The Inflamed Brain

Calling the brain "inflamed" is a figure of speech, but it points at something real: feeds are built to keep us reacting and coming back. Social media works on old neural circuits, many times a day. Researchers are still measuring the effects on attention, mood and behavior, and still arguing over them.

This essay covers the rewards built into the feed, the pull of moral outrage, the strain on careful thinking, and whether digital stress reaches the immune system.

Reward and habit

At the heart of social media's grip lies the brain's reward system. Platforms are not neutral conduits; they are engineered to capture attention, and they deliver social rewards on the kind of unpredictable schedule long known to build habits. The logic resembles a slot machine's: rewards arrive unpredictably, and unpredictable rewards are especially good at building habits.

A like, a comment, a notification: each one can condition users to check again in anticipation of a payoff that may or may not come.

Brain scans show the same reward regions at work. In adolescents, seeing their own photos with many likes activates the nucleus accumbens, a core reward region. People with problematic use differ, on average, in the structure and function of reward and control regions, though no one has shown which came first.

Dopamine itself is harder to pin down. People often say the feed rewires it. That is an inference; we found no study that measures dopamine directly during social media use. In monkeys, dopamine neurons respond to cues that predict a reward and carry a sustained signal when the outcome is uncertain, which may be part of why the next scroll feels so promising.

Some researchers see parallels with substance addiction, but a systematic review of the imaging evidence calls that conclusion premature. A deeply ingrained habit is the better description, and one 2025 study found that users tend to overestimate how addicted they are.

Short-form video adds to this. TikTok, Instagram Reels and YouTube Shorts serve rapid, emotional clips chosen by algorithm, and a meta-analysis of 71 studies and more than 98,000 people links heavier use to weaker attention (r = -.38) and inhibitory control (r = -.41). Those are correlations. It is unclear which drives which.

One plausible result, not yet well tested, is a preference for low-effort, high-feedback activity. Scrolling is cheap and stimulating, and sustained thinking feels unrewarding next to it. If so, ordinary life would feel slow by contrast.

Outrage

Platforms rank posts by engagement, and moral anger gets engagement. An audit of Twitter's algorithm with 806 users found that it amplified anger and hostility toward the other side compared with a chronological feed, and that users did not prefer the political tweets it chose.

Reward shapes what people post. In a study of 12.7 million tweets, people who got more likes and shares for expressing outrage expressed more of it the next day. Doubling the feedback raised expected outrage by 2 to 3 percent. That is small per post, and there are a lot of posts. Misinformation sources evoke more outrage than trustworthy ones, and users are more willing to share outrage-evoking misinformation without reading it first.

The neuroscience is suggestive. In the lab, punishing a wrongdoer engages the striatum, part of the brain's reward-learning circuitry, which has led researchers to propose that outrage can become habitual. Punishing a wrongdoer can feel good, and doing it in public earns approval whether or not the anger is deeply felt.

Engagement is also how platforms make money. Across 2.7 million posts from news media and members of Congress on Facebook and Twitter, language about the opposing political side was the strongest measured predictor of sharing, with each such term raising the odds of a share by 67 percent. The authors caution that the result is correlational.

Many people assume feeds seal us inside a bubble of views we already hold, but a Reuters Institute review finds echo chambers less common than often assumed. Recommendation systems may do something different, showing people the most provocative versions of the views they disagree with.

Whether any of this hardens politics is contested. In a ten-day field experiment with 1,256 people on X during the 2024 U.S. presidential campaign, reranking feeds to reduce partisan animosity warmed feelings toward the opposing party by about two points on a 100-point scale, and increasing it cooled them by about as much. The researchers estimated that as equal to about three years of the long-run rise in hostility between U.S. parties. Whether the shift lasts is unknown.

Other experiments find almost nothing. When nearly 20,000 Facebook users and more than 15,000 Instagram users were paid to deactivate for six weeks before the 2020 election, the effects on polarization were close to zero. A three-month switch to chronological feeds also had no significant effect.

Hostility toward other groups is far older than the internet. Platforms scaled it up and found a way to sell ads against it.

Attention and misinformation

Researchers have also studied what the feed does to thinking. A systematic review of short-form video research in young adults found consistent associations between heavier use and attentional disruption, reduced executive functioning, and emotional dysregulation, though it relied largely on cross-sectional studies and self-report. Evidence on memory and reasoning is much thinner.

Kahneman's distinction between fast, reflexive thinking and slow, effortful thinking is one way to read this: frequent use builds habits that run on the former, though no one has shown it weakens the latter.

This matters for misinformation. Fake news spreads partly because people don't stop to consider accuracy when they share, and partly because sharing itself has become a habit. One fMRI study of 38 people implicates the left dorsolateral prefrontal cortex in both accuracy judgments and sharing decisions.

Interfaces built for speed make it easy to share without asking whether something is true. False information, once encountered, gains credibility through mere exposure, and belief in it can keep shaping people's reasoning even after it has been corrected.

Stress and inflammation

Some researchers take "inflammation" literally, though the evidence is thin. One recent opinion article proposes that chronic media stress can strain the immune system. The blood-brain barrier and neuroimmune effects it cites come mainly from animal studies of chronic stress, with no social media involved.

Social media, 24-hour news and algorithm-amplified crisis content are always on. Repeated exposure to emotional content about war, pandemics and climate disasters can cause real psychological stress, and chronic stress in general is known to affect immune function.

The proposed mechanism runs through the hypothalamic-pituitary-adrenal axis. In chronic stress generally, impaired cortisol feedback can leave inflammatory mediators elevated, including interleukin-6, tumor necrosis factor-alpha, and C-reactive protein. Whether social media produces this is not established. A small experiment found that using Facebook after a stressor impaired cortisol recovery; chronic effects have not been shown.

The direct evidence is mixed. In a national sample of 863 adults, social media use was not significantly correlated with C-reactive protein or interleukin-6 overall, though a positive link emerged among people with lower self-esteem. Later work from the same group reported a positive association with C-reactive protein and found that it predicted more social media use weeks later, which suggests the arrow may also run from inflammation to use.

For adolescents, sleep may be the most plausible route: social media use is linked to poorer sleep, and sleep loss during adolescence may affect prefrontal development. The amygdala, involved in emotional control, and the hippocampus, implicated in memory, are sensitive to stress hormones. No one has shown that the feed delivers enough stress, for long enough, to change them.

The developing brain

The brain reshapes itself in response to experience. Social media is a strong, repeated experience, and for adolescents it arrives during a sensitive period of neural development.

One study scanned 169 adolescents annually for three years. Those who checked social media more than fifteen times a day started out less responsive when anticipating social feedback, then grew more responsive, while non-checkers moved the other way. The authors could not say whether checking caused the difference. The change may promote compulsive use, or may simply be an adaptive response to a more social world; they raise both possibilities.

Small cross-sectional scans link excessive use to lower gray-matter volume in regions such as the amygdala and ventral striatum, without prefrontal differences. Studies of problematic internet use more broadly do report prefrontal differences, but none shows that use caused them.

Heavy use may be training people to react before they reason. So far that is a worry. No one has demonstrated it.

Conclusion

If "inflamed" fits, it fits as a metaphor. Rewards build habits, ranking systems profit from outrage, short video competes with sustained attention, and stress systems may stay switched on. The effects vary widely from person to person.

Polarization moved in some experiments and not in others. Effects on well-being can be real but small: six weeks off Facebook, against one week for a control group, improved emotional state by 0.06 of a standard deviation. Some scientists reject the claim that social media rewires the brain.

The feed may also be shaping how much patience we have for slow, effortful thought. We do not yet know how much. We do know it was engineered for engagement, and that we can put it down.