Where it sits
Strongest immediately, gone soonest — which makes it dangerous in the room and harmless by next week.
Tell it apart
| Aspect | Recency bias | Primacy effect |
|---|---|---|
| Favours | The last items | The first items |
| Mechanism | Short-term memory | Rehearsal |
| Survives delay | No | Yes |
| Best for | Immediate decisions | Later ones |
| Use it for | Immediate decisions | Lasting impressions |
The two ends of the same U-shaped curve, differing in mechanism and durability. Primacy is attributed to rehearsal into longer-term memory and survives a delay; recency reflects still-active short-term memory and collapses once you're distracted. What you say first shapes the lasting impression; what you say last shapes the impression in the room.
| Aspect | Recency bias | Availability heuristic |
|---|---|---|
| Scope | Narrow: about order and time | Broad: about ease of recall |
| Silent rule | 'What came last matters most' | 'If I can think of examples, it must be common' |
| Example | Judging a whole year on the last month | Overrating the risk of plane crashes |
| Named by | No single coiner | Tversky and Kahneman, 1973 |
The availability heuristic is the broad shortcut: judging how common or likely something is by how easily examples come to mind, as Tversky and Kahneman described in 1973. Recency bias is a narrower slant about time: giving the latest information more weight than everything before it. Recent events are usually easy to recall, so the two often push the same way. But availability also covers what is vivid, dramatic or heavily reported, however old, and recency bias covers order effects that have nothing to do with memorability.
Opposite of Recency Bias
What it doesn't mean
The opposite: recency is the fragile one. It's wiped out by a filled delay of as little as 15 to 30 seconds, while the primacy advantage survives untouched. It dominates immediate recall and then decays fastest.
It shifts moral verdicts. Participants gave harsher punishment ratings when harmful intent appeared at the end of a scenario and milder ones when a neutral outcome came last — with order effects strongest exactly where judgement rather than a rule is required, such as jury decisions.
What it really means
Give someone a list of twenty words and ask them to recall as many as they can, and the results form a U. The first few come back well. The last few come back best of all. The middle is a wasteland.
The last-few advantage is the recency effect, and Glanzer and Cunitz found the detail that makes it useful in 1966. Insert a short distraction between the list and the recall — count backwards from a hundred for thirty seconds — and the recency advantage vanishes completely. The primacy advantage survives untouched. Two different mechanisms, sitting at two ends of one curve: primacy is material rehearsed into longer-term storage; recency is material still sitting in the mental equivalent of an open window.
Which means recency bias is dangerous in a very specific way: strongest immediately, gone soonest. The last candidate interviewed is vivid while you're deciding, and by next month you won't remember why. The same asymmetry runs through annual reviews, where six weeks of November crowd out eight months of everything else.
And it isn't confined to word lists. In studies of moral judgement, punishment ratings moved depending on whether harmful intent appeared at the beginning or the end of the same scenario — with order effects strongest where a judgement rather than a rule is required, which is a fair description of a jury.
Where it comes from
Latin — fresh, new
The root literally means 'fresh' — which is exactly what the bias mistakes for important.
- 1962
Bennet Murdock maps the serial position curve in detail, finding a recency effect extending across roughly the last eight words of a list.
- 1966
Glanzer and Cunitz establish the crucial boundary: insert a delay with a distractor task between study and recall, and the recency effect disappears while primacy survives. This dissociation made recency a short-term memory phenomenon rather than a general 'last is best' rule.
In real life
- An annual review leans heavily on the last six weeks, so a strong spring and a weak December add up to a disappointing year.
- A football club wins its last three matches and the pundits talk about a title challenge, forgetting the ten games before.
- After two warm days in March, someone decides spring has arrived for good and puts the winter coats away.
Questions
Q01Why is the last candidate interviewed often rated highest?
Because their answers are still in active memory when you score the panel, while candidate one has faded to a general impression. The standard defence is structural: score each candidate immediately after their interview against a fixed rubric, before the next starts, so you're comparing written scores rather than freshness of recall.
Q02Is recency bias why investors chase last year's winners?
A large part of it. A fund's recent three-year run is vivid and its fifteen-year record is abstract, so the recent stretch dominates. This is a real-money version of overweighting the tail of a sequence, and it's why long-horizon figures are required in performance disclosures. Nothing here is investment advice — the point is only about how sequences get weighted.
Q03How do I counter it when reviewing someone's year?
Keep a running log rather than reconstructing from memory in December. Without notes, an annual review disproportionately describes the last six weeks — a strong November can outshine eight solid months, and a single late stumble can erase them.
Sources
- Serial position, output order, and list length effects for words presented on smartphones — PMC.
- Leloup, L., Meert, G. & Samson, D., 'Moral Judgments Depend on Information Presentation: Evidence for Recency and Transfer Effects' — Psychologica Belgica 58(1), 2018, via PMC.
- recens (etymology) — Wiktionary.
- Recency bias — Wikipedia.
- Availability heuristic — Wikipedia.