Essentialized Social Categories I: Gender Essentialism

Psychological essentialism is the belief that kinds have an underlying, probably unseen essence that makes them what they are. We may, for example, believe that tigers have an underlying essence that gives them their stripes and makes them carnivores. We could represent that essence as a particular underlying feature of tigers, such as their genetic makeup, or we could believe that such an essence exists without having any ideas about what it might be. Whether we know what the essence of a particular kind might be doesn't matter. We are psychological essentialists about a particular kind if we believe that kind has an underlying nature that makes it what it is, even if we don't know what that nature is. Psychologists and anthropologists have produced a great deal of evidence that people are, across cultures, psychological essentialists about natural kinds. We aren't, however, indiscriminate psychological essentialists. The evidence indicates that we don't believe that artifacts have essences (or even that they belong to kinds)1. The belief that natural kinds have essences seems to be justified, philosophically (as Brandon noted in this excellent post), but essentialist beliefs about human-made artifacts would not be (for the most part - human-made chemicals might have essences). An interesting is whether we are psychological essentialists about concepts that might be considered as falling somewhere in between natural kinds and artifacts, like social concepts. Are we psychological essentialists about concepts such as gender, ethnicity, political orientation, or mental illness? Do we treat these concepts like natural kinds, or like artifacts, or as something in between? In this and subsequent posts, I'm going to discuss evidence indicating that we are, in fact, psychological essentialists about many social concepts, and that we treat many of them much as we would treat natural kinds. As the title indicates, this first post is about gender.

The first clue that we may be essentialists about social concepts comes from research suggesting that our representations of these concepts are composed of theory-like beliefs (naive theories, as they're often called)2. These naive theories about the relationship between social group membership and behavior allow us to see group members as similar on many dimensions, and imply that we believe that something about being a member of a social group causes that similarity. But we can have theory-like beliefs about a concept without essentializing that concept (treating that concept as a kind with an essence), so in order to show that our social concepts are in fact essentialized concepts, we need more direct evidence. And since there are different types of social concepts, it may turn out that we're essentialists about some, and not others.

Direct evidence that we are intuitive gender essentialists comes from a study by Haslam et al.3. They compiled 40 social categories derived from 20 different social concepts. Examples of the concepts they used, and the respective categories (in parentheses) include age (old, young), class (lower, middle), intelligence (average, smart), personality (introvert, extrovert), religion (Catholic, Jewish), and of course gender (male, female). In their study, they gave each participant 20 of the 40 categories, and had them rate each category on nine different dimensions that corresponded to properties associated with essentialism in the literature. These dimensions were (p. 118):

  • Discreteness: How clear are the category's boundaries?
  • Uniformity: How similar members of the categories are to each other?
  • Informativeness: Does knowing that an individual falls into this category tell you a lot about the individual?
  • Naturalness: Is the category natural or artificial? [Added in an Update, because I left it out. Thanks to Blar for pointing this out in comments.]
  • Immutability: Once a tiger, always a tiger.
  • Stability: Do the properties associated with the category change over time, or remain stable?
  • Inherence: Do the similarities and differences of a category members "correspond to an underlying reality?"
  • Necessity: Are their properties associated with the category that are necessary for category membership?
  • Exclusivity: If you belong to the category, can you also be a member of other categories?

Using principle component analysis, Haslam et al. were able to divide participants ratings into two dimensions: entativity (uniformity, informativeness, inherence, and exclusivity) and naturalness (discreteness, naturalness, immutability, stability, and necessity). They argue that these correspond to two different kinds of essentialism, and their data show that most of the forty categories had high scores on one of the two dimensions. For our purposes, the naturalness dimension is the important one, for two reasons. The first is that it is, as Haslam et al. argue, it is the dimension that corresponds to our essentialist beliefs about natural kinds. The second is that, of the 40 categories, the two gender categories scored the highest on this dimension. Male and female were treated as a natural kinds.

Drawing on this research, Prentice and Miller4 attempted to explore the implications of treating gender as a natural kind. In their two studies, they first had participants complete a task that they were told would measure their "perceptual style." The task involved watching slides flash on the screen for a fraction of a second, and after each slide, estimating the number of dots in it. When the task was completed, they were randomly (i.e., without regard to their actual performance) told that they were "overestimators" or "underestimators."

In the first study, participants either completed the perceptual style task alone (alone condition), or with a member of the opposite gender. If they completed the task with another participant (together condition), they received their feedback (whether they were overestimators or underestimators) together. In one together condition, both participants were told that they had the same perceptual style (together-same condition), and in a second together condition (together-different condition), they were told that they had different styles. Thus, in both of the together conditions, participants had one example of the perceptual style of a member of the opposite gender, which was either the same or different from their own. Finally, participants in each condition (alone, together-same, and together-different) were asked what percentage of each gender shared their perceptual style.

Since people treat male and female as natural kinds, with essences, Prentice and Miller predicted that participants would be willing to generalize from one individual to the entire gender, particularly when they have received information that males and females differ on perceptual style (in the together-different condition). Consistent with these predictions, participants in the together-different condition said that about 60% of the members of their own gender would share their perceptual style, while only 40% of the other gender would. They found the same pattern in the alone condition, though the difference was much smaller (but statistically significant). In the together-same condition, participants' percentage estimates were about 50% for both their own and the opposite gender.

The second study was a bit more complex. In the first part of the study, participants completed the perceptual style task alone, and were then randomly told that they were overestimators or underestimators. They were then given the task again. When taking the task the second time, participants attempted to correct for their perceptual style by either making higher estimates if they were told they were underestimators, or lower estimates if they were told they were overestimators. This shows that under normal conditions, people will attempt to correct for their reported perceptual style. In the second part of the study, another group of participants completed the perceptual style task in male-female pairs or same-gender pairs. After completing the task, they were either told that they both had the same style or that they had different styles, and then completed the pereptual style task again.

The prediction for this study was that if participants believed that their style was a result of their gender, they would be less likely to try to correct for their reported style in the second perceptual style task. Since participants would be more likely to believe that their perceptual style is a result of their gender if the member of the opposite gender in their pair had a different perceptual style, the tendency not to correct would be most pronounced in the different style condition in male-female pairs. In same-gender pairs, participants should try to correct regardless of the style attributed to the other member of the pair. This is in fact what they found. In all conditions but the male-female pair, different style condition, participants tried to correct for their style by either estimating more dots or fewer dots. In the male-female pair, different style condition, participants made almost no effort to correct for their style, indicating that they believed their style was associated with their gender.

To wrap up, while the Prentice and Miller and Haslam et al. studies paint a picture of intuitive gender essentialism that is far from complete, they do tell us two important things: gender is treated as a natural kind, with an essence, and as a result, people are perfectly willing to make generalizations about an entire gender based on one individual if they believe that a characteristic is associated with a person's gender. Furthermore, and perhaps most strikingly, they're willing to believe that a characteristic is associated with a person's gender simply by learning that one male and one female differ in that characteristic. Now, I don't want to get into a debate about the biological basis of gender (as opposed to sex), but the fact that in the Prentice and Miller studies, participants had no a priori reason to believe that perceptual style (which was in fact a made up characteristic) had any association with gender, but still assumed that it did when their own perceptual style was different from that of a member of the opposite gender could have profound implications. It certainly shows that we don't need to have any information about the possible biological basis of gender differences to assume that they exist. It may also underlie the difficulty in overcoming sexism. Using evidence to overcome beliefs that are based in essentialism, and for which there is little or no evidence in the first place, can be very difficult.

1Sloman, S.A., & Malt, B.C. (2003). Artifacts are not ascribed essences, nor are they treated as belonging to kinds. Language and Cognitive Processes, 18(5/6), 563-582.
2Yzerbyt, V., Corneille, O., & Estrada, C. (2001). The interplay of subjective essentialism and entitativity in the formation of stereotypes. Personality and Social Psychology Review, 5(2), 141-155.
3Halam, N., Rothschild, L., & Ernst., D. (2000). Essentialist beliefs about social categories. British Journal of Social Psychology, 39, 113-127.
4Prentice, D.A., Miller, D.T. (2006). Essentializing differences between women and men. Psychological Science, 17(2), 129-135.

More like this

A couple weeks ago, a couple Science Bloggers, sparked by Jessica of Feministing, discussed the potential dangers of discovering the biological causes of homosexuality. Jessica expressed a common attitude in her post, writing: And naturally the larger question with all these why-are-you-gay studies…
Originally posted on the old blog on 3/8/2005, and reposted here out of laziness. The Importance of Names What's in a name, for a concept I mean? Cognitive psychologists studying concepts and categorization have, notby and large, treated concept names (often called "category labels") as just…
Originally posted on the old blog on 3/12/2006 My contribution to Darwin Day was pretty weak for a staunch supporter of science. Sure, I think the name is a bad idea, and want to rename it "Evolution Day," or at least something other than Darwin Day (I thought about maybe suggesting "Variation…
There's an interesting short paper by Paul Bloom and Susan Gelman in the July issue of Trends in Cognitive Science with that title. Unfortunately, it's not yet available without a subscription (though Bloom tends to put his papers on his website once published, so it might show up there sometime in…

It looks you forgot to include the definition of "naturalness" as one of the nine dimensions.

In the Haslam et al. study, did they make it clear that they were asking about gender rather than sex? Because, it looks like all four of the naturalness dimensions that you define (discreteness, immutability, stability, and necessity) would get very high ratings for sex, and if they didn't clearly identify male/female as cultural gender categories rather than biological sex categories then that could account for the high naturalness ratings of gender.

Blar, good point. They simply provided the names of the categories. However, I think the Prentice and Miller study addresses that directly: people have no reason to assume that a characteristic they've never heard of (and which doesn't actually exist!), perceptual style, is associated with biological sex. It appears that people actually equate gender and sex.

I agree that the Prentice & Miller study is nicely done, and good indirect evidence for gender essentialism. I'm not sure if the Haslam et al. study adds much to the evidence, though. Maybe the point just is that people don't distinguish sex and gender, but their method seems to depend too much on the words "gender", "male", and "female", and all three of those words are often used to refer to biology (especially the last two). Just think about how frequently people fill out forms that ask them "Gender: M F " and are expected to find this a ridiculously easy question. Maybe if subjects were asked to rate the category sex (male, female) on those 9 dimensions before they were asked to rate gender (masculine, feminine), then the meaning of high naturalness scores for gender would be less ambiguous.

Blar, that's actually a good idea, though I still think the Prentice and Miller study covers some of the problems in the Haslam et al. study.

You know, it would be pretty easy to do a study like the Haslam et al. one online, and get quick data. Maybe I'll try it with "masculine" and "feminine."

This is a very interesting tool. This is a valuable analysis for dealing with contorversial issues about which conventional thinking is at odds with science. If one has already found themselves asking questions about the exact nature and make-up of some "kind" are they, in effect, beyond the pall of essentialism with respect to that kind?

Can this analysis be used to clarify how we think about outgroups, minorities, enemies, foreigners...or perhaps it already has and you could point me to the write-up?

Greensmile, there's actually a growing literature on just the stuff you mention. Haslam is really the main researcher so far, and you can search using his name (or use the cites here and in the post on homosexuality). Also, Gil-White wrote an interesting paper on essentialism and ethnicity, though I don't have the exact cite at the moment (the paper is online, I think, without a subscription). I'm away from my own computer, but when I get back to it, I can send you a longer bibliography.

Thanks Chris. That is good news. Bora left me a few more things to read than I have time for today but I will go a googling and check back later.