Serial position effect
Claim assessed: From a list we remember the last and the first items best.
Published · how we work
In short
The serial position effect means that, after you hear or read a list once, you remember the items at the end and the beginning better than those in the middle. The advantage of the last items is called the recency effect, and that of the first, the primacy effect. The two seem to have different causes: the last items are still “at hand” in short-term memory, while the first ones were rehearsed more in your head.
What it says
The classic curve. Murdock (1962) read students lists of common words chosen at random, from 10 to 40 words long, at 1 or 2 seconds per word. After each list, the students had a minute and a half to write down all the words they remembered, in any order. 103 students took part, each with 80 lists. All six versions gave the same shape: a strong rise for the last words, a flat stretch in the middle and a smaller but steeper rise for the first ones. Murdock estimated that the recency effect spans the last eight positions and the primacy effect the first three or four, regardless of list length and speed.
The last ones vanish, the first ones stay. Glanzer and Cunitz (1966) had participants do an arithmetic task between the end of the list and recall. The task removed the recency effect but barely touched the primacy effect. The result was taken as evidence that the last words are recalled from short-term memory, which an extra task empties. We haven’t read the study; we describe it from Rundus and Atkinson.
Why the first ones. Rundus and Atkinson (1970) asked eight students to rehearse aloud, while learning a list of 20 nouns, any words they wanted from the list. They recorded the rehearsals. The first words were rehearsed many more times than the others, and a word was recalled more often the more it had been rehearsed. Words still being rehearsed just before the test were recalled 92% of the time. The authors conclude that primacy comes from the extra rehearsal of the first words, and recency from the last ones still being in mind at the test.
Outside the lab. Bestue et al. (2020) followed 117 spectators of a magic show with eight tricks. Some saw the tricks in one order, others in the reverse order. Right after the show, the last trick seen was recalled more often, whichever it was. After 10 days, 1.5 months and 4.5 months, this advantage no longer appeared. What was remembered in the long run depended mostly on how memorable the trick itself was. No advantage appeared for the first trick seen.
Example
At a meeting, ten proposals are presented one after another. At the end, everyone mostly discusses the last two, which are fresh, and the first, which opened the discussion. Proposals five and six, perhaps the best ones, are almost forgotten. The example is an illustration, not a study cited here.
How to apply it
The steps below are a practical way to apply it that we propose, based on the studies cited.
- Put what matters most at the start. The first items get more rehearsal, and in the lab their advantage doesn’t disappear after a short delay, as that of the last items does.
- Don’t count on the last-place advantage if the decision comes later. The recency effect lasts while the list is still in mind. A break filled with something else erases it.
- When you learn a list, give the middle extra attention. That is where most is lost. Rehearsing and testing yourself later helps more than a single pass.
- When you compare options presented in a row, write them all down. That way you don’t judge them by how easily they come to mind.
Limits and nuances
- The curve depends on conditions. Murdock notes that another study from that period had found a primacy effect stronger than the recency effect, and that instructions, such as being asked to recall the words in order, change the shape of the curve. In Rundus and Atkinson, where participants rehearsed aloud, primacy was more pronounced than usual.
- The two-memory explanation is debated. Recency sometimes appears in the long term too, under conditions that rule out short-term memory. Davelaar et al. (2005) review the arguments and propose that immediate and long-term recency have different properties, so two memory components are needed to explain them.
- Most studies use lists of unrelated words. In real life, content matters: in the magic study, how memorable a trick was mattered more than its position, except for the last one, right after the show.
- The entry covers recall only. Order effects in judgements, for example in contests or interviews, are a separate subject that we haven’t researched here.
- What we read. Murdock (1962), from the copy used in a Dartmouth College course, we read in full, and we computed the curve from the raw data published by the same lab. Rundus and Atkinson we read in full in the 1969 Stanford technical report version, from the University of California’s eScholarship archive. Bestue et al. (2020), from PubMed Central, we read in full. For Davelaar et al. (2005) we read the abstract. We haven’t read Glanzer and Cunitz (1966); we describe it from Rundus and Atkinson.
Sources
- Bennet B. Murdock (1962). The serial position effect of free recall
- Murray Glanzer, Anita R. Cunitz (1966). Two storage mechanisms in free recall
- Dewey Rundus, Richard C. Atkinson (1970). Rehearsal processes in free recall: A procedure for direct observation
- Eddy J. Davelaar, Yonatan Goshen-Gottstein, Amir Ashkenazi, Henk J. Haarmann, Marius Usher (2005). The demise of short-term memory revisited: Empirical and computational investigations of recency effects
- David Bestue, Luis M. Martínez, Alex Gomez-Marin, Miguel A. Gea, Jordi Camí (2020). Long-term memory of real-world episodes is independent of recency effects: Magic tricks as ecological tasks
See also: Availability heuristic, Peak-end rule, Spacing and retrieval practice, Zeigarnik effect