Social Robot Companionship: Understanding Loneliness to Improve Interventions
Social robots, or robotic agents designed to interact with their users through social behavior, are slowly making inroads as consumer products. One user group the social robot industry is currently targeting are lonely or socially isolated individuals. The overarching presumption is that either through the nature of the artificial social interaction itself or through the functionality of the robot, feelings of loneliness can be reduced, leading to improvements in quality of life. There is certainly some evidence to support social robots’ potential to reduce loneliness, but current social robot development often fails to account for the complexity of loneliness. While we tend to focus on engineering-related topics such as natural language processing (NLP), embodiment, computer vision, facial recognition, and more when designing robotic interventions for loneliness, understanding the individual experience of loneliness is vital if the industry is to succeed on this front.
What Exactly is Loneliness Anyway?
While this seems like a simple question, given that most of us can instinctively identify the feeling of loneliness when we experience it, confusion often arises between the terms social isolation and loneliness. Social isolation refers to the objective state of having little to no social engagement. When we design technology to help older users who live alone and rarely see other family members or friends, we are designing social isolation interventions. In these designs, we might be interested in implementing features which help connect users with other people or assist in safety-related monitoring tasks. Additionally, a socially isolated individual may not be feeling any psychological distress.
Loneliness, on the other hand, is the subjective feeling that our social relationships are not what we want them to be. Loneliness is a lot more complex than social isolation. A person can be socially isolated and feel lonely, just as they can feel lonely without being socially isolated at all. Similarly, loneliness can occur for a multitude of reasons, with most of them having to do with the perceived quality of social relationships, rather than any quantity of social relationships. Loneliness is often about perceiving that “something” is missing from a social relationship and if we are to design proper robotic interventions for loneliness, we need to understand what this “something” could be among a social robot’s users.
The State… and Trait…of Loneliness
The design of robots and virtual agents for loneliness, as well as the research that leads to these designs, often justifies the motivation for the work on two commonly observed and accepted psychological effects. First, lonely individuals are more likely to anthropomorphize non-human objects, such as robots (Epley et al., 2008). In other words, people who feel lonely are more likely to see a robot, virtual agent, or even a pet as more human-like than a person who is not feeling lonely. Second, lonely individuals are more likely to evaluate a social robot positively than someone who is not lonely (Lee et al., 2006). Both these effects are thought to be the result of increased sociality motivation, or a drive for social engagement, among the lonely (Epley et al., 2007).
Although these prosocial effects have been widely replicated, there is some doubt as to the nature of these relationships to the temporal aspect of the experienced loneliness. To understand this, we must first understand the difference between state and trait loneliness. State loneliness refers to the temporary moment in which we feel lonely (Overland, 1991). For most people, state loneliness occurs in moments in which we feel loss, miss others, or experience social exclusion. This could occur after the loss of a close other, end of a romantic relationship, or if we feel left out of an activity among friends. State loneliness can be a normal part of life experiences and under healthy circumstances, goes away within a reasonable timeframe.
Trait loneliness, on the other hand, refers to the predisposition an individual has towards feeling lonely (Overland, 1991). High trait loneliness indicates that an individual experiences loneliness at regular intervals throughout their lives or even constantly. This kind of chronic loneliness is more concerning as its long-term effects include decreased life expectancy, faster progression of dementia, and other health and quality of life related problems (Holt-Lunstad et al., 2015; Sutin et al., 2020). Due to its long-term nature and continuous negative effects on mental and physical health, it is individuals with trait loneliness who are better situated to benefit from social robots or virtual agents designed as loneliness interventions.
Although previous research showed that lonely individuals anthropomorphize more and ascribe higher potential for social companionship to robots and virtual agents, these results rarely disambiguate between state and trait loneliness. That is, it is assumed that individuals with high state loneliness (but low trait loneliness) feel exactly the same as those who have high trait loneliness. This is a risky assumption that could negatively influence the design and development process of social robotic interventions. Individuals high in trait loneliness generally develop perceptual and cognitive biases towards social cues which lead to negative appraisal of others (Käll et al., 2020). This negative appraisal then creates counterproductive behavior and negative emotional responses, leading to further impaired cognitive processes and interpersonal difficulties. A by product of this process is that an individual high in trait loneliness perceives less value from their social contacts, further driving continued long-term loneliness. Some recent studies exploring state vs. trait loneliness have shown that trait loneliness may reduce anthropomorphic tendencies and acceptance of a social robot (Li et al., 2020). Additionally, individuals with higher trait loneliness could potentially ascribe lower social presence to a robot (Montalvo et al., 2019). Social presence is an important metric in social human-robot interaction as it measures the degree to which a user perceives the robot is a real social agent that is present in an interaction.
These and other recent studies serve as notice to developers that state and trait loneliness may require different design approaches. For example, trait loneliness may require greater attention given to the portrayal of robotic social cues, with an emphasis on reducing any potential perceptual and cognitive biases that an individual may experience. Rather than focus on simply increasing social interaction, an additional approach in these designs should be to implement features which interrupt the self-perpetuating cycle of trait loneliness.
Loneliness Typology
The differences between state and trait loneliness are just one aspect of loneliness that social robot designers need to understand if they are to properly design robotic interventions. A crucial aspect of loneliness which must also be understood is that loneliness experiences cannot be defined under a one-size-fits-all approach. Loneliness comes in many different flavors and while there is still some lack of consensus as to just how many different types there are or even a clear understanding of how each type should be defined, below are a few common types of loneliness social robot developers should be aware of (please note that this list is not exhaustive).
Social Loneliness
The perception that one does not have friends, “good” friends, is not part of a group of friends, or does not have anyone with whom to share common interests and activities (Weiss, 1973). Most robot interventions addressing loneliness are aimed at this type of loneliness. Example approaches for this type of loneliness include direct companionship provided by the robot, providing a connection between the user and other people (i.e., telepresence), or serving as a point of entertainment between multiple people.
Situational Loneliness
A subcategory of social loneliness, this type of loneliness refers to the perception of social isolation which comes from changes in a person’s social or actual environment (Beck & Young, 1978). For example, if a senior moves to another state after retirement, a concept known as retirement migration, they may find it difficult to meet new people at the new location. This could lead to a transient state of loneliness until a strong social network is reestablished. Social robotic interventions aimed at this type of loneliness should be similar to those of social loneliness, with a stronger emphasis on providing telecommunication capabilities to help connect the user with distant or new friends.
Emotional Loneliness
The subjective feeling that one does not have a close or intimate attachment to another person (Weiss, 1973). This is not necessarily romantic in nature (see below) as it pertains more to the sense of security and support we get from close relationships. People who experience emotional loneliness often state that there is nobody in their life to care for them or whom they can care for. This type of loneliness is harder to address with robots, but developers can take some cues from human-animal interactions. Pets are able to reduce loneliness given that they often exhibit emotional attachment towards their owners, as well as dependency. A robot can be designed to mimic such attachment, but it is unlikely that such a device can address the missing intimate emotional needs provided by close interpersonal bonds. A potential approach here is to use entertainment and other approaches as ways to mitigate the negative effects from emotional loneliness.
Romantic Loneliness
A subcategory of emotional loneliness, romantic loneliness is related to the lack of an intimate romantic relationship (DiTommaso & Spinner, 1997). Individuals may feel romantic loneliness after a break-up or if they have lacked a romantic bond with another person for a long period of time. Romantic loneliness is difficult to address via robotics. While it is difficult to address romantic intimacy needs with robotics, social robotics can be used to mitigate the negative emotional effects of feeling romantic loneliness through entertainment or therapy.
Family Loneliness
Another subcategory of emotional loneliness, family loneliness refers to the subjective experience of a lack of close, intimate ties to family members (DiTommaso & Spinner, 1997). Undergraduate students may feel this type of loneliness when moving away from home. Seniors may feel this kind of loneliness if they have little to no contact with other family members. Social robots can be used to bridge this gap, by serving as a communication platform to help connect family members.
Existential Loneliness
This category of loneliness, also referred to as existential isolation, is somewhat different in experience from most of the others. Individuals who experience existential loneliness feel a disconnect with society in general, experience a lack of meaning in life, and often report feeling empty inside (Yalom, 1980). This type of loneliness is also reported by individuals who are dying, such as patients in hospice care. Existential loneliness is difficult to address and should be treated via professional therapists given its link to suicidal ideation (Helm et al., 2020). However, through positive affect displays and therapeutic techniques, social robots could aid with existential loneliness by improving a user’s mood and perspectives about their life.
7 Tips for Designing Robots for Loneliness
Of course, the above suggestions are just some examples as to how robots can be used to address different types of loneliness. Many other strategies not mentioned here can be employed to address loneliness. However, taking the typology of loneliness into account when developing social robots is an important step at discovering and addressing the “something” that a lonely person perceives is missing from their lives. This user-centered design approach is critical if social robots are ever to consistently address loneliness in home settings.
Here are some other tips for social robot development teams to consider when designing loneliness interventions:
- Design for parasocial presence: In robotics, parasocial presence refers to the sense of companionship we may feel from a robot or virtual agent. The term was previously used to describe the one-way relationship viewers often feel with characters in TV series or movies. This construct looks at elements such as psychological connection, intimacy, involvement, and more (Kumar & Benbasat, 2002). While designing robots to exhibit high social presence is a step in the right direction, the ultimate goal should be to establish a sense of companionship between the user and robot. Special emphasis should be placed on robots which exhibit qualities of parasocial presence and build interpersonal relationships through rapport building techniques.
- Exaggerate positive cues. Loneliness is a negative affect state and one of the best ways to combat negative affect is through the contagious nature of positive affect. Given that lonely individuals generally evaluate positive social cues as less positive and negative social cues as more negative (Lodder et al., 2015), social robots used for loneliness intervention should clearly display positive emotions, keeping any negatively valenced interaction to a minimum. Positive cues can be reinforced through positive body language, happy facial expressions, humor, and more.
- Reduce ambiguity. Loneliness causes difficulties interpreting social stimuli. As such, the use of neutral intonation or unclear social cues should be avoided if possible to ensure that social information portrayed by the robot is clearly understood. Additionally, any messaging or video conferencing platforms available through the robot or associated apps should ensure that social cues can easily be transmitted between multiple users. For example, messaging portals should include emojis which can help provide alternative avenues through which social cues can be communicated. Any video or voice communication technology should be of sufficient quality that it does not negatively impact the transmission of visual or verbal social signals.
- Consider implementing therapy. The fields of clinical and counseling psychology, as well as other associated mental health fields, have been helping individuals reduce loneliness for decades. Human factors psychologists and other technology development professionals often fail to consider established mental health therapies when designing technology-based loneliness interventions. Finding ways to integrate existing therapies, such as cognitive behavioral therapy or emotional training, into a social robots capabilities could potentially lead to significant and meaningful decreases in loneliness among users.
- Don’t forget that “something”. Loneliness can occur for a variety of reasons and be experienced in multiple ways. Social robot developers should carefully think about which type of loneliness their robot will address in order to design a proper intervention. Understanding the target loneliness can help developers know what that “something” that needs to be addressed is, helping establish specific design requirements and likely leading to better design. Important here is that we don’t forget the user. Focus groups and participatory design with individuals who are experiencing specific types of loneliness will greatly aid the development process.
- Interpersonal skill training. Social robots are currently used to help children develop both social and emotional skills. These kinds of interventions can also be useful to help individuals correct social cue perception biases or improve social skills. For example, a robot can be used to train an individual to engage in small talk, which may improve rapport building in real world relationships.
- Help with the negative consequences of loneliness. Generally speaking, it is still unclear if the present iteration of robots can meaningfully reduce loneliness over the long-term simply by serving as direct social companions themselves. I suspect this is unlikely the case given the nature of loneliness and current artificial intelligence technology. However, this does not mean that robots cannot mitigate the effects of loneliness. Through entertainment functions, social games, humor, and other means, robots and virtual agents can help reduce negative feelings related to loneliness, such as depression, anxiety, and rumination. Additionally, by implementing exercise protocols, robots can counter some of the negative physiological effects of chronic loneliness, such as increases in heart rate or pulmonary respiration. As much importance should be placed on reducing the effects of loneliness as is placed on reducing loneliness itself.
Conclusion
Robots do not really need to be that complex to reduce loneliness; at least over the short-run. Animal like robots such as Paro seals and the Joy for All toy-like pet series have shown that loneliness can be reduced to some extent. But to really attack loneliness we need to move beyond simply improving social presence or anthropomorphism towards purposeful design. As robot development professionals, we need to develop a full understanding of the loneliness experience and implement capabilities in our robots that specifically target determinants and consequences of loneliness.
References
Beck, A. T., & Young, J. E. (1978). College blues. Psychology Today, 12(4), 80–92.
DiTommaso, E., & Spinner, B. (1997). Social and emotional loneliness: A re-examination of Weiss’ typology of loneliness. Personality and Individual Differences, 22(3), 417–427. https://doi.org/10.1016/S0191-8869(96)00204-8
Epley, N., Waytz, A., Akalis, S., & Cacioppo, J. T. (2008). When we need a human: Motivational determinants of anthropomorphism. Social Cognition, 26(2), 143–155. https://doi.org/10.1521/soco.2008.26.2.143
Epley, N., Waytz, A., & Cacioppo, J. T. (2007). On seeing human: a three-factor theory of anthropomorphism. Psychological Review, 114(4), 864–886. https://doi.org/10.1037/0033-295X.114.4.864
Helm, P. J., Medrano, M. R., Allen, J. J., & Greenberg, J. (2020). Existential isolation, loneliness, depression, and suicide ideation in young adults. Journal of social and clinical psychology, 39(8), 641–674. https://doi.org/10.1521/jscp.2020.39.8.641
Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). Loneliness and social isolation as risk factors for mortality: a meta-analytic review. Perspectives on Psychological Science, 10(2), 227–237. https://doi.org/10.1177%2F1745691614568352
Käll, A., Shafran, R., Lindegaard, T., Bennett, S., Cooper, Z., Coughtrey, A., & Andersson, G. (2020). A common elements approach to the development of a modular cognitive behavioral theory for chronic loneliness. Journal of Consulting and Clinical Psychology, 88(3), 269–282. https://doi.org/10.1037/ccp0000454
Kumar, N., & Benbasat, I. (2002, January). Para-social presence: A re-conceptualization of’social presence’to capture the relationship between a web site and her visitors. In Proceedings of the 35th annual Hawaii international conference on system sciences (pp. 106–112). IEEE.
Lee, K. M., Jung, Y., Kim, J., & Kim, S. R. (2006). Are physically embodied social agents better than disembodied social agents?: The effects of physical embodiment, tactile interaction, and people’s loneliness in human–robot interaction. International Journal of Human Computer Studies, 64(10), 962–973. https://doi.org/10.1016/j.ijhcs.2006.05.002
Li, S., Xu, L., Yu, F., & Peng, K. (2020). Does Trait Loneliness Predict Rejection of Social Robots? The Role of Reduced Attributions of Unique Humanness (Exploring the Effect of Trait Loneliness on Anthropomorphism and Acceptance of Social Robots). Paper presented at the Proceedings of the 2020 ACM/IEEE International Conference on Human-Robot Interaction. https://doi.org/10.1145/3319502.3374777
Lodder, G. M., Scholte, R. H., Clemens, I. A., Engels, R. C., Goossens, L., & Verhagen, M. (2015). Loneliness and hypervigilance to social cues in females: An eye-tracking study. PloS one, 10(4), e0125141. https://doi.org/10.1371/journal.pone.0125141
Montalvo, F., Alves, G. M., Payne, C. A., Sasser, J. A., Burn, A., McConnell, D. S., & Smither, J. A. (2019, May 25). Loneliness and social presence: Human-robot interaction mimics human-human interaction. Paper presented at the 31st Association for Psychological Science Annual Convention, Washington DC.
Overland, M. (1991). Differentiating state and trait loneliness. New School for Social Research,
Sutin, A. R., Stephan, Y., Luchetti, M., & Terracciano, A. (2020). Loneliness and risk of dementia. The Journals of Gerontology: Series B, 75(7), 1414–1422. https://doi.org/10.1093/geronb/gby112
Weiss, R. S. (1973). Loneliness: The experience of emotional and social isolation. Cambridge, Massachussets: MIT Press.
Yalom, I. (1980). Existential Psychotherapy. New York: Basic Books.
