Expectations and Mental Rotation: Different effects on gender explored through the (unlikely) case of female superiority in mental rotation rate of large objects

The present study aims to examine the relationship between expectation about performance in a mental rotation task of 3-D geometrical shapes on a small (standard) and large (novel) display and for each sex. In Experiment 1 we explicitly induced expectations about whether participants would perform better on either the large or small displays. Results revealed that women mentally rotate the 3-D shapes faster than men on the large display, but both genders were affected by the expectation variable. Apparently, only females expecting to do better in the large display condition experienced superior performance relative to men. Males appeared affected by expectations toward task difficulty (superior conditions), where they scored lower when expecting a superior condition. Experiment 2 confirmed the effect of display size on gender and female superiority was again observed in the large display condition – however without any explicit manipulation in the experimental design of expectations (for superiority of a display condition over the other). Expectation and Mental rotation 1 Expectations and Mental Rotation: Different effects on gender explored through the (unlikely) case of female superiority in mental rotation rate of large objects Introduction Spatial ability would seem essential in order to interact with objects in our daily environment, not only to locate them, manipulate them, and navigate around them and within the space containing them, but also for several technical and specialized activities like designing and building a house (e.g., architecture design and engineering), medical procedures and analysis of radiology images (e.g., brain scans used by neurologists). Moreover, spatial ability may be important for high-level cognition, since there has been observed a high correlation between spatial ability and the ability to solve problems in geometry, mathematics, chemistry, and physics (Delgado & Prieto, 2004). Spatial ability could be measured in an infinity of ways, but within psychology, the Mental Rotation Tests (MRT) have been established as a relatively robust and easyto-collect measure of at least one important component of spatial ability: The ability to mentally hold the image of a three-dimensional figure and rotate it in space so that one can decide whether it is identical to another object (or previous event). This test has shown a consistent behavioral pattern consisting in a strongly linear increase in response time and error rates as the angular disparity between objects in the stimulus pair increases. In other words, the efficacy of the mental rotation process can be operationalized by the slope of the linear function described above (Shepard & Metzler, 1971). Shepard and Metzler measured mental rotation in two planes; “picture plane” and Expectation and Mental rotation 2 in “depth” (corresponding to “X-Y” plane and “Z-Y” plane) and found no difference in regards to response times and linear slopes. The original motivation for their study was to reveal a human “trait” that is able to identify two identical three-dimensional objects from two-dimensional drawings (representations), despite the fact that they may be seen in very different orientations. Hence, in order to be able to “mentally rotate” the objects, an internal representation would have to be created from a two-dimensional image before the process of mental rotation could be executed. The finding that objects that were rotated in “depth” took no longer to rotate than those that were rotated in the picture plane was interpreted as support to the claim that a three-dimensional representation of the two-dimensional portrayed image was created by the participants because “they could imagine the rotation around whichever axis was required with equal ease” (ibid.). Remarkably, since its discovery as a test and mental ability, the mental rotation is the spatial trait that has shown the largest and most consistent sex difference in human subjects, favoring male performance in speed and accuracy compared to what is observed in other spatial tasks (Hirnstein, Bayer, & Hausmann, 2009; Linn & Petersen, 1985; Daniel Voyer, Voyer, & Bryden, 1995). In the attempt to explain such a gender difference in mental rotation tasks, several accounts have been put forward. One is the “biological” account, which actually comprises a family of accounts that can more or less stress the existence of evolutionarily-selected traits. For example, it has been suggested that the difference originates already in the mother’s womb, where the fetus is exposed to specific hormones, and afterwards post-natally to the exposure of the fetus’ sex hormones, all of which we know affect brain development (Kerns & Berenbaum, 1991). Also, the timeExpectation and Mental rotation 3 lines of sexual maturation (which differ in the two sexes) would seem to be crucial for performance in MRT; specifically, those who mature later then the average will score higher than those who mature earlier (Sanders & Soares, 1986). Differences in brain volumes, within target brain regions of each sex, have also been observed and there is evidence favoring female in grey matter volume in the parietal lobe, which was shown to be disadvantageous for women regarding performance on mental rotation task. The parietal lobes are considered to play an important role in spatial processing, especially for mental rotation. Men show predominantly parietal activation whereas females show additional inferior frontal activation, and their activation in the parietal lobe seems to be more symmetrically organized than in males in MRT e.g. females use more brain regions which require more activation, and hence more time (Koscik, O'Leary, J.Moser, Andreasen, & Nopoulos, 2009). Further examples are lateralization, where subjects with low spatial ability showed a left field advantage, medium ability showed no field advantage and high ability showed right field advantage (Daniel Voyer & Bryden, 1990), which is in line with the expectation that mental rotation should be one of the expressions of a right hemisphere’s general superiority in spatial ability (Linn & Petersen, 1985). A sex difference in mental rotation in human infants has also recently been reported by two studies (Moore & Johnson, 2008; Quinn & Liben, 2008), interpreted as further support for a biological explanation. An alternative explanation for gender differences in the MRT is based on a hypothetical cultural influence. One frequent proposal is that, from early on boys are either driven towards or given toys and activities that foster development of spatial abilities, like Lego toys, sports activities, and video and computer games (I. D. Cherney Expectation and Mental rotation 4 & Voyer, 2010). This difference in training and experience of spatial abilities accelerates as the time goes by, since boys also choose more mathematics classes and therefore experience other spatially-relevant mental activities and training. These real life experiences, enhanced by the school curriculum, may foster the development of spatial abilities in boys while girls may perceive most of the spatial tasks as “masculine.” Possibly, girls are even discouraged and/or feel intimidated by the thought of engaging in tasks that demand excellence in basic spatial skills (Parsons et al., 2004). Although a girl who chooses math classes in school has likely received a level of training that demonstrates her excellence in math, studies have shown that outside the school environment, the same girl may engage in fewer spatial experiences then the boys (I.D Cherney, 2001). Boys on the other hand may perceive themselves as generally competent and would tend to do well in tasks which include spatial abilities. In a study by Moè (2009), men were affected positively when they believed that they were better than women and also expected the task to be easy, while women’s performance were more linked to the expectations based on the self image than on the difficulty of the task. The outperformance of males in mental rotation can also be accounted for by the choice of strategies that each gender takes when they perform the actual mental rotation task. It seems likely that women would prefer using an analytical strategy, which involves breaking apart the stimulus object and then comparing, in a sequential process, certain features from the sample figure with certain features on the target figure. In contrast, men may prefer a holistic approach such as they would image the whole object and rotate the whole representation in their mind (M. M. Peters et al., 1995). The genders can also vary in their response strategies, in the situation where they compare the sample Expectation and Mental rotation 5 figure to the target figure. Men seem to have a leaping strategy, they continue immediately with the next item as soon as they have discovered whether it is a match or a non-match, without verifying it. Women tend to have a conservative response strategy, so that they might do a “double check” of whether the item is a match or non-match, which is more time consuming (Hirnstein, Bayer, & Hausmann, 2009). Evidence from eventrelated brain potentials research have shown that response preparation begins before mental rotation is finished, and thereby will a conservative strategy cost enormously (Heil, Rauch, & Hennighausen, 1998). Certain features of the mental rotation test stimuli has also been proven more difficult for women, e.g. long trajectories multiline or multispotted (Birenbaum, Kelly, & Levi-Keren, 1994). When alternated the standard block figures in MRT with three –dimensional human figures performance improved more for women then men (Alexander & Evardone, 2008). Interestingly, mental rotation tasks have also yielded robust and reliable sex differences for decades already, with men typically outperforming women in the task (Hedges & Nowell, 1995; Linn & Petersen, 1985; M. M. Peters et al., 1995; D Voyer & Hou, 2006; Daniel Voyer, Voyer, & Bryden, 1995); although such a male superiority has not been observed consistently with all types of stimuli (Jansen-Osmann & Heil, 2007)) or any mental rotation task (M. Peters, 2005; Daniel Voyer, Voyer, & Bryden, 1995)., and as shown above from different perspectives, there is evidence enough to question if men always outperform women. For example, Moè (Moè, 2009) used a version developed by Vandenberg and Kuse, which is an adaptation of the original mental rotation test developed by Shepard & Metzler (Vandenberg & Kuse, 1978; Daniel Voyer & Saunders, 2004). Moè’s research Expectation and Mental rotation 6 show that women’s performance was affected by positive instructions about gender (e.g. that they were “better than men”), so much that they reached men’s scores in MRT, regardless if they expected a easy or difficult task. Men, on the other hand, were affected by instructions about gender and the task difficulty combined. When the instructions were that they were better than women and the task was easy, men improved their performance but in the condition with the opposite instructions (“men are better than women” and that the the task is “difficult”) the female superiority effect disappeared. This suggests that an increase in women’s performance in MRT is possible when given positive beliefs about self, to the point of reaching men’s performance, while men’s performance is more effected by the complexity of the task than the belief about self. Thus, it seems that women’s spatial abilities are vulnerable and easy to modify trough attitudinal and experimental factors (Quaiser-Pohl & Lehmann, 2002). In fact, the magnitude of sex difference has decreased in recent years (Daniel Voyer, Voyer, & Bryden, 1995). Scali et al (2000) found that men outperformed women in MRT only when scored in a strict manner combined with the instruction to focus on accuracy. Men did not outperform women when the focus was on speed (Scali, Brownlow, & Hicks, 2000). Quaiser-Pohl & Lehmann found that even though there was an effect size of gender differences in MRT, it varied the most with students in arts, humanities and social sciences and smaller in computational visualistics. They also showed a correlation with spatial abilities and computer experiences and between test performance and achievement related self-concept which depended on gender e.g. evident mostly in the female sample. Expectation and Mental rotation 7 Even tough research has shown that men outperform women, because of biological, cultural, strategically or situational factors, there is the suggestion that it is easier to undermine cognitive performance then to improve it, even for men (Wraga, Duncan, Jacobs, Helt, & Church, 2006). The power of effort attribution was proven by Moè and Pazzaglia (2010) in an experiment where they told their participants that performance could either be “effort-dependent” or “genetic-dependent”; namely, e.. “anyone can succeed on this task by putting in effort” versus “performance on this test depends on genetic determinants”. As their findings showed, the ‘effort’ instruction had a substantial positive effect on performance, which was significantly higher than in the other two conditions. It also occurred independently of gender, although males scored higher than women. This finding indicates that it is possible to improve on MRT performance as long as there is a belief that what matters above all is effort and practice and, hence, there is a chance to improve performance (Moè & Pazzaglia, 2010). Human-computer interaction and spatial abilities From an applied point of view, scientific interest for the topic of spatial ability and visual perception has received “renewed” interest, resulting in an increasing attention from fields of study related to technological advances for computer displays. This has been partly motivated by an increase of several orders of magnitude in computational power over the last decades, but specifically in the relatively recent advent of much higher quality of displays, particularly with the introduction of the Liquid Crystal Display (LCD) in the late 1990s. Qualitative improvement in display technology has basically happened along two dimensions: an increase in the pixel count (and smaller size of pixels); and a considerable increase in the size of the displays. A novel display technique Expectation and Mental rotation 8 that facilitates “tiling” of displays, making many displays “work together” as one display surface basically eliminates restrictions regarding size of the display – making us technologically capable of producing “unlimited sized displays”. In light of this development, from a technological – and economical perspective rather obvious questions have been raised; how big is “enough”; how big is useful – and the consequential: what role does size play (in visual perception)? A larger area seems quite useful for many work applications, and a large display has e.g. been shown to yield some productivity-gains over smaller (Czerwinski, 2003). Interestingly, Czerwinski, Tan, and Robertson (2002) found that the “gender gap” in 3Dworld navigation (spatial tasks) diminishes and seem to almost disappear when users are given a wider field of view – meaning a broader display than what we are typically used to. Specifically, they set up two projection displays side by side and minimized the “seam” between them in order to have a continuous and coherent display that was twice as wide as an ordinary one. They made a virtual 3D-environment in which the participants engaged in way-finding tasks in a virtual world, a task in which males typically outperform females. They found significant effects of field of view and display size resulting on trial times. There was also a significant effect of gender in the small screen setup where males outperformed females, while in the large screen setup there was no observed difference and men and women performed equally. In a follow-up study Tan, Czerwinski, & Robertson (Tan, Czerwinski, & Robertson, 2003) found that with a large display setup of 100 degrees field of view (normally around 30 degrees) and with the presence of optical flow cues ( i.e., the effects of moving within the environment has on the change in image on the retina of immobile Expectation and Mental rotation 9 objects residing within environment), female performance improved to the point of evening out gender differences. They suggested that a wider field of view and optical flow cues could separately contribute to female performance enhancements. In a recent study, Tan et al. hypothesized that large displays result in immersing the users in the 3D-world, making them feel more “in” the world, and may bias them to adopt more efficient cognitive strategies when performing spatial tasks (Tan, Gergle, Scupelli, & Pausch, 2006). Specifically, they hypothesized that a large screen biases users in adopting egocentric strategies for those tasks (i.e., “moving” their own person/head within the virtual world or “being there”, as opposed to exocentric strategies – rotating the environment around them). In testing this hypothesis they also used the classical mental rotation (MR) test. However, they found no sex differences in task accuracy or response time. Interestingly, in this study the relative size of the display was constant (i.e., only manipulating physical size as variable), so that the retinal size of the object was also a constant. In a quite small study Suzuki and Nakata (Suzuki & Nakata, 1988) investigated whether the size of objects (retinal size) can affect RTs. Participants (N= 6; 2 females) viewed either small, medium, or large object stimuli, where the retinal objective sizes, measured between the centers of the objects, ranged from 2.9 degrees to 5.7 degrees to 11.5 degrees. They found that angular differences in figures had a clear effect on RTs, as did the differences in retinal size. In accordance with Tan and colleagues (Tan et al., 2006), they found no effect of constant retinal size of objects (i.e. manipulating both the size of the display and distance from it in order to keep retinal size constant). Furthermore, Suzuki and Nakata found that RTs increased with the smaller sized objects, compared to the medium and large sized objects, arguing that this might be explained that Expectation and Mental rotation 10 the objects might have to be mentally “scaled up” (going through a “size-normalization process”) before rotation on the smaller object. In the first study to our knowledge to investigate the issue of object size and mental rotation, Shwartz (1979) examined whether the size of an object and its complexity affect rotation rates. Shwartz had 20 subjects (no mention of their sex) making comparisons between two-dimensional objects, which were shown briefly and in succession on the screen, the first stimulus being also followed by an orientation cue in the direction of which the subject were to visualize the object being rotated. When the second stimulus was given, the subject was to make a same/different judgment about the two objects. Shwartz found that increasing size of the objects resulted in increased rotation time. He also found that larger objects required increasingly more time to rotate farther. One should note that the amount of pixels on a screen generally affects the amount of “inherent information” within the displayed image. That is: the more pixels, the more information can be displayed simultaneously. The physical size of the display however affects the visual angle with which we perceive the image. A small display gives a narrow angle, meaning that the displayed image occupies a smaller fraction of our field of view – or the image “captured” on our retina; a larger display provides a larger visual angle. In our work, we have concentrated on the physical size of the display and how this affects the cognitive aspects of our perception Hypothesis Based on the relative scant previous research on mental rotation tasks of large objects and/or large displays, we wanted to investigate both the effect of display size and how this factor may affect the gender bias in the task of mental rotation. Since the Expectation and Mental rotation 11 previous findings suggested that the size of the objects should influence rotation times (in some direction), and that display size seems to have a positive effect on female performance, we expected to find effects of screen size on gender. Specifically, we predicted an increase in female performance relative to male performance, given the above mentioned effects of field of view on other spatial tasks. We assume that females prefer an analytic approach to the MR task; for instance, by comparing the figures piecewise, this strategy requires comparing at least two parts of the objects in order to make a call whether they are the same. The preferred male approach of holistic processing requires making an internal representation of the complete objects, rotating one of them in order to compare with the other. Females, on the other hand, would encode only parts of the stimulus (at a time), and then mentally rotate another part. If encoding an object requires visual focus (thought to occupy a minimum 1o of our visual field is in focus at any time; (Eriksen & Hoffman, 1972)), this implies longer MRT response for large objects, given the use of a holistic strategy. The larger the object, the more shifting of focus is required in order to encode the whole object – in order to subsequently perform the mental rotation. A piecewise strategy would require focus-shift behavior by default, and, hence, would presumably not be as affected by object-size, as a holistic approach. In addition, visual working memory could impose restrictions on object size, requiring a down-scaling of larger-than-allowed stimulus objects. The effect should be similar to that predicted by the focus-shift hypothesis. If we can confirm a positive effect on gender performance regarding display size, then this will be further investigated to establish whether this effect can be a result of Expectation and Mental rotation 12 expectancies, e.g. do men or women perform better when they are told they are expected to do so, either in the small display condition or the large display condition? Expectation and Mental rotation 13 Method Experiment 1 Participants Forty-one participants, 23 men and 18 women participated in Experiment 1, with an age range from 18 to 45 years and a mean age of = 29.9, SD= 6.2. Four participants were excluded (one woman and three men) because they didn’t respond above the criterion level (i.e., 70% correct). Thus, the descriptive statistics and analyses shown below will consist of responses of the remaining 38 participants. All participants came in voluntarily and were compensated with two lottery tickets, worth 50 NOK. Written inform consent was also obtained from all participants, and at the end of the experiment all were presented with a short questionnaire. Three persons were not native Norwegian, and therefore two were given instruction in English and care was taken to make sure that they understood the task. The third participant comprehended Norwegian well. All participants had normal, or corrected-to-normal, eyesight. Stimuli and apparatus The large display wall consisted of 28 projectors, which projected an image onto a screen surface. There were 7 x 1024 horizontal pixels and 4 x 768 vertical pixels appearing onto the screen, all together forming a display of 22 millions pixels of red, green and blue. The screen size was 230 inches (diagonally). In contrast, the small screen’s size was 14.1 inches on a Dell D600 laptop computer, with the rather standard native resolution of 1400x 1050 pixels and a 24 bit color spectrum. The SuperLab software was running natively (i.e., the local processor and graphics hardware was running and displaying the SuperLab software and programExpectation and Mental rotation 14 interface) on the laptop computer, while the image was transferred to the display wall using a 100MB Ethernet interface and a Java implemented display-server running on the Virtual Network Computing (VNC) server on a Dell PowerEdge 2800, with 2 Xeon 3.8GHz/2MB 800FSB, 8GB Dual Rank DDR2 Memory (4x2GB), 146GB SCSI Ultra320 (15,000rpm) 1in 80 pin Hard Drive x 2 with the RedHat Linux operating system. The computer cluster feeding the projectors was comprised of 28+1 Dell 370 PCs with P4 Prescott, 3.2GHz, 2GB RAM, 1Gbit Ethernet and a 48 port HP switch. See (Jensen, 2006) for a current description of the equipment. The SuperLab interface (with the stimuli) was transferred to the display and enlarged to fit the larger display area of the wall. As a consequence, the number of (perceived) pixels was held constant between the displays, along with the aspect-ration (4:3). As for the screen width and consequential retinal size of the images projected (visual angle of screen), the projected screen (display area covered by SuperLab) was measured using a laser-meter to 404cm and 28,5cm for the small screen. Note that projectors working together to produce a single coherent and continuous image (one “desktop”, if you will) have the unique feature that, if aligned correctly, they can produce one image without the bezels that ordinary displays (LCDs) do when aligned in a matrix. There will, however, be small color-variations between the different projectors, but the resulting “image” can be remarkably coherent. Stacking either projectors or LCD-displays together like this, one can produce a display of almost unlimited sized with a number of pixels proportional to the number of display devices in the configuration. Expectation and Mental rotation 15 Measuring an exact viewing distance was not possible, since the participants were instructed to maintain “comfort viewing distance” from their chair and table. Nevertheless, the table remained at the same point at all times, and was placed 370 cm from the large screen. As a result, viewing distance from large and small screen respectively, hence, was about 370 cm and ca 65 cm. The viewing angle for large and small screen in our experiment is shown in Table 1. ‘Total visual angle’ means the visual angle provided by the display in question, while ‘angle between objects’ refers to the approximate angle from the person to the midpoints of the objects. Figure 1 shows the display setup with corresponding visual angles. Table 1 Visual angles in the two display setups Angle explanation Large display Small Display Total visual angle display 57.0° 24.7° Angle Between objects 27.3° 11.8° 25° 57° Large Display Small Display