Personalized Social Feeds Linked to Mood-Related Brain Activity in Small Study

The study included 60 young adults with an average age of 20. Before viewing videos, participants completed validated questionnaires that measured symptoms of anxiety and depression.

They then watched two sets of videos. One set came from recommendations on the participants’ own Instagram or TikTok accounts. The other set consisted of generalized trending videos. Researchers also categorized the personalized recommendations by content type, although the supplied report does not describe those categories in detail.

During viewing, the team used electroencephalography, or EEG. EEG records the brain’s electrical activity in real time, within milliseconds. The researchers focused on a measure called frontal alpha asymmetry, which looks at differences in activity across the front areas of the brain.

According to the researchers, greater left-sided activity on this measure is generally associated with more positive emotional processing. Greater right-sided activity is generally associated with more negative processing and withdrawal-related behavior. This is a measure of moment-to-moment brain activity, not a diagnosis of depression.

The researchers accounted for participants’ overall screen time, treating it as a possible outside factor that could affect the results. Their focus was not simply on how long people spent on social media, but on the kind of material their feeds recommended.

They found that participants reporting more depressive symptoms also received recommendations containing more depressive content. Those participants’ EEG activity while viewing personalized recommendations was described as characteristic of more negative emotional processing. The report says this pattern differed from their brain activity while they watched generalized trending videos.

That comparison is important because it suggests personalized material may be related to a person’s emotional response in a different way than broadly popular content. Still, the study measured patterns at one point in time. It does not establish which came first: depressive symptoms, engagement with depressive material, or recommendations containing more depressive material.

The study also cannot determine whether an algorithm itself caused a change in participants’ mood. People may seek out, pause on, share or otherwise engage with content that fits how they already feel. Those actions can shape future recommendations. It is also possible that several factors work together.

The research is described as a proof-of-concept study, meaning it is an early effort to test whether this kind of real-time brain measurement can help examine social-media experiences. Its 60 participants were young adults, so the results may not apply to people of different ages or to all social-media users.

The supplied report does not give effect sizes, detailed statistical results, or information about how long any emotional effects lasted. It also does not provide a follow-up period. Larger studies that track people over time would be needed to test whether particular recommendation patterns predict changes in mood, and whether changing what is recommended changes those outcomes.

For now, the findings add to the idea that the content people see may matter alongside the amount of time they spend online. They do not show that personalized feeds inevitably harm mental health, but they raise a question researchers can study more closely: how do platforms’ recommendations interact with a person’s existing emotional state?