New study finds depression heightens the behavioural pull of Twitter likes

People living with depression appear to get a stronger behavioural nudge from the likes they receive on social media than other users, according to a new analysis of more than 7,700 Twitter accounts.

Across the data, receiving more likes on one day was linked with posting more the following day. Crucially, that reinforcement effect was reliably larger among users with depression.

This finding cuts against decades of laboratory research, which has typically suggested that depression dampens responsiveness to rewards rather than amplifying it.

Overall, the results point to a specific way social media design may interact with depression - and to which users might be most susceptible to that interaction.

The pull of social media likes

The research was led by Dr Dan-Mircea Mirea at Princeton University, working with colleagues at Princeton and Trinity College Dublin.

The team treated each post as a behaviour that can be shaped by social media affirmation. They quantified how strongly likes received on one day predicted posting behaviour the next day, referring to this relationship as an individual’s reinforcement tendency. To avoid leaning on a single sample, they did not rely on just one dataset.

Testing across datasets

To check that the pattern was genuine rather than an artefact of one unusual group, the researchers combined three distinct Twitter datasets containing more than 17 million posts.

The largest dataset matched just over 1,000 users who had disclosed a depression diagnosis in a tweet with around 5,000 randomly selected accounts.

The remaining two samples were recruited directly and involved participants reporting their own mental health. One group came via adverts on the platform, while the other was recruited through an online work marketplace and completed a validated depression questionnaire.

Likes mattered more than retweets

Because depression was assessed differently in each dataset, a result that appeared in all three was difficult to dismiss as a quirk of any single method.

The broader premise is familiar. Earlier work has suggested that the same sorts of rules used to describe how animals pursue food can also describe how people pursue approval online.

What was notable here was the consistent individual difference: in each dataset, the link between yesterday’s likes and today’s posting became stronger as depression increased.

The effect was also specific in an informative way. The association showed up for likes but not for retweets, implying that explicit approval - rather than simple visibility - was what mattered most.

In one sample, additional likes did not raise posting levels overall, yet they still exerted a stronger influence on users with depression.

A surprising reversal

For many years, lab studies have characterised depression as involving blunted reward processing.

In controlled tasks, people with depression tend to learn less from small monetary rewards or points. This aligns with reduced motivation and a diminished capacity for pleasure.

That muted response has also been observed with social rewards. One study reported that people with more depressive symptoms learned more slowly from approving faces in a laboratory task.

Online, a like is intended to serve as a comparable signal. Mirea’s results point in the opposite direction, and the authors argue that the real-world setting helps explain why.

Real life is different from a lab

A pound earned in a lab task may not carry much emotional weight, whereas a like from an actual person can deliver a social reward that token incentives cannot replicate. The surrounding context may therefore account for the difference.

This reversal also resembles earlier patterns seen in depression research. Some initial laboratory findings indicated that people with depression respond less to emotional events.

Yet studies that followed people in everyday life have often found the contrary - more intense reactions to stress and to experiences such as social exclusion.

A separate pattern linked to habit

In one of the samples, participants completed a broad mental health survey, allowing the researchers to group symptoms into clusters that cut across standard diagnostic categories.

This approach revealed two distinct online behaviour patterns mapping onto two different symptom profiles.

Greater reinforcement from likes aligned with a symptom cluster the researchers described as anxious depression, which included low mood, apathy and long-standing anxiety.

A separate pattern also emerged: some users posted at a relatively steady pace regardless of how many likes they received. This habit-like posting was associated with compulsivity and intrusive thoughts.

In other words, reward sensitivity and persistent habit appeared as separate traits, each tied to a different psychological profile.

The link between likes and depression

Because mental health was measured at only a single point in time, the study cannot show which direction the relationship runs.

It is possible that heightened responsiveness to likes is part of depression. Equally, it may be that relying heavily on likes contributes to worsening depression over time.

There are reasons to take the latter possibility seriously. One computational study found that younger users both increase posting in response to likes and experience a larger drop in mood when likes are scarce.

If not receiving likes feels painful while a steady flow of them feels soothing, people may start pursuing that reward, potentially creating a self-reinforcing cycle. Small, repeated nudges of this kind can accumulate.

Why depression seems to intensify this pull remains uncertain. People who receive less warmth or support offline may turn to social media as a more accessible source of it.

Depression has been linked with greater feedback-seeking online, so a like may simply matter more to someone experiencing fewer other rewards.

Changes on the horizon

The effect sizes for individuals are small, which the authors note is expected in noisy real-world data.

Even so, social media is used by more than 60% of people, and about one in 25 lives with depression, which makes the connection relevant to public health.

The study offers a narrow but novel contribution: until now, there had been no demonstration that depression is associated with a stronger real-world response to social rewards.

The next step is to test whether this pull helps drive depression or whether it merely mirrors it.

The same like button might function as support for some people, while acting as a lever on a vulnerable mind for others.