Egocentric Bias¶

Trace a family of social-judgment miscalibrations to one asymmetry — the self's mental contents are available by default while other minds must be inferred at cost — so the self serves as an anchor from which adjustment is systematically insufficient.

Core Idea¶

Egocentric bias is the family of systematic miscalibrations that arise from a reasoner's privileged, effortless access to her own mental contents combined with the cognitive cost of modelling another's: because first-person information is available by default and other-perspective information must be inferred, the self's knowledge, experience, preferences, and contributions function as a default anchor in social and joint judgments, with adjustment away from that anchor being systematically insufficient. The bias manifests across four empirically distinct paradigms that share this underlying asymmetry. In joint-credit attribution (Ross and Sicoly 1979), spouses' and team members' independent estimates of their own contributions to shared tasks sum to more than 100% because each person's own contributions are more vivid in memory than the partner's, producing over-claiming that is not primarily motivated but memorial. In false-consensus estimation, people rate the prevalence of their own preferences, attitudes, and behaviors in the broader population as higher than it is, treating the self as a representative sample. In curse-of-knowledge effects, knowing a fact makes it difficult to recover the perspective of not knowing it — experts over-estimate how much novices can infer, teachers over-estimate how readily students will follow explanations, and communicators over-estimate how much background is shared — because the known fact is permanently available and the naive state must be reconstructed at cognitive cost. In the spotlight effect (Gilovich, Medvec, and Savitsky 2000), individuals overestimate how much observers notice and remember their appearance and behavior, because one's own appearance is salient to oneself and the inferential effort required to model another's inattention is not automatically performed. Across all four, the corrective move is the same: supply explicit external information — a base rate, a third-party estimate, a contemporaneous record — that bypasses the self-anchored default and requires the reasoner to weight it against self-generated information rather than substituting for it.

Structural Signature¶

Sig role-phrases:

the privileged first-person access — the self's own mental contents (knowledge, memory, preference, self-attention) available by default with speed and vividness
the inferred other-mind model — another's perspective that must be reconstructed from external cues at cognitive cost, not observed directly
the perspective-taking cost — the load that makes full modeling of the other expensive, so the inference is underperformed (the source is information cost, not motivation)
the self-as-default-anchor — the self's own perspective serving as the anchor in any social or joint judgment
the insufficient adjustment — movement away from the self toward the other systematically too small, the self-generated content left over-weighted
the axis-selected symptom — the surface error selected by which judgment axis is in play: over-100% joint-credit sums, false consensus, curse of knowledge, spotlight over-estimation
the memorial-not-motivated signature — the over-claim driven by accessibility (own contributions more vivid) rather than self-interest, so it appears even in an other-regarding agent
the external-information corrective — supplying a base rate, third-party estimate, contemporaneous record, or salient other-perspective that bypasses the self-anchored default and must be weighted against it rather than substituting for it

What It Is Not¶

Not egotism or selfishness. Egotism is a normative claim about valuing oneself over others; egocentric bias is a descriptive claim about which perspective is the cognitive default. The error arises from the asymmetric accessibility of self-versus-other information, not from a preference for the self, so an agent can be entirely other-regarding in what she wants and still over-claim, over-project, and over-assume shared knowledge.
Not a motivated, self-serving distortion. The over-claim is memorial, not motivated: own contributions are simply more vivid in memory than the partner's, which is why independent contribution estimates sum to more than 100% even with no self-interest at stake. (Self-serving attribution — crediting success to ability, blaming failure on circumstance — is a sibling that adds a motivational component; egocentric bias proper does not require it.)
Not a single quirk. Egocentric bias is the root of a family, not one isolated effect: false consensus, the curse of knowledge, the spotlight effect, and joint-credit over-summing are the same self-anchor read on different judgment axes. Equating it with any one of them (e.g., "it's just the spotlight effect") mistakes a leaf for the architecture that predicts all of them co-occurring in the same reasoner.
Not fixable by moral exhortation. Because the source is information cost rather than motivation, urging someone to "be less self-centered" is the wrong remedy; it leaves the default access untouched. The corrective is to supply external information — a base rate, a third-party estimate, a contemporaneous record, a salient other-perspective — that bypasses the self-anchored default and must be weighted against self-generated information.
Not present in any system with merely local information. The bias requires an agent with privileged first-person access and an inferred model of other minds under perspective-taking cost; a control system whose sensor is local does not "over-attribute," it simply lacks remote data. With no default self and no other-model, there is nothing to over-anchor on — the construct has no referent outside perspective-bearing minds (its group-level analogue, with the in-group as reference, is ethnocentrism).

Scope of Application¶

Egocentric bias lives in one domain — the social and joint judgment of perspective-bearing minds — and is unusually the root of a family: the contexts below are the empirical paradigms it parents within that single substrate, not structurally distinct systems. Its component parents (availability heuristic, anchoring-and-adjustment) carry the cross-context lesson, and the same anchor-and-under-adjust move recurs with the in-group as reference in ethnocentrism; outside a first-person/other-mind asymmetry the construct has no referent and stays out of this map.

Joint-credit attribution — independent self-estimates of contribution to shared work sum to more than 100% (chores, co-authored papers, team projects), because one's own contributions are more vivid in memory (Ross and Sicoly).
False-consensus effect — own preferences, attitudes, and behaviors are estimated as more prevalent in the population than they are, the self treated as a representative sample.
Curse of knowledge — knowing a fact makes the not-knowing state hard to recover, so experts, teachers, and communicators over-estimate what naive others can infer (Camerer's tappers-and-listeners), biasing pedagogy, marketing, and expert testimony.
Self-serving attribution — success credited to ability and failure to circumstance, the motivated sibling that adds a self-esteem component to the same self-anchor.
Spotlight effect — one's own appearance and behavior are over-estimated in how much observers notice and remember them (Gilovich, Medvec, Savitsky).
Theory-of-mind development — the egocentric baseline (failure to suppress one's own true belief in false-belief tasks) that decentering progressively overcomes.

Clarity¶

Naming egocentric bias does its main work by collapsing a scatter of paradigms that the literature studies separately — over-100% joint-credit sums, false consensus, the curse of knowledge, the spotlight effect — into surface readings of one underlying asymmetry: first-person mental contents are available by default, other minds must be inferred at cost, and so the self serves as an anchor from which adjustment is systematically insufficient. Without the label these look like four unrelated quirks of social judgment; with it, they become predictable outputs of a single architecture, differing only in which self-generated content is over-weighted (one's own remembered contributions, preferences, knowledge, or self-attention) and which judgment axis is in play. That unification is what lets a researcher predict the co-occurrence of these effects in the same reasoner and apply one corrective logic — supply external information that bypasses the self-anchored default — across all of them.

The label's second clarifying act is to separate two things ordinary language fuses under "self-centered": egotism, a normative claim about valuing oneself over others, and egocentric bias, a descriptive claim about which perspective is the cognitive default. The distinction is consequential because it relocates the source of the error from motivation to information cost — an agent can be entirely other-regarding in what she wants and still over-claim, over-project, and over-assume shared knowledge, because the bias arises from the asymmetric accessibility of self-versus-other information, not from a preference for the self. This blocks the wrong remedy (moral exhortation to be less selfish) and points to the right one (make the other's perspective salient or supply a base rate), and it makes the sharper question not "is this person self-serving?" but "has the other's perspective been modeled at cost, or substituted for by the self's default access?"

Manages Complexity¶

The payoff of the single self-access asymmetry is that the whole sprawling design space of this corner of social cognition — the empirical paradigms, the debiasing interventions, the pedagogical and theory-of-mind-development findings — reduces to two coordinates plus one fixed corrective. Any new instance is specified by stating which self-generated content is over-weighted (remembered contributions, preferences, knowledge, self-attention) and which judgment axis carries it (joint-credit share, population prevalence, a novice's inference, an observer's notice), and its qualitative direction follows: over-claim, over-project, over-assume shared knowledge, over-estimate scrutiny. A researcher therefore predicts that the same reasoner who over-claims credit will also fall to false consensus and the curse of knowledge, because all are the same anchor read on different axes, rather than cataloguing each as a separate quirk to be measured and remediated on its own. The remedy space collapses identically: every effective intervention is one move — supply external information (a base rate, a third-party estimate, a contemporaneous record, a salient other-perspective) that bypasses the self-anchored default — so the analyst chooses an instrument from one family rather than designing a bespoke fix per paradigm. The high-dimensional question "how do all these social-judgment errors interrelate and how is each corrected?" becomes a two-parameter lookup over one architecture with a shared corrective logic.

Abstract Reasoning¶

Egocentric bias licenses moves organized around one architectural asymmetry — first-person contents available by default, other minds inferred at cost — and the self-anchor-with-insufficient-adjustment it produces. Diagnostic: confronted with a miscalibrated social judgment, the analyst specifies it by two coordinates — which self-generated content is over-weighted (remembered contributions, preferences, knowledge, self-attention) and which judgment axis carries it (joint-credit share, population prevalence, a novice's inference, an observer's notice) — and reads the qualitative direction off them: over-claim, over-project, over-assume shared knowledge, over-estimate scrutiny. The unifying diagnostic inference is that these are not four unrelated quirks but one anchor read on different axes, which licenses a prediction the separate-paradigm view cannot make: the same reasoner who over-claims credit will also fall to false consensus and the curse of knowledge, because all three are the same self-access asymmetry on different judgment dimensions. A signature read on the joint-credit axis is especially clean — independent self-estimates of contribution summing to more than 100% positively identifies the bias, since the over-claim is memorial (own contributions more vivid) rather than motivated.

Interventionist: the corrective move is one family, applied across every paradigm. Supply external information — a base rate, a third-party estimate, a contemporaneous record, a salient other-perspective — that bypasses the self-anchored default and must be weighted against self-generated information rather than allowed to substitute for it; predicting that the over-claim, over-projection, or over-assumption of shared knowledge attenuates as the external anchor displaces the self as default. The analyst therefore chooses an instrument from one family rather than designing a bespoke fix per effect: a base rate for false consensus, a disaggregated task record for joint-credit inflation, an explicit model of the naive state for the curse of knowledge, an observer's actual report for the spotlight effect — all the same move, bypass the default. To shift the judgment, then, one makes the other's perspective available at lower cost or supplies the missing external datum, rather than exhorting better intentions.

Boundary-drawing: the concept fixes the error's source as information cost, not motivation, and that placement is itself the central boundary. It separates egocentric bias (a descriptive claim about which perspective is the cognitive default) from egotism (a normative claim about valuing oneself over others), and the consequence is diagnostic and prescriptive: an agent entirely other-regarding in what she wants still over-claims, over-projects, and over-assumes shared knowledge, because the bias arises from asymmetric accessibility, so moral exhortation to be less selfish is the wrong remedy and making the other's perspective salient or supplying a base rate is the right one. The discriminating question becomes "has the other's perspective been modeled at cost, or substituted for by the self's default access?" rather than "is this person self-serving?" The boundary also bounds where the bias exists at all: it requires an agent with privileged first-person access plus an inferred (not observed) model of other minds operating under load that makes full perspective-taking expensive, so where the other's contents are directly available, or no other-model is in play, there is nothing to over-anchor on. Predictive: given a social or joint judgment, the analyst forecasts in advance the direction of error from the two coordinates — that contribution estimates will over-sum, that one's own preference will be projected onto the population, that a known fact will be assumed inferable by the naive, that one's own appearance will be assumed noticed — and predicts their co-occurrence in the same reasoner, before measuring any single effect.

Knowledge Transfer¶

Within cognition the bias transfers as mechanism, intact, and unusually it is itself the root of a family rather than a leaf: the single self-access asymmetry (first-person contents available by default, other minds inferred at cost, self-anchor with insufficient adjustment) carries without translation across the paradigms it parents — joint-credit attribution (Ross and Sicoly's over-100% contribution sums), the false-consensus effect (own preference projected onto the population), the curse of knowledge (a known fact assumed inferable by the naive — Camerer's tappers-and-listeners, biasing pedagogy, marketing copy, expert testimony, and economic-modeling assumptions), self-serving attribution, the spotlight effect (own appearance assumed noticed), and even theory-of-mind development (the egocentric baseline that decentering progressively overcomes). The two-coordinate specification (which self-generated content is over-weighted × which judgment axis carries it), the diagnostic (the same reasoner shows all of these because they are one anchor read on different axes; over-summing contribution estimates positively identify the bias as memorial rather than motivated), and the one-family corrective (supply external information — base rate, third-party estimate, contemporaneous record, salient other-perspective — that bypasses the self-anchored default) all move freely, because the same self-versus-other accessibility asymmetry underlies each.

Beyond a perspective-bearing mind the bias has no referent — and this boundary is harder than "metaphor." Egocentric bias requires the cognitive architecture of an agent with privileged first-person access, an inferred (not observed) model of other minds, and perspective-taking cost under load — humans paradigmatically, with weaker analogues in some primates and possibly social cetaceans. Outside such minds there is nothing to over-anchor on: a control system whose sensor is local does not "over-attribute," it simply lacks remote information; there is no default self, no other-model, no insufficient adjustment. So the construct does not transfer to non-mind-bearing substrates at all, by analogy or otherwise. What genuinely travels is the more-general structure the bias composes and instantiates. As a composite it is built from parents that each carry within cognition — the availability heuristic (self-information most accessible) and anchoring-and-adjustment (self the default anchor, adjustment insufficient) — plus a substrate-specific fact about perspective-taking cost; when the cross-context lesson is wanted, those component primes are the cargo. And the structural move itself — anchor on a privileged-access reference and under-adjust toward an inferred alternative — recurs as a genuine co-instance when the reference unit is shifted: its group-level analogue is ethnocentrism, the same self-anchor-and-insufficient-adjustment architecture with the in-group, rather than the self, as the default reference. That shifted-reference recurrence is the honest sense in which the pattern travels — to other reference units within perspective-bearing systems — not to thermostats or markets. So: as mechanism it stays inside (and roots a family within) perspective-bearing cognition; its component primes (availability, anchoring-and-adjustment) travel as the real structure; the move recurs with a changed reference unit in ethnocentrism; and it has no purchase on systems without a first-person/other-mind asymmetry (see Structural Core vs. Domain Accent).

Examples¶

Canonical¶

Ross and Sicoly's 1979 study is the founding demonstration. They asked married couples to estimate, independently, their own share of responsibility for a list of household activities — making breakfast, cleaning the house, planning leisure, even causing arguments. When each spouse's self-estimated contributions were added to the partner's, the totals systematically exceeded 100%: both partners claimed more than half. Crucially, the over-claiming appeared even for undesirable activities like starting fights, which rules out simple self-flattery. The explanation is memorial, not motivated: one's own contributions — the mornings you made breakfast — are encoded and retrieved more vividly than a partner's, so when each person reconstructs the shared total from memory, self-generated instances dominate the sample. The self's own actions, available by default, serve as an anchor from which adjustment toward the partner's less-accessible contributions is insufficient.

Mapped back: Each spouse's vivid memory of their own chores is the privileged first-person access functioning as the self-as-default-anchor; the partner's fainter contributions get insufficient adjustment, so the estimates over-sum — the axis-selected symptom on the joint-credit axis. That over-claiming survives for arguments too is the memorial-not-motivated signature.

Applied / In Practice¶

Clinical medicine has operationalised a corrective for the curse-of-knowledge form of the bias. Physicians, fluent in their own knowledge, systematically overestimate how much of a diagnosis, medication regimen, or discharge instruction a patient has understood — the known fact is permanently available to the clinician while the patient's naive state must be reconstructed at cost, and often is not. Studies of patient comprehension repeatedly find gaps the clinician does not anticipate. The standard remedy is the "teach-back" method: rather than asking "Do you understand?", the clinician asks the patient to restate the plan in their own words. This supplies exactly the external datum the concept prescribes — the patient's actual output — which bypasses the clinician's self-anchored assumption of shared understanding and forces it to be weighted against evidence rather than substituted for.

Mapped back: The clinician's fluency is the privileged first-person access; overestimating patient comprehension is the curse-of-knowledge axis-selected symptom, driven by the perspective-taking cost of reconstructing a naive state. Teach-back is the external-information corrective — a third-party datum (the patient's restatement) bypassing the self-as-default-anchor.

Structural Tensions¶

T1: Information cost versus motivation (the memorial signature and its motivated sibling). The concept's load-bearing placement is that the error's source is accessibility, not selfishness — the over-claim is memorial (own contributions more vivid) rather than motivated, which is why it survives even for undesirable acts like starting arguments and why moral exhortation is the wrong remedy. But motivation does not vanish in the real world: self-serving attribution is a sibling that adds a self-esteem component to the same self-anchor, and in any given over-claim the memorial and motivated contributions are entangled and hard to separate. The tension is that the concept earns its precision by isolating the information-cost mechanism, yet the phenomena it explains often co-occur with the motivated distortion it defines itself against — and mis-attributing a memorial error to motivation prescribes the wrong fix. Diagnostic: Does the over-claim persist even for undesirable contributions (memorial, information-cost) or only for flattering ones (motivated, self-serving)?

T2: Ecologically valid self-anchor versus systematic bias (the self is a cheap, usually-serviceable sample). Using one's own knowledge, preference, or experience as the default anchor is not a malfunction bolted onto reason — it is often a serviceable heuristic, because absent other information the self genuinely is some evidence about others, and first-person access is fast and free. The bias is the same move evaluated in a different ecology: it misleads precisely where the self diverges from the other (an expert's knowledge from a novice's, a minority preference from the population's). So calling every self-anchored judgment a bias misreads a reasonable prior, while trusting it everywhere ignores the divergence that makes it fail. There is no separate "good" and "bad" self-anchoring to pry apart — the useful default and the systematic error are one architecture read in two situations. Diagnostic: In this judgment, is the self a roughly representative sample of the other (anchor serviceable), or does self diverge from other on the axis in play (anchor misleads)?

T3: Root-of-a-family unification versus paradigm-specific mechanism (one architecture or four). The concept's power is collapsing false consensus, the curse of knowledge, joint-credit over-summing, and the spotlight effect into one self-access asymmetry read on different axes — licensing the prediction that the same reasoner shows all of them. But the unification can be pressed too far: the paradigms may have partly distinct proximate mechanisms (the spotlight effect involves an observer-model, the curse of knowledge a difficulty un-knowing a fact, joint-credit a memory-sampling asymmetry), and treating them as pure surface readings of one anchor can obscure differences that matter for measurement and remediation. The architecture that predicts co-occurrence also risks over-homogenizing effects that diverge in detail. Diagnostic: Is the claim about the shared self-anchor architecture (co-occurrence, one corrective family) or about the specific proximate mechanism of one paradigm (where the effects may differ)?

T4: The corrective attenuates versus the anchor persists (supplying information does not evict the default). The one-family remedy — supply a base rate, third-party estimate, contemporaneous record, or salient other-perspective — is meant to bypass the self-anchored default. But the concept's own core is anchoring with insufficient adjustment, and that insufficiency does not disappear when an external datum is provided: the self remains the default, the external information is weighted against it rather than substituting for it, and adjustment stays systematically too small. So the corrective attenuates the error without eliminating it — a base rate reduces false consensus but the residual self-tilt survives, and the external anchor must be made to displace the default rather than merely sit beside it. The fix works partially, by the same mechanism that causes the bias. Diagnostic: Does the external information actually displace the self as the default anchor, or is it merely present alongside a self-tilt that still under-adjusts?

T5: Autonomy versus reduction (the root of a cognitive family or its availability/anchoring parents). Within cognition egocentric bias is unusually the root of a family, and as mechanism it stays inside — and parents — perspective-bearing minds, with no referent in a system that merely has local information (a local-sensor control system does not "over-attribute," it simply lacks remote data). What genuinely travels is the structure it composes: the availability_heuristic (self-information most accessible) and anchoring_and_adjustment (self the default, adjustment insufficient), plus a substrate-specific perspective-taking-cost fact. And the move itself recurs as a genuine co-instance when the reference unit shifts from the self to the in-group — that is ethnocentrism, the same anchor-and-under-adjust architecture. The tension is between a named cognitive architecture that roots its own family and the recognition that its portable content is those component primes (and its shifted-reference recurrence). Diagnostic: Resolve toward availability_heuristic/anchoring_and_adjustment (or ethnocentrism when the reference is a group) when carrying the lesson; toward egocentric bias when a first-person/other-mind asymmetry is diagnosing a social judgment in situ.

Structural–Framed Character¶

Egocentric bias sits on the framed side of the spectrum — best read as framed-leaning: a genuine accessibility-driven cognitive mechanism, but one bound to perspective-bearing minds, labeled as a miscalibration, and stated in judgment-and-decision-making idiom. On evaluative_weight it is the softened but real presence that keeps it framed rather than structural: "bias," "miscalibration," "over-claim," "insufficient adjustment" are all deviation-from-calibration terms, so the concept convicts a judgment of being wrong relative to a normative standard — yet the entry deliberately blunts that charge, insisting the error is memorial, not motivated, descriptive not normative, and often (T2) a serviceable ecological prior, so it labels an error without the moral verdict a fallacy like ad hominem carries. Its human_practice_bound character is unusual: the bias is not constituted by an institution the way an argumentation fallacy is — it runs automatically, so a spouse over-claims credit whether or not any psychologist names it — but it is constituted by a cognitive substrate, requiring an agent with privileged first-person access plus an inferred model of other minds under perspective-taking cost, and it has "no referent" at all outside such minds (a local-sensor control system does not over-attribute, it simply lacks remote data). That hard substrate-boundedness, short of full observer-free naturalness, is what places it framed-leaning rather than mixed-structural. Its institutional_origin is mild-to-moderate: the four-paradigm taxonomy (false consensus, curse of knowledge, spotlight effect, joint-credit over-summing) and its measures are furniture of cognitive psychology (Ross–Sicoly, Gilovich–Medvec–Savitsky, Camerer), though the underlying accessibility asymmetry is a fact of how minds work rather than an artifact of the theory. On vocab_travels it scores low and import_vs_recognize is bounded: within cognition the self-access asymmetry transfers as mechanism and roots a family of paradigms (recognized across axes), and its move recurs as a genuine co-instance under a shifted reference unit (ethnocentrism), but off perspective-bearing minds it does not transfer at all — not even by analogy.

The one portable structural skeleton is anchor on a privileged-access reference and under-adjust toward an inferred alternative. That skeleton is real and recurs (with the in-group swapped for the self, it is ethnocentrism) — but it does not pull egocentric bias off the framed side, because it is precisely what the bias composes and instantiates from its parents, not what makes "egocentric bias" itself travel: the portable content belongs to the availability_heuristic (self-information most accessible) and anchoring_and_adjustment (self the default anchor, adjustment insufficient) it is built from, plus a substrate-specific perspective-taking-cost fact, while the four-paradigm taxonomy, the memorial-not-motivated signature, and the first-person/other-mind framing are the parts that stay home. Its character: a substrate-bound, error-labeled but non-motivated cognitive-judgment architecture whose distinctive content is the perspective-cost asymmetry and its paradigm family, structural only in the anchor-and-under-adjust skeleton it borrows from availability_heuristic and anchoring_and_adjustment and reads as a calibration failure of social judgment.

Structural Core vs. Domain Accent¶

This section settles why egocentric bias is a domain-specific abstraction and not a prime — the boundary between the anchor-and-under-adjust skeleton that lifts and the perspective-cost apparatus that stays is the whole of the ruling.

What is skeletal (could lift toward a cross-domain prime). Strip the four paradigms and a thin relational structure survives: a reasoner anchors on a privileged-access reference and adjusts systematically too little toward an alternative that must be inferred at cost. The portable pieces are abstract: a default reference available fast and free, an alternative reachable only by expensive inference, an anchoring move, and an adjustment that stops short. That skeleton is genuinely substrate-portable — it is precisely what the bias composes from the availability_heuristic (the reference is the most accessible information) and anchoring_and_adjustment (the reference is the default anchor, adjustment insufficient), and it recurs as a genuine co-instance when the reference unit is swapped from the self to the in-group, which is ethnocentrism. But that skeleton is the core it shares with those parents and with its shifted-reference sibling, not what makes it egocentric bias.

What is domain-bound. Nearly all the worked content is judgment-and-decision-making furniture that does not survive extraction: the privileged first-person access to one's own mental contents, the inferred other-mind model reconstructed under perspective-taking cost, the memorial-not-motivated signature that distinguishes it from self-serving distortion, and the four-paradigm taxonomy the construct roots (false consensus, curse of knowledge, spotlight effect, joint-credit over-summing) with its measures (Ross–Sicoly's over-100% sums, Camerer's tappers-and-listeners, Gilovich–Medvec–Savitsky). Decisively, the construct's defining commitment is that the reference is the self and the alternative is another mind — a fact about first-person/other-mind cognitive architecture. The decisive test: remove that substrate — a perspective-bearing agent with privileged first-person access and an inferred model of other minds under load — and there is nothing to over-anchor on; a control system whose sensor is local does not "over-attribute," it simply lacks remote data, so what remains is only the bare anchor-and-under-adjust pattern with no self and no other-mind in it. The naming vocabulary renames every component off the social-cognition substrate.

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. Egocentric bias's transfer is unusually sharp-edged. Within perspective-bearing cognition the self-access asymmetry travels intact as mechanism and is itself the root of a family — joint-credit attribution, false consensus, curse of knowledge, self-serving attribution, spotlight effect, theory-of-mind development are one anchor read on different judgment axes, recognized across all of them, with a single corrective (supply external information that bypasses the default). Beyond perspective-bearing minds it does not transfer even by analogy: outside a first-person/other-mind asymmetry the construct has no referent at all. And when the bare structural lesson is what is needed cross-domain — anchor on a privileged-access reference and under-adjust toward an inferred alternative — it is already carried, in more general form, by availability_heuristic and anchoring_and_adjustment (and, with a group reference, by ethnocentrism). The cross-domain reach belongs to those parents; "egocentric bias," as named, carries the perspective-cost asymmetry, the memorial-not-motivated signature, and the four-paradigm taxonomy that should stay home.

Relationships to Other Abstractions¶

Current abstraction Egocentric Bias Domain-specific

Parents (2) — more general patterns this builds on

Egocentric Bias presupposes Theory Of Mind Prime

Egocentric Bias requires an agent to model another's attention, knowledge, preference, or contribution before its self-anchored model can be miscalibrated.
Egocentric Bias is a decomposition of Anchoring Prime

Egocentric Bias is anchoring-and-insufficient-adjustment with the reasoner's own mental state fixed as the anchor and another perspective as the target.

Children (4) — more specific cases that build on this

Comparative Self-Assessment Crossover Domain-specific is a kind of Egocentric Bias

The crossover is the comparative-ability species of Egocentric Bias, using one's own felt skill as the readily available anchor while under-modeling the abilities of others.
Curse Of Knowledge Domain-specific is a kind of Egocentric Bias

Curse of Knowledge is the Egocentric Bias species in which an informed state anchors simulation of an uninformed other.
False Consensus Effect Domain-specific is a kind of Egocentric Bias

False Consensus Effect is the Egocentric Bias species in which an estimator projects its own state onto a population.

▸ Show 1 more

Hierarchy paths (4) — routes to 3 parentless roots

Egocentric Bias → Theory Of Mind → Mental Model → Representation → Abstraction

Show alternative paths (3)

Not to Be Confused With¶

Egotism / selfishness. A normative disposition to value oneself over others — a matter of what one wants. Egocentric bias is a descriptive fact about which perspective is the cognitive default, driven by asymmetric accessibility of self-versus-other information. An entirely other-regarding agent still over-claims, over-projects, and over-assumes shared knowledge. Tell: does the error stem from a preference for the self (egotism), or from the self's information simply being cheaper to access regardless of motives (egocentric bias)? The wrong-remedy tell: moral exhortation touches egotism but not the accessibility default.
Self-serving bias / self-serving attribution. The motivated sibling: crediting success to one's ability and blaming failure on circumstance, to protect self-esteem. It adds a motivational component the pure egocentric bias does not require — which is why egocentric over-claiming survives even for undesirable acts (starting arguments), ruling out self-flattery. Tell: does the over-claim appear only for flattering contributions (self-serving, motivated) or also for unflattering ones (egocentric, memorial)?
A single paradigm (false consensus / curse of knowledge / spotlight effect). Each of these is one leaf — the same self-anchor read on a particular judgment axis (population prevalence, a novice's inference, an observer's notice). Egocentric bias is the root architecture that predicts all of them co-occurring in the same reasoner. Equating it with any one ("it's just the spotlight effect") mistakes a leaf for the tree. Tell: is the claim about one axis's specific effect (a paradigm), or about the shared self-access asymmetry that generates the whole family (egocentric bias)?
Fundamental attribution error. The tendency to over-attribute others' behavior to disposition rather than situation. It concerns how we explain other people's conduct; egocentric bias concerns anchoring on the self's accessible contents and under-adjusting toward another mind. Different asymmetry, different target. Tell: is the error over-weighting disposition when judging another's behavior (FAE), or over-weighting one's own perspective as the default in a social/joint judgment (egocentric bias)?
Ethnocentrism. The group-level analogue — the same anchor-and-under-adjust architecture with the in-group, rather than the self, as the default reference. It is a genuine co-instance of the identical structural move under a shifted reference unit, not a rival mechanism. Tell: is the privileged-access reference the self (egocentric bias) or the in-group (ethnocentrism)? Same move, changed reference.
Availability heuristic and anchoring-and-adjustment (the parent primes it composes). The substrate-neutral skeleton — anchor on a privileged-access reference and adjust systematically too little toward an inferred alternative — is built from availability_heuristic (self-information most accessible) and anchoring_and_adjustment (self the default anchor, adjustment insufficient), plus a perspective-taking-cost fact. These carry the portable content. Tell: for the anchor-and-under-adjust lesson, use those parents (or ethnocentrism for a group reference); "egocentric bias" has no referent outside a first-person/other-mind asymmetry — a local-sensor system doesn't over-attribute, it just lacks remote data. (Treated fully in earlier sections.)

Neighborhood in Abstraction Space¶

Egocentric Bias sits in a crowded region of the domain-specific corpus (12^th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Social Perception & Self-Referential Bias (23 abstractions)

Nearest neighbors

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