The experimental protocol utilized in this study was adapted from Cook et al.’s work [28], entitled “Incorporating Game in Hippotherapy A Companion Book to the Brown Pony Series”, which has been shown effective in improving various aspects of cognitive function, social interaction skills, neural control and coordination in riders. The content of the 12 weeks of EAA training is shown in Table 3.

The 12-week EAA comprised 2 training sessions per week for 45–55 minutes each. The experimental intervention took place every week 1 and 3 from 15:50–17:50 and was performed by 6 MingLiu Horse Club instructors, each responsible for 2–3 children. An example of the lesson is shown in Fig. 2 [28].

Fig. 2.

Flow diagram of an EAA Intervention lesson.

Each EAAG and CG subject underwent a pre-test within 1 week prior to training and a post-test within 3 days after training.

2.3.2 fNIRS Measurements

In this study, a portable NirSmart63 was used to collect raw signal light intensities from the prefrontal and motor cortex of the brain during the Flanker task. Then, we calculated the mean value concentration change of oxyhemoglobin (Oxy-HB) in the prefrontal and motor cortex using the absorbance difference based on the improved Beer-Lambert law [30]. The experimental acquisition system comprised 14 signal sources and 14 detectors, forming the following 35 channels: 1=S1-D1, 2=S1-D6, 3=S2-D2, 4=S2-D7, 5=S3-D2, 6=S3-D3, 7=S3-D8, 8=S4-D3, 9=S4-D4, 10=S4-D9, 11=S5-D4, 12=S5-D10, 13=S6-D5, 14=S6-D11, 15=S7-D1, 16=S7-D6, 17=S7-D12, 18=S7-D13, 19=S8-D2, 20=S8-D7, 21=S8-D8, 22=S9-D3, 23=S9-D8, 24=S9-D9, 25=S10-D4, 26 =S10-D9, 27=S10-D10, 28=S11-D5, 29=S11-D11, 30=S12-D12, 31=S12-D13, 32=S13-D11, 33=S13-D14, 34=S14-D11 and 35=S14-D14. According to Bordmann’s partition, they were placed in the prefrontal and motor cortex, respectively (Fig. 3: Top, Front, Right and Left). The distance between the two probes on the signal acquisition cap was 3 cm, and they were arranged in a certain pattern to monitor the signals of the 35 channels covering the prefrontal and motor cortex. In addition, the more pronounced the activation of brain areas, the darker the color on the brain area map (pink $>$ purple $>$ blue).

Fig. 3.

Distribution of signal sources and detectors in the prefrontal and motor cortex.

Descriptive results are reported as means $\pm$ standard deviations. The assumption of normality was verified using the Shapiro-Wilk test. The t-test was used to perform the difference test between the EAAG and CG and the Oxy-Hb of each channel. The Flanker task data were analyzed using repeated measures Analysis of Variance (ANOVA). All statistical analyses were performed using the SPSS software version 26.0 (IBM Corp., Armonk, NY, USA), and significant differences are indicated by p 0.05, whereby ${}^{*}$ represents p 0.05 and ${}^{**}$ represents p 0.01.

The demographic variables included age, height, weight, body mass index (BMI), and the Flanker task, which included congruent and incongruent Accuracy and RT. The data were assessed for the EAAG and CG samples to reduce the effect of these factors on the experimental results by excluding demographic and IC differences between the two groups (Table 4).

Before conducting the independent samples t-test and chi-square test, all data underwent normal distribution testing, with a significance level of p $>$ 0.05. We observed no significant difference between the EAAG and CG groups in terms of height (cm) (126.6 $\pm$ 6.1 vs. 128.6 $\pm$ 8.2; T = –0.94, p = 0.35), weight (kg) (26.7 $\pm$ 3.6 vs. 26.5 $\pm$ 3.5; T = 0.45, p = 0.96), BMI (kg/m${}^2$) (16.6 $\pm$ 1.4 vs. 16.0 $\pm$ 1.6; T = 0.23, p = 0.82), congruent task accuracy (0.86 $\pm$ 0.06 vs. 0.86 $\pm$ 0.06; T = –0.22, p = 0.83), incongruent task accuracy (0.84 $\pm$ 0.07 vs. 0.86 $\pm$ 0.07; T = –1.12, p = 0.25), congruent task RT (718.1 $\pm$ 100.3 vs. 708.6 $\pm$ 81.6; T = 0.36, p = 0.72) and incongruent task RT (771.5 $\pm$ 87.9 vs. 762.9 $\pm$ 90.7; T = 0.34, p = 0.74). Altogether, these results indicated no difference in demographic variables and Flanker task between the two groups in the pre-test.

To investigate the changes in EAA on the IC of children aged 7–8 years, this study used 2 (group: EAAG, CG) $\times$ 2 (time: pre-test, post-test) repeated measures ANOVA to determine the Flanker task data pre-and-post and the results (Table 5).

In the congruent task accuracy (%), no significant main effect of group was observed, F(1,46) = 0.58, p = 0.45 $>$ 0.05, $\eta$${}^2$ partial = 0.01, while there was a significant main effect of time, F(1,46) = 12.91, p = 0.001 0.01, $\eta$${}^2$ partial = 0.22, indicating a significant interaction effect of group*time, F(1,46) = 9.21, p = 0.004 0.01, $\eta$${}^2$ partial = 0.17 (Table 5). In the analysis of simple effects, a highly significant difference (p 0.01) pre-and-post experiment was observed in EAAG, with a congruent task accuracy (%) pre-and-post experiment of 0.85 $\pm$ 0.06 and 0.89 $\pm$ 0.05, indicating an increase by 0.04, as well as in CG, which was 0.84 $\pm$ 0.07 and 0.87 $\pm$ 0.05, with an increase of 0.03 (Table 6).

In the incongruent task accuracy (%), there was no significant main effect of group, F(1,46) = 0.46, p = 0.51 $>$ 0.05, $\eta$${}^2$ partial = 0.10, while there was a significant main effect of time, F(1,46) = 5.85, p = 0.02 0.05, $\eta$${}^2$ partial = 0.11, as shown by post hoc multiple comparisons, whereby the EAAG of pre-and-post experiment measurements were highly significantly different (p 0.01), while no significant interaction effect for group*time was observed, F(1,46) = 3.29, p = 0.08 $>$ 0.05, $\eta$${}^2$ partial = 0.07 (Table 5).

In the congruent task RT (ms), there was no significant main effect of group, F(1,46) = 0.87, p = 0.36 $>$ 0.05, $\eta$${}^2$ partial = 0.20, but there was a significant main effect of time, F(1,46) = 11.52, p = 0.001 0.01, $\eta$${}^2$ partial = 0.20. In addition, there was also a significant interaction effect of group*time in the congruent task RT, F(1,46) = 19.42, p = 0.000 0.01, $\eta$${}^2$ partial = 0.30 (Table 5). In the analysis by simple effects, a highly significant difference was observed in EAAG pre-and-post experiment (p 0.01), with a congruent task RT of 718.07 $\pm$ 100.34 and 663.52 $\pm$ 71.65 (an improvement of 54.55), and in CG, with a congruent task RT pre-and-post experiment of 708.89 $\pm$ 90.61 and 715.63 $\pm$ 74.57 (a decrease of 6.55). The congruent task RT in the EAAG improved by 61.29 compared to the CG (Table 6). However, in terms of targeting between groups, there was a highly significant difference between the EAAG and CG post-experiment (p 0.01).

In the incongruent task RT (ms), there was no significant main effect of group, F(1,46) = 0.89, p = 0.35 $>$ 0.05, $\eta$${}^2$ partial = 0.02, while there was a significant main effect of time, F(1,46) = 40.57, p = 0.000 0.01, $\eta$${}^2$ partial = 0.47. Additionally, we also observed a significant interaction of group*time effect, F(1,46) = 24.49, p = 0.000 0.01, $\eta$${}^2$ partial = 0.35 (Table 5). In the analysis by simple effects, there was a highly significant difference in EAAG pre-and-post experiment (p 0.01), with an incongruent task RT of 771.07 $\pm$ 87.88 and 699.86 $\pm$ 73.69 (an improvement of 71.21), and in the CG, with an incongruent task RT pre-and-post experiment of 762.89 $\pm$ 90.68 and 753.91 $\pm$ 92.94 (an improvement of 8.21). The incongruent task RT of the EAAG was 62.23 higher than that of the CG (Table 6).

After 12 weeks of EAA intervention, the raw signals from the congruent task test were collected using NirSmart63 during the Flanker task in the two groups, followed by transformation using the NirSpark software (version 1.7.5, NirScan, HuiChuang, Beijing, China) to derive the change in Oxy-Hb concentration in each channel in the prefrontal and motor cortex in the EAAG and CG, respectively. The mean values of each channel in each of the 24 subjects were processed separately and combined with the paired-samples t-test to yield significant differences in the changes of each channel pre-and-post EAAG and CG, respectively (Table 7).

In this study, a pair of adjacent light sources and a detector were used to form a channel, and we calculated the intra-group average of the mean at the channel level. Then, the image was generated by the interpolation method of inverse distance [31], as shown in Figs. 4,5.

Fig. 4.

Brain activation of prefrontal and motor cortex congruent task test in EAAG. The color reflects the mean of each region on the time scale, and the purple color represents higher activation than the blue–colored regions. The gray node represents the channel formed by each light source probe and detector probe, and the channel number is the label of the channel.

Fig. 5.

Brain activation of prefrontal and motor cortex congruent task test in CG.

The values of changes in Oxy-Hb in each channel in the prefrontal and motor cortex in the pre-and-post congruent task in EAAG and CG were examined by a paired-sample t-test. The results showed a statistical difference (p 0.05) between the EAAG pre-and-post in channels 7, 21 and 23 (pink areas), as shown in Fig. 6. However, there was no statistical difference in CG. According to Bordmann, channels 7, 21 and 23 belong to the right dorsolateral prefrontal cortex (R-DLPFC).

Fig. 6.

Brain activation of significant channel regions in EAAG.

In the Flanker task test, the paired-sample t-test for the incongruent task individual channels data was not significantly different between EAAG and CG (p $>$ 0.05).

Our findings suggest that 12 weeks of EAA increased IC to some extent in children aged 7–8 years. Previous studies have shown that IC development rates vary at different stages. Anderson et al. [32] found that 6–7 years was a sensitive period for development, with growth slowing down after age 7 and leveling off after age 10. In this study, the subjects belonged to this age group. Training load has always been an important aspect of sports and must be emphasized, as the amount of exercise intensity and time may also directly affects IC. A previous study showed that a chronic exercise cycle of 12–24 weeks, with an exercise time of 50–90 min/week and frequency of 2–3 times/week, significantly affected IC [33]. The training load requirements of our experiment were largely consistent with these findings and may have a positive impact. Additionally, supportive evidence was found in a study on open- and closed-skill sports [34]. It is well known that closed skills sports are performed in a relatively stable and predictable environment, in which motor actions are repetitive and unrelated to the external environment, whereas open skills sports are performed in a dynamic and changing environment that requires constant adaptation to external stimuli [35]. EAA is an open-skill sport that requires overcoming the distractions of the external environment, maintaining a high level of concentration, and ultimately automating a complex physical activity. These complex movements require increased cognitive involvement [36], avoiding incorrect technical movements and activating more brain systems [37]. Thus, the nervous system regulates and controls both simple and complex movements, with the brain playing the most significant and prominent regulatory role. Related studies have found that sports with greater cognitive involvement are more effective in exerting further positive effects on children’s cognitive abilities than sports without cognitive involvement [37]. In addition, riders are required to maintain a high degree of attention and tension during the whole process, which helps improve their attention retention. Studies have shown that EAA can effectively improve children’s attention [38, 39] and may further improve their IC.

In this study, it was observed that 12-week EAA activated the R-DLPFC in the Flanker congruent task. The changes in the Oxy-Hb in the R-DLPFC of the brain indicate an increase in local cerebral blood flow triggered by the activation of neuronal cells in the cerebral cortex. According to the principle of neural cell coupling, increased local brain oxygenation suggests that the brain needs to recruit more neural resources [40], which in turn indicates that EAA training requires constant activation of neural processing associated with the R-DLPFC to adapt to the more complex cognitive demands of the relevant training task. It has been shown that high accuracy during the execution of attentional network tasks may originate from the high neural activation of the right frontoparietal network [41], which showed a clear right hemisphere dominance [42, 43]. Moreover, meta-analysis correlation results also showed that the brain activation sites associated with the Flanker task were mainly found in the R-DLPFC [44] and that different types of cue conditions were associated with frontoparietal network activation. Comparatively, R-DLPFC is thought to be an important component of neural processing in the right frontoparietal network [45]. Significant activation in this region also reflected reliable performance in extrinsic behavior, as indicated by the observed increased accuracy and decreased reaction time in the Flanker task. The dorsolateral prefrontal cortex coordinates the rest of the human brain and plays a crucial role in arithmetic and executive functions. Numerous investigations closely related to the activation model of brain regions in the Flanker task show that the activation is mainly present in the R-DLPFC [44]. In addition, fNIRS has also been used to monitor prefrontal cortex oxygenation in other sports and showed that testing different sports effectively activated R-DLPFC. Miao Yu et al. [46] found significant activation of the R-DLPFC in all three groups through a comparison between elite, expert and novice ice hockey athletes, which was concordant with our present study.

Neuropsychological and neuroimaging findings link dorsolateral prefrontal activation and top-down attentional control [47]. In addition, recent findings reveal an important role of R-DLPFC in task preparation [48], which plays an irreplaceable and important role in human development. In this present research, the children selected were 7–8 years old and did not undergo relevant training, and there was a greater need to enhance top-down attentional control to always pay attention to the outside world and changes in the trainer’s commands. This continuous state probably enhanced the attentional control of the riders, signifying that EAA training might lead to greater stimulation of the right R-DLPFC, which is related to attentional control in cognitive function. Previous studies have shown the importance of R-DLPFC activation in top-down attention [49]. Thus, it could be deduced that EAA training activated the R-DLPFC and improved the ability to process conflicting information, which in turn enhanced participants’ cognitive functioning. The result of better performance in completing cognitive tasks provides neurophysiological evidence that EAA training promotes the development of cognitive function and further supports the behavioral findings of this study.

Other prefrontal and motor cortex regions did not show significant activation changes. One possible explanation could be that cognitive changes depend to a substantial extent on prefrontal cortex function, which leads to a possible reason for the lack of significant changes before and after activation of the motor cortex. In addition, important subcortical areas, such as the hippocampus, are difficult to detect and could be considered a limitation of the present study.