Source
- Turner, Oakes, Haslam & McGarty (1994), LIT-1794, read in full in NOTE-1284 from a scan of the published article.
- Hornsey (2008), LIT-1742, read in full in NOTE-1253.
- The founding statement, Turner et al. (1987), LIT-1723, is held but Unavailable, and is not read. Its chapter 3 statements of meta-contrast, salience and depersonalization (pp. 42–56) are quoted with page numbers by Skorich (LIT-1765, NOTE-1295) and León (LIT-1721, NOTE-1309). The record therefore has the book’s own words at second hand, but not its argument or its evidence.
What the account says
A self-category groups self with some class of others as equivalent, against some other class. Personal identity is self-categorization by differences from other in-group members. Social identity is self-categorization by shared membership against other categories. Both are equally authentic (Turner et al. 1994, p. 454).
Which category is salient is an interaction of perceiver readiness (accessibility, from experience, goals and motives) and fit (p. 455). Comparative fit is the metacontrast principle: a collection is more likely to be categorized as an entity to the degree that the average differences among its members are less than the average differences between them and the rest of the frame of reference. Normative fit requires those differences to lie in the direction background beliefs about the category predict.
From metacontrast the account derives four kinds of variation (pp. 456–458):
- level: more inclusive categories as the frame widens;
- kind: which category, by normative fit;
- content: the traits that make up the category;
- internal structure: who is most prototypical.
Salience depersonalizes self-perception: self and others are seen as interchangeable exemplars of the prototype, and group phenomena such as cohesion, influence and polarization follow (p. 455; Hornsey 2008, pp. 208–211).
What was actually shown
Nothing in the record shows it at first hand. Both sources report results from the theory’s originating group without presenting data. Two of the primary studies were filed unread on 2026-10-29: Oakes, Turner & Haslam 1991 (LIT-1726) and Haslam et al. 1992 (LIT-1767). The reported results are:
- arts and science identities most salient under both comparative and normative fit (Oakes, Turner & Haslam 1991);
- more sex-stereotyping of self under intergroup than intragroup disagreement (Hogg & Turner 1987);
- the content of “Americans” shifting with the comparison country and over the Gulf War (Haslam et al. 1992);
- extremists’ greater polarization reversible by their position relative to the majority (Haslam & Turner, in press in 1994).
Hornsey adds that in the SCT programme, high salience or identification increased self-stereotyping and perceived in-group homogeneity (2008, p. 213). Each of these could have come out otherwise, since content could have stayed fixed as the frame changed. None has been read here.
What this does not say
- Not that the self is generated rather than stored. Turner et al. (1994, pp. 458–459) go on to “speculate” that a self-concept is defined by reflexive use and that a fixed self-structure may be neither “meaningful or necessary”. That is a further thesis argued by analogy to category research, and it is not filed. A stored repertoire with context-dependent selection (Markus & Wurf’s working self-concept) predicts much of the same variation.
- Not that fit determines salience. The authors call fit “part of the explanation but not the complete account” (p. 456). Readiness takes in motives and goals, so the account predicts a direction only once readiness is held fixed or measured.
- Not social identity theory’s motivational claims. Positive distinctiveness, the self-esteem hypotheses and the socio-structural variables are SIT’s (Hornsey 2008, pp. 206–207, 214). They are not part of this account, and the self-esteem strand is contested.
- Not that variability is veridical. The authors claim it (p. 458) but give no criterion.
- Not a strict functional antagonism. The account allows competing categorizations at several levels at once (Turner et al. 1994, p. 456). Hornsey reports a later softening (2008, p. 216).
Bearing
- QUESTION-119 and CASE-104. If an intergroup frame makes extreme members more prototypical, a fanbase’s apparent stance against its owner can be set by its extremes without any sampling bias. That is a derivation for the record to test, not a finding about fans.
- Accounts of the self. The account is psychological. It sits beside THEORY-406 (audience-segregated self-presentation), THEORY-650 (sameness relative to a level of abstraction) and THEORY-691 (the self as a posited abstractum) without entailing any of them.
- Rival: THEORY-1112 (Skorich). It accepts that self-perception varies with context but denies that meta-contrast forms groups. It argues that exhaustive meta-contrast over n people is a 2ⁿ subset-selection problem, and that context effects belong to inference, not perception. It also predicts that comparison-dependent effects can appear with no out-group present, which this account does not allow. An experiment that removes the comparison group would separate the two.
- Scope limit: THEORY-1115 (León). The we-identities of close relationships arise by mutual individualization, not depersonalization, so the depersonalization clause here is not the only route to a “we”.
Note of 2026-10-29: what holds it back
No study that varies only the comparative frame has been read. Oakes, Turner & Haslam 1991 (LIT-1726) and Haslam et al. 1992 (LIT-1767) are filed and Unavailable: no lawful copy was reached, and every result the record holds reaches it through the theory’s authors without data. The promote_when also asks for a result from outside the originating group, or a preregistered replication there, and none is filed. Lea et al. 2007 (NOTE-1268) has the right design but comes from the authors’ group and is not filed.