What Is the Serial Position Effect? Why the Middle of a List Disappears

Plotting the probability of recalling an item against its position in the list produces a distinctive U-shaped curve. The advantage for early items is the primacy effect, and the standard explanation is that those items received more rehearsal, since a person hearing the first item has nothing else to attend to and repeats it, while later items compete for the same attention. The advantage for late items is the recency effect, and the explanation is that they are still available in a short-term store when recall begins. The middle items get neither benefit, having been rehearsed little and having been displaced from the short-term store by everything that followed. The shape is extremely robust, appearing with words, numbers, pictures, faces and events, and across ages and cultures.

The evidence that these are distinct mechanisms comes from manipulations that affect one and not the other:

• •A delay filled with a distracting task between the list and recall removes the recency advantage and leaves primacy intact
• •Presenting items faster reduces the primacy advantage, since there is less time to rehearse, and leaves recency intact
• •Making items harder to pronounce or less familiar affects primacy more than recency
• •Patients with impaired short-term memory show primacy without recency
• •Patients with impaired long-term memory show recency without primacy
• •That double dissociation is the strongest form of evidence for two separate systems

The effect operates well outside memory experiments and has practical consequences. Candidates interviewed first and last are remembered better than those in the middle, and evidence on whether that translates into better outcomes is mixed but the recall advantage is clear. Items at the start and end of a presentation, a meeting agenda or a document are retained better, which is an argument for putting what matters at the ends rather than burying it. Advertising placement within a break follows the same logic. Witnesses recall the beginning and end of an event more reliably than the middle, which matters in legal contexts. Voting order on a ballot has measurable effects. In each case the effect is modest relative to the content itself, and it is consistent enough to be worth designing around when the arrangement is under anyone's control.

The standard account has been challenged in ways worth knowing, since the textbook version is tidier than the evidence. Recency effects appear over long delays as well as short ones, with people recalling recent items better even when weeks have passed and no short-term store could plausibly be involved, which suggests a more general principle about distinctiveness in time rather than a separate storage system. On that account an item stands out when it is separated from its neighbours in relative terms, which explains both ends of the curve and the long-delay findings with one mechanism. Rehearsal accounts of the early advantage have also been questioned, since the effect survives conditions that should prevent rehearsal. The basic pattern is not in doubt, and the explanation is less settled than its presence in every introductory textbook suggests, which is a common situation for robust findings.

For anyone trying to learn material rather than exploit the effect, the implications run the other way. The middle of any study session is the weak point, so breaking a long session into several shorter ones creates more beginnings and endings and reduces the amount of material that falls into the middle. Varying the order between sessions moves different material into the advantaged positions, which is why working through a list from the same starting point repeatedly leaves the later items permanently weaker. Retrieval practice reduces the effect, since actively recalling material is not subject to the same rehearsal asymmetry. And knowing the pattern is itself useful, since a person who assumes their recall is uniform will misjudge what they know, and checking the middle specifically catches the gap that the curve predicts.

Early items are recalled well because they were rehearsed more and late items because they are still in a short-term store, which is why a filled delay removes the late advantage and faster presentation removes the early one. Patients with damage to one system or the other show one effect without the other. Breaking study into shorter sessions creates more beginnings and endings and shrinks the weak middle.