Fan Effect¶

The memory finding that verifying a stored fact about a concept gets slower and less accurate as more other facts are attached to it, because a fixed activation budget spread across a concept's associations ('fan') leaves each one less to cross the recognition threshold.

Core Idea¶

The fan effect is the cognitive-psychology finding, established by John R. Anderson (1974) and elaborated through the ACT-R memory architecture, that the time required to verify a stored fact about a concept increases — and accuracy decreases — as a function of the number of other facts stored about that same concept, where the number of associations is termed the concept's "fan." If a participant has memorised that a hippie is in the park, in the church, and in the bank, then verifying any one of those propositions is slower and less accurate than verifying a proposition about a concept that appeared in only one studied sentence, even when the target proposition is identical across conditions. The mechanism is spreading activation with a fixed activation budget per concept: when a cue activates a concept node, that node's activation spreads to its associates, but the available activation is divided across however many associates the node has. A high-fan concept distributes activation thinly, so each associate receives less and must wait longer to cross the recognition threshold against competing associations in the retrieval race. The effect is the canonical experimental signature of resource-division in declarative memory — the demonstration that knowing more facts about something can make retrieving any one fact about it slower and less reliable. The paradox of expert performance reverses this prediction in practice: domain experts show faster and more accurate retrieval despite knowing far more, which is accounted for by chunking — hierarchical integration of multiple associations into single higher-order units that collapse the effective fan. The fan effect is therefore also the theoretical foundation for predicting when knowledge organisation does and does not help: accumulated unstructured facts increase the fan and slow retrieval, while structured, chunked knowledge reduces the effective fan and restores or improves it.

Structural Signature¶

Sig role-phrases:

the declarative store — a memory of concept-fact associations indexed by cue
the concept's fan — the number of associations co-attached to a concept, the countable divisor the effect scales with
the fixed activation budget — a roughly constant pool of activation per concept that must be shared across all of its associates
the cue and spread — a retrieval cue activating the concept node, whose activation spreads divisively to its associates
the competitive retrieval race — the target association having to out-activate its co-attached competitors to cross the recognition threshold
the fan cost — more associates meaning thinner activation each, so latency and error rise with the summed fan over the cues
the effective-vs-raw distinction — chunking folding many associations into single higher-order units, collapsing the effective fan and resolving the expert-memory paradox
the chunking intervention — hierarchical organization or distinctive disambiguating cues lowering the effective fan and restoring retrieval

What It Is Not¶

Not confirmation that knowing more always helps recall. The effect's whole point overturns that folk assumption: under unstructured accumulation, knowing more facts about a concept makes retrieving any single one slower and less accurate, because the added associations share — and thin out — a fixed activation budget.
Not memory decay, weak encoding, or waning attention. The slowing is synchronic associative interference under a divided activation budget, not a fading or poorly-laid trace. That localization is what lets the fan be isolated as its own variable and counted.
Not proactive or retroactive interference. Those turn on the temporal order in which material was learned; the fan effect is competition among associations co-attached to one concept regardless of learning order. The slowing scales with how many other facts share the cued concept, not with what was learned before or after.
Not a list-length effect. List-length slowing scales with the total size of the studied set; the fan effect scales with the fan of the cued concept specifically. Two designs with identical total list length can differ in fan, and it is the fan that predicts the latency.
Not an unconditional law that more facts always slow retrieval. Experts retrieve faster despite knowing far more, because chunking folds many associations into single higher-order units, collapsing the effective fan. The effect bites only on raw, unstructured accumulation; once chunked, added knowledge restores or improves retrieval. "Does knowing more hurt?" resolves to "is the fan accumulated or chunked?"
Not any case of competition for a shared resource. A router's fan-out or a thrashing cache instantiates the parent contention/dilution pattern, not the fan effect — which presupposes cued retrieval from a declarative store, a spreading-activation architecture, and a measurable retrieval race among co-active associations. The database analogue is the parent reaching a new substrate, not the named effect with its activation-division and chunking vocabulary.

Scope of Application¶

The fan effect lives across the memory and learning subfields of cognitive psychology; its reach is within that domain, confined to cued retrieval from a declarative store where co-active associations divide a fixed activation budget. The general "more competitors for a fixed resource degrade service" pattern — which covers the database-indexing analogue the literature itself draws — belongs to its contention/dilution parents, not to this named effect.

ACT-R cognitive architecture — a foundational empirical fit for the activation-based declarative-memory module, which divides each chunk's activation across the contexts in which it appears.
Expertise and chunking research — supplies the resolution of the expert-memory paradox: experts know more yet retrieve faster because hierarchical chunking collapses many associations into single higher-order units, reducing the effective fan.
Eyewitness and reality-monitoring research — predicts that a witness with many associations to a person or scene will be slower and less accurate in recognizing any one particular detail.
Instructional design — predicts that teaching many disconnected facts without chunking produces slow, error-prone recall, providing the theoretical backing for meaningful-organization techniques.
Interference research — the synchronic member of the interference family, distinguished from temporal-order (proactive/retroactive) interference and from total-list-length effects by the variable it scales with.

Clarity¶

The fan effect makes legible a result that the folk model of memory cannot accommodate: knowing more about a concept can make retrieving any single fact about it slower and less accurate. Naming it does two things at once. It supplies a counterintuitive regularity that overturns the assumption that added knowledge can only help recall, and it localizes the cause precisely — associative interference under a divided activation budget, not memory decay, waning attention, or weak encoding. That localization is the heart of the clarity, because it separates the fan effect from neighbors that produce superficially similar slowing: proactive and retroactive interference, which turn on the temporal order in which material was learned, and list-length effects, which turn on the total size of the studied set. The fan effect is none of these; it is synchronic competition among the associations co-attached to one concept, and isolating that as its own variable — the concept's "fan" — converts a vague sense that cluttered knowledge retrieves poorly into a countable quantity that predicts reaction time.

This precision reframes the apparent paradox of expertise into a sharp, answerable question. If accumulating facts raises the fan and slows retrieval, why do experts, who know vastly more, retrieve faster? The effect forces the analyst to ask not "how much does this person know?" but "how is that knowledge organized?" — distinguishing raw association count from effective fan after chunking has folded many associations into single higher-order units. That distinction tells a researcher or instructional designer exactly when added knowledge will help and when it will hurt: unstructured accumulation widens the fan and degrades retrieval, while hierarchical chunking collapses the effective fan and restores it. The sharper diagnostic becomes whether a given body of knowledge is merely accumulated or genuinely structured.

Manages Complexity¶

A whole class of memory-performance questions — why a particular fact retrieves slowly, why one witness is reliable about a detail and another is not, why a body of taught material recalls poorly, why the same person who knows ten facts about a topic is slower on each than someone who knows one — looks, before the fan effect, like a set of separate phenomena demanding separate explanations across attention, encoding, and decay. The fan effect compresses them to a single countable parameter: the concept's fan, the number of associations sharing its fixed activation budget. Reaction time and error then scale, to first approximation, with that one number (summed over the cue concepts), so a researcher predicts retrieval performance for a novel item by counting its associations rather than modelling the encoding history of each fact. The same parameter resolves the expertise paradox without adding machinery — the apparent contradiction (experts know more yet retrieve faster) reduces to the gap between raw association count and effective fan after chunking has folded associations into higher-order units, so "does added knowledge help or hurt?" collapses to "is the fan accumulated or chunked?" And it fixes the intervention: anything that lowers the effective fan — chunking, hierarchical organization, distinctive disambiguating cues — speeds retrieval, predictably. A diffuse catalog of retrieval slowdowns and the standing puzzle of expert memory both reduce to one divisor that an analyst can count and manipulate.

Abstract Reasoning¶

The effect licenses inferences that run through one countable quantity — a concept's fan, the number of associations sharing its fixed activation budget — and the divided-resource mechanism behind it. Diagnostic: when verification of a stored fact is slow and error-prone, do not infer decay, weak encoding, or waning attention as the default; infer synchronic associative interference under a divided activation budget — the cue's activation spread thin across many co-attached associates, so each crosses the recognition threshold late in the retrieval race. The signature is specific and distinguishes the effect from its neighbors: the slowing scales with how many other facts share the cued concept, not with the temporal order of learning (proactive/retroactive interference) and not with the total studied-set size (list-length effects). From a measured reaction-time cost one reads back the hidden divisor — the fan of the cue concepts — because, to first approximation, latency and error scale with the summed fan over the cues. The diagnostic also resolves the expertise paradox by re-localizing the question: an expert who retrieves faster despite knowing more is not a counterexample but a sign that raw association count has diverged from effective fan — so the inference is not "how much is known?" but "is the knowledge accumulated or chunked?"

Interventionist: because performance is governed by the divisor, the predicted lever is anything that lowers the effective fan — hierarchical chunking that folds many associations into single higher-order units, distinctive disambiguating cues that let the target win its race, retrieval practice that strengthens one association over competitors. Each predicts faster, more accurate verification, and the prediction is quantitative: collapse N associations into one chunk and the effective fan (hence the latency cost) should drop accordingly. The mechanism equally predicts which interventions fail to help or actively hurt: adding more unstructured facts about a concept widens the fan and is predicted to slow retrieval of every fact already attached to it — so "learn more about the topic" without organization is predicted to degrade, not improve, recall of any single fact. That added-knowledge-hurts prediction, contrary to the folk model, is the effect's sharpest interventionist claim, and it is reversible exactly by chunking.

Boundary-drawing: the effect bites in the regime of unstructured accumulation — many associations co-attached to one concept, sharing the budget, none integrated — and is suppressed in the regime of structured knowledge, where chunking has collapsed the effective fan, or where strong contextual disambiguation lets one association dominate. This draws the line on when added knowledge helps versus hurts: below integration it widens the fan and degrades retrieval; once chunked it restores or improves it. The boundary also separates the fan effect from same-surface slowdowns it is not — temporal-order interference and total-list-length effects live in different regimes and respond to different manipulations. Predictive: the order is fixed and countable — more associations on a cue, thinner activation per associate, longer time to threshold, more errors — so one forecasts retrieval performance for a novel item by counting its associations rather than reconstructing each fact's encoding history, and predicts that expert advantages will be largest precisely in domains where the expert's knowledge has been chunked rather than merely amassed.

Knowledge Transfer¶

Within cognitive psychology and memory research the fan effect transfers as mechanism, with the concept's fan — the count of associations sharing a fixed activation budget — as the portable quantity. It is the foundational empirical fit for the ACT-R activation-based memory module, and across the subfields it touches the diagnostics carry intact: in interference research it is the synchronic member of the family, isolated from temporal-order (proactive/retroactive) interference and from total-list-length effects by the variable it scales with; in expertise and chunking studies it supplies the resolution of the expert-memory paradox (raw association count versus effective fan after chunking); in eyewitness and reality-monitoring work it predicts that a witness with many associations to a person or scene is slower and less accurate on any one detail; in instructional design it is the theoretical backing for meaningful-organization techniques. The vocabulary — fan, spreading activation, activation budget, effective versus raw fan, the chunking intervention — travels across these subfields without translation, and the quantitative predictions (latency and error scale with summed fan over the cues; collapse N associations into one chunk and the latency cost drops accordingly) hold throughout. The within-domain transfer is the mechanism itself, applied from list-learning to expertise to courtroom recall.

Beyond associative memory the picture is the third category, and the structural shadow is unusually clean. A genuinely more general pattern recurs across substrates — a fixed resource divided across more competing recipients yields degraded service per recipient — and it travels as the parent primes: scarcity, interference_and_contention, bandwidth_and_throughput, and dilution. Those recur as co-instances far from cognition: a shared communication channel slows as more senders contend for it, a router's fan-out forces more candidates to disambiguate per query (the database-indexing analogue the literature itself draws), a cache thrashes as more keys compete for fixed lines. That divided-resource structure genuinely repeats, and where the cross-domain lesson is needed it is already available in those primes. But the fan effect's own machinery does not ride along: it presupposes cued retrieval from a declarative memory store, a spreading-activation (or similar divisive-attention) architecture, and a measurable competition among co-active associations in a retrieval race — properties of a particular family of cognitive memory models and their biological substrate. The database fan-out is a real instance of the parent (contention over a fixed budget), but calling it "a fan effect" would import the activation-division, retrieval-race, and chunking vocabulary that belong to declarative memory and have no referent in an index — analogy at the level of shape, mechanism only at the level of the parent. Strip "fan," "association," "cue," and "retrieval" and what remains is "more competitors for a fixed resource means worse service per competitor," which is the contention/dilution prime, not the named effect. So the honest move is to let the cross-domain lesson ride on interference_and_contention / bandwidth_and_throughput / scarcity, which recur as mechanism everywhere a fixed resource is shared, and to keep "fan effect," as named, for the associative-memory corner where activation-division and chunking make its specific, countable predictions hold (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

John Anderson's 1974 experiment established the effect. Participants studied a set of simple sentences of the form "A [person] is in the [location]" — for example, "A hippie is in the park," "A hippie is in the church," "A lawyer is in the park" — where each person or location concept was engineered to appear in one, two, or three studied sentences (its fan). After memorizing the set to criterion, participants performed a speeded recognition test: shown a probe sentence, they judged whether it had been among the studied ones. The signature result was that verification time rose systematically with fan — a probe about a person who had appeared in three sentences was verified more slowly, and with more errors, than one about a person appearing in a single sentence, even though the target proposition itself was equally well learned. Retrieving one fact about a concept was degraded precisely by how many other facts shared that concept.

Mapped back: The memorized person-in-location sentences form the declarative store; the number of sentences sharing a concept is the concept's fan. The recognition probe is the cue and spread, and the target proposition must win the competitive retrieval race against its co-attached rivals. Rising latency with fan is the fan cost — thinner activation from the fixed activation budget per associate.

Applied / In Practice¶

Expert memory research shows the effect's reversal through chunking, and chess is the classic case (de Groot; Chase and Simon, 1973). Presented briefly with a real game position and asked to reconstruct it, master players vastly outperform novices — seeming to violate the fan effect, since masters "know" enormously more chess than beginners. The resolution is that experts do not store the position as ~25 independent piece-square associations (high raw fan) but recognize it as a handful of familiar configurations — pawn chains, castled-king structures, known openings — each a single higher-order chunk. Their effective fan is therefore low, so retrieval is fast and accurate. The decisive control confirms the mechanism: on randomly arranged pieces, where no chunks apply, the expert advantage largely disappears and masters perform near novice level, because the raw associations can no longer be folded.

Mapped back: The masters' vast chess knowledge would imply a crippling concept's fan if stored as raw piece-square links; instead the chunking intervention folds configurations into units, invoking the effective-vs-raw distinction that collapses effective fan and speeds the competitive retrieval race. Random boards, defeating chunking, restore high raw fan and the fan cost — the expert edge vanishing is the reversal made visible.

Structural Tensions¶

T1: Accumulation degrades versus organization restores (the folk model overturned). The effect's sharpest claim is that under unstructured accumulation, knowing more facts about a concept makes retrieving any single one slower and less accurate, because the added associations share and thin out a fixed activation budget. The tension is that this is double-edged and contingent, not a blanket law: the same act of learning more can help or hurt depending entirely on whether the knowledge is merely amassed or genuinely integrated. Below integration, added facts widen the fan and degrade retrieval; once chunked, they collapse the effective fan and restore or improve it. So "should I learn more about this?" has no context-free answer — the intervention that helps and the one that hurts are the same input distinguished only by its organization. Diagnostic: Is the added knowledge raw, unstructured accumulation (widens the fan, slows retrieval) or hierarchically chunked (collapses effective fan, speeds it)?

T2: Raw fan versus effective fan (the expertise paradox's hinge). The countable quantity is the concept's fan — the number of associations sharing the budget — yet experts retrieve faster despite knowing vastly more, which reads as a direct counterexample. The resolution is that raw association count has diverged from effective fan after chunking folds many associations into single higher-order units. The tension is that the effect's predictive power rests on a quantity that is not simply countable from the surface: two learners with the same raw association count can have wildly different effective fans, so the naive count both drives the core prediction and, taken as the operative variable, generates the paradox. The concept is only saved by insisting the load-bearing number is the one you cannot directly see. Diagnostic: Is the operative divisor the raw count of associations, or the effective fan that remains after chunking has folded them into higher-order units?

T3: Synchronic competition versus same-surface slowdowns (isolating the right variable). Slow, error-prone retrieval has several candidate causes that look alike: memory decay, weak encoding, temporal-order interference (proactive/retroactive), total-list-length effects, and the fan effect. The fan effect is specifically synchronic competition among associations co-attached to one concept, scaling with how many facts share the cued concept — not with learning order, not with total set size. The tension is that the observable symptom (slowing) underdetermines the mechanism, and the interventions differ: chunking fixes a fan problem but not a list-length problem, and two designs with identical total list length can differ in fan. Misattributing the slowdown selects the wrong lever. The effect earns its keep only by being isolated from the neighbors it superficially resembles. Diagnostic: Does the slowing scale with how many other facts share the cued concept (fan), with the order in which material was learned (temporal interference), or with the total size of the studied set (list-length)?

T4: Countable divisor versus association strength (the first-approximation limit). Latency and error scale with the summed fan over the cues — but only to first approximation. Retrieval practice that strengthens one association over its competitors, and distinctive disambiguating cues that let the target win its race, both improve retrieval without lowering the fan count: they act on relative strength, not on the number of associates. The tension is that the clean, countable predictor (how many associations) is not the whole mechanism; the retrieval race is decided by activation, which count governs only when strengths are equal. So an intervention that adds an association would ordinarily widen the fan and hurt, yet a distinctive cue "added" to the concept helps — because it changes who wins the race, not how many run in it. Diagnostic: Is retrieval here governed by how many associations share the cue (count), or by their relative strengths — one practiced or distinctively cued association dominating despite a high fan?

T5: Autonomy versus reduction (named associative-memory effect or the parent contention pattern). The "fan effect" is a named, canonically measured finding with cargo bound to declarative memory: cued retrieval from a store, a spreading-activation architecture, a measurable competition among co-active associations, and the chunking intervention. A genuinely more general pattern lies beneath it — a fixed resource divided across more competing recipients yields degraded service per recipient — and that travels as the parents interference_and_contention, bandwidth_and_throughput, scarcity, and dilution, recurring as a contended channel, a router's fan-out, a thrashing cache. The database fan-out the literature itself cites is a real instance of the parent, not the effect: calling it "a fan effect" would import activation-division and retrieval-race vocabulary with no referent in an index. Diagnostic: Resolve toward interference_and_contention / dilution when carrying the divided-resource lesson to non-cognitive substrates (routers, caches, channels); toward the fan effect when diagnosing retrieval latency in a declarative memory store where activation-division and chunking hold.

Structural–Framed Character¶

The fan effect sits at mixed-structural on the structural–framed spectrum — the cleanest structural case among the cognitive effects in this cluster, because the mechanism it names is bare resource-division rather than an interpretive or motivated distortion, yet it stops short of the structural side by being stated wholly in declarative-memory vocabulary. Four of the five criteria point structural. Its evaluative weight is nil: a concept's retrieval slowing as its fan rises is neither good nor bad, and "fan effect" convicts nothing — it is a latency law, not a verdict (unlike the relevance fallacies filed nearby, which render defect-judgments). Its institutional origin is none: the effect is a fact about how an activation-based declarative memory behaves under load, isolated by Anderson and formalized in ACT-R, not an artifact of a survey, agency, or reporting convention — Anderson counted and named a regularity minds already run. It is not human-practice-bound in the constituting sense: the fan cost fires in any memory system with a spreading-activation architecture and a fixed activation budget whether or not anyone studies it, so it does not dissolve when a human practice or institution is removed; it needs a cognitive substrate, not a tradition. And within its proper range cross-domain reuse is recognition rather than import: moving from list-learning to eyewitness recall to expert chess memory, the same activation-division mechanism is recognized intact, not borrowed as a frame. These four marks place it firmly on the structural side and make it closely analogous to how isostasy is characterized — a real, evaluatively neutral, recognized-in-nature mechanism.

What keeps it off the structural pole is the remaining criterion, vocab-travels, which it fails. The operative vocabulary — fan, spreading activation, fixed activation budget, cue, retrieval race, effective versus raw fan, chunking — is irreducibly the vocabulary of associative-memory models and does not float free of that substrate the way "growing quantity" or a divisor does in a pure structural prime; carry it to a router or a cache and every term must be renamed. The portable structural skeleton it shares — a fixed resource divided across more competing recipients yields degraded service per recipient — is genuinely and cleanly substrate-portable (a contended channel, a fan-out index, a thrashing cache all instance it), but that is exactly the part the catalog already carries as the contention/dilution primes interference_and_contention, bandwidth_and_throughput, scarcity, and dilution that the fan effect instantiates; the cross-domain reach belongs to those parents, while the activation-division, retrieval-race, and chunking machinery is the domain-accented expression that stays home. Its character: structural in skeleton — a real, evaluatively neutral, recognized-in-cognition instance of fixed-resource contention — but stated in spreading-activation memory vocabulary that pins it to its home domain, leaving it mixed-structural rather than a free-floating prime.

Structural Core vs. Domain Accent¶

This section decides why the fan effect is a domain-specific abstraction and not a prime — and it carries the case for its domain-specificity.

What is skeletal (could lift toward a cross-domain prime). Strip the memory model and a clean relational structure survives: a fixed resource is divided across a set of competing recipients, so that adding recipients thins the share of each and degrades the service any one receives. The portable pieces are abstract — a bounded budget, a divisor (the count of contenders), and a per-recipient cost that rises with that count. That skeleton is unusually cleanly substrate-portable: it recurs as a contended communication channel that slows with more senders, a router's fan-out that forces more candidates to disambiguate per query, a cache that thrashes as more keys compete for fixed lines — none of them cognitive. Precisely because it recurs, it is carried by the parents the entry instantiates — interference_and_contention, bandwidth_and_throughput, scarcity, and dilution. But that is the core the effect shares, not what makes it distinctive.

What is domain-bound. Everything that makes this specifically the fan effect is declarative-memory furniture and none of it survives extraction. It presupposes cued retrieval from a declarative store, a spreading-activation architecture in which a concept node's fixed activation budget spreads divisively to its associates, and a measurable competition among co-active associations in a retrieval race against a recognition threshold — properties of a particular family of cognitive memory models (Anderson, ACT-R) and their biological substrate. Its worked content is discipline-internal: the fan as the countable divisor, the raw-versus-effective-fan distinction, the chunking intervention that folds associations into higher-order units, the resolution of the expert-memory paradox, and the empirical cases (the hippie-in-the-park sentences, the chess-reconstruction studies). The decisive test: remove the retrieval race and the activation budget — a router's fan-out or a thrashing cache — and there is no fan effect, only bare contention over a fixed resource; calling that "a fan effect" would import activation-division, retrieval-race, and chunking vocabulary that have no referent in an index. The spreading-activation machinery is the part that stays home, and it is what makes the effect's countable, chunking-reversible predictions hold.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. The fan effect's transfer is bimodal. Within cognitive psychology and memory research it moves intact as mechanism — the same activation-division machinery is literally at work from list-learning to eyewitness recall to expert chess memory, and the diagnostics (latency scales with summed fan, not with learning order or total list length) and interventions (chunking collapses the effective fan) carry without translation. Beyond associative memory it travels only by analogy: the database fan-out the literature itself cites is a genuine instance of the parent contention pattern, but naming it "a fan effect" borrows the shape while dropping the retrieval-race machinery that gives the original its content. And when the bare structural lesson is needed cross-domain — more competitors for a fixed resource means worse service per competitor — it is already carried, in more general and substrate-neutral form, by interference_and_contention, bandwidth_and_throughput, scarcity, and dilution. The cross-domain reach belongs to those parents; "fan effect," as named, carries the spreading-activation baggage that should stay in the associative-memory corner where activation-division and chunking make its specific, countable predictions.

Relationships to Other Abstractions¶

Current abstraction Fan Effect Domain-specific

Parents (2) — more general patterns this builds on

Fan Effect is part of Threshold Prime

A recognition Threshold is a constituent of the Fan Effect because a thinned association is retrieved only after its divided activation accumulates enough to cross the decision boundary.
Fan Effect is a decomposition of Interference and Contention Prime

The Fan Effect is the declarative-memory application of Interference and Contention in which concurrent associations divide one concept's fixed activation budget and degrade retrieval.

Hierarchy paths (4) — routes to 3 parentless roots

Fan Effect → Threshold

Show alternative paths (3)

Not to Be Confused With¶

Proactive and retroactive interference. Memory competition set by the temporal order in which material was learned — earlier learning impeding later (proactive) or later impeding earlier (retroactive). The fan effect is synchronic: competition among associations co-attached to one concept regardless of learning order, scaling with how many facts share the cued concept. Tell: does the slowing track when the competing material was learned (proactive/retroactive) or how many other facts share the cued concept (fan)?
List-length effect. Slowing that scales with the total size of the studied set. The fan effect scales with the fan of the cued concept specifically — two designs with identical total list length can differ in fan, and it is the fan that predicts the latency. Tell: does the cost rise with the whole set's size (list-length) or with the number of associations on the probed concept (fan)?
Hick's law. The finding that choice reaction time rises (logarithmically) with the number of response alternatives presented in a decision task. It concerns selecting among options laid before the subject; the fan effect concerns retrieval from a declarative store, where a target association must out-activate co-attached competitors in a recognition race. Tell: is the delay in choosing among presented alternatives (Hick's law) or in verifying a stored fact whose concept carries many associations (fan)?
Working-memory load / cognitive-load limits. Degradation from exceeding the capacity of the transient workspace that holds items for active processing. The fan effect is a property of long-term declarative retrieval — a fixed activation budget divided across a concept's associates — not of a working-memory bottleneck. Tell: is the constraint the number of items held active right now (working memory), or the number of associations permanently attached to the cued concept in long-term store (fan)?
Semantic (associative) priming. The facilitation side of spreading activation — a related prime speeding recognition of a target. The fan effect is the cost side of the very same architecture: activation divided across many associates leaves each one less, slowing retrieval. Tell: does a related cue speed the target (priming, activation converging on it) or does a crowd of co-attached associates slow it (fan, activation divided among them)?
Memory decay / weak encoding. Slowing attributed to a fading or poorly-laid trace. The fan effect's slowing is competition under a divided budget with the trace fully intact — which is exactly why the fan can be isolated as its own countable variable, and why adding well-learned facts (weakening nothing) is what produces it. Tell: is the trace itself faded or poorly formed (decay/encoding), or intact but competing with many co-attached rivals (fan)?
The contention/dilution parents (interference_and_contention, bandwidth_and_throughput, scarcity, dilution). The broad, substrate-neutral pattern — a fixed resource divided across more competing recipients degrades service per recipient — that the fan effect instantiates. The database fan-out and thrashing cache the literature itself cites are instances of these parents, not of the named effect. Tell: strip the cue, the association, the retrieval race, and chunking and what remains — more competitors for a fixed budget means worse service each — is the parent, with no activation-division machinery. (Treated fully in the sections above.)

Neighborhood in Abstraction Space¶

Fan Effect sits in a crowded region of the domain-specific corpus (21^st 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 — 0.86
Retrieval Practice Effect — 0.86
Self-Reference Effect — 0.86
Levels-of-Processing Effect — 0.85
Retrieval-Induced Forgetting — 0.85

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