Women and men have a similar potential for malevolent creativity – But their underlying brain mechanisms are different

Research interest in gender differences in aggression and creative ideation cumulates in the phenomenon of malevolent creativity. Taking another critical step in understanding malevolent creativity, we investigated gender differences in brain activation and functional coupling of cortical sites in the EEG alpha band while n = 88 women and men purposefully generated malevolent creative ideas for taking revenge on others. Results showed that malevolent creativity performance between the genders was similar; however, their underlying EEG patterns were markedly different. While women exhibited a steep decrease of task-related alpha power from frontal to left central-temporal, men ’ s malevolent creative ideation was characterized by a more diffuse pattern of task-related alpha power changes, along with decreased frontal-central coupling. Per interpretation, women ’ s malevolent creative thinking may more strongly rely on controlled semantic memory retrieval and novel re- combination of social/relationship information, while men may utilize more automatic motor-related imagery that may predominantly facilitate physical revenge ideation. Our findings add novel evidence to the idea that women and men engage different neurocognitive strategies to achieve similar creative performance and may help to further illuminate the darker side of creative ideation.


Introduction
Research has long been intrigued with gender differences in cognitive, social, and affective functions. Putting the spotlight on maladaptive behavior, aggression research usually finds men to be more physically aggressive than women (Card et al., 2008). Comparatively, studies focused on more positive, thriving aspects of human behavior are facing more difficulties in disentangling subtle gender differences in complex cognitive phenomena like creativity (Abdulla Alabbasi et al., 2022). The present study extends both aggression and creativity research avenues by examining gender differences in functional patterns of brain activity during malevolent creativitywhere creative ideas are intentionally generated to damage others (Cropley et al., 2008). We provide a combined EEG analysis of functional patterns of brain activation (taskrelated alpha power changes; TRP) and connectivity in terms of functional coupling of cortical sites (task-related phase locking; TRPL) to elucidate which neurocognitive processes facilitate harmful creative ideation in women and men.

Gender differences in aggression
With increasing knowledge on the different, at times highly subtle manifestations of aggression, literature largely reports more direct, physical aggression in men compared to women (Archer, 2004;Card et al., 2008;Nivette et al., 2019). This male physical aggression bias is often associated with learned social roles, but also gender-dependent variations in testosterone levels (Björkqvist, 2018;Book et al., 2001). By contrast, women are often found to prefer indirect (social, relational) aggression like social manipulation and social exclusion (Björkqvist, 2018;Hess & Hagen, 2006); however, this latter finding is markedly less robust (Card et al., 2008). Both direct aggression (getting beaten up) and indirect aggression (being the victim of vicious rumors) may be equally severe, depending on their frequency or the perpetrator's relationship to the target. Crucially however, their damaging potential may be further amplified if aggressive behavior is original and unexpected, and thus, difficult to detect and to defend against (Harris & Reiter-Palmon, 2015;Perchtold-Stefan et al., 2022b) which happens in the case of malevolent creativity.

Malevolent creativity as original aggressive behavior
Similar to positive creative ideation, individuals show striking creativity when aggressing against others in daily life, as evident in instances of creative lies, bullying, blackmailing, theft, assault, defamation, or malicious pranks (Harris & Reiter-Palmon, 2015;Perchtold-Stefan et al., 2022b). Malevolent creativity implies that creative ideas are explicitly and intentionally generated to physically, mentally, or financially harm others (Cropley et al., 2008), which distinguishes it from negative creativity, where harmful ideas emerge as unintended byproducts of divergent thinking (Kapoor & Khan, 2016). While naturally correlated with general creative potential (Hao et al., 2016(Hao et al., , 2020Perchtold-Stefan et al., 2021a), malevolent creativity is also consistently linked to antagonistic and exploitative personality traits (Kapoor and Kaufman, 2022;Jonason et al., 2017;Perchtold-Stefan et al., 2022a). The positive relationship of malevolent creativity with Dark Triad traits (narcissismgrandiosity and self-orientation; psychopathycallous social attitudes and lack of empathy/remorse; Machiavellianismmanipulative behavior and exploitation of others; Furnham et al., 2013) is particularly interesting, as men consistently score higher on the Dark Triad than women, with gender differences being most robust for psychopathy and Machiavellianism (Jonason et al., 2020(Jonason et al., , 2022. Together with the extensively investigated gender differences in aggression as well as general creativity, gender differences in the Dark Triad provide another interesting basis for speculations on gender differences in malevolent creativity.

Gender differences in malevolent creativity
Still subject to substantial heterogeneity, malevolent creativity assessment can be broadly distinguished into three approaches, which provide a fragmented picture on gender differences in malevolent creative ideation. Regarding (1) the spontaneous occurrence of malevolent ideas in classic divergent thinking tests (e.g., alternate uses test, using a fork as a murder weapon), men seem to produce a larger number of harmful ideas than women (Dumas & Strickland, 2018;Lee & Dow, 2011; but see Kapoor, 2019). Findings for (2) self-reported malevolent creativity behavior (Malevolent Creativity Behavior Scale, MCBS; Hao et al., 2016) are more mixed, with some studies reporting higher malevolent creativity in men (Zhao et al., 2022;Xu et al., 2022), while others report higher scores in women (Al-Mahdawi et al., 2022) or no significant gender differences (Kapoor and Kaufman, 2022). Finally, (3) performance-based approaches thatin line with common psychometric creativity testsinstruct participants to generate different creative ideas to take revenge on others partially find higher malevolent creativity in men (Harris & Reiter-Palmon, 2015;Perchtold-Stefan et al., 2021a,b), but have also reported that malevolence of ideas (i.e., harmfulness) may be more easily increased in women depending on social context (Perchtold-Stefan et al., 2022b).
Given this divergence in behavioral findings, the present study took a neurophysiological EEG approach to investigating gender differences in the generation of malevolent creative ideas. We follow recent insights from creativity research that women and men may use different neurocognitive strategies to achieve similar creative performance, even in more real-life domains of creative thinking (Rominger et al., 2020a).

Gender differences in general creativity and associated brain activation
Similar to malevolent creativity, gender differences in general creative performance have been inconclusive, with a large variety of studies denoting slight advantages for either women or men, depending on creativity domain, or the investigation of creative potential vs creative achievement (Abdulla Alabbasi et al., 2022;Abraham, 2016;Baer & Kaufman, 2008;He & Wong, 2021;Hora et al., 2021;Taylor & Barbot, 2021). Interestingly, while behavioral studies support a gender similarity account of creative performance, neuroscientific studies usually report significant differences in activation and communication of different brain areas during creative idea generation (Abraham et al., 2014;Fink & Neubauer, 2006;Razumnikova et al., 2009;Rominger et al., 2020a;Ryman et al., 2014;Silberstein et al., 2019;Zhang et al., 2022).
Using the well-established EEG method of analyzing task-related alpha power changes (TRP), Fink and Neubauer (2006) found that creative ideation in men was accompanied by greater posterior alpha power increases in the right hemisphere, while women (particularly those with higher verbal IQ) displayed greater bilateral frontal and central alpha power increases. Similarly, Rominger et al. (2020a) reported stronger prefrontal involvement when women generated creative soccer moves, while men exhibited stronger parietal-occipital involvement. While the investigation of alpha TRP changes has become a central pillar of EEG creativity research Stevens & Zabelina, 2019), studies further draw on the informative potential of connectivity-based approaches to disclose neuronal integration and functional coupling of brain regions during creative ideation (Beaty et al., 2015(Beaty et al., , 2016Jaušovec, 2000;Rominger et al., 2019Rominger et al., , 2020aRominger et al., , 2022Zhou et al., 2018). Here, various structural and functional imaging studies suggest that during creative idea generation, women may recruit more brain networks than men, like the default mode network, or the executive control network (Rominger et al., 2020a,b;Takeuchi et al., 2017;Tomasi & Volkow, 2012; but see Ryman et al., 2014). Moreover, differently lateralized frontal-parietal connectivity changes in women (left parietal to frontal) compared to men (right parietal to frontal) have also been reported (Silberstein et al., 2019).
The existence of (various) neurophysiological, but not behavioral gender differences in creative ideation supports the assumption that women and men follow different neurocognitive routes that result in similar creative outcomes (Abraham et al., 2014;Abraham, 2016;Rominger et al., 2020a). Applied to malevolent creativity, this could mean that although both genders are equally capable of generating original ideas for destructive purposes, their involved processes and strategies may be differentwhich may critically advance our understanding of the emergence of malevolent ideation in daily life.

EEG approach of the present study
The present study investigated both functional EEG activation (taskrelated power changes) and connectivity patterns (task-related phase locking across different cortical sites) while women and men generated creative ideas for the purpose of taking revenge on unfair peers/associates in the Malevolent Creativity Test (MCT; Perchtold-Stefan et al. 2021a). Participants are shown close-to-real-life, anger-evoking situations and are instructed to generate as many original revenge ideas as possible within 3 min time per item. In line with psychometric creativity tests, ideas are rated for fluency and originality, but also malevolence (degree of harmfulness). The MCT has proven highly suitable for elicitation of malevolent creativity in the general population (Perchtold-Stefan et al., 2021a,b,c;Perchtold-Stefan et al., 2022a,b).

Expectations for malevolent creativity performance
As our first expectation (1), we hypothesized that women and men would in general show similar malevolent creativity performance (fluency, originality of ideas). However, based on the idea of different neurocognitive strategies for creativity (Abraham, 2016) and insights from aggression research (Archer, 2004), men and women may differ in which types of malevolent ideas they generate. Therefore, we implemented a novel MCT scoring scheme that in addition to creativity ratings, also differentiated between ideas featuring different types of aggression (see 2.4.2). We expected that men may generate a greater number of ideas focused on physical aggression, while women generate a greater number of ideas focus on social aggression (Archer, 2004;Card et al., 2008).

Expectations for TRP changes
As the only study examining task-related alpha power changes during malevolent creativity, Perchtold-Stefan et al. (2023) reported similar brain activation patterns as observed during general verbal divergent thinking. Using the very same task as the present study (MCT), the authors observed prefrontal alpha power increases at the beginning of the creative ideation process, followed by posterior-temporal alpha power increases at later stages of creative ideation (Agnoli et al., 2020;Benedek et al., 2014a,b;Fink et al., 2009a,b;Rominger et al., 2020a,b), which were all more pronounced in the right hemisphere. These TRP changes were also directly linked to malevolent creativity performance and were interpreted in terms of conceptual expansion from prosocial to more antisocial perspectives as well as semantic inhibition of obvious associations to generate novel revenge ideas. For the gender differences focus of the present study, (2) we hypothesized that women in contrast to men may show stronger alpha power increases at frontal sites, which would indicate greater engagement of executive control functions in creative ideation (Fink & Neubauer, 2006;Rominger et al., 2020a).

Expectations for TRPL changes
Functional connectivity between default mode and executive brain networks is believed to reflect the intricate interplay of spontaneous idea generation and top-down idea evaluation during creative ideation (Beaty et al., 2015;2016). Recently, increased coupling of the executive and default mode network were also found during malevolent creativity, providing further validity of this concept (Gao et al., 2022). Several EEG studies also report increased frontal-parietal coupling during different creative tasks, generally suggesting greater top-down inhibition and internally directed attention during creative thinking (Petsche, 1996;Rominger et al., 2019Rominger et al., , 2020a. However, it has been noted that frontaltemporal coupling and frontal-central coupling may also facilitate specific creativity-relevant functions like top-down control of semantic processes (Luft et al., 2018;Zhou et al., 2018) or planning of movement and motor imagery (Wang et al., 2006;Rominger et al., 2020a). Following previous suggestions that women may show higher integration of information during creative ideation (Razumnikova et al., 2009;Rominger et al., 2020a, Takeuchi et al., 2017, we speculate that (3) women would show higher functional coupling between frontaltemporal and/or frontal-parietal/occipital sites during malevolent creativity compared to men.

Participants
For the analysis of gender differences in brain activation during malevolent creativity, data from the study of Perchtold-Stefan et al. (2023) were used. In total, n = 100 participants were recruited. The sample size was chosen based on previous TRP and TRPL studies investigating creative thinking (also in more real-life domains), which reported average effect sizes of np 2 = 0.15 (between np 2 = 0.05 and 0.24) for e.g., AREA × HEMISPHERE interactions (Agnoli et al., 2020;Fink et al., 2017;Rominger et al., 2019). However, after data collection, n = 9 participants had to be excluded due to malfunctioning of the audio device recording participants' MCT answers; another n = 3 participants were excluded due to an excessive amount of artifacts. The final sample comprised n = 88 participants (51 women, 37 men), aged between 18 and 46 years (M = 25.41, SD = 5.05), which still exceeded the n = 84 required by a priori power analyses with G*Power (np 2 = 0.15, α = 0.05, 1-β = 0.80). A total of 95.4 % of our sample reported owning a college degree (51.1 %) or an undergraduate bachelor's degree (44.3 %). All participants were right-handed, and reported no psychiatric or neurological illness, drug use or medication intake, and came to the EEG session well rested. Written informed consent was obtained from all participants. The study was approved by the authorized local ethics committee (GZ. 39/73/63 ex 2020/21).

Self-report measures prior to EEG recording 2.2.1. Current mood
The German version of the Profile of Mood States (Dalbert, 1992) assesses participants' current positive (6 items) and negative mood (13 items) prior to engaging in malevolent creativity. Items are rated on a scale from 1 (not at all) to 7 (very strong; α positive mood = 0.90; α negative mood = 0.87).

Antagonistic personality
The German version of the Personality Inventory for DSM-5 -the PID-5 - (Zimmermann et al., 2014) was used to assess the personality domain of Antagonism (manipulativeness, deceitfulness, grandiosity). Forty-three items are rated on a 4-point Likert scale from 0 (very false or often false) to 3 (very true or often true; e.g., α = 0.93; e.g., "It is easy for me to take advantage of others).

Self-reported malevolent creativity behavior
On the Malevolent Creativity Behavior Scale (MCBS; Hao et al., 2016), participants report how often they engage in malevolent creativity behaviors in everyday life (e.g., deceptions, tricks, lies, revenge, etc.). The MCBS consists of 13 items rated on a 5-point Likert scale from 0 (never) to 4 (usually; α = 0.87; e.g., "How often do you engage in original forms of sabotage?").

Malevolent creativity task during EEG
The adapted EEG version of the Malevolent Creativity test (MCT; Perchtold-Stefan et al., 2021a;Perchtold-Stefan et al., 2023) consists of six negative social scenarios in which participants are treated unfairly by others and have to generate as many original ideas as possible to take revenge on these wrongdoers. In Item 1 (neighbor) for example, participants face an unfair neighbor promising them money for helping with flat renovations, but after the work is done, insists that this agreement never existed (see Fig. 1).
Each EEG trial started with a reference period of 10 s (fixation cross), followed by a 30 s item presentation featuring the description of a negative social scenario together with a picture. Participants were told to immerse themselves in the situation and imagine it as vividly as possible. During the subsequent ideation phase (3 min per item), participants voiced as many original (high quality) ideas as possible to exert revenge on the wrongdoers. Per default, participants saw a white question mark on screen that changed to green when they pressed the space bar, which indicated the start of idea vocalization. Participants voiced their idea into a microphone in one or two short sentences and then pressed the space bar again (question mark returned to white) to resume the task until the 3 min had elapsed. Comparative analyses confirmed that this think-aloud approach did not diminish sample means of malevolent creativity compared to pen-and-paper protocols in previous behavioral studies (Perchtold-Stefan et al., 2021a,b,c, Perchtold-Stefan et al., 2022a, which speaks against the idea that participants may have been inhibited by social desirability concerns.

Quantification of malevolent creativity performance 2.4.1. Standard MCT scoring
Applying standard MCT scoring (Perchtold-Stefan et al., 2021a,b, Perchtold-Stefan et al., 2022a,b, Perchtold-Stefan et al., 2023, ideational fluency was scored as the total number of generated ideas meeting the criterion of (at least slight) malevolence, based on the assessment of two independent raters (ICC two-way random, absolute = 0.93).
Similar to known psychometric creativity tests like the alternate uses test (Fink et al., 2009;Perchtold-Stefan et al., 2020;Rominger et al., 2022), originality (creative quality) of ideas was rated by five independent raters (3 female, 2 male) using a scoring scheme from 1 (low originality, everybody would come up with this idea) to 6 (high originality, only very few would come up with this idea). Interrater reliability was ICC = 0.87 (two-way random, consistency).
In separate data sets, the same five raters rated malevolence (harmfulness) of ideas with a time-delay of two weeks to minimize rating-bias from scoring originality. Here, a score of 1 indicated low malevolence (very little damage), while a score of 6 indicated high malevolence (extreme damage). Interrater reliability was ICC = 0.99 (two-way random, consistency). Examples for ideas high and low in originality and malevolence are illustrated in Fig. 1.
The total score of malevolent creativity was computed by summing up the number of malevolent ideas with an average originality rating of ≥ 3 (=moderately original). This approach reflects the notion that for "true" malevolent creativity, ideas should be both, malevolent as well as original (Harris & Reiter-Palmon, 2015;Perchtold-Stefan et al., 2021a,b, c, Perchtold-Stefan et al., 2022a.

New approach: Content categorization of MCT answers
Since research reports greater use of direct, physical aggression in men and greater use of indirect (social, relational) use of aggression in women (Archer, 2004;Björkqvist, 2018;Card et al., 2008), we adopted a similar categorization approach to the MCT that allowed us to examine gender differences in different types of malevolent creativity. Two independent raters classified the generated creative ideas into five different categories: a) physical aggression (ideas that focused on bodily harm to the wrongdoer, examples include pushing someone, beating someone up, assault, murder), b) property aggression (ideas that focus on destroying the property of others, examples include damage to the wrongdoers' car, furniture, clothes, home, etc.), c) social aggression (lying, defamation, public humiliation, damage to relationships, negative social media posts, etc.), d) threat-related aggression (making threats, blackmailing, harassment like nightly calls, stalking, etc.) and. e) tricks (smaller deeds like tricks or pranks with only minor physical, social, or property damage like putting glue in the mailbox, hiding things, etc.).
Interrater reliability (ICC, two-way random, absolute) was excellent for physical aggression (ICC = 0.96), social aggression (ICC = 0.96), and property aggression (ICC = 0.94), good for threat-related aggression (ICC = 0.88), and moderate for tricks (ICC = 0.74). For analyses, the number of creative ideas from each category was divided by overall fluency of ideas to gain percentage scores, i.e., relative shares of specific aggression themes among all generated ideas.

EEG recording and analysis
EEG was recorded from 19 electrode positions (FP1, FP2, F3, F7, FZ, F4, F8, C3, CZ, C4, T7, T8, P3, PZ, P4, P7, P8, O1, O2) using a Brainvision ActiCHamp Plus Research Amplifier (Brain Products TM ; 1000 Hz sampling rate). Recording took place in a quiet, shielded examination room. The ground electrode was located on the forehead, the reference electrode was positioned on the nose Rominger et al., 2018, Rominger et al., 2020a. Vertical and horizontal electrooculograms (EOGs) were measured with two bipolar channels for horizontal and vertical eye movements. Electrode impedances were kept below 30 kΩ for EEG electrodes, and below 10 kΩ for the ground and reference. The EEG signal was preprocessed by removing drifts (0.1 Hz) and low pass filtering (50 Hz) and re-referenced to an averaged ears reference.
The g.BSanalyze software (g.tec™, Graz, Austria) was used to manually check for artifacts and to calculate the band power values (μV2) by squaring the filtered EEG signals (10-12 Hz; FFT-filter with a window size of 250 samples and an overlap of 249 samples was used). Only band power and phase-locking values from artifact free timeintervals were averaged. For the TRP and the TRPL analyses, the 8-s interval from 1 s after onset of the fixation cross until 1 s before its offset served as the reference interval and the period starting 500 ms after stimulus onset until 500 ms before the IDEA button was pressed served as the activation interval (Rominger et al., 2019).
Similar to other EEG studies on divergent thinking, TRP scores were quantified for the upper alpha band (10-12 Hz) for an electrode i by subtracting the log-transformed power of the reference period (Pow i , reference ) from that of the activation period (Pow i , activation ) for each trial j separately, according to the formula: TRP i = Median (log (Pow i,activiation ) j ) -Median (log (Pow i,reference ) j ). Consequently, negative values indicate a decrease of TRP from the reference to the activation period, while positive values indicate an increase (Pfurtscheller and Da Silva, 1999).
Phase locking values were obtained by computing the convolution of each of the two signals by means of a wavelet centered at a frequency of 11 Hz and a bandwidth of 2 Hz (10-12 Hz; Grabner et al., 2007;Rominger et al., 2019;2020a,b). We extracted the phase of the convolution and used it as the index of phase-locking, which varies between 0 (independent signals, no functional coupling) and 1 (constant phase lag between two signals, perfect functional coupling; see Lachaux et al., 1999). The phase-locking value separates the effect of the phase component from the amplitude component for a specific frequency band and represents a measure of neuronal integration and functional coupling (Lachaux et al., 1999;Varela et al., 2001). The PLVs were normalized using Fisher's z transform (Dyke et al., 2014;Gallicchio et al., 2017;Miskovic & Schmidt, 2010;Reiser et al., 2012).
Changes in functional coupling (TRPL; Rominger et al., 2019) were quantified for the upper alpha band (10-12 Hz) for each electrode pair by subtracting the median of Fisher's z transformed phase-locking values (PLV) of the reference periods from the activation periods of all trials. The same formular as used for calculating the TRP scores was applied.
Negative TRPL values express a decrease of functional coupling or a decrease of band power from the reference to the activation period, while positive values express an increase in these indices (Rominger et al., 2019;Rominger et al., 2020a,b). Changes in TRPL were calculated between all intra-hemispheric pairs of frontal-central electrodes (left: Previous research found these cortical sites and especially the increase of their functional coupling meaningfully related to creative cognition (Beaty et al., 2016;Jaušovec, 2000;Petsche, 1996;Pidgeon et al., 2016;Rominger et al., 2019;Zhou et al., 2018).
The utilized task-related design (subtracting phase-locking values of the reference period from the activation period) is especially suited to control for potential spurious/artificial synchrony in signals and volume conduction artifacts, as PLVs of activation and reference contain a similar amount of spurious/artificial synchronization (Bastos & Schoffelen, 2016;Palva et al., 2018;Rominger et al., 2020a,b).
Only activation periods with at least 250 ms and reference periods with at least 500 ms of artifact-free EEG recording were used. The length of the activation intervals ranged from a minimum of 1428 ms to a maximum of 12590 ms (M = 4490.17 ms, SD = 2201.96 ms). There were no gender differences in length of artifact-free activation intervals (t86 = -0.20, p =.838; women: M = 4531.28, SD = 2279.08; men: M = 4433.50, SD = 2120.77).

Statistical analyses
Preliminary analyses were run to test for gender differences in age, current positive and negative mood, as well as antagonistic personality (independent sample t-tests).
Independent sample t-tests were further calculated to test for gender differences in self-reported malevolent creativity (MCBS), performance measures of the MCT (total malevolent creativity, originality, malevolence, fluency), and different MCT aggression categories (physical, property, social, threat, tricks).
EEG analyses were conducted separately for TRP and TRPL scores. First, the gender specific effect of TRP in the upper alpha band during the MCT was investigated by a 2 × 8 × 2 ANOVA involving the betweensubjects factor GENDER and the within-subjects factors AREA (eight electrode positions in each hemisphere) and HEMISPHERE (left vs right). Second, to investigate whether women and men showed differences in neuronal integration of information during the MCT (TRPL), a 2 × 3 × 2 ANOVA with the between-subjects variable GENDER and the within-subjects factors COUPLING (frontal-parietal/occipital, frontaltemporal, and frontal-central) and HEMISPHERE was calculated.
In case of violations of the sphericity assumption, the multivariate approach was used (Vasey and Thayer, 1987). Significant interactions of interest were followed-up with selected independent t-tests as well as paired-sample t-tests. Estimates of effect sizes are given in terms of partial eta-squared measures (ηp 2 ) and Cohen's d (d). All statistical tests were performed with α = 0.05 (two-tailed).
Finally, we computed Pearson correlations between self-reported malevolent creativity on the MCBS, MCT performance scores and MCT aggression categories from participants' MCT answers. All statistical tests were performed with α = 0.05 (two-tailed).

Gender differences in malevolent creativity categories
Men incorporated a greater amount of physical aggression into their malevolent ideas compared to women, however, this effect was only observed as a statistical trend, t(86) = 1.72, p =.090, d = 0.37. We observed no significant gender differences for individuals' generation of ideas related to property aggression, t(86) = 0.41, p =.687, d = 0.09, threat-related aggression, t(86) = -1.32, p =.190, d = -0.29, or tricks, t (86) = -1.26, p =.210, d = -0.27. While women generated more malevolent ideas related to social aggression than men, results were not statistically significant, t(86) = -1.44, p =.153, d = -0.31. See Table 1 for descriptive statistics of MCT scores and categories by gender.

Discussion
The present study took a combined neuroscience approach to gender differences in aggression and creative ideation by examining brain activation and functional connectivity patterns while women and men engaged in malevolent creativity. Despite showing a similar malevolent creativity potential, female and male participants exhibited different patterns of alpha power changes (TRP) and functional connectivity changes (TRPL) during the generation of original ideas for harming others. By tapping into a darker context of creative thinking, our findings contribute further evidence to the idea that men and women achieve similar creative performance through diverging neurocognitive processes (Abraham, 2016;Abraham et al., 2014;   Note. M = mean value; SD = standard deviation, t = t-value, p = p-value, d = Cohen's d, n = 88. Looking at the overall pattern of TRP and TRPL changes, women's malevolent creative thinking was underlined by a steep decrease of alpha power from frontal (alpha synchronization) to left centraltemporal sites (alpha desynchronization). Importantly, this specific decrease was not found in men, who presented a topographically more diffuse, and undifferentiated pattern of TRP changes. However, compared to women, men showed lower functional coupling of frontalcentral sites.  As a general interpretation, the observed activation differences may suggest that in producing malevolent creative ideas, women more strongly rely on executive control, which matches our previous hypothesis. Additionally, we found that malevolent creativity in women seems to be linked to potentially effortful episodic and semantic memory retrieval. In this respect, it was reported that left temporal and parietal alpha desynchronization during creative ideation may indicate retrieval of task-relevant information from long-term memory (Agnoli et al., 2020;Klimesch et al., 2007;Hanslmayr et al., 2011;Schwab et al., 2014), which provides a crucial first step in expanding different semantic recombination possibilities to achieve originality of ideas. Agnoli et al. (2020) argued that posterior desynchronizationand more precisely, the absence of right parieto-temporal synchronizationcould signal ineffective inhibition of close semantic associations, only yielding ordinary, but not unique ideas (Camarda et al., 2018). However, this is clearly not the case in the present study, as women achieved similar originality of malevolent ideas to men. Much rather, the female pattern of steep left central-temporal desynchronization, along with right temporal synchronization (see Fig. 2) suggests that women's focused memory search in semantic network may be an effective neurocognitive strategy for recombining distant semantic elements into novel malevolent ideas (Agnoli et al., 2020;Camarda et al., 2018;Luft et al., 2018;Perchtold-Stefan et al., 2020). This interpretation is also fitting in light of the observation that women compared to men, tended to incorporate a greater degree of social aggression in their malevolent creativity (e.g., creative lies, vicious rumors, destruction of social hierarchies, online mobbing/bullying). While not reaching statistical significance, this female preference for social/indirect aggression not only matches aggression literature (Björkqvist, 2018;Card et al., 2008;Hess & Hagen, 2006), but may also require more intricate schemes and plotting, and accordingly, draw more strongly on representations of social networks and hierarchies. In brain imaging studies, neural representations of social hierarchies, including competition, intergroup exclusion, and retaliatory punishment scenarios, activated the temporo-parietal junction along with lateral and medial prefrontal cortices (Bellucci et al., 2020;Lasko et al., 2022;Zink et al., 2008). In sum, it appears plausible that our findings point to a female semantic route to malevolent creativity through increased top-down control over semantic memory search (Ovando-Tellez et al., 2022;Rominger et al., 2020a).
Compared to women, men displayed a much more diffuse pattern of TRP changes during performance of the MCT, along with lower functional coupling of frontal-central sites. These findings suggest that men's malevolent creative ideation may have more strongly depended on spontaneous and more automatic modes of thinking, which was less reliant on executive control processes. This interpretation aligns with findings by Rominger et al. (2020a) who reported lower frontal-central coupling when men mentally generated creative soccer moves to score a goal (also see Fink et al., 2018). The authors proposed that men needed less top-down control over motor imagery and motor planning processes when generating creative ideas for tasks that next to essential creativity requirements (prepotent response inhibition, memory retrieval, etc.) also integrate physical, visual, and spatial navigation components (e.g., anticipation of others' physical movements). A similar explanation may apply to our study as well, especially since men's malevolent creative ideas were rated as significantly more harmful than women's, and (at trend level) incorporated a greater amount of physical violence. This suggests that when it comes to ideas for creative revenge, men strongly rely on strategies centered around physical punishment (e.g., beatings, mutilation, physically harmful pranks, murder), which may require a greater amount of motor-related processing and mental visualization. However, given their natural preference for physical aggression (Archer, 2004;Card et al., 2008), and greater tendency for violent revenge fantasies (Goldner et al., 2019), this physical-activity related mental imagery may be less controlled, more automatic, and potentially, less effortful, which is reflected in decreased frontal-central coupling compared to women. Accordingly, it may be argued that men's creative thinking towards revenge purposes relies on more automatic imagination and less executive control processes, which may give rise to ideas that are not necessarily more unique but may have a more damaging impact.
Given the strong link of prefrontal alpha power increases with executive top-down control in creative thinking (Agnoli et al., 2020;Benedek et al., 2014a,b;Zabelina et al., 2019), our finding of lower (left) frontal alpha power increases in men compared to women fits the interpretation that men require less top-down inhibition to switch to more taboo, revenge-oriented mindsets, while women more strongly rely on conceptual expansion (Abraham et al., 2012). Interestingly, brain imaging studies also proposed that more activity in the left inferior frontal gyrus during creative ideation is linked to more conservative, rule-based behavior and thus leads to greater rejection of unconventional, unprecedented ideas (Kleinmintz et al., 2018;Silberstein & Camfield, 2021). Since creative physical aggression is arguably more taboo than social aggression, this may provide a further neurophysiological explanation for the different neuronal patterns and malevolent strategies applied by the women and men.
Taken together, our obtained neurophysiological activation patterns suggest different neurocognitive routes to malevolent creativity in women and men. When generating creative ideas for malevolent purposes, women seem to more strongly rely on executive control processes and semantic memory retrieval to re-combine remotely associated  social/relational information into novel revenge ideas, underlined by a TRP pattern of prefrontal alpha increases, along with left centraltemporal decreases. By comparison, men's malevolent creative thinking appears less controlled and more reliant on spontaneous imagery, as indicated by a more diffuse pattern of TRP changes and lower frontal-central coupling, which together, may facilitate the generation of physical revenge fantasies that are also rated as more harmful.

Limitations and future directions
By integrating measures of neuronal activation (TRP) and functional connectivity (TRPL), the present study offers comprehensive, novel information about different neurocognitive strategies involved in the generation of malevolent creativity. However, some limitations must be noted. Like all EEG studies, we cannot make definitive inferences regarding the exact source localization of our obtained TRP and TRPL patterns. Accordingly, all comparisons of this study's findings (e.g., left frontal alpha power) with findings from previous fMRI studies (e.g., left inferior frontal gyrus) only tentatively hint at shared brain mechanisms and functions, and should be taken as inspiration for future studies to conduct conjoint EEG/fMRI investigations to fully integrate spatial and temporal properties of malevolent creative ideation. Further, with increasing knowledge on the brain mechanisms of creative ideation, cross-frequency functional connectivity analyses may constitute another pioneering approach advancing the in-depth network exploration of malevolent creativity with EEG (Bose et al., 2018(Bose et al., , 2019. Next, some of our pair-wise comparisons (e.g., gender differences in physical aggression) failed to reach statistical significance. However, they may still provide meaningful patterns of behavioral and neurophysiological differences that should be followed-up by studies using larger, more diverse samples that also control for aspects like menstrual cycle (Galasinska & Szymkow, 2022) and a broader spectrum of antisocial traits (Kapoor and Kaufman, 2022;Perchtold-Stefan et al., 2022b). Despite genderbalanced recruiting efforts, due to auditory device malfunctions, women and men were not evenly matched in the final study sample. While t-tests and ANOVAs are considered to be fairly robust against moderate departures from the equal variance assumption, oversampling of men may be recommended for future studies investigating gender differences in malevolent creativity. Here, age and IQ may be further influencing factors. Yet, the relationship between IQ and creativity is complex (e.g., Plucker et al., 2020), and IQ differences may only critically effect gender-based analyses of malevolent creativity if data is merged from samples with vastly different backgrounds (e.g., students vs general population vs clinical patients vs prison populations). Further, there are no major gender differences in verbal fluency or verbal IQ (see Sokołowski et al., 2020;Weiss et al., 2006), which are most likely to influence creative idea generation in psychometric tests with time restrictions. Finally, it is plausible that in daily life, different social situations elicit different motivation for malevolent creativity in men and women. While the MCT has been sufficiently validating in provoking malevolent creativity in younger student samples (Perchtold-Stefan et al. (2022c)), a more personalized approach (having individuals self-select social scenarios that would most likely make them malevolently creative) could be eminently helpful in disentangling subtle gender differences in real-life contexts of malevolent creativity.

Conclusion
Adding a novel perspective to previously reported gender differences in malevolent creativity, the present study demonstrated that women and men are similarly capable of malevolent creative ideation, but seem to utilize different neurocognitive processes in their efforts to exact creative revenge on others: While men rely on less controlled motor imagination that may initialize ideas geared towards physical punishment, women more strongly engage executive and semantic processes that may yield more intricate social aggression. In light of the reciprocal linkup of neurophysiology and behavior, further research on gender differences in malevolent creativity may have crucial practical implications for the detection and prevention of the infinite variety of malevolent creative acts in real-life.

Funding
This research received no funding.

Ethics approval
Written informed consent was obtained from all participants in this study. The study was reviewed and approved by the ethics committee of the University of Graz, Austria (GZ. 39/73/63 ex 2020/21).

Consent to participate
All participants gave written informed consent to participate in this study.

Consent for publication
Not applicable.

Availability of Data and Materials
Data and materials are available from the corresponding author under the condition of a formal data sharing agreement and the submission of a formal project outline.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.
request to the corresponding author under the condition of a formal data sharing agreement and the submission of a formal project outline. The experiment was not preregistered.