Full citation is:   Ó Libin, A., Libin, E., Ojika, T., Nishimoto, Y., Takeuchi, T., Matsuda, Y., Takahashi, Y. (2002). On Person – Robot Interactions: Cat NeCoRo Communicating In Two Cultures (Phase 1. USA – Japanese study). Proceedings of the 8th International Conference on Virtual Systems and Multimedia, Creative Digital Culture, VSMM Society, Seoul, pp. 899 – 905.

Person – Robot Interactions:

NeCoRo Cat Communicating In Two Cultures

(Phase 1. USA – Japan study)

Alexander LIBIN^1,2, Elena LIBIN³, Takeo OJIKA⁴, Yutaka NISHIMOTO⁵, Tomiko TAKEUCHI⁵, Yoshimi MATSUDA⁵, Yukiko TAKAHASHI⁵

1.Department of Psychology, Georgetown University, Washington DC, U.S.A.; 2.Research Institute on Aging, Rockville, MD, U.S.A.; 3. Robotherapy Institute at the Psychological Research for Complex Interactive Systems, Inc., U.S.A.; 4. Institute of Dream Systems Development, Japan; 5. Nursing Course, Gifu University School of Medicine, Japan

libina@georgetown.edu

Abstract. This is a pilot study on person–robotic cat NeCoRo interactions aimed at a psychological examination of human reactions emerging from communication with an artificial creature imitating an animal’s behavior. We implemented a research methodology based on the concept of congruence between living and artificial systems [6]. Each of the 32 participants (16 Americans and 16 Japanese) of both genders and two age groups (19-28 and 60-79) was engaged in a 15-minute session with an artificial creature with a cat-like personality. The robotic cat NeCoRo was designed by Dr. Shibata, and its commercial version was manufactured by Omron Corporation (Japan) in 2001. An assessment of person – robotic cat interactions was performed via the Person – Robot Complex Interactions Scale [11]. Certain age and gender differences, as well as differences across American and Japanese cultures are discussed.

1 Introduction.

       A person’s interactions with an artificial environment have many various aspects. Nevertheless, one determining factor remains the same. Willingly or unwillingly, humans create an artificial world and its inhabitants by projecting their own image and characteristics of the world they live in. In the same way that people and other living beings differ from each other by various parameters, such as weight and height, behavioral reactions and character, abilities and coping strategies [1] [2], artificial creatures have their own distinct individuality [3] [4] [5] [6]. Therefore, analyses of  person – robot interactions should account for the essentials of both human perceptual, cognitive and emotional processes and robot’s design and behavioral configuration.

      The factors influencing differences in human performance are biological (i.e., sex and age), psychological (i.e., past experiences – tactile, emotional, cognitive etc.), and social (i.e. culture). Robots’ physical (i.e., hardware), aesthetical  (i.e., appearance), and behavioral (i.e., functionality) parameters also vary – and here we would like to follow Rodney Brooks’ eloquent definition [7] – according to their existence described as a “robot-being”.

        We present an approach to studying person – robot communications aimed at the analysis of an interface between general tendencies in human interactions with the artificial creature and its cultural specifics. The culture is understood as a system of created social meanings and artifacts, which have symbolic value and reflect the logic of human mentality derived from norms shared by given community [1] [8] [9].

2 Method

Objectives of the study

     The goal of the conducted pilot study was to show how general (i.e., age, sex, past experiences) and cultural factors (i.e., based on social traditions, communication style) influence the specifics of interactions between a person and the communication-type robot NeCoRo (which means ‘companion’ in Japanese) [10]. Our research aimed to examine preferences in behavior and features of the robotic cat NeCoRo through the analysis of: 1) age and gender differences in person-robot interactions; 2) cross-cultural specifics of interactive patterns; 3) the influence of a person's past experiences with alive pets and modern technology on the type of person-robot communication.

      The conducted pilot study is the first step of a bigger international project on psychological analyses of human – robot interactions.       

Subjects

     Two groups of participants of different ages were recruited for the person – robotic cat study in the USA and Japan. The USA group included 9 persons with the mean age of 28.4 years, and 7 persons with the mean age of 67.5 years. Sixty-seven percent of younger participants and 28% of older persons were females. The Japanese group included 10 persons with the mean age of 21.6 years, and 6 persons with the mean age of 72.5 years. Ninety percent of participants in the younger group and 33% of older participants were females. Informed consent was obtained from all participants. The following Schema 1 illustrates the recruitment procedure.  

Instruments

Person-Robot Complex Interactive Scale (PRCIS). The PRCIS [11] includes two subsections for participant's evaluation and two subsections for instructor's evaluation. The first part of the participant's evaluation scale has five subscales for the assessment of a person's likes and dislikes regarding the robot's behavior, features and interactions. The five-points Lickert scale was employed and ranged from "not at all" to "greatly " for the evaluations of (1) tactile, (2) manipulative, and (3) communicative behaviours. Robot's features (4) and quality of the person-robot interactions (5) were evaluated via multiple-choice subscales. Also, participants were given an opportunity to express their opinion about robotic cat advantages and disadvantages through open-ended questions. The second part of participant’s section includes a special subscale about the person’s experiences with both real pets and modern technologies. Instructor evaluation (part 3) concerns verbal and non-verbal, emotional, and animated aspects of participant's communication with the robot during the session. All four subscales measure behaviours using a five-point Lickert-type assessment. The instructor’s overall evaluation (part 4) of person – robot interactions consists of a multiple-choice scale that has to be completed after each session.   

Procedures

     The study was conducted at two locations: the Institute of Robotherapy at the Complex Interactive Systems, Inc. (Chevy Chase, USA) and Gifu University (Gifu, Japan). Informed consent was obtained from every participant. Each participant received a standardized introduction to the robotic cat, followed by a 15-minute interactive session with the robotic cat NeCoRo. During the session the instructor observed the participant at all times and evaluated the quality and intensity of the interactions via PRCIS, Part 3. Direct observations were performed twice -- during the first and last three minutes of the session. In addition, each session was videotaped.

Once an interactive session was completed, the participant evaluated his\her own experience with the robotic cat as well as assessed NeCoRo`s features, advantages and disadvantages (PRCIS, Part1). Also, a scale measuring participant's past experiences with real pets and modern technologies was administered (PRCIS, Part2). At the same time the instructor performed an overall assessment of the session using PRCIS, Part4.

3 Results

    A between-subjects research design was used. For studying age and gender specifics of the participants’ interactions with NeCoRo, both Pearson correlations (r) and Levene’s test for equality of variances as a stastistical measure (F) of the differences between the groups were used. Analysis of the cross-cultural differences was performed via t-criteria (an independent samples test). Only outcomes with the confidence level no less than p < 0.05 were considered for the interpretations. Distributions of the used variables were fairly symmetric and had no outliers.

3.1. Age and gender differences in person-robot interactions

Overall age differences

     Older people enjoy it more than younger people when the robotic cat `meows` (r = 0.44, p = 0.13) and when the robotic cat listens if spoken to (r = 0.47, p = 0.008). With age the evaluation of the interactive session as “interesting” and “exciting” progressed (r = 0.40, p = 0.03). Interestingly enough, younger participants were less active in tactile interactions with the cat, such as stroking its body, rubbing behind the ears, playing with the paws, tail, or ears, stroking the back with their palm open (average r = 0.55, p = 0.002). Analysis of the past technological experience scale showed that younger participants use cellular phones, computers and household devices more intensively then older adults and enjoy more technology in general (average r = 0.58, p = 0.002).

Overall gender differences

     No differences between sexes were found on the scale of tactile and manipulative interactions. However, males from both cultures like the cat's active behavior. They enjoy it when the cat turns its head around, closes and opens its eyes, changes its posture (F = 9.57 p = 0.004). The highest gender differences were found for the item “I like when the cat opens and closes its eyes”(F = 12.60, p = 0.001). Males also enjoy it a little bit more while stroking cat's back with their palm open (F = 5.60, p = 0.03). Surprisingly, females enjoy it less when the cat cuddles (F = 5.62, p = 0.002) and moves its ears (F = 5.12, p = 0.003). Overall, males describe interactions with the cat NeCoRo as “exciting to play with”(F = 9.72, p = 0.004), whereas females were tend to evaluate the cat as “boring to interact with”(F = 9.78 p = 0.004).

3.2.  Cross-cultural specifics of interactive patterns 

     It seems that Americans enjoy touching the robotic cat a little bit more than the Japanese. Overall tactile interactions were more intensive for USA. participants (F = 4.03, p = 0.05). The highest cultural differentiation was found for the behavior from the tactile interaction scale indicated as ‘touching cat's body’ (F =  7.02, p = 0.01). However, for other items of tactile scale no significant differences were found.

     On the scale, that evaluates the cat's behavior, cross-cultural specifics were established for the item “looking in the cat's eyes” (F =  7.8, p = 0.009). It seems that Japanese people do not like it when the cat looks at them. Americans like it a lot more than Japanese participants when the cat cuddles while being stroked (F =10.7, p = 0.003). This feature has the most distinct cultural connotation. The second subsection of the instructor’s evaluation, which measures overall satisfaction with the interactive session, shows that Americans in general consider interactions with the robotic cat more exciting and interesting (F = 8.76, p = 0.006).

3.2.1.Cross-cultural specifics of interactive patterns for younger adults

       Some interesting results were obtained after the comparison of two groups (Japanese and American) of younger adults. Young Japanese adults have an extreme dislike for when a cat cuddles while they stroke it (F = 12.0, p = 0.003). Americans like when a cat makes sounds when it moves, while Japanese people do not like it (F = 5.30, p = 0.03). Slight differences were found for the item “exciting to play with”. Young Japanese participants were not as exited about playing with the robotic cat as their oversea counterparts (F = 4.60, p = 0.04).  In general Japanese youngsters were less interested in interacting with the robotic cat than American young people (F = 5.80, p = 0.03).

         Experience with new technology, such as the use of a cellular phone had the most distinct cultural influence. Japanese youngsters enjoy cell-phone applications far more than their American counterparts (F = 26.76, p = 0.0001).     

3.3. Influence of person's past experiences with real pets and modern technology on 

       communication with the robot.

       Among the general sample more than fifty persent of participants in both cultures admit that they adore pets. No age or sex differences were found on the item ”I like pets”. However, as expected, the degree of a persons’ liking pets influences their style of interactions with the robotic cat. The more people like real pets, the higher score they have on a subscale of tactile interactions on such behaviours as rubbing behind the ears or strocking the back with their palm (r = 0.37, p = 0.04). And, of course, pet-lovers evaluate NeCoRo as an exiting robot to play with (r = 0.47, p = 0.007).

     We also looked at how past experiences in interactions with real cats influenced participants’ communication with the cat NeCoRo. Those participants in both cultures who had rich experiences of interactions with real pets enjoyed more manipulative activities with the robotic cat, such as picking it up (r = 0.45, p = 0.01) and keeping on their lap (r = 0.37, p = 0.03). The more experience the participants had, the more they liked how the robotic pet cuddles (r = 0.37, p = 0.04). Greater experience with real pets is associated with the highest scores on the item “cat…makes unnecessary noise”(r = 0.38, p = 0.03).

      Analyses of differences with independent sample t-criteria showed that pet-lovers like stroking NeCoRo (F = 5.28, p = 0.03). They also like the robotic cat’s responses to touch (F = 5.01, p = 0.03), but at the same time they consider that the cat NeCoRo is heavy to pick up and to hold (F = 7.60, p = 0.01). Participants with greater real pet experience are more likely to kiss the robotic cat (F=5.15, p=0.03). They also see the discrepancies between a real and artificial cat more clearly, indicating that NeCoRo is “heavy to pick up”(F = 11.1, p = 0.002) and that it produces “unnecessary noise” (F = 20.3, p = 0.0001). At the same time, those who show greater activity with real pets are more inclined to consider that the robot's behavior is very interesting to observe (F = 4.86, p = 0.03).

3.3.1. Cross-cultural specifics of the interface between past experience with a real

          pet and communication with the robotic creature

American pet-lovers

     With the American group, the degree of liking a real pet was associated with how the robotic cat moves its tail (r = 0.72, p = 0.002). Overall, Americans who love cats have higher scores on scales evaluating NeCoRo’s behavior and responses (r = 0.54, p = 0.03). They like it when the cat is active and interactive with them. Those who indicate extensive past experience in communicating with real pets do not consider that NeCoRo is good to touch (r = - 0.58, p = 0.02).

Japanese pet-lovers

     A slightly different scenario was encountered with the Japanese group. Those who love real pets also liked the color of the NeCoRo cat (r = 0.5, p = 0.02). Japanese people with greater experience in interacting with real pets enjoyed it more when interacting with the NeCoRo cat by stroking the cat's back with their palm (r = 0.51, p = 0.04), picking the cat up (r = 0.60, p = 0.01), and keeping the cat on their lap (r = 0.53, p = 0.003). Liking a real pet is associated with  liking the robotic cat’s cuddling in response to stroking (r = 0.59, p = 0.001).

     Overall, American pet-lovers like picking up the robotic cat more than the Japanese pet-lovers (F = 10.5, p = 0.004). Also they enjoy keeping the cat on their lap more (F = 6.6, p = 0.01) and turning the cat over (F = 6.0, p = 0.002). Japanese pet-lovers like it less when the robotic cat turns its head around  (F = 14.3, p = 0.001). Japanese participants like it less than Americans when the robotic cat changes its posture (F = 6.7, p = 0.01) and looks at them (F = 10.1, p = 0.005). Americans enjoy it more when the cat cuddles while being stroked  (F = 7.2, p = 0.01) and makes a sound when it moves  (F = 11.5, p = 0.003, n = 22). Japanese people, more than Americans, consider that the cat makes unnecessary noises  (F = 13.0, p = 0.002). Americans enjoy touching the robotic cat far more than the Japanese participants (F = 29.6, p = 0.0001). Also, Americans consider NeCoRo’s character to be friendlier than the Japanese do, who evaluated the robotic cat as a bit aloof (F = 9.1, p = 0.007).  

4 Conclusion

        The higher level of appeal of interacting with robotic cat by older participants shows that this animal-like artificial creature meets their needs and is a more desirable companion for them than for the younger participants. This result confirms the intentions of the NeCoRo cat developers to create a robotic pet who might be a partner, a friend, and a companion for the older person. Our data also show that the intensity of interactions with the technology, predetermined by past experience, is not among the parameters that make the prognosis of interest in the robot. Though young people receive more pleasure from technology in general, their evaluation of the interactive session with NeCoRo as exciting and interesting was lower than in older adults.

           Preliminary results showed that past experience with real pets were positively associated with intensity and variety of tactile communication with the robotic cat. The degree of liking real pets influenced the way people interact with the robotic cat. Experience with a real pet also influenced the evaluation of robotic pet features by participants. Presumably subconsciously comparing an artificial cat with a real one, participants named heavy weight and unnecessary noise as unattractive features of NeCoRo.

     Although pet-lovers in both cultures enjoy touching and manipulating the robotic cat, Americans have higher scores on both scales. The robotic cat's expressive behaviors (i.e., direct looking, cuddling) seem more enjoyable for American participants.

References

[1] Libin, A. (2001). Differential Psychology: On the Crossroad of Russian, American, and European

      Traditions. 2^nd edition. Moscow: Open Society Institute/Smysl. 580 p. (in Russian)

[2] Libin, E. (2001). Coping with life crises and difficult situations. Moscow: PerSe. 610 p. (in Russian)

[3] Shibata, T., et al. (1996). Emotional robot for intelligent system – Artificial Emotional Creature

      Project,  Proceedings of the 5^th IEEE International Workshop on ROMAN, pp. 466-471.

[4] Brooks, R.A. (2002). Flesh and Machines. New York: Pantheon  Books. 315 p.   

[5] Breazeal, C. L. (2002). Designing sociable robots. MIT: MIT Press, 325 p.

[6] Libin, A. (2001). Virtual reality as a complex interactive system: A multidimensional model of a person

      – artificial partner co-relations. In: H. Thwaites, & L. Addison (Eds.), Proceedings of the Seventh

      International Conference on Virtual Systems and Multimedia.  Enhanced realities: Augmented and

      unplugged, pp. 652-657.  Los Alamitos, CA: IEEE Computer Society.