IRo was developed by the research department of a Dutch profit-based company that operates on an international market. The company focuses mostly on developing technology in the area of health and wellbeing. Within this company, there is an increasing focus on (health-related) technologies for elder users. This increasing focus on elder users is a world wide trend, as exemplified by MIT’s Age Lab, the Georgia Institute of Technology’s Human Factors and Aging Laboratory, the European Ambient Assisted Living project and the creation and marketing of robots intended for, or tested with, elders, like Aibo, Ifbot, Wakamura and Paro.3 These robots range from robotic animals (Paro) or pets (Aibo) to robotic companions and assistants (Ifbot, Wakamura). Although most robots are not designed solely with elder users in mind, companies do imagine these robots as health technologies for elders and hope they will help prevent cognitive decline, combat loneliness, or provide reminders to take medication as people age. Robot use is hoped to help individuals age at home, and thus alleviate the healthcare costs associated with the move of elders into nursing homes or hospitals.
IRo as a tool to gain knowledge about elder users
The researchers initially developed iRo to understand the ways in which users interact with socially intelligent robots. Though iRo has a polished appearance, Johan stressed that iRo was not a product and was solely intended ‘as a research prototype’, at least for now. However, based on the iRo research project, the researchers envisaged the future development of robots which could be sold to aged consumers as both preventative technologies that help maintain cognitive health and as assistive technologies to be used when cognitive skills or mobility decline.
Depending on iRo’s programming, it can perform tasks itself but it can also serve as an interface for digital technologies. For example, in other tests iRo was used to set alarm clocks and operate audio equipment. It is easy to communicate with iRo. No commands need to be learned. Instead, a user can speak plain Dutch to iRo and iRo responds likewise. Like Kismet (an anthropomorphic robot that mimics human facial expressions)4, iRo has the ability to blink, nod, frown, smile and grimace. The development of iRo is thus in line with trends in social robotics towards ordinary conversation and displaying and reacting to emotions (Weber 2008).
Though iRo is built out of plastics, electronics, and servomotors, iRo’s most fluid component is its software. The researchers can change iRo’s software, which turns it into a robot with a different application. For the tests with elder participants, the researchers programmed several cognitively challenging games and puzzles into iRo and wrote conversational software which allowed iRo to make game-related comments like: ‘I’ll have to think of a clever move now’ or ‘you have to jump’. The researchers also added a digital game board, which was connected to iRo and placed between iRo and the participant. Together, this turned iRo into a game companion for elder users. The board games and puzzles were designed to be both entertaining and cognitively stimulating in order to help maintain elders’ cognitive health.
The researchers went to considerable lengths to increase their understanding of elder users. This is in line with the research department’s guiding vision (which is stated in various publications and vision statements) in which technology should ‘know’ and adapt to the user. Adhering to this vision, the researchers’ focus was directed towards understanding the elder user, which was considered integral to technical research. This is reflected in their user-centred design approach,5 the use of scientific literature on elder users, the interdisciplinary composition of the research team, and the design of their laboratory. By making the laboratory look like a middle-class elder persons’ apartment, it was hoped that participants would feel relaxed and behave naturally.
The researchers’ attention to the importance of users can further be seen in their decision to conduct two week-long tests in people’s homes. Concerned about methodological shortcomings of laboratory testing, Johan mentioned, ‘if people come for two hours – like with Marjolein’s tests now – they all like it, but what happens after longer periods of time?’ Therefore, field tests – placing iRo in elders’ homes – were deemed necessary. With every research project, the researchers gathered more information about user behaviour in relation to iRo, and they thus progressively built up an image of elder users. Interestingly, despite this concern for understanding users, the researchers would struggle to incorporate the participants’ alternative views on iRo which emerged during the tests.
Initial ideas of elder users and their health
The researchers initially developed their views about (health) robots and elders by carrying out an exploratory literature review and organising a workshop in which ideas were generated for specific tasks or activities which iRo could perform, such as being a game companion. In both the literature review as well as the workshop, researchers imagined old people as varied; elders were imagined as having differences in preferences, needs, lifestyles, mental and physical abilities and social environments. They were thus not simply imagined as frail, lonely, or forgetful.
Based on the workshop findings, all ideas for applications were evaluated with the help of experts in the area of aged care6 together with other experts: staff members of a nursing home. These two groups considered the application that would make iRo function as an opponent or a companion in playing games the most promising, so it was developed further. These games would be entertaining, but Johan explained that they were ‘not just games for fun, which is what you would focus on with children, but [the games had] the goal of stimulating their cognitive abilities’. The game application was thus designed to support the cognitive health of iRo’s elder users. The long-term view for developing iRo was also health related, in particular to a possible transition to frailty or illness in later life. Marjolein explained: ‘The idea is to get an iRo into their homes as early as possible, so that they get used to it, but also so that it can be adapted to their skills. A person of 65 perhaps doesn’t need anybody to keep his agenda, but five years later it may be necessary for him to check what he should do every day, or for him to be reminded that he should take his medication. Research has shown that it is better if [iRo] has a different function at the beginning and that he can adapt to the skills of people. In that case, people will be more likely to accept it than if you say at 70: here’s your iRo, it will help you to remember now’.
Thus, even though the board games and puzzles would be fun and cognitively stimulating, in the future, iRo could play additional assistive roles in the maintenance of the health of elders. IRo could, for instance, remind them to measure their blood pressure or take their medication. The combination of assistive and cognitively stimulating applications illustrates the researchers’ idea of health — one that includes physical, mental, and cognitive dimensions and imagines these as interrelated and changing. In addition to these ideas about health, the researchers also reflected on acceptance and rejection of (health) technologies and devised strategies to entice elders to accept a health robot in their lives.
The criteria for selecting participants were strict. Marjolein explained: ‘There are a number of requirements. First, 65 years and older, to be sure they are retired and home all day, so they have time to play with it [iRo]. Secondly, living in a single-person household, to make sure that no one else can play with it’.7 (The researchers focused on single-person households as multiple users would render their data unusable.) The age of participants was evenly spread between 62 and 79 years and the number of male and female participants was equal. The researchers screened for physical disability, colour blindness, and early signs of dementia. The focus of this test was on cognitively healthy elder users, as the researchers wanted to develop and test technologies for maintaining cognitive health. In addition, the researchers judged people with declining cognitive health less capable of completing the tests.
The researchers were mainly interested in the long-term interactions between ageing users and iRo, as they had previously only conducted short-term research projects in the laboratory. In addition, they wanted to know whether, as Marjolein put it, ‘relationship building strategies, that have been shown in literature from human-human, human-robot and human-animal interaction to foster relationships’ could be used to create and deepen relationships between iRo and elders. In order to answer their questions they chose to do two rounds of tests; one round in their laboratory and another more extensive round of field testing a couple of months later in which the participants would have iRo in their home for several weeks. The objectives of the first round of tests were to check whether iRo functioned well technically, whether the participants perceived iRo as appealing and easy to use, and whether they enjoyed playing games with iRo. However, as the final report mentioned, the most important objective of the laboratory test was to ‘recruit participants for the field test’.
On the whole, the elder test users turned up early at the laboratory. Marjolein would welcome them and chat with them over a cup of coffee. Following completion of a consent form, a questionnaire, and a test for colour-blindness, Marjolein retreated to the ‘bedroom’ of the laboratory and left the participant to interact with iRo. She would only intervene if the interaction ground to a halt. This happened, for instance, when a test user inadvertently made a wrong move during a game and subsequently did not understand iRo’s suggestions for correcting this. When the games were finished, Marjolein returned to interview the participant. While some participants experienced some difficulties interacting with iRo, they were generally enthusiastic about iRo. They said that they liked the games, and thought iRo looked fun. One woman said, she liked ‘the way he flutters his eyelids’.
All but one of the participants who were asked decided to take part in the field test. This test comprised two periods of two weeks in which the participants would have two variations of iRo – a version with and a version without relation-building software. The test users could play as much as they liked. On average they played for about 40 minutes a day, mostly when there was a gap in their schedules. Marjolein reported that, irrespective of the version, test users enjoyed having iRo in their homes. ‘They admitted that they talked to it’, noted Marjolein, something she had not expected. Some participants got emotionally attached to iRo. ‘Most of them said that they were going to miss it. It was a question in the interview, but a few said that without me asking the question’, Marjolein explained. After two sessions of two weeks, the researchers picked up the iRos from the participants’ homes. The data logged by iRo’s programs were analysed along with other types of data, including test user diaries in which participants were free to write down relevant events and thoughts. Subsequently, the researchers’ report concluded that interactions with iRo could be sustained for longer periods if the applications were varied and compelling. Furthermore, strategies to build relationships between users and the robot were relatively effective.
Researchers imagine elders as needing and wanting iRo
The researchers’ views linked up with, but also went beyond, the discourse mentioned in the introduction. Referring to demographic data provided by the World Health Organisation and the United Nations, the researchers acknowledged that the number of old people is growing, especially in developed countries, that the cost of (health) care will increase, and that there will be a dearth of qualified staff to care for ageing people. A research department publication concluded that elders want to live independently as long as possible, but also that their quality of life is reduced when they live alone. Robots were presented as a technology that can play an assistive role in the everyday lives of elders, providing practical assistance that allowed users to stay in their homes and that could potentially function as an emotional and intellectual companion. In a preparatory document the researchers recognised that loneliness was a taboo, and the final report concluded that elders felt positive about a robotic companion if they could develop a bond with it; furthermore, it is possible for researchers to devise strategies to develop these bonds. The researchers agreed with Reeves and Nass (1996) that people tend to treat computers and robots as social actors and react emotionally to them, despite the fact that they may rationalise their behaviour afterwards. A research project leading up to the tests concluded that acceptance of technology by elders was also improved if they understood that the technology could help them live independently. The researchers had learned in another research project that the combination of entertainment and a cognitively stimulating activity benefits elders; they also feel positive about this, as they worry about their mental health, creating a need for the application which was tested with iRo.
Thus, ageing users were portrayed in a specific way. Though the researchers were employed by the company, they had a scientific approach to design; using scientific methods, fostering links with universities and publishing their results in scientific journals. The researchers methodically built an image of elders and their relation to technologies like iRo. This image consisted of two main parts. First, the researchers found that old people needed (health) robots, if not now, then certainly in the future. The second part went beyond the discourse mentioned in the introduction, stating that elders would accept (health) robots, see their benefits and even bond with them. In effect, the researchers believed that elder users not only needed such a robot but that eventually they would want it too. Despite complexities like diversity of elder users and the taboo of loneliness, this combination of societal and personal need for robotic technologies and the elder users’ willingness to accept such technologies, allowed the researchers to make a strong case for developing (health) robotic technology for elders.
This ‘need and want’ representation was largely in place before the tests were conducted and the tests mainly reinforced it. This representation thus provides one answer to the question of how the user is represented in these tests. Leaving it there, however, as other studies have, would render invisible another group actively making their own user representations: the elder participants themselves.
Test users’ representations: how not to be old and lonely
Generally, it is assumed that only researchers, designers, engineers and other professionals involved in creating technology form user representations. Test users apparently do not play an active role in this process, only passively providing input for the work of researchers, designers and engineers. However, users are seldom passive and test users have been shown to be no exception (Epstein 1995, Feenberg 1995, Oudshoorn 2003). The elder participants in the user studies of iRo were also active. They reacted to images of old people, and, as will be shown, actively created their own images of the prospective user of the technology.
During conversations with Marjolein, the test users would state that they liked iRo, liked playing games with it and thought it worked very well. However, they also repeatedly, and often without provocation, mentioned that iRo was not a robot for them. For instance, one woman said: ‘I still lead too much of an active life; I’ve always been amongst people. I don’t need an iRo, not yet anyway’. One of the men related it to his hobby: ‘At home I am still very active, I really don’t have spare time. I collect old army radios, you see. That takes time and I like doing it. It’s creative work, so that’s good as well. It keeps me busy.’ Later he added: ‘Maybe if I was all washed up [incapacitated], but I don’t need to be kept busy. But I think there are plenty of old people sitting behind their geraniums who need to be kept busy, for them it would be good’. Another man said: ‘If you were, say, old and growing demented, than I could imagine this being a good thing, but for me?’ and ‘You’d have to be a lonely old person, chained to your home with few contacts. I still go to my checkers club.’ Another woman reported: ‘I play games with people, with my friends (...). Therefore I don’t have an iRo. I am not lonely, so I’m not planning on getting one of these. If people are lonely and have difficulties getting in touch with other people, they’d go out and get one, but I don’t need it.’
While Marjolein mentioned that ‘there was one woman who said: this is something for me, but she was very housebound’, in the end, 10 out of 12 participants in the laboratory test reported that iRo was not for them. Most participants felt that the intended user was housebound, old, lonely, feeble and in need of care and attention, and they did not want to be equated with this person.
In part, this equation resulted from the participants knowing that they were selected to test this prototype on the basis of their age and the fact that they lived alone. They thus assumed that iRo was intended for older people with few social contacts. But Johan and Marjolein attributed a substantial part of this interpretation to the way in which robots for old people were portrayed in newspapers and on television: ‘This positioning of iRo for only old and lonely people could have been due to the fact that iRo was in the media a bit before the test [as] a robot exclusively for elders’, Marjolein said. She added: ‘they had just seen on television that robots, similar to iRo, are placed in retirement homes, for some entertainment’. There was also a newspaper article about Aibo in retirement homes; thus, according to Marjolein, ‘the image of a robot for lonely older people, they just got that from the newspaper’. This portrayal of robots meant that the researchers were faced with a problem. Marjolein explained, ‘You’ve got to change that image. The fact that they saw on television that iRo is just for old people, is just an unfortunate coincidence. Otherwise, they perhaps would not have kept so stubbornly to this notion, that this is only for really old people. People who are very lonely, can’t do anything and just sit at home.’
Despite the researchers’ attempt to explain away the association of iRo with dependence and loneliness as a short term effect of media stories, test users felt that iRo was a signifier of old age, loneliness and needing care. The elder test users did not see themselves as such. Instead, they made it clear to the researchers that they were (still) active and independent. Furthermore, they presented themselves as possessing sound physical and cognitive health and as (still) having a wide social network. Minichiello et al. (2000) have shown that ageing people can create an image of ‘old’ people and actively dissociate themselves from that group. The test users did exactly that; they made the prospective user of iRo into somebody else, somebody older, lonelier and in need of care, and they actively dissociated themselves from this user.
Test users imagine themselves as helpful and not old
Interestingly, when asked at the end of the laboratory test, all but one of the test users agreed to take part in the field test. This seems counter-intuitive: why would you want a robot in your house for four weeks if it is a signifier of old age, poor health and loneliness to you, and you actively want to position yourself as not old, frail and lonesome?
A first indicator is that nearly all the elder test users, as Marjolein explained, ‘had participated in prior research or were persuaded [to participate] by their children because they work here’. Thus, this particular set of test users was inclined to participate because they knew the company and perhaps because they did not want to disappoint their children. In addition to this, the elder test users mentioned that they participated in the test to help other elders. One woman said ‘I think it’s fun and interesting, not because I want one but to help somebody else. I like that (...) I think I am helping people with this.’ One of the men told the researchers that he thought he was ‘still too good’ for an iRo, but he made it clear that he knew ‘enough people who would benefit from one’. The elder test users positioned iRo as ‘a good thing’ which they wanted to associate themselves with as a test user (though not as the prospective user). The elder participants thus displayed altruistic motives for participating in the tests, which has also been observed for test users in drug testing for AIDS patients (Epstein 1996) and for male contraceptives (Oudshoorn 2003). Thus, the participants created another image – one of themselves as helpful elder test users.
Thus, they embraced the idea of ‘successful aging’ (Calasanti and King 2005), as they tried to position themselves as active, healthy, altruistic, helpful old people, seemingly (still) untouched by the negative consequences of ageing. This image of themselves as successfully ageing elder test users, allowed them to further widen the gap between themselves and the perceived prospective user of iRo, while simultaneously allowing them to use and enjoy iRo.
Researchers’ responses to elder test users’ ideas
Test users thus also created user representations, either in accordance with who they thought the prospective user of the technology was, or in accordance with what they thought a good test user should do. The researchers had to deal with these alternative images of the (test) user. On the whole, the researchers tried to remain focused on their initial research questions, in spite of the participants’ additional responses. There was, for instance, no mention in the final report of how test users’ distanced themselves from the perceived prospective user of iRo. In interviews, the defiant responses were acknowledged as relevant, but were simultaneously seen as a relatively small incident, caused by media attention on robots for elders and as something that could be addressed by an early introduction of iRo into the lives of people, meaning when people were in their late fifties or early sixties, and iRo would then still only be, in Johan's words, ‘just for fun’. In the end, the representations made by test users added little to the researchers’ understanding of potential meanings attached to the technology.
Instead of taking the participants’ user representations into account, the researchers saw these responses as a side issue that presented them with some difficulties and additional work. Marjolein found that getting ‘helpful’ elder test users to provide the desired information could prove difficult. She mentioned that the participants gave ‘please-me answers’ or that they reasoned not from their own perspective, but ‘tried to reason from a consumer perspective, so: “Oh, if iRo does this, then this offends the consumer, so you should maybe change this”’, assuming that this perspective was most valuable to the researchers. Also, the researchers made changes to iRo between the two phases of the field test, to give the impression to the elder participants that their input was taken into account, because the researchers, as Marjolein said, ‘did not want to disappoint them’. The researchers had to perform additional forms of work to cope with elder participants behaving according to what the elders thought were helpful test users.
Thus, the researchers held another user representation; an image of what old people are like as test users and how to work with them. According to Marjolein and Johan, working with elder test users is different from working with other age categories. Marjolein explained: ‘First of all, I think you can find more test users in other target groups. It is simpler, they’re more open to something like this.’ Johan thought they might be ‘insecure’; Marjolein added that the word research might make them ‘a little scared’ and she also felt that ‘they might be afraid that we’re going to sell them something’. The researchers reserved extra time for the elders who were willing to partake in the tests. Johan explained: ‘They all show up early, sometimes even half an hour. With people aged, say, up to 50, that just doesn’t happen’. He added: ‘you’ve got to put more time in, making contact, making appointments, phoning them again. But also during the experiment, (...) the people like to talk, it’s an outing for them.’ Nevertheless, the researchers made sure the tests would not be too long, as they did not want to overburden the participants. Explanations for using iRo were also modified. Marjolein explained: ‘you’ve got to do it slowly, give an example, demonstrate it first, and then let them do it’. Johan added: ‘These are just people who interact less easily with technology than younger generations. You’ve got to take that into account’ and ‘people gradually forget the instructions’. Marjolein continued: ‘But you can’t underestimate them. They’ll also see how far they can go.’ She recalled one elder man who ignored the instructions to find out ‘what iRo can and can’t do’. Coping with these differences meant work for the researchers. The researchers had to find elder participants, schedule more time, persuade them to stick with what they were asked to do and interpret and make selections in the information put forward by the participants to stay focused on their research questions. The researchers thus also held a ‘craft representation’, a set of images of what old people are like as test users and the specific skills and changes in testing practices required to work with them. They thus created a setting in which they could negate the test users’ ideas about the imagined user and their rejection of him/her.