Serial-Position Effect¶

The memory phenomenon in which items in a sequence are recalled in a U-shaped curve — high for the beginning (primacy, from extra rehearsal) and end (recency, from the transient buffer), sagging in the middle — with the two limbs mechanistically dissociable.

Core Idea¶

The serial-position effect is the memory phenomenon in which items presented sequentially are recalled at rates that depend systematically on their position in the list, producing a characteristic U-shaped curve: items from the beginning of the list (primacy) and items from the end (recency) are recalled at reliably higher rates than items from the middle. The two ends of the U arise from mechanistically distinct processes. Primacy reflects the advantage that early items receive from extra rehearsal — they can be rehearsed repeatedly before subsequent items fill the available short-term buffer, producing stronger consolidation into long-term memory. Recency reflects the advantage that late items hold by still being resident in the short-term buffer at the time of recall, available without consolidation. These two mechanisms can be selectively abolished: inserting a distractor task between list presentation and recall eliminates the recency advantage (the short-term buffer is cleared) while leaving primacy intact; increasing the presentation rate eliminates primacy (early items no longer receive extra rehearsal) while leaving recency intact. The mechanistic dissociation, documented systematically by Murdock (1962) among others, was a primary source of evidence for the two-store model of memory — the distinction between a fast-decay short-term store and a slower-access but more durable long-term store that eventually became central to working-memory architecture. In applied contexts, the curve predicts that middle content in a sequential presentation — a lecture, a list of candidates, a legal argument — will be retained at lower rates than opening or closing content regardless of its importance, which has direct design implications: load-bearing material placed in the middle of a sequence suffers a structural retention disadvantage that repetition, explicit recap, or repositioning to opening and closing slots can counteract. This pattern appears reliably in free-recall studies of word lists, in mock-jury research on opening and closing statements, in ballot-order effects on candidate vote shares, and in interview-sequence ratings of sequentially-seen candidates.

Structural Signature¶

Sig role-phrases:

the sequential presentation — a list of comparable items shown one after another, position defined by order of arrival
the dual stores — a transient short-term buffer (fast-decay) and a durable long-term store (slower-access), the two-store architecture the effect rests on
the rehearsal window — the opportunity early items get to be rehearsed before the buffer fills, consolidating them into the durable store
the retrieval probe — the recall test, whose timing relative to presentation determines whether the buffer still holds the late items
the primacy limb — the recall advantage of early items, sourced from extra rehearsal and consolidation
the recency limb — the recall advantage of late items, sourced from still residing in the transient buffer at probe
the U-shaped curve — the characteristic retention-versus-position profile: high ends, sagging middle, summarized by two scalars (primacy strength, recency strength)
the selective dissociation — each limb answers to a different lever (a filled delay abolishes recency; fast presentation abolishes primacy), so recency is bounded to immediate recall

What It Is Not¶

Not a single mechanism. The two ends of the U are mechanistically distinct: primacy is rehearsal-and-consolidation into the durable store; recency is residence in the transient buffer at probe. They are dissociable — a filled delay abolishes recency while sparing primacy, fast presentation weakens primacy while sparing recency — so the curve is two processes, not one, and that double dissociation is its whole diagnostic value.
Not endpoint advantage earned by importance. Position, not content or emphasis, governs the curve. Load-bearing material placed in the middle suffers the sag regardless of how important it is; the penalty is structural and unearned, which is precisely why a presenter who orders by logic alone loses the middle.
Not a stable property of the items. The recency limb requires the probe to arrive while the transient buffer still holds the late items. Insert a filled delay before recall and recency vanishes, leaving a primacy-dominated curve — so the characteristic U is a property of immediate sequential recall, not a permanent ranking of the positions.
Not anchoring or recency bias in judgment. Anchoring is first information disproportionately shaping a later judgment; recency bias is over-weighting recent evidence in a decision. The serial-position effect is about recall of sequentially presented items — a memory output, not a judgment-weighting one, even though primacy and recency share names with those biases.
Not the peak-end rule. The peak-end rule concerns retrospective evaluation of an experience, dominated by its most intense moment and its ending. The serial-position effect concerns memory for a list of items by input position; both emphasize ends, but one is about how an episode is judged and the other about which items are retrieved.

Scope of Application¶

The serial-position effect lives across cognitive psychology and the applied design fields that rest on the same human-memory substrate — a dual-store architecture plus the rehearsal opportunities of sequential presentation; its reach stays within that human-audience substrate, since any cross-substrate version would belong to a thin, still-speculative endpoint-salience pattern, not to this effect.

List-learning research — the home turf: the canonical free-recall U across word lists, picture sequences, and digit strings, with Murdock's distractor-and-delay dissociation that evidenced the two-store model.
Witness and jury research — opening and closing statements remembered better than middle testimony, shaping how arguments are sequenced.
Ballot and survey design — position-based vote shares for first- and last-listed candidates, and order effects in question sequences.
Hiring and interviewing — first- and last-seen candidates in a day of interviews recalled more vividly and rated differently from middle ones.
Presentations and pitch decks — the durability of the opening framing and the final ask over the sagging middle, with load-bearing content repositioned or recapped.

Clarity¶

Naming the serial-position effect makes legible a gap that uniform-effort intuition hides: that even attention by a presenter yields uneven retention by an audience. Position, not just content or emphasis, governs what survives, so the curve reveals that load-bearing material placed in the middle of a sequence carries a structural retention penalty it did not earn — a fact invisible to anyone who assumes importance translates into memorability. This lets a designer of any sequential presentation — lecture, argument, candidate slate — ask where the durable slots are and place consequentially, rather than ordering by logic alone and losing the middle.

The deeper clarity is mechanistic. Because primacy and recency arise from distinct stores — early items advantaged by rehearsal and consolidation, late items advantaged by still residing in a transient buffer — the two ends of the curve are dissociable, and that dissociability turns the curve's shape into a diagnostic instrument. A researcher can read which store is operative from how the curve responds to manipulation: insert a filled delay and the recency limb collapses while primacy holds (the recency component was short-term); speed presentation past the rehearsal window and primacy weakens while recency holds (primacy was rehearsal-driven). The sharper question the effect licenses is therefore not "do endpoints get remembered better?" but "which memory store is producing this end of the curve, and what manipulation would selectively abolish it?" — the reasoning that made serial position a primary evidence base for the two-store architecture of memory.

Manages Complexity¶

Raw recall data from a sequential presentation is, in principle, a separate retention number for every position — twenty items, twenty probabilities, and no obvious reason any one should resemble another. The serial-position effect collapses that position-by-position table into a single shape, the U, parameterized by just two scalars: the strength of the primacy limb and the strength of the recency limb. An analyst no longer reasons about each slot but reads the whole curve off those two quantities and the list length between them, knowing the middle will sag and the two ends will rise. That alone turns a thicket of per-position data into a two-number summary that predicts the qualitative profile across free-recall lists, juror exposure to opening and closing statements, ballot order, and a day of sequential interviews without re-measuring each.

The compression is more than descriptive because the two scalars are mechanistically anchored to distinct stores, which gives the curve a branch structure the analyst can drive with manipulations. Primacy is rehearsal-and-consolidation into the durable store; recency is residence in the transient buffer at the moment of probe. So each limb responds to a different lever, and which lever moves which limb is fixed in advance: insert a filled delay before recall and the recency limb collapses while primacy holds (the transient buffer was cleared); speed presentation past the rehearsal window and primacy weakens while recency holds (early items lost their extra rehearsal). The analyst therefore tracks two things — rehearsal opportunity for early items and probe timing relative to presentation for late ones — and reads off not just the curve's shape but how it will deform under any given change to the presentation. This is what lets the same effect serve as a diagnostic of memory architecture (the dissociation that evidenced the two-store model) and as a design rule (durable slots are the ends; load-bearing middle content carries a structural penalty that repetition or repositioning offsets): one U-shaped regularity, two mechanistically separable parameters, and a definite prediction for how each end answers to manipulation — in place of an unstructured position-by-position retention map re-derived per presentation.

Abstract Reasoning¶

The serial-position effect supports a dissociation diagnostic that reasons from how the recall curve deforms under a manipulation back to which memory store produced a given limb. The signature inference is a double dissociation read off the curve's shape: observe that inserting a filled distractor task between presentation and recall collapses the recency limb while leaving primacy intact, and infer that recency was residence in a transient buffer the distractor cleared; observe that speeding presentation past the rehearsal window weakens primacy while leaving recency intact, and infer that primacy was rehearsal-and-consolidation into a durable store. The reasoning is which manipulation selectively abolishes which end — and because the two ends answer to different levers, an experimenter can target one component without touching the other and use the selective collapse as evidence for two stores rather than one. This is the move that made the U-curve a primary evidence base for the two-store architecture: not "endpoints are remembered better" but "the endpoints are mechanistically separable, and here is the manipulation that pulls them apart."

The interventionist and predictive moves reason in the design direction, from a planned presentation to the retention profile its ordering will produce. Given a sequence — a lecture, a slate of candidates, an argument, a ballot — the analyst predicts a sagging middle and rising ends regardless of content importance, and reasons that load-bearing material placed in the middle carries a structural retention penalty it did not earn. The corrective move follows directly: reposition the consequential item to an opening or closing slot, or, where it must stay in the middle, counteract the penalty by repetition or explicit recap — each a prediction that retention of that item should rise. The two limbs also suggest targeted design levers, since each is driven by a different cause: to strengthen the front of the curve, protect early items' rehearsal opportunity (slow the pace, pause after openers); to strengthen the back, shorten the gap between the closing item and the moment of recall. The boundary-drawing move fixes where the curve applies and where its ends decouple: the recency limb requires that the probe arrive while the transient buffer still holds the late items, so any filled delay before recall predicts recency should vanish while primacy survives — meaning the effect's characteristic U is a property of immediate sequential recall, and a delayed test predicts a different, primacy-dominated shape. Reasoning from probe timing alone, the analyst forecasts which limb will be present: short delay, both ends; filled delay, the front end only.

Knowledge Transfer¶

Within cognitive psychology the effect transfers as mechanism, across every setting that rests on the same human-memory substrate — a dual-store architecture plus the rehearsal opportunities afforded by sequential presentation. Both faces of the concept carry intact: the dissociation diagnostic (read which store produced a limb from how the curve deforms under manipulation) and the design rule (the ends are the durable slots; load-bearing middle content carries a structural penalty offset by repetition or repositioning). In list-learning research it is the canonical free-recall U across words, pictures, and digit strings, and Murdock's distractor-and-delay dissociation that evidenced the two-store model. In witness and jury research it is opening and closing statements outweighing middle testimony. In ballot and survey design it is position-based vote shares for first- and last-listed candidates. In hiring it is the vividness of first- and last-seen interviewees. In presentations it is the durability of the framing and the final ask over the middle. Across all of these the two-scalar summary (primacy strength, recency strength) and the two levers (rehearsal opportunity for early items, probe timing for late ones) move without translation — because in every case the substrate is a human listener or reader with the same memory architecture.

Beyond that substrate the honest reading is a boundary case, not a transfer — and the entry is unusually firm about it. The cross-domain reach across ballots, interviews, juries, menus, and pitches is genuine, but it is the same mechanism operating on human listeners and readers in each case, not the pattern recurring in structurally distinct substrates. Stripped of psychology vocabulary the concept becomes "human memory for a sequence shows endpoint advantages" — a specific empirical fact about the cognitive system, not a portable structural template that would apply in physics, ecology, or formal computation. And the applied interventions ("put the important thing first or last," recap the middle) are tactics, not a structural move that travels. So unlike entries that hand a clean parent prime to cross-domain reasoning, this one does not: the most that might recur across genuinely different substrates is a thin, still-speculative endpoint-salience pattern (a more general perceptual-weighting tendency for the ends of any ordered set to draw extra weight), which would need its own evidence base and is not an established prime. Until such a parent exists, the cross-domain lesson the serial-position curve seems to offer is carried by nothing more specific than that vague endpoint-weighting intuition; the curve's own machinery — the two-store dissociation, the rehearsal-vs-buffer decomposition, the immediate-recall boundary on recency — stays home-bound. Invoking "the serial-position effect" for a non-memory ordered system (items in a pipeline, positions in a schedule) is therefore (A) analogy: it borrows the U-shape while dropping the dual-store mechanism that produces and explains it. The discipline is to treat the design heuristic as a human-audience tactic, to reserve "serial-position effect" for memory of a sequence, and to reach for a general endpoint-salience pattern — not this effect — if and when a cross-substrate version is wanted (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

Murdock's (1962) free-recall study is the systematic demonstration. Participants heard lists of unrelated words and then recalled them in any order; plotting recall against input position produced the signature U — the first few and last few words recalled far better than the middle. The mechanistic proof came from Glanzer and Cunitz (1966): interposing a brief filled distractor task (counting backward) between the last word and recall selectively wiped out the recency limb while leaving primacy untouched, whereas slowing or speeding presentation moved primacy without touching recency. The two ends of the same curve could be knocked out independently.

Mapped back: the word list is the sequential presentation, its positions the input order. The early-item advantage is the primacy limb (extra rehearsal into the durable store), the late-item advantage the recency limb (still resident in the transient buffer at the retrieval probe). That a filled delay abolishes only recency is the selective dissociation, evidence for the dual stores behind the U-shaped curve.

Applied / In Practice¶

Ballot-order effects are the design consequence taken seriously by election administrators. Empirical studies of real elections find that candidates listed first on the ballot gain a small but measurable share of votes — often on the order of a couple of percentage points, and larger in low-information or nonpartisan races — an advantage owed to position, not merit. Because that structural bias can exceed the margin in close contests, many jurisdictions rotate candidate order across precincts or randomise it, so no candidate reliably occupies the favoured slot.

Mapped back: the candidate list is the sequential presentation, and the first-position vote boost is the primacy limb surfacing in choice — the endpoint advantage of the U-shaped curve rather than any content difference. Rotation and randomisation are the applied corrective the effect's design rule prescribes: neutralise the unearned positional penalty by repositioning, since the favoured slots are structural, not merited.

Structural Tensions¶

T1: One curve versus two mechanisms (the U's unity is a coincidence of testing condition). The serial-position effect presents as a single regularity — the U — and both its diagnostic value and its design rule trade on that one shape. But the two limbs are mechanistically independent: primacy is rehearsal-and-consolidation into a durable store, recency is residence in a transient buffer, and they answer to opposite manipulations. The U is not a unified phenomenon but two processes that merely co-occur in immediate free recall; change the probe timing and the shape decomposes — a filled delay yields a primacy-only curve with no U at all. So "the serial-position effect" names a composite that looks like one thing only under one condition, and reasoning about it as a single endpoint law obscures that its two ends have nothing mechanistic in common beyond both raising recall. Diagnostic: Is the U being treated as one phenomenon, when the observed shape is actually two dissociable processes whose co-occurrence depends on the recall being immediate?

T2: Fragile recency versus durable primacy (the two "good slots" are not equally good). Both ends of the curve rise, inviting the symmetric advice "put important material first or last." But the ends are not equivalent in durability: recency evaporates under any filled delay before recall, while primacy has consolidated into long-term memory and persists. So the closing slot's advantage is real only when the probe arrives immediately; whenever the outcome is deferred — jurors who deliberate hours after the closing statement, a hiring decision made the next week — the recency boost is gone and only the opening (primacy) genuinely lasts. The design heuristic that treats first and last as interchangeably durable is right for immediate recall and wrong for delayed decisions, and the designer often cannot control the gap between presentation and the moment that matters. Diagnostic: Will the decision be made while the transient buffer still holds the late items (recency counts) or after a delay (only primacy survives)?

T3: Positional durability versus logical order (repositioning fights the sequence's own reasons, and the ends are scarce). The design rule says move load-bearing content to the ends, since the middle carries an unearned structural penalty. But sequences usually have their own order for good reasons — an argument builds premise to conclusion, a lecture respects dependencies, a narrative has a shape — and repositioning content into durable slots can break the coherence that ordering served, trading memorability for intelligibility. Worse, the fix is zero-sum: there are only two ends, so if every consequential item "belongs" at an endpoint they cannot all be placed there, and privileging one item's retention necessarily consigns another to the sag. The heuristic optimizes recall against the very structure the content's order exists to convey, and cannot be applied to more than a couple of items at once. Diagnostic: Does moving this item to an endpoint improve its retention more than it costs the sequence's logical coherence — and which other item is being pushed into the middle to make room?

T4: Diagnostic rigor versus design tactic (one name lends the tactic the lab's authority). The effect has two faces: a rigorous memory-architecture diagnostic, where the double dissociation was primary evidence for the two-store model, and an applied design rule for lectures, ballots, and pitches. The applied face borrows the prestige of the experimental one, but the real-world effects are small and noisy — ballot order shifts a couple of points, interview position nudges vividness — heavily moderated by content, motivation, and delay, and easily conflated with neighbours the effect is not (anchoring, recency bias in judgment, the peak-end rule). Holding both faces under one name lends "put it first or last" more authority than its field effect sizes warrant, and imports the clean lab curve into messy settings where many of its preconditions (immediate recall, comparable items, no strong content differences) do not hold. Diagnostic: Is the design claim resting on the effect's measured magnitude in this setting, or on the borrowed credibility of the laboratory dissociation that does not carry the applied conditions?

T5: Autonomy versus reduction (a memory phenomenon with no clean parent to reduce to). The serial-position effect is a named cognitive finding with proprietary cargo — the two-store dissociation, the rehearsal-versus-buffer decomposition, the immediate-recall boundary on recency — that transfers as literal mechanism across every human-audience setting sharing that memory architecture. Unusually, it hands cross-domain reasoning no established parent: stripped of psychology it is only "human memory for a sequence shows endpoint advantages," and the most that might recur in genuinely different substrates is a thin, still-speculative endpoint-salience pattern that is not a catalogued prime. So invoking "the serial-position effect" for a pipeline, a schedule, or any non-memory ordered system is analogy — it borrows the U-shape and drops the dual-store machinery that produces it. This tilts the resolution toward autonomy: the effect is genuinely its own named thing, because there is no good parent to which its explanatory content reduces, and its applied uses are human-audience tactics, not a portable structural move. Diagnostic: Resolve toward the named effect (and its two-store machinery) whenever a human listener's memory for a sequence is at issue; treat any non-memory "serial-position" invocation as analogy to a still-unformed endpoint-salience pattern, not a transfer of this effect.

Structural–Framed Character¶

The serial-position effect sits toward the structural end of the structural–framed spectrum — best read as mixed-structural, a sibling of the self-reference effect: a genuine, evaluatively-neutral memory mechanism running in the mind, wearing two-store vocabulary — but an unusually autonomous one that hands cross-domain reasoning no clean parent. On evaluative_weight it scores structural: a U-shaped recall curve is neither good nor bad and renders no verdict (the design maxim "put it first or last" is a downstream tactic, not a charge the effect carries). On human_practice_bound it is largely structural: the curve arises in any mind encoding a sequence — a person recalling a word list, a juror hearing arguments — whether or not a researcher plots it; like all cognition it requires a perceiving mind, but it is not constituted by any institution. On institutional_origin it is largely structural: it names a real regularity of memory architecture (Murdock's U, the Glanzer–Cunitz dissociation), a scientific construct rather than an artifact of a tradition — though the two-store model it evidences is itself a theoretical idealization. On vocab_travels it fails: the dual stores, primacy/recency, the rehearsal window, and the transient buffer are irreducibly memory-psychology vocabulary. On import_vs_recognize it patterns strongly toward recognition within the human-memory substrate (ballots, juries, interviews are the same mechanism on human audiences), but the entry is unusually firm that beyond that substrate it hands over no established parent — invoking "serial-position effect" for a non-memory ordered system is analogy that borrows the U-shape and drops the dual-store machinery producing it.

Unusually among these entries, there is no clean portable structural skeleton to name: stripped of psychology the concept is only "human memory for a sequence shows endpoint advantages," and the most that might recur across genuinely different substrates is a thin, still-speculative endpoint-salience pattern (a general perceptual-weighting tendency for the ends of an ordered set to draw extra weight) that is not a catalogued prime and would need its own evidence base. So what the effect instantiates is the human dual-store memory architecture itself, and its distinctive machinery — the two-store dissociation, the rehearsal-versus-buffer decomposition, the immediate-recall boundary on recency — stays home-bound, tilting the entry toward autonomy: it is genuinely its own named thing because there is no good parent to which its explanatory content reduces. Its character: a real, evaluatively-neutral, unusually self-contained memory mechanism — an endpoint-advantaged recall curve produced by two dissociable stores — structural in kind but pinned to human memory architecture by two-store vocabulary and lending cross-domain reasoning only a vague endpoint-weighting intuition, not a prime.

Structural Core vs. Domain Accent¶

This section decides why the serial-position effect is a domain-specific abstraction and not a prime — an unusual case, because it hands cross-domain reasoning no clean parent prime at all, which is exactly what pins it home rather than lifts it.

What is skeletal (could lift toward a cross-domain prime). Strip the psychology and only a very thin residue survives: in an ordered set, the endpoints draw extra weight and the middle sags. That is a candidate endpoint_salience pattern — a general perceptual-weighting tendency for the ends of any ordered sequence to be favoured over its interior. But note the diagnostic honesty the entry insists on: this residue is not a catalogued prime, it is speculative, and it would need its own evidence base to become one. Unlike entries whose stripped core lands cleanly on contrast, accumulation, or attention, the serial-position effect's abstract shape is only a vague endpoint-weighting intuition. So there is barely any portable core to speak of — and what little there is does not yet exist as a structure the catalogue can carry.

What is domain-bound. Almost the entire concept is human-memory-architecture furniture that does not survive extraction — and, crucially, the two things that make the U explanatory rather than merely descriptive are the most home-bound of all. The endpoints are not one advantage but two mechanistically dissociable limbs: primacy from extra rehearsal and consolidation into a durable long-term store, and recency from residence in a transient short-term buffer at the moment of the retrieval probe. The signature evidence is a double dissociation — a filled distractor delay abolishes recency while sparing primacy; fast presentation abolishes primacy while sparing recency — which is what made the effect a primary evidence base for the two-store model. And recency is bounded to immediate recall. The decisive test: remove the human listener with a dual-store memory and there is no serial-position effect — a pipeline stage, a schedule slot, a queue position does not rehearse, does not sit in a buffer, and cannot be probed. What survives is only the bare endpoint-weighting shape; the entire dual-store machinery that produces and explains the U stays home.

Why this does not clear the prime bar. A prime's vocabulary travels and its cross-domain transfer is recognition of the same mechanism, not analogy. The serial-position effect's transfer is unusual: broad within one substrate, and essentially nil beyond it. Within the human-memory substrate it travels intact as full mechanism — the two-scalar summary (primacy strength, recency strength), the two levers (rehearsal opportunity, probe timing), the dissociation diagnostic, and the design rule apply identically across free-recall lists, jury statements, ballots, interviews, and pitch decks, because each is the same mechanism operating on human listeners and readers, not the pattern recurring in structurally distinct substrates. Beyond human audiences the named effect does not go: invoking "the serial-position effect" for a non-memory ordered system borrows the U-shape while dropping the dual-store mechanism that produces it — pure analogy. And when the bare endpoint-weighting lesson genuinely is wanted cross-substrate, there is no established parent to carry it: the most that recurs is the still-speculative endpoint_salience pattern, which is not a prime and would need its own evidence. This tilts the entry toward autonomy: it is genuinely its own named thing precisely because its explanatory content — the two-store dissociation, the rehearsal-versus-buffer decomposition, the immediate-recall boundary on recency — reduces to no good parent. The applied heuristics ("open and close with what matters," "recap the middle") are human-audience tactics, not a portable structural move. So the discipline is to reserve "serial-position effect" for a human listener's memory of a sequence, and to reach for a general endpoint-salience pattern — not this effect — if and when a cross-substrate version is ever established.

Relationships to Other Abstractions¶

Current abstraction Serial-Position Effect Domain-specific

Parents (2) — more general patterns this builds on

Serial-Position Effect is part of Primacy Effect Domain-specific

The serial-position effect contains the primacy effect as its durable beginning-of-sequence limb.
Serial-Position Effect is part of Recency Effect Domain-specific

The serial-position effect contains the recency effect as its delay-sensitive end-of-sequence limb.

Hierarchy paths (3) — routes to 2 parentless roots

Serial-Position Effect → Primacy Effect → Positional Advantage → Asymmetry

Show alternative paths (2)

Not to Be Confused With¶

Anchoring. The judgment bias in which an initial value disproportionately shapes a later estimate. It shares the word "first-item advantage" with primacy, but anchoring is about a judgment being pulled toward an anchor, whereas the serial-position effect is about recall of sequentially presented items. Tell: is early information biasing a subsequent numerical or evaluative judgment (anchoring), or are early items simply remembered better (primacy limb)?
Recency bias (in judgment/decision). The tendency to over-weight the most recent evidence when forming a decision. Serial-position recency is about late items being retrieved better because they still sit in the short-term buffer — a memory output, not a weighting of evidence in a verdict. Tell: is recent information counting for more in a decision (recency bias), or late items being recalled more often on a memory test (recency limb)?
Peak-end rule. The finding that retrospective evaluation of an experience is dominated by its most intense moment and its ending. Both emphasize ends, but the peak-end rule concerns how an episode is judged, while the serial-position effect concerns which items in a list are retrieved by input position. Tell: is the subject an overall rating of an experience driven by peak and finish (peak-end), or item-by-item recall across a sequence (serial position)?
Sequential clarity / von Restorff distinctiveness. The recall advantage that comes from an item's contrast with its immediate neighbours — a distinctive item pops out regardless of where it sits. The serial-position effect is driven by list position (the boundaries), not by neighbour contrast; a mid-list item can be well-recalled if distinctive (von Restorff) despite its poor serial position. Tell: is the advantage explained by an item's position at the list ends (serial position), or by its feature contrast with adjacent items independent of position (sequential clarity / von Restorff)?
The two-store model / working memory (the architecture, not the effect). The theoretical account of memory as a fast-decay short-term buffer plus a durable long-term store. The serial-position effect is the empirical phenomenon whose double dissociation supplied evidence for that model — it is not the model itself. Tell: is the subject the theorized dual-store architecture (two-store model), or the U-shaped recall curve that evidenced it (serial-position effect)?
Endpoint salience (the speculative "parent"). The thin, still-uncatalogued perceptual-weighting tendency for the ends of any ordered set to draw extra weight. This is at most what might recur in non-memory substrates — and it is not an established prime, nor does it carry the effect's explanatory dual-store machinery. Tell: for a non-memory ordered system (a pipeline, a schedule) the most that applies is the vague endpoint-salience intuition; "serial-position effect" applies only to human memory for a sequence. (Treated fully in an earlier section.)

Neighborhood in Abstraction Space¶

Serial-Position Effect sits in a crowded region of the domain-specific corpus (20^th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Memory Encoding & Retrieval Effects (22 abstractions)

Nearest neighbors

Primacy Effect — 0.88
Retrieval Practice — 0.86
Temporal Distinctiveness — 0.86
Retrieval Practice Effect — 0.85
Telescoping Effect — 0.85

Computed from structural-signature embeddings · 2026-07-12