The Effects of Moderate Physical Exercise on Adult Cognition: A Systematic Review

Background: Physical exercise is a systematic sequence of movements executed with a predefined purpose. This muscular activity impacts not only on circulatory adaptations, but also neuronal integration with the potential to influence cognition. The aim of this review was to determine whether the literature supports the idea that physical exercise promotes cognitive benefits in healthy adults. Methods: A systematic search for relevant articles was performed according to the Preferred Reporting Items for Systematic Review and Meta-Analysis criteria using available databases (PubMed, LILACS, Scopus, Web of Science, The Cochrane Library, OpenGrey, Google Scholar and CENTRAL). The search terms included “humans” or “adults” or “cognition” or “awareness” or “cognitive dissonance” or “cognitive reserve” or “comprehension” or “consciousness” and “motor activity” or “exercise” or “physical fitness,” and not “aged” or “nervous system diseases,” with the purpose of finding associations between moderate physical exercise and cognition. A methodological quality and risk of bias unit assessed the eligibility of articles. Results: A total of 7179 articles were identified. Following review and quality assessment, three articles were identified to fulfill the inclusion criteria. An association between moderate physical exercise and cognition was observed. Improvements in cognitive parameters such as reduced simple reaction time, improved response precision and working memory were identified among the included articles. Conclusion: This systematic review found that moderate physical exercise improves cognition.


INTRODUCTION
Physical exercise consists of a systematic sequence of movements executed with a predefined purpose (American College of Sports Medicine, 2009;Bowden et al., 2011) and is described by its effect on oxygen consumption as a percentage of maximum oxygen consumption (VO 2 max) (Barstow et al., 1990;Drummond et al., 2005). The concept of physical exercise is a subcategory of physical activity (Caspersen et al., 1985). Since the seminal paper by Morris and Crawford (1958) which evaluated health in bus drivers, several papers have reported an association between regular exercise activity and a reduced risk for the development of cancer and cardiovascular disease. However, its effect on brain function seems less explored.
Neuronal integration involves multiple cognitive functions and neuronal networks, including work memory and spatial memory (Tucker and Stern, 2011;Moon et al., 2017). Both neurons and astrocytes need a constant supply of nutrients (glucose and lactate) and oxygen, and regional cerebral oxygenation may be limited during exercise (Nielsen et al., 1999;Rooks et al., 2010;Braz and Fisher, 2016) with a change in the balance of cerebraI metabolism (Avola et al., 2004;Dalsgaard and Secher, 2007;van Hall et al., 2009). In addition, specific neurotrophic growth factors seem to be released as a result of moderate physical exercise, which increases the expression of brain-derived neurotrophic factor (BDNF) and neuronal growth factor (NGF) (Tyler and Pozzo-Miller, 2003;Dietrich et al., 2008;Seifert et al., 2010;Bonini et al., 2013;Coelho et al., 2014;Hashimoto et al., 2018). These factors contribute to increased cell survival and differentiation, as well as resistance to oxidative stress (Dietrich et al., 2008;Coelho et al., 2014). Furthermore, they can modulate neuroplasticity with an effect on brain function (Erickson et al., 2013). Another possible brain mechanism mediated by physical exercise could be related to the orexin system, increasing the release of orexin and orexin b. These are neuropeptides that modulate synaptic plasticity, neurogenesis, and cognition (Chieffi et al., 2017b,c).
Physical exercise may influence an individual's cognitive ability through its integrative effect on circulation and cerebral activity (van den Berg et al., 2016;Best et al., 2017); however, the scientific literature on this topic is limited. This systematic review aims to find evidence pointing to the effects of moderate physical exercise on cognitive functions in healthy adults.

Study Design
A systematic review was conducted according to the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) guidelines (Moher et al., 2009). The review was registered in PROSPERO, the international prospective register of systematic reviews (CRD42016049663).

Participants, Interventions, and Comparators
This research followed the PICO strategy, as recommended by PRISMA group (Moher et al., 2009), to develop the search criteria and determine which relevant articles to include or exclude. Interventional studies focused on the effects of moderate physical exercise on healthy adults were included, presented in Table 1.

Systematic Review Protocol
Two evaluators (MGC and RMF) performed an independent search, evaluated the validity of the publications and extracted duplicates of documents. The studies were evaluated based on the  (Moher et al., 2009 relevance of their titles and abstracts, and those that did not meet the PICO criteria were excluded. The evaluators then read the full-text versions of the selected articles. In case of disagreement, the two authors would give a joint evaluation. If any doubt remained, a third member of the research team was consulted (NCFF). All articles that met the study requirements were included in this review. The exclusion criteria included in vitro experiments, literature reviews, laboratory animal studies, letters to the editor, case reports, opinions and guides. Studies that did not include cognitive tests were also excluded.
The contact consisted in sending a weekly email, for up to five consecutive weeks.

Risk of Bias
The same evaluators also independently analyzed the methodological quality of all articles. Randomized trials were evaluated according to the Cochrane Collaboration's Risk of Bias tool. This tool evaluates the risk of bias as "low, " "high, " or "unclear" (Higgins et al., 2011) according to the items described in Table 2.
For non-randomized studies, the Methodological Index for Non-Randomized Studies (MINORS) was used (Slim et al., 2003). This ranked the articles according to a list of 12 items, and each reviewer scored the articles according to a standard checklist where a score of 0 represented "not reported, " 1 was "reported but inadequate" and 2 was "reported and suitable" for the items described in Table 3.
The quality of the article was evaluated by the sum of the values obtained in the MINORS qualifier, with a score of 0-6 indicating very low quality, 7-12 low quality, 13-18 moderate quality and 19-24 representing high quality (Khan et al., 2015).

Description of Studies
There were a total of 7,179 citations about this topic identified in the different databases. Scopus had the most (n = 3,666), followed by Central Register of Controlled-CENTRAL (n = 2,027), Lilacs (n = 654), Web of Science (n = 430), PubMed (n = 300), Cochrane Library (n = 76), Google Scholar (n = 26) then finally OpenGrey (n = 0). Figure 1 summarizes the study selection process. From the 7,179 citations, 414 duplicates were withdrawn and 6,765 remained.
A total of 155 papers were selected according to their title and abstract, of which seven (Biddle and Ashford, 1988;Brown et al., 1995;Bue-Estes et al., 2008;Douw et al., 2014;Lee et al., 2014;Tsai et al., 2014;Olson et al., 2016) were chosen based on the eligibility criteria for full-text reading. Four articles were excluded, two (Douw et al., 2014;Lee et al., 2014) were excluded due to the absence of a control group, and the other two (Biddle and Ashford, 1988;Brown et al., 1995) due to the absence of specific cognition tests. Three articles (Bue-Estes et al., 2008;Tsai et al., 2014;Olson et al., 2016) were selected and submitted to a qualitative synthesis.

Study Selection and Characteristics
The main characteristics of the included articles are listed in Table 4. In the three studies there was an association between moderate physical exercise and cognition. While the included articles each used a different evaluation tool, they all evaluated reaction time as a common parameter to measure cognition. Reaction time is the time elapsed between a stimuli and the response. Each study showed a positive correlation between moderate physical exercise by a different parameter improvement as tested by the following methods: Automated neuropsychological assessment metrics, which evaluates the reaction time, memory, attention, concentration, thinking speed 2 | Criteria for risk assessment of bias according to "The Cochrane Collaboration's tool for assessing risk of bias (Higgins et al., 2011).

RANDOM SEQUENCE GENERATION
Criteria for judgment of "Low risk" of bias The articles that appropriately described the method of randomization Criteria for judgment of "High risk" of bias Articles that presented a methodological failure in the randomization criterion or the difficult reproducibility method Criteria for judgment of "Unclear" of bias When the articles did not describe the method of randomization

ALLOCATION CONCEALMENT
Criteria for judgment of "Low risk" of bias When the allocation sequence of samples were concealed in the randomization Criteria for judgment of "High risk" of bias When the sequence of allocation of samples were not concealed at randomization Criteria for judgment of "Unclear risk" of bias When the allocation sequences were unreported

BLINDING OF PARTICIPANTS AND RESEARCHERS
Criteria for judgment of "Low risk" of bias When the sample was blind Criteria for judgment of "High risk" of bias If the methodology could not be blinded for whatever reason (sample/appraiser) Criteria for judgment of "Unclear risk" of bias When the sample was not reported either way

BLINDING OF OUTCOME ASSESSMENT
Criteria for judgment of "Low risk" of bias When the evaluators reported that the blinding in the evaluation was effective Criteria for judgment of "High risk" of bias If the study informed the evaluators how the blinding was done Criteria for judgment of "Unclear risk" of bias When the blinding was not reported

INCOMPLETE OUTCOME DATA
Criteria for judgment of "Low risk" of bias When there was an exhaustive description of the main data Criteria for judgment of "High risk" of bias If there was a loss due to an incomplete description of the main results regardless of quantity, nature and manipulation Criteria for judgment of "Unclear risk" of bias When the results were not reported

SELECTIVE REPORTING
Criteria for judgment of "Low risk" of bias When the discussion excluded some of the results Criteria for judgment of "High risk" of bias When the article discussed the data completely Criteria for judgment of "Unclear risk" of bias When the organization of the results in the discussion were unclear and decision making; the Eriksen flanker task test, which evaluated the reaction time and precision of the response; Visuospatial attention, which assess the reaction time.

Risk of Bias
The risk of bias in randomized articles was determined based on the Cochrane Collaboration's tool (Figure 2). When evaluated, the "blinding of outcome assessment" was classified as high risk in two studies. For the topics "incomplete outcome data" and "selective reporting" the included articles were classified as low risk. Among the included articles, the article by Tsai et al. (2014) (Slim et al., 2003).

CLEARLY STATED AIM
Criteria for judgment of "0" of bias When the objectives were not expressed Criteria for judgment of "1" of bias When a clear relationship between the objectives, results and conclusion was not found Criteria for judgment of "2" of bias When there was a relationship between the objectives of the article and its results and conclusions

INCLUSION OF CONSECUTIVE PATIENTS
Criteria for judgment of "0" of bias When the objectives were not expressed Criteria for judgment of "1" of bias When only the inclusion criteria was described or when it was not clear how the sample was selected Criteria for judgment of "2" of bias When the terms led to the samples' inclusion and exclusion and were clearly described

PROSPECTIVE DATA COLLECTION
Criteria for judgment of "0" of bias When there was no information reported Criteria for judgment of "1" of bias When there were changes Criteria for judgment of "2" of bias When the data collection was already established at the beginning of the study

ENDPOINTS APPROPRIATE TO STUDY AIM
Criteria for judgment of "0" of bias When unreported Criteria for judgment of "1" of bias If the techniques were not well explained or leave doubt regarding the methodology Criteria for judgment of "2" of bias If the evaluation techniques are well explained and were already referenced and answered the purpose of the study

UNBIASED ASSESSMENT OF STUDY ENDPOINT
Criteria for judgment of "0" of bias When it was not reported Criteria for judgment of "1" of bias When, due to some limitation, the methodology was not a blind sample Criteria for judgment of "2" of bias If the existence of the double-blind analysis was described Criteria for judgment of "0" of bias When the results were not reported Criteria for judgment of "1" of bias When the loss was higher than 5% and justified by the study Criteria for judgment of "2" of bias When the loss in the sample was reported as <5% (Continued)

CONTEMPORARY GROUPS
Criteria for judgment of "0" of bias When these details were not reported Criteria for judgment of "1" of bias When one of the groups already had a training routine Criteria for judgment of "2" of bias When the groups started the training together

BASELINE EQUIVALENCE OF GROUPS
Criteria for judgment of "0" of bias When these details were not reported Criteria for judgment of "1" of bias When the groups did not have the same demographic characteristics Criteria for judgment of "2" of bias When the groups had similar demographics

ADEQUATE STATISTICAL ANALYSES
Criteria for judgment of "0" of bias When the tests were either unsuitable for the type of study or not reported Criteria for judgment of "1" of bias When the descriptions of the performed tests were not clear Criteria for judgment of "2" of bias When the test was used according to the type of study and was clearly described presented the highest index. In contrast, the lowest risk of bias was observed in the article by Olson et al. (2016).
When the article by Bue-Estes et al. (2008) was evaluated ( Table 5), it was revealed to be moderate quality with a score of 13 out of a maximum of 24 points (Khan et al., 2015). For the topics "clearly stated aim, " "inclusion of consecutive patients, " "endpoints appropriate to study aim, " "adequate control group, " and "baseline equivalence of groups, " the study achieved the maximum score (2). The article did not report information for the topics "follow-up period appropriates to study aim, " "<5% lost to follow-up, " "prospective calculation of study size" and "contemporary groups, " therefore, these were scored null (0).

DISCUSSION
This systematic review investigated the effects of moderate physical exercise on cognition. All three studies included in this review (Bue-Estes et al., 2008;Tsai et al., 2014;Olson et al., 2016) concluded that physical exercise seems to improve some cognitive parameters in adults, such as simple reaction time, response accuracy and working memory.
The systematic review could be an effective tool to answer specific questions from primary studies on a subject (Cook et al., 1997), in addition to evaluating the quality of the information and the risk of bias in the articles (Sanderson et al., 2007).
In this systematic review, all studies shown an improvement of cognitive parameters related to moderate physical exercise, such as reduced simple reaction time and improved precision of response to a visual motor stimulus and working memory may be related to exercise intensity (Macpherson et al., 2011), which was measured by VO 2 max (Bue-Estes et al., 2008;Tsai et al., 2014;Olson et al., 2016). This methodological variable of physical training (intensity) was the same in the three selected articles. In the selected articles, exercise intensity was quantified by the VO 2 max test, which is considered a gold standard to quantify exercise intensity and provides a basis for ensuring that the obtained results are in accordance by the proposed training.
High-intensity exercise may depress cognitive ability (van den Berg et al., 2016), but moderate-intensity exercise improves this function when the simple reaction time is evaluated (Tsai et al., 2014). On the other hand, studies have shown that sedentary groups who do low-intensity exercise show no difference when compared to sedentary groups (Voss et al., 2011;Olson et al., 2016). The current evident indicates that exercise prescriptions to improve cognition should not be at the extremes of intensity (Voss et al., 2011;Tsai et al., 2014;Olson et al., 2016;van den Berg et al., 2016), and thus should have a linear relationship with this variable, corroborating the findings of this review.
Among the included studies, different forms of exercise performed by individuals from different age groups and both genders, showing similar results on Improvement of cognitive function. This may be related to the characteristics of the exercise performed, which was primarily aerobic in all of the analyzed studies. Quantification of exercise by VO 2 max (Koutlianos et al., 2013) can be used to individually quantify the intensity according to independent physical exercise variables such as the type of exercise, gender and age.
The cognitive improvements reported in the selected articles may be associated with the capacity of moderate exercise to modulate the central nervous system. Aerobic physical exercise is related to elevated levels of orexin-A and orexin-B (Messina et al., 2016), both neuropeptides synthesized in the hypothalamus that increase neurogenesis and connections between hippocampal neurons, especially in the dentate gyrus, an area involved in cognition (Oomen et al., 2014;Chen et al., 2015;Chieffi et al., 2017a;Trinchero et al., 2017). Orexin-A has neuroprotective and anti-apoptotic effects, and is essential for better performance in terms of attention and working memory (Deadwyler et al., 2007).
Orexin-B/hypocretin-2 (OxB/Hcrt-2) increases the expression of BDNF mRNA leading to increased production of BDNF (Chieffi et al., 2017b), a neurotrophic factor which plays an essential role in cognition (Lee et al., 2015). High levels of BDNF in the hippocampal region modulates long-term potentiation and synaptic plasticity, stimulating learning and memory (Nettiksimmons et al., 2014), in addition to increasing working memory in the prefrontal cortex (Yeom et al., 2016).
Another growth factor that is increased with physical exercise is vascular endothelial gowth factor (VEGF), which promotes angiogenesis (Morland et al., 2017) and is directly associated with neurogenesis and improvement of synaptic function. Thus, VEGF can influence cognitive function through neurogenesis, cerebral blood flow and modulation of long-term potentiation; however, the underlying mechanism has not yet been fully elucidated .
New blood vessels formed through angiogenesis are unstable (Darland and D'Amore, 1999). The presence of insulin-like growth factor 1 (IGF-1) is essential as it is responsible for the maturation and stability of these neovessels (Jacobo and Kazlauskas, 2015). It is interesting to note that studies have suggested that physical exercise contributes to stabilizing neovessels due to the associated increase in IGF-1 production FIGURE 2 | Cochrane Collaboration's tool for assessing risk of bias (adapted from Higgins and Altman).  (Carro et al., 2000;Nakamura et al., 2010;Jacobo and Kazlauskas, 2015).
As suggested by studies and animals and humans, the moderate physical exercise is related to an increase in different neutrophins, as well as to the homeostasis and energetic regulation of nervous system (Kerr et al., 2010;Rhyu et al., 2010;Swain et al., 2012;Marosi and Mattson, 2014;Bathina and Das, 2015;Jacobo and Kazlauskas, 2015;Camerino et al., 2016;Chieffi et al., 2017b,c;Morland et al., 2017). These changes contribute to an increase on plasticity, the formation of new synapses and the integration of neurons to neurons circuits (Gomez-Palacio-Schjetnan and Escobar, 2013;Lu et al., 2014;Phillips, 2017). The improvement of cognition is suggested as a result of this modulation (Colicos and Syed, 2006).
Cognition is a set of processes which functions to classify, recognize and understand information by reasoning through learning and executing responses (Vaynman et al., 2004). However, the three chosen articles only evaluated reaction time parameters, working memory and the accuracy of response. Executive function is an example of a higher cognitive process and includes capabilities such as goal setting, planning and executing action plans and effectively performing actions (Jurado and Rosselli, 2007). Cognition can be modulated throughout life under various stimuli, and varies according to educational level , nutritional quality (Moody et al., 2017), activity of antioxidants (Farah et al., 2016), and the practice of physical exercise (Atherton et al., 2016;Cadenas-Sanchez et al., 2016).
Several tests are available to the functional evaluation of cognitive parameters (Sanders and Lamers, 2002;Stins et al., 2007;Tsoi et al., 2015). Among them, the primary example of these tests are: STROOP Task (Erdodi et al., 2018), Simon Test (Stoet, 2017), Visual Spatial Attention (Tsai et al., 2016;Byun et al., 2017) and Eriksen Flanker Task (Swatridge et al., 2017;Baumgartner et al., 2018). The Visual Spatial Attention and Eriksen Flanker Task were used among the included studies. All of these tests are used to study the interaction between processing speed, executive functions and working memory, important cognitive domains, which allow us to evaluate selective attention ability and ability, as well as planning, decision making and interference management in real world (Zeischka et al., 2010;Chen et al., 2013;Diamond, 2013).
Another test used in one of the included articles (Bue-Estes et al., 2008) is the Automated Neuropsychological Assessment Metrics (ANAM). This tool is composed by computerized cognitive tests and behavioral questionnaires (Vincent et al., 2017). This tool was previoulsy described as a time-cost efficient feature and presenting a moderate sensitivity and a high specificity when compared to traditional neuropsychological tests, (Jones et al., 2008;Xie et al., 2015;Cole et al., 2017;Paech et al., 2017;Vincent et al., 2017).
In the articles included in this review (Bue-Estes et al., 2008;Tsai et al., 2014;Olson et al., 2016), reaction time as a cognitive parameter was found to be improved, as seen in Table 4. This is an important parameter that can help evaluate the speed of information processing (Batra et al., 2015). These processes were analyzed for their speed and accuracy in terms of response (van Ede et al., 2012), thus evidencing a possible association between moderate physical exercise and improvement of the synaptic network integration involved in this action.
Working memory was another altered cognitive parameter described in one of the articles, which is characterized by the ability to reorganize lists, organize thoughts to form meaningful sentences, incorporate new information, consider options and relate ideas and thoughts (Baddeley, 2012). This was only evaluated in one of the studies (Bue-Estes et al., 2008), in which the authors found an improvement in the group that performed moderate physical exercise compared to the control group and the intensity of the moderate physical exercise group, evidencing a specific effect of exercise intensity.
Cognition can be modulated throughout life under various stimuli, and varies according to educational level , nutritional quality (Moody et al., 2017), activity of antioxidants (Farah et al., 2016) and the practice of physical exercise (Atherton et al., 2016;Cadenas-Sanchez et al., 2016).
To evaluate the methodological soundness on individual studies regarding their validity and risk of bias, two tools were used: the MINORS tool, a non-randomized tool, and the Cochrane collaboration's tool, used for randomized studies. Among the two randomized studies, one was classified as low risk (Olson et al., 2016) and the other as uncertain risk (Tsai et al., 2014), taking into account the critical key domains chosen for this review.
From the MINORS tool, one of the articles was considered moderate quality (Bue-Estes et al., 2008) while the other had a high risk of bias regarding parameters related to blinding. This was not a critical parameter for this review due to its non-viability, as blinding ensures ignorance (study participants, researchers, medical staff, statisticians) regarding the allocation of participants to one group or another (Higgins et al., 2011).
It is also worth mentioning that during the recruitment of individuals, all the included studies reported that they received information about the procedures and purpose of the research (Bue-Estes et al., 2008;Tsai et al., 2014;Olson et al., 2016). Therefore, this was not one of the major domains for evaluating the quality of the articles included in this systematic review as it interferes with the reliability of this information (Higgins et al., 2011).

LIMITATIONS
Despite the moderate methodological quality and risk of bias of the included articles, moderate physical exercise appears to play a relevant role in the improvement of some cognitive parameters such as simple reaction time, response precision and working memory. However, these are just some of the existing cognitive parameters, and it is important that additional investigations with greater methodological accuracy be carried out by elucidating unexplored parameters in the selected articles.

CONCLUSION
Practicing moderate physical exercise may improve cognition in individuals who seek to improve their routinely used cognitive functions and for those who want to prevent or delay cognitive decline.

AUTHOR CONTRIBUTIONS
RF and MC contributed equally to this work. RF and MC designed the systematic review and supervised the entire program. RL, MdS, and LM reviewed all the studies and extracted the information from the eligible trials. NF and AA analyzed the data and prepared the figures and table. RF and MC wrote the paper. RL, MS, and LM revised the manuscript. All authors reviewed and approved the manuscript.