Introduction

Mental rotation is a cognitive spatial ability defined as the ability to rotate two- or three-dimensional objects quickly and accurately in one’s mind (Shepard & Metzler, 1971). Mental rotation is a complex ability influenced by biological as well as psychosocial factors (Hausmann, 2021). A recent review article suggests that the underlying processes in mental rotation include motor rotation, visuospatial processing, and higher-order visual object recognition (Hiew et al., 2023). Tests with simple rotational material in a habituation task displayed that even infants as young as 5 months can perform mental rotation (Erdmann et al., 2018). Elementary school children (between 8 and 10 years) can already perform above chance level when they try to solve the mental rotation test (MRT) according to Peters et al., (1995), in which three-dimensional cube figures have to be rotated in depth (Moè, 2018a, b; Titze et al., 2010). Research usually reports a male advantage in mental rotation ability already from infancy on (Enge et al., 2023). At the same time, mental rotation ability appears to be malleable, for instance, through (spatial) training (Uttal et al., 2013). Boxing could serve as such a spatial training that also influences children’s self-concept (Annesi, 2006; Liu et al., 2015; Pfister et al., 2023; Schranz et al., 2014), which can again affect mental rotation performance (Rahe et al., 2023). Accordingly, the present study investigates the effect of boxing training, self-concept, and mental rotation ability in fourth-graders.

The Malleability of Mental Rotation Abilities

Uttal et al., (2013) showed in a meta-analysis that mental rotation is a malleable ability that can be trained by video games and spatial task practices. For children younger than 13 years, they found an effect size of g = 0.61 (SE = 0.09, m = 53, k = 226). This training can also transfer to other spatial abilities (Uttal et al., 2013). Studies with adults (Jansen & Lehmann, 2013; Ozel et al., 2002; Pietsch and Jansen, 2012a; Schmidt et al., 2016) and adolescents (Feng et al., 2017) also found that athletes outperformed non-athletes in a mental rotation task. Reasons for that could be similar brain regions (intraparietal sulcus) that are activated during motor coordination (Draganski et al., 2004) and mental rotation (Jansen & Heil, 2010; Jordan et al., 2001). Erickson et al., (2011) found that the hippocampus volume increased in elderly participants after aerobic training and that this brain region is important for spatial memory function.

Furthermore, participants with faster coordinative skills solved more items of the MRT correctly than slower participants (Pietsch and Jansen, 2012b). Weigelt and Memmert, (2021) found that basketball players with high expertise outperformed novice players in a mental rotation test using basketball plays as rotational material. In children, participants with better motor abilities (Jansen & Heil, 2010) and normal weight (Jansen et al., 2011) performed better than those with worse motor abilities and overweight. Jansen and Heil concluded that motor development and spatial abilities might be related when motor ability includes motor coordination.

These results indicate that physical movement could be favorable for people’s spatial cognition. Rather than comparing groups of athletes and non-athletes (Voyer & Jansen, 2017) or calculating the influence of sports performance on mental rotation performance, some studies measured the effect of specific training on participants’ mental rotation performance. After a 10-month training of either running or wrestling in undergraduate students, Moreau et al., (2012) found that wrestlers improved their mental rotation abilities more than runners. In 6- to 14-year-old girls, a 3-month juggling training compared to light strength training led to faster reaction times in a chronometric mental rotation test (Jansen et al., 2011). Wiedenbauer and Jansen-Osmann, (2008) trained one group of 10 and 11-year-old children with a task where they manually rotated objects on a screen with a joystick. These children had a significant decrease in reaction time (posttest – pretest) compared to a control group that solved a quiz game. Blüchel et al., (2013) trained a group of 8- to 10-year-old children with a specific coordination practice and compared this group to children who did not receive training. The training group improved significantly more than the control group between a pre- and a post-training MRT. The effect size of the improvement in the trained group was d = 0.92. The exercise contained some ball tasks, a motor memory game, orientation, skateboarding, and juggling tasks. Another study with 7- to 9-year-old children who either received a 5-week training in creative dancing or normal physical education (control group) found that the creative dancing group improved more in their mental rotation performance than the control group (η2 = 0.08, Jansen et al., 2013). Pietsch et al., (2017) trained one group of children with a mean age of 9 years with a specific motor co-ordinative (Life Kinetik, Lutz, 2017) training program (experimental group) while another group received physical education (control group) according to their curriculum in school. The Life Kinetik practice consisted of ball tasks, lateral and cross-over motions, finger games, and course running. Comparing both groups with pre- and post-training MRTs, results confirmed higher improvement of the experimental group (d = 1.52). Furthermore, Pietsch and Jansen, (2018) illustrated that specific laterality training with the non-dominant foot during soccer training improved the performance of 10- and 11-year-old children more than the same training with the dominant foot. The effect size of the training was η2 = 0.41 with a significant interaction between time and group, η2 = 0.31. It can be summarized that manual and athletic practice improves mental rotation performance in children and that specific practices seem to be more favorable than normal physical movement. Furthermore, coordinative training seems to be more suitable to improve spatial abilities than simple aerobic training (Moreau et al., 2012). The first research question of the present study, therefore, was whether boxing training with fourth graders would improve their mental rotation performance.

Boxing belongs to the interceptive sports, which require coordination between body (parts) and an object and/or the environment (Davids et al., 2002). It takes place in a dynamic visual environment, involving situations in which an athlete’s spatial performance is critical for success (Pfister et al., 2023). Similar activities like wrestling (Moreau et al., 2012), manual training (Wiedenbauer & Jansen-Osmann, 2008), coordination training (Blüchel et al., 2013; Pietsch et al., 2017), creative dancing (Jansen et al., 2013), or laterality training (Pietsch & Jansen, 2018) can have positive effects on mental rotation performance, and a meta-analytic review by Voss et al., (2010) points out that particularly in interceptive sports, training can have large effects on cognition. Moreover, physical activity can improve children’s self-concept (Liu et al., 2015), leading to the second research question.

Self-Concept

The second research question of the present study is whether boxing training could also improve children’s self-concept. A meta-analysis found that physical activity had a positive effect on children and adolescents’ self-concept (g = 0.49, p = 0.014) and self-worth (g = 0.31, p = 0.005) (Liu et al., 2015).

Self-concept is defined as people’s knowledge on what or who they are and their evaluation on how they feel about themselves (Wehrle & Fasbender, 2019). It is malleable but stable over time, influenced by someone’s experiences, and controls how we process information about ourselves (Wehrle & Fasbender, 2019). Shavelson et al., (1976) state that the definition of self-concept is multifaceted and identifies academic, social, and physical self-concept as different aspects. Academic self-concept is associated to people’s achievement in different subjects, e.g., history, math, science, while the physical self-concept focuses on physical ability and physical appearances (Shavelson et al., 1976). Social self-concept relates to social situations where peers and significant others are involved.

A meta-analysis found that physically oriented interventions for children positively affected their self-concept (g = 0.34, O’Mara). Focusing on the physical self-concept, Annesi, (2006) trained 9 to 12-year-old children in a 12-week program. Two groups were enrolled in training that included cardiovascular activities, resistance training, self-regulatory skills, and general health and nutrition information. Children of the control group engaged themselves in unstructured and voluntary physical activity. An improvement in physical self-concept was found only for one of the training groups (d = 0.22, p = 0.013), but not for the control group (d = 0.16, p = 0.358) and the second training group (d = 0.31, p = 0.080). In both training groups, the physical activity frequency was positively associated with the change in physical self-concept (Annesi, 2006). The self-concept of overweight male adolescents also improved after a 6-month resistance training program compared to a control group that did not receive any training (Schranz et al., 2014).

Regarding the social self-concept, Jones et al. (2014) found better social self-concept—measured with a popularity scale of a self-concept questionnaire—in children with epilepsy after an intervention study. Moreover, Peens et al. (2008) examined 7- to 9-year-old children with developmental co-ordination disorder and trained them with different programs (self-concept enhancing intervention, motor intervention, psycho-motor intervention, and a control group). Results showed that both motor interventions and the self-concept enhancing intervention significantly enhanced children’s self-concept compared to the control group and this effect was found for the social and the academic self-concept.

A possible mechanism for the positive effects of physical exercise on children’s and adolescents’ self-concept could be their perceived fitness level which improves due to the training and is then related to their self-worth (Daley, 2002). Motor ability is also linked to self-esteem and self-concept partly via perceived social acceptance (Schmidt et al., 2015). Boxing training with elements of coordination, communication, and playful duels could enhance children’s perceived social acceptance which could then improve their self-concept. Another mechanism can be the mediating role of body satisfaction: Physical activity can reduce body dissatisfaction which can then enhance (physical) self-concept (Fernández-Bustos et al., 2019). Moreover, self-concept can be a mediator in the relationship between motor ability and physical activity in both directions (Jekauc et al., 2017). On the one hand, more active children develop a better self-concept which then leads to better motor abilities. On the other hand, better motor abilities due to training enhance children’s self-concept, and that can then lead to more physical activity. In children with a migration background who might not always be successful in school, physical training could be particularly favorable to enhance their self-concept. The present study focuses on the effects of boxing training on children’s self-concept as well as on their mental rotation abilities.

The Goal of the Study

The present study analyzes the effects of boxing training on children's self-concept and mental rotation abilities. We predicted that children who participated in the training would report increasing physical (Annesi, 2006) (H1a), social (H1b), and academic self-concept (H1c) compared to the children who did not participate in the training. For children’s mental rotation abilities, we hypothesized that the training group would increase more in mental rotation performance than the control group (H2).

Methods

Participants

The participants were 45 fourth graders (26 boys and 19 girls) from three classes between 9 and 12 years (M = 9.62, SD = 0.71). The children were cluster-randomized by classes. One class with seventeen children participated in the training group and two classes with 28 children were used as a control group. Children in the control group did not receive any specific training and were significantly older (M = 9.79, SD = 0.78, p = 0.048) than those in the training group (M = 9.35, SD = 0.49), t(43) = -2.03, p = 0.048, d = -0.62. A Chi2-test revealed no deviation from an even distribution of gender between both groups, Chi2(1) = 3.09, p = 0.121. The study was carried out in a school with a high percentage of parent unemployment. 85% of the children spoke a different language than German with their family members.

For effects of the intervention on children’s self-concept (Liu et al., 2015), a G*Power (Faul et al., 2007) analysis for the repeated measure ANOVA with a medium effect size of f = 0.25 (α = 0.05, 1 – β = 0.80, two groups, two measurement points) revealed a total sample size of 34 participants.

Because effect sizes for mental rotation found in the literature varied between moderate (Jansen et al., 2013) and large (Blüchel et al., 2013; Pietsch & Jansen, 2018), we predicted a moderate effect size (Jansen et al., 2013). A G*Power (Faul et al., 2007) analysis for the repeated measure ANOVA with a medium effect size of f = 0.25 (α = 0.05, 1 – β = 0.80, two groups, two measurement points) revealed a total sample size of 34 participants.

ANOVAs could not be calculated because of the violation of homogeneity or normal distribution. Therefore, post hoc Power analyses for the used tests were calculated. For the three interactions with large effect sizes, the power was > 0.88, only for the moderate effect size (social self-concept), the power was 0.65.

Material

Mental Rotation Test

The paper–pencil mental rotation test consisted of 12 items (according to Peters et al., 1995; Vandenberg & Kuse, 1978). For each item, children had to compare one main cube figure on the left side to four comparison figures on the right side. The children had to cross out the two comparison figures that were identical but rotated versions of the main figure. The remaining two figures were mirrored versions of the main figure. Children received one point only if both identical figures were crossed out. Internal consistency for the MRT was critical for t1 and good for t2 (Cronbach’s alpha: t1: 0.69, t2: 0.85). Guttman Split-Half Coefficient was acceptable (t1 = 0.79, t2 = 87). A sum score of all 12 items was calculated.

Self-Concept

To assess children's self-concept, we used the self-concept questionnaire (SKF, Selbstkonzeptfragebogen für Kinder, Engel, 2015) which consists of 27 items that had to be answered on a 4-point Likert scale ranging from 1 = not true at all to 4 = totally true. Answers were pictured as smileys and a fifth option (I don’t know) was presented as a question mark. Three aspects of the self-concept were measured with the SKF: Physical self-concept was measured with eight items (“I often have a stomach ache”), the academic self-concept with nine items (“I can do many things”), and the social self-concept with ten items (“I like to help others”). One item of the academic self-concept and two items of the physical self-concept had to be eliminated because of low corrected item-total correlations (< 0.3). Internal consistencies for the remaining items were acceptable and good for both measurement points: Physical self-concept (six items, Cronbach’s alpha: t1: 0.74, t2: 0.78), social self-concept (ten items, Cronbach’s alpha: t1: 0.81, t2: 0.81), academic self-concept (eight items, Cronbach’s alpha: t1: 0.76, t2: 0.79). Mean values were calculated for each subfactor.

Training

The boxing training consisted of 12 units divided over ten weeks with at least one training session per week. Each unit lasted for 90 min and was planned by one trainer with a fitness trainer license and a degree in sports studies. A second trainer with many years of experience as a boxer supported the practice. The presence of two trainers ensured that one trainer could help individual children who had problems with a specific task. Besides some individual support, all children received the same training at the same time.

Each unit included the repetition of rules, warm-up, training of coordination, simulation of boxing moves, playful form of duels, exercises in motion (around a partner), and a concluding discussion of the practice. Children were not allowed to apply what they learned outside of the lessons. There was no drop-out of the study. Besides normal missed sessions due to illness, all children completed the training sessions.

Procedure

One week before the first training session, children were tested in mixed-gender groups in their classrooms (Moè, 2018b) with paper–pencil tests and questionnaires. First, the children filled out the self-concept questionnaire. Afterward, two items of the mental rotation test were solved with the whole group to ensure that all children understood the task. The difference between a rotated image and a mirror image was explained. Then, all children had five minutes to solve the 12 items of the test. After the last training session, the same procedure was performed for data collection.

One parent of each child gave his/her written informed consent. The study was approved by the Supervisory and Service Directorate (Aufsichts- und Dienstleistungsdirektion, ADD) and conducted according to the ethical guidelines of the Helsinki Declaration. Ethical approval for this study was not required following the conditions outlined by the German Research Foundation where research that carries no additional risk beyond daily activities does not require Research Ethics Board Approval.

Statistical Analyses

For some variables, the assumption of normal distribution was violated (MRT at t1 and t2 and physical self-concept at t2). Therefore, we report the main effects of time using the paired sample t-test when the variable was normally distributed or the Wilcoxon test when this assumption was violated.

To test the interaction effects of time and group, we first calculated changes over time (t2-t1) for each variable (MRT, academic, social, and physical self-concept). Here, the assumption of normal distribution was violated for the chance (t2-t1) in MRT and physical self-concept. To compare both groups, we then ran the Wilcoxon-Mann–Whitney test if the data were not normally distributed. In the case of normal distribution, we used the t-test in case of variance homogeneity and the Welch test in case of heterogeneity. The assumption of homogenous variances was violated by comparing age, social self-concept at t1 and t2, and academic self-concept at t2 and for the change. The results of our hypotheses (interaction effects of time and group) are reported in Table 1.

Table 1 Descriptive data of the sample as a function of group, effect sizes, and internal consistencies of the used tests and questionnaires
Full size table

To give a better overview, other group differences for age, gender, and all variables at t1 and t2 separately can be found in Table 1 as well and the main effects of time for both groups are reported in the text.

Because children in the control group were about five months older, we checked the effects of age. Age did not significantly correlate with any of the study variables. Adding age as a covariate only slightly enhanced or reduced the effect sizes of the interaction effect of group and time. Furthermore, no significant main effects of age or interaction effects of age and group appeared.

Results

Testing effects for the physical self-concept (H1a), we found a moderate non-significant main effect of time, z = 2.01, p = 0.044, r = 0.31 (Fig. 1). The interaction between time and group was significant with a large effect size, t(43) = 3.27, p = 0.002, d = 1.01. Simple main effects revealed a large significant improvement for the training group, z = 3.22, p = 0.001, r = 0.78, but not for the control group, z = 0.01, p = 0.99, r < 0.01.

Fig. 1
figure 1

Physical self-concept as a function of group and time

Full size image

We then tested effects for the social self-concept (H1b) and found no significant main effect of time, z = 1.87, p = 0.062, r = 0.28 (Fig. 2). The interaction between time and group was significant with a moderate effect size, U = 123.00, Z = -2.70, p = 0.007, r = -0.40. Simple main effects revealed a significant large improvement for the training group, z = 3.32, p < 0.001, r = 0.81, but not for the control group, z = -0.18, p = 0.86, r = 0.03.

Fig. 2
figure 2

Social self-concept as a function of group and time

Full size image

Testing effects for the academic self-concept (H1c), we found no significant main effect of time, z = 1.49, p = 0.136, r = 0.22 (Fig. 3). The interaction between time and group was significant with a large effect size, t(42.82) = 3.36, p = 0.002, d = 0.91. Simple main effects revealed a significant improvement for the training group, z = 3.31, p < 0.001, r = 0.80, but not for the control group, z = -0.52, p = 0.60, r = 0.10.

Fig. 3
figure 3

Academic self-concept as a function of group and time

Full size image

For mental rotation performance (H2), we found a significant main effect of time, z = 4.13, p < 0.001, r = 0.62 (Fig. 4). The interaction between time and group was significant with a large effect size, U = 42.00, Z = -4.65, p < 0.001, r = -0.69. Simple main effects revealed a significant improvement for the training group, z = 3.64, p < 0.001, r = 0.88, but not for the control group, z = 1.42, p = 0.16, r = 0.27.

Fig. 4
figure 4

Mental rotation performance as a function of group and time

Full size image

Discussion

The present study analyzed the effects of boxing training on children’s mental rotation performance and their self-concept. For the academic, physical, and social self-concept, moderate to large interaction effects of time and group were found. The self-concept of trained children improved more than that of children without training. The effects of time on children with boxing training were large whereas untrained children only had non-significant small or very small improvements in self-concept. Regarding mental rotation performance, the experimental group improved a lot more than the control group with a large effect size for trained and a moderate, non-significant effect for untrained children.

For the three subscales of self-concept, children who participated in the boxing training improved more than untrained children. This is in line with meta-analyses showing that physical activity leads to a higher self-concept and self-worth (Liu et al., 2015) and to a higher physical self-concept (Babic et al., 2014). Liu et al., (2015) found moderate effects of physical activity on the general self-concept in children. Our results revealed a moderate interaction between the training group and time for the social self-concept and a large effect size for the academic and the physical self-concept. The training in the present study consisted of ten-week boxing training for fourth graders with a high percentage of migrant children and unemployment of parents. Similarly, in the meta-analysis, most studies analyzed marginalized groups, as well, albeit focusing on overweight children or participants with special needs (disabilities, asthma, cerebral palsy, sedentary children).

Regarding the improvement during the intervention for the trained group, Jones et al. (2014) found a moderate effect size while our effect size was large for the boxing training group. Their trained children were slightly older and the intervention lasted 12 weeks with 50 to 60 min of training each week. In contrast, our children received 90 min of training each week. Annesi, (2006) found only a small effect size for children of similar age for a 12-week intervention on their physical self-concept while the effect size for our training group was large. This could be due to the diverse components of the training and the small group with two instructors.

One reason for the large effects of the training on children’s self-concept could be the training itself. Each training was carried out in a rather small group of 17 children with two trainers and lasted 90 min. It included not only boxing matches but the teaching of rules and behavior for the matches. Children were told they were not allowed to apply what they had learned in the boxing training outside of the lessons. All conflicts must be resolved without violence. Furthermore, they were taught a stopping rule: They had to stop boxing if the partner called to stop. Moreover, respect concerning the training was taught: children learned that the training started on time, that they had to meet others with respect, and that they were not allowed to insult others. Before and after a match, children were touching gloves which symbolized respect for each other and the rules, a willingness not to want to hurt the partner, and a thanks to the opponent for practicing together. That could have improved especially the social self-concept (Peens et al., 2008). The training also consisted of motor co-ordinative practices. Children had to alternately bounce a ball with the dominant and the non-dominant hand, touch a partner’s shoulder and knee, or simulate boxing moves. This could have improved children’s physical self-concept (Daley, 2002). During the training, the children learned many new movements, rules, and practices and made the experience that they could succeed. Furthermore, the concluding discussion at the end of each lesson demonstrated to the children what they had learned. This could have improved children’s academic self-concept (Peens et al., 2008). It can be assumed that different components of boxing training may have influenced various aspects of children’s self-concept.

Another reason why the boxing training was successful could be the socio-economic background of the children and their families: 80% of the children spoke a first language that differed from German and a high percentage had unemployed parents. Studies found that children in families with a low income had lower self-concepts than those in high-income families (Garg, 2021; Rana, 2020) and that socio-economic status positively correlated with self-concept (Fin & Ishak, 2013; Li et al., 2020). Other studies found higher self-concepts in Spanish and Asian language-speaking children compared to natives (Niehaus & Adelson, 2013) but regarding social-emotional self-concept, Spanish-speaking children had higher values in internalizing and externalizing problems. We did not compare our sample with children from a higher socio-economic background. Hence, we can only assume that their self-concept at t1 could have been low. This could have provided more scope for improvements.

The large positive effect of our boxing training on children’s mental rotation performance is in line with studies showing positive effects of wrestling (Moreau et al., 2012), manual training (Wiedenbauer & Jansen-Osmann, 2008), coordination training (Blüchel et al., 2013; Pietsch et al., 2017), creative dancing (Jansen et al., 2013), or laterality training (Pietsch & Jansen, 2018). All of these practices included coordinative training of participants’ hands or bodies. The training in the present study included simulation of boxing moves, exercises in motion around a partner (rotation around the partner’s axis), playful forms of duels, and training of coordination (bouncing a ball alternately with the right and the left hand or touch a partners’ shoulder and knee). The physical training could have several effects on children: working memory, cognitive flexibility, and attention could have been improved (Alvarez-Bueno et al., 2017; de Greeff et al., 2018; Pfister et al., 2023) or better strategies for solving mental rotation tasks could have been used (Heil & Jansen-Osmann, 2008; Meneghetti et al., 2017). To the best of our knowledge, no study has yet investigated the specific effects of boxing training on children’s mental rotation abilities. The meta-analyses included all types of physical intervention programs (Alvarez-Bueno et al., 2017) and physical activities (de Greeff et al., 2018) on children’s cognition and academic performance.

Moreover, the coordinative practices in particular could affect children’s brain connectivity (Demirakca et al., 2016). Using event-related potentials, a study could show that the hemispheric laterality during a mental rotation task was significantly lateralized towards the left hemisphere in second graders, non-significantly in sixth graders, and completely bi-lateralized in adults (Jansen-Osmann & Heil, 2007). For the intraparietal sulcus, Kucian et al., (2007) found stronger activation in the left hemisphere in adults compared to children. Special coordinative training could accelerate these brain developments.

Children’s self-concept is a good predictor of their future academic performance (Möller et al., 2020) and their mental rotation ability is a good predictor of adolescents’ choice of STEM subjects in higher education (Wai et al., 2009). Therefore, it is important to train these abilities and improve the knowledge about own skills and characteristics. As our boxing training was rather short, it could be included in the elementary school program or as an extra-curricular workshop. This would be especially important for children from lower socio-economic backgrounds and for children with low social self-concept and low academic self-concept.

Regarding specific boxing training, one study found positive effects of a short 3-min boxing training on a decision-making task (Wollseiffen et al., 2016). Another study compared Ukrainian adolescent boxers with non-boxers and found advantages for the boxers regarding fast reaction and indicators of attention capacity (Solovey et al., 2021). Other studies investigating boxing mostly focus on the negative effects of professional and amateur boxers (Kim et al., 2019).

One limitation of the study is that the control group did not receive training at all. Part of the effects of the training—particularly on children’s self-concept—could be that they got attention from two trainers for ten weeks. Above that, the children were assigned to the groups at the classroom level. We cannot be sure that the class of the trained group did not receive any other lessons that might have influenced their self-concept or mental rotation performance. Furthermore, we did not analyze long-lasting effects by investigating children’s mental rotation abilities and their self-concept months after the training. As the sample size was rather small, no additional group effects of gender or native language were analyzed. Further studies should examine the effects of practices on girls and boys depending on their socio-economic background.

To conclude, a ten-week boxing training is suitable to improve children’s self-concept and their mental rotation abilities. Such a coordinative practice can simultaneously improve children’s cognitive and self-perceived knowledge while they can try out physical exercise. Because of the cooperative practice and the communication of rules and behavior, it might also improve their character and their social skills while the children are having fun practicing and learning.