Exploring the Influence of Co-Present and Remote Robots on Persuasiveness and Perception of Politeness

Politeness is a crucial aspect of human social interactions. While the influence of politeness is well understood in human groups, it remains underexplored in group interactions with robots. Therefore, in this paper, we conduct an initial exploration into the influence of the presence of humanoid robots on their persuasiveness and perceived politeness in small groups. We conducted a user study (N = 119) with co-present and remote robots that invited participants to join the group using six politeness behaviors derived from Brown and Levinson's politeness theory. It requests participants to join them at the furthest side of the group, even though a closer side is also available to them, but would ignore the robot's request. The results show that co-present robots are perceived to be less persuasive than remote ones. However, co-presence enhances the clarity of the robot's requests and the perceived freedom of action while decreasing the perceived friendliness and offensiveness.

Figure 1: (left) a participant is standing at the initial location in the co-present condition; (right) a participant is watching a video of the robots in the remote condition.In both conditions, R1 invites them to join the group at the furthest side F by conducting verbal and non-verbal behaviors.Closest (green area, C) and furthest sides (blue area, F ) are marked.The "Left" and "Right" markers in green (right fgure) were included in the videos to disambiguate participants' perspective of the side to join.

ABSTRACT
Politeness is a crucial aspect of human social interactions.While the infuence of politeness is well understood in human groups, it remains underexplored in group interactions with robots.Therefore, in this paper, we conduct an initial exploration into the infuence of the presence of humanoid robots on their persuasiveness and perceived politeness in small groups.We conducted a user study (N = 119) with co-present and remote robots that invited participants to join the group using six politeness behaviors derived from Brown and Levinson's politeness theory.It requests participants to join them at the furthest side of the group, even though a closer side is also available to them, but would ignore the robot's request.The results show that co-present robots are perceived to be less persuasive than remote ones.However, co-presence enhances the clarity of the robot's requests and the perceived freedom of action while decreasing the perceived friendliness and ofensiveness.

CCS CONCEPTS
• Human-centered computing → User studies; Empirical studies in HCI; Empirical studies in interaction design.

INTRODUCTION
Politeness is a crucial element of human social interactions, as it plays a vital role in shaping individuals' decisions and fostering positive relationships.According to Brown and Levinson's research [5], politeness refers to the efort to prevent or mitigate actions that could harm an individual's public self-image or "face" [9].In the Human-Robot Interaction (HRI) context, designing polite behaviors for humanoid robots helps to create positive impressions for users.To join a group politely, a newcomer should consider the group dynamics [15] and the personal space of each member [12].Generally, when a group member invites a newcomer to join the group, the choice of politeness strategies can signifcantly afect the execution of the invitation and its level of persuasiveness and perceived politeness since that invitation can be perceived as a face-threatening act.Persuasion involves infuencing beliefs, attitudes, or behaviors through communication to convince others to adopt a specifc viewpoint, take action, or make a decision [8].
In our scenario, a robot seeks to persuade participants to join an inconvenient side of a group that they might not choose without the persuasive attempt.While these dynamics are well understood in human groups, they remain underexplored in group interactions with robots.
Previous work has shown that the presence of artifcial agents, whether they are physically embodied such as robots or exist virtually like virtual agents, is signifcant in interactions between humans and agents [4,18,34].A robot that shares the same physical space as the user is referred to as "copresent" and a robot that is displayed through a live or recorded video feed on a screen or projection is referred to as "remote" [18,35].Understanding the impact of presence on social interaction is important for a number of applications, for example, determining what social interaction behaviors can be evaluated or designed without a present physical robot platform, which may be difcult or costly to transport.Specifically, in this study, we work towards determining the potential infuence of robots' presence on their ability to efectively communicate their requests as well as how their requests are perceived by humans in terms of their politeness.
In this paper, we undertake an initial investigation to examine the impact of presence of humanoid robots on their persuasiveness and the perception of politeness of their requests.We conducted a controlled experiment with co-present (Figure 1 left) and remote (Figure 1 right) robot experiences, in which participants were invited by a Pepper robot to join a group of two Pepper robots at the furthest side of the group (from the participant's perspective) using diferent politeness behaviors.Our results show that co-present robots are not as persuasive as remote.However, copresence improved the clarity of the robot's requests and the perceived freedom of action (i.e., negative face) while reducing the robot's perceived friendliness (i.e., positive face) and ofensiveness (i.e., face loss).

RELATED WORK
Brown and Levinson [5] have identifed fve strategies for communicating needs while minimizing face-threatening acts: not doing the act (NOT), of-record or indirect (IND), using indirect language; negative politeness (NEG), focusing on avoiding imposition; positive politeness (POS), emphasizing friendliness; and bald on-record or direct (DIR), using clear and direct language.Our design for this study was based on these strategies, and we utilized them to create six distinct combinations of verbal and non-verbal behaviors, which will be elaborated on later.Several studies have also investigated how individuals regulate their personal space within a group, with Kendon's [15] theory of "F-formation" describing the spatial organization of a group engaged in a face-to-face conversation where all members have unobstructed and equal access to the space.One of the main concepts of this study is how humans apply this concept to a group of robots when joining it.
Previous work has explored social behaviors and norms in individual or group scenarios [2,6,24,30], aiming to understand how humans interact with robots and how these interactions can be improved to enhance the user experience.These studies have examined the impact of social cues such as gaze [1], facial expressions [16], and body posture on users' perceptions of robots' social presence and their willingness to interact with them [26,28].Other studies have focused on investigating the role of robot behavior's role in shaping users' attitudes and behaviors toward them, such as politeness.Overall, the research in HRI has shown that social behaviors and norms play a critical role in shaping the human-robot interaction and that incorporating such elements into robot design can enhance the user experience and promote positive attitudes towards robots [19,29,31,32].Moreover, similar studies have been conducted to investigate the impact of the politeness of humanoid agents on the trajectories taken by human participants when joining a group of humanoids and the persuasiveness of those strategies [14,36,[38][39][40].In experimental studies, the term "presence" refers to how an agent, such as a robot, is presented to participants.According to Zhao et al., [35], an agent is considered to be "copresent" when it shares the same physical space as the user.In contrast, an agent displayed on a screen or projection using a video feed is referred to as "remote".This presence aspect can signifcantly afect the participants' perception and reaction toward the agent and must be considered in designing HRI studies.

METHODOLOGY
We conducted a controlled experiment to investigate the infuence of robots' presence and politeness on users' behavior.Therefore, for this experiment, we had the following research question: How does a robot's physical and remote presence infuence its persuasiveness and users' perception of its politeness?
The experiment was designed with two independent variables: (1) robot's presence (between-subject) and (2) politeness behaviors (within-subject).The presence contained two levels: (1) co-present with a robot physically placed in a room and (2) remote with a robot being observed remotely via a video.The co-present condition involved two Pepper robots placed at the center of a motion capture (MoCap) room (5.2 x 4.4 m) (Figure 1 left).Although the participants were given information that the robots were controlled automatically, they were controlled by a human using the Wizard of Oz methodology [23], which was shared with the participant during the debriefng.Participants started each trial at a distance (6 meters) from the group center and when the robot invited them to join their group, they were free to move in the environment.In each trial, the participant's initial location was switched between the right-hand side and left-hand side of the group.The remote condition involved pre-recorded videos of two Pepper robots placed at the center of the MoCap room, and the participant's task was to watch each video and decide where they wanted to join in after watching each video (Figure 1 right).
The politeness behaviors of the robot included six levels (Table 1).Participants experienced all six behaviors as experimental conditions for one level of presence in a counterbalanced order created using a Balanced Latin Square.The primary robot (R1) faced participants and invited them to join the group using a combination of verbal and non-verbal politeness behaviors.R1 always invited participants from their viewpoint at the furthest side of the group, except in the BSL condition.R1 also maintained eye contact with the participant throughout the whole experiment.
After obtaining informed consent, we collected participants' demographic data and provided a brief overview of the experimental procedures.We introduced participants to the MoCap room and the robots and provided them with a tablet to provide feedback at each trial's end.They had to start at the initial location at the beginning of each trial, looking at the group of robots in front of them and hearing a beep signal.After joining the group, the participant answered the following questions using a 7-point Likert scale (1 -"strongly disagree", 7 -"strongly agree"): (1) "I could precisely understand the robot's wants."This question was intended to assess the clarity of the robot's requests.(2) "I got ofended by the robot's action."This question was designed to measure the degree of face loss or ofensiveness of the robot's requests.(3) "The robot wanted to increase intimacy with me."This question aimed to determine the level of satisfaction of a positive face, which relates to friendly or warm behavior.(4) "The robot respected my freedom of action."This question was designed to evaluate the level of satisfaction of a negative face, which relates to respecting the other's choice, freedom of action, or cold behavior.
For the conditions with co-present robots, we recruited 59 participants (32 F, 27 M), aged between 19 and 63 (32 ± 10) and profcient in English were recruited from the general public in Sweden.Sixteen had no previous knowledge or familiarity with AI systems (robots, virtual agents), 32 had basic, and 11 had intermediate or advanced familiarity with AI systems.
For the conditions with remote robots, we recruited 60 participants (28 F, 32 M), aged between 18 and 58 (32 ± 9) and profcient in English were recruited online from the general public using Ques-tionnairePro 1 .Sixteen participants had no previous knowledge or familiarity with AI systems (robots, virtual agents), 34 had basic, and 10 had intermediate or advanced familiarity with AI systems.

RESULTS
We found that co-present robots are less persuasive than remote.Moreover, co-presence improved the clarity of the robot's requests and the perceived freedom of action while reducing the robot's perceived friendliness and ofensiveness.For the data analysis, we used mixed repeated measures ANOVA.

Persuasiveness
Overall, there was a signifcant main efect of presence of the robots on Persuasiveness (F (1, 117) = 21.546,p < 0.001).Our results suggested that the co-present robots were signifcantly less successful in persuading participants to join the group at the furthest side than the remote robots.Figure 2 provides a detailed illustration of persuasiveness, including both experimental conditions and their corresponding associated politeness behaviors.

Politeness
Figure 3 illustrates the results regarding the perceived politeness of the robots.Overall, there was a signifcant main efect of presence of the robots on Clarity (F (1, 117) = 17.321, p < 0.001).Our results suggested that co-presence of the robots signifcantly enhanced the clarity of the robot's requests.Moreover, there was a signifcant main efect of presence of the robots on Face loss (F (1, 117) = 11.137,p = 0.001).Our results suggested that co-present robots were seen as the most efective in minimizing the loss of face for participants.
Additionally, there was a signifcant main efect of presence of the robots on Positive face (F (1, 117) = 24.753,p < 0.001).Our results suggested that the requests from the co-present robots were perceived as less friendly in comparison to the ones from the remote robots.Lastly, there was a signifcant main efect of presence of the robots on Negative face (F (1, 117) = 43.311,p < 0.001).Our results suggested that co-present robots made the participants feel less constrained in terms of their freedom of action.

DISCUSSION & FUTURE WORK
Although the requests from the co-present robots were perceived as signifcantly clearer, co-present robots showed a lower success rate in persuading participants to join the group at the furthest location compared to the remote.This contradicts the fndings of several studies in the literature [4,18,34], which indicate that robots have more infuence physically than remotely.Participants were more easily persuaded to join the group at the furthest side with remote robots, as they did not need to exert themselves to reach that location physically.However, the persuasion rate signifcantly declined when participants were required to walk to the location in the co-present condition.Consistent with psychological research [13], participants will likely choose options requiring less efort when presented with similar rewards.The results of this study have signifcant implications for the development of persuasive systems [20], as it suggests that mere physical presence alone may not always lead to positive responses from participants.Moreover, it highlights the necessity of considering additional factors such as efort in the equation.Consistent with previous studies [4,18,34], our fndings indicate that physical presence enhances the clarity of robot requests.Specifcally, when the robots were physically present, participants perceived the requests more clearly.Additionally, the presence of robots made participants feel less restricted and more free to decide how they wanted to join the group.The decreased persuasiveness of robots in the co-present condition may also be due to the participants' sense of greater freedom and less constraint in their decision about where to join the group.The robots in the remote condition showed a higher level of perceived friendliness, i.e., positive face, while a higher level of ofensiveness, i.e., face loss.It shows the contraction in the literature towards online and digital communication versus face-to-face.Some studies have found that digital communication can evoke more negative emotions compared to in-person interactions [3,21].In contrast, others showed paradoxical, i.e., positive and negative, outcomes of online communications [17].Consistent with their results, Roos [25] found that extroverts and individuals with more social support had better outcomes using social media.In comparison, introverts and those with less support had worse outcomes.This shows the potential impact of personality on digital communication.Moreover, several studies [10,11] have investigated politeness in communication and found that perceptions of politeness can difer between cyber and in-person communication.
The concept of politeness, whether verbally, non-verbally, or through para-verbal behavior, may be defned diferently across various cultures and personalities.Participants' interpretations of politeness in terms of clarity, loss of face, positive face, and negative face may difer signifcantly.Hence, there is a question about whether the fndings from this study can be generalized across cultures.Moreover, several studies have shown that robot embodiment [7,33], and its motion [27] impact human-robot interaction.This embodiment allows robots to convey emotions, intentions, and social cues through nonverbal cues such as facial expressions, gestures, and body language.We utilized the Pepper robot, which has a specifc embodiment and range of motions and voice.This factor may have infuenced our results, as the robot embodiment can afect how humans perceive and interact with it.Thus, future research should explore the infuence of agent embodiment, i.e., robot or virtual agent, on the perception of politeness and persuasiveness of the agent.Lastly, the remote condition constrained participant movement within the environment, prompting future research in telepresence experiments to simulate participant mobility.

CONCLUSION
We explored the impact of presence on humanoid robot's persuasiveness and politeness in a small group scenario.The results showed that while copresent robots were perceived to be less persuasive than remote ones, copresence enhanced the clarity of the robot's requests and the perceived freedom of action while decreasing the perceived friendliness and ofensiveness.Overall, the study's fndings contribute to our understanding of the complex dynamics involved in human-robot interactions within small groups, and they highlight the importance of further research on the infuence of presence and politeness on the persuasive abilities of humanoid robots, and perceived politeness for building rapport and long-term collaborations in various scenarios like education [37] or healthcare [22].The results of this study ft into a broader context of research we are engaged in to determine which aspects of human-robot social interaction may be robustly investigated or designed via video or virtual reality demonstrations without the immediate need for present physical robot embodiments.

Figure 2 :
Figure 2: Persuasiveness of the robot's request in both experimental conditions and their associated politeness behaviors.

Figure 3 :
Figure 3: Perceived politeness of the robot's request in both experimental conditions and their associated politeness behaviors.Note that only in the particular case of the Face loss, lower values are preferred.

Table 1 :
Robot's behaviors and their associated politeness strategies, verbal and nonverbal behaviors derived from the theory.Nonverbal behaviors: open palm up (UP), open palm sideways (SIDE), open palm sideways and partly downwards (DOWN), pointing directly at a specifc place with a straight arm (POINT).