A study to Measure the Effect of Embodiment on Acceptability and Engagement Rate Amongst Elderly in Pakistan.

Methods: Systematic Literature Review, Qualitative & Quantitative Data Analysis
Project For: Course Project for Topics in Interactive Computing
Collaborators: Anam Tahir, Meher Zaidi
Project Date: Spring 2017


Mental and physical health is a multi-million dollar industry that avows to increase life expectancy but we constantly face the challenges and consequences of growing older. Pakistan is a third world country with lack of proper health infrastructures and adequate facilities for the elderly. The quality of life greatly diminishes as a person ages in Pakistan due to the constant physical, psychological and socio economic hardships they have to face. The UN-backed Global Age-watch Index 2016 has ranked Pakistan among one of the worst countries in the world to grow old in due to the low life and health expectancy after the age of sixty. One of the major concerns in Pakistan is the lack of companionship the elderly have as their loved ones are either caught up in the fast pace of the world or are estranged from them due to familial or social issues. The elder care facilities in Pakistan are a safe haven for those senior citizens who have nowhere else to go or no one to take care of them. While living in such assistive care facilities, the elderly rarely get any visits from their families and are very often subjected to bouts loneliness and depression. The old age homes in Pakistan are not state facilities and the private homes often run on donations due to which there is not much funding given to recreational activities for the residents. Time is also a critical issue for the woefully understaffed facilities and there is shortage of volunteers as well. Improving the quality of life of the elderly is a major area of research and assistive technology is being seen as one of the most promising solutions. Social Robotics is a fast growing field that has not yet been deeply explored in Pakistan, especially in the elderly context. It has been proved through several researches that conversational robots can elicit a positive emotional change in the user. Our goal was to introduce appropriate technology in the form of a tablet PC and a robot equipped with a tablet to help understand the needs of the elderly in old age homes and examine the difference between the acceptability, interaction and engagement rates of the two technologies while determining their preference of physical attributes , embodiment, audiovisual interfaces and tasks that they would accept help for from a robot or tab. The within-subject study was conducted over a period of one week with three sessions. The first session was an introductory talk with all the participants and a pre-test survey to gather the socio economic and technological fluency background of the users. The second session saw 14 participants take part in each of the three tasks controlled through Wizard of Oz on a tablet PC whereas the third session was conducted with 10 users on the robot and tab combination. Each session was concluded with a post-test semi-structured interview and the study was ended with a post sessions focus group to help us gain a better understanding of the elderly’s needs, preferences and acceptability rates.


The basis of a systematic review was to critically analyze the research that has already been done in the area of our interest. This helped in examining what can be improved, what was missing or what has not yet been covered in the previous work when mapping it to the framework of our research topic. The below explained protocol was developed using the PRISMA checklist for protocol guidance. The steps involved in conducting a thorough systematic review involved planning the research questions first and narrowing down the topic to a single focus. The methodology to select the research papers was then identified, after which the inclusion and exclusion criteria was determined so that only relevant papers are selected for the review. Once the data was gathered, it was systematically analyzed and the evaluation is presented in the paper.


These studies attempted to provide an in-depth description of how older people’s thoughts and  attitudes influence the use of technology. Technological barriers often prove a hindrance for the elderly when interacting with new technologies. Elderly show some mistrust towards robots that are likely to be unsafe. To avoid possible adverse effects, they prefer to limit the autonomy of the machine, preferring one that is pre-programmed in a fixed way, not able to improve its competence and not free to move around. Further, they try to overcome their anxiety by attributing specific features, such as small size, slow motion, feminine voice, and the possibility to execute only collaborative tasks, to the robots. In the case of conversational robots elderly would like to tell stories to and discuss the weather, their family, and their future plans with the robot. Also, it was found that the robots can offer many advantages over a computer tablet in healthcare.  In the extensive literature review we conducted, there has not been one study done in Pakistan or in a similar social or cultural environment. There is no data to evaluate how elderly in Pakistan will react to a robot or tablet PC keeping their social, religious and cultural perspectives in mind. We do not know how religion and culture affect the interaction and engagement rates or do they even play an important role in this scenario. The elderly are very rigid in their behaviors and not very technology savvy due to which it is important that the interaction device utilized in such a system has to be selected carefully. 


Replicating the methodology of [19], we extended their research question and tailored the tasks and questions according to the Pakistani context and our above findings from other previous work done in the social robotics field. The within-subject study was conducted over a period of one week with three sessions and was designed to adhere to the cultural, social and religious issues. The sessions with the male and female participants were conducted in different rooms of the old age home and the tasks were fashioned in a way so as not to appear insensitive to the participants’ situation. 


An old age home, Dar-ul-Kafala was selected for the study. They had 19 residents, all above the age of 65, however not all were not willing to participate in our study. We conducted the pre-test survey and first session with 14 residents, 8 female and 6 men whereas the second session had only 5 female and 5 male participants and the post sessions focus group involved 7 male and 7 female users. Majority of the participants had not studied further than secondary school and could not speak, read or write in english. They all were fluent in Urdu which was our primary language in the tasks. 75% had been living at the old age home for more than a year and many had been abandoned or were there because they had no one left to care for them. None of them had ever seen a robot in reality but 13% had heard about robots whereas 4 % had seen them in movies. This distorted their perception of what a robot should look like or behave.


The first session was conducted with 14 participants, 8 female and 6 male. A pre-test survey was given to each participant, however due to low literacy rates the surveys were filled out by the researchers who read out each question and recorded the answers manually. The survey was used as a tool to gather the participants’ socio-economic background, technological literacy and a baseline. They were also asked if they knew what a robot was, had they seen or heard about it and whether they would like to interact with Aisoy after a picture of it was shown to them. 

6.1. TASKS

We designed three different tasks to understand what the elderly preferred and which technology was better suited to them. 


The first task was a communication based task. We used the voice of the cat from the “Talking Tom” app in all our tasks to give it a robotic cat like feel. The tablet PC showed the Tom Cat speaking whereas the display on the Robot+Tab was a pitch waveform whereas the audio was of the Tom Cat. Both the videos and audios for both technologies were operated through Wizard of Oz and asked the user questions and replied with pre-defined appropriate answers. It asked them their names, how they were, their favorite food, hobbies, places etc. to establish some common ground for a conversation.


The second task was about playing games. We used two non strategic existing apps on the tablet PC and the robot+tab so it wasn’t redundant and wouldn’t bore the participants. The first was just a bricks breaking game where a cluster of the same colored bricks had to be popped. The second was a balloon popping game where the balloon had to be popped before it flew away as shown in figure 2.


The last task was a physical exercise task where we showed the participants a video demonstrating how to the simple exercise while sitting. The instructions were given using the Tom Cat voice in the background. It also counted with the participants the number of times they had to perform the exercise. Both the exercises on the robot + tab and the tab were slightly different to reduce redundancy.

This image has an empty alt attribute; its file name is zI7KxBinrnqoKrgEqFlh-R2wi_y1oPhL98O610cgm7Bjvx7iZ2VPduLIcGZnRqi1F9woASAiexyWaXMaEH_NMwDj5529GAa4meLVswafSYEQoay1LK3AblZ_3cwvJsMRu6yo_luZ


After each of the second and third session, a posttest semi-structured interview was conducted with each participant where we asked them how they liked the activities, their likes and dislikes about the tasks with the tab or robot, their preference for physical attributes and what they would like the robot to help them in. 


After the sessions and semi-structured interviews were ended. We conducted two focus group sessions, one with 7 males and one with 7 females. Even though some had not participated in the robot+tab experiment, they had observed the others’ interaction and their opinions were recorded and observed by us. We asked them their preference of the two technologies, which one was more user friendly and which one was more entertaining. We also asked questions about religion based tasks as we could not incorporate those in our activities as we did not know how the elderly would react to them and whether or not it would be a sensitive matter. 

The interviews were transcribed word by word. Deductive analysis approach was then used to group the data using our research question. Data was then explored for relationships between the two studies. The videos of the studies were encoded to figure out the average gazing time, number of time participant uses gestures while interacting and the frequency of cross questioning.


7.1. Baseline

There were total 14 participants, 8 female, 6 male. Baseline results for male and female are displayed in Table 1 and indicate that there were no significant differences at baseline.

7.2. Observations

During interactions, the number of time participants used hand gestures while communicating was significantly greater in case of embodiment for both male and female. Participants pointed, waved hand, nodded and touched the robot. Whereas while performing the communication task on tablet the only gesture observed was nodding when the participant agreed on something.

Similarly, the average gazing time was greater in case of embodiment. As the study was performed using wizard of oz there was a delay in the response time. Due to this delay participants got distracted when communicating with the tablet and started looking at the researcher. However, in case of robot despite the delay in response they did not lose eye contact with the robot and kept looking at it anxiously waiting for a reply.

In case of game playing task, participants completed the particular level of the game when robot was added to the equation. Without the robot they played for an average of 45 seconds and returned the tablet. They were more negligent of the audio praise while playing on the tablet. However with the robot they looked up at the robot for a praise whenever they performed good.

There was no significant difference observed in interaction for the exercise task. Participants completely followed the instructions in both the scenarios.

7.3. Technology Ratings

7.3.1 Enjoyment

Participants rated robot higher in terms of enjoyment because of the movement of its eyes, eyelids, mouth and neck when it responded. P4 said: “Yh bolta ha na is ki batain piyari lagti hain.” They cross questioned the robot more as compared to the tablet and smiled at the robot throughout. The tone of responses of participants was also very different for the two cases. In case of tablet their responses were straightforward whereas in the case with embodiment they used satire and adopted a more jolly tone. For example during the communication task, for both cases, when the robot asked: “apko muj sy batain kr k maza aya?” In case of tablet P11 replied “haan aya.” Whereas in case of robot he replied “haan buht maza aya mujy toh sari raat neend nahin aay gi.” This variation in tone also proved that they enjoyed interacting with the robot more as compared to the tablet.

7.3.2 Likability

Tablet was rated higher in terms of likeability and ease of use as compared to the robot. P4 shared: “phone sahi ha hath mein ajata ha jaisy marzi lait k istemal krlo. Isski (robot)  toh tarain hi buht hain, itni machninery ha sath.” As some participants had hearing issues they could easily bring the tablet closer to their ear or hold it close to their eyes if they couldn’t see it properly. This comfort of interacting with the tablet at their ease made them rate tablet higher in terms of likeability. Moreover, some prefered tablet over robot because they thought of robot as a toy. P6 said: “phone bahtar ha yh (robot) toh bachon k lia ha.” 

7.3.3 Trust

Participants rated both the technologies equally in terms of trust because they believed that the technology is not intelligent enough, it speaks or does what the researcher wants it to say or do. P8 said “Yh khud toh nahin kuch kahta  na jo ap kahalwatay hain wohi kahta ha. Quran ki ayat bhi khud toh nahin bolay ga na ap jo feed karain gy wohi bolay ga toh galat kaisy ho sakta ha.” 


Embodiment did affect the engagement level as the average gazing time was greater (86.76 seconds) in case II (with embodiment) as compared to case I (30.43 seconds). Participants found the robot more engaging because of its movements. About the robot P13 shared: Iss ny mujy shabash zayada dee ha yh zayada acha ha.” The frequency of praise where as was constant. While playing the game on the tablet participants lost focus and  didn’t pay attention to the background voice because of which they failed to realise that they were being praised throughout as they played the game on the tablet as well. However, in case of robot the mouth of the robot moved whenever it praised the participant while playing the game and this is what engaged the participants and made the praises noticeable. The engagement was so original that participants often cross questioned the robot while playing using phrases like “Ab kyun nahin bol raha? Ab mein acha nahin khail raha?”, “Aur bolo. Aur shabash do”, “Abhi bhi toh acha khaila ha ab bhi bolo na shabash”, “yh wali shot py shabash nahin zabardast bolna tha” and “shabash hi bolay jaa rahay ho mujy bhi toh zabardast kaho na.” In excitement, they even repeated what robot said trying to mimic his accent.  Another important indicator of the level of engagement was the use of gestures while communicating. Participants pointed, waved hand, nodded and touched the robot. In case of tablet however they were just responding to the audio. Many participants didn’t even look at the tablet. They were looking around and answering the questions and some even held the tablet next to their ear. This effect of greater engagement with the robot however was greatly due to the novelty factor. It was observed that engagement and excitement level decreased in subsequent levels of the games and in subsequent tasks of the study. The number of times gestures were used and the number of times they smiled at the robot also decreased as the tasks progressed. The frequency of cross questions decreased aswell as the participants realised that the robot gives generic pre programmed answers. Regarding this point in the post interview P7 said: Mein ny daikha k yh jawab nahin daita. Udhar sy daita ha. Jab ap button dabati hain tb daita ha. Jo khud ba khud jawab daita toh baat karnay ka dil karta.”  This point of novelty factor also ties up with the fact that participants were not willing to interact with the technology in the future. After the interaction all the participants labeled the robot and the tasks on the tablet as childish. P2 said: ““>Yh toh bachon k lia ha P4 stated: “Bachon ka khail bachon ko acha lagay baron ka khail baron ko acha lagay aur yh bachon ka ha.” Similarly P6 said: “Haan acha ha lakin yh bachon ko do gy na unko buht acha lagay ga. Mein toh pareshan haal mein bezar aurat hun chup rahna pasand ha mery yh kis kaam ka.” In the post interview when questioned about the future usability P7 stated: “Nahin mein toh nahin istemal karun gi. Yh toh chottay bacahy jo nadan hottay hain unk lia buht acha ha. Baron ki aur batain hotte hain aur shook hottay hain.” Interfered from the post interviews there are three possible explanations of this result: 1) the tasks were very simple and easy, 2) the voice of the robot and the tablet was childish instead of robotic and 3) the robot looked like a toy.   Considering how all the participants rated the tasks and the appearance of the robot childish they were questioned about the changes they would like in terms of appearance of the robot or in terms of tasks that the robot can perform. A few participants suggested for the robot to have arms or for it to be able to move. P14 for instance suggested “Iss k hath nahin hain wo chez thora tang karti ha. Iss k hath hottay thora hill jull sakta jaisy filmon mein hotta ha toh bahtar hotta.” While others believed that no matter what changes you make it will remain a toy.  P7 in this regard said: “kuch bhi kr lain bayshak bilkul insan wali shakal aur awaz hi kyun na ho rahay ga toh yh khilona hi na.” Majority however were indifferent about the appearance. They were only concerned with the audio. “Awaz saaf honi chahiya.” was a common remark.  Participants wanted a more seamless integration of technology in their lives. P4 said: “Meray lia koi Mp3 wali chez ho. Koi aisi cheez jis k lia mujy hila na paray jo mujay pakar k na baithni paray bss mein jaisy abhi kaam kr rahi hun mein usko challa dun aur sunti rahun.” All the participants proposed to have a audio based interface which is easy to operate and can be kept anywhere. They were opposed to the idea of holding the interface for long or to sit up in order to interact with it. There health issues restricted them from having long interactions.  Generally female liked performing the communication task whereas men preferred the exercise task. Both of them found the game tiring and childish. Apart from theses participants suggested the robot to perform health related tasks. P10 for instance suggested “yh BP check kar day, sugar check kr day toh acha ho.” P13 suggested: “yh humain latest news bata dy sports ki. TV py sara din intezar karna parta ha. Khass time hotta ha jb sports ki khabrain lagti hain wo bhi kabhi rah jati hain toh agr yh wo suna dy roz toh kitna acha ho humain intezar na karna paray.”  Another factor explored in the post interview session for the sake of future work is the acceptability of involvement of robotics in religion. Participants were questioned if it would be socially, culturally and religiously acceptable for the robot to recite quranic verses or narrate Islamic teachings. There were two factors which were to be explored: (1) if the acceptability has anything to do with the appearance of the robot and (2) if they would trust the robot in regard to what it says. It turned out that the response to both the scenarios was positive. Participants were again indifferent about the appearance in this context as well. P5 said: “Billi ha toh kia hua. Hum tottay ko bhi toh ayat aur kalmay yaad karatay hain ha. Wo bhi toh janwar hain. Shakal ka kia ha. Shakal sy koi fark nahin parta.” In regard to the trust factor P8 said: “Haan toh kyun nahin yakeen karain gy issny apny dimag sy thori na bolna ha. usko toh jo ap feed karain gy wohi bolay gi yh.” This provided a baseline for future study on this aspect.


The study suffers from some limitations which limits the generalizability of our findings. The sample size was small because of very few residents in each oldage home and conducting the study across multiple old age homes was not economical.  Another major limitation was that because of the age difference participants were really empathetic towards the researchers. Considering us as their grandchildren they agreed to perform every task and gave positive remarks for most of the interview questions regarding the study.  This infused a lot of biases in the results. P4 for instance stated: “app loogon ka kaam ha ap itni mehnat karty ho toh acha hi hoga na, bura toh nahin kahain gy hum” Therefore presence of researcher was in itself a limitation. Misconception about the purpose of the study was another factor that lead to biased results. Participants thought that if they perform well and show a positive attitude towards the technology they will actually be handed the technology. P2 said: “Mujy day dana mujy acha laga meray bachay hain”, similarly P10 said: “yh billi mujy dyna mein ghar ly jaun gi chupkay sy inko pata nahin chalay ga. Inko do gi toh yeh larain gy.” Thinking on similar lines P12 said: “Mujy dy do mujy smjh agai mein roz istemal karun gi issay Inshallah.”   


Overall, this study found that though elderly in Pakistan find new technologies such as smartphones and robots engaging but this is not something they would want to use each day for the rest of their life. It’s only the novelty factor that motivated them to give it a try. Upon interaction they found it too complex and a hassle for daily use. Even though the participants found the robot to be more interactive, they liked the simpleness of the tablet PC more. They also proposed audio based solutions as the physical attributes and embodiment did not have a major effect on their preference. Some suggested the robot should have had limbs and be able to move around but it was again just a minor detail as they emphasised more on audio than visual. 

Longitudinal studies could be conducted in future for generalisability of results. This study can be replicated eliminating the presence of researcher from the equation to see what effect it has on elderly’s willingness to engage with the technology. 

An important result of this study was that elderly are more inclined towards the idea of audio interaction rather than visuals or 3D interactions. In line with this outcome a future study can be designed to gauge in what context audio interactions are favoured over visuals and vice versa for this particular user group. 

Moreover, this study fails to substantiate the factor of trust in both technologies because only one scenario was presented to the participants. The study can thus be extended to evaluate the element of trust contextually. If the context is changed to playing a game against the tablet and getting a biased result every single time or giving advice regarding health, will the trust factor still be the same.  

Qualitative data suggested that participants are quite open to the idea of teaching religion through technology. This hypothesis can be evaluated by a research study to see how demographic and cultural differences affect this stance.