In recent years, the pervasive influence of social media in our lives has sparked heated debates on its potential impact on mental health, particularly concerning loneliness. A new study published in the Journal of Social and Personal Relationships explored this issue by examining how the algorithms driving social media platforms like Instagram affect feelings of loneliness among users. The findings provide evidence that perceptions of social media algorithms are linked to feelings of loneliness.
The research was motivated by the increasing prevalence of loneliness alongside the rise of social media use. Over the past decade, the U.S. has seen a steady increase in both loneliness and the use of platforms like Instagram. Some researchers argue that social media contributes to a loneliness epidemic by displacing face-to-face interactions, while others believe it provides new opportunities for connection.
The relationship between social media use and loneliness remains highly debated, with mixed findings suggesting both positive and negative effects. This study aimed to clarify this complex relationship by focusing on the role of social media algorithms, which curate content based on user engagement and are integral to the user experience on platforms like Instagram.
“Social media companies, especially Meta, make claims that their personalization algorithms are designed to aid the formation and maintenance of social relationships. My interest is understanding how personalization can promote feelings of closeness to friends and family, because the ideas personalization and personal relationships are often presented as opposites,” said study author Samuel Hardman Taylor, an assistant professor at the University of Illinois Chicago.
The primary goal was to investigate how users’ perceptions of social media algorithms influence their feelings of loneliness. The researchers hypothesized that users who perceive these algorithms as responsive and supportive would experience less loneliness, while those who see them as insensitive or dismissive would feel more isolated. The research was conducted in two parts: a cross-sectional study examining a broad age range of Instagram users and a longitudinal study focusing on young adults.
The first study involved 468 Instagram users, aged 19 to 68, recruited from Amazon’s Mechanical Turk. Participants were surveyed about their awareness of Instagram’s algorithms and their perceptions of how responsive these algorithms were to their needs and identity.
The researchers measured two dimensions of perceived algorithm responsiveness: perceived algorithm responsiveness and perceived algorithm insensitivity. Perceived algorithm responsiveness refers to the extent to which users feel the algorithms understand and support their goals, while perceived algorithm insensitivity indicates the degree to which users feel ignored or undermined by the algorithms.
The cross-sectional analysis revealed that higher levels of perceived algorithm insensitivity were associated with greater loneliness across all age groups. Interestingly, the relationship between perceived algorithm responsiveness and loneliness varied by age.
Among younger adults, higher perceived algorithm responsiveness was linked to lower loneliness, suggesting that supportive algorithmic curation could enhance social connectedness. Among older adults, higher perceived algorithm responsiveness was associated with increased loneliness, possibly because these users might substitute online interactions for face-to-face connections.
In the second study, 155 undergraduate students aged 18 to 29 participated in a three-wave longitudinal survey conducted over four weeks. This study aimed to explore the bidirectional relationship between algorithm perceptions and loneliness over time. Participants reported their loneliness, perceptions of algorithm responsiveness, and their engagement in different types of Instagram relational maintenance (active public interactions like commenting, active private interactions like direct messaging, and passive public interactions like browsing).
While perceptions of algorithm responsiveness did not predict future loneliness, loneliness did predict subsequent perceptions of algorithm responsiveness. Specifically, lonelier individuals tended to perceive Instagram’s algorithms as less responsive over time. This suggests that loneliness can color users’ perceptions of social media algorithms, potentially creating a feedback loop where loneliness begets more loneliness due to negative perceptions of algorithmic curation.
The researchers also found that active public and passive public relational maintenance mediated the relationship between perceived algorithm responsiveness and loneliness. Users who perceived the algorithms as supportive were more likely to engage in commenting, liking, and browsing content from their social network, which in turn was associated with lower loneliness. However, perceived algorithm insensitivity’s relationship with loneliness was not explained by relational maintenance behaviors, indicating that other factors might be at play.
“Two main takeaways from this study are (1) when people felt that Instagram’s recommendation algorithms were supporting their goals and identity, they were more likely use the platform to communicate with their friends and family, (2) however; people who were lonely reported less supportive from algorithms. Thus, loneliness may beget more loneliness because of how algorithms are perceived the user,” Taylor told PsyPost.
While the study provides valuable insights, it also has limitations. The cross-sectional nature of Study 1 makes it difficult to establish causality, and the longitudinal analysis in Study 2 was limited by a short time frame. Additionally, the reliance on self-reported data introduces the possibility of demand effects, where participants might alter their responses based on their awareness of the study’s aims.
Future research should consider longer study durations to better understand the long-term effects of algorithm perceptions on loneliness. Exploring other potential mechanisms beyond relational maintenance, such as the content of algorithmically curated interactions, could also provide deeper insights. Additionally, experimental designs could help mitigate demand effects and offer more robust conclusions.
“I’m working on building this into research on the lonely algorithm problem, which scrutinizes the claims from social media companies about the interpersonal consequences of their algorithms,” Taylor said. “My goal in this research is to identify how humans and algorithms work together to produce outcomes.”
The study, “Lonely Algorithms: A Longitudinal Investigation Into the Bidirectional Relationship Between Algorithm Responsiveness and Loneliness,” was authored by Samuel Hardman Taylor and Mina Choi.