Content, not screen time, tied to depression

Clinical takeaway: Ask patients what their feeds show them, not just how long they scroll. The former tracked with depressive symptoms even after accounting for screen time. 

Counseling about social media still tends to start and end with a single question: how many hours a day? But screen time has proven a blunt instrument. Studies linking it to depression keep returning small effects. That leaves clinicians with little to say beyond a generic advisory to cut back, advice that lands poorly with young patients and may miss what matters about their time online.

Part of the problem is that an hour of scrolling is not one exposure. Two users with identical screen time can inhabit entirely different feeds because recommendation algorithms tailor content to each user's interests and past behavior. And because the algorithms are proprietary, researchers have had to rely on after-the-fact self-report about what people saw and how it felt. A new pilot study takes a more direct approach. It pairs the videos young adults' own accounts served them with a real-time neural measure of emotional processing, asking whether what the feed delivers, and how the brain receives it, tracks with depressive symptoms.

Young adults with more depressive symptoms saw a different feed and processed it differently. After accounting for sex and total screen time, participants with more symptoms viewed fewer positive, more negative, and more neutral videos about social relationships in their personalized feeds, and their brain activity while watching showed the pattern of negative emotional processing associated with depression. That neural pattern did not appear when the same participants watched generalized trending videos. Seeing fewer positive relationship videos was itself linked to more negative neural processing.

The effect was not that personalized feeds felt worse across the board. On average, neural processing did not differ between personal and trending videos, and participants reported liking their own recommendations more. Relationship content, videos about friendships and romantic partners, was the most common category served, making up roughly 20% of what participants' accounts recommended.

The study enrolled 60 young adults, mean age 20, recruited from undergraduate psychology courses; most were female. Each watched two sets of short videos while electroencephalography recorded frontal alpha asymmetry, a measure of the balance of activity between left and right frontal regions in which a rightward tilt reflects more negative emotional processing. One set was pulled directly from the participant's own TikTok or Instagram account; the other came from new accounts with no viewing history, capturing what the platforms serve a generic user. Recommended videos were content-coded and relationship videos rated positive, neutral, or negative.

The researchers are now collecting data on a cohort of teens ages 13 to 16 for a potential long-term study of how algorithm-driven recommendations shape mood over time. In the meantime, the team's working intervention is educational rather than restrictive: teaching frequent TikTok and Instagram users to curate their feeds by following and engaging with creators who serve positive content, and to reset a stale account to its defaults so the algorithm can be retaught. That stands in contrast to blanket screen-time limits, which the investigators argue do little to show young users how to respond to what their feeds serve them.

"What the algorithm recommends to you matters, too," said first author Carole Leung, a doctoral student in cognition and neuroscience at The University of Texas at Dallas. "The emotional content in your personalized feed is linked to your brain's real-time emotional responses and to depressive symptoms."

Source: Leung C, et al. (2026 Jun 25) Comput Human Behav. Neural emotional processing of personally recommended short-video content and depressive symptoms