FMRI to probe sex-related differences in brain function with multitasking

Background Although established as a general notion in society, there is no solid scientific foundation for the existence of sex-differences in multitasking. Reaction time and accuracy in dual task conditions have an inverse relationship relative to single task, independently from sex. While a more disseminated network, parallel to decreasing accuracy and reaction time has been demonstrated in dual task fMRI studies, little is known so far whether there exist respective sex-related differences in activation. Methods We subjected 20 women (mean age = 25.45; SD = 5.23) and 20 men (mean age = 27.55; SD = 4.00) to a combined verbal and spatial fMRI paradigm at 3.0T to assess sex-related skills, based on the assumption that generally women better perform in verbal tasks while men do better in spatial tasks. We also obtained behavioral tests for verbal and spatial intelligence, attention, executive functions, and working memory. Results No differences between women and men were observed in behavioral measures of dual-tasking or cognitive performance. Generally, brain activation increased with higher task load, mainly in the bilateral inferior and prefrontal gyri, the anterior cingulum, thalamus, putamen and occipital areas. Comparing sexes, women showed increased activation in the inferior frontal gyrus in the verbal dual-task while men demonstrated increased activation in the precuneus and adjacent visual areas in the spatial task. Conclusion Against the background of equal cognitive and behavioral dual-task performance in both sexes, we provide first evidence for sex-related activation differences in functional networks for verbal and spatial dual-tasking.

Introduction However, it is not clear yet whether men and women process verbal and spatial dual-tasks differently. The aim of the present study thus was to identify such potential sex-related differences in cerebral activation patterns during dual-tasking. For this purpose, we generated a combined verbal and spatial fMRI paradigm not targeting on identifying differences on single task level.

Subjects
Forty age-matched (mean = 26.5 years; SD = 4.73; range = 18-36), right-handed, German native speaking healthy participants (20 female, 20 male) took part in this experiment. The ethics committee on human experimentation of the Medical University of Graz had approved the study. Subjects gave written informed consent.

Neuropsychological assessment
The following neuropsychological tests were used to control for cognitive performance: three subtests of the verbal "Intelligenz Struktur Test" (IST-2000R, German versions of an intelligence test involving completing sentences, analogy and similarities) and spatial IST-2000R (cubes, figure selection and matrices) [27], the Stroop Task (color word interference test) [28] and the Trail Making Test (version A and B) [29].

FMRI dual task paradigm
The paradigm consisted of 30-second active blocks (single task and dual task), alternately presented with 24-second rest phases characterized by a crosshair in the middle of the display. A verbal single task and a spatial single task were each combined with a one-back task, yielding a total of four conditions: (1) a verbal single task, (2) a spatial single task, (3) a verbal dual task, and (4) a spatial dual task. In addition, a working memory task was implemented as a single task (Fig 1). Participants completed five blocks of each dual task-condition and four blocks of each single-task condition in pseudo-randomized order (duration: 1230 seconds).
Participants had to decide on equality or disparity of letters during the verbal single task (e.g. D and d: different appearance but same meaning; P and q: different meaning) [24]. The following letters were used: D, d, P, p, Q, q, B, b.
During the spatial task, participants had to compare two arrows and decide whether they were matching or not after mental rotation (Fig 1, A = single task; B = dual-task; C = example) [24].
During the working memory one-back-task (secondary task), participants had to decide whether the current presented frame on the display had the same color as the frame presented one image before.
For all three single-tasks, participants had to press with both index fingers simultaneously using an fMRI compatible response box for a 'Yes, same meaning/color' and both middle fingers for a 'No, not the same meaning/color' response. This approach was chosen to take into account increased motor-related activation changes during dual task blocks. In the dual-task condition, people had to perform the primary task with the right response box and the secondary task with the left response box (index or middle finger, respectively). Stimuli were presented for 3000 ms and in the dual task condition the secondary task was presented with a fixed SOA of 200ms.
High pass temporal filtering was selected for 63s. FILM provided time-series with statistical analysis with local autocorrelation correction. FLIRT was used for registration to high-resolution/standard images. Higher-level-analyses were done using FLAME 1. A Z statistic threshold of z = 2.3 (verbal and spatial single-tasks; verbal and spatial dual-tasks) was used. Correction for multiple comparisons was conducted by using FDR in FSL. During first level analyses, all active conditions (single-task and dual-task conditions) were compared to baseline. Motion parameters were used as a covariate of no interest. Higher level analyses were conducted for Dual-Task < Single Task. Furthermore we conducted the contrast Men > Women, as well as Women > Men for both "Dual-Task < Single-Task"-contrasts (verbal and spatial). Two participants had to be excluded from the analyses due to severe movement artifacts (pre-defined as motion > 3mm). For higher-level statistics, the sample thus comprised 19 men and 19 women.

Contrast masking
Activation patterns of all study participants (i.e. of both men and women) for the verbal and the spatial dual-task condition (dual-task > single-task) were used to mask the activation patterns for the group comparison between men and women in order to ensure that only regions involved in the given task can yield significant sex differences. A Z statistic threshold for the masking of z = 2.3 was used.

Statistical analyses
The Statistical Package of Social Sciences (SPSS, Version 21) was used for inter-group comparison of behavioral data. Normal distribution was assessed with the Shapiro-Wilk Test. Whenever appropriate t-test was used for group comparisons. If data were not normally distributed Mann-Whitney U and Wilcoxon Test was used.

Neuropsychological results and demographics
Verbal and spatial IQ measured by six subtests of the IST2000R were comparable for men and women (verbal IQ: t(38) = 0.173, p = 0.863; spatial IQ: t(38) = -0.746, p = 0.461; compare online supplemental S1 Table). Also the executive functions measured by the Stroop Task, subtest interference and the Trail Making Test B did not show differences between men and women (U = 137.0, p = 0.091) U = 146.5, p = 0.149). Moreover, attention, as measured by two subtests of the Stroop task (Reading words: U = 187.0; p = 0.738, Naming colors: U = 163.5, p = 0.327) and the Trail Making Test A (U = 160.0, p = 0.289) did not reveal differences between sexes. Men and women did not differ regarding age (U = 136.0, p = 0.086) or educational level measured in years of education (U = 170.0, p = 0.429, Table 1).

Behavioral performance in the scanner
Overall, RT and decreasing accuracy in the dual-task conditions compared to the single-task conditions for primary and secondary tasks. There is no sex-related difference in the reaction time or the accuracy in dual-task or single-task ( Table 1). The reaction time for the working memory task increased significantly in the verbal and spatial condition compared to the single-task condition (verbal: Z = -5.256, p<0.001, spatial: Z = -5.202, p<0.001). Accuracy for the working memory task was lower in the dual-task conditions compared to the single-task condition (verbal: Z = -5.512, p<0.001; spatial: Z = -5.196, p<0.001). Also, the primary tasks showed increased reaction time and less accuracy in the dual-task condition in the whole sample. The verbal dual-task reaction time increased compared to the single-task (Z = -5.377, p<0.001), and the accuracy decreased (Z = -5.499, p < 0.001). The spatial dual-task reaction time (Z = -2.61, p = 0.009) was significantly higher and the accuracy was lower (Z = -4.489, p < 0.001) than in the spatial single-task condition.
Considering sex, no differences in behavioral performance were found. Men and women showed no significant differences in reaction times and accuracy in all single-tasks (working memory task, verbal single-task, spatial single-task) and dual-task conditions (verbal dual-task, spatial dual-task). Also the secondary task demonstrated no differences between the sexes during the dual-task condition (see Fig 2 for the actual distribution of the behavioral performance).

Functional MRI data: Task specific activation
Whole brain analyses independent of sex for the verbal dual-task vs. verbal single-task revealed a network comprising the right superior frontal gyrus, the right medial frontal gyrus, the right precuneus, the lateral occipital cortex, the right occipital fusiform gyrus, the left superior frontal gyrus and the left middle frontal gyrus.
The comparison of the spatial dual-task vs. spatial single-task indicated increased activation for the dual-task condition in the following regions: the left inferior frontal gyrus, the bilateral middle frontal gyrus, the left orbitofrontal gyrus, the right anterior cingulate and left anterior paracingulate as well as the bilateral precuneus and left lateral occipital cortex. Subcortical regions such as the left thalamus, the left caudate, the left insula and left putamen were also identified as active for the spatial dual-task condition compared with the spatial single-task  Tables 2 and 3 for the whole brain analyses of both dual-task conditions in comparison to the single-task conditions.

Group comparisons
Compared to men, women showed regional increased network activation in the right anterior paracingulate gyrus as well as in the right prefrontal cortex and the left orbitofrontal cortex in the verbal dual-task (masked with the mean dual-task activation of women). In contrast, men did not show increased activation in any area when compared to women in the verbal dualtask (see Fig 3B and Table 4).
Considering the spatial dual-task, men compared to women showed increased activation within the left intracalcarine cortex and left lateral occipital cortex in the spatial single-task and masked with the mean activation of men for the spatial dual-task. In contrast, women did not show increased activation in any region compared to men for the spatial dual-task related to the spatial single-task and masked with the mean activation for women in the spatial dualtask condition (Fig 3B, Table 5).

Group differences in single-task conditions (compared to baseline)
Contrasts between single-tasks and baseline activation of the verbal single-task condition showed increased activation for men in the right postcentral gyrus, the right superior parietal lobe, the right planum temporale, the right precuneus, the left prefrontal cortex, the left superior occipital lobe, the left superior temporal lobe, the left orbitofrontal cortex, the left superior parietal lobe, the left amygdala, and the left hippocampus. Women did not show increased activation in any region for the verbal single-task. Women showed increased activation for the spatial single-task in the left inferior frontal lobe (pars triangularis). Men did not show any region of increased activation. For the working memory task, men showed increased activation in the precuneus and intracalcarine cortex compared to women. Women did not show increased activation for the working memory single-task (see Table 6 for the local maxima).

Discussion
The aim of this study was to investigate differences in behavioral performance and functional cerebral activation between men and women during dual-tasking. For that, a sex-specific fMRI paradigm was developed. After appropriately developing the tasks and selecting participants in a fashion to eliminate possible difference in behavioral performance, differences in fMRI activation between verbal and spatial dual-tasks were observed between men and women. Based on fMRI data, we here thus provide first evidence for sex-related neurophysiological differences during dual-tasking, which are fully independent of the behavioral performance level.
Offer and Schneider [30] demonstrated that women are forced to complete two tasks at the same time ten hours more often a week than men what suggest a better ability to perform multitasking. General knowledge regarding sex differences is assumed because of differences in Sex-related differences in multitasking daily living between men and women concerning housework and child care [30]. Compared to this, behavioral multitasking studies already suggest no differences in multitasking [1].
One way to operationalize multitasking and possible sex differences is a dual-task paradigm [3]. As a methodological strength, in the current study age, education, attention, and executive function were matched between sexes, strongly suggesting that the observed sex-related activation differences with the tasks exist merely due to processing differences in the brain. Consistent with existing literature, we failed to identify differences in behavioral performance between women and men for dual-tasking [1]. Men and women also showed equal behavioral performance on the single-task level, in line with the study by Neubauer et al. [24] who used a similar paradigm.
We could not replicate the behavioral sex-related differences in working-memory tasks described by previous empirical studies, what is inconsistent with existing literature [18]. However, as others, we noted increased reaction times and decreasing accuracy in the dualtask conditions compared to the single-task conditions [4,9]. Comparable to the bottleneck theory we found increased reaction times for the secondary tasks. However, increased reaction times were also found for the primary task and not just for the secondary task, as predicted by Pashler [31]. Stelzel et al. [9] and Herath et al. [4], who reported similar results as the present study, interpreted this effect by motor interference as determinants of increased reaction times and errors, for the primary as well as for the secondary task. Importantly, these effects were not sex-specific.
In fMRI, women compared to men showed increased activation for the spatial single task condition, while men demonstrated higher activation in the verbal single task condition. These activation differences for the verbal and spatial single-task conditions are comparable to previous studies [22,25,32]. However, for dual-task conditions, women compared to men demonstrated increased activation for the verbal dual-task compared to the verbal single-task. Such difference between single and dual-task conditions was not observed in men. Parts of the Sex-related differences in multitasking frontal gyrus and the paracingulate yielded increased activation for women over men while performing two tasks simultaneously, if the primary task required processing of verbal stimuli. This greater regional activation lead to the same behavioral performance, supporting the interpretation that women activate the inferior frontal gyrus much more for the processing of verbal stimuli if they appear in a dual-task condition [33]. This result appears consistent with previous observations that women show increased and more bilateral activation patterns for verbal tasks [22,25].
For the spatial dual-task, activation in the spatial dual-task compared to the spatial singletask was increased in men over women, but vice versa there were no differences. This activation in areas of the occipital cortex implicated in visual processing might be related to analyzing the spatial stimuli [34]. One interpretation might refer to a more scrutinized analysis of spatial stimuli by men compared to women, if spatial processing is needed for dual-tasking, consistent with the findings of Semrud-Clikeman et al. [32], who have shown that visual attention plays a greater part for men in mental rotation tasks. This study also showed increased activation in the left middle occipital cortex in men while performing mental rotation tasks. In contrast to the present findings, this activation difference was associated with a better performance of men in mental rotation tasks, similar to previous studies showing associations between better performance and decreased activation [35].
These results partly also could contribute to the entities of neuronal efficiency vs. expertise. Parallel to comparable behavioral performance between sexes, men showed increased activation during the verbal single task, while women showed increased activation in the spatial single task, which may lead to the assumption that men process spatial tasks more efficiently while women do so for verbal tasks to achieve a similar cognitive performance (neuronal efficiency). However, the opposite was found for the more complex task, which included the Sex-related differences in multitasking additional load for working memory. Here, women showed increased activation in the frontal cortex during the execution of the verbal dual-task, while men showed increased activation in occipital areas during the spatial dual-task, despite comparable behavioral performance. Whilst for the performances in the single task condition a ceiling effect can be determined (verbal single-task ! 9.25 of max. 10 correct responses, mean = 9.74±0.22), a similar effect does not hold true for the dual task condition (verbal dual-task !2.40 of max. 10 correct responses, mean = 7.72±1.94). Therefore, it is thinkable that both sexes evolve different mental strategies to solve the more complex task, which might be based on prior expertise. Some limitations of our study also need to be considered. Clearly, verbal and spatial dualtasking should not be regarded as the only approach to assess sex-related differences in multitasking. It is therefore important to emphasize that this study only explains the difference between men and women in processing verbal and spatial tasks in combination with a secondary task. Further investigations thus are necessary to more precisely elaborate on sex related differences in multitasking. Sex-related differences in multitasking

Conclusion
We here implemented verbal and spatial tasks together with a working memory task in a dualtask paradigm to assess sex-related differences in dual-tasking. No behavioral differences in accuracy or reaction times were noted, but activation differences for the verbal dual-task and the spatial dual-task were found between the sexes. Men and women varied regionally in functional activation in several areas. The fMRI findings suggested that women need an increased involvement of frontal regions to analyze verbal stimuli in a verbal dual-task condition, while men resort to an increased activation of visual areas to analyze the spatial stimuli in a spatial dual-task condition.
Supporting information S1 Table. Minimal raw data of the study for all participants. (XLSX)