Image familiarity influences sex differences in mental rotation

Males generally achieve higher levels of performance than females on mental rotation tasks (Linn & Petersen, 1985, Voyer, Voyer, & Bryden, 1995); though the factors which contribute to this sex difference are poorly understood. The present research aimed to investigate the cognitive mechanisms that underlie sex differences in mental rotation task performance using event-related potentials (ERPs). Specifically, the effects of image type, task repetition and image familiarity on the cognitive processes used by males and females in mental rotation were explored. This research also investigated similarities and differences in the way that males and females performed specific cognitive operations within the task (e.g., forming a mental representation of the stimulus, accessing an existing representation from memory and mental rotation per se). In Experiment 1, 18 males and 18 females completed three mental rotation tasks. The three tasks were the same, except that stimulus category was manipulated between tasks (letters, familiar symbols or unfamiliar abstract figures). Males and females were found to achieve similar performance levels. For both sexes, performance was highest for the letters condition and lowest for the unfamiliar abstract figures condition. ERP analysis revealed sex differences in the P2 and P3 components and for mental-rotation related negativity (e.g., P2 was left-lateralised in females and symmetrical in males). This indicates that males' strategic approach to the task was different to females', but the ultimate performance outcome did not differ between the sexes. In Experiment 2, the effects of image familiarity on sex differences in mental rotation were further explored. Participants were trained to reach a determined level of familiarity with initially unknown polygons which ensured that familiarity with the images was reasonably similar for all participants. Nineteen males and 20 females completed three mental rotation tasks, which were largely the same, except that image familiarity was manipulated between tasks (trained, or initially unfamiliar/repeated during the task, or unfamiliar/never repeated). Again, males and females demonstrated similar overall performance levels, except during the unfamiliar condition. In this condition, there was a trend for males' accuracy to be higher than females'. ERP results revealed the P2 component to be larger for females than for males. Females showed differentiation in P2 amplitude between mirrored and normal images when presented upright rotated by 45°, while males did not demonstrate any parity differentiation in this component. P2 indexes early stimulus processing, including feature detection processes (Crowley & Colrain, 2004; Luck & Hillyard, 1994). The P2 sex difference indicates that early stimulus processing was undertaken differently by males and females. It was also found that mean amplitude (300-600ms) was symmetrical for males and rightlateralised for females. Interpreted within a resource allocation framework (Isreal, Wickens, Chesney, & Donchin, 1980; Kok, 1997; 2001), this result indicates that females probably allocated a high proportion of visuospatial resources to character classification and the formation of an elaborate mental image while males did not demonstrate this same visuospatial resource recruitment strategy. In both experiments, ERP rotation-related negativity was demonstrated differently by males compared to females. Give that this negativity is regarded as a direct index of mental rotation (Heil, 2002), this indicates that males and females used different mental rotation strategies. Females showed strong left hemisphere rotationrelated negativity and weak negativity in the right hemisphere. Males showed symmetrical rotation-related negativity. These lateralisation patterns indicate that females adopted a piecemeal strategy and males used a holistic strategy. For females, performance generally improved as each mental rotation task progressed (i.e., performance was higher in the second half than in the first half of most tasks). Males' performance remained consistent within each mental rotation task. (These effects were observed when data for each task was split into two equal halves and compared.) Within-task performance enhancements were present for females when the mental rotation task contained a set of four stimuli which were repeated throughout but not when the task contained a large set of stimuli which were never repeated. Thus females' performance improved when stimuli became familiar, but did not improve when there was no opportunity for stimuli to become familiar. Although females showed marked performance changes, there were no similar, consistent ERP changes over time. This indicates that performance changes were probably not caused by a strategy shift in the way that repeated images were processed. It appears that both sexes adopted relatively consistent strategies throughout the task which, for females, were more successful when the images were repeated and became familiar. In summary, males' and females' strategic approach to the mental rotation task differed during early stimulus processing and mental rotation. Because sex differences were not limited merely to mental rotation, it is recommended that future investigation of the root(s) of sex differences on the task explores each stage of processing independently. Additionally, image familiarity was found to be an important influence on females' processing, particularly when they were familiar with rotated and upright versions of the stimuli. Females showed clear performance benefits when they had mental representations of images at multiple orientations but these effects were not present for males. These benefits for females were interpreted with reference to sex difference frameworks which emphasise the underlying processes recruited to perform cognitive tasks (Colom, Escorio, & Rebollo, 2004; Halpern, 1992; Halpern & Wright, 1996; Vecchi & Girelli, 1998). Females were thought to have excelled at mental rotation of familiar images because of their high skill in retrieving previously stored mental representations from memory. The present research provides a basis for understanding, at least in part, why the magnitude of performance sex differences has varied so widely between studies (Voyer & Hou, 2006; Voyer & Saunders, 2004; Voyer et al., 1995): image familiarity has differed markedly between studies. Image familiarity differences between studies subsequently alter the underlying cognitive requirements of the mental rotation task and, ultimately, the likelihood that sex differences will occur.