Establishing and maintaining long-term human-computer relationships. ACM Transactions on Computer-Human Interaction (TOCHI), 12(2), 293–327.

Capurro

(2019). Ethical issues of humanoid-human interaction. In Goswami

Vadakkepat

(Eds.), Humanoid robotics: A reference (pp. 2421–2435). Springer. https://doi.org/10.1007/978-94-007-6046-2_127

Chatley

A. R.

Dautenhahn

Walters

M. L.

Syrdal

D. S.

Christianson

(2010, February 10–15). Theatre as a discussion tool in human-robot interaction experiments—A pilot study [Conference session]. 2010 Third International Conference on Advances in Computer-Human Interactions, Saint Maarten, The Netherlands.

Diederich

Brendel

A. B.

Kolbe

L. M.

(2019). On conversational agents in information systems research: Analyzing the past to guide future work [Conference session]. Proceedings of Internationale Tagung Wirtschaftsinformatik, Siegen, Germany [VHB 3: C].

10.

Dou

C.-F.

Lin

K.-C.

Gan

Tseng

T.-M.

(2021). Effects of different types of social robot voices on affective evaluations in different application fields. International Journal of Social Robotics, 13, 615–628.

11.

Dou

C.-F.

Niu

Pan

K.-R.

(2022). Effect of voice type and head-light color in social robots for different applications. International Journal of Social Robotics, 14(1), 229–244.

12.

Dou

C.-F.

Wang

Niu

(2020, July 19–24). User expectations of social robots in different applications: An online user study [Conference session]. HCI International 2020—Late Breaking Papers: Multimodality and Intelligence: 22nd HCI International Conference, HCII 2020, Copenhagen, Denmark.

13.

Etzrodt

(2021, November 23–24). The ontological classification of conversational agents: An adaptation of Piaget’s equilibration theory [Paper presentation]. Paper presented at the Chatbot Research and Design: 4th International Workshop, Conversations 2020, Virtual Event, 2020.

14.

Etzrodt

(2022). The third party will make a difference—A study on the impact of dyadic and triadic social situations on the relationship with a voice-based personal agent. International Journal of Human-Computer Studies, 168, 102901.

15.

Etzrodt

Engesser

(2021). Voice-based agents as personified things: Assimilation and accommodation as equilibration of doubt. Human-Machine Communication, 2, 57–76.

16.

Fortunati

(2018). Robotization and the domestic sphere. New Media & Society, 20(8), 2673–2690.

17.

Fortunati

Cavallo

Sarrica

(2020). Multiple communication roles in human–robot interactions in public space. International Journal of Social Robotics, 12(4), 931–944.

18.

Gong

(2008). How social is social responses to computers? The function of the degree of anthropomorphism in computer representations. Computers in Human Behavior, 24(4), 1494–1509.

19.

Hancock

P. A.

Billings

D. R.

Schaefer

K. E.

Chen

J. Y.

De Visser

E. J.

Parasuraman

(2011). A meta-analysis of factors affecting trust in human-robot interaction. Human Factors, 53(5), 517–527.

20.

Hoff

K. A.

Bashir

(2015). Trust in automation: Integrating empirical evidence on factors that influence trust. Human Factors, 57(3), 407–434.

21.

Jarrasse

Sanguineti

Burdet

(2014). Slaves no longer: Review on role assignment for human–robot joint motor action. Adaptive Behavior, 22(1), 70–82.

22.

Jeong

Yang

Baltes

(2022). Robot magic show as testbed for humanoid robot interaction. Entertainment Computing, 40, 100456.

23.

Jian

J.-Y.

Bisantz

A. M.

Drury

C. G.

(2000). Foundations for an empirically determined scale of trust in automated systems. International Journal of Cognitive Ergonomics, 4(1), 53–71.

24.

Keizer

Foster

M. E.

Lemon

Gaschler

Giuliani

(2013). Training and evaluation of an MDP model for social multi-user human-robot interaction [Paper presentation]. Paper presented at the Proceedings of the 14th Annual SIGdial Meeting on Discourse and Dialogue.

25.

Keizer

Kastoris

Foster

M. E.

Deshmukh

Lemon

(2014). Evaluating a social multi-user interaction model using a Nao robot [Symposium]. The 23rd IEEE International Symposium on Robot and Human Interactive Communication.

26.

Kim

Kwak

S. S.

Kim

M.-S.

(2013). Am I acceptable to you? Effect of a robot’s verbal language forms on people’s social distance from robots. Computers in Human Behavior, 29(3), 1091–1101.

27.

Kraus

Scholz

Baumann

(2021). What’s driving me? Exploration and validation of a hierarchical personality model for trust in automated driving. Human Factors, 63(6), 1076–1105.

28.

Kraus

J. M.

Nothdurft

Hock

Scholz

Minker

Baumann

(2016). Human after all: Effects of mere presence and social interaction of a humanoid robot as a co-driver in automated driving [Conference session]. Adjunct Proceedings of the 8th International Conference on Automotive User Interfaces and Interactive Vehicular Applications.

29.

Kraus

Baumann

Minker

(2018). Effects of gender stereotypes on trust and likability in spoken human-robot interaction [Conference session]. Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018).

30.

Lee

J. D.

See

K. A.

(2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1), 50–80.

31.

Lee

M. K.

Makatchev

(2009, April 4–9). How do people talk with a robot? An analysis of human-robot dialogues in the real world [Conference session]. Proceedings of the 27th International Conference on Human Factors in Computing Systems, CHI 2009, Extended Abstracts Volume, Boston, MA, United States (pp. 3769–3774).

32.

Lenz

(2009). Soziologie der Zweierbeziehung: Eine Einführung. Springer.

33.

Ling

Z.-H.

Qin

Gao

Dai

L.-R.

Wang

R.-H.

Jiang

Zhao

Z.-W.

Yang

J.-H.

Chen

G.-P.

(2007). The USTC and iFlytek speech synthesis systems for Blizzard Challenge 2007 [Paper presentation]. Paper presented at the Blizzard Challenge Workshop.

34.

Looije

Neerincx

M. A.

Cnossen

(2010). Persuasive robotic assistant for health self-management of older adults: Design and evaluation of social behaviors. International Journal of Human-Computer Studies, 68(6), 386–397.

35.

Lopatovska

Williams

(2018). Personification of the Amazon Alexa: BFF or a mindless companion [Conference session]. Proceedings of the 2018 Conference on Human Information Interaction and Retrieval.

36.

Matthews

Lin

Panganiban

A. R.

Long

M. D.

(2019). Individual differences in trust in autonomous robots: Implications for transparency. IEEE Transactions on Human-Machine Systems, 50(3), 234–244.

37.

McCroskey

J. C.

Hamilton

P. R.

Weiner

A. N.

(1974). The effect of interaction behavior on source credibility, homophily, and interpersonal attraction. Human Communication Research, 1(1), 42–52.

38.

McCroskey

J. C.

Richmond

V. P.

Daly

J. A.

(1975). The development of a measure of perceived homophily in interpersonal communication. Human Communication Research, 1(4), 323–332.

39.

Miller

Kraus

Babel

Baumann

(2021). More than a feeling—Interrelation of trust layers in human-robot interaction and the role of user dispositions and state anxiety. Frontiers in Psychology, 12, 592711.

40.

Nass

Steuer

Tauber

E. R.

(1994). Computers are social actors [Conference session]. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems.

41.

Nomura

Kanda

(2015). Influences of evaluation and gaze from a robot and humans’ fear of negative evaluation on their preferences of the robot. International Journal of Social Robotics, 7, 155–164.

42.

Nørskov

(2017). Social robots: Boundaries, potential, challenges: Taylor & Francis.

43.

Ohanian

(1990). Construction and validation of a scale to measure celebrity endorsers’ perceived expertise, trustworthiness, and attractiveness. Journal of Advertising, 19(3), 39–52.

44.

Oleson

K. E.

Billings

D. R.

Kocsis

Chen

J. Y.

Hancock

P. A.

(2011, February 22–24). Antecedents of trust in human-robot collaborations [Conference session]. 2011 IEEE International Multi-Disciplinary Conference on Cognitive Methods in Situation Awareness and Decision Support (CogSIMA), Miami Beach, FL, United States (pp. 175–178). IEEE. https://doi.org/10.1109/COGSIMA.

45.

Paradeda

R. B.

Hashemian

Rodrigues

R. A.

Paiva

(2016, November 1–3). How facial expressions and small talk may influence trust in a robot [Conference session]. Social Robotics: 8th International Conference, ICSR 2016, Kansas City, MO, United States.

46.

Parasuraman

Riley

(1997). Humans and automation: Use, misuse, disuse, abuse. Human Factors, 39(2), 230–253.

47.

Porcheron

Fischer

J. E.

Reeves

Sharples

(2018). Voice interfaces in everyday life [Conference session]. Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems.

48.

Purington

Taft

J. G.

Sannon

Bazarova

N. N.

Taylor

S. H.

(2017). “Alexa is my new BFF” social roles, user satisfaction, and personification of the Amazon echo [Conference session]. Proceedings of the 2017 CHI Conference Extended Abstracts on Human Factors in Computing Systems.

49.

Raveh

Siegert

Steiner

Gessinger

Möbius

(2019). Three’s a crowd? Effects of a second human on vocal accommodation with a voice assistant [Paper presentation]. Paper presented at the INTERSPEECH.

50.

Reinhardt

Pereira

Beckert

Bengler

(2017, October 5–8). Dominance and movement cues of robot motion: A user study on trust and predictability [Conference session]. 2017 IEEE International Conference on Systems, Man, and Cybernetics (SMC), Banff, AB, Canada.

51.

Salem

Ziadee

Sakr

(2013, October 27–29). Effects of politeness and interaction context on perception and experience of HRI [Conference session]. Social Robotics: 5th International Conference, ICSR 2013, Bristol, UK.

52.

Sanders

Oleson

K. E.

Billings

D. R.

Chen

J. Y.

Hancock

P. A.

(2011). A model of human-robot trust: Theoretical model development [Paper presentation]. Paper presented at the Proceedings of the Human Factors and Ergonomics Society Annual Meeting.

53.

Simmel

(1923). Soziologie: Untersuchungen über die formen der vergesellschaftung (Vol. 2). Duncker & Humblot.

54.

Tay

Jung

Park

(2014). When stereotypes meet robots: The double-edge sword of robot gender and personality in human–robot interaction. Computers in Human Behavior, 38, 75–84.

55.

Thompson

L. F.

Gillan

D. J.

(2016). Social factors in human-robot interaction. In Thompson

L. F.

Gillan

D. J.

(Eds.), Human-robot interactions in future military operations (pp. 87–102). CRC Press.

56.

Topp

E. A.

Huettenrauch

Christensen

H. I.

Eklundh

K. S.

(2006, October 9–15). Bringing together human and robotic environment representations—A pilot study [Conference session]. 2006 IEEE/RSJ International Conference on Intelligent Robots and Systems, Beijing, China.

57.

Van Der Laan

J. D.

Heino

De Waard

(1997). A simple procedure for the assessment of acceptance of advanced transport telematics. Transportation Research Part C: Emerging Technologies, 5(1), 1–10.

58.

Wheeless

L. R.

Grotz

(1977). The measurement of trust and its relationship to self-disclosure. Human Communication Research, 3(3), 250–257.

59.

Zhao

Rau

P.-L. P.

(2020). Merging and synchronizing corporate and personal voice agents: Comparison of voice agents acting as a secretary and a housekeeper. Computers in Human Behavior, 108, 106334.