Sex differences in visuo-spatial processing: An fMRI study of mental rotation

Abstract

Following the theoretical framework of coordinate and categorical principals for visuo-spatial processing, originally formulated by [Kosslyn, S. M. (1987). Seeing and imagining in the cerebral hemispheres: AQ computational approach. Psychological Review, 94, 148–175], we present data from an fMRI study on mental rotation, using the classic [Shepard, R. N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science, 171, 701–703] task, comparing males and females. Subjects were presented with black-and-white drawings of 3-D shapes taken from the set of 3-D perspective drawings developed by [Shepard, R. N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science, 171, 701–703], alternated with 2-D white bars as control stimuli. The drawings were presented pairwise, as black and white drawings against a black circular background. On half of the trials, the two 3-D shapes were congruent but portrayed with different orientation, in the other half the two shapes were incongruent. Analysis of response accuracy and reaction times did not reveal any significant differences between the sexes. However, clusters of significant neuronal activation were found in the superior parietal lobule (BA 7), more intensely over the right hemisphere, and bilaterally in the inferior frontal gyrus (BA 44/45). Males showed predominantly parietal activation, while the females, in addition, showed inferior frontal activation. We suggest that males may be biased towards a coordinate processing approach, and females biased towards a serial, categorical processing approach.

Introduction

¹In the Shepard and Metzler mental rotation task, subjects are shown pairs of perspective drawing of 3-D regular shapes. The task of the subject on each trial is to decide whether the two shapes are identical, or if one is a mirror-image of the other (see also Cooper, 1976, Kosslyn, 1980). The typical finding is that response times increase as the angle of disparity between the two shapes increases, with requirements for cognitive processing in order to determine if they are the same shape or not. A common explanation of these findings is that an image of the shape has to be mentally “rotated” to be superimposed on the reference shape in order for the subject to decide whether the shapes are identical or not (Shepard & Cooper, 1982; Tagaris et al., 1997).

Based on behavioral responses (response accuracy) several studies have found males to perform better than females (e.g. Astur, Tropp, Sava, Constable, & Marcus, 2004; Crucian & Berenbaum, 1998; Fisher & Pellegrino, 1988; Parsons et al., 2004, Peters et al., 1995). This has been explained with reference to differences between the sexes in hemisphere functioning, with males performing better than females on typical right hemisphere processing tasks (e.g. Levy & Reid, 1978). However, not all studies have found sex differences in mental rotation. Cohen and Polich (1989) found no differences between males and females for mental rotations involving letter and polygons, i.e. different tasks than the Shepard and Metzler 3-D task.

Hooven, Chabris, Ellison, and Kosslyn (2004) measured salivary testosterone in 27 young males on a 2-day testing schedule. The results showed that increased testosterone levels had a negative effect on reaction time and error rate on the mental rotation task. The importance of the Hooven et al. (2004) study is that it identifies a biological marker for the typical male superior performance on mental rotation tasks. Similarly, Aleman, Bronk, Kessels, Koppeschaar, and van Honk (2004) found that a single administration of testosterone improved performance in young women on a mental rotation task.

The objective of the present study was to further investigate sex differences in mental rotation by means of mapping the crucial neuronal mechanisms involved. Few studies of brain activation during mental rotation thus far have systematically evaluated differences between males and females. A clear majority of published studies on brain activation to mental rotation tasks have not separated their findings for males and females (e.g. Belin, Moroni, Gelbert, Cordoliani, & Delaporte, 1998; Cohen et al., 1996, Tagaris et al., 1997, Tagaris et al., 1998, Watson et al., 1998), or only one sex has been included in the study (e.g. Jäncke et al., 1998; Richter, Ugurbil, Georgopoulos, & Kim, 1997; Taira, Kawashima, Inoue, & Fukuda, 1998). The first study that reported separate results for males and females was the study by Tagaris et al. (1996). In two more recent studies, Weiss et al. (2003) and Jordan, Wustenberg, Heinze, Peters, and Jäncke (2003), it was found that males had increased activation in the inferior parietal lobule compared to females, while females showed increased activations in frontal lobe and fusiform areas. Finally, Seurinck, Vingerhoets, de Lange, and Achten (2004), using rotated hands, found that females recruited left ventral premotor cortex. A left hemisphere effect in females was also observed by Alexander, Packard, and Peterson (2002) for memory of the locations of objects briefly presented in the left or right visual half-field.

Mental rotation paradigms are particularly suitable for studying visuo-spatial processing strategies in the right and left cerebral hemispheres. Mental rotation of abstract 3D objects are hard to verbalize, and would thus present as pure spatial tasks. Mental rotation tasks may also be conceptualized in the Kosslyn (1987) theoretical framework of categorical and coordinate metric space, pointing towards a right hemisphere processing superiority for such tasks. This is also what has been found. For example, Ditunno and Mann (1990) reported faster reaction times when the shapes were presented in the left visual field.

Similar results were reported by Corballis and Sergent (1988) when testing a patient with a surgical split of the corpus callosum, thus separating the right and left hemispheres. A more recent study by Harris, Harris, and Caine (2002) reported that patients with damage to the right basal ganglia showed significant performance deficits on a mental rotation task. However other studies on patients with unilateral, left or right hemisphere lesions (e.g. Kosslyn, Holtzman, Farah, & Gazzaniga, 1985) have found impaired performance particularly for left-hemisphere damaged patients, indicating a left hemisphere basis for mental rotation (cfr. Hellige, 1995, Kosslyn and Brown, 1995). Thus, there seem to be some contradictory points in the literature that beg for explanation. On one hand, Kosslyn's original theory posits that categorical representations are abstract and qualitatively disjoint; therefore they cannot be used to describe the continuous rigid transformations of parts and spatial rotations. In contrast, continuous rigid transformations can easily be described within a coordinate framework. Hence, mental rotation would be expected to be mainly accomplished by using coordinate, metric, representations, in a continuum of space where two spatial forms can transform into one another through intermediate orders of points. A possible solution to the paradox could be that mental rotation cannot be seen as a simple (one mechanism) process and it is in fact likely to be carried out by a system of operations working together. Particularly, comparing views of a multipart object (like the Shepard and Metzler's cubes) might require at the outset also a multipart object's structural description (i.e., representing the individual parts and their relations). In other words, although the mechanics of the rotation (the continuous transformation) might well be taking place in the right hemisphere, verbalizations or the use of verbal strategies may require the cooperation of the left hemisphere. Although the Shepard and Metzler cubes may not lend themselves too easily to verbal labels, or descriptions, it cannot be ruled out that a subject may apply some verbal labels to parts when processing the stimuli (e.g., one could label parts of the cubes shown in Fig. 1 as ‘the bottom, nearest arm’ and ‘the top, left-pointing arm’). This would explain why neuronal activations (in fMRI and PET studies) during rotation tasks are typically bilateral and why a left hemisphere lesion might compromise – in some cases more than a right hemisphere lesion – the mental rotation task.

Thus, although bilateral activations typically are found in brain imaging studies (PET and fMRI), almost all studies have also reported the right parietal lobule as involved in mental rotation, suggesting a right hemisphere processing dominance (e.g. Belin et al., 1998, Cohen et al., 1996). Other studies have, however, reported bilateral parietal lobule activation (Tagaris et al., 1996, Tagaris et al., 1997, although these authors used a different task), or inconsistent laterality across subjects (Cohen et al., 1996). Thus, the issue of hemispheric asymmetry in brain activation studies of mental rotation is unresolved yet. Part of the explanation for the variability across studies and subjects in hemispheric asymmetry for mental rotation may be the kind of baseline or control stimulus used. Both Cohen et al. (1996) and Richter et al. (1997) used the same Shepard and Metzler (1971) shapes during the experimental (mental rotation) and control conditions, subtracting fMRI images between the two conditions. However during the control condition, only non-rotated shapes were used. Although this controls for the effect of visual perception, it may induce “carry-over” processing effects from the experimental to the control stimulus condition, with the subject trying to mentally rotate also the control stimuli, since they are similar in shape and outline as the experimental stimuli. Therefore, in the present study we used different control shapes in order to optimize differences in mental rotation demands between the experimental and control conditions, while controlling for overall visual perception and for motor responses. The study of sex differences in neuronal activation in a mental rotation task is relevant also from the point of view that this has not been the topic in studies on categorical/coordinate processing.