the animal as a means of encoding a nuanced dis-cussion about the objectification of beings especially in Atlantic slavery.

This code is much easier to understand as it do not add levels of indentation and follows the principle where the 0 indentation level is the principal path of the application where other paths are exceptions or rare cases.

This is raising questions about the future of police robotics and what kind of oversight there should be.

Thus, our research contributes to aburgeoning literature on cross-group friendships by showingthat the positive effects of friendship can extend beyond inter-group attitudes per se to institutional attitudes, and by directlytesting causal links from cross-race friendships to positiveintergroup outcomes (cf. Pettigrew, 1998).

To the extent that a friend’s perceivedmembership in the university in-group is salient, cross-groupfriendship may increase the likelihood that minority-groupstudents will eventually incorporate a university identity as partof themselves.

Therefore, friendships with majority-group peers may be key in the development of dual identityamong minority-group students, and may provide a route towardrelational diversity within institutions of higher education.

Our research underscoresthe importance of the interpersonal climate for addressing issuesof access and diversity within such institutions, and shows thatthe development of affiliative ties across group boundariesprovides an important vehicle for achieving relational diversity.

Together, thefindings of these studies suggest that efforts to increase cross-group friendship are not incompatible with institutional effortsto clearly communicate acceptance of the minority group bysupporting organizations or activities centered on the ethnic orracial background of that group.

This analysisrevealed the predicted three-way interaction, b 5 0.80, F(1, 126)5 6.10, p < .02.

The model for university satisfaction revealed a significant in-teraction between number of majority-group friends and RS-race, b 5 0.46, F(1, 34) 5 7.19, p 5 .01.

The analysis for belonging revealed a significant main effect ofRS-race, b 5
0.25, F(1, 34) 5 6.17, p < .02.

In a longitudinalstudy of African American students (Study 1), cross-groupfriendships with majority-group peers buffered studentshigh in RS-race from lack of belonging and dissatisfactionat their university. An experimental intervention (Study 2)that induced cross-group friendship replicated the findingsand established their specificity for minority-group stu-dents.

These findings provide experimental evidence that cross-group friendship is beneficial forpeople who are likely to experience anxiety in intergroup contexts.

Cross-group friendship led to decreases incortisol reactivity (a hormonal correlate of stress; W. R. Lovallo & T. L. Thomas, 2000) over 3 friendshipmeetings among participants high in race-based rejection sensitivity (R. Mendoza-Denton, G. Downey,V. J. Purdie, A. Davis, & J. Pietrzak, 2002) and participants high in implicit prejudice (A. G. Greenwald,B. A. Nosek, & M. R. Banaji, 2003). Cross-group partners’ prior intergroup contact moderated therelationship between race-based rejection sensitivity and cortisol reactivity. Following the manipulation,participants kept daily diaries of their experiences in an ethnically diverse setting. Implicitly prejudicedparticipants initiated more intergroup interactions during the diary period after making a cross-groupfriend. Participants who had made a cross-group friend reported lower anxious mood during the diaryperiod, which compensated for greater anxious mood among participants high in race-based rejectionsensitivity.

she was only Anne.

Others have called these“Main Notes” or “Permanent Notes” or “EvergreenNotes”. I called them Point Notes to remind myself thatwhen I write them I should be making a point.

the ethics of using AI ineducation are political

To address these limitations, in Experiment 1, we requiredparticipants to rate their confidence after each individual response(rather than after each pair of responses).

In Experiment 4, we manipulated participants’ confidence priorto administration of the MRT. All participants first completed aline judgment task that was intentionally difficult, so that par-ticipants would be unable to gauge their performance.

Upon completion of the line judgment task, participants wererandomly informed that their performance on the line judgmenttask was either above average (‘‘high confidence’’condition) orbelow average (‘‘low confidence’’ condition).

Ifconfidence mediates mental rotation performance, then partic-ipants in the high confidence condition should outperform theircounterparts in the low confidence condition.

Cooke-Simpson and Voyer (2007) provided tentative evi-dence that confidence predicted MRT performance, but thatstudy had several critical limitations.

correlation was comparable to that observed in prior studies(r = ?.69; Cooke-Simpson & Voyer, 2007).

As illustrated in Fig. 2, confidence predicted accuracy acrossboth sexes, r(67) = ?.56, p\.001, and the strength of this

Thisresult replicates the general pattern observed in Experiment 1(Fig. 3), and indicates that both males’ and females’ confi-dence was indeed calibrated to their performance.

Experiment3 therefore supported the hypothesis that mental rotation ismediated by confidence.

Experiment 3 provided a further test of whether the sex differencein performance is better explained by confidence or by omissions.

Results are summarized in Table 1. The sex difference in accu-racy was replicated in the omission condition but not in the com-mission condition.

Thus, the sex difference in mental rotationwas attributable to confidence rather than omissions.

mediates mental rotation. If confidence was unrelated to mentalrotation, then the sex difference should be equivalent acrossgroups (i.e., no interaction should occur).

Critically, an inter-action in either direction would suggest that confidence

.

.

In Experiment 2, we sought to attenuate the sex difference inmental rotation performance by rendering confidence irrelevantto the task.

Fig. 4

In conclusion, Experiment 1 corroborated the finding thatconfidence predicted mental rotation performance both acrossand within sexes (Cooke-Simpson & Voyer, 2007). Experiment1 further demonstrated, for the first time, that confidence pre-dicted mental rotation performance within individuals: Partic-ipants were more accurate on trials for which they were moreconfident. These results thus provide the most precise evidenceto date of the relation between confidence and mental rotation.Finally, Experiment 1 also provided the first evidence of thedirection of this relationship: Mediation analyses revealed thatconfidence mediated the sex difference in mental rotation per-formance whereas mental rotation performance did not mediatethe sex difference in confidence.

If confidence mediates mental rotation performance,then confidence ought to predict accuracy on the MRT acrosssexes, within each sex, and possibly even within individuals.

In Experiment 1, we tested whether confidence predicted men-tal rotation performance between sexes, within each sex, andwithin individuals.

Thus, mental rotation performance did notmediate the sex difference in confidence, but rather confidencestrongly mediated the sex difference in mental rotation per-formance.

So, together, Figs. 2 and 3 reveal thatmales and females were similarly successful at calibrating theirconfidence to their accuracy (Fig. 3), though males tendedtoward the upper part of the distribution on both confidence andaccuracy (Fig. 2).

As expected, an independent samples t-test revealed that males(M = 5.61, SD = 1.02) were more confident than females (M =4.62, SD = 1.41), d = .74, t(65) = 3.26, p\.01.

Specifically, if confidence medi-ates mental rotation, then (1) confidence should predict mentalrotation scores not only between sexes, but also within sexes, (2)rendering confidence irrelevant to the task should attenuate thesex difference, and (3) manipulating participants’ confidenceshould affect their mental rotation performance.

.

Because confidence appears tobe an important component of the MRT, it stands as a plausiblemediator of performance.

The most common measure of mental rotation performanceis the Mental Rotations Test (MRT; Vandenberg & Kuse, 1978),which is based on the 3-dimensional block figures introduced byShepard and Metzler (1971) and updated by Peters et al. (1995).

.

Preliminary evidence suggests that confidence might indeedunderlie the sex difference in mental rotation.

.

We therefore tested whether confidence mediatedmental rotation performance.

basic cognitive skills, such as attention, memory, and judgment(e.g., Schmader et al., 2008), which ultimately would affectperformance.

The beliefthat one (or one’s social group) is skilled or poor at a given taskmay well affect one’s confidence when approaching that taskand this effect on confidence may have cascading effects on

Much of the research on gender role and sex stereotype effectsassumes confidence as a potential cognitive mechanism bywhich those social factors exert their effect.

So, in summary, beliefs about and aware-ness of sex stereotypes are both related to the sex difference inmental rotation performance. But how exactly might sex ste-reotypes affect performance?

Mental rotation performance may also be affected by mereawareness of, rather than belief in, the stereotype that men aresuperior to women on spatial tasks.

.

Performance on spatialtasks thus is clearly related to gender role beliefs and traits.

Gender role beliefs and traits may partially explain the sex dif-ference in mental rotation performance.

Here, we examined whether one suchsociocognitive factor, namely participants’ confidence, contrib-uted to this sex difference in mental rotation performance.Although this presumed relation between confidence and men-tal rotation performance has received little empirical attention,related research on gender roles, sex stereotypes, and stereotypethreat provides a rich source of supportive evidence.

.

Given the complexity of the taskand the magnitude of the sex difference, it likely has multi-ple causes or mediators.

Of all cognitive sex differences, the mental rotation of abstractfigures in 3-dimensional space is the most robust (Halpern, 2000;Hines, 2004; Linn & Petersen, 1985; Maccoby & Jacklin, 1974).

Thus, confidence medi-ates the sex difference in mental rotation performance and hencethe sex difference appears to be a difference of performancerather than ability. Results are discussed in relation to otherpotential mediators and mechanisms, such as gender roles,sex stereotypes, spatial experience, rotation strategies, work-ing memory, and spatial attention.

On tasks that require the mental rotation of 3-dimensional figures, males typically exhibit higher accuracythan females. Using the most common measure of mental rota-tion (i.e., the Mental Rotations Test), we investigated whetherindividual variability in confidence mediates this sex differ-ence in mental rotation performance.

.

Thus, the third approach to the issue of Sex differencesand the time factor in the MRT examined the sex differ-ence in a RT paradigm where only two mental rotationfigures were used.

.

For this reason, Study 2, which manipulatestime directly, compares sex differences under the stan-dard condition with sex differences which are observedwhen time is increased, but within limits.

Here, we administered the MRT under identical con-ditions, but allowing two durations.

.

If time pressure is a significant factor in per-formance, we expect to see two indicators in the data.First, we expect to see that females attempt significantlyfewer problems than males and, second, we expect to seethat the magnitude of the sex difference increases as theproblem position increases.

In the present study, three different approaches to theproblem of time constraints are taken, each examining adifferent aspect of how time might affect the sex differ-ences on the MRT.

Thus, our understanding of the role of time con-straints in MRT sex differences remains inconclusive.

Several studies find that when enough time is pro-vided for the V&K MRT, sex differences on the MRTdisappear (Goldstein, Haldane, & Mitchell, 1990;Voyer, 1997), leading to the conclusion that sex differ-ences arise because the sexes differ in the amount of timetaken to perform the mental rotation.

The idea that the activational effects of hormones mightlead to relatively faster mental rotations touches uponthe time factor in MRT sex differences.

That males and females differ in performance on Van-denberg and Kuse (1978) mental rotation task is wellknown (Voyer, Voyer, & Bryden, 1995). The causes areless well understood.

.

We conclude that performancefactors may play a role in sex difference on mental rotation tasks, but do not account for all of the differences.

In accounting for the well-established sex differences on mental rotation tasks that involve cube stimuli of the Shepard and Met-zler (Shepard & Metzler, 1971) kind, performance factors are frequently invoked.

.

Gender differences in favor of men in spatial abilities are one of the most commonly replicated finding inpsychology research (Hedges & Nowell, 1995; Linn & Petersen, 1985; Voyer, Voyer, & Bryden, 1995). Among thetasks that produce such differences, the Mental Rotations Test (MRT), developed by Vandenberg and Kuse (1978),produces the largest effects (Linn & Petersen, 1985; Peters, 2005).

Eighty undergraduate students (40 males, 40 females) completed the MRT while ratingtheir confidence in the accuracy of their answers for each item. As expected, gender differences in favor of men were obtained.Results also indicated a positive correlation between confidence ratings and scores on the MRT, as well as negative correlationsbetween confidence ratings and MRT outcomes presumed to reflect propensity to guess. More elaborate analyses using a measureof accuracy of predictions (the Brier score) indicated that men have a more accurate perception of their performance on the MRTthan women do.

Further support for a constant anduniform biological explanation comes from studies that findsimilar effect sizes for gender across cultures (Silverman,Phillips, & Silverman, 1996).

The distinct hormonal environments of men and womenmay play a role.

.

.

When brain imaging studies are done on subjects per-forming mental rotation tasks, there is some evidence ofactivation in motor areas of the brain.

.

Biological factors have also been discussed as potentialcauses for this difference. Biological theories stress theimportance of genetics, hormonal influence, brain organi-zation, and maturational factors.

There is,however, no consensus as to the role of task complexity ongender differences in mental rotation.

Task difficulty is one potential performance factor thathas been explored (Bryden, George, & Inch, 1990; Collins& Kimura, 1997).

.

.

A number of explanations have been advanced for theexistence of the gender difference in mental rotation.

Voyer, Voyer, & Bryden’s (1995) meta-analysis of sexdifferences in spatial abilities, found that the average differ-ence (using Cohen’s d = (M1 − M2)/σ) between men andwomen on the (MRT; Vandenberg & Kuse, 1978), was 0.94(this represents a very large effect), indicating that men per-form nearly one standard deviation above the average per-formance of women.

The ability to mentally rotate an object has been foundto produce one of the largest sex differences in the cogni-tive literature (Linn & Petersen, 1985).

The visuospatial ability referred to as mental rotation has been shown to produce one of the largest and most consistent sex differences,in favor of males, in the cognitive literature.

Sex differences were also seen in the patterns of correlations between rotation tasks and other neuropsycho-logical measures. Current results suggest men may rely more on left hemisphere processing than women when engaged in rotational tasks.© 2003 Elsevier Ltd. All rights reserved.

.

.

.

Pre-vious results with similar stimuli and mixed groups of participants(Han, Humphreys, & Chen, 1999; Kimchi, 1994) showed the globaladvantage that is typically observed with hierarchical stimuli (e.g.Kimchi, 1992; Navon, 1977), as well as faster classification by clo-sure than by line orientation. In addition, the global advantage wasmore pronounced when local classification was based on line ori-entation than on closure (Han et al., 1999; Kimchi, 1994).

Thus, in this study, we examined gender differences in global–local processing with a visuospatial judgment task (line orienta-tion) and a shape judgment task (open vs. closed shape).

.

The investigation of gender differences in global–localprocessing in the context of a spatial and a non-spatial task will al-low us to examine whether these differences, if they exist, charac-terize visual perception of women and men in general, or whetherthey are related only to visuospatial performance.

The main purpose of the present study was to systematicallyexamine gender differences in global–local processing.

The inconsistencies in the results of these studies are mostprobably due to the differences in stimulus and task variables,which are known to affect global–local performance, as discussedearlier.

Direct examinations of gender differences in global–local pro-cessing are sparse, and the results are equivocal.

.

.

The gender differences in navigation and way-finding also ap-pear to suggest a global bias for men versus a local bias for women.

.

.

Gender differences in spatial abilities have been reported in anumber of studies over the years (see Voyer, Voyer, & Bryden,1995, for a review).

Direct examinations of gender differences in global–local processing are sparse, and the results are incon-sistent. We examined this issue with a visuospatial judgment task and with a shape judgment task.Women and men were presented with hierarchical stimuli that varied in closure (open or closed shape)or in line orientation (oblique or horizontal/vertical) at the global or local level. The task was to classifythe stimuli on the basis of the variation at the global level (global classification) or at the local level (localclassification).

This finding suggests that women aremore distracted than men by misleading global oriented context when performing local orientation judg-ments, perhaps because women and men differ in their ability to use cognitive schemes to compensatefor the distracting effects of the global context. Our findings further suggest that whether or not genderdifferences arise depends not only on the nature of the visual task but also on the visual context.

.

Nevertheless, we conducted linear mixedmodels to statistically test for developmental effects to assess whether the magnitude of genderdifferences in confidence increases (or decreases) with grade and found no evidence ofdevelopmental effects in gender differences in confidence (Appendix 4).

Finally, our analyses included data from participants ranging from early childhood toadulthood, but we chose not to focus on developmental trends in our outcomes of interest.

.

.

Across allparticipants, we found the mean values of gamma were positive for both genders (boys/men: n =309, M = .18, SD = .31; girls/women: n = 400, M = .20, SD = .28), which suggests that participantshave some ability to monitor the accuracy of their estimates. However, no reliable gender differencewas observed for relative accuracy (forest plot and analyses are displayed in Appendix 3). Takentogether, the present outcomes suggest that although girls/women (as compared to boys/men) areless confident in their task performance, they are equally able to discriminate between estimates thatare more versus less precise.

Although we cannot calculate measures of absolute metacognitive accuracy, we were able toassess whether gender differences exist for relative accuracy, or the degree to which participants candiscriminate between number-line estimates that are more (vs. less) precise.

Because of these issues, assessing the degree of over- or under-confidence for the number-line estimation task will require future advances in measurement ofjudgments and performance (so they can be made on comparable scales) for this task.

In the current study, we found gender differences in confidence, even when controlling forperformance. We also found that the magnitude of the gender differences observed forconfidence were smaller than those for performance (g = .30 versus .52).

.

.

That is, as compared toboys/men, girls/women may generally have less task-specific efficacy for number-line esti-mation (either because of its math component, spatial component, or both) and may also haveless perceived familiarity with the numbers. In fact, lower self-efficacy may push some girls/women away from engaging in math tasks, which in turn could reduce their actual familiaritywith those numbers.

As mentioned in the Introduction, confidence judgments can be influ-enced by multiple theory- and experience-based factors (e.g., Undorf et al. 2018). The observedgender gap in confidence could be explained by differences in judgment cue use by girls/womenand boys/men.

Although considerable debate exists over the causes of such gender differences when they areobserved (e.g., Hyde 2014), we imagine that psychological (e.g., differences in math attitudes;Sidney et al. 2019), social (e.g., differences in early spatial experiences, such as exposure to spatiallanguage, media, and toys; Caldera et al. 1989; Doyle et al. 2012; Pruden and Levine 2017; orgendered stereotypes about math and spatial ability, McGlone and Aronson 2006; Moè andPazzaglia 2006), and possibly even biological factors (e.g., sexual dimorphism in the parietalcortex; Goldstein et al. 2001) could contribute to the gender differences observed in the number-line estimation task (as is the case for performance; e.g., Tosto et al. 2018).

.

However, gender remained astatistically significant predictor of confidence: Even when controlling for trial-level estimationprecision, girls/women were .038 points less confident in their estimates than were boys/men.

Thus, it appears that the average girl/woman was about 7%less confident than the average boy/man (.048 / .688 = .0697; Model A).

Using the fixed-effects estimates, girl’s/women’s confidence wasestimated to be .048 points lower than boy’s/men’s (p = .001; Model A).

Using linearmixed-effects models to predict confidence, we again found that girls/women were slightlyless confident than boys/men (replicating the findings from the Hedges’ g meta-analysis).We also found that these gender differences remained when estimation precision wasaccounted for.

To summarize the main outcomes, we again found thatgirls/women were less precise in their number-line estimates than were boys/men (replicatingthe findings from the Hedges’ g meta-analysis).

The fixed-effects model resulted in a similar overall mean effect size,g = .26, 95% CI [.10, .41], p = .002.

A gender difference occurred in confidence favoringboys/men (Fig. 2). The overall weighted effect size was g = .30, 95% CI [.12, .47], p = .002.No significant heterogeneity was observed among the effect sizes, Q(17) = 19.03, p = .33.

This indexrevealed a small, non-significant amount of heterogeneity among the effect sizes; I2 = 10.69%.

Our analyses revealed medium gender differences innumber-line estimation performance favoring boys/men (g = .52; Appendix 2).

In summary, the current study indicates that sex differences inglobal self-assessments of performance do not always coincide with sexdifferences in moment-to-moment spatial performance monitoring.Even though female students were in most cases less confident thanmale students in their general spatial ability, their trial-by-trial meta-cognitive monitoring accuracy was not impaired in either an absoluteor relative sense. Thus, female students appear to have relatively ac-curate perceptions of their spatial performance for spatial orientationand spatial visualization tasks.

.

Regardless, the current results show that even female studentspursuing STEM degrees are less confident than male students in theirability to reason spatially about some STEM related content. It is un-clear if these differences reflect true underconfidence or if they are dueto actual differences in academic spatial ability.

.

.

Given that female students have lower visual-spatial working memoryspans than male students (Voyer et al., 2017), they may be more sus-ceptible to monitoring errors in dynamic spatial domains that they arenot susceptible to in static spatial tasks.

Although we observed limited sex-related differences in monitoringaccuracy in the current study, more substantial sex-related differencescould be present in qualitatively different tasks that require dynamicspatial processing.

The cues people attend to can also vary when monitoring is pro-spective vs. retrospective (Nelson, 1990).

However, the sex differences in relativeaccuracy we observed for the PSTV:R, suggest that the quality of thecues students used to monitor their performance may have differed formen and women. It is unclear whether these differences are task-spe-cific or reflect sex differences in cue utilization during prospectivemonitoring that are less prevalent in retrospective monitoring tasks.

.

.

Furthermore, female STEMmajors had lower self-evaluations than their male counterparts of vi-suospatial abilities needed for scientific reasoning.

Across multiple spatial measures, female students displayedlower confidence in their item-level monitoring and global assessmentsof performance than did male students, even when no actual differencesin spatial performance were present (e.g., Paper Folding Test andSpatial Relations Test).

3.6. Self-perceptions of everyday and academic spatial ability

bsoluteaccuracy measures for each task were significantly positively correlatedfor male students. The same pattern was present for female studentswith the exception that their absolute accuracy for the Paper Foldingtest and PSVT:R were not significantly correlated. However, Fisher r-to-z tests indicated that there were no significant differences between themagnitudes of the absolute accuracy correlations for male or femalestudents. Relative accuracy measures for the Spatial Relations test andthe Paper Folding test were also significantly positively correlated. Noother correlations between relative accuracy measures were significantand no sex differences were present for these correlations. Correlationsbetween the absolute and relative accuracy measures were not sig-nificant with the exception that female student's relative and absoluteaccuracy for the PSVT:R was significantly negatively correlated. AFisher r-to-z test indicated that this negative correlation for femalestudents was significantly different from the correlation for male stu-dents, Z = 2.31, p < .05. T

Table 3 shows that both maleand female students were underconfident in their global predictionsand postdictions for their performance on the Spatial Relations Test andPSVT:R but their estimates were well calibrated for the Paper FoldingTest. There were no sex differences in the degree of underconfidencestudents displayed for either predictions or postdictions on the PSVT:Ror for predictions of performance on the Spatial Relations Test. Abso-lute accuracy for global postdictions of performance for the SpatialRelations Test were significantly lower for females than male studentswhich indicates that female students were more underconfident in theirperformance than male students after completing the Spatial RelationsTest. There were no sex differences for either global predictions orpostdictions for the Paper Folding test.

no reliable effects.

It shows that both female and malestudent performed consistently across each spatial measure, with po-sitive moderate to high correlations between most tests. Fisher r-to-ztests testing sex differences in the magnitude of each correlation found

Fisher r-to-z tests indicated that there were no sexdifferences in the relative accuracy for global prediction of performancefor the Spatial Relations Test (Male: r = 0.22, p = .01; Female:r = 0.37, p = .001), Z = 1.2, p = .23, Paper Folding Test (Male:r = 0.40, p = .001; Female: r = 0.29, p = .01), Z = 0.91, p = .36, orPSVT:R (Male: r = 0.48, p = .001; Female: r = 0.56, p = .001),Z = 0.90, p = .37. There were also no sex differences in relative accu-racy for global postdictions of performance on the Spatial Relations Test(Male: r = 0.44, p = .001; Female: r = 0.60, p = .001), Z = 1.61,p = .11, Paper Folding Test (Male: r = 0.71, p = .001; Female:r = 0.68, p = .001), Z = 0.42, p = .68, or PSVT:R (Male: r = 0.67,p = .001; Female: r = 0.59, p = .001), Z = 0.97, p = .32.

. Table 2 shows that male andfemale students were equally accurate in terms of both their absoluteand relative accuracy for all spatial measures except for the relativeaccuracy of performance on the PSVT:R. Surprisingly, female studentsdisplayed higher relative accuracy on the PSVT:R than did male stu-dents. Apparently they were better able to discriminate correct fromincorrect responses, even though males performed better on the test.

There were no significant sex differences in performance onthe Paper Folding test or the Spatial Relations test, but male studentsdid outperform female students on the PSVT:R. Although females per-formed as well as male students on several spatial tasks, female studentsconsistently predicted lower performance than male students on allspatial tasks. They generated lower mean CJs, global predictions, andglobal postdictions for each test, producing significant sex differences inevery task and for each prediction type except for global predictions forthe Paper Folding test.

Consistent with previous findings, male students per-formed significantly better on the symmetry span task than femalestudents. There were no sex differences in performance on the Raven'sProgressive Matrices task.

.

.

.

Absolute accuracy (also referred to as calibration) refers towhether the average magnitude of an individual's judgments corre-sponds to their overall level of performance. Relative accuracy refers toone's ability to discriminate between correct and incorrect spatial taskdecisions (i.e., manifest higher confidence for correct than for incorrectitem responses). In the current experiment, we compared sex differ-ences for both absolute and relative accuracy on measures of spatialorientation and spatial visualization.

.

Sex differences in spatial strategy use could also cause sex differ-ences in item-level monitoring accuracy. Metacognitive monitoring isan inferential process that involves evaluating cues (e.g. item char-acteristics, processing fluency, etc.) that are present at the time of amonitoring judgment and applying beliefs or heuristics to infer thequality of these processes (Dunlosky & Tauber, 2014; Koriat, 1997;Schwartz, Benjamin, & Bjork, 1997).

These differences instrategy preference may be due to differences in the accuracy of mon-itoring strategy effectiveness.

They also adopt differentstrategies than male students to solve spatial problems (Allen &Hogeland, 1978; Goldstein, Haldane, & Mitchell, 1990; Kail, Carter, &Pellegrino, 1979; Lohman, 1986; Miller & Santoni, 1986; Peña,Contreras, Shih, & Santacreu, 2008; Prinzel & Freeman, 1995; Raabe,Höger, & Delius, 2006; Tapley & Bryden, 1977).

The limited research examining sex differences in monitoring spa-tial cognition is especially surprising because sex differences in spatialcognitive performance have been indirectly linked to metacognitivevariables

.

Only a few studies have explored whether there are sex differencesin metacognitive monitoring accuracy in non-spatial domains (Hertzog,Dixon, & Hultsch, 1990; Lichtenstein & Fischhoff, 1981; Lundeberg,Fox, & Punćcohaŕ, 1994).

This reveals an interesting pattern:for the left ROI, both ipsi- and contralateral attention conditionslead to a decrease in alpha power compared with pre-cue values.However, in the right ROI, we observed a contralateral decreasebut a slight ipsilateral increase.

For the left ROI, the observed decreases were signifi-cant for both conditions (Fig. 6, p
0.05; significant time sam-ples indicated). For the right ROI, only the contralateral attentioncondition lead to a trend ( p  0.064).

A lin-ear regression analysis showed that with decreasing cue reliabil-ity, there was a strong trend toward decreasing differences in thealpha-lateralization index between correct and incorrect trials(R 2  0.043, p  0.081). This effect was significant for reactiontimes: with decreasing cue reliability, the difference in the alpha-lateralization index between low- and high-RT trials becomessmaller and eventually flips from positive to negative values(R 2  0.135, p
0.01)

As expected, in the 50% condition alpha-lateralization index val-ues were rather low and did not correlate with performance (data notshown): both correct and incorrect, and low- and high-RT trialsshowed similarly low alpha-lateralization index values (in the rangeof 0.01– 0.02; correct vs incorrect: t(17)  0.436, p  0.669; low vshigh RT: t(17)  0.375, p  0.712).

For discrimination rate this effect was not sig-nificant (t(17)  0.638, p  0.532), for RT a near significant trend wasobserved (t(17)  2.048, p  0.056).

This was further substantiated by analysis of the invalid cue trialsfrom the 75% condition, on which an opposite pattern was ob-served: high alpha lateralization was detrimental for performance oninvalid cue trials (Fig. 4C,D).

This analysis showsthat a higher alpha-lateralization index precedes better perfor-mance: both correct trials and fast RTs are related to high alphalateralization values whereas incorrect trials and slow RTs showless alpha lateralization. Paired-sample t tests confirmed thatthese differences were significant (correct vs incorrect: t(17) 2.187, p
0.05; low vs high RT: t(17)  2.556, p
0.05).

alpha-lateralization index showed no significant effect (R 2 0.000, p  0.964).

This decrease could not be explained by a difference in overallipsilateral plus contralateral alpha power between the conditions(data not shown), as a similar test on the denominator (normal-ized per subject using the average power over conditions) in the

There was a significant parametric decrease of thealpha-lateralization index with decreasing cue reliability (Fig.3B) as assessed by linear regression (R 2  0.150, p
0.01).

The alpha lateralization was significanton sensor level both in the 100% (see before) and 75% condi-tion ( p
0.01 for two clusters above left and right sensori-motor regions). For the 50% condition the effect was muchweaker, however, a significant cluster was found in sensorsover left sensorimotor regions ( p
0.05).

A cluster-based randomization test overthe 3D source space showed that the lateralized difference inalpha activity between attention-left and attention-right was sig-nificant for the right somatosensory source ( p
0.01) andshowed a trend for the left source ( p  0.062). Note that data ofonly 17 subjects was used for the source analysis.

A time-frequency analysis of the lower frequencies (5–35Hz) showed that alpha lateralization was sustained through-out the 1 s before stimulus onset (Fig. 2 B) and that none of theother lower frequencies between 5 and 35 Hz showed a sub-stantial modulation.

Spectral analysis revealed a lateralized pattern of alpha power. Acluster-based randomization test over the sensors further showedthat the alpha lateralization had two significant clusters of sensorsabove left and right sensorimotor regions ( p
0.05 for bothclusters) (Fig. 2 A).

To summarize, the behavioral results confirmed the expectedoutcome: performance on invalid trials was significantly worsethan on valid trials, both in terms of discrimination rate and RT.Invalid cues had a more detrimental effect on RT for the 75%condition than for the 50% condition. Subjects were faster on the100% condition than on the 75% or 50% conditions.

In terms of discrimination rate there were no differencesbetween the reliability conditions (100% vs 75%, t(17)  0.687,p  0.502; 100% vs 50%, t(17)  1.208, p  0.244), but subjectswere faster on the 100% condition compared with the other twoconditions (100% vs 75%, t(17)  2.445, p
0.05; 100% vs50%, t(17)  2.960, p
0.01).

Therewas neither a significant effect of reliability on discrimination rate(F(1,17)  0.847, p  0.370), nor on RT (F(1,17)  1.479, p 0.241). There was a significant effect of validity both on discrim-ination rate (F(1,17)  6.534, p
0.05) and on RT (F(1,17) 23.239, p
0.001), with higher discrimination rates and lowerRTs for validly cued trials. Furthermore, the interaction effectbetween reliability and validity was not significant for discrimi-nation rate (F(1,17)  0.458, p  0.508), but showed a highlysignificant effect for RT (F(1,17)  11.715, p
0.01).

.

.

Wehypothesized that prestimulus somatosensory alpha powerwould modulate with respect to attention and that thestrength of this modulation would increase parametricallywith cue reliability. Since we posit that alpha activity plays adirect role in modulating neuronal processing, we further hy-pothesized that prestimulus alpha would be predictive of so-matosensory discrimination performance.

These results indicate that the somatosensory alpha rhythmserves the same functional role as posterior alpha.

In support of such an alpha mechanism, visual attention isknown to modulate alpha activity over parieto-occipital cortex asmeasured with electroencephalography (Foxe et al., 1998).

.

In particular,alpha activity might serve to direct the flow of information throughthe brain and allocate resources to relevant regions (Jensen andMazaheri, 2010). This is consistent with previous work suggestingthat sensory alpha activity is involved in directing focal attention(Pfurtscheller and Lopes da Silva, 1999; Suffczynski et al., 2001).

Here, we investigated whether somatosensory alpha activity istop-down modulated according to the anticipation of sensoryinput. Further, we asked whether the alpha modulation has con-sequences for somatosensory discrimination performance.

Thisstudy demonstrates that prestimulus alpha lateralization in the somatosensory system behaves similarly to posterior alpha activityobserved in visual attention tasks. Our findings extend the notion that alpha band activity is involved in shaping the functional architec-ture of the working brain by determining both the engagement and disengagement of specific regions: the degree of anticipation modu-lates the alpha activity in sensory regions in a graded manner. Thus, the alpha activity is under top-down control and seems to play animportant role for setting the state of sensory regions to optimize processing.

The brain receives a rich flow of information which must be processed according to behavioral relevance. How is the state of the sensorysystem adjusted to up- or downregulate processing according to anticipation? We used magnetoencephalography to investigate whetherprestimulus alpha band activity (8 –14 Hz) reflects allocation of attentional resources in the human somatosensory system.