This ALE meta-analysis has been conducted following a strict protocol by using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher et al., 2009).

A literature search was conducted by PubMed,¹ ISI Web of Science,² Elsevier³ and Cochrane Library.⁴ All included articles were published in the English language until March 2021. The following search keywords were used: “athlete” OR “expert” OR “novice” OR “non-athlete,” “decision-making” OR “decision,” “functional magnetic resonance imaging” OR “fMRI” OR “positron emission computed tomography” OR “PET.” The retrieval formula was (“athlete” [Mesh Terms] OR expert OR novice OR non-athlete) AND (“decision-making” [Mesh Terms] OR decision) AND (“functional magnetic resonance imaging” [Mesh Terms] OR “fMRI” [Mesh Terms] OR “positron emission computed tomography” [Mesh Terms] OR PET).

The titles and abstracts were independently screened by two trained reviewers (YD and LXH). Two independent reviewers completed the entire article inclusion process, and only articles that both reviewers reached an agreement on were finally included. When two independent reviewers disagreed on whether to include articles, resolved the differences through discussion, and seed the assistance of a third reviewer (DBL) if they still could not be resolved.

We selected studies considering the following inclusion criteria: (1) the samples included a group motor expert and a group novice; (2) motor experts with extensive experience were awarded the title of experts by their country, region, or school; (3) subjects performed motor decision-making task stimulus; (4) study with fMRI or PET; (5) the peak coordinates of brain activation areas were explicitly reported of motor experts and novices; (6) the reported results of activated foci were normalized to the Montreal Neurological Institute (MNI) (Collins et al., 1994) or the Talairach standardized stereotactic spaces (Paus et al., 1996); (7) whole-brain voxel analysis or at least one whole-brain analysis was performed; (8) subjects were healthy; (9) subjects were adults; (10) the language of studies was limited to English; (11) original research articles.

Exclusion criteria: (1) report coordinates were incomplete, or complete results have not been obtained after contacting the corresponding author; (2) research published in the form of conference reports, abstracts, etc.

These criteria identified 10 studies including 350 people (168 experts and 182 novices, 411 activation foci). Figure 1 shows the outcomes of the search process.

The two reviewers (YD and LXH) extracted the following information from each included study: authors, age, year of publication, study name, type of coordinates, brain analysis method, number of participants, the ratio of male to female participants, comparison conditions, handedness, Foci.

ALE is a quantitative meta-analysis method that uses a spatial variance model (3D Gaussian) to calculate the likelihood of each voxel being activated under a certain condition, thus obtaining the consistency of brain activation across multiple experiments (Laird et al., 2005; Eickhoff et al., 2009). Data processing was performed using GingerALE software (version 3.0.2)⁵ (Eickhoff et al., 2012). The difference in coordinate space (MNI vs. Talairach space) could be explained by converting the coordinates in Talairach to MNI space using icbm2tal in GingerALE (Lancaster et al., 2007). Finally, all activation coordinates were displayed in MNI space. We adopted a threshold for the map of the final ALE score graph with a familywise error (FWE) at p < 0.05 and a minimum cluster size of k > 10 mm³. For visualization, the ALE whole-brain maps were imported into Mango software (version 4.0.1)⁶ overly on a standardized anatomical MNI template (Colin27_T1_seg_MNI) (Dehghan et al., 2016).

A total of 1,323 records were initially retrieved, and only 1,191 studies remained after deleting duplicates (n = 132). We screened potentially relevant articles by applying the inclusion and exclusion criteria (see Supplementary Material 1). Finally, 10 articles focusing on the brain activation of experts and novices while performing decision-making tasks were included in the present study. All included articles tested decisions made by experts and novices under the same stimuli. Among those papers, six (Wright et al., 2011; Wu et al., 2013; Balser et al., 2014; Wimshurst et al., 2016; Qiu et al., 2019; Blazhenets et al., 2021) determined the accuracy of the ball’s flight direction, three (Bishop et al., 2013; Wright et al., 2013; Xu et al., 2016) tested to determine the accuracy of opponent behavior, one (Guo-Zheng, 2016) tested response accuracy rate of ball block. Regarding the types of sports including basketball, tennis, hockey, volleyball, handball, badminton, soccer, etc. Table 1 described the detailed information of the included study.

Experts had 93 foci in 5 different experiments, the 3 regions activated included the right inferior temporal gyrus (ITG) (BA 37), right sub-gyral (BA 37), right middle occipital gyrus (MOG) (BA 19).

Whereas novices had 124 foci in 7 different experiments. After completing the single data ALE analysis, novices activated the left precuneus (BA 7), left superior parietal lobule (SPL) (BA 7), left inferior parietal lobule (IPL) (BA 40), left middle frontal gyrus (MFG) (BA 6), left precentral gyrus (BA 4, BA 6), left cingulate gyrus (BA 24), left postcentral gyrus (BA 2), left supramarginal gyrus (BA 40), right middle temporal gyrus (MTG) (BA 37, BA 39), right superior temporal gyrus (STG) (BA22). Table 2 and Figure 2 showed the results of the single-dataset ALE analysis.

Experts did not have significant activation compared to novices. In contrast, novices were activated in the left middle temporal gyrus (MTG) (BA 19, BA 39), bilateral MOG (BA 18, BA 19), left posterior lobe, right lingual gyrus (BA 18), left precuneus (BA 31), right fusiform gyrus (BA 19), bilateral inferior occipital gyrus (IOG) (BA 17). Table 2 and Figure 3 showed the results of the ALE analysis for experts and novices comparisons.

Handedness might affect an individual’s skills and athletic performance (Raymond et al., 1996). Some studies also have found that handedness was associated with cognitive function (Buckingham et al., 2011; Somers et al., 2015). The results of an EEG study found that left-handed athletes had greater P300 wave amplitude than right-handed athletes, suggesting that left-handed athletes had an advantage in cognitive perception and observation, but slower processing than the right-handed athletes (Nakamoto and Mori, 2008). Almost all of the subjects in the existing fMRI literature of motor decision-making were right-handed since there were fewer left-handed athletes. Among the 10 articles included in this study, only one article did not report whether the subjects were right-handed or not (Wimshurst et al., 2016), and the remaining nine articles reported that the subjects were right-handed. Therefore, we considered the articles included in this study to be general.

Through systematic review and ALE meta-analysis of the included papers investigating brain activation in decision-making tasks by experts and novices, we found that experts activated the right ITG, right sub-gyral, and right MOG, but novices activated more brain regions including the left middle temporal gyrus (MTG) (BA 19, BA 39), bilateral MOG (BA 18, BA 19), left posterior lobe, right lingual gyrus (BA 18), left supramarginal gyrus (BA 40), right fusiform gyrus (BA 19), bilateral inferior occipital gyrus (IOG) (BA 17). Results of this ALE meta-analysis showed significant clusters in novices, and they were located in the bilateral occipital lobe (including the MTG, MOG, IOG, lingual gyrus, fusiform gyrus, and precuneus), left posterior cerebellar lobe, and left MTG compared to experts. This also confirmed that long-term motor training could improve the neural efficiency of motor experts, reflecting the automated process of motor skills.

There was substantial evidence that the occipital lobe played an important role in the processing of visuospatial information (Todorov and Sousa, 2017; Buening and Brown, 2018). Motor cognitive psychologists argued that decision-making was based on information, of which visual information was particularly important (Darren et al., 2016). The results of the present study showed that the occipital lobe had in novices compared to experts, suggesting that the brain required more visual information processing when performing decision-making tasks. The results also corroborated previous research (Jie, 2014; Filgueiras et al., 2017; Guo Z. et al., 2017; Blazhenets et al., 2021). When the brain was stimulated by complex motor scenarios, motor experts first transferred the stimulus information into the brain’s perceptual system. The brain recognized the information and matched it with tactical information extracted from long-term memory and then made a final decision based on the learned motor skills (Smits et al., 2014). In contrast, novices performed the same task needing to find usable information and invalid information, which led to more visual information systems being activated to process and analyze information. Therefore, we believed that the primary mechanism for decreased activation in the expert’s occipital lobe was that experts had higher neural efficiency, which implied task-specific brain function plus sparing (Neubauer and Fink, 2009; Ludyga et al., 2016). This suggested that visual information processing played a key role in motor decision-making behavior.

Compared with experts, the results of ALE analysis showed that novices activated the left posterior cerebellar lobe during the decision-making process. The cerebellar was more involved in motor regulating and motor learning (Ashida et al., 2019; Schmahmann et al., 2019). Several studies have indicated that the activation of the cerebellum during performing decision-making tasks could enhance decision-making planning, initiation, and control (Yarrow et al., 2009; Nowrangi et al., 2014; Kim et al., 2015). Previous studies have found that long-term professional motor training could alter cerebellar activation. For example, a study comparing brain activation in elite archers and non-archers during archery found that non-archers had more cerebellar activation than elite archers (Chang et al., 2011), which was consistent with the findings of Guo Z. et al. (2017). Authors indicated that experts through years of sports skill learning and training might develop precise and professional sports skills, which included the ability to rapidly regulate motor information and motor learning, whereas novices did not have the ability. Imamizu et al. (2000) suggested that motor could be accurately controlled by using internal models of the body and that after the cerebellum acquired internal models through long-term motor training, athletes were able to perform tasks in a relatively automated, energy-conserving processing mode. Based on this view we believed that the reduced activation of the expert cerebellum was due to the formation of internal models of the cerebellum.

The MTG (BA 39) was already known to be significantly involved in cognitive functions such as motor planning and information processing (Guo L. et al., 2017; Xu et al., 2019). The results of this study found that the left MTG was activated when novices made decision-making in sports, which reflected the fact that experts did not need to activate more cognitive areas of the brain. Motor psychologists have suggested that cognitive mechanisms might be the main reason for the different decision-making levels between experts and novices (Dew et al., 2009). Experts with extensive motor experience could facilitate sports memory and attention, especially the ability to increase the depth of attention and reduce the waning of attentional information in complex motor scenarios. However, when novices made motor decisions in face of unfamiliar motor conditions, activation of the left MTG might be due to further processing of motor behavior information which reconfirmed the neural areas involved in motor information processing (Woods et al., 2014). Therefore, we argued that long-term motor training in athletes led to the more efficient cognitive neural network to adapt to the demands of high intensity and high correctness, which induced the brain to automatically process information. This was consistent with previous research findings (Babiloni et al., 2010; Zhang et al., 2019; Dobersek and Husselman, 2021).

Potential limitations in this study should be mentioned. ALE analysis inevitably ignored the variation in each study and belonged to the statistical inference of fixed effects. Unlike a meta-analysis with complete activation maps, the data used in ALE were based on the reported data peak activation coordinates (Zhang et al., 2016). Therefore, it was unable to take into account studies without any significant categorical reporting, which might lead to systematic overestimation of biased results. Another limitation of this article was the gap between the designed motor situation and the real situation in included studies, which might make the findings not necessarily a true reflection of the actual movement, but rather an exploration of brain activation for decision-making tasks in a laboratory setting. Further development of realistic experimental designs or fMRI techniques could facilitate the direct exploration of decision-making in real motor situations, which was a direction of future research. Last, it was difficult to assess whether these activation regions were related to sports types due to the lack of comparisons between different motor experts, so further imaging studies were needed to compare brain activation alterations of different professional motor experts.