moderates the relationship between adaptive narcissism and performance under pressure

Narcissism-performance research has focused on grandiose narcissism but has not 22 examined the interaction between its so-called adaptive (reflecting over-confidence) and 23 maladaptive (reflecting a domineering orientation) components. In this research, we tested 24 interactions between adaptive and maladaptive narcissism using two motor tasks (basketball 25 and golf in Experiments 1-2, respectively) and a cognitive task (letter transformation; 26 Experiment 3). Across all experiments, adaptive narcissism predicted performance under 27 pressure only when maladaptive narcissism was high. In the presence of maladaptive 28 narcissism, adaptive narcissism also predicted decreased pre-putt time in Experiment 2 and 29 an adaptive psychophysiological response in Experiment 3, reflecting better processing 30 efficiency. Findings suggest that individuals high in both aspects of narcissism perform better 31 under pressure thanks to superior task processing. In performance contexts, the terms 32 “adaptive” and “maladaptive” – adopted from social psychology – are over-simplistic and 33 inaccurate. We believe that self-inflated narcissism and dominant narcissism are better 34 monikers for these constructs. 35


Introduction 38
Performing to a high standard is important in sport and in many facets of life. One's 39 desire to perform well under high pressure typically evokes performance anxiety that often 40 harms performance (Woodman & Hardy, 2001). Conversely, while performance pressure may 41 be detrimental to those who are worried about the uncertainty of success (Eysenck,42 Derakshan, Santos, & Calvo, 2007), it may be beneficial for individuals who seek glory and 43 pursue admiration from performance success. In the context of performing under pressure, 44 one relevant personality trait is narcissism, especially in its grandiose form (see Roberts, 45 Woodman, & Sedikides, 2018). 46 Here we conceptualize narcissism as a non-clinical personality trait that can be 47 assessed on a continuous scale. We adopt the definition of narcissism as a self-centered, self-48 aggrandizing, entitled, dominant, and manipulative interpersonal orientation (Morf & 49 Rhodewalt, 2001). Such a conceptualization focuses on grandiose narcissism from an agentic 50 perspective and does not include communal narcissism (Gebauer,Sedikides,Verplanken,& 51 Maio, 2012). Further, our conceptualization of grandiose narcissism does not consider 52 vulnerable aspects of narcissism (e.g., Miller et al., 2011). From this point forward, when we 53 use the term narcissism we refer to grandiose narcissism. 54

Narcissism and performance: An overview 55
Individuals high in narcissism are thought to have the ability to perform well because 56 they possess attributes that are essential for performance success, such as confidence 57 (Campbell, Goodie, & Foster, 2004), optimistic expectations 58 1998), and a strong desire for dominance (Morf & Rhodewalt, 2001). Indeed, narcissists 59 believe they are superior to others and consider themselves as exceptional performers 60 (Gabriel, Critelli, & Ee, 1994). This grandiose belief is unfounded, however, as evidenced by 61 research revealing no effect of narcissism on performance. For example, although narcissists maladaptive narcissism). Equally, as there are no widely accepted alternative terms, we have 114 retained the use of adaptive and maladaptive narcissism in this research 1 . In the next section, 115 we focus more on the psychological attributes of these different components of narcissism 116 rather than their presupposed outcomes. We then propose our theoretical position regarding 117 how these components of narcissism may influence performance under pressure. 118

Adaptive and maladaptive narcissism and performance under pressure 119
Despite a plethora of work in the social domain, researchers have yet to consider the 120 adaptive/maladaptive narcissism distinction in the context of performance. Equally, although 121 both adaptive and maladaptive narcissism are relevant to performance (Roberts,Woodman,et 122 al., 2018), these components may not necessarily predict performance under pressure. 123 Typically, adaptive narcissism reflects high levels of confidence (Emmons, 1984), and 124 confidence is commonly linked to better performance under pressure (Woodman & Hardy, 125 2001). Conversely, excess confidence can be detrimental to performance, as individuals may 126 be overly assured of their potential and thus fail to allocate appropriate resources to facilitate 127 performance (e.g., Beattie, Dempsey, Roberts, Woodman, & Cooke, 2017). As such, adaptive 128 narcissism on its own is unlikely simply to lead to optimal performance. 129 Similarly, maladaptive narcissism, which reflects a strong sense of personal control 130 and a willingness to dominate (e.g., Washburn et al., 2004), may not yield clear performance 131 effects. Indeed, although maladaptive narcissism is linked to internalizing symptoms (e.g., Rather than exploring in parallel the performance effects of adaptive and maladaptive 140 narcissism, we propose a more nuanced position; that the influence of adaptive narcissism on 141 performance under pressure depends on the relative degree of maladaptive narcissism. Given 142 that overconfidence can be detrimental to performance (e.g., Beattie et al., 2017), performers 143 who hold an inflated self-view (i.e., high in adaptive narcissism) may only perform well 144 when they also have the willingness to dominate (i.e., high in maladaptive narcissism). As 145 such, we hypothesized that adaptive narcissism, reflecting (over)confidence, would not 146 predict performance under pressure when maladaptive narcissism was low. However, when 147 maladaptive narcissism is high, reflecting a strong willingness to dominate and have control 148 over situations, we expected adaptive narcissism to predict performance because of the 149 precise combination of confidence and willingness to dominate. We tested such an 150 overarching hypothesis across three different experimental settings. 151

Mechanisms underlying narcissism and performance 152
Beyond examining the hypothesized interaction between adaptive and maladaptive 153 narcissism on performance under pressure (Experiments 1-3), we also explored the 154 mechanisms that might underlie this performance effect (Experiments 2-3). Recent research 155 offers two accounts for why narcissists perform better in some situations than in others (see 156 ; one where narcissists improve performance as a result of 157 investing greater effort for self-enhancement (hereafter trying harder), and one where 158 narcissists improve as a result of a more efficient allocation of resources (hereafter trying 159 smarter). The trying harder position rests on a prediction of Processing Efficiency Theory 160 (Eysenck & Calvo, 1992); that performers can maintain or even improve performance under 161 pressure if they invest substantial amounts of effort (at a cost to processing efficiency). Such a position, that effort can aid performance under pressure, has received considerable empirical 163 support in the sport domain (e.g., Wilson, 2008). The trying smarter position is based on 164 tenets of Attentional Control Theory (Eysenck et al., 2007); that performers can maintain or 165 improve their performance under pressure via excellent regulation of processing resources 166 within the capacity-limited working memory system (improved processing efficiency; see 167 Wilson, 2008 for an overview of research investigating the effects of Attentional Control 168 Theory in the context of sport). 169 Embracing the trying harder hypothesis, Wallace and Baumeister (2002) argued that a 170 greater opportunity for glory drives narcissists to invest extra effort to perform. Providing 171 evidence for this position, in a dart throwing task and a muscular endurance task, Roberts, 172 Cooke, et al. (2018) found that effort invested on the task mediated the influence of 173 narcissism on performance. The finding indicates that narcissists perform better when there is 174 a self-enhancement opportunity (e.g., in a competition) because they try harder. 175 While the trying harder position has received some attention, the trying smarter 176 position has yet to receive empirical support. Nonetheless, the trying smarter position is 177 promising in explaining why narcissists perform better especially under high performance 178 pressure. Eysenck et al. (2007) suggest that performance pressure impairs the goal-directed 179 system and overly activates the stimulus-driven system, which disrupts task processing via 180 shifting attention to task-irrelevant thoughts (e.g., worry) and impairs performance. However, 181 narcissists' greater focus on success as opposed to failure make them more likely to remain 182 goal-driven and less likely to be overwhelmed by task irrelevant thoughts (Elliot & 183 Covington, 2001). Such an achievement orientation would ensure superior attentional control, 184 enabling narcissists to perform well under pressure. 185 Although promising, these conceptualizations of the trying harder and the trying 186 smarter positions are too simplistic as they fail to consider the potential adaptive × that whether narcissists exert increased effort to perform under high pressure or not depends 189 on the combination of adaptive and maladaptive narcissism. More specifically, the overly 190 inflated self (associated with adaptive narcissism), in the absence of maladaptive narcissism, 191 is unlikely to yield greater effort (cf. Woodman et al., 2011). Instead, high levels of 192 maladaptive narcissism may drive the inflated self to strive for desirable states because of the 193 willingness to dominate. Consequently, based on the trying harder position, adaptive 194 narcissism will predict effort during task processing when maladaptive narcissism is high. 195 Equally, while narcissists may have the potential to achieve superior attentional 196 control under pressure, adaptive narcissism in the absence of maladaptive narcissism may 197 prevent this potential being realized. This is because narcissistic individuals believe their 198 attentional control is already excellent. As maladaptive narcissism provides a strong desire to 199 dominate, however, the link between adaptive narcissism and attentional control will likely 200 strengthen. As such, the trying smarter position suggests that adaptive narcissism will predict 201 better efficiency during task processing when maladaptive narcissism is high. 202

Present research 203
In sum, our theoretical stance suggests that maladaptive narcissism will moderate the 204 relationship between adaptive narcissism and performance under pressure, and increases in 205 effort and/or more effective task processing will help to explain such performance benefits. 206 We tested these predictions across three laboratory experiments. In Experiment 1, we used a 207 basketball free throw task to test the interaction between adaptive and maladaptive narcissism 208 on performance under pressure. In Experiment 2, we used a golf-putting task to examine the 209 replicability of the Experiment 1 results and employed self-report and behavioral measures to 210 test both the trying harder and the trying smarter positions. In Experiment 3, we used a letter 211 transformation task to test the generalizability of the results from the first two experiments.

Task 227
We used a basketball free throw task. Participants completed the free throw task (see 228 Experimental conditions section) using a regulation basketball (24.60cm in diameter) from 229 the free throw line, 4.33m from the basket (45.00cm in diameter) at a regulation height of 230 3.05m. We assessed performance using Hardy and Parfitt's (1991) point system designed for 231 this task. Participants scored "5" for a "clean" basket shot, "4" for rim and in, "3" for 232 backboard and in, "2" for rim and out, "1" for backboard and out, and "0" for a complete 233 miss. We summed participants' scores. 234

Design 235
We used a within-group design to reduce sampling error and to allow a better 236 understanding of how performers respond to high-pressure environments. Participants completed the same experimental procedures in groups of ten. All participants completed 238 experimental tasks under two conditions: low pressure (i.e., individual session) and high 239 pressure (i.e., competition in front of audience, opportunity for monetary reward, public 240 recognition). The individual session took place seven days before the competition. 241

Experimental conditions 242
Low-pressure condition. This condition consisted of twenty non-recorded warm-up 243 throws and five recorded testing throws (Hardy & Parfitt, 1991). Each participant attended an 244 individual session in an indoor sports hall. We introduced the scoring system and instructed 245 participants to perform at their normal pace. 246 High-pressure condition. This condition consisted of twenty non-recorded warm-up 247 free throws followed by five recorded free throws performed in front of an audience as part 248 of a competition. We informed participants that the top three performers would receive cash 249 prizes of £30, £20, and £10, and that we would place a congratulatory poster on the sports 250 hall news wall, highlighting the winning participants. We also asked participants to watch 251 other participants when they were not performing the task. We asked our 'audience' 252 participants to stay in a pre-set audience zone that surrounded the free throw area and 253 provided them with whistles and inflatable sticks to make similar noises to those during 254 basketball matches. Before starting the free throws, we asked participants to perform the free 255 throws at their normal pace. 256

Measures 257
Narcissism. We assessed narcissism using the Narcissistic Personality Inventory-16 258 Campbell, 2012). The NPI-16 manifests identical nomological networks to the most widely 261 used measure of narcissism (i.e., NPI-40; Raskin & Hall, 1979), especially in relation to (Ames et al., 2006). It also demonstrates good test-retest reliability (r = .85). Given its 264 reliability and convenience, the NPI-16 has been well used in sport narcissism research (e.g., 265 Beattie et al., 2017). The NPI-16 contains sixteen forced-choice items from the NPI-40 and 266 asks participants to choose between one narcissistic and one non-narcissistic statement (e.g., 267 "I will be a success" vs "I am not too concerned about success"). Following Barry  items (e.g., "I am concerned that I may not do as well in this competition as I could") rated 273 from 1 (not at all) to 4 (very much so). Cronbach's alpha was .90 in the current experiment. 274

Procedure 275
With institutional ethical approval, we recruited participants from a university 276 basketball club. With the agreement from the club manager, we provided study information 277 sheets to club members in a briefing session after a weekly club meeting. After the briefing 278 session, club members who decided to participate provided consent, signed up for their 279 sessions, and completed the NPI-16. On the day of the individual session, participants 280 completed the CSAI-2R before starting their free throws. On completion of the throws, we 281 thanked participants and reminded them of the group competition a week later. On the 282 competition day, following the instructions (see High-pressure condition section) participants 283 drew lots to decide the order of performance. They completed the CSAI-2R immediately 284 before their individual performance. After the competition, we thanked and debriefed 285 participants, and awarded prize money to winners. 286

Main analyses 295
To create a performance variable for analysis, we regressed the high-pressure 296 performance on the low-pressure performance, with higher residual scores reflecting better 297 performance under pressure. This residualized approach (see Castro-Schilo & Grimm, 2018) 298 allowed us to account for participants' performance capacity in low-pressure situations when 299 considering their performance under pressure. Hereafter, we use the term performance to 300 denote residualized performance. 301 To test our hypothesis that adaptive and maladaptive narcissism would interactively 302 predict performance, we performed moderated hierarchical regression with 5,000 bootstraps 303 and reported unstandardized regression coefficients and the ΔR 2 for each step of the 304 hierarchical regression. Lower and upper bound 95% confidence intervals (CI) that do not 305 encompass zero indicate significance at .05 for all effects. We probed significant interactions 306 using z-score transformation before the moderated regression analyses. Such an approach 315 helps mitigate the potential collinearity issue in moderation analyses (Hayes, 2013) and is 316 useful to check for univariate extreme values (i.e., three standard deviations from the mean). 317 Further, we used Cook's distance (Cook & Weisberg, 1982) and leverage (Stevens, 2002) to 318 screen multivariate outliers. We used the recommended cut-off value of greater than 1 Cook's 319 distance and larger than 3*(k+1)/n leverage (where k is the number of predicators in the 320 model and n is the sample size) as the criterion for multivariate outliers. We found no case 321 with undue influence. Further, we calculated Cohen's f 2 (Cohen, 1977) as an effect size index 322 for the interaction, with .02, .15, .35 reflecting small, medium, and large effects, respectively. 323 The regression models satisfied the normality and homoscedasticity assumptions.  We chose medium-handicap golfers because they are particularly sensitive to pressure 344 manipulations (Mullen & Hardy, 2000). All participants reported that they had played 345 competitions on a weekly basis over the previous 12 months (unless weather or illness/injury 346 prevented participation). 347

Task and Apparatus 348
Participants performed a putting task on a 4.5 × 1.6-meter indoor putting green. We 349 provided a standard (90cm) steel-shafted blade style putter and competition white golf balls 350 (4.27cm diameter). We used a half-size target hole (5.5cm diameter) to increase the accuracy 351 demands. We disguised a digital camera in a box at the end of the putting green, facing 352 directly toward participants. The camera had a 10mm diameter lens and a shutter speed of 353 1/2000 second. We used the digital camera to measure pre-putt time and introduced the 354 camera to participants as an additional source of pressure (see High-pressure condition). 355

Performance 356
We used an automated measuring system for putting performance, which we 357 conceptualized as the distance between the center of the golf ball and the center of the hole. 358 We took the mean distance of the balls from the target hole (in mm) to generate the mean 359 radial error (MRE), with lower MRE representing higher accuracy. We recorded each 360 successful holed putt as 0mm. 361 Design participant attended an individual session to complete all experimental conditions. 364

Experimental conditions 365
Practice. This condition consisted of five blocks of nine putts (i.e., 45 putts in total) to 366 familiarize participants with the task. Participants received the standardized instruction that 367 the objective of the experiment was to examine the effect of using different putting positions 368 in golf putting skills training and that they had been randomly assigned to the group that 369 would follow a specific putting sequence. In reality, all participants followed the same 370 randomized sequence of the three starting points within each putting block -1.6, 2.2, 2.8, 2.8, 371 2.2, 1.6, 1.6, 2.2, and 2.8m from the target. The purpose of this training-related instruction 372 was to blind participants from the real objectives of this experiment and to help achieve 373 experimental manipulation. Before each putting block, we instructed participants to "relax 374 and take your time to perform the putt as you want; try to acclimatize yourself with the task 375 and get the ball ideally holed or make it as close to the hole as possible." 376 Low-pressure condition. This condition consisted of a single block of nine putts, with 377 the same putting sequence as in practice. To minimize pressure, we reminded our participants 378 of the experimental purpose we provided at practice. Prior to putting, we asked participants to 379 "relax and take your time to perform the putt as you want; try to get the ball ideally holed or 380 make it as close to the hole as possible". 381 High-pressure condition. This condition consisted of a final block of nine putts, using 382 a putting sequence different from the previous blocks. To start, we informed participants that 383 based on their putting performance in previous blocks they were to receive prize money of 384 £5. However, to secure the £5, participants needed to achieve a "reasonable level of 385 performance", which in reality was participants' MRE in the low-pressure condition minus a 386 half standard deviation. We informed participants that they would lose the £5 if they failed to extra prize money if they achieved a "superior" performance standard, which in reality was 389 their respective MRE in the low-pressure condition minus one standard deviation. 390 Furthermore, we informed participants that they would compete against each other in 391 the final block. We asked participants to draw one of twelve task cards from an envelope we 392 prepared. We explained that different task cards provided different levels of task difficulty. 393 For example, repeating nine putts from the same starting point represents an easy task; 394 completing three mini-blocks of three putts whilst repeating the same starting point in each 395 mini-block represents a medium-level task; putting from a randomized sequence of the three 396 different starting points represents a difficult task. We reminded participants that regardless of 397 the level of difficulty, the participant who improved most from the previous block to the final 398 block would win £50 and be recognized in congratulatory posters posted on the news boards 399 in the golf club of which they were members. Additionally, we informed participants that we 400 would release the top-ten and the bottom-ten rankings to all participants through emails based 401 on their performance change from the previous block to the final block. 402 Despite instructing participants that different task cards provided different putting 403 sequences, in reality, everyone completed the same task order: 2.2, 1.6, 2.8, 2.8, 2.2, 1.6, 2.2, 404 2.8, and 1.6m. After drawing the task card, we checked a pre-printed document in front of 405 participants to provide a fake historical record revealing the likelihood of obtaining a prize. 406 We told participants that about 50% of people had secured £5 and about 10% of people had 407 earned the £15 extra prize, but that nobody had gained any prize when putting the same 408 sequence as them. 409 Finally, we made participants aware of the video camera we had disguised. We 410 informed participants that the recorded video materials would be assessed by an external 411 expert, and selected records would be edited and used for promotional and educational purposes. We further reminded participants that they were free to withdraw from completing 413 the final block if they were unhappy with anything. After participants confirmed their 414 willingness to participate, we instructed them to "take your time, concentrate on the task in 415 hand, try to get the ball ideally holed or as close as possible to the target to win a prize."  Mental effort. We used the Rating Scale for Mental Effort (RSME, Zijlstra, 1993) to 431 examine the trying harder position. The RSME is a vertical axis scale that asks participants to 432 rate their mental effort from 0 to 150, with increments of 10 displayed on the left side of the 433 scale and nine descriptive indicators from 3 (no mental effort at all) to 114 (extreme mental 434 effort). The RSME is an effective measure of mental effort during the performance of various 435 tasks, with a test-retest reliability of .78-.88 (Zijlstra, 1993). efficiency, in order to examine the trying smarter position. This approach was recommended 438 by Eysenck and Calvo (1992) and has been adopted in performance-related research (see 439 Zhang et al., 2018). Although longer pre-putt time was previously interpreted as greater 440 effort, the relationship between pre-putt time and effort is not evidenced (Wilson et al., 2007). counted video frames (50Hz field rate) from the moment that participants prepared for the 452 putting posture to the moment that participants initiated a "real" putt with the putter touching 453 the golf ball. We transformed these video frames into pre-putt-time (in seconds). 454

Procedure 455
The experiment took place in a golf-putting laboratory. With institutional ethical 456 approval, we advertised the study in local golf clubs and recruited club members given their 457 informed consent. After welcoming participants to the laboratory, we asked participants to 458 provide consent and to complete the NPI-40. Next, participants completed the experimental 459 conditions of five blocks of practice, one block of low-pressure putts, and one final block of 460 high-pressure putts. We asked participants to complete the MRF-L after our manipulations in 461 the low-and high-pressure conditions and the RSME on completion of each condition. At the

Preliminary analyses 466
There were no missing data. A paired t test revealed a significant increase in cognitive 467  Table 2 provides descriptive 469 statistics and correlations between study variables. 470

Main Analyses 471
As with Experiment 1, we generated the residualized scores for all of our outcome 472 variables including performance (MRE), mental effort, and pre-putting time (hereafter we use 473 the variable name to refer to the residualized scores, e.g., "performance" refers to 474 residualized performance). We performed moderated regression analyses as in Experiment 1. 475 There were no univariate or multivariate outliers. All assumptions for regression were met.  Experiments 1 and 2 consistently demonstrated that increased adaptive narcissism was 509 related to better performance under pressure only when maladaptive narcissism was high. 510 The data from Experiment 2 did not support the trying harder hypothesis because adaptive 511 narcissism failed to predict effort regardless of the levels of maladaptive narcissism. Results improved efficiency and performance only when maladaptive narcissism was high. 514 In Experiment 3, we employed a letter transformation task to examine the 515 generalizability of findings from Experiments 1 and 2. This task requires participants to 516 transform a random letter a given distance to obtain another letter under low-and high-517 pressure conditions. For example, the instruction 'A + 4' requires participants to transform 518 the letter A to E. This process directly tests the functions of working memory (Hamilton et  519   al., 1977), which is known to play a vital role in motor execution and performance under 520 pressure (see Furley & Memmert, 2010). Another advantage of this task is that it permits 521 recording of psychophysiological indices of processing efficiency such as heart rate 522 variability. More specifically, r-MSSD (a time domain measure of heart rate variability) 523 provides an index of cardiac vagal control (Achten & Jeukendrup, 2003), which is positively 524 associated with affective regulation, attentional control, and goal-directed executive function 525 (Thayer & Brosschot, 2005). We therefore employed r-MSSD as a measure of processing 526 efficiency in Experiment 3. 527 In the interests of parsimony, we report much of Experiment 3 (i.e., method, analyses, 528 tables) in the online supplement. We encourage readers who are interested in this innovative 529 pressure manipulation (via a computerized testing program) to scrutinize those materials. We 530 report the results below to evidence the replicability of the performance effect and to provide 531 additional support for the underlying mechanism using psychophysiological data. 532
Step 1 of the analysis revealed that adaptive 536 narcissism was significantly related to better performance (reduced time taken), ΔR 2 = .05, F Consistent with Experiments 1 and 2, adaptive narcissism was only associated with 566 improved performance under pressure when maladaptive narcissism was high. In accord with 567 Experiment 2, the effort data did not support the trying harder hypothesis. The r-MSSD data 568 from the letter transformation task provide further support for the trying smarter hypothesis, 569 as adaptive narcissism protected processing efficiency and predicted improved performance 570 only when maladaptive narcissism was high. 571

General discussion 572
Although global grandiose narcissism as measured by the NPI has been the main 573 focus of the narcissism-performance research, the performance effects of its so-called 574 adaptive and maladaptive components had previously been unexplored. In the present 575 research we examined the adaptive × maladaptive narcissism interaction on performance 576 under pressure and tested potential mechanisms to explain these performance effects. 577 Across two motor tasks and one cognitive task, we provide the first evidence that 578 adaptive narcissism is beneficial to performance under pressure only in the presence of 579 maladaptive narcissism. The findings demonstrate that a one-dimensional conceptualization 580 of grandiose narcissism is inadequate to explain the effects of narcissism on performance. We 581 also investigated the mechanisms underlying these findings and provide the first support for 582 the trying smarter proposition over the trying harder viewpoint (see Roberts, Woodman, et 583 al., 2018). In the golf-putting and letter transformation tasks (Experiments 2 and 3), results 584 consistently demonstrated that adaptive narcissism was unrelated to effort regardless of the 585 levels of maladaptive narcissism. Conversely, adaptive narcissism predicted better efficiency 586 and performance only when maladaptive narcissism was high. These findings suggested that

Trying harder vs Trying smarter 590
While evidence for the trying harder hypothesis has emerged in the existing 591 narcissism-performance research (e.g., Roberts, Cooke, et al., 2018) throwing) and imposed high levels of physical demand (i.e., the muscular endurance task). 603 However, our tasks involved participants with higher levels of task-related expertise (i.e., 604 basketball players and skilled golfers) and imposed mental (i.e., letter transformation) rather 605 than physical demand. Indeed, skilled performance requires less mental control (Masters & 606 Maxwell, 2008), and cognitive compared to muscular endurance tasks are less physically 607 demanding. Therefore, effort quantity plays a less critical role in our tasks compared to 608 performance climate to manipulate experimental conditions, but a performance climate does 611 not necessarily create high pressure. Conversely, our tasks combined a range of stimuli to induce pressure during task performance. According to distraction theories of anxiety and 613 performance (Eysenck & Calvo, 1992;Eysenck et al., 2007), additional effort is less likely to 614 compensate for performance as performance pressure increases. As such, it is possible that 615 trying harder could help achieve desired performance under relatively low levels of pressure 616 and that trying smarter could optimize performance when pressure is higher. Such a position 617 is worthy of consideration. 618

Theoretical and applied implications 619
The findings have several important implications. First, in performance contexts, it 620 appears that maladaptive narcissism is sometimes adaptive because it can contribute to better 621 performance under pressure. Given that adaptive narcissism was beneficial to performance 622 under pressure only in the presence of maladaptive narcissism, the so-called adaptive and 623 maladaptive monikers of the corresponding components in the NPI are misleading. We 624 recommend the use of different terms to describe these aspects of narcissism and suggest 625 The finding that maladaptive narcissism plays an adaptive role in performance Indeed, performance environments operate within an intrapersonal and interpersonal context 639 such that one would explore the potential benefits of maladaptive narcissism to best effect 640 beyond the performance setting in isolation. For example, if narcissists behave aggressively 641 and violently in a social environment because they do not recognize any alternative ways to 642 eliminate any ego-threats and re-establish dominance (Baumeister et al., 1996), creating 643 performance environments and fostering performance goals are likely to be particularly 644 beneficial for those high in maladaptive narcissism. Although such a position requires 645 empirical support, it provides an alternative route for alleviating the potential adverse 646 influences of narcissism in social and interpersonal settings. 647 Additionally, the present data offer an insight into the mechanism that underlies 648 optimization of narcissists' performance under pressure. Specifically, individuals high in both 649 adaptive and maladaptive narcissism performed better under pressure thanks to their superior 650 regulation of task processing rather than simply by investing greater effort during task 651 performance. As such, we recommend that performance-focused practitioners consider 652 interventions to enhance performers' regulation of task processing. Furthermore, considering 653 the adaptive × maladaptive narcissism interaction on performance under pressure, it appears 654 that high levels of confidence and performance motivation are equally important for 655 achieving optimal performance. 656

Limitations 657
Although the findings are clear and offer important implications, we note several 658 limitations that warrant attention. First, although our sample estimations aimed to provide 659 sufficient power for detecting performance effects, they may have been imprecise for 660 examining the underlying mechanisms of the performance effects. Indeed, some of our 661 analyses, especially the examination of the trying harder hypothesis in Experiments 2 and 3 were subject to low statistical power. This is because the effect sizes in mental effort was 663 smaller than our a priori estimations. However, the analyses on efficiency (i.e., pre-putt time, 664 r-MSSD) achieved sufficient power and demonstrated larger effect sizes. As such, the trying 665 smarter perspective likely plays a more vital role in performance under pressure over the 666 trying harder perspective for those high in both adaptive and maladaptive narcissism, at least 667 in tasks that require fine motor control (e.g., golf-putting) and working memory (e.g., letter 668 transformation). Second, the cognitive task used in Experiment 3 might invite concern about 669 the generalizability of the findings to sport contexts. However, such a concern is less of an 670 issue because we used a letter transformation task that relies on the functions of working 671 memory, which play a vital role in sport performance (see Furley & Memmert, 2010). As 672 such, Experiment 3 findings have relevant performance implications for sport settings. under high pressure tend to reinvest attention to task processing through the use of explicit 680 task-relevant knowledge (e.g., Mullen & Hardy, 2000) or step-by-step monitoring (e.g., 681 Beilock & Carr, 2001) to avoid undesired performance. However, such reinvestment will 682 regress effortless skilled performance to a de-automatized and more effortful form of control 683 which results in performance failure (Masters & Maxwell, 2008). From a reinvestment 684 perspective, since individuals high in narcissism are confident in their ability and seek to 685 approach rather than to avoid performance settings (Zhang et al., 2018), they likely see 686 themselves as so capable as to have no need for reinvestment to ensure good performance. when performing in high-pressure environments. Our data support this position, especially 689 that adaptive narcissism in the presence of maladaptive narcissism predicted reduced pre-690 putting time in golf-putting and less of a decrease in r-MSSD in letter transformation, which 691 indicates automated task execution and lower levels of interference (see also Lam et al., 692 2010). This position clearly warrants further research attention. 693

Conclusions 694
The current research demonstrated that adaptive narcissism (reflecting assurance and 695 over confidence) was related to better performance under pressure only when maladaptive 696 narcissism (reflecting a strong willingness to dominate) was high. In the specific context of 697 high-pressure performance, there is thus nothing maladaptive about maladaptive narcissism -698 quite the contrary. We thus urge researchers to abandon the use of adaptive and maladaptive 699 narcissism in favor of self-inflated and dominant narcissism, respectively. The findings 700 further support that the precise combination of adaptive and maladaptive narcissism 701 contributes to the efficient use of processing resources such that individuals high in both 702 components of narcissism perform well under pressure because they try smarter rather than 703 try harder. Future research would do well to examine different forms of narcissism in 704 performance settings, and beyond. 705

1.
Based on the data reported in this paper, we suggest in the General Discussion that 862 adaptive narcissism would be better labeled self-inflated narcissism and that 863 maladaptive narcissism would be better labeled dominant narcissism. We believe 864 these alternative monikers better describe the psychological attributes of the so-called 865 adaptive and maladaptive components of narcissism, at least in the contexts of sport 866 and performance. 867

2.
This research is the first to examine the effect of adaptive and maladaptive narcissism 868 interaction and thus no previous studies provide possible effect size of such an 869 interaction. However, as we were interested in examining the effects of these aspects 870 of narcissism on performance under pressure, we used the effect sizes for the 871 previously reported interaction between narcissism and pressure on performance for 872 the power analysis. 873 Table 1 Descriptive statistics and correlations between study variables in the basketball set shot (n = 80) Note. Experience = Years of Experience; NPI-16 = 16-item Narcissistic Personality Inventory    Radial Errors (in millimeter). * p < .05; ** p < .01 Figure 1. The interaction between adaptive and maladaptive narcissism on performance scores in basketball free throw (top) and mean radial errors (middle) and pre-putt time (bottom) in golf-putting. Regression slopes were derived from one standard deviation below the mean (low) and one standard deviation above the mean (high). All variables were standardized. Figure 2. The interaction between adaptive and maladaptive narcissism on time taken in the letter transformation (top) and the r-MSSD during the letter transformation (bottom). Regression slopes were derived from one standard deviation below the mean (low) and one standard deviation above the mean (high). All variables were standardized.