Why Do We Fall for Things We See Online?

What the data actually shows

Repetition is one of the most robust findings in the field. The illusory truth effect — that simply encountering a statement more than once makes it feel more accurate — has been demonstrated repeatedly since the late 1970s and shown to apply even to implausible claims and even when people know better. In a feed that surfaces the same claim again and again, familiarity quietly does the work of credibility.

Source memory is weaker than content memory. People routinely retain a claim while forgetting where it came from — sometimes misremembering a dubious source as a trustworthy one. The effect of a 'this is unreliable' warning fades faster than the claim it was attached to, so over time a thinly sourced assertion can detach from its origin and float free as something that just feels known.

Susceptibility looks more like laziness than bias. Work by Pennycook and Rand (for example, 2019 and 2021) argues that people who fall for fake news tend to be those who engage in less reflective, more intuitive thinking — and that analytic thinking predicts better truth discernment across the political spectrum, rather than simply reinforcing one's prior beliefs. They also find a striking gap between believing and sharing: people will share headlines they could have identified as false if asked, because the social context of sharing pulls attention toward engagement and away from accuracy.

And false content can travel faster than true content. A widely cited 2018 study (Vosoughi, Roy and Aral, in Science) of news cascades on Twitter found that false stories spread further and faster than true ones, plausibly because novelty and emotional charge drive sharing. The structure of the platform rewards what spreads, not what is accurate.

Why this feels different from how it actually is

It feels like the problem is other people — the gullible, the biased, the people on the other side — because we experience our own beliefs from the inside as obviously reasonable. But the mechanisms here are general human ones. The illusory truth effect and source forgetting operate on everyone; the question is only which claims happen to repeat in front of you.

It also feels like falling for misinformation must require strong motivation or ideology, when much of the evidence suggests the opposite. A lot of the time we are simply not thinking hard — scrolling quickly, half-attending, reacting. The 'lazy not biased' framing is uncomfortable precisely because it implicates ordinary inattention rather than a villain.

And the speed of sharing hides the gap between feeling and judgment. In a calm moment, asked directly, most people can tell a lot of false headlines from true ones. But the feed is not a calm moment; it nudges toward a quick reaction, and the share button is right there. So the same person who 'knows better' shares something they would have flagged if the question had been put to them plainly.

What the research says to do about it

The single most supported intervention is almost embarrassingly simple: pause and consider accuracy before reacting. Pennycook and Rand's 'accuracy nudge' studies find that briefly prompting people to think about whether a headline is true noticeably improves the quality of what they subsequently share — because the default failure is inattention to accuracy, not an inability to judge it.

Checking the source and the date, and reading past the headline, target the specific weaknesses in how we process online claims. Because source memory is fragile and repetition breeds false confidence, deliberately asking 'where is this actually from, and when?' reinserts the context the feed strips away. Lateral reading — opening a new tab to see what other sources say about a claim — is a habit professional fact-checkers use and that research supports.

'Prebunking' shows promise as well. Studies on inoculation — briefly exposing people to the techniques of manipulation before they encounter them in the wild — find it can build durable resistance to misleading content. Knowing the moves (emotional language, fake experts, decontextualized images) makes them easier to spot.

What the research says does not help

Assuming this is only a problem for the biased or the uneducated does not help, and the evidence runs against it. Susceptibility tracks reflective thinking more than political alignment or intelligence, so 'smart people don't fall for it' is false comfort that mainly lowers your own guard.

Repeating a myth in order to debunk it can backfire if done carelessly, because the repetition itself feeds the illusory truth effect and the correction may fade faster than the false claim. Effective corrections lead with the truth and attach a clear, memorable explanation, rather than amplifying the falsehood while swatting at it.

Trying harder to spot bias in others while skipping the simple accuracy check on yourself misses where the leverage is. The research suggests the highest-value move is the boring one — slowing down before you share — not constructing elaborate theories about who is being manipulated.