Contribution
A short theoretical statement, from the theory’s authors, of what self-categorization theory says about the self. It adds to the 1987 book (LIT-1723, not read) a catalogue of four ways self-categories vary with context, drawn from the group’s studies of 1987–1994. It also adds an argument from that variability: the self is better treated as a product of judgment in context than as a fixed, stored structure that context activates.
Key insight
Being a “we” is not a departure from the self but one of its normal settings. Whether I see myself as “I” or as one of “us”, which “us” it is, what that “us” means and who in it is most typical all depend on whom I am being compared with. They are computed afresh from the comparison at hand by one principle, metacontrast. A woman who is the only female psychologist in a room may define herself as a woman against the men, or as an individual against the other psychologists (p. 455). Neither definition is the real one.
Assumptions
- Categorization is comparative. A self-category is a grouping of self and some class of stimuli as identical, in contrast to some other class (p. 454). It is always relative to a frame of reference.
- Salience = readiness × fit, following Bruner (p. 455). Readiness
(relative accessibility) reflects past experience, present
expectations, motives, values, goals and needs. Fit has two parts:
- comparative fit, the metacontrast principle: a collection of stimuli is more likely to be categorized as an entity to the degree that the average differences perceived between them are less than the average differences perceived between them and the remaining stimuli in the frame of reference;
- normative fit: the differences also lie in the direction, on the content dimensions, that background beliefs about the category predict. Catholics must differ from Protestants in the right way, not just differ.
- Fit is part of the explanation, not all of it (p. 456). The authors’ work is mostly on fit. They say outright that fit is not a sufficient explanation of salience.
- Personal versus group level is a continuum of competing categorizations (p. 456, after Turner & Oakes 1989). What is salient is “rarely a single category”; talk of one dominant self-category is shorthand.
Key results
The paper reports results; it presents none. Each is cited to the group’s own work, several of it in press in 1994.
- Fit predicts salience (p. 455–456). Oakes, Turner & Haslam (1991): subjects watched six arts and science students discuss. Arts/science identities were most salient, judged by attributions to academic field, when the students disagreed along category lines and the views matched the stereotype, that is, under both comparative and normative fit.
- Four forms of context dependence (pp. 456–458).
- Level. Self-categorization becomes more inclusive as the comparative context widens to include people more different from self, so prior others are recategorized as “us” (Gaertner et al. 1989; Haslam & Turner 1992). Social identity is more salient in intergroup than in intragroup comparison: in staged discussions, men and women self-stereotyped more by sex when the disagreement ran between the sexes (Hogg & Turner 1987). The out-group homogeneity effect is explained the same way, and relative in-group homogeneity is obtained when judgments are intergroup on in-group dimensions (Haslam, Oakes, Turner & McGarty, in press). People extreme within a context categorize self and others more sharply and polarize, and the classic extremist/moderate pattern can be reversed by changing their position relative to the majority (Haslam & Turner, in press).
- Kind. Which category becomes salient depends on normative fit (Oakes et al. 1991).
- Content. What a category means is selectively constructed for the comparison. In the “Iraq” study (Haslam et al. 1992), Americans were seen as more aggressive than the Soviet Union but less aggressive than Iraq, and their rank for ambition against Britain, Australia and Iraq reversed between the start and the end of the Gulf War. Australians judged alone were “happy-go-lucky, straightforward, and sportsmanlike”; judged beside Americans, more sportsmanlike, less happy-go-lucky, and “pleasure loving” appeared. “There is no one-to-one correspondence between the long-term knowledge one has about what different kinds of people are like and the actual social category that is constructed” (p. 457).
- Internal structure. Relative prototypicality shifts with context (cf. Barsalou 1987). As a group becomes more extreme in an intergroup context, its more extreme members gain prototypicality over its moderates (p. 458). In-groups can define themselves in opposite ways on different occasions (Hogg, Turner & Davidson 1990).
- Three conclusions about the self (p. 458). Its variability is lawful, not arbitrary; self-categories are “social contextual definitions of the perceiver”; and they are veridical, since variability is “necessary if self-perception is to be veridical”.
- The self as generated (pp. 458–459). “We can speculate that self-categories are reflexive judgments”. The authors “doubt whether the idea of self as a relatively fixed mental structure is meaningful or necessary”. What makes a concept a self-concept may be its use, reflexively to define the perceiver, not where it is stored. Stability is explained by stable social reality, stable knowledge frameworks, stable groups and institutions, and influence processes (p. 460).
- Influence (p. 461). Private acceptance is “private category change”. The same information is accepted from an in-group source and rejected from an out-group one. A minority’s latent influence depends on its being categorized as in-group, and the same minority can be either, depending on the comparison (David & Turner 1992).
- Cognition (pp. 461–462). The variable self sets what is relevant to the perceiver, which background knowledge is normative to apply, and whether another’s judgment counts as agreement or disagreement. Hence the closing claim that “all cognition is social cognition”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Personal and social identity are different levels of self-categorization, and the social level is as valid and authentic an expression of self as the personal | definitional; argued, not tested | pp. 454, 460 |
| C2 | Shared social identity depersonalizes self-perception (self seen as an interchangeable representative of the category), and this explains group phenomena | moderate as reported: rests on the 1980s programme the paper summarizes | p. 455, citing Turner et al. 1987, Hogg 1992, Turner 1991 |
| C3 | Salience of a self-category is an interaction of perceiver readiness and fit; comparative fit is metacontrast, normative fit is content-consistency | moderate as reported: one study described (Oakes et al. 1991); the authors call fit only part of the account | pp. 455–456 |
| C4 | The level, kind, content and prototype structure of self-categories all vary lawfully with the comparative frame of reference | moderate as reported: several studies by the authors’ group, summarized without data; not independently checked here | pp. 456–458 |
| C5 | Variation in self-categorization is veridical, tracking changes in social reality, not a distortion of a true self | weak: argued from C4; no criterion of veridicality is given | p. 458 |
| C6 | The self is not a stored structure that context activates but is generated in context; a self-concept is defined by its reflexive use | weak: explicitly speculation, supported by analogy to Barsalou 1987 and Murphy & Medin 1985 | pp. 458–459 |
| C7 | Social influence is private category change, and its success depends on the source being categorized as in-group | moderate as reported: group-polarization and minority-influence studies cited | p. 461 |
Concepts
- self-category: a cognitive grouping of self and some class of stimuli as identical and different from some other class (p. 454).
- personal identity / social identity: self-categories that define the individual by differences from other (in-group) persons, versus by shared similarities with members of a category in contrast to other categories (p. 454).
- depersonalization: defining and seeing oneself less as a differing individual and more as an interchangeable representative of a shared category (p. 455). Not a loss of self.
- perceiver readiness: relative accessibility of a category (p. 455).
- metacontrast / comparative fit: see Assumptions.
- normative fit: match between category specifications and the content of the instances (p. 455).
- private category change: the authors’ name for what the influence literature calls private acceptance, conversion or informational influence (p. 461).
- cognitive choice: selective representation of phenomena from the vantage point of the perceiver (p. 461).
Connections
- The book. It restates and extends Turner et al. 1987 (LIT-1723), which the record holds but has not read.
- Recategorization. Its first form of variation cites Gaertner et al. 1989, “the benefits of recategorization”, the source of the common in-group identity model. That is the model Biel (LIT-1133) names, and the recategorization strategy in Pettigrew’s review (LIT-1408). Both are unread.
- Categories in general. It borrows Barsalou’s (1987) instability of graded structure and Murphy & Medin’s (1985) theory-based categories. Neither is held.
- Rivals it names (pp. 459–460). Social-cognitive self-schema models (Markus & Wurf 1987), which also allow the working self to vary but keep a stored repertoire. Symbolic interactionism and Stryker’s identity theory, where self comes from reflected appraisal and roles. The paper reverses the order and subsumes role under group.
- The record’s accounts of the self. It is a psychological, not a metaphysical, thesis, but it sits near several Proposed THEORYs. THEORY-406 holds that self-presentations differing by audience collide when audiences merge; this paper would say the frame of reference has changed. THEORY-650 makes self-sameness relative to a level of abstraction. C6 is close in spirit to THEORY-691 (the self as a posited abstractum) and opposed to accounts of a stored self-model (THEORY-064). The paper argues none of these, and nothing here settles them.
Bearing on the record
- It is the source of THEORY-1114, filed Proposed from this reading and Hornsey’s review (NOTE-1253). That THEORY takes C3–C4, the testable core. C6 is left as a note here, since the authors offer it as speculation.
- CASE-104 and QUESTION-119. The prototype result (p. 458) bears on the case of a fanbase credited with a stance. In an intergroup context, such as fans against an owner, the more extreme members are predicted to become more prototypical, and depersonalized members converge on the prototype. On this theory a vocal minority setting the public stance is not just a sampling bias; it is what the intergroup frame does to whom the group treats as typical. That is an inference from the paper, not a claim it makes about fans.
- THEORY-126. Smith’s group-based emotions, one route in that account, are built on this theory (Hornsey 2008, p. 216).
- No ML instruction.
Limitations
- No data. Every result is reported from the authors’ own programme, and several were in press. The reading checks what the paper says, not what the studies found.
- “Lawful” and “veridical” are claimed, not specified. No criterion says when a self-categorization is veridical; and since readiness takes in motives, goals and needs, fit alone does not fix the prediction.
- The theory explains its findings after the fact more easily than it predicts them. With readiness, two kinds of fit, and categories competing on a continuum, many outcomes can be fitted. Hornsey (2008, p. 217) reports the unfalsifiability charge, against the wider approach.
- C6 is speculation by the authors’ own word. It is argued from analogy to category research, not from self data.
Open questions
- Do the content and prototype shifts replicate outside the authors’ group, with the comparison frame manipulated and the target held fixed? A primary study read in full would settle it for the record.
- Can C6 be distinguished empirically from a stored-repertoire account (Markus & Wurf’s working self-concept) at all? The paper concedes the similarity and claims “real and substantial differences” it has no room to spell out (p. 459).