How the Internet Is Quietly Breaking People’s Minds

There’s a person in every comment section you’ve ever scrolled through. You know the one. They’re speaking in all caps about something that happened weeks ago. They’re connecting dots that don’t connect. They’re absolutely certain, with a seething, radioactive certainty, that everyone else is lying, blind, or complicit. They’re announcing and demanding agreement. There’s a quality to it that feels less like disagreement and more like... broadcast. Like they’re transmitting from somewhere you can’t quite reach.

You probably shook your head and kept scrolling. Maybe laughed at them.

But here’s the thing I keep coming back to: what if that person didn’t start that way? What if something happened to them that moved them from a functional human being to whatever that is?

I think we need to talk about what the internet is doing to the brain. Not in the vague, hand-wavy “screens are bad” way we’ve been having this conversation for twenty years. I mean mechanistically. I want to talk about cognitive bandwidth, information saturation, and the uncomfortable reality that for a meaningful subset of people, constant exposure to the internet’s information firehose is producing something that looks, clinically, a lot like psychosis.

The Brain Was Not Built for This

Let’s start with the basics, because they’re more radical than they sound.

The human brain has a fixed capacity for processing information. This is a hard architectural limit: your working memory can hold roughly seven items at once, and your prefrontal cortex, the part responsible for filtering signals from noise and making rational decisions, can only do so much before it starts to buckle under the load.

For most of human history, information arrived slowly. You learned what you needed to survive your immediate environment, and even as recently as a few generations ago, the information ecosystem was relatively bounded: a newspaper, the evening news, what your neighbors told you. The brain evolved for that and it was calibrated for that.

Now consider that researchers have estimated the average person today receives more information in a single day than someone living a century ago encountered in an entire lifetime. That’s the scale of what we’ve done to ourselves, and we’ve taken a cognitive system designed for a specific throughput and we’ve flooded it, continuously, from the moment we wake up until the moment we go to sleep.

What happens when a system runs past its capacity, persistently, with no relief valve?

It distorts.

Cognitive Fragmentation Is Now a Clinical Category

In November 2025, researchers published a paper in Frontiers in Digital Health that should have been front-page news. The study introduced a new framework for understanding mental health in the digital era, identifying what they called emergent psychopathologies or disorders that traditional diagnostic manuals were simply not built to capture. At the top of the list were two conditions endorsed by over 80% of the 75 licensed mental health clinicians surveyed: Continuous Partial Attention Disorder and Digital Anxiety Disorder.

Continuous Partial Attention Disorder is exactly what it sounds like. It’s the state of never being fully present anywhere and always monitoring, always scanning, always half-attending to the stream. It’s what happens when your nervous system adapts to constant input by never fully committing its resources to any single thing. You become, in a very real sense, cognitively fragmented. The ability to hold a sustained thought, to follow a long argument, to sit with complexity degrade.

Cognitive fragmentation doesn’t just make you scattered. It makes you vulnerable. When you can’t sustain careful reasoning, you become more susceptible to the thing the brain falls back on when it’s overwhelmed. Things like pattern matching, emotional response, and the path of least cognitive resistance.

That path leads directly into the arms of the algorithm.

The Algorithm’s Not Your Friend and It’s Not Neutral.

We’ve had the social media conversation so many times that most people have developed a kind of immunity to it. Yes, the algorithm maximizes engagement. Yes, that means outrage. We know. We’ve accepted it as background toxicity.

But I don’t think we’ve fully reckoned with what this means when you combine it with a cognitively depleted user because the algorithm isn’t just showing you content for a reaction. It’s actively learning your cognitive vulnerabilities and exploiting them with increasing precision.

Research published in PNAS analyzing more than 100 million pieces of content across Facebook, Reddit, and the artist formerly knkown as Twitter (X) found that the aggregation of users into homophilic clusters dominates online dynamics.

Homophilic = groups of people who share beliefs and reinforce each other.

The algorithms detect your confirmation biases, amplify them, and feed them back to you. Every engagement, every hover, every lingering glance at a post signals the system to narrow the aperture further.

What you end up with is what researchers call a filter bubble: a personalized information environment that functions less like a window onto the world and more like a mirror. And mirrors, as anyone who’s spent too long staring into one knows, can start to do strange things to your sense of reality.

The particularly insidious part is that the mirror doesn’t reflect you accurately. It reflects the most emotionally activated version of you. The version that clicked on the scary headline, or the one that lingered on the post about betrayal, about enemies, about hidden truth. The algorithm doesn’t know you, but it knows your worst hour, and it keeps showing you more of it if you keep showing that version of yourself to the algorithm. A snake eating itself metaphor would be applicable to the circumstance.

When Confirmation Bias Becomes Something More

Here’s the clinical bridge I want to walk you across carefully, because I think it’s important and easy to overstate.

Psychosis, in the clinical sense, involves a break from shared reality and typically through hallucinations, delusions, or severely disorganized thinking. It has biological substrates, and it runs in families. It’s not something you can scroll your way into if you’re neurologically healthy and your life is otherwise stable.

The research is clear, and clinicians are increasingly saying it out loud: the internet doesn't cause psychosis, but for people carrying latent vulnerability, the sleep-deprived, the chronically stressed, those with preexisting conditions that already strain their grip on reality, it acts as an accelerant. Psychosis fuel.

A 2025 review published in Frontiers in Psychiatry proposed what researchers called the Delusion Amplification Model. The argument is that for people prone to unusual beliefs, things like paranoid thinking, grandiosity, suspiciousness, or a fragile grip on consensus reality, curated feeds, echo chambers, and anonymous online communities don’t just reflect those tendencies. They actively reinforce and elaborate them.

Think about what an echo chamber does to a nascent delusional belief. You believe, tentatively, that something is being hidden from you. You search for information confirming that belief. The algorithm, detecting your engagement pattern, shows you more. The community you find online reflects your belief back to you, validated by strangers who feel like the only people who get it. The belief grows more elaborate, more certain, more insulated from contradiction. Every piece of disconfirming information becomes, within this framework, further evidence of the conspiracy.

This is not metaphorical. This is a mechanistic description of how certain people’s relationship to reality deteriorates in digital environments.

Doomscrolling Is the Gateway Drug

Before we get to the full collapse, there’s an intermediate state worth naming. You’ve probably been in it, and I know I’ve been in it.

It starts with a notification, then a headline, and then a cascade of headlines, each one more alarming than the last. Before you realize it, you’ve lost forty minutes and you feel worse than when you started. You are anxious, agitated, vaguely convinced that everything is burning down, and somehow unable to stop.

This is doomscrolling. The term entered Merriam-Webster’s dictionary because it was describing something real and widespread enough to need a name. But the mechanics of it are worth sitting with. Mayo Clinic psychologist Craig Sawchuk explains that the brain is hardwired to orient toward novelty and threat and this wiring kept humans alive for tens of thousands of years. Social media exploits it with ruthless efficiency. The scroll is infinite. The content is algorithmically selected for maximum emotional activation. Crucially, the compulsion to keep going is rooted in the same cognitive system that drove our ancestors to check the tree line for predators: the desperate need to feel informed and therefore safe.

Except the tree line never empties. All rivers run to the sea, but the sea is never full.

The threats never resolve, and unlike a predator you can either flee or accept has moved on, the internet’s threat environment replenishes itself constantly, in real time, faster than any human nervous system can process and discharge.

Prolonged exposure to this, and we’re talking about the average adult spending multiple hours a day in this state, produces measurable outcomes: elevated cortisol, disrupted sleep, sustained anxiety, depression, and what researchers have documented as social media fatigue. The emotional exhaustion that comes from a cognitive system running on overdrive with no recovery period.

Sleep disruption alone creates a feedback loop worth noting. A 2022-2023 study of university students found that heavy late-night online activity disrupted sleep, and that insomnia fully mediated the link between compulsive internet use and paranoid ideation. Sleep deprivation makes your prefrontal cortex less effective at distinguishing signal from noise and the paranoid ideation increases with continuous exposure.

The Comments Section as Diagnostic Window

So let’s talk about what you see when you look at the comments.

There’s a specific behavioral signature that I think most of us recognize intuitively but haven’t fully named: the person whose online presence has decoupled from ordinary social reality. They’re performing certainty for an invisible audience that, in their mind, they are either exposing or rallying. The content of their certainty almost doesn’t matter, and it could be political, medical, conspiratorial, religious. What matters is the quality: absolute, indignant, sealed against contradiction, often grandiose and definitely delusional.

This maps, uncomfortably precisely, onto the clinical description of psychotic thinking because the cognitive and emotional architecture is similar. The capacity for epistemic humility or for holding a belief while acknowledging you might be wrong has been completely overwhelmed. The person is no longer in dialogue with reality and, instead, they are broadcasting a version of it.

What the internet has done is provide an audience for this. Before social media, the person having this kind of cognitive experience might have been visible only to the people immediately around them. The friction of the real world, the faces looking at you blankly, the social signals of confusion or concern, provided a kind of reality check. Online, that friction is removed and you can find, within minutes, a community that reflects your beliefs back at you. The more extreme the belief, the more dedicated the community, because extreme beliefs select for committed adherents.

The algorithm amplifies this because outrage and certainty generate engagement, and engagement generates revenue. The platform has no interest in your epistemic health: only in your attention.

Who’s Most Vulnerable

It would be comforting to believe this only happens to people who were already unstable. The research suggests the picture is more complex and less comfortable than that.

Yes, people with existing psychotic-spectrum conditions are at elevated risk. The study published in the Schizophrenia Bulletin found that among adults with psychosis, those experiencing paranoia and grandiosity were more likely to use the internet for mental health information in ways that could reinforce their symptoms through a feedback loop where the illness drives the search behavior and the search behavior drives the illness. We are back to the snake metaphor, again…

But vulnerability is a spectrum, not a binary. Stress, sleep deprivation, grief, isolation, chronic anxiety, all of these things reduce cognitive bandwidth and increase susceptibility to the distortions the algorithm offers. The pandemic years were a controlled experiment in this: isolation drove people online, doomscrolling exploded, and the WHO coined the term “infodemic” to describe the parallel epidemic of misinformation that emerged alongside COVID. Rates of anxiety and depression increased 25% globally. Conspiracy theories spread at viral scale, and these weren’t all people with pre-existing psychiatric conditions. They were ordinary people whose cognitive systems were overwhelmed by circumstances, and the internet was waiting.

Please Don’t Misunderstand

I want to be clear about what I’m saying.

What I am saying is that we’ve built an information environment that human cognition was not designed to handle, and we’ve handed it to people with no instruction manual, no cognitive training, and in many cases no awareness of what it’s doing to them. We’ve created systems that profit from emotional activation and have optimized them relentlessly for that purpose. We’ve removed the social friction that used to provide a reality check for distorted thinking, and we’re watching the results play out in real time, in the comments, in the threads, in the increasingly wild gap between how people talk online and how they function in the actual world.

Mental health professionals are scrambling to categorize what they’re seeing. The diagnostic frameworks we have don’t cleanly capture what happens when a person’s relationship to reality is shaped primarily by an algorithmic feed designed to make them emotionally charged, so we’re going to need new language.

Moving Forward…

The next time you see someone in a comment section behaving in a way that seems, frankly, unhinged, I’m not suggesting you feel sorry for them in a way that obligates you to engage because I’ve tried that and it just gets worse. I’m suggesting you consider the possibility that you’re not watching a moral failure, but you’re engaging in a cognitive one. You’re watching what happens when a human nervous system, possibly already under pressure, possibly already vulnerable, gets fed into a machine that has been specifically optimized to find its cracks and pull.

That doesn’t excuse the behavior. But it might change how you understand it. And understanding it is the first step toward doing something about the machine itself.

If you found this useful, share it with someone who needs to hear it, and maybe close the app when you’re done.