(PDF) Sex Differences in Semantic Categorization

Arch Sex Behav DOI 10.1007/s10508-011-9764-y ORIGINAL PAPER Sex Differences in Semantic Categorization Vickie Pasterski • Karolina Zwierzynska • Zachary Estes Received: 2 November 2010 / Revised: 21 March 2011 / Accepted: 28 March 2011 Springer Science+Business Media, LLC 2011 Abstract Sex differences in certain cognitive abilities, includ- Introduction ing aspects of semantic processing, are well established. How- ever, there have been no reports investigating a sex difference in Sex differences in certain aspects of cognition have been estab- semantic categorization. A total of 55 men and 58 women each lished across domains and cultures (Kimura, 2002). Men and judged 25 exemplars of natural categories (e.g., FRUITS) and 25 of women display, on average, different patterns of task performance artifact categories (e.g., TOOLS) as a nonmember, partial member, in domains such as spatial orientation and verbal or perceptual or full member of the given category. Participants also rated skills (Kimura, 2002). For instance, men tend to excel on tasks of confidence for each judgment. Women provided a greater num- spatial ability such as mental rotation (Linn & Petersen, 1985; ber of vague (partial member) judgments whereas men provided Voyer, Voyer, & Bryden, 1995), whereas women tend to excel more inclusive (full member) judgments of artifacts but more on tasks of linguistic ability such as verbal fluency (Kolb & exclusive (nonmember) judgments of natural categories. The sex Wishaw, 1985; Spreen & Strauss, 1991; for review, see Hines, difference in vagueness was observed across domains (Cohen’s 2009). Although there is substantial overlap between male and d = .56). Confidence predicted categorization among both men female performance and cognitive abilities may be better char- and women, such that more confident participants exhibited acterized by between-sex similarities rather than differences fewer vague category judgments. However, men and women (Hyde, 2005), studying sex differences in cognition is important were equally confident in their category judgments, and confi- as it may elucidate mechanisms underlying sex-differentiated dence failed to explain the sex difference in categorization. Men behavior, which, in turn, may inform our understanding of sex and women appear to categorize the same common objects in sys- stereotypes (Hyde, 2007). tematically different ways. One aspect of cognition that has not been investigated in terms of a sex difference, but could have broad implications, is semantic Keywords Artifacts and natural kinds categorization. Indeed, several studies indicate a sex difference in Semantic categorization Sex differences processing natural categories (categories occurring indepen- dently of human production or intention, such as FRUITS) as well as artifact categories (categories occurring by human production or intention, such as TOOLS). Whereas women name natural objects faster and more fluently, men name artifacts faster and more flu- ently (Capitani, Laiacona, & Barbarotto, 1999; Laws, 1999). Women also recognize natural objects more accurately, whereas men recognize artifacts more accurately (Barbarotto, Laiacona, V. Pasterski (&) Macchi, & Capitani, 2002). Similar effects have also been Department of Paediatrics, Addenbrooke’s Hospital, University of observed in semantic priming, where natural category names facil- Cambridge, Box 116, Level 8, Hills Road, Cambridge CB2 0QQ, UK itate subsequent judgments of their exemplars (e.g., FISH ? trout) e-mail:

more strongly for women than for men, but artifact category V. Pasterski K. Zwierzynska Z. Estes names facilitate judgments (e.g., TOOL ? hammer) more strongly Department of Psychology, University of Warwick, Coventry, UK for men than for women (Bermeitinger, Wentura, & Frings, 123 Arch Sex Behav 2008). Given this sex difference in semantic processing, we disagreement between participants and uncertainty within par- hypothesized that categorization would also exhibit a sex differ- ticipants. Thus, all targets were‘‘borderline items’’for which cat- ence. egory membership is unclear and which were most likely to elicit A further finding with respect to natural and artifact categories vague judgments (Hampton, 2007). Stimuli are shown in the is that they differ in terms of vagueness, i.e., whether membership Appendix. Note that because most of these categories lack a clear in the given category is absolute (all-or-none) or graded (a matter and consensual definition, participants’ judgments about these cat- of degree; see Hampton, 2007). Artifact categories tend to be egories are subjective and hence are not classifiable as‘‘correct’’ judged with more vagueness than natural categories (Diesend- or ‘‘incorrect.’’ For example, because FISH is not a biologically ruck & Gelman, 1999; Estes, 2003; Hampton, 1998; Rhodes & defined category, judging an item to belong or not belong in the Gelman, 2009). For example, a computer may partially belong in category is subjective. the category TOOLS, but a tomato tends to be judged either com- pletely in or completely out of the category FRUITS. So given the Procedure sex difference in processing artifacts and natural objects, and given that artifacts and natural objects differ in vagueness, we The items and their target categories were presented in random tested whether men and women differ in the vagueness of their order, and participants judged each item as a nonmember, partial category judgments. member, or full member of the given category. Instructions were In the present study, participants judged whether each of 50 based on those used by Estes (2003). Using the example bil- items (e.g., tomato—FRUITS, computer—TOOLS) is a full member, liards:SPORTS, the instructions read:‘‘If you believe that billiards partial member, or nonmember of its target category, and they is not a sport, then you should check the nonmember box. Or if were informed that‘‘Partial membership means that the item does you think that billiards is only somewhat a member of the cat- belong in the category, but not to the same extent as some other egory, then you should check the partial member box. But if you items.’’ The percentage of partial membership judgments is a believe that it’s just as much a member of the category as any measure of category vagueness (Estes, 2003), whereas the per- other sport, then you should indicate that it’s completely a centages of full member and nonmember judgments respectively member by checking the full member box…Partial membership are measures of category inclusivity and exclusivity. Because means that the item does belong in the category, but not to the vague category judgments are associated with less confidence same extent as some other items.’’ After each category judg- (Estes, 2004), we also tested whether confidence was related to ment, participants also rated their confidence in that judgment the predicted sex difference in vagueness. on a scale ranging from 1 =‘‘not at all confident’’to 5 =‘‘com- pletely confident.’’ Method Results Participants Confidence A total of 113 students and staff at the University of Warwick participated for £3 each. The mean age was 21.9 years for men We first examined whether participants’ confidence (i.e., mean (N = 55, SD = 3.4) and 24.1 years for women (N = 58, SD = confidence ratings) predicted the vagueness of their category 8.3). This difference in age was not significant. To check the judgments (i.e., percentage of‘‘partial member’’judgments). sample’s representativeness of men and women, participants’ Indeed, more confident participants produced fewer vague cat- sex-typedness was assessed via a retrospective version of the egory judgments (r = -.33, p\.001), and this correlation was Pre-school Activities Inventory (PSAI; Golombok & Rust, observed among both men (r = -.27, p\.05) and women 1993). The PSAI is a robust 17-item measure of self-reported (r = -.41, p\.01). This significant negative relationship masculinity and femininity. As expected, men (M = 69, SD = between confidence and vagueness corroborates prior research 9.55) were significantly more masculine than women (M = 40, (Estes, 2004). It also indicates that any sex difference observed in SD = 15.13), t(111) = 12.08, p\.001. In this case, we can vagueness might be due to a concomitant sex difference in confi- assume typicality for sex-typedness in the current sample. dence. To test for such a sex difference, participants’ mean con- fidence ratings were analyzed via a 2 (sex) 9 2 (domain) ANOVA. Measures Only the expected main effect of domain was significant, F(1, 111) = 11.15, p\.001; artifact categories (M = 4.07, SE = .05) A total of 25 target items from seven artifact categories and 25 were judged with greater confidence than natural categories target items from six natural categories were sampled from Estes (M = 3.95, SE = .05) (see also Estes, 2004). For instance, par- (2003). These target items were sampled on the basis of prior ticipants were more confident in their judgments of whether a studies in which the selected items were the most likely to elicit computer is a TOOL than of whether a tomato is a FRUIT. More 123 Arch Sex Behav importantly, men and women were equally confident in their p = .07. As evident in Table 1, men exhibited significantly more judgments (M = 4.01, SD = .52), F\1. Thus, if there were a sex inclusive judgments of artifacts, t(111) = 2.36, p\.05, but sig- difference in categorization, it would not be attributable to a sex nificantly more exclusive judgments of natural categories, difference in confidence. t(111) = 2.29, p\.05. For instance, men were more likely than women to judge that a computer is a full member of the TOOL Categorization category, and that a tomato is not at all a FRUIT. These sex dif- ferences were also medium, with effect sizes of .44 and .43 in the Mean percentages of exclusive (‘‘nonmember’’), vague (‘‘partial artifact and natural categories respectively. member’’), and inclusive (‘‘full member’’) category judgments Finally, we also tested the generality of these results across are summarized in Table 1 (including effect sizes in d; Cohen, items rather than participants. Mean percentages of exclusive, 1988). Participants’ mean percentages of vague category judg- vague, and inclusive judgments were analyzed via a 2 (sex) 9 2 ments were analyzed via a 2 (sex) 9 2 (domain) ANOVA. In (domain) MANCOVA, with confidence included as a covariate. corroboration of prior research (e.g., Estes, 2003), artifact cate- The pattern of results replicated that described above: Vague gories were judged with more vagueness than natural categories, judgments exhibited significant main effects of sex, F(1, 47) = F(1, 111) = 118.01, p\.001. For example, participants were 5.78, p\.05, and domain, F(1, 47) = 150.16, p\.001, without more likely to judge that a computer is a partial member of the interaction, whereas significant interactions were observed in TOOL category than to judge that a tomato is a partial member of both exclusive judgments, F(1, 47) = 8.35, p\.01, and inclu- the FRUIT category. More importantly, however, sex also pre- sive judgments, F(1, 47) = 8.47, p\.01. dicted vagueness, F(1, 111) = 8.89, p\.01. Women provided more vague category judgments than men across domains, i.e., sex and domain did not interact. The effect size of this sex dif- Discussion ference in vagueness was medium (d = .56; see Table 1 for effect sizes within each domain). To be conservative, we also analyzed This study yielded three novel findings. First, these results these data with participants’ mean confidence ratings included revealed a domain-general sex difference in vagueness (d = as a covariate (ANCOVA). With confidence statistically con- .56). Women provided more vague judgments than men in both trolled, the sex difference in vagueness remained significant, F(1, artifact and natural categories. Although prior studies have dem- 110) = 10.01, p\.01. Thus again, the sex difference in vagueness onstrated sex differences in naming (Capitani et al., 1999; Laws, was not attributable to confidence. 1999), recognition (Barbarotto et al., 2002), and semantic prim- Having observed a robust sex difference in vagueness, we ing (Bermeitinger et al., 2008) of artifact and natural objects, the next examined whether it was complemented by a sex differ- present study provided the first demonstration of a sex difference ence in exclusivity, inclusivity, or both. That is, given that men in the categorization of such objects. So whereas prior studies exhibited fewer ‘‘partial member’’ judgments, did they exhibit have revealed differences in the speed and/or accuracy of partic- more ‘‘nonmember’’ judgments or more ‘‘full member’’ judg- ipants’ responses, the present study showed a sex difference in ments than women? Participants’ mean percentages of exclu- actual judgments. Men and women categorized the same com- sive (nonmember) and inclusive (full member) judgments were mon objects in systematically different ways. analyzed via a 2 (sex) 9 2 (domain) MANOVA. Sex and domain Second, we also found a domain-specific sex difference in interacted significantly in exclusive judgments, F(1, 111) = 4.75, absolute judgments. Relative to women, men provided more p\.05, and marginally in inclusive judgments, F(1, 111) = 3.28, inclusive judgments of artifacts and more exclusive judgments of natural categories. This finding is broadly consistent with the prior demonstrations of a sex difference in semantic processing, with men and women exhibiting superior processing of artifacts Table 1 Percentages of exclusive (nonmember), vague (partial mem- ber), and inclusive (full member) category judgments of artifact and and natural objects respectively (Barbarotto et al., 2002; Capi- natural categories by men (N = 55) and women (N = 58) tani et al., 1999; Laws, 1999). For example, in the semantic fluency task administered by Capitani et al. (1999), men pro- Domain Judgment Men Women p d duced more instances of the category TOOLS, whereas women M SE M SE produced more instances of FRUITS. Men’s superior fluency with Artifactual Exclusive 20.15 2.10 20.55 1.96 ns .03 artifact categories may, in fact, be related to the greater inclu- Vague 45.82 2.75 52.76 2.19 .05 .37 sivity of their artifact categories. That is, if men have more Inclusive 34.04 2.42 26.69 1.99 .02 .44 inclusive artifact categories than women, it follows that they would have more instances upon which to draw in the semantic Natural Exclusive 44.07 2.78 35.10 2.76 .02 .43 fluency task. And, conversely, the greater exclusivity of men’s Vague 21.45 2.29 30.28 2.39 .01 .50 natural categories might also be related to their inferior fluency Inclusive 34.47 2.29 34.62 2.45 ns .00 in naming natural objects. 123 Arch Sex Behav A third novel finding of this study was that participants’ con- affect performance within one’s chosen profession. For instance, fidence predicted their categorization, but confidence failed to male doctors may be more or less likely than female doctors to explain the observed sex differences in categorization. Confi- diagnose a given set of symptoms as a disease. The potential dence negatively predicted vagueness among both men and consequences are manifold, and further studies may be fruitful. women, such that more confident participants provided fewer vague category judgments (cf. Estes, 2004). However, men and Acknowledgment This study was funded by the Undergraduate Research Scholarship Scheme (URS) at the University of Warwick. women were equally confident in their category judgments, and statistically controlling participants’ confidence failed to eliminate the sex difference in vagueness. Thus, the sex difference in cate- Appendix Stimuli gorization was not attributable to a sex difference in confidence. Domain Rather than indicating a sex difference in semantic categori- zation per se, this result could instead reflect a general tendency Artifactual Natural for women to choose more moderate responses than men. That is, Category Exemplar Category Exemplar women might simply be more likely to select a middling or mid- scale response, whereas men might be more likely to select an Clothing Headband Animals Bacterium extreme or endpoint response, regardless of the task. If so, then the Pocket Fungus aforementioned sex difference would say little about categori- Furniture Clock Virus zation in particular. We used participants’ confidence ratings to Piano Yeast test this potential explanation. We calculated for each participant Refrigerator Fish Clam the percentage of confidence ratings that were moderate, opera- Shelves Crab tionally defined as any response of 2, 3, or 4 on the 1-to-5 confi- Ships Canoe Lobster dence scale. Contrary to the moderation hypothesis, women were Kayak Octopus no more likely than men to select moderate confidence ratings Raft Plankton (t\1). Thus, the sex difference in category judgment appears to Spacecraft Seahorse be a genuine sex difference in semantic categorization rather than Tools Computer Shrimp a sex difference in scale use. Funnel Squid The sex difference in categorization is also consistent with a Paint Fruits Avocado sex difference in the use of tentative language. In general, women Toys Backgammon Coconut tend to use more tentative language such as hedges (e.g.,‘‘sort of’’) Cards Cucumber and disclaimers (e.g.,‘‘I’m not sure’’; Carli, 1990). This tendency Guitar Rhubarb is particularly evident when discussing masculine topics such as String Tomato sports; when discussing feminine topics such as fashion, how- Vehicles Horse Insects Caterpillar ever, men use more tentative language than women (Palomares, Roller skates Leech 2009). This sex difference in tentative language may reflect the Tricycle Scorpion sex differences in exclusive and inclusive judgments demon- Wheelchair Spider strated here. Weapons Car Worm The present study also replicated a domain difference in cate- Chair Mammals Goose gorization (Diesendruck & Gelman, 1999; Estes, 2003; Hampton, Drugs Vegetables Pumpkin 1998, 2007; Rhodes & Gelman, 2009), such that artifact catego- Fingernails Rice ries were judged with greater vagueness than natural categories. This effect is typically attributed to a domain difference in cat- egory representation. Whereas artifact categories are primarily represented according to their functions, which are mutable, References natural categories tend to be represented according to their appearance and biological features, which are less mutable (e.g., Barbarotto, R., Laiacona, M., Macchi, V., & Capitani, E. (2002). Picture Hampton, Storms, Simmons, & Heussen, 2009). More impor- reality decision, semantic categories and gender: A new set of pictures, with norms andan experimental study.Neuropsychologia,40, 1637–1653. tantly for our purposes, this replication supports the validity of our Bermeitinger, C., Wentura, D., & Frings, C. (2008). Nature and facts sample. about natural and artifactual categories: Sex differences in the seman- This sex difference in categorization could have important social tic priming paradigm. Brain and Language, 106, 153–163. implications in career choices and performance. The differential Capitani, E., Laiacona, M., & Barbarotto, R. (1999). Gender affects word retrieval of certain categories in semantic fluency tasks. Cortex, 35, tendency for absolute judgments may partially explain the gender 273–278. gap in fields that allow more or less precision or vagueness, such Carli, L. (1990). Gender, language, and influence. Journal of Person- as sciences and humanities. Furthermore, categorization may also ality and Social Psychology, 59, 941–951. 123 Arch Sex Behav Cohen, J. (1988). Statistical power analysis for the behavioral sciences Hyde, J. S. (2007). New directions in the study of gender similarities (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates. and differences. Current Directions in Psychological Sciences, 16, Diesendruck, G., & Gelman, S. A. (1999). Domain differences in absolute 259–263. judgments of category membership: Evidence for an essentialist Kimura, D. (2002). Sex hormones influence human cognitive pattern. account of categorization. Psychonomic Bulletin & Review, 6, 338– Neuroendocrinology Letters Special Issue, 23(Suppl. 4), 67–77. 346. Kolb, B., & Wishaw, I. Q. (1985). Fundamentals of human neuropsy- Estes, Z. (2003). Domain differences in the structure of artifactual and chology (2nd ed.). New York: W. H. Freeman and Co. natural categories. Memory & Cognition, 31, 199–214. Laws, K. R. (1999). Gender affects naming latencies for living and nonliv- Estes, Z. (2004). Confidence and gradedness in semantic categorization: ing things: Implications for familiarity. Cortex, 35, 729–733. Definitely somewhat artifactual, maybe absolutely natural. Psy- Linn, M. C., & Petersen, A. C. (1985). Emergence and characterization chonomic Bulletin & Review, 11, 1041–1047. of sex differences in spatial ability: A meta-analysis. Child Devel- Golombok, S., & Rust, J. (1993). The Pre-school Activities Inventory: A opment, 56, 1479–1498. standardized assessment of gender role in children. Psychological Palomares, N. A. (2009). Women are sort of more tentative than men, Assessment, 5, 131–136. aren’t they? How men and women use tentative language differ- Hampton, J. A. (1998). Similarity-based categorization and fuzziness of ently, similarly, and counterstereotypically as a function of gender natural categories. Cognition, 65, 137–165. salience. Communication Research, 36, 538–560. Hampton, J. A. (2007). Typicality, graded membership, and vagueness. Rhodes, M., & Gelman, S. A. (2009). Five-year-olds’ beliefs about the Cognitive Science, 31, 355–384. discreteness of category boundaries for animals and artifacts. Psy- Hampton, J. A., Storms, G., Simmons, C. L., & Heussen, D. (2009). Feature chonomic Bulletin & Review, 16, 920–924. integration in natural language concepts. Memory & Cognition, 37, Spreen, O., & Strauss, E. (1991). A compendium of neuropsychological 1150–1163. tests. New York: Oxford University Press. Hines, M. (2009). Gonadal hormones and sexual differentiation of human Voyer, D., Voyer, S., & Bryden, M. P. (1995). Magnitude of sex dif- brain and behavior. In D. Pfaff, A. P. Arnold, A. M. Etgen, S. E. ferences in spatial abilities: A meta-analysis and consideration of Fahrback, & R. T. Rubin (Eds.), Hormones, brain and behavior critical variables. Psychological Bulletin, 117, 250–270. (2nd ed., Vol. 3, pp. 1869–1909). New York: Academic. Hyde, J. S. (2005). The gender similarities hypothesis. American Psy- chologist, 60, 581–592. 123