What the data actually shows
Neil Weinstein's foundational work on unrealistic optimism in the 1980s found that when people rated their own chances of various misfortunes against the average person's, they consistently judged themselves less at risk — for everything from disease to being a crime victim. The pattern held across a wide range of events and was remarkably consistent.
Tali Sharot's research suggests this optimism is pervasive and partly hardwired. A central finding is asymmetric belief updating: when people receive good news (their risk is lower than they thought) they update their beliefs readily, but when they receive bad news (their risk is higher) they update much less. The brain, in effect, learns more eagerly from information that supports a rosy outlook.
Crucially, the research frames this as largely adaptive rather than a flaw. A mild optimism bias is associated with lower anxiety, more forward planning, and the motivation to act — believing things will probably work out is part of what makes people try. The same studies note it has a clear downside when it tips into ignoring real and addressable risks.
Why this feels different from how it actually is
It feels reasonable rather than biased because you have rich, vivid access to your own intentions and habits — you know you drive carefully, eat reasonably, choose well — and only a thin, statistical sense of everyone else. So your own risk feels controllable and others' feels abstract, which tilts the comparison in your favour.
The asymmetric updating Sharot describes means the belief is self-reinforcing. Reassuring information slides in easily and sticks; alarming information meets more resistance and fades. Over time this quietly tunes your sense of risk downward without any deliberate wishful thinking.
There is also a comfort dimension. Living with a clear-eyed sense of every statistical danger you face would be anxiety-inducing, so a degree of 'it won't be me' is psychologically protective. The feeling of relative safety is doing a real job — it just isn't a measurement.
'It won't happen to me' is not a personal quirk or a failure of imagination — it is a default setting of human cognition.
What the research says to do about it
The research does not suggest replacing optimism with pessimism — both extremes have costs. What it points to is targeted realism for the small number of high-stakes, addressable risks: the handful of areas where being under-prepared is genuinely expensive, such as basic insurance, emergency savings, health screening, and obvious safety steps. The aim is to keep optimism's benefits while closing its most costly blind spots.
Because we update poorly from bad news, it helps to make protective steps automatic and decided in advance rather than left to in-the-moment judgement, which the bias will quietly skew. Setting up an emergency fund, default insurance, or scheduled check-ups removes the need to feel at risk in order to act.
Distinguishing controllable from uncontrollable risks is also useful. The bias is strongest for events that feel within our control, where our sense of personal competence does real work. For those, modest, concrete precautions tend to be worth more than the reassurance that we are careful people.
What the research says does not help
Trying to scare yourself into preparedness with vivid worst-case stories tends not to work well and can backfire — because of asymmetric updating, alarming information is exactly what the mind resists absorbing, and heavy fear appeals often produce avoidance rather than action.
Swinging to chronic pessimism is not the fix either. The same body of research associates a mild optimism bias with lower anxiety and more initiative, so abandoning it wholesale tends to cost wellbeing without reliably improving decisions. The goal is calibrated preparation, not gloom.
Simply knowing about the optimism bias does little on its own. Awareness rarely shifts the underlying tilt, which is why pre-committed, automatic safeguards outperform resolutions to 'be more realistic about risk' — they protect you without requiring you to feel personally endangered first.
The feeling of relative safety is doing a real job — it just isn't a measurement.
What this looks like in real life
'I'll deal with insurance later'
Someone skips renters' or income insurance because a break-in or a lost job feels like the kind of thing that happens to other people. This is the optimism bias doing its quiet work: their own risk feels controllable, everyone else's abstract. The research doesn't say to swap the optimism for dread — it says to make the protective step automatic and decided in advance, so acting doesn't depend on first feeling personally at risk.
Good news sticks, bad news slides off
Told a screening lowers their risk, a person adjusts their outlook happily and remembers it. Told a habit raises their risk, they find reasons to discount it and it fades. That asymmetric updating is why certain risks are so hard to take seriously even when the odds are known — not wishful thinking so much as a brain that learns more eagerly from reassurance than from warnings.
Real numbers in context
There isn't a single headline number for the optimism bias; it is a direction that shows up reliably across studies rather than a fixed percentage. The consistent finding is comparative: when large samples rate their own risk of negative events against the average person's, the majority place themselves below average — a statistical impossibility if everyone is right, which is the whole point.
Treat any precise 'people underestimate risk by X%' figure with caution, since the size varies by event and population. What is robust is the shape: optimism about one's own future is widespread, appears across cultures, is reinforced by asymmetric updating from good versus bad news, and is strong enough that researchers describe it as a default rather than an exception.