What comes to mind easily gets treated as common

A shortcut with a sensible design

When somebody asks how common a thing is, the mind does not consult a tally. It tries to retrieve instances, notices how easily they arrive, and converts that ease into an estimate of frequency. This is the availability heuristic, and calling it a bias undersells it: for most of human experience the shortcut worked, because things you encountered often were things you could recall easily. Retrieval fluency was a decent proxy for the underlying rate.

The proxy breaks when something other than frequency drives retrieval. Vividness, emotional charge, recency and sheer repetition in the surrounding conversation all make an example easier to summon without making the underlying event any more common. The estimate moves anyway, because the machinery cannot tell why the example came quickly.

What is well established, and what is shakier

The core phenomenon — that judged frequency tracks ease of retrieval rather than actual counts — is one of the older results in the field and has held up across many demonstrations. The classic pattern is that people misrank causes of death, overestimating dramatic and heavily reported ones and underestimating the quiet, statistically much larger ones. That basic ordering error reproduces well.

A more specific claim has fared less well. A line of research argued that it is the subjective feeling of difficulty during retrieval that does the work, tested by asking people to recall many examples versus few and observing that struggling to produce many made people judge the trait less common. Replication of these ease-of-retrieval manipulations has been mixed, with several large attempts finding weaker or inconsistent effects. So the heuristic is solid and one popular explanation of its mechanism is not settled.

Why the news is the worst input

Reporting selects for the unusual by definition, since an event that happens constantly isn’t news. This means the flow of information most people use to calibrate their sense of risk is filtered by a rule almost exactly opposite to frequency. The rarer and more shocking an event, the more coverage per occurrence it receives, and the more retrievable it becomes.

The result is a fairly predictable distortion in public risk perception: dramatic, sudden and involuntary hazards are rated far above chronic, cumulative and voluntary ones that account for much more harm. Nobody is being lied to. Every individual report can be accurate while the aggregate impression is badly wrong, which is a harder problem than misinformation and gets much less attention.

The personal version is quieter and harder to see

Availability also runs on your own memory, where the sampling is at least as biased. Ask two people who share a household to estimate what proportion of the chores they each do, and the numbers will usually sum to well above the total, because your own contributions are easier to recall than someone else’s. The same asymmetry appears in teams estimating their share of a joint effort.

This is worth noticing because it does not feel like an estimate at all. It feels like a memory of how things are. The correction is not to trust the other person’s number instead, but to recognise that both figures were generated by a process that had privileged access to one side of the record.

What actually reduces it

Replacing the retrieval with a count works whenever a count exists, which is more often than people check. Where no count exists, deliberately searching for the cases you would not have recalled — the quiet failures, the ordinary weeks, the times nothing happened — brings the sample closer to the population, though it never gets there.

A second corrective is to notice the source of the vividness. If an example arrives with unusual force, ask whether it arrived because it is typical or because it was memorable, and those two are frequently in tension. Anything you can picture in detail should be treated with slightly more suspicion than something you cannot, which is an uncomfortable inversion of how conviction normally works.

Living with a heuristic you can’t switch off

None of this is removable. The retrieval happens before any deliberate reasoning gets a turn, and the feeling of frequency arrives already formed. What can change is what you do next, and the useful habit is small: treat the first estimate as a reading from an instrument with a known bias, then ask what would have to be true for it to be right.

Sometimes it’s right. Availability is not systematically wrong, it is systematically sensitive to something other than frequency, and often the two coincide. The skill is in recognising the situations where they come apart — rare events, heavily reported events, anything you have a personal stake in remembering.