دستیار متقاعد کننده روباتیک برای سلامت خودمدیریتی افراد مسن: طراحی و ارزیابی رفتارهای اجتماعی

Daily health self-management, such as the harmonization of food, exercise and medication, is a major problem for a large group of older adults with obesity or diabetics. Computer-based personal assistance can help to behave healthy by persuading and guiding older adults. For effective persuasion, the assistant should express social behaviors (e.g., turn taking, emotional expressions) to be trustworthy and show empathy. From the motivational interviewing method and synthetic assistants’ literature, we derived a set of social behaviors, and implemented a subset in a physical character, a virtual character and a text interface. The first behavior type concerns conversing with high-level dialogue (semantics, intentions), which could be implemented in all 3 assistants. The other behavior types could only be implemented in the characters: showing natural cues (e.g., gaze, posture), expressing emotions (e.g., compassionate face), and accommodating social conversations (e.g., turn taking). In an experiment, 24 older adults (45–65) interacted with the text interface and one of the characters, conform a “one-week diabetics scenario”. They experienced the virtual and physical character as more empathic and trustworthy than the text-based assistant, and expressed more conversational behavior with the characters. However, it seems that the preference of interacting with the character or the text interface was influenced by the conscientiousness of the participant; more conscientious people liked the text interface better. Older adults responded more negative to the characters that lacked the social behaviors than to the text interface. Some differences between the virtual and physical character probably occurred due to the specific constraints of the physical character.

In the year 2000, one in ten individuals in the world was 60 years or older and one in fourteen was at least 65. It is expected that these numbers will increase to one in every five persons being 60 or older and nearly one in six people 65 or older in 2050 (United Nations, 2002). On top of that, the prevalence of chronic diseases is rising amongst older people because of urbanization and an unhealthy lifestyle (Wild et al., 2004) (e.g., diabetes, COPD, obesities). A major problem is that only 50% of the chronically ill adheres to their treatment advice (WHO, 2003). For older adults, this problem is particularly hard because of their health illiteracy and deep-rooted daily routines—or lifestyles. Information and Communication Technology (ICT) might provide the required support to better cope with the personal health constraints, such as doing exercises (Kidd and Breazeal, 2006; Ruttkay et al., 2006; Bickmore et al., 2004; Goetz et al., 2003; Gockley and Mataric, 2006), giving social support (Kidd et al., 2006; Kriglstein and Wallner, 2005), and helping with lifestyle change (Bigelow et al., 2000; Looije et al., 2006). Research on persuasive technology (Fogg, 2002) and affective computing (Picard, 1997) is providing technological (partial) solutions for the development of this type of assistance, e.g., for the realization of social behavior, such as social talk and turn taking (Bickmore et al., 2004; Kidd et al., 2006; Goetz et al., 2003; Gockley and Mataric, 2006), and of empathic behavior, such as attentiveness and compliments (Kidd et al., 2006; Bigelow et al., 2000; Looije et al., 2006; Kriglstein and Wallner, 2005). In our view, psychological techniques for behavioral change, such as motivational interviewing, should be accommodated by this type of ICT support (e.g., Rogers, 1951; Looije et al., 2006; Miller and Rollnick, 1991). However, a concise and coherent set of behaviors – worked-out in specific user-interface behaviors – for such an accommodation is lacking. Derived from relevant literature of psychology, persuasive technology and affective computing, this paper presents a first set of behaviors (e.g., compassion) that map support objectives (motivating, educating, and supporting) on specific – social – user-interface behaviors for the intended ICT support. In an experiment, we tested how far these social behaviors help to make the interface more empathetic and trustworthy, which are preconditions for long-term use. It should be noted that older adults experience specific hindrances to actually make use of ICT support, due to relatively limited computer skills (partly due to limited sensory, physical and cognitive abilities), and motivation to use a standard Windows, Icons, Menu and Pointing (WIMP) device (Czaja and Lee, 2007). An important question is how to realize user interfaces with which older adults feel at ease and which are pleasant in use (Czaja and Lee, 2007). Different user-interface types for older adults’ health assistance have been developed, such as text or WIMP-based questionnaires and feedback providers (Blanson-Henkemans et al., 2007; Bigelow et al., 2000), and character educators and buddies (Looije et al., 2006; Kidd et al., 2006; Goetz et al., 2003; Gockley and Mataric, 2006; Kriglstein and Wallner, 2005). A speech interface may be more natural to use than a text interface for people who are not experienced with computer technology (Neerincx et al., 2008). A user interface with a character-like appearance (e.g., animal) could further improve the feeling of comfort of older adults with technology (Kidd et al., 2006; Kriglstein and Wallner, 2005). Social behavior increases when a character-like appearance is used (Kidd et al., 2006; Kriglstein and Wallner, 2005) and as a result the resistance towards the interface might decrease. However, empirical research on possible benefits of these types of user interfaces on older adults’ appreciation and conversational behavior is lacking. This paper presents an experiment, comparing a text-based interface with a – virtual and physical – character providing health assistance for older adults. For this comparison, we used the currently most common dialogue: written input and output (keyboard and window) for the text (WIMP) interface, and spoken input and output for the character. Overall, the expectation was that social behaviors improve users’ appreciation and enriches their interaction behavior; i.e., more talking, looking, laughing will occur. Furthermore, it is expected that such user-interface behaviors can be best incorporated in characters in comparison with virtual characters and text interfaces. However, we assume that characters which do not show the required social behavior evoke relatively large negative responses of the users. Finally, we expect that the personality of the user influences his/her preference for a specific interface.

In this study, our aim was to find behavior for an electronic personal assistant that improves the self-care capabilities of older adults. The psychological method of “motivational interviewing” provides a sound basis to persuade adults to behave healthy. Based on this method and recent research on affective computing, we derived two objectives of a persuasive assistant, to show empathy and establish trust, and found four social behaviors that could be mapped onto these objectives: high-level dialogue, natural cues, use of emotions, and social dialogue competencies. As far as possible, these behaviors were implemented in three assistants – a (classical) text interface, virtual character and a physical character – incorporating, increasingly natural, social behavior. The results of our experiment partially supported three of the four hypotheses on the positive effects of these behaviors. First, the social characters were found to be more empathetic than the text interface, and the characters evoked more conversational behavior, both as expected. It should be noted that typing with the text interface took substantially more time (more than twice) than talking with the character without increasing the content. In this experimental setting the participants were inclined to make the effort, but it is unclear whether they will still do so with long-term use. Speech, as implemented in the character interfaces, removes some hindrances that older adults might experience when they use classical input devices such as keyboard and mouse. Regarding the second hypothesis, the non-social characters were perceived as less trustworthy than a text interface, and were appreciated less, exactly as we expected. Regarding the third hypothesis, social characters were perceived as more empathetic and trustworthy than non-social characters, and evoked more conversational behavior, also as expected. The non-social physical iCat was seen as annoying by the participants, because it ignored them (e.g., looked past them) and in that way ignored conversational conventions. This indicates that it is of utmost importance that behaviors are social behaviors, because otherwise the behaviors are contra-productive. Regarding the fourth hypothesis, the physical characters were indeed found to be more trustworthy but less empathic than the virtual character, which was not expected (cf. (Lee et al., 2006), who found little differences between virtual and embodied characters). This negative outcome on empathy might be due to specific constraints of the iCat: it makes relatively much noise when it moves, and the head and body movements are perhaps not fluent enough. Another technical constraint was the (occasionally) appearing of errors in the movements and speech, such as skipping choices of the multiple choice questions. Furthermore, it may be that our three character roles did not capture important advantages of a physical character that can act in the real environment. For instance, more positive outcomes might show up with a character that helps to attend to a medicine box with a specific location in house, compared to a virtual character that is not a real actor in this house (c.f. Shinozawa et al., 2005). It was promising to notice that almost 90% of our participants asserted that they would use a personal assistant if they needed one, and that the iCat and text interface were equally preferred (both 50%). It seemed that conscientious persons particularly prefer a text interface. Furthermore, it was promising that most participants had no trouble learning to interact with the iCat assistants, perhaps indicating that there may be little hindrances to make use of the iCat for older adults. The dialogue should be made more varied in the next version, both for the text and character interfaces: participants who disliked the assistant thought it was repeating too many questions and gave too often similar responses. By using a speaking and hearing, emotional, physical character in this experiment, we choose for the most “natural” human-character communication and compared it with a conventional text chat. Subsequent experiments should study more in detail what the costs and benefits are of the different modality choices. Daily health self-management, such as the harmonization of food, exercise and medication, is a major problem for a large group of older adults with obesity or diabetics. Persuasive assistants that are emphatic and trustworthy may help solve this problem if they apply principles from the psychological method of “motivational interviewing”. Although our experiment was limited in its methodology by the fact that we used healthy people rather than unhealthy people, it provided at least partial evidence that applying these principles might indeed be a fruitful approach. An experiment using the virtual iCat with people with obesity provided further support for this approach (Blanson-Henkemans et al., 2009). In this experiment, patients with obesity used either a web interface with the proposed – virtual-iCat mediated – motivational support or without. The support resulted in better adherence to the personal goals for diet, exercises and diary-notations, and a significant improvement of the body mass index.