We recruited 40 skilled tennis players (M = 20.0 yrs, SD = 1.2) from a university tennis team. All participants had more than 3 years of tennis experience, with the majority having competed in intercollegiate tournaments. Participants were randomly assigned to either the contraction group (11 males and nine females) or the no-contraction group (12 males and eight females). Each group comprised 17 intercollegiate-level and 3 regional-level players. They were right-handed, as assessed by the Edinburgh Handedness Inventory (M = +92.7, SD = 16.9; Oldfield, 1971). With a sample size powered at 0.80 to detect significance at an alpha level of 0.05, G*Power (Faul et al., 2007) determined a required sample size of 34 for a repeated measures analysis of variance (f = 0.25), corresponding to a small to medium effect size (Cohen, 1992). This study was approved from the local ethics committee (approval number: 2020–077).

The participants were instructed to serve a ball from a predetermined position (i.e., 1 m to the right of the center mark), targeting the wide corner of the deuce service court. We quantified serving scores based on the ball landing position, as classified by the lines drawn on the court (Figure 1). We established scoring zones across the entire service court. Following assessment by a skilled tennis player, we adjusted the scoring zones to align with difficulty levels, as proficient serves typically target the near side-line in the wide corner. The participants were told to “perform second serves as if they were in a real match, aiming for maximum accuracy to achieve the highest possible score.” We presumed that the second serve would be executed under pressure because its failure would result in losing a point in real tennis competitions as a double fault (Weinberg, 2005). The tasks were performed using their own rackets. Faults received a score of 0 points. If a let occurred, an additional serve was provided, adhering to official tennis rules. Serving motions were recorded using a high-speed camera positioned 9 m away that captured the right sagittal plane. The courts were recorded at a height of 3 m.

The participants attended a 30-min session. After receiving instructions for the experiment, they completed the serving task, which comprised 25 servers. This consisted of 5 serves in the practice session, 10 on the pre-test, and 10 on the pressure test. Before each test, we assessed state anxiety (Figure 2).

On the pre-test, participants in both groups served balls without hand contractions. On the pressure test, participants in the contraction group served a ball immediately after contracting their left hand by grasping a material called “expanded polyethylene foam” (0.1 m × 0.1 m × 0.01 m, density 30 kg/m³, compressive stress 50 kPa, P0030, manufactured by Fuji Gomu Co., Japan, Figure 3). This material was embedded in a flat pocket that could be sewn into the sportswear. They repeatedly grasped the material with their left hand for 20 s immediately before serving. Participants in the no-contraction group did not grasp the material during the same period (20 s). We set the grasping pace at 60 beats per minute (bpm) and recorded the sounds using a metronome.

Before the pressure test, the participants were told the following, which was effective for inducing state anxiety as in previous studies (Cooke et al., 2014; Lo et al., 2019): (a) the serving motion would be recorded with two cameras; (b) performance outcomes would be evaluated by the head coach of the tennis club; (c) performance results would be disclosed to all tennis club members; (d) if the participants performed significantly worse than others, they would be asked to participate in another experiment later; and (e) the reward money would be decreased by 300 yen per serve based on the results (down from a maximum of 3,000 yen). In (e), if the score was 5 or 6, the reward would be maintained. However, if the score were 4 or less, the reward would be reduced to 300 yen per serve (down from 3,000 yen). Therefore, the maximum reward would be 3,000 yen and the minimum 0 yen. The experimenter announced the score results and the amount earned aloud. All participants were debriefed after the experiment.

We subjected all measurements—including psychological states, serve performance (i.e., scores, the number of point 6 achievements, VE, and BVE), and kinematics (i.e., ball speed, racket speed, and impact height)—to two-way mixed-design analysis of variance (ANOVA) with a group factor (no-contraction/contraction) and a repeated measures factor of the test (pre−/pressure). When we found an interaction, we conducted multiple comparison tests by applying the Bonferroni correction. The effect size was expressed as partial eta squared (η_p²). We performed statistical analyses using SPSS Statistics ver. 28 (IBM Corp. NY, Armonk, United States), with a significance level of 5%.

A two-way ANOVA confirmed an increase in somatic anxiety during the pressure test (M = 5.38, SD = 2.40) compared to the pre-test (M = 3.45, SD = 2.40) [F (1, 38) = 23.912, p < 0.001, η_p² = 0.442]. The interaction between group and test was also significant [F(1, 38) = 4.392, p = 0.043, η_p² = 0.442]. Simple main effect analyses demonstrated that somatic anxiety increased from the pre- to the pressure test and was significant for the contraction group (p < 0.001) and marginally significant for the no-contraction group (p = 0.055). Cognitive anxiety did not differ between the two tests [F(1, 38) = 2.533, p = 0.120, η_p² = 0.062] or groups [F(1, 38) = 1.561, p = 0.219, η_p² = 0.039].¹

Figure 4 outlines the performance scores. A two-way ANOVA revealed no effects of group [F(1, 38) = 0.008, p = 0.929, η_p² = 0.000] or test [F(1, 38) = 0.720, p = 0.401, η_p² = 0.019]. The interaction between group and test was not significant [F(1, 38) = 1.167, p = 0.287, η_p² = 0.030].

Figure 5 presents the number of maximum scores achieved. Neither a main effect of group [F(1, 38) = 0.310, p = 0.581, η_p² = 0.008] nor test [F(1, 38) = 1.724, p = 0.197, η_p² = 0.043] was found. However, the interaction between group and test was significant [F(1, 38) = 4.414, p = 0.042, η_p² = 0.104]. Simple main effect analyses demonstrated no group difference in the pre-test (p = 0.159). However, the contraction group improved the number of maximum score achievements in the pressure test (M = 1.25 times, SD = 0.94) compared to the pre-test (M = 0.6 times, SD = 0.49) (p = 0.021). The no-contraction group did not demonstrate any improvement in the number of maximum scores (pre-test, M = 0.9 times, SD = 0.76; pressure test, M = 0.75 times, SD = 0.94, p = 0.581).²

Table 1 presents the VE and BVE results. For VE (x), a two-way ANOVA revealed neither a main effect of group [F(1, 38) = 2.720, p = 0.107, η_p² = 0.067] nor test [F(1, 38) = 0.002, p = 0.968, η_p² = 0.000]. There was no interaction [F(1, 38) = 1.056, p = 0.311, η_p² = 0.027]. For VE (y), a two-way ANOVA indicated neither a main effect of group [F(1, 38) = 0.016, p = 0.901, η_p² = 0.000] nor test [F(1, 38) = 0.135, p = 0.715, η_p² = 0.004]. There was no interaction between group and test [F(1, 38) = 0.151, p = 0.700, η_p² = 0.004]. For BVE, neither an effect of group [F(1, 38) = 1.959, p = 0.170, η_p² = 0.049] nor test [F(1, 38) = 0.009, p = 0.923, η_p² = 0.000] was found. There was no interaction [F(1, 38) = 1.307, p = 0.260, η_p² = 0.033].

The kinematics data are summarized in Table 1. Both the ball and racket speeds were greater on the pressure test than on the pre-test [ball speed, F(1, 38) = 33.489, p < 0.001, η_p² = 0.468; racket speed, F(1, 38) = 6.763, p = 0.013, η_p² = 0.151]. These did not differ between the two groups [ball speed, F(1, 38) = 0.171, p = 0.681, η_p² = 0.004; racket speed, F(1, 38) = 0.137, p = 0.713, η_p² = 0.004]. No interactions were found [ball speed, F(1, 38) = 0.187, p = 0.668, η_p² = 0.005; racket speed, F(1, 38) = 0.537, p = 0.468, η_p² = 0.014].

For impact height, neither main effect of group [F(1, 38) = 0.550, p = 0.463, η_p² = 0.014] nor test [F(1, 38) = 2.079, p = 0.158, η_p² = 0.052] was found. There was no interaction [F(1, 38) = 0.003, p = 0.958, η_p² = 0.000].

We hypothesized that the left-hand contraction could prevent choking under pressure based on previous findings in a series of experiments of Beckmann’s group (Beckmann et al., 2013; Gröpel and Beckmann, 2017; Beckmann et al., 2021). However, serve performance did not deteriorate with pressure manipulation, suggesting a failure of choking induction. Therefore, we failed to test the choking intervention, although participants perceived somatic anxiety during the pressure test.

Interestingly, the left-hand contraction group increased the number of maximum score achievements, supporting the assertion that the left-hand contraction may not only prevent choking but also enhance performance under pressure (Beckmann et al., 2013, Study 2; Gröpel and Beckmann, 2017, Study 2; Mesagno et al., 2019, Study 2). Our results reconfirmed the beneficial effects of the left-hand contraction on tennis serve performance, apart from the context of choking intervention. Beckmann et al. (2021) reported the effectiveness of left-hand contraction in preventing choking among skilled tennis players. Therefore, our data further support the evidence of the beneficial effects of left-hand contraction on enhancing tennis serve performance.

Performance enhancement under pressure is referred to as a “clutch” (Otten, 2009) that is conceptually opposite to the choking phenomenon. The occurrence of a clutch can be ascribed to a high level of athletes’ perceived control (Otten, 2009), which is an element of implicit knowledge. Based on Seger’s (1994) description, Otten (2009) defined perceived control as knowledge derived from the accurate prediction of subsequent stimuli or the ability to control the values of variables. The left-hand contraction may be associated with an improved sense of perceived control.

Given the occurrence of clutches, participants may have perceived increased somatic anxiety as optimal tension rather than a hindrance to their actions. An optimal level of anxiety may increase motivation properly (Eysenck and Calvo, 1992). Cheng et al. (2009) pointed out that success relies on whether individuals can interpret current pressure as a positive event. It is possible that the left-hand contraction also influences the interpretation of anxiety.

The occurrence of the clutch in our study may be due to the right dominant hemispheric asymmetry induced by the left-hand contraction (Harmon-Jones, 2006; Hirao and Masaki, 2019), which is known to be an appropriate brain state for expert athletes (e.g., Hatfield et al., 1984). Inhibition of the left hemisphere and prefrontal coactivation (i.e., weak T7-Fz connectivity, indicative of unconscious motor control) occurs after left-hand contractions (Hoskens et al., 2020). Furthermore, T7-Fz inhibition occurred more strongly for experts than for novices in golf (Gallicchio et al., 2016), and more for experts than for near experts in shooting (Deeny et al., 2003). Therefore, the inhibition of left hemispheric activity that manifests as the right-dominant hemispheric asymmetry is thought to be an ideal brain state for high performance.

We posited that the left-hand contraction would conserve the serving kinematics even under pressure, whereas no contraction would disrupt the kinematics. However, this was not the case in this study. We observed no differences in movement kinematics between the two groups. Instead, we found improved performance accompanied by kinematic changes, although the direction of the changes was opposite to our prediction (i.e., the clutch). The kinematic analysis indicated that both groups exhibited faster ball and racket speeds during the pressure tests. In contrast to previous findings that have reported performance deterioration under pressure due to disrupted kinematics (e.g., decreased ball speed) (e.g., Sekiya and Tanaka, 2019), we noted better kinematic changes underlying the clutch phenomenon. Considering that the contraction group exhibited greater accuracy in serving toward the maximum scoring area, left-hand contraction may facilitate optimal serving attributes, such as increased ball speed and accuracy (e.g., Hernández-Davó et al., 2019), particularly under pressure.

The lack of differences in kinematics and serving variability (i.e., VE and BVE) between the groups can be attributed to analytical issues. The achievement of a successful tennis serve requires the coordination of several factors such as speed, impact angle, spin direction, and precision (Colomar et al., 2022). Mirifar et al. (2022) pointed out that the effect of unilateral hand contraction was fairly small and potentially insufficient to affect behavioral levels. Therefore, 3D kinematic analysis might be a good way to detect the beneficial effects of left-hand contraction (e.g., coordination among shoulder rotation, wrist and elbow flexion, or extension). Furthermore, in this study, the participants aimed at the target area instead of a specific target point, which was limited to the evaluation of several measures, including the mean radial error (MRE) and constant error (CE) from the target (Hancock et al., 1995). These indices are commonly used to assess tennis serve performance (Delgado-García et al., 2019; Hernández-Davó et al., 2019; Yamamoto et al., 2019; Beckmann et al., 2021). It would be fruitful to investigate the effects of hand contractions on these indices in future studies.

Our findings indicate that grasping a polyethylene material with the left hand would be helpful to enhance serving performance during critical, high-pressure matches in tennis. Previous studies have used soft balls (Beckmann et al., 2013; Gröpel and Beckmann, 2017; Mesagno et al., 2019) or tennis balls (Beckmann et al., 2021) for unilateral hand grasping. However, bringing these materials to real sports games is impractical, and grasping the ball per se likely interferes with motion. The developed wearable grasping material is likely to make the left-hand contraction more practical in real situations.

According to previous studies, left-hand contraction may be effective for expert or semi-expert players (Beckmann et al., 2013; Gröpel and Beckmann, 2017; Mesagno et al., 2019; Beckmann et al., 2021) but not for novices (Hoskens et al., 2020). Therefore, the effectiveness of our approach might be limited in skilled tennis players.

Our study has several limitations. First, we could not assess the intervention effects during choking. Therefore, pressure manipulation must be reconsidered. Our participants were driven to achieve success through pressure manipulation. We manipulated pressure using a reward of 3,000 yen (approximately 20 USD). While this amount was set to increase the likelihood of participants seeking the reward, it may enhance the clutch phenomenon rather than induce choking. Second, we should point out the technical limitations of EEG recordings in real sporting situations. EEGs are vulnerable to muscular activities. Consequently, in our study, we did not know whether T7-Fz connectivity was reduced by the left-hand contraction. Future studies should employ a certain task (e.g., golf putting) during which EEGs can be recorded.