Your Instagram profile might be revealing more about your emotional well-being than you realize. A new study suggests that everyday activity on the platform, from the pictures you post to the time you spend scrolling, could provide useful clues about how lonely someone feels.
Researchers at the University of California, San Diego analyzed Instagram data from 415 college students and discovered that different patterns of online behavior were associated with different types of loneliness.
The study, published in npj Mental Health Research, examined posting habits, likes, comments, captions, and facial expressions in photographs. Researchers then used machine-learning models to compare these digital traces against five established measures related to loneliness.
Your Instagram activity might reveal two kinds of loneliness
The findings specifically described two distinct kinds of loneliness. One is emotional, and the other is social. Both of these are visible from different digital footprints. The researchers explain the distinction in their paper:
The researchers explain the distinction in their paper:
We also asked whether emotional and social loneliness showed distinct digital signatures. Compared to EL, SL was more strongly associated with indicators of broad social participation (e.g., liking and commenting frequency, weekend posting, and the proportion of posts with faces), and higher engagement in these behaviors tended to accompany lower SL. In contrast, EL was more strongly associated with negative affect (negative sentiment in language; sad or angry facial expressions) and late-night posting, consistent with evidence that loneliness co-occurs with affective distress and sleep disruption.
In simpler terms, people experiencing greater social loneliness tended to interact less with others on Instagram. Those reporting greater emotional loneliness were more likely to exhibit negative emotions in their posts and remain active late at night.
Researchers also found that pictures containing fewer faces were associated with higher loneliness across multiple measures. This was the only individual feature that retained a significant association after adjusting for depression and anxiety.
Instagram knows plenty, but it doesn’t paint the entire picture
The researchers’ machine-learning models achieved an area under the curve (AUC) score of 0.77 for the UCLA Loneliness Scale, indicating moderate ability to distinguish participants with higher and lower loneliness scores. Performance was similarly moderate for emotional and social loneliness, both reaching 0.74. However, the models were a lot less effective at identifying structural social isolation, which concerns the actual availability of friends, family, and social support.
Keep in mind that the participants were all university students, predominantly women, and the study cannot establish whether Instagram behavior causes loneliness or simply reflects it. The researchers also emphasize that their work relied on data willingly shared by participants, rather than any existing Instagram feature that automatically monitors users’ mental health.
Even so, the findings suggest that our digital habits might offer researchers another way to understand loneliness. The challenge will be ensuring these insights are used responsibly, with consent, rather than turning people’s personal social media activity into an unwelcome psychological assessment.
