21 Oct 2026 · 10 min read

12 Cognitive Biases Shaping How We Think in 2026

What each one is, what it looks like in a feed, a group chat or an AI answer, and what it costs us.

A cognitive bias is a systematic tilt in how we judge and decide. A mental shortcut pulls the conclusion away from what the evidence supports. Biases aren't flaws in a few careless people. They're part of how every human brain handles too much information with too little time.

The twelve below come from judgment and decision-making research, the field Daniel Kahneman made famous in Thinking, Fast and Slow. Each one has a plain-language version of the thought, a definition, an example from life in 2026, and what it tends to cost.

The 12 biases

1. Confirmation bias

Sounds like"I only notice evidence supporting what I already believe."

People selectively seek, interpret and remember information that supports what they already think.

2026 example: Someone believes AI will eliminate all software engineering jobs. They share every layoff announcement and skip past evidence of new roles or rising demand for particular engineering skills.

Impact: Polarisation, misinformation, poor decisions.

2. Negativity bias

Sounds like"One bad thing outweighs ten good things."

Negative experiences and information get more attention than their share of reality.

2026 example: A video gets 100 supportive comments and one insult. The creator spends the evening thinking about the insult.

Impact: Rumination, a heightened sense of threat, strain in relationships.

3. Social comparison bias

Sounds like"Everyone is doing better than me."

People judge their success, looks or worth by comparing themselves with others. Comparison is a normal psychological process, not always an error. It becomes a bias when the comparison set is distorted, which is what feeds tend to do.

2026 example: A stream of 25-year-old founders, luxury lifestyles and influencers leaves someone feeling unsuccessful, despite a stable career and real achievements.

Impact: Envy, dissatisfaction, distorted expectations, pressure to achieve.

4. Availability bias

Sounds like"If I see it everywhere, it must be happening everywhere."

People mistake what is easy to recall for what is statistically common.

2026 example: After weeks of videos about cheating, dating scams or violent crime, someone concludes these experiences are nearly universal.

Impact: Exaggerated fear, pessimism, distorted sense of risk.

5. In-group bias

Sounds like"People who think like me are more trustworthy."

People tend to favour members of groups they identify with.

2026 example: Someone accepts a weak argument from their own political community and rejects a stronger one from the other side.

Impact: Tribalism, stereotyping, distrust, polarisation.

6. Authority bias and automation bias

Sounds like"It sounds confident and intelligent, so it must be right."

People can give too much credibility to experts, influencers or automated systems. The two biases are distinct, but they increasingly meet in the same place: an AI answer that reads like an expert wrote it.

2026 example: Someone accepts an AI-generated medical explanation without checking its citations, because it's written in a calm, professional tone.

Impact: Over-reliance on AI, poor judgment of real expertise, misplaced trust.

7. Fundamental attribution error

Sounds like"Their mistake reflects their character. Mine reflects my circumstances."

People tend to explain others' behaviour by personality and underweight the situation they're in.

2026 example: A colleague replies late, so you decide they're disrespectful instead of considering that they're overloaded.

Impact: Conflict, resentment, misunderstanding.

8. Survivorship bias

Sounds like"Successful people did this, so doing it leads to success."

People draw conclusions from the successes they can see and overlook the failures they can't.

2026 example: An influencer says dropping out of university to build AI apps is the fastest path to wealth. They show the founders who made it, not the thousands whose businesses failed.

Impact: Unrealistic expectations, financial risk, a distorted picture of success.

9. Anchoring bias

Sounds like"The first number I saw decides what feels reasonable."

The first piece of information has an outsized effect on later judgments.

2026 example: An online course shows an "original" price of $999, discounted to $99. The first number makes $99 feel like a bargain, with no independent evidence the course is worth either.

Impact: Overspending, weak negotiations, distorted valuations.

10. Sunk cost fallacy

Sounds like"I've invested too much to quit now."

Past investment that can't be recovered still drives the decision about what to do next.

2026 example: A founder keeps building an app nobody wants because they've already spent two years on it.

Impact: Staying in failing projects, unhealthy relationships and commitments that no longer serve you.

11. Overconfidence bias and the Dunning–Kruger effect

Sounds like"I understand this better than I actually do."

People can overestimate their knowledge or accuracy. The Dunning–Kruger effect is a narrower and still debated finding, that the least skilled tend to overestimate themselves the most. It isn't just another name for overconfidence.

2026 example: After three short videos about neuroscience, someone starts confidently correcting clinical researchers.

Impact: Poor decisions, inaccurate claims that spread.

12. Illusory truth effect

Sounds like"I've heard this so many times that it feels true."

Repetition makes a claim feel more credible, even when no new evidence has been added.

2026 example: A false nutrition claim moves through AI-generated posts, podcasts and influencer videos until its familiarity is mistaken for science.

Impact: Persistent myths, misinformation, susceptibility to repeated messaging.

Why these biases matter in 2026

The Reuters Institute's Digital News Report 2026, published in June and based on a survey of about 97,500 people across 48 markets, shows how the information environment is shifting.

54%use social media or video networks for news
62%are concerned about fake news online, up 4 points
10%use AI chatbots for news weekly, up from 7%

These are averages across survey respondents in the markets covered, not a census of the world's population.

The numbers don't show that cognitive biases have increased. They show that a growing share of the information people see arrives in settings where repetition, selective exposure, popularity cues and persuasive presentation can sway judgment. That is the territory several of the biases above, including the illusory truth effect, availability, confirmation and automation bias, work best in.

It's also easy to overstate social media's effects. A 2024 review in Nature by Budak and colleagues cautioned that common stories about the harms of online misinformation often go further than the evidence supports. Both things can be true: the setting rewards biased thinking, and the size of the damage is still being measured.

The societal list isn't the personal one

A bias can matter a great deal for society without being the most useful thing to work on in your own head. The patterns that drive a bad evening, like mind reading, catastrophising or all-or-nothing thinking, mostly come from a different tradition: the cognitive distortions taught in CBT. Cognitive biases vs cognitive distortions explains the difference and which ones to work on first. Five thinking patterns to catch first gives the question that breaks each one.

Questions people ask

What is a cognitive bias?

A systematic pattern in how people judge and decide, where a mental shortcut pulls the conclusion away from the evidence. Biases are normal features of human thinking, not signs that something is wrong with you.

Has social media made cognitive biases worse?

There's no evidence the biases themselves have grown. The setting has changed: more news now arrives through feeds built on repetition and popularity, which are the conditions several biases thrive in.

What is automation bias?

Trusting an automated system's output, such as a chatbot's answer, more than the evidence warrants, especially when it's written confidently. It overlaps with authority bias.

Is the Dunning–Kruger effect the same as overconfidence?

No. Overconfidence is the broad tendency to overestimate what you know. Dunning–Kruger is a narrower, debated claim about who overestimates most.

A note on limits

This article explains common thinking patterns for general education. It isn't a diagnosis or a treatment. If thoughts like these are tied to persistent low mood, anxiety or distress, a mental health professional can help. If you're in crisis, contact your local emergency service or a crisis line.

Sources

Catch the pattern in your own thoughts

Reframed helps you notice the thinking pattern behind a thought, check it against the evidence, and write a more accurate version. Everything stays on your device.

Download on the App Store