Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Psychology. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 16 June 2024

Social Categorizationfree

Social Categorizationfree

  • Craig McGartyCraig McGartyDepartment of Psychology, Western Sydney University


Categorization is a process whereby we make sense of the world around us by separating things into different classes or groups. When we learn which categories that objects belong to, we also learn about relationships between those objects. Social categorization involves applying that same process to people, including ourselves. It is not only a cognitive process for understanding and explaining the world, but it is part of the way we organize the world. That is, the groups we belong to such as genders, ethnicities, religions, and nations are based on social categories, and thus phenomena such as stereotyping and person perception rest on social categorization. The study of social categorization has drawn heavily on the study of object categorization and many of the core insights from that field are relevant, but there are also some important differences that suggest social categorization is more, indeed much more, than object categorization. The first key difference is that social categories, unlike object categories, are made up of people who can choose to unite or divide. Social categorization can help us not only to understand why other people are similar to each other and different from us but also to predict when they will be similar and different to us. The second key difference is that when we categorize ourselves, we learn who we can cooperate with, who shares our goals and interests, and who we might cooperate with. It is hard to imagine effective human functioning without the abilities that social categorization grants us.


  • Social Psychology

Social Categorization: “You Know It Makes Sense”

We understand the world and explain it to ourselves, and to those around us, by working out how, why, and when things are the same as, and different from, other things. Indeed, until we take this step and unleash the powerful cognitive process of categorization we struggle to do any more than call things, “things” really. If all we know is that things are things then, in effect, we know nothing. As we do start to find out that one thing is like another and different from a third, our stock of knowledge grows rapidly. If we learn that two things have a feature we do not just gain two pieces of information (one about each thing): We gain, at the very least, the additional information that both those things are the same in at least one way. That is, relationships between those things can begin to be detected, and we can use that knowledge to predict the property of other things that we have never seen.

Imagine for a moment that those things are berries growing on plants that might be collected by hunter-gatherers. What chance is there that any of us would be here now if our ancestors had treated all berries as equally good to eat or equally dangerous? In quite direct terms, your capacity to sit and read this article rests on the good sense of your ancestors in telling apart good berries from bad berries and remembering which ones were good and which ones were bad. In other words, the survival and emergence of human beings has rested on the ability of humans to categorize and to retain information in categorical form. The same is true of birds and other creatures, but looking at this example a little bit more deeply we can see one of the reasons why social categorization is such an important topic. If you were dropped in some isolated, forested part of the world in the early 21st century, you might find plenty of berries. Your chance of surviving in a lot of those places would be massively increased if somebody told you which berries were poisonous. The point, of course, is that human beings have an extraordinary capacity not only to generate and store categorizations but also to communicate them so that we come to share the same categorizations.

Let us take a step further. Imagine in your new hunter-gatherer existence you received different advice on safe berries from two different people. Who should you listen to? Should you pick one adviser at random (they are both people) or ignore both? If you had some way of distinguishing between the sources in terms of their knowledge of berries, their trustworthiness, and the extent that they were genuinely interested in helping us (that they were on our team) then that could be incredibly helpful. If we can categorize people into the group of people we can trust to advise us on which berries to eat, and which people cannot be trusted, we can use that social categorization to help us decide that. The point here is that we categorize people in much the same way that we categorize other things. These social categorizations are used to form stereotypes that shape understandings and expectations that can inform our interactions with other people across time. There are divisions and barriers between people, and these divisions and barriers are no less real than the divisions that exist among physical objects. Berries may not have a lot of individuality; they may seem easy to categorize; but unlike people individual berries do not work with other berries to follow rules, form alliances, and exclude outsiders. Furthermore, the ability to categorize people into groups (especially into groups of people we can trust to accept information from that can help us to stay alive) is useful for every person who has ever lived. We should expect that the ability to create meaningful categories of people will be universal across cultures because the ability is as useful for hunter-gatherers as it is for people living in industrialized societies.

There is, however, an even deeper layer. People do not just categorize other people, they categorize themselves. We acquire a sense of who we are at times by contrasting ourselves from some others, and at other times by recognizing the similarities.

Hence these are the three faces of social categorization. Categorization is the process of coming to understand a thing by perceiving it to be similar to some things and different from other things (McGarty, 1999). Social categorization is the process of categorization applied to people and/or as shaped by interactions with people. Categorization, even the categorization of physical objects, can be a social process that rests on learning and sharing categories. As I noted (McGarty, 1999), the precondition for this socially mediated sharing is mental flexibility. If categories and categorizations were fixed and unchangeable, people could never reach consensus because they would be locked in irreconcilable disagreement. Kurzban, Tooby, and Cosmides (2001) show that even categories that seem highly established such as race can rapidly be cast aside when they do not help people solve the problems they confront. However, if categories did not have the prospect of enduring over time, there would be no point in sharing them through language. Second, social categorization is the cognitive process of perceiving the groupings that exist in society (Hamilton & Sherman, 1996). Third and finally, social categorization includes self-categorization: finding out who you are.

Lessons From Object Categorization

The study of categorization is normally traced back to the so-called classical view attributed to Aristotle. Categories in the world were widely believed to have fixed all or none or defining properties. This was the logical basis of taxonomy, the classification of living organisms, whereby tree diagrams could be drawn separating out kingdoms, phyla, genera and species, and so on in an orderly manner. In the early 20th century the philosopher Wittgenstein (1974/1953) illustrated that even seemingly simple, widely used categories did not have all or none defining features or properties. Wittgenstein’s thought experiment used the example of a game. Nobody has been able to list a single set of properties that unite all the things that we call games (think of things as varied as preschool children’s amusements and the Olympic Games).

In the 1970s, through the work of Rosch (1978) and others, it became apparent that categories have graded structure. Specifically, not all of the things that are regarded as members of a category tend to be seen to be equally good examples of the category. Some members are seen to be more typical members of a category, and there are also family resemblances—members of categories share features or positions on a dimension (Rosch & Mervis, 1975). For what it is worth, this seemed to be true of many categories of things in the world and as well as categories of the mind. Bats and penguins are less good examples of mammals and birds than are dogs and sparrows. Knowing that categorical systems could have a hierarchical structure and varying levels of inclusion much like the tree diagram of biological taxonomy provides an incredibly powerful basis for logical inference. If I know that all the members of some subcategory have a property (such as poison in a berry), that members of other subcategories do not, then I can use that knowledge to interact reliably with the world.

In other words, categories turn out to be fuzzy (the logical basis for this was established by Zadeh, 1965), and that fuzziness can be expressed in two main ways. The first was the prototype view most closely associated with Rosch (1978). Categories were believed to be based around an abstract representation of a typical member (the prototype). The question of what belongs to the category reduces to asking which things match or are close to the prototype. The other is what has come to be known as the exemplar view (most closely associated with the work of E. E. Smith & Medin, 1981, but clearly prefigured in other work, and observed in social psychology in the work of E. R. Smith & Zarate, 1992, and others). Categories can be thought of as sets of discrete instances of which traces of information about things and events in the world are stored in memory. Under this view the problem of defining what a category is reduces to specifying which instances belong to it.

The debate over the prototype and exemplar views is ongoing and both views are highly relevant in social psychology. At the risk of oversimplifying some complex issues, it is likely that both views are relevant under different circumstances (analogous to the consideration of light both as a wave or a particle).

The second insight from object categorization is that categorization is a motivated process. The New Look in perception championed by Bruner (1957) demonstrated that perception was governed by goals and past experience. What people saw was determined by what they were ready to see and with what fitted with their expectations. Unlike the prototype-exemplar debate where the concepts were imported into social psychology from cognitive psychology, social psychologists were working at the core of these insights before and during the New Look (as can be seen in the work of Sherif, Campbell, and Tajfel and others who drew on the Gestalt perspective).

The third and final great insight from the study of object categorization was that there is more involved in binding together categories than the proximity of members on a set of dimensions. People use knowledge about categories in their judgment of category membership. For example, research shows that a disc of 12cm diameter is more likely to be judged to be a pizza rather than a coin even though it is equally distant from the mean diameter of each category, presumably because people know that coins are regulated by law to be specific sizes whereas pizza size can vary with choice (Rips, 1989). Research by Tversky and Gati (1978) showed that people regarded Mexico as being like the United States but that the United States was not seen to be like Mexico. It seemed that establishing the order of comparison changed the nature of the comparison by activating different knowledge bases. This idea that similarity is not a constant fixed relational property but changes with knowledge has been championed in various ways in cognitive psychology in the work of Medin (e.g., Medin, Goldstone, & Gentner, 1993) and others.

Why Do We Make Sense Through Social Categorization?

There are two key answers to the question as to why we categorize. The first most clearly associated with the work of Taylor and Fiske (1978) is that people categorize in order to conserve meagre cognitive resources by simplifying. That is, rather than retaining and storing all the information that is required to treat aspects of the world as unique exemplars it is easier, quicker, and simpler to group things, to deal in generalities. Such generalization comes at the cost of distortion and inaccuracy but is preferable to cognitive overload.

The alternative championed by Oakes, Haslam, and Turner (1994) is that in social perception we are confronted not by too much information but by too little. The power of categorization lies in the word of Bruner to go beyond the information given. Generalizations under this view are not simplifications involving a loss of information but additions to the stock of knowledge that come about by identifying relationships that, once created, allow deductions to be made.

These views have come into sharp contest over the years in relation to a variety of topics such as person perception, outgroup homogeneity, and stereotype formation (for a review, see McGarty, 1999). The first cognitive miser view has widespread support in the pages of most social psychology textbooks and remains the orthodox view. There have been relatively few tests of the competing views (most recently by Skorich & Mavor, 2013) but those published suggest that there are grounds for further and more detailed tests.

How Do We Make Sense Through Social Categorization?

A proposal by Turner, Hogg, Oakes, Reicher, and Wetherell (1987) that is based on ideas of Campbell (1958) and Rosch (1978) is that social categorization proceeds through a process of meta-contrast. When we look at a set of stimuli (including social stimuli such as people), we are most likely to perceive them to be a group when the differences between those things are less than the differences between those things and other things. Turner et al. postulated that people used a meta-contrast ratio (technically a quotient): Within some comparative context (i.e., a set of things to be compared to each other), we can calculate the average difference between things classed in one group and divide that by the average difference between things within that group. The idea is that we are constantly performing this kind of mental arithmetic (unconsciously) and that the average meta-contrast ratio of various group members is informative of the degree to which those members are seen to be prototypical (representative or typical) of the group.

According to self-categorization theory (SCT), prototypicality of this sort underlies a vast array of processes and effects that are relevant to social categorization. Just to name a few, the most prototypical group members of our own groups are expected to be the ones who are most preferred or liked, and they are also the members of a group who are most likely to be influential.

SCT used the idea of meta-contrast to reboot Bruner’s accessibility X fit formulation. SCT argued that meta-contrast defined the degree of (comparative) fit of stimuli to the category specifications. Comparative fit was complemented by normative fit or the degree to which behavior was consistent with the social meaning of the category. For example, if we see various members of the Ku Klux Klan all behaving supportively toward an African American, that could be an example of high comparative fit but it is low normative fit (we do not expect Klan members to behave in that way). SCT also identified an additional factor (that came to be termed perceiver readiness; Oakes et al., 1994; Turner, Oakes, Haslam, & McGarty, 1994) to update the concept of accessibility. There are some categories we are ready to see on the basis of expectations derived from needs, values, and past experience. We might expect many young, underemployed, male minority members in a lower socioeconomic inner-city neighborhood to be organized into violent criminal gangs. We are unlikely to expect the same organization in a retirement village or preschool. Are we being unfair in the first case? Are we not just stereotyping in a way that reflects cultural biases and political standpoints? It is to these issues that we now turn.

Making Sense of Other People: Stereotyping

Stereotypes involve using social categories to predict behavior and explain social structure. As Haslam, Turner, Oakes, Reynolds, and Doosje (2002) observe, stereotypes are tools, and potentially powerful tools at that. This is because categories of human beings have shared properties and we can use those properties to make predictions about members. Thus, if we know that (U.S.) Americans tend to speak English, then we use that to predict that an American will understand our communications in English and that Americans can communicate to each other in English. We may also know that it is true that some Americans (e.g., young children, some immigrants) do not speak English, but our stereotype is going to help us function pretty well when we interact with Americans (and if we visit the USA, we will not need to learn a language other than English or carry a phrasebook). The stereotype that people in Japan speak Japanese (and are less likely to be fluent in English than Americans) would serve equally well in Japan. The alternative to stereotyping here would be to avoid making assumptions, to treat each person as a unique individual, and to ask each new American we encounter if they spoke English and each new Japanese person if they spoke Japanese. If we were to do this we would rapidly find that many of the people we interacted with became puzzled or infuriated. Stereotyping helps us avoid that sort of problem, so what is there not to like about stereotyping?

Stereotypes can also have negative consequences and connotations. Imagine that we knew that nearly all Japanese people learn English at school and that they were likely to have some command of the language. Imagine then that we spoke English to all Japanese people we met in Japan and assumed that all Japanese people, being courteous, would be willing to reply to us in English. There is a good chance that the stereotypes described in the previous sentence could be misleading and offensive. The stereotypical expectations could lead us to make significant mistakes because they would oversimplify the complexities of language learning and competency, and would violate standards of cultural appropriateness. Many people in many parts of the world regard it as rude to address people in that country in English without first attempting to communicate in the local language.

This example points to the core concern in research on stereotyping. It has long been believed that stereotypes are rigid distortions of reality. Although they may contain a kernel of truth, they tend to be gross oversimplifications that tend to lead to unduly negative views of other groups. There is extensive evidence of negative and destructive stereotypes and the foundational work of M. Sherif, Harvey, White, Hood, and C. Sherif (1961) suggested that the process of group formation (especially under conditions of competition for scarce resources) could trigger the rapid formation of negative stereotypes. Dramatic earlier evidence of this was shown by Meenes (1943) who found that the surprise attack on Pearl Harbor in 1941 triggered a change in stereotypes held by Americans of Japanese descent (who had in 1935 been seen as polite and reserved and were now regarded as deceptive and aggressive).

How can we reconcile the functional and adaptive and sense-making aspects of categorization that we see in object categorization with the distorted excesses of social stereotyping? Botanists use categories to describe plants. We do not have any reason to believe that botanists become less fond of some plants and more fond of others simply because they have applied the (categorizing) principles of taxonomy to the plants. Part of the answer is to ask whether the categorization of people really is distorting perceptions of people or it is accurately reflecting the relations between groups in a dynamic way. Japanese people in 1942 really had just launched an attack on the United States without effectively delivering a prior declaration of war. For Americans in 1942 it was true that Japanese people were sneaky and aggressive.

Stereotypes therefore are powerful applications of the categorization process because they capture social dynamics. Human social groups can diverge and cohere. The members can become more similar to each other by regulating their behavior through the development of social norms, rules, laws, and customs and by excluding outsiders. Stereotypes help capture these complex relations by drawing on sophisticated categorical systems that need to be open to change and modification in the face of new information.

Stereotyping, or the way we make sense of other groups, can also be more or less automatic. Some of our stereotyping processes appear to be implicit and not available to consciousness or to conscious control (Greenwald, McGee, & Schwartz, 1998). A plethora of studies of the implicit association test, on the one hand, suggest that there are unconscious biases against members of outgroups even by people who are low in prejudice (Greenwald & Banaji, 1995). On the other hand, some stereotypes seem to be readily available to consciousness and thus it should be possible to actively suppress stereotypes (Bodenhausen & Macrae, 1998).

The study of unconscious stereotypes and bias has been a major activity in relation to social categorization. The available evidence remains at the point of accumulation rather than integration. There has been a rise in interest in subtle and implicit forms, but that should not be taken as an indication that these phenomena are new as opposed to merely being undiscovered, or that blatant hostility to members of other categories does not exist. There is plenty of contemporary evidence of blatant and hostile prejudice that seems very deliberate.

Making Sense of Who You Are: Self-Stereotyping

The final powerful aspect of social categorization is making sense of self (Leonardelli & Toh, 2015, have recently proposed that self-categorization itself comes in three forms). SCT argued that people stereotyped themselves. That is, thinking of yourself as a group member entails stereotyping yourself as a member of that group. Self-stereotyping was believed to underpin the perception of interchangeability between self and other group members (a phenomenon that was named “depersonalization” in SCT, inviting unfortunate confusion with the unrelated construct of deindividuation, Zimbardo, 1969, that is replete with negative connotations that are beyond the scope of this article).

Self-stereotyping was postulated to occur when the salience of a social categorization (determined by the perceiver readiness X fit interaction) was high. That is when a specific social categorization was salient a person would think of themselves as a group member, and act as one, and when no social categorization was salient then that same person would think of themselves, and act as, unique individuals. The hydraulic operation of these two levels was named the principle of functional antagonism.

It is a powerful and relatively simple idea. People act in the same way as others not just because they are in the same situation, respond to the same stimuli, share the same attitudes, or encounter the same rewards and punishment as those others, but because they perceive themselves as interchangeable with those others. This idea has underpinned mountains of work in areas such as intergroup emotions (Mackie, Smith, & Ray, 2008), and collective action (van Zomeren, Postmes, & Spears, 2008).

Despite these productive research outcomes, functional antagonism has encountered a range of theoretical alternatives and challenges. Brewer (1991) proposed optimal distinctiveness theory that postulates that the level of categorization adopted by social perceivers is determined by the balance of pressures to be distinctive and to be inclusive. The theory and much of the evidence it has generated is inconsistent with functional antagonism. Another line of research has consistently pointed to the role of self-anchoring as an alternative process to self-stereotyping (see van Veelen, Otten, Cadinu, & Hansen, 2016). That is, there is evidence that the personal self may serve as a reference point for judgment even under conditions of high salience. More generally, Postmes and Jetten (2006) and others have pointed to the plurality and diversity of individual and group relations. One illustration is the work of Hornsey and Imani (2004), showing that internal criticism within an ingroup has a different force, and is perceived in a different way, from criticism from outside the group. Other work suggests that multiple categorizations (see Crisp & Hewstone, 2007), rather than a single salient categorization, can be beneficial for health and well-being (e.g., Steffens, Cruwys, Haslam, Jetten, & Haslam, 2016). Generally speaking, it appears that there are consequences for self and social perception that ride on the personal self. This is, perhaps, not surprising given the qualities of agency and consciousness associated with the continuity of the person as a biological-psychology entity across fluctuations in salience. These are complex issues, and it is apparent that a deeper and more sophisticated level of theorizing of social categorization is required.

Another ongoing controversy is the issue of distinguishing social categories from groups. These terms are often used interchangeably. This is semantically problematic, but there is a deeper conceptual problem if there is a functional disjunction between social categories and groups. Gaertner, Iuzzini, Witt, and Oriña (2006) have assembled evidence that groups can form in the absence of categorization (reference to the exclusion of people outside the group). The question this raises is whether there are categorization processes occurring but were not detected by Gaertner and colleagues (perhaps because the processes were implicit) and if not whether categorization-free “group” formation can also occur for aggregations of nonsocial objects (not containing the perceiver). If so this has wide ramifications for categorization. Balanced against this, Postmes and colleagues’ interactive model of identity formation (2006) emphasizes emergent self-categorization. The key element of interest here is that group formation can be both top-down and bottom-up. That suggests that groups can both follow from the ascription of an existing identity and from the crafting of new identities through social interaction. Postmes et al. propose that the latter path is critical for social change, a point followed up in the INN-formation model of Smith, Thomas, and McGarty (2015).

Future Directions

I will conclude by identifying a few areas of growth and future development. Not all of these developments are new by any means, but in several cases they exist as fresh opportunities waiting for development.

The first is the development of what can be termed ecological categorization, for example, in the work of Fiedler, Freytag, and Meiser (2009). In various ways these authors have demonstrated that effects that look like the outcome of a categorization process can arise from the ways that people sample from complex arrays of stimulus information. It is difficult to do justice to the richness of this approach in the space available but to put it simply: Sometimes we have difficulty telling forests from trees because of the context provided by the stimuli. Lots of trees (or people) afford different bases for perceiving contingencies than a few people, and some of those contingencies can be pseudo-contingencies that might never be perceived when encountering small samples. It can be tempting to interpret such perceived contingencies as outcomes of active, thoughtful categorization processes when they are not.

Secondly, the question of the stability of categories and category systems across time remains underexplored. Are categories stored and retained or are they generated afresh online as required? If the latter view is true, it implies that categories may be best thought of as epiphenomena that reflect processes of information integration. This would, however, seem to be overstated for some key reasons. Categories help form concepts, and there are powerful reasons to believe that concepts will be retained in ways that involve their rapid deployment in symbolic forms. The most obvious reason is that we need to talk about categories with other people, especially if those categories reflect who we (and they) are and who we are not.

Third, there is expanding scope for considering the different building blocks of social categories. Most attention has focused on nominal sociological categories such as sex, age, ethnicity, nation, and religion, but there are also categories based on institutional membership, shared opinions, and shared emotions. Categorizations across time and situation are also logically possible.

Fourth, Skorich et al. (2016) have recently proposed a central role for social categorization in autism. That is, the social interactional difficulties experienced by people on the autism spectrum may be explained by difficulties in categorizing people.

Finally, the limits of hierarchical systems as bases for categorical systems continue to demand further inquiry (McGarty, 1999, 2006). Orderly, inclusive hierarchies are powerful tools for logical inference, but, in reality, categories in the world and categories in the mind rarely seem to have such a neat structure, and even when they do people do not necessarily use them in ways that reflect that structure. This issue is most marked for social categories. Many categorizations are cross-cutting so what is a subgroup in one context is a higher level category in another. Thus, in an international meeting of feminists the category “American” will be a subgroup but in a meeting in America the subgroup may be “feminists.” If there is one point that captures the intent of this article, then it is to say this: Whether the meeting participants see themselves as “Americans” or “feminists” is unlikely to be rigidly determined by context, but rather that it will be flexibly determined by the participants in line with their goals and assumptions. If it is clear why I say that, then my work here is done.