The Sex Difference on Mental Rotation Tests Is Not Necessarily a Difference in Mental Rotation Ability

The largest consistent sex difference in human cognition is found on mental rotation tests, which require participants to compare pictures of three-dimensional objects and decide whether they depict the same object or different objects. Across cultures, males score up to one standard deviation higher than females. We administered two standard rotation tests to 123 participants and found that these higher scores likely do not reflect superiority in the process of mental rotation per se, but rather in other aspects of task performance. Our results show that males decide more accurately when two objects are different, a situation in which women are more likely to claim incorrectly that they are the same, and that individual differences in confidence are responsible for part of the male advantage found on this test, whereas differences in spatial encoding ability are not. These results have implications for evolutionary theories of sex differences in spatial cognition. Sex differences on mental rotation tests – p. 3 of 26 The fact that males score higher than females on tests that require rotating objects in mental images is widely accepted in psychology. The most popular mental rotation test, developed by Vandenberg and Kuse (1978; hereafter referred to as “VK”) and used in nearly half of all studies of sex differences in mental rotation, consistently yields the largest male advantage: 0.75–1.0 standard deviations according to a meta-analysis (Voyer, Voyer, & Bryden, 1995). Although it may vary by socioeconomic background (Levine et al., 2005) and sexual orientation (Peters, Manning, & Reimers, 2007), this effect occurs robustly across cultures (Silverman, Phillips, & Silverman, 1996; Silverman, Choi, & Peters, 2007) and age ranges (Linn & Petersen, 1985; Maylor et al., 2007), and it is a critical empirical pillar of theories of human sex differences in spatial abilities, including evolutionary theories (Silverman & Eals, 1992; Kimura, 1999). The VK test is a paper-and-pencil adaptation of a task developed by Shepard and Metzler (1971; hereafter “SM”) that first demonstrated that people “mentally rotate” imagined objects to compare them and decide whether they are identical. The consistent behavioral signature of this mental rotation process is a strongly linear increase in response time and error rate as the angular disparity between objects in a stimulus pair increases. The slope of this linear function measures the efficacy of the rotation process; a lower cost (in time and accuracy) for each additional degree of rotation indicates better mental rotation ability. In the VK, however, response time and accuracy are not measured as a function of the angle of disparity on individual trials; instead, the participant completes a series of trials within a given time limit, and performance is computed simply as the total number of correct answers. Figure 1 illustrates the SM and VK. In a previous study (Hooven et al., 2004) we assessed the relationship between performance on a version of the SM task and testosterone level in men. We partitioned Sex differences on mental rotation tests – p. 4 of 26 performance into two components, one that reflects primarily the mental rotation process itself (i.e., the slope) and one that reflects primarily other processes, including visual encoding, preparation for rotation, decision making, and response (i.e., the intercept of the rotation function). We found that for error rate, testosterone level was related only to the intercept, and only when the two objects are different (“Different” response trials). Importantly, as Shepard and Metzler (1971) explained, these distractor trials are included in the task not to measure the mental rotation process but to ensure that participants must engage in this process by making the task difficult to perform in any other way. Based on this within-sex result—and the theory that differences between men and women in spatial ability are caused by differences in preand/or post-natal testosterone levels (e.g., Kimura, 1999)—we hypothesized that the between-sex difference in performance on the SM task would also be confined to the intercept of Different trials, and that this component of an individual’s performance would be the best predictor of that individual’s score on the VK test. Accordingly, we administered both tests, in counterbalanced order, to a group of male and female participants, following standard procedures from previous studies. To explore possible mediators of the relationship between sex and mental rotation performance, we added tests of spatial relations encoding, impulsiveness, and confidence. Methods Participants We tested 123 volunteers (60 male, 63 female; ages 18–60, mean 26 years), who participated for pay after being recruited via advertisements (that did not mention sex differences or spatial ability). Approximately two thirds of the participants were students and one third were Sex differences on mental rotation tests – p. 5 of 26 local residents. All reported not using drugs or psychoactive medications, no history of psychiatric or neurological illness, and at least a high-school education. Our male and female samples did not differ significantly in age, years of education, handedness, general cognitive ability (measured using a short form of Raven’s Advanced Progressive Matrices; Bors & Stokes, 1998), vividness of experienced mental imagery (Vividness of Visual Imagery Questionnaire; Marks, 1972), frequency of imagery use in daily life (Spontaneous Use of Imagery Scale; Reisberg, Pearson, & Kosslyn, 2003), or length of time awake before testing (all p > .10). Materials and Apparatus The SM was administered using a computerized adaptation of the three-dimensional mental rotation test described by Shepard and Metzler (1971). Stimuli were presented and responses were recorded by an Apple Macintosh computer running OS 9 with a 40.5 cm monitor. Keypresses and response times (in milliseconds) were automatically recorded by PsyScope 1.2.5 software (Cohen et al., 1993). Each trial consisted of two circles, presented side-by-side, with each containing a block stimulus. We presented a subset of the stimuli used in the original Shepard and Metzler (1971) study. As illustrated in Figure 1a, each circle (diameter 5.6 cm, or 10.8° of visual angle) contained one two-dimensional representation of a three-dimensional block object (approximately 5.7 cm x 0.64 cm, or 7.5° x 4.2°). An equal number of objects in each pair were presented at angles that differed by 0, 40, 80, 120, or 160 degrees. Half of the stimuli at each angle were Same pairs and half were Different pairs. Accordingly, there were 5 angles x 2 response types x 8 standard objects = 80 total trials. The VK (Vandenberg & Kuse, 1978) is a paper-and-pencil adaptation of the original SM task. Figure 1b shows a sample trial from this test, which consists of four practice and 20 test trials, with five on each page. Sex differences on mental rotation tests – p. 6 of 26