EMPIRICAL RESULTS ON DETERMINANTS OF ACCEPTANCE AND EMOTION ATTRIBUTION IN CONFRONTATION WITH A ROBOT RABBIT

Abstract

As robots increasingly enter people's everyday lives, it becomes ever more important to explore the conditions and determinants of acceptance of human interactions with these devices. Moreover, a positive feeling associated with the interaction with robots is a precondition for the user's willingness to engage in further interactions and establish long-term relationships. This article presents empirical results from two studies that focus on the user's perception of the robot rabbit Nabaztag, a small WiFi-enabled device with movable ears, integrated RFID reader functionality, and speech-synthesis capability. In the first study, 53 participants were confronted with a range of the Nabaztag's functionality, and, using RFID cards, they interacted with the rabbit. The analysis of people's answers concerning Perceived Ease of Use and Perceived Usefulness as well as hedonic and pragmatic aspects, showed gender-accorded differences regarding the evaluation of the device. Neither the degree of familiarity with computers nor the fact of whether technical disfunctionality occurred during the trial influenced the evaluation of the robot, while ownership of a robotic toy let people evaluate the Nabaztag more positively. The second study took a more detailed observance of the effect of the rabbit's expression by its ears. In a within-subjects setting, a German (N = 100) and a U.S. American sample (N = 111) were asked to rate the rabbit's current emotional status from pictures that showed the rabbit with a variety of six different ear positions. Results indicate that people infer specific emotional states from the robot rabbit's different ear positions. Also illustrated is that observers' attribution of feelings to the rabbit depends on their cultural backgrounds. Implications and questions for future research are discussed.

INTRODUCTION

Although for a long time the average person encountered robots only when watching TV or as industrial robots in factories, in the last few years, robots and agents in various shapes can be met more and more in the private sphere of our everyday lives. As robots enter our private lives there is an ever increasing need to develop systems that are accepted by the user and positively co-notated (also compare Breazeal 1999; Dautenhahn et al. 2005; Friedman, Kahn, and Hagman 2003). Ideally, these systems should not only be engaging for a short time but should accompany their users over longer periods.

As the following examples will show, the presence of robots in people's homes is not a future scenario; they are indeed already present in the average person's household. Many off-the-shelf robots like Roomba, a small flat disk, hovering over the floor and supporting the housework, or zoomorphic robots such as AIBO, the small robot dog (cp. e.g., Kahn et al. 2004; Melson et al. 2005), Pleo, a robot toy dinosaur (e.g., Krämer et al. 2010), Paro, a small robot seal (cp. e.g., Wada and Shibata 2006, 2007; Kidd, Taggart, and Turkle 2006), or the Nabaztag (a small robot rabbit) have been used in research settings to explore the effect of their presence or (social) attribution and interaction, confirming that they can be more than just vacuum cleaners or toys. Research “suggests that people do form strong intimate attachments to these technologies” (Sung et al. 2007, p. 145). In a study conducted by Forlizzi and DiSalvo (2006), for example, Roomba has been found to be characterized (unlike traditional vacuum cleaners) by its “functionality, aesthetics, and symbolic merit” (p. 263). And Paro, which has sensors giving it rudimentary capabilities of perceiving its surroundings and reacting, has been successfully used for therapeutic purposes, for example, in nursing homes (Wada and Shibata 2006, 2007; Kidd et al. 2006). All of the robots hold at least basic potential of becoming a companion to the human owner and foreshadow the role robots may play as technical advancements progress.

The EU Project SERA (Social Engagement with Robots and Agents), in the context of which the following studies were conducted, focused on long-term data collection with the robot rabbit Nabaztag that was set up in elderly people's homes. During the three iterations of the project, data were collected that have been analyzed from a variety of perspectives (also refer to the other articles in this volume). Participant's utterances in confrontation with the Nabaztag have been taken into consideration in order to shed light on the question in how far conversations are comparable to human-human conversations. By looking at extracts from the video data, Payr (2010) discusses the prerequisites and functions of interaction rituals. Other insights are based on qualitative interview data investigating how aspects of use and acceptance can be important determinants for the establishment of long-term relationships with a social robot (Klamer and Ben Allouch 2010). Besides these qualitative approaches, quantitative analyses of the data collected in the iterations have focused on aspects like smiling or gaze behavior of the participants to deduce specific indicators of social behavior towards the robot. These data also show, that (social) relationships can be established with a robot since “natural” interactions with facial expressions displayed by the user can be observed (von der Pütten, Eimler, and Krämer 2010, von der Pütten, Eimler, Ganster, Hoffmann, and Krämer 2010).

Altogether, data of these previous studies give insights into how people behave towards the robot rabbit that is established in their homes for some time. Because of the small sample size (three participants in iteration 1, five participants in iteration 2, and six in iteration 3), however, the results of the quantitative analyses, especially, have to be interpreted with care. Since the settings and functionality are changed in the course of the iterations and the number of participants is fairly low, additional empirical testing of how users perceive the robot are worthwhile and give useful additional insights.

This paper presents two studies, both conducted with the Nabaztag robot to inform not only the design of the robot's interaction in the iterations, but also to be of relevance for future scenarios with the rabbit and of overall importance to research regarding social engagement with robots. The Nabaztag is a WiFi-enabled ambient device that can process a variety of services via a server connection. It is 23 cm high and equipped with five colored LED lights, a microphone, built-in speakers, as well as rotatable ears and an RFID reader (see Figure 1). Moreover, it features a text-to-speech synthesizer. The two studies that are presented in the following especially explore aspects of user's experience in the sense of pragmatic and hedonic aspects as well as aspects of user acceptance and the rabbit's ear positions as signals of nonverbal communication.

FIGURE 1 Participant in interaction with the Nabaztag.

STUDY I: GENERAL EVALUATION OF USER EXPERIENCE ASPECTS

Beyond Usability: Positive User Experience as a Precondition for Relationship Building/Affiliation

In recent years, researchers in the field of Human-Computer Interaction (HCI) have drawn attention to the necessity of considering emotional and motivational aspects associated with a product and its use. In general, there has been a shift from a mere concentration on pragmatic or functional qualities of a product and the prevention of usability problems, to the additional consideration of the emotional role and importance of positive experiences associated with the objects or systems (Norman 2004; Overbeeke et al. 2003, Karpanos et al. 2009; Seligman and Csikszentmihalyi 2000; Jordan 1998). More specifically, Weiss et al. (2009) also observe an increased interest in the field of Human-Robot Interaction to elicit positive experiences for humans interacting with robots. They emphasize that this is especially relevant in home settings (Weiss et al. 2009; Fong, Nourbakhsh, and Dautenhahn 2003). As has also been pointed out by Klamer and Ben Allouch (2010), several studies have been conducted with zoomorphic robots (e.g., Paro, a robot seal or AIBO, a small dog) in the last few years; few, however, focused on the acceptance of these robots specifically (e.g., Heerink et al. 2008; de Ruyter et al. 2005).

As a definition of user experience that can also be applied to the field of human robot interaction, Alben (1996) describes it as “aspects of how people use an interactive product: the way it feels like in their hands, how well they understand how it works, how they feel about it while they're using it, how well it serves their purposes, and how well it fits into the entire context in which they are using it” (p. 13). To extend this view with regard to emotional facets, Overbeeke et al. (2003) state that products do not necessarily have to be easy to use to be attractive for users, but that humans look for products that are “challenging, seductive, playful, surprising, memorable or even moody, resulting in enjoyment of the experience” (Overbeeke et al. 2003, p. 9). In line with this, the HCI researcher community works toward a more holistic view on human-technology interaction (e.g., Gaver and Martin 2000; Jordan 2000; Wright, Wallace, and McCarthy 2008; Olivier and Wallace 2008; Hassenzahl 2001; Hassenzahl 2003; Burmester, Hassenzahl, and Koller, 2002; Kaptelinin and Nardi 2006; Norman 2004).

With regard to the robot rabbit Nabaztag, certain emotional and sensual values for the user can be presumed that might trigger a positive user experience. The robot is an embodied object; it is tangible. It has a surface, a texture, and a certain weight. Adding to this, the Nabaztag addresses many different senses and invites the user to experience it not only on the haptic, but on the audio and visual level as well. This multi-sensual experience is provided by the rabbit's capability of moving its ears, flashing its LED lights, and speaking. All in all, the small rabbit matches up to the emotional and physical pleasure tangible objects might provide (Norman 2004).

It is also imaginable that people are impressed by the novelty of the robot rabbit's appearance and its cuteness. That the beauty of an object is essential for its evaluation was shown by Tranctinsky, Katz, and Ikar (2000). Furthermore, the rabbit might have a certain appeal because of its reference to a real rabbit, since according to Jordan (2000), robotic products might be conceived as “living objects with which people have relationships” (p. 7). In addition to tangibility, novelty, and aesthetic appeal, interacting with the rabbit might help people to express and communicate a specific image of themselves to others (Hassenzahl 2003; Burmester, Hassenzahl, and Koller 2002); for example, using a robotic device may define a person as up-to-date or technically skilled. In sum, the Nabaztag's appearance and functionality have a high potential of exerting a specific appeal on its user. Against the background of previous results, chances are good that the interaction will be perceived to be pleasurable and that if preconditions are fulfilled, the Nabaztag might, in the long run, become an emotionally meaningful object to its user.

Objective

In this study we investigated the user's general evaluation of the robot rabbit as well as a number of person variables that might be seen as determinants for the evaluation. In doing this, several aspects of the rabbit's functionality were tested by the user, for example, use of RFID stamps (small colored RFID tags in the shape of stamps that were attached to small envelopes, see Figure 2), radio, speech-synthesis, and ear movement. The whole interaction was evaluated for its usability; pragmatic aspects, as well as emotional aspects, in other words, hedonic aspects of interacting with the Nabaztag. Furthermore, a number of person variables (e.g., gender, computer expertise) were assessed to see whether these variables determine the appeal the Nabaztag would have for the person. Data collected in this study can be used not only to find out whether people like interacting with the rabbit in general (voice, lights, ears, etc.), but also to analyze how they evaluate the use of RFID tags and whether the interruptions or failures in usage influence the users' feeling toward and evaluation of the robot rabbit.

FIGURE 2 Stimulus material, Nabaztag with six different ear positions.

Method

Before filling in the questionnaire, participants interacted with the Nabaztag for about 15 minutes. Each section of the questionnaire was preceded by instructions informing the participants on how to fill in the different parts (e.g., “In the following you will see a list of statements. Please indicate for each statement how much you agree/disagree”). The Perceived Usefulness (PU) Scale and the Perceived Ease of Use (PEU) Scale, taken from the Technology Acceptance Model (TAM)(Davis 1989; Davis, Bagozzi, and Warshaw 1989), were used to assess participants' acceptance of the Nabaztag and the applications triggered during the interaction. While the P U Scale contains seven statements on how useful or useless the student considers the Nabaztag to be (e.g., “All in all, I consider the rabbit useful” or “I think the rabbit could help me to perform better”), the P E U scale consists of six items for the evaluation of how easy or complicated the use of the Nabaztag is perceived to be (e.g., “The rabbit is easy to use and behaves as expected”). Both scales are measured on a 7-point Likert scale ranging from “I don't agree at all” to “I fully agree.”

Because we were interested in how the user experiences the interaction with the Nabaztag, the AttrakDiff 2 scale (Hassenzahl, Burmester, and Koller 2003) was included in the questionnaire. This instrument consists of a 21-item, 7-point semantic differential containing, for example, opposing pairs: complicated vs. simple, or unimaginative vs. creative. The central notion behind the AttrakDiff is that products not only fulfill pragmatic (usability) needs of their users in the sense of the users' wishes to predictably manipulate their surroundings, but that rather emotional, hedonic needs are important, too. A product's usability is generally understood as being fulfilled when it provides a specific, appropriate functionality (utility, usefulness) and can be (smoothly) operated at the same time. A user's pragmatic claim toward a product is thus, according to Hassenzahl (2004), “connected to the users' need to achieve behavioral goals” (p. 322). An object that ensures effective and efficient goal achievement is perceived as having pragmatic quality. Hedonic attributes are, in contrast, related to the user's self. They can be distinguished as stimulation; for example, novelty, challenge, fun, (Hassenzahl 2004, Csikszentmihalyi 1975) and identification; for example, self-presentational values (Wicklund and Gollwitzer 1982). According to these scholars, people's need for stimulation arises from the striving for personal development and advancement in knowledge and skills. Products with new features and interesting functionality or presentation styles can fulfill these stimulation needs (Hassenzahl et al. 2003). Identity and identification needs are mainly connected to people expressing themselves though objects (Prentice 1987). They attribute certain ideas or values to the product that they think will help to form the impression others will build of them (Hassenzahl et al. 2003). Since we were interested in seeing to what extent the Nabaztag would be able to fulfill these needs, we used the AttrakDiff2.

As a potential moderator variable, we assessed participants' general degree of experience with computers on a 4-point Likert scale (1 = none, 2 = basic, 3 = sufficient, 4 = extensive). Moreover, they were asked if they would use the Nabaztag for private purposes, and if they would recommend it to a friend (5-point Likert scale, 1 = not at all, 5 = absolutely), and if they owned a Nabaztag or similar artificial entities (e.g., AIBO, Tamagotchi, Furby, Nintendogs, Irobi, Genibo, etc.). In the final section of the questionnaire, age and sex of the participant were assessed.

Participants

53 people took part in a laboratory study in which they were confronted with a computer and the rabbit, as well as with three small envelopes with RFID stamps and their function indicated by words written below the stamps (see Figure 1).

Procedure

The study was announced on several university message boards and was conducted in laboratories at the University of Duisburg-Essen, Campus Duisburg, Germany. On arrival, people were lead into a room equipped with a computer, keyboard, mouse, and monitor as well as the Nabaztag and small envelopes with several colored RFID stamps. Since we aimed at exploring participants' evaluations in a rather natural setting, we decided to instruct people to inform themselves about the daily news, choosing from one of three German online newspaper Web pages (e.g., www.spiegelonline.de) while the Nabaztag would perform a number of actions (moving its ears, speaking, light choreography) and encourage the user to use the RFID cards to trigger specific functionalities (see Table 1 for details on the Nabaztag's actions with their temporal order of occurrence).

TABLE 1 Details on Actions Performed by the Nabaztag

Schedule	Actions
After appr. 3 min.	“Oh, what nice weather, isn't it? Use the envelope with the label ‘weather,’ and Iwill tell you the weather forecast. Just hover over my nose”
After appr. 5 min.	Nabaztag says, “ähm.”
Minute 6	“Would you like to hear a fairy tale? Hold the envelope with the yellow stamp near my nose, and you can hear a fairy tale. If you want me to stop, press the button between my ears for some seconds.”
Minute 8	Please take the small yellow rabbit and hold it with its bottom at my nose so that I can play the radio for you. If you want to stop the radio, please press the button between my ears for approximately 4 seconds and it will stop.”
Minute 10	Lights flash, Nabaztag says, “lalala.”
Minute 12	“I need to reflect for a while and practice some Thai-Chi.” à ear and light choreography
Minute 14	RSS-Feed from “Tagesschau,” using the Ztamp. “Please press the envelope labeled ‘news’ with the stamp at my nose so that I can inform you about today's news.”
Minute 16	“You're done. I will go to sleep now. In a minute, the instructor will pass by to give you new instructions.”

In the course of this interaction, some participants experienced problems, for example, in the sense of technical problems with the server connection, the rabbit needs to compute the commands (which was the main reason for technical problems), leading to a delay of the display of specific functionality, or problems resulting from an incorrect handling of the Nabaztag or the stamps. Since these problems cannot be prevented, they were documented in order to differentiate users who experienced difficulties from those who did not. After they had interacted with the Nabaztag, people were asked to fill in a questionnaire containing demographic questions, the AttrakDiff2 and the P EU and P U Scales from Davis' Technology Acceptance Model. Participants were not financially rewarded, but could choose from a variety of sweets.

Results

This section focuses on the results of the study. It starts with results concerning the correlations between the PEU and PU Scales as well as the AttrakDiff2 items and proceeds with the descriptive details regarding these constructs and the items relating to the recommendation of the device. After that, results assessed on the influence of situational factors will be described. Finally, the focus will be on the analysis of person variables, in other words, gender, computer experience, and ownership of robotic toys, helping to identify determinants of the evaluation of the rabbit.

Correlations between PU/PEU and AttrakDiff2

Correlations between PU and PEU and the AttrakDiff 2 constructs were calculated. Results show that perceived usefulness and perceived ease of use correlate with all three AttrakDiff constructs.

Hedonic identification (r = .564, p < .001) as well as the hedonic stimulation (r = 475, p < .001) and pragmatic quality (r = .490, p < .001) correlated strongly positively with the PU. The pragmatic quality of the AttrakDiff 2 questionnaire correlated with the PEU (r = .729, p < .000). Also, hedonic identification (r = .334, p = .015) and hedonic stimulation correlated (r = 329, p = .016) with PEU. Also, correlations with beauty and goodness and PU and PEU showed to be significantly positive. Beauty correlated positively with both PU (r = .352, p = .010) and PEU (r = .327, p = .017). Also, goodness correlated positively with PU (r = .493, p < .001) and PEU (r = .471, p < .001).

AttrakDiff2: Hedonic Identification, Hedonic Stimulation and Pragmatic Quality

Reliability measures for all dimensions of the AttrakDiff2 (namely, pragmatic quality, hedonic identification/stimulation) showed to be satisfying. Pragmatic quality reached the lowest, but still satisfying, consistency value with α = .75. Internal consistency for the items of hedonic stimulation were α = .81 and hedonic identification α = .86.

Hedonic stimulation reached the highest average value with a mean value of M = 4.85 (SD = .926). The global item beauty was rated M = 4.75 (SD = 1.731) and goodness (M = 4.74, SD = 1.583), highest possible value being 7. Items adding up to the hedonic identification of the Nabaztag reached slightly lower mean values with M = 4.43 (SD = 1.105). Finally, the pragmatic quality of the application reached an average value of M = 4.07 (SD = 1.020). This shows that, with regard to the evaluation of the rabbit, the attribution of, for example, novelty and challenge and the overall ratings of beauty and goodness were most prominent. The aspect of personal values and self-presentation was slightly less important. Compared with the other items, usability was rated rather low.

Technology Acceptance: Perceived Ease of Use and Perceived Usefulness

The constructs of Perceived Usefulness (PU) and Perceived Ease of Use (PEU) from the TAM were tested for predicting user acceptance. Since Davis (1989; Davis, Bagozzi, and Warshaw 2004) does not state reliability measures for the scales, Cronbach's alpha values were calculated for both scales. Cronbach's alpha showed to be very high (α = .89) for the items constructing the PU Scale and the PEU Scale (α = .76). Items of both scales were summed up to obtain an average value for the PU (M = 2.67, SD = 1.25) and the system's ease of use (M = 4.90, SD = 1.03). While people considered the system usefulness to be rather low, the ease of use was comparably high.

Recommendation

People were confronted with two additional questions. The mean values for the question of whether people could imagine using the Nabaztag for their own private purposes were M = 2.60 (SD = 1.51). Almost the same value was obtained for the question of whether participants would recommend the robot to a friend (M = 2.58, SD = 1.45). The maximum approval being 6, both mean values can be interpreted as medium consent.

Effects of Technical (Dis)Functionality

As mentioned above, some participants experienced problems in the interaction with the robot rabbit that were documented. For the analysis, participants were subdivided into two groups. One group, in which part of the rabbit's functionality did not work out correctly (e.g., stamps were not read, some applications were not triggered, etc.) (N = 38) and one group in which all intended applications functioned correctly (N = 15). ANOVAs (Analysis(es) of Variance) were conducted concerning the AttrakDiff2 and the TAM constructs. No significant differences between the two groups were found.

Effects of Participants' Gender on Evaluation

ANOVAs were conducted for several (socio-demographic) person variables in order to identify determinants for the evaluation. The analysis showed significant gender-accorded differences concerning the evaluation of pragmatic quality (F(1,52) = 4,244, p = .045, np ² = .078) and hedonic identification (F(1,52) = 5.870, p = .019, np ² = .105). Women rated the rabbit's pragmatic quality higher than men did (Mm = 3.7453, SDm = 1.045; Mw = 4.32, SDw = .9686) and reported higher values for hedonic identification. The ANOVAs including gender as independent and hedonic stimulation as dependent variables did not show any significant differences. However, the global items, beauty (F(1,52) = 5.332, p = .025, np ² = .096) and goodness (F(1,52) = 4.245, p = .045, np ² = .078) differed across sexes. Women considered both to be more positive than men did (Beauty: Mm = 4.13, SDm = 2.007; Mw = 5.21, SDw = 1.346/Goodness: Mw = 5.10, SDw = 1.372; Mm = 4.22, SDm = 1.731).

All in all, this means that while men and women attribute rather similar values concerning novelty and challenge to the use of the rabbit, women seem to see higher self-presentational potential in the rabbit and rate it as more useful.

An ANOVA was conducted using gender as an independent and PU as a dependent variable. Results indicate a significant main effect of gender, F(1,52) = 6.205, p = .016, np ² = .110. Women evaluated the usefulness of the Nabaztag to be higher than men did (Mw = 3.00, SDw = 1.254; Mm = 2.186, SDm = 1.088). The ANOVA conducted with PEU as a dependent variable also resulted in a significant effect of gender: F(1,52) = 7.740, p = .008, np ² = .134. Women considered the ease of use to be higher than men did (Mw = 5.221, SDw = .7904; Mm = 4.465, SDm = 1.6474). These results are in line with results regarding the pragmatic quality measured by means of AttrakDiff2, which was also evaluated higher by women than by men.

Additionally, two factorial ANOVAs were conducted, including participant's gender and the variable of functionality/disfunctionality, in order to test whether women are more forgiving of the rabbit's failures. The analysis did not show any significant interactions effects of these two variables with regard to the evaluation of the Nabaztag.

Effect of General Degree of Computer Experience

45.3% (N = 24) of the participants indicated having extensive experience with computers, 43.4% (N = 23) said they had sufficient knowledge and only six people were found to have only basic computer knowledge. An ANOVA including these three groups (extensive experience vs. sufficient knowledge vs. basic knowledge), conducted to find out whether the degree of computer experience would influence the evaluation of the Nabaztag, did not find any significant main effects of the degree of computer knowledge on the evaluation of the robot rabbit.

Effect of Experience with Artificial/Robotic Toys

An ANOVA was conducted including people's experience with artificial or robotic toys like AIBO, Furby, or Irobi as an independent variable (distinguishing between people who had no prior experience whatsoever with artificial entities, N = 33, and those who had, N = 20) and the AttrakDiff2 constructs as dependent variables. Results did not show a significant effect of the independent variable on pragmatic quality or on hedonic stimulation. However, owning a device such as the Nabaztag, AIBO, Tamagotchi, Furby, Nintendogs or Irobi showed to have significant effects on the attribution of hedonic identification to the Nabaztag (F (1, 51) = 4.191, p = .046, np ² = .076) since people without such a device attributed less hedonic quality (M _with  = 4.82, SD = .986; M _without  = 4.2, SD = 1.12).

ANOVAs conducted including the TAM constructs did not show a significant effect of ownership on the PEUbut did on PU (F (1, 51) = 10.337, p = .002, np ² = .169). People owning one or more of the devices listed above considered the Nabaztag's usefulness to be higher (M _with  = 3.32, SD = 1.27; M _without  = 2.28, SD = 1.06).

Discussion

The aim of this study was to test people's attribution of usefulness and ease of use in interacting with the Nabaztag, as well as to find out about the hedonic and pragmatic quality that is assigned to this robotic device. In this regard, person-specific determinants of the attribution of positive value were identified and the effect of technical failures explored.

Generally, this study produces some very interesting results with regard to the acceptance of the rabbit and the role it might play for different user groups. With regard to the more emotional values of the product, we found that people attributed a mediocre level of novelty and challenge, beauty and goodness to the rabbit. Slightly less important were self-presentational aspects and usability (average value between 4 and 5 of a maximum of 7). While the usefulness as such reached comparably low values, the Ease of Use was rated positively. With regard to specific person variables that can hardly be explored in qualitative analyses, results show that women evaluated the Nabaztag generally more positively than did men. Women rated the rabbit's pragmatic quality, the PEU, and the PU to be higher, as well as the product's beauty and goodness. Their hedonic identification with the Nabaztag was also higher than that of men. One might have expected that women would tend to be more forgiving toward technical problems with the rabbit's functionality, which may have influenced the above mentioned difference between the evaluation of men and women. However, no significant interaction effect could be found. Results may be explained by either women's higher tendency for socially desirable answers, or by women potentially being more receptive to the beauty and cuteness of the rabbit than men. Alternatively, women might see the Nabaztag to be more influential with regard to their self-presentation due to the fact that women are stereotypically not associated with robotic toys.

No differences with regard to the evaluation were found with respect to the general degree of computer expertise. Interestingly, in contrast to that, people owning a robotic device indicated more positive values with regard to the hedonic quality of the rabbit. People without such a device attributed less hedonic identification to the rabbit. These people generally might not be especially interested in such kind of entertainment or may not see how to fulfill their needs by using the Nabaztag. In line with this, people owning one or more of the devices listed also considered the Nabaztag's usefulness to be higher.

With regard to the situational determinants of positive evaluation, there was no significant difference whatsoever between those people who experienced a failure and those who did not. It may well be that people forgive the Nabaztag these functional errors without evaluating it as being less useful. The question that remains unanswered for the moment, however, is whether people excuse these technical complications only in a laboratory setting, or whether this effect endures in a long-term relationship when they own such a robot and it inhabits their home. Another limitation of this study is the fact that the user group comprised mainly students, rather than older people who seldom encounter computerized interfaces or toy robots. Also, people's reactions were not assessed after a long-term interaction, but after only 20 minutes, during which the participants got to know the rabbit and its functionality. However, the goal of the study was to analyze which factors influence the impressions people form of artificial companions, and for this, first hints were identified that might be addressed in greater detail in future long-term studies. Also, the application of the chosen questionnaires may be problematic. Although these methods are certainly useful, it is not clear whether self-report measures in general are useful in a novel application area such as the one we targeted. In line with this, Weiss et al. (2009) put forward, “[e]specially when interacting with a robot, individual user experiences might be heavily influenced by the individual's general attitude and the overall societal opinion” (p. 150). A more advantageous possibility might have been think-aloud protocols to assess in greater depth what people think and feel while interacting with the rabbit, especially in those situations in which the Nabaztag did not react as expected (e.g., by not reading the RFID tag).

STUDY II: EMOTIONS EXPRESSED BY “NONVERBAL” EAR COMMUNICATION

Nonverbal Communication: Functions and Implications

Because humans are involved in human-robot interactions, it is necessary to know about the (pre)conditions of (successful) human-human communication in order to be able to deduce specific design paradigms or hypotheses regarding the impression that is generated by the robot (also see Eimler, Krämer, von der Pütten 2010). Human communication can be described as a “multichannel reality” (Poyatos 1983, p. 175) consisting of language, paralanguage (i.e., vocal aspects such as intonation), and kinesics. The latter two are referred to as nonverbal behavior. Especially kinesics (i.e., the visual aspects of communication), constitutes a complex system of channels we encounter in our everyday experience: facial expressions, gazes, gestures, postures, and head and body movements (Wallbott 1994). Nonverbal communication fulfills specific communicative functions. Bente and Krämer (in press), unifying several approaches, suggest three functional levels: (1) Discourse functions; behaviors such as pointing, or illustrative gestures that are closely related to verbal behavior (Efron 1941; Ekman and Friesen 1969), (2) Dialogue functions; behaviors that serve the smooth flow of interaction when exchanging speaker and listener roles (Duncan 1972), and (3) Socio-emotional functions; behaviors affecting person perception, evaluation, and interaction climate. Highlighting the importance of socio-emotional functions, Burgoon (1994) suggests that, overall, approximately 60-65% of social meaning is derived from nonverbal behavior. Thus, nonverbal components of human communication have sustainable but subtle impact on our evaluation of people and situations (Argyle et al. 1970; Burgoon 1994; Mehrabian and Wiener 1967; Schneider, Hastorf, and Ellsworth 1979).

Related Work: Nonverbal Expressiveness in Artificial Entities

As a starting point for the generation of specific hypotheses about the elicited impression of a robot's behavior, and for the deduction of guidelines for the design of future robots that humans will want to interact and build relationships with, research from the field of human-human relationships can provide useful insights. In support of an approach that is oriented along human “modalities” as an anchor point, the following research results can be used: it can generally be observed that artificial characters such as robots elicit social behavior by their mere presence (Reeves and Nass 1996), they can evoke attraction (Lee et al. 2006) or socially desirable behavior (Schermerhorn, Scheutz, and Crowell 2008) in a user.

There is anecdotic evidence that nonverbal positive and negative feedback of a virtual audience has a strong impact on a human speaker (Pertaub, Slater, and Barker 2001). As indicated by von der Pütten et al. (2008), numerous agent systems feature nonverbal behavior that is meant to support the conversation between user and agent (see e.g., Cassell and Thórisson 1999; Gratch et al. 2007; Kopp et al. 2008). This includes gazes, body and head movements, gestures, and facial expressions such as raising eyebrows or smiling. In line with this, a study by von der Pütten et al. (2008) found that an agent giving emotional feedback increases the users' interest. Also, mimicry seems to have an essential impact. Bailenson and Yee (2005), for example, demonstrated that a virtual character that mimics the user is more influential. In studies conducted by Sommer (2007; Sommer et al. 2008), participants tended to mimic the agent: smiling behavior increased as a result of the smiling behavior of the agent.

Although the larger part of the results listed in the following refer to human-agent communication, it may well be transferrable to human-robot communication. In this regard, studies comparing robots and agents have shown that often robots are able to elicit even stronger effects with regard to closeness, familiarity, and enjoyment than agents (Yamato et al. 2001; Powers et al. 2007).

To conclude, because research results generally indicate a positive evaluation and impact of expressive behavior, nonverbal aspects in robots are important and should be taken into account. For a system intended to trigger a positive feeling in the user and to establish a (natural and social) long-term relationship, the capability of adapting to the user's communication modalities, at least in a rudimentary form (e.g., generation of any nonverbal expression) may be of importance. With regard to the robot rabbit Nabaztag, it can be stated that it generally has the capability of (rudimentary) nonverbal expression. It can move its ears and flash the lights that are located across its body. However, since nonverbal expressions are a very complex phenomenon, it is not clear how users interpret these ear movements and how they might be used to create specific impressions in the user.

Objective

Consequently, in this explorative study, we tested which impression of the robot's emotional status would be generated by each of six different ear positions that are one of the major components making up the robot rabbit's expressivity. Since robotic devices are designed for an international use by users with a variety of cultural backgrounds, an essential question to be answered is whether cultural differences in the interpretation of the robot's nonverbal behavior exist that may influence the perception of the Nabaztag and its acceptance. Indeed, there is evidence that culture influences the way we produce and perceive nonverbal signals (Matsumoto et al. 1989; Matsumoto 2006). However, it is unclear whether this holds for robots as well and whether it can be applied for the interpretation of basic nonverbal signals, too. In this study, we compared a U.S. American and a German sample with regard to their attribution of the emotions shown by the Nabaztag.

Method

In an online experiment, participants were confronted with six pictures of the rabbit in a within-subject design (see Figure 2). For each picture, people were asked to evaluate the robot rabbit's current emotional state. Verbal and visual methods were used. To assess the attributions made about the rabbit's mood, the Positive Affect and Negative Affect Schedule (PANAS) by Watson, Clark, and Tellegen (1988) was used. The PANAS is a psychometric scale measuring the positive and negative affects of 20 verbal items such as: afraid, proud, jittery, guilty, and interested. In this study, participants did not use it to indicate their own emotional state but to give an evaluation of the rabbit's emotional state. Participants rated the ear positions on a 5-point Likert scale (1 = very slightly or not at all; 5 = extremely) to indicate how much they thought the rabbit represented the specific emotion as they considered these different verbal items.

A factor analysis of the PANAS items resulted in three factors for the German sample that explain 68.44% of variance. Factors were named according to their constituting items: “Positive Evaluation” (31.45%; Cronbach's α = .936), “Embarrassment” (22.38%; Cronbach's α = .892) and “Hostility” (14.61%; Cronbach's α = .871). Table 2 shows the items and factor loadings for the three factors.

TABLE 2 German Sample, Factor Loadings Based on a Principal Components Analysis with Varimax Rotation for 20 Items Regarding the Attributed Mood of the Rabbit (N = 100)

Item	Positive evaluation	Embarrassment	Hostility
Interested	.853
Alert	.827
Inspired	.824
Attentive	.819
Enthusiastic	.802
Excited	.788
Active	.782
Strong	.659
Determined	.650
Proud	.621
Afraid		.869
Jittery		.789
Nervous		.746
Distressed		.736
Scared		.730
Guilty		.726
Ashamed		.725
Hostile			.857
Annoyed			.828
Irritable			.815

A factor analysis of the PANAS items resulted in two factors for the U.S. American sample. The factors explained 55.85% of variance and were named according to their constituting items: “Positive Affect” (29.470%; Cronbach's α = .909) and “Negative Affect” (26.384%; Cronbach's α = .895) (Table 3).

TABLE 3 U.S. American Sample, Factor Loadings Based on a Principal Components Analysis with Varimax Rotation for 20 Items Regarding the Attributed Mood of the Rabbit (N = 111)

Item	Enthusiasm	Fear
Enthusiastic	.829
Excited	.813
Active	.758
Interested	.753
Strong	.731
Determined	.715
Attentive	.710
Alert	.698
Proud	.697
Inspired	.861
Afraid		.831
Nervous		.817
Scared		.801
Distressed		.755
Guilty		.735
Ashamed		.705
Annoyed		.684
Irritable		.676
Hostile		.574
Jittery		.566

Moreover, the Self-Assessment Manikin (SAM) (e.g., Bradley and Lang 1994, Fischer, Brauns, and Belschak 2002) was used to evaluate the degree of Pleasure, Dominance and Arousal displayed by the rabbit. With regard to the PANAS, in contrast to the original use of this nonverbal pictorial technique for the assessment of affects while processing emotional stimuli, participants did use it to give an evaluation of the rabbit's emotions.

The three dimensions of this affective rating system that was originally designed by Lang (1980) have repeatedly been identified as important in human emotional judgment. According to Bradley and Lang (1994), “Empirical work has repeatedly confirmed that pleasure, arousal and dominance are pervasive in organizing human judgments for a wide range of perceptual and symbolic stimuli” (p. 49). For each dimension, several graphic figures, or manikins, represent specific manifestations of the respective emotional reaction. For example, the arousal dimension is represented by figures that range from an excited figure to a relaxed, almost sleepy figure. For the dominance dimension, the SAM uses changes in the size of figures; it ranges from a large figure, representing the feeling of being in control, to a small figure, referring to the feeling of being dominated (Bradley and Lang 1994). Completing these instruments, demographic data (e.g., age, gender) were assessed.

Participants

Participants were recruited via Social Network Sites (e.g., www.facebook.com), as well as mailing lists, newsgroups, and university message boards for the German sample. The American sample comprised 111 students from Michigan State University, who were informed about the study in lectures and who received credits for participating. The sample consisted of 100 German (57 female) and 111 U.S. American (28 female) students.

Procedure

Proceeding from the introductory screen, which referred to data handling and duration of the study, people were confronted with pictures of the rabbit showing six different positions of the rabbit's movable ears, as depicted in Figure 2. These pictures, which were shown in randomized order, alternated with the PANAS and the SAM to assess the emotional state people attributed to the robot, in other words, what they felt to be the actual emotion the rabbit showed in the respective picture.

Results: German and U.S. American Samples

Results: German Sample

Repeated-measure ANOVAs showed significant evaluation differences among the varying ear positions with regard to the factors Positive Evaluation (F (4.17, 413.25) = 56.028, p < .001; ηp ² = .361), Embarrassment (F (4.29, 424.44) = 14.154, p < .001; ηp ² = .125) and Hostility (F (3.63, 358.99) = 9.120, p < .001, ηp ² = .084). Also, significant differences emerged with regard to Pleasure (F (5, 470) = 29.913, p < .001, ηp ² = .241), Dominance (F (4.23, 397.78) = 23.456, p < .001; ηp ² = .200) and Arousal (F (3.97, 373.35) = 9.826, p < .001; ηp ² = .095). The results of the post-hoc tests for the different dependent measures listed in Table 4 show in detail which ear positions significantly differ from each other. For a more detailed differentiation of the respective positions with regard to the emotion attribution, it is also useful to look at the descriptive values given in Table 5 and 7.

TABLE 4 List of Significantly Differing Ear Positions Ordered by Dependent Variables (German Sample)

	Significant differences in post-hoc test
Ear position	Positive evaluation (PANAS)	Embarrass-ment (PANAS)	Hostility (PANAS)	Pleasure (SAM)	Dominance (SAM)	Arousal (SAM)
1	2,3,4	3,4,6	3	2,4,6	3,4,6	2,3,4
2	1,4,5,6	3,4,6	3,6	4,6	3,4,6	1,4,5,6
3	1,4,5,6	all	all	4,6	all	1,4,5,6
4	all	all	3	all	all	1,2,3
5	2,3,4	3,4	3,6	4	3,4	2,3
6	2,3,4	1,2,3,4	2,3,5	1,2,3,4	1,2,3,4	2,3

TABLE 5 Mean Values and Standard Deviations for the Attributions Made Regarding the Different Dependent Variables (German Sample)

	M (SD)
Ear position	Positive evaluation (PANAS)	Embarra-ssment (PANAS)	Hostility (PANAS)	Pleasure (SAM)	Dominance (SAM)	Arousal (SAM)
1	−.04 (.935)^3	−.05 (975)^3	−.12 (986)^3	3.20 (.846)^3	3.12 (1.184)^5	2.92 (1.028)^4
2	.64 (.894)^6	−.11 (.961)^2	.01 (.929)^4	3.40 (.949)^5	3.08 (1.191)^4	3.43 (1.373)^6
3	.51 (.874)^5	−.47 (.809)^1	.48 (1.349)^6	3.44 (.953)^6	3.54 (1.119)^6	3.35 (1.165)^5
4	−.91 (.699)^1	.46 (.991)^6	−.14 (.676)^2	2.17 (.921)^1	1.96 (.933)^1	2.55 (2.21)^1
5	−.00 (.931)^4	−.01 (1.058)^4	.03 (981)^5	3.24 (1.049)^4	2.91 (1.102)^3	2.82 (1.22)^3
6	−.19 (852)^2	.17 (.968)^5	−.24 (.810)^1	2.96 (.849)^2	2.75 (1.021)^2	2.76 (1.00)^2
N	100	100	100	95	95	95

TABLE 6 List of Significantly Differing Ear Positions Ordered by Dependent Variables (U.S. Sample)

	Significant differences in post-hoc test
Ear position	Positive affect (PANAS)	Negative affect (PANAS)	Arousal (SAM)	Pleasure (SAM)	Dominance (SAM)
1	2,3,4	n.s.	2,3,4,5	2,3,4	2,3,4
2	1,4,5,6	n.s.	1,4,5,6	1,4,5,6	1,4,5,6
3	1,4,5,6	n.s.	1,4,5,6	1,4,5,6	1,4,5,6
4	all	all	all	all	all
5	2,3,4,6	n.s.	all	2,3,4	2,3,4
6	2,3,4,5	n.s.	2,3,4,5	2,3,4	2,3,4

TABLE 7 Mean Values and Standard Deviations for the Attributions Made Regarding the Different Dependent Variables (U.S. Sample)

	M (SD)
Ear position	Positive affect (PANAS)	Negative affect (PANAS)	Arousal (SAM)	Pleasure (SAM)	Dominance (SAM)
1	−.197 (.771)^3	−.068 (.894)^3	2.82 (1.014)^3	3.05 (.848)^2	2.92 (.987)^4
2	.6449 (.933)^6	−.049 (.967)^5	3.82 (1.248)^6	3.62 (1.060)^5	3.40 (1.245)^5
3	.5493 (.934)^5	−.247 (.909)^1	3.59 (1.060)^5	3.71 (.997)^6	3.63 (.983)^6
4	−.774 (.855)^1	.539 (1.046)^6	2.29 (1.180)^1	2.29 (1.054)^1	2.20 (1.172)^1
5	.0106 (.934)^4	−.058 (1.108)^4	3.17 (1.229)^4	3.18 (1.142)^4	2.92 (1.133)^3
6	−.234 (.824)^2	−.118 (.885)^2	2.72 (.915)^2	3.08 (.817)^3	2.70 (1.020)^2
N	111	111	99	105	105

Displaying ear position 4, which, according to Table 4, significantly differs from all other ear positions, the Nabaztag is evaluated least positive and most embarrassed when its ears are horizontally oriented (see Table 5). Also, people attributed the lowest degrees of pleasure, dominance, and arousal to the rabbit when it displayed this ear position. This position also is associated with a relatively low degree of hostility. Only ear position 6, in which one ear is upright and the other one horizontally oriented (see Figure 2) is evaluated to express less hostility.

Ear position 2 (see Figure 2), in which the rabbit's ears are in upright position, is associated with the highest degree of positivity and arousal. When the rabbit had its ears in the upright position, slightly bent forward (position 3), participants attributed the highest values of pleasure and the lowest degree of embarrassment. People interpreted this ear position as the strongest expression of hostility and dominance. All in all, ear position numbers 1 and 6, which show the rabbit with one upright ear and one horizontal ear, are not associated with extremely high or low intensities of positive feelings, hostility, pleasure, or arousal (see Table 5 for mean values). The only two exceptions are ear position 6 that reaches the second highest attribution of embarrassment, and ear position number 1, reaching the second highest attribution of dominance. Since both positions differ only in the sense that their ear position is mirrored (see Figure 2), one would have expected to find comparable results. However, considering all the rest of the evaluations of ear position 1 and 6, there is not a single match in the results.

Results: U.S. American Sample

Repeated-measure ANOVAs were also calculated for the American sample. As with the German sample, they showed significant mood evaluation differences among the ear positions with regard to the Positive Affect (F (4.353, 478.779) = 49.499, p < .001, ηp ² = .310) and Negative Affect (F (4.302, 473.243) = 12,731, p < .001, ηp ² = .104). Also, significant differences among the ear positions emerged with regard to Pleasure (F (4.416, 459.305) = 27.932, p < .001, ηp ² = .212), Dominance (F (4.288, 445.985) = 23.673, p < .001, ηp ² = .227) and Arousal (F (4.402, 431.364) = 28.760, p < .001, ηp ² = .227). Table 6 shows in detail which ear positions differed significantly in the impression they evoked.

Looking at the descriptive values, it can be said that compared to all other ear positions, the horizontal ear position (position 4, see Figure 2) let people perceive the rabbit's emotional status least positive with regard to Positive Affect, and let them attribute the highest values regarding Negative Affect. In this ear position, the rabbit was thought to feel the lowest degree of arousal. Also people attributed the lowest intensities of pleasure and dominance to the rabbit when its ears were horizontally oriented.

When the rabbit's ears were in upright position (position 2, see Figure 2) people attributed the highest values with regard to Positive Affect. Ear position number 3, which shows ears upright and bent forward, let people perceive the rabbit's emotional status to be least intense with regard to Negative Affect. Also, the highest degrees of pleasure and dominance in the rabbit were seen when it displayed ear position number 3 (ears upright and bent forward). Arousal was strongest with ear position number 2 (ears upright).

As in the German sample, ear positions numbers 1 and 6 that show the rabbit with one upright ear and one horizontal ear are not associated with extremely high intensities of any feeling in the rabbit. However, ear position 1 reaches the second-lowest mean value with regard to pleasure. Except for pleasure, people attributed the second-lowest mean values in enthusiasm, factor 2/fear, arousal, and dominance to the rabbit when showing ear position 6.

Comparison and Discussion

Results show that people indeed attributed specific emotional states to the rabbit. Most of the ear positions differed with respect to what people thought to be the emotional state or the intensity of the emotionthe rabbit expressed. Ear position number 4, in which the rabbit's ears are horizontal, has led people to attribute lowest values in Pleasure, Arousal, and Dominance compared with all other ear expressions. This result applies to both the German and the English samples.

With regard to the factors that resulted from the factor-analyzed PANAS items, differences between the German and the American samples were found. While the data collected for the German sample resulted in three factors, the U.S. data built only two factors. Considering differences and similarities regarding Pleasure, Dominance, and Arousal, it can be observed that the German and the U.S. samples have a number of similarities. Both samples considered ear position 2 (ears upright) to express the highest degree of arousal, followed by ear position 5 (ears bent backwards). The lowest degree of arousal was assigned to ear position number 4 (horizontal ears) by both samples, followed by ear position 6 (one horizontal ear, one upright). Both groups saw the highest intensity of pleasure in ear position 3 (ears bent forward) and the lowest in ear position 4 (horizontal ears). With regard to dominance, both samples assigned the highest dominance values to ear position number 3, the lowest to position 4, followed by positions 6 and 5 (please refer to Figure 2 for pictures of the ear positions).

Generally, there are differences between German and English users, however, these differences are not extremely prominent. On the one hand, it is astonishing that differences emerged at all, particularly because, when it comes to nonverbal communication, many signals are interculturally transferrable. Also, similarities in interpretations might be explained by the fact that, at least when resorting to Hofstede's (1980, 2001) cultural dimensions, Germans and Americans are not extremely different from one another.

Results from the U.S. American and the German samples show that observers interpret the ear positions of a rabbit-shaped robot. Results may, to a certain extent, be attributed to the Nabaztag's reference to a real animal or pet. Humans also infer specific intentions and emotions from their pets' ears' positions (apart from other indicators such as dogs wagging their tails).

It can be concluded that nonverbal communication is not only an essential part of human-human communication and human-agent communication, but is also important in human-robot interaction, even when the robot has only rudimentary capability to express itself. Ear positions, or capabilities of expression that may seem unimportant at first, need to be carefully considered because, as these results show, they contribute to the impression the user forms about his or her artificial counterpart.

With regard to enrichment of the user's experience (also see the Beyond Usability section) and also as a precondition for the establishment of relationships and long-term appeal, these nonverbal expressions are important and can be pointedly used to support verbal messages and to express or mirror emotions, attitudes, or moods. Ear movements have also been implemented in the third iteration of the SERA data collection as a sign of empathy with the user. In response to the question “How are you feeling after today's activity?” or “What do you think you would give that on a scale of 1 to 5?” the rabbit would say “Oh. Perhaps it'll be better tomorrow” while its ears would go down if the participant shows the rabbit a card indicating that his or her mood is worse than the day before.

However, despite providing useful insights, these results are not without limitations. Another aspect that should be mentioned here is of methodological nature. In this study, we confronted people with pictures of the rabbit rather than with the real robot in a laboratory setting. We cannot be absolutely sure whether the same results would be obtained in a setting where people would have to interpret the rabbit's emotional status in direct confrontation.

Furthermore, it has to be kept in mind that the interpretation of nonverbal expressions has to be considered in connection with the verbal context in which they appear. It has been demonstrated that the effects of nonverbal signs are modulated by their verbal context (Krämer 2008). Also, from human-human communication, we know that people tend to rely on the nonverbal signals when verbal and nonverbal information is inconsistent. Future studies should concentrate on the exploration of the question of whether this is the same with robots. Do people rely on the rudimentary nonverbal signals provided by, for example, the Nabaztag, when nonverbal information is accompanied by verbal messages? Does this depend on the impression that users have about the robot's agency? However, not only the productive capabilities of robots displaying nonverbal behavior need to be taken into account; equally important for the development of future systems and the development of long-term relationships is the robot's capability of interpreting its counterpart's nonverbal expression. This may, on the one hand, increase the chance of an adequate reaction and, on the other hand, provide the possibility of rudimentary forms of mimicry that might in turn increase the acceptance of the robot companion by its user.

GENERAL DISCUSSION

This article introduced two studies that explore the reception of the robot rabbit, Nabaztag. Both studies were intended to collect data from a larger sample of potential users in order to inform not only the design of the scenarios for the iterations of the SERA project but especially to deduce specific aspects worth considering in future settings with robots. Study I looked into the overall evaluation of interacting with the Nabaztag and explored the personal and situational determinants of a positive user experience. Findings show that the attribution ofnovelty and challenge were most prominent in the evaluation. The expression of personal values and self-presentation were considered to a lesser extent; usability was considered comparably low. These results show that, at least with regard to the initial interaction with the Nabaztag, a rather low level of usability does not negatively influence the positive emotional attribution. Also, the fact that even when the interaction did not flow smoothly, people did not evaluate significantly differently nor more negatively, respectively. Future studies should therefore focus on the conditions and the limits of this positive emotional attribution and the role of “forgiving.”

According to the results obtained in this study, women evaluate the interaction more positively than men do. This connection between women and technology is an aspect that should be explored in more detail. Future studies should identify the role of the robot's shape and the displayed functionality in this regard.

Also, people owning a robot toy evaluated the Nabaztag more positively, although the degree of computer experience did not significantly influence the evaluation. Besides the influence of gender, this may be another indicator of the need to provide individualized robots. Very specific characteristics of a user seem to determine the appeal springing from a robotic device; these need to be identified in order to increase the potential of robots to be accepted by the potential user and to engage especially current nonusers as well as those people who are robot users and owners already.

The second study has specifically looked into the importance of nonverbal communication and has illustrated that ear positions of the Nabaztag mean something to the user and thus have to be used with care. Also, this study has drawn attention to cultural differences in the interpretation of a robot's nonverbal expressions, finding that there are only minor differences between the U.S. and the German samples.

All in all, the results draw attention to the fact that emotional aspects are essential when it comes to the interaction and acceptance of robots and may, in the long run, determine the likeliness of the establishment of a human-robot relation. Although culture seems not to be a key factor in the evaluation, individual differences such as gender and previous experience with similar technology determine how appealing a robotic device like the Nabaztag is to the user. As still very little is known about the socio-demographic characteristic of present robot owners, future studies should more specifically look into person variables and usage patterns of specific groups against the background of the robot's shape and functionality. Finally, both studies have also shown that there is a need for a questioning of existing methods and the development of new instruments that are tailored for research in the field of human-robot interaction.

Acknowledgments

The authors would like to thank Lars Bollen, Lobna Hammadi, and Tina Ganster. This study was supported by funding from the European Community's Seventh Framework Programme [FP7/2007-2013] under grant agreement no. 231868 and project name Social Engagement with Robots and Agents (SERA).

Notes

Note: Factor loadings <.6 are suppressed.

Note: Factor loadings <.5 are suppressed.

Note: Superscripted numbers behind the SD values indicate the order of the mean values, beginning with 1 indicating the lowest degree of the respective manifestation.

Note: Superscripted numbers behind the SD values indicate the order of the mean values, beginning with 1 indicating the lowest degree of the respective manifestation.