The Serial Position Effect: Why You Remember Firsts and Lasts

Try this. Have someone read you sixteen unrelated words at a steady pace, then write down as many as you can in any order. Almost everyone produces the same shape of result. The first few words are there. The last few are there, often first on the page. The middle is a fog with two or three survivors.

That shape has a name, the serial position curve, and it is one of the most reliable findings in the study of memory. It also has unusually direct implications for how you should arrange a study session, which is why I keep returning to it.

Two effects, one curve

The curve is the sum of two separate advantages.

• Primacy: items at the start of a list are recalled better than items in the middle.
• Recency: items at the end of a list are recalled better still, and usually best of all when recall happens immediately.

Hermann Ebbinghaus, working on himself with nonsense syllables in the 1880s, already observed that position within a series mattered. The modern mapping came from Bennet Murdock's 1962 free-recall studies, which varied list length and presentation rate and produced the characteristic U across conditions. Later work by Glanzer and Cunitz added the detail that makes the whole thing interpretable: if you insert a filled delay between the last item and recall, asking people to count backwards for half a minute, the recency advantage largely disappears while the primacy advantage survives.

That dissociation is the reason the curve is usually read as evidence of two contributions rather than one. Something short-lived is holding the last items, and something more durable has already captured the first ones.

Why the ends win

The standard account of primacy is rehearsal. When the first item arrives there is nothing competing with it, so you repeat it to yourself. The second item gets rehearsed alongside the first. By item eight you are trying to juggle a growing set while new material keeps arriving, so each individual item receives a smaller share of attention. Early items simply get more processing, and more processing means a better chance of being encoded durably.

The standard account of recency is availability. The last items were presented seconds ago and are still active, so recalling them takes little more than reading off what is currently in mind. Because that store is fragile and easily displaced, anything that occupies you between presentation and recall erases the advantage.

Both accounts point at the same underlying constraint. Working memory is small and attention is finite, which is the theme we develop in cognitive load explained. Middle items are not intrinsically less memorable. They arrive at the worst possible moment, when rehearsal capacity is already committed and there is still enough new material coming to push them out of the active store.

What changes the curve

The fourth row is the most encouraging one. The curve is at its most brutal with arbitrary lists. Give the material a narrative or a causal spine and the middle stops being a void, because each item is now cued by the one before it. That is a large part of why stories are easier to remember than bullet points.

How to use this when you study

1. 1Front-load or back-load what matters. Put the definition you must not forget in the first minute or the last minute, never at minute seventeen of twenty-five.
2. 2Make more edges. Four eight-minute sessions have eight positions of advantage. One thirty-two-minute session has two.
3. 3Protect the ending. Do not fill the last minute with admin. Close by restating the one thing you want to keep, then stop.
4. 4Reorder on review. Shuffle the sequence between passes so the same unlucky items do not sit in the middle every time.
5. 5Give the middle a structure. Turn a list into a chain of causes, a timeline or a story so that middle items inherit cues from their neighbours.

The second point is the one people underuse. Splitting a session is not a compromise for busy people, it is a change to the shape of the recall curve. You are converting a long stretch of low-advantage middle into several short stretches that are almost entirely edge.

Position effects and forgetting are different problems

It is worth keeping two things apart. The serial position effect is about where material sat within a session. Forgetting is about what happens to it over the following days, which is a separate curve entirely and the reason spacing exists at all. We go through that one in the forgetting curve explained.

The two interact usefully. Position determines what gets encoded well enough to be worth reviewing, and spacing determines whether it survives the week. A well-placed item that is never revisited still fades. An item you revisit at good intervals but only ever encountered in a saggy middle had a weak start to recover from.

Why short lessons dodge the problem

Follow the logic to its end and you arrive at a structural argument for very short learning sessions. If the middle of a session is where memory goes to die, then a session with almost no middle has almost no such territory. A two-minute lesson is a beginning, a small amount of content and an ending. Everything in it is near an edge.

Disclosure, since this is where I mention our own product: MindSnap is our app. It was built around short daily lessons for exactly this family of reasons, with five flagship story collections plus unlimited further topics you can browse from topics, and each lesson ends on a question rather than trailing off, which puts a retrieval attempt at the strongest position on the curve.

I would make the same recommendation to someone using paper flashcards or a library book. Keep the unit short. Put the important thing at an edge. Break the long middle into several short middles. The curve has been the same shape since Ebbinghaus was reading nonsense syllables to himself in the 1880s, and it is not going to change to suit your schedule.