Learn it → Apply it → Save as Skill
People remember the beginning and the end of almost any sequence far better than the middle. Whatever you put first tends to earn real rehearsal and settle into durable memory. Whatever you put last is still sitting in short-term memory when the listener is asked to recall anything. Everything in between competes for attention it rarely gets, and quietly falls out of what people can report back later, even when every item was, on paper, equally important.
The Dr. Wallace’s Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
Read someone a list, show them a slide deck, or introduce them to eight people in a row, and a predictable, U-shaped pattern shows up in what they can recall afterward. The first few items get rehearsed while there is little else competing for attention, which tends to push them into long-term memory. The last few items are still sitting in short-term memory at the moment of recall. The middle items get neither advantage: not enough rehearsal to stick, and no longer fresh enough to be sitting there waiting.
Important: this is Glanzer and Cunitz's actual 1966 finding, described at the level their published result supports, not an invented set of percentages. Inserting a 30-second backward-counting task between the end of a list and recall makes the recency effect disappear almost entirely, while leaving the primacy effect fully intact. That single manipulation is strong evidence the two effects come from two different memory systems, not one.
The model is most useful the moment you are deciding what order to present information, candidates, or options in, and you want to know, in advance, which position in that order will quietly disadvantage whatever lands there.
The goal is not to cram everything into the first and last slot. It is to notice, on purpose, which position each item of a sequence will land in, and make sure the middle is not quietly absorbing your most important material by accident.
List out the actual order: candidates, agenda items, slides, or options, exactly as a listener, reader, or interviewer will experience them.
Which items currently sit away from the start and the end, in the position most likely to be remembered the least, regardless of their actual importance?
Can the item you most need remembered move to the start, the end, or both, without misrepresenting its real priority?
For anything that must stay in the middle, what extra signal, repetition, a pause, a visual, a written summary, can give it the attention position alone will not?
Open a use case to see what the model changes.
Audiences remember the opening frame and the closing ask far better than whatever sits in the middle third of a deck. The strongest point, the number that should stick, or the single objection you most need to pre-empt belongs near the start or the end, not buried where attention is weakest.
Interviewers tend to remember the first and last candidates of the day noticeably better than whoever was scheduled in the middle, independent of actual performance. Deliberately varying interview order across panelists, or scoring immediately after each interview rather than at the end of the day, reduces this purely positional advantage.
Long agendas suffer the same curve: items in the middle get the least attention and the weakest recall afterward. Moving the item that truly needs a decision earlier, or explicitly flagging it before the meeting, protects it from simply disappearing into the middle.
Designers who understand this model deliberately place the items they most want noticed, premium dishes, featured products, the highest-margin option, near the top and bottom of a list, and often visually separate anything is stuck in the middle rather than leaving it to compete on position alone.
Witnesses reconstructing a sequence of events from memory typically recall the opening and closing moments with more confidence and detail than what happened in between, independent of how important the middle events actually were. Investigators and attorneys who know this probe the middle of a timeline deliberately rather than treating thin middle recall as evidence that nothing happened there.
Speakers who front-load a strong opening and close with a strong, memorable ending are working with this effect rather than against it. The weakest material, not the strongest, is what should be allowed to sit in the middle, since that is the position memory protects the least.
This is an illustrative example designed to teach the method.
Expand each item for the research behind the model and what it does not claim.
In the 1880s, German psychologist Hermann Ebbinghaus ran extensive self-experiments memorizing lists of nonsense syllables and tracking exactly how well he could recall items by their position in the list. His work was among the first to document that position in a sequence, not just the content of an item, systematically affects how well it is remembered, decades before the term "serial position effect" was coined.
Bennet Murdock tested 103 college students using word lists ranging from 10 to 40 words, presented at either one word per second or one word every two seconds, then asked participants to recall the words immediately, in any order. The results showed the now-classic U-shaped curve: words from the beginning of the list (primacy) and the end of the list (recency) were recalled noticeably more often than words from the middle. The recency advantage held consistently across roughly the final eight serial positions, regardless of how long the overall list was.
Glanzer and Cunitz, in 1966, had participants complete a brief counting-backward task for about 30 seconds between the end of a word list and the recall test. That short delay was enough to make the recency effect disappear almost entirely, while the primacy effect remained fully intact. This is strong evidence the two effects come from two different memory systems: primacy reflects items that have already been rehearsed into durable long-term memory, while recency reflects items still sitting in a fragile, limited-capacity short-term store that a brief distraction can empty.
It does not claim that simply repeating a list in the same order, over and over, is enough to overcome the effect; research shows repetition alone does not eliminate it. It does not claim the effect is specific to one language or culture; bilingual speakers show the same pattern in both of their languages. And some later studies of delayed or long-term recall have found recency-like advantages re-emerging under conditions the simple two-store explanation cannot fully account for, which is why many memory researchers now treat the clean primacy-is-long-term, recency-is-short-term story as a strong first approximation rather than the final word on the mechanism.
A candidate, argument, or agenda item remembered poorly because of its position is not the same as a candidate, argument, or agenda item that actually performed poorly. Position and quality are two separate variables that get conflated constantly.
Overcorrecting by front- and back-loading every priority creates its own problems: a sequence with no credible middle, and a listener who starts discounting whatever you place at the edges because it happens so often.
Reading or presenting the same list in the same order again does not meaningfully improve middle-item recall. The fix is changing position, adding a distinct retrieval cue, or scoring and capturing judgments immediately, not simply repeating the sequence.
Waiting until the end of a full day of interviews, pitches, or presentations to form a judgment relies on memory that has already been reshaped by position, rather than on an accurate read of what actually happened.
A short break, a side conversation, or a different task inserted between the end of a sequence and the decision point can wipe out the recency advantage specifically, while leaving the primacy advantage untouched, skewing the comparison further.
Blaming "the middle always gets forgotten" for a confusing, poorly structured, or overly long list lets a real design problem hide behind a real memory effect. The two issues call for different fixes.
a list of two or three items has barely any "middle" for the effect to act on, so reordering rarely changes much.
if each item is scored, decided on, or written down right after it is presented, rather than recalled later from memory, the effect has little room to operate in the first place.
fatigue, genuine disinterest, unclear content, or simply not paying attention are separate, well-documented explanations for forgetting that have nothing to do with position in a sequence.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply the Serial Position Effect, what questions to ask, what evidence to seek, what mistakes to avoid, and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
The pattern is extremely well replicated for free recall of lists, but its strength varies with list length, presentation speed, how meaningful or distinct each item is, and whether judgments are captured immediately or only recalled later. A short, highly distinct sequence may show little of it; a long, repetitive one often shows it strongly.
Murdock's 1962 work found the recency advantage held across roughly the final eight positions regardless of overall list length, which means in a longer list, a larger share of items fall into the unprotected middle. A 30-item list has far more "middle" to lose than a 6-item one.
Capturing a judgment, a score, a note, a decision, immediately after each item rather than waiting until the full sequence is finished and relying on memory to compare everything at once. This single change removes most of the advantage or disadvantage that position alone would otherwise create.