Evaluation of cognitive load in team sports: literature review

In team sports, load management has become one of the most common areas of investigation, given that effective control over load is the key to being able to optimize performance and avoid injuries. Despite the constant evolution and innovation in the latest theories, we can see a clear tendency in load management that focuses on physiological and mechanical aspects and neglects its cognitive character, which generates the variability inherent in the performance of athletes in a changing environment. Indicators of response that inform methods of control over cognitive load can include cognitive, physiological and behavioral indicators. However, limited investigations exist to support the reliability of each indicator regarding cognitive load. For this reason, the objective of this literature review is to present strategies used to manage cognitive load in team sports, as well as the indicators utilized for such a proposition and their relationships in specific contexts.


INTRODUCTION
Team sports have a dynamic and complex nature (Lord et al., 2020). The players perform intermittent, high-intensity activities such as repeated sprints, jumps, impacts, fights, changes of direction, accelerations and decelerations during training and competition (Paulauskas et al., 2019). In this regard, cognitive requirements are also considered crucial for optimal performance. These short, intense actions generally last less than 3 s, with moderate and long recovery periods (Ben Abdelkrim et al., 2007), and should be integrated into training strategies to promote adaptations that efficiently optimize athletic performance (Schelling & Torres-Ronda, 2013) through interaction with the training load.
Traditionally, planning for training and competition has centered on the management of physiological and mechanical aspects, defined as internal load (IL) and external load (EL), respectively. IL and EL are related, the first being defined as the individual psychophysiological response and the second as the external physical stimulus applied to the athlete during training or competition (Soligard et al., 2016). This process can be assessed by the control and management of certain indicators that could offer information about the load effect on the athletes (Soligard et al., 2016). In this context, however, the other elements that can affect the success of the athlete's training are neglected; for example, the quantification of the cognitive effort (Mujika et al., 2018), defined as the volitional assignment of resources in order to respond to the demands imposed by a task. Depending on the degree of effort involved, cognitive resources can be exhausted which can cause mental fatigue, a psychobiological state provoked by prolonged periods of demanding cognitive activity (Job & Dalziel, 2001). Furthermore, Van Cutsem, Marcora & De Pauw (2017) and Brown, Graham & Innes (2020) conclude in their respective systematic reviews that cognitive effort causes a negative effect on physical performance. Therefore, we consider that the cognitive load will be the relative amount of available mental resources that are invested in the resolution of a task. Cognitive load can be varied by changing the complexity of the assigned task but increasing its complexity will not involve a real increase of the individual's load if they do not employ enough effort to solve it (Cárdenas, Conde-González & Perales, 2015). For this reason, various authors describe how cognitive load is closely related to the emotional state of the athlete (Camacho et al., 2020;Cárdenas et al., 2013;Cárdenas, Conde-González & Perales, 2015;Rottenberg & Gross, 2003;Vaughan, Laborde & McConville, 2018). The regulation of these emotions involves processes through which individuals influence their emotions. These processes require effort, thereby draining cognitive resources (Schmeichel, 2007).
Team athletes perform in a high-entropy environment that causes them to constantly expend mental resources to respond to the demands of the tasks, which requires significant mental effort (Cárdenas, Conde-González & Perales, 2015). Furthermore, deliberate constant evaluation of the possible alternatives during a game can consume the resources of the system and promote fatigue, impairing performance (Marcora, Staiano & Manning, 2009). This cognitive demand combined to the effects of the internal and external load, can cause a maladaptation or rejection of the suggested load. This will directly influence whether the athlete is physically and/or mentally prepared for exposure to another training stimulus, known as the readiness to train/compete (Gabbett et al., 2017).
Cognitive load, together with its emotional dimension (Alarcón et al., 2018), requires structured planning and management that is complimentary with that of the physical load, providing information about how the planned sessions are being received. Other authors, such as Camacho et al. (2020), Cárdenas, Conde-González & Perales (2015), Gabbett, Jenkins & Abernethy (2010) and García-Calvo et al. (2019) agree with this reasoning. The main reason is that mismanagement of this load generates not only short-term effects, such as loss of performance (Sansone et al., 2020), reduction of technical abilities (Gantois et al., 2019) or loss of control over the real impact of the session (McLaren et al., 2016a); but also, long-term effects, such overtraining or burnout (Goodger et al., 2007).
Including this parameter in training plans will be key to balancing the readiness of the athletes, achieving optimal performance and preventing injuries. The team sports environment requires intense and constant cognitive activity (Figueira et al., 2019), employing various brain mechanisms to adapt to a changing environment (Coutinho et al., 2017). In fact, there is a need to include specific, suitable and objective information for successful planning (Gabbett, Jenkins & Abernethy, 2010), as well as the control of the cognitive load of training (Gonçalves et al., 2016) and to define properly the variables involved on this process (Soligard et al., 2016). To this end, the current review aims to show, in a practical way, the strategies for the management of cognitive load in team sports, as well as the indicators used to monitor the load and the ways they are related in specific contexts.

SURVEY METHODOLOGY Information sources
A search of the literature between 1970 and 2020 was performed using the following online databases: Medline (PubMed), SPORTDiscus, PsycINFO and Google Scholar system. The following keywords were used: team sports, cognitive load, mental load, psychological load, workload, mental fatigue, cognitive fatigue, psychological fatigue, together with Boolean operators such as "AND" and "OR". Furthermore, this literature review was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Stewart et al., 2015).

Criteria of inclusion of the study
The titles and abstracts of all the articles were analyzed to determine the relevance of the publications for inclusion. Selection criteria for the articles were followed. The complete text of the publication was obtained to determine if it met the criteria for inclusion. In addition, the bibliographies of the selected articles were analyzed to find other relevant articles. Finally, for the current review, only those studies that are centered on the management of cognitive, mental or psychological load (in relation to the workload) and/ or the ones that analyzed cognitive, mental or psychological fatigue (the effect on work and performance) in the specific context of team sports, were included. Therefore, those articles that try to quantify the individual cost of mental resources, given some capabilities, while achieving a given level of performance in a task with specific demands were included.

Criteria of exclusion of the study
Individual sports and studies in situations far from sports reality were excluded, as well as analyses in laboratory situations. Duplicate articles were eliminated and summaries, non-peer reviewed works, book chapters and opinion articles were also excluded. As a first step, the titles, abstracts and key words screening of the literature was carried out by the authors. In the second step, full-text articles of the relevant studies were screened while in the third step, the reference lists of the suitable articles and the review articles on the management of cognitive, mental or psychological load and/or fatigue were searched for additional articles. Any disagreement was discussed until consensus was reached.

RESULTS
The initial literature search found 28,851 articles related to cognitive load, but only 1,947 were selected based on their title and abstract. After determining the content of the complete text, 1,926 articles were excluded for being unrelated or incompatible with the inclusion criteria, or both (Fig. 1). A total of 22 studies were included, which described the management of cognitive load in different situations. These, following the classification of indicators described by Capdevila (2001), were classified depending on the type of indicator of response analyzed, discriminating between cognitive (Table 1), a set of subjective indicators collected in a sporting situation, usually from written tests such as questionnaires, tests or self-reports; physiological (Table 2), indicators obtained with the aid of apparatus allowing the analysis of physiological or biochemical parameters; and behavioral (Table 3), indicators of the analysis of the observable behavior of the athlete, usually external motor behavior or verbal behavior.

DISCUSSION
The authors use different methods to manage cognitive load, analyze specific and applicable information, exercise better control over training and optimize performance.  To do so, they try to quantify, through indicators, the individual cost of mental resources, given certain capacities, while reaching a certain level of performance in a task with specific demands. Cognitive load management methods can be used before, during or after task performance. These cognitive load management methods are based on the analysis of these response indicators, usually classified as cognitive, physiological and behavioral indicators (Capdevila, 2001), so this review will address all three types.

NASA-TLX
The NASA-TLX questionnaire has proven sensitive to the mental load in a variety of cognitively demanding tasks, such as piloting an aircraft or laboratory tasks (Luque-Casado et al., 2016). This scale analyses the subjective perceived work according to six dimensions: mental demand (perceptive and mental effort), physical demand (degree of physical activity), temporal demand (perceived pressure related to decision-making speed), effort (the combination of the mental and physical effort needed) and frustration (the negative emotions perceived). This offers an overall score of the workload (from 0 to 100 points, a.u.) based on the average of the six dimensions. In all studies analysed, the NASA-TLX questionnaire was significantly sensitive to changes in the cognitive load in different team sports situations. In tasks that involve mental fatigue, significant differences are shown between the control and experimental groups (Alarcón, Ureña & Cárdenas, 2017;Alarcón et al., 2018). In the study by Camacho et al. (2020) performance on training tasks under temporal and quantity-of-technical-movement restrictions was analysed and the most specific management tool for the increased cognitive load was the NASA-TLX. Likewise, García-Calvo et al. (2019) also used this tool to identify the increase of the cognitive load as a function of the modification of the scoring system. Measurement Period: After the session/competition.

mm VAS (MF and ME)
The VAS scale (Visual Analogue Scale) offers specific information about a characteristic or attitude which is identified along a continuum of values and cannot easily be directly measured (Gould et al., 2001). It has a unidimensional format, charted in a 100 mm straight line, with limits identified as the perceived minimum (0) and maximum (100). The subjects only need to make a mark on the line that indicates their relative perception of their current situation. The distance will be measured in millimeters (from left to right) and this will be the subjective reference value. This tool has been used frequently in the bibliography to describe the fatigue and lack of energy caused by a cognitively demanding activity (mental fatigue, MF) and/or the degree of effort employed to perform a cognitively demanding activity (mental effort, ME). It has been proved that the scale is a valid, reliable method to measure both MF and ME (Lee, Hicks & Nino-Murcia, 1991).
The first investigations demonstrated that the increase of perceived indicator VAS MF damaged specific technical performance (Smith et al., 2016;Badin et al., 2016). In Badin's study (2016), effects of VAS MF on physical abilities were not found, in contrast to subsequent studies which related VAS MF to the loss of both technical and physical abilities (Veness et al., 2017;Coutinho et al., 2017;Coutinho et al., 2017). Sansone et al.
(2020) adds that within this loss of abilities, offensive tasks cause more significant VAS MF and ME than defensive tasks. Differing a bit from what has been previously described, Kunrath et al. (2020) shows that VAS MF reduces technical, tactical and cognitive abilities, provoking a compensatory improvement of the performance.
Measurement Period: After the session/competition.

CSAI-2
CSAI-2 (Competitive State Anxiety Inventory 2) is a test that aims to estimate the cognitive and somatic anxiety levels of the players, as well as their levels of self-confidence. Cognitive anxiety refers to the negative feelings that the subject has about his or her performance and the consequences of the outcome. Somatic anxiety, on the other hand, refers to the perception of physiological indicators of anxiety such as muscle tension, increased heart rate, sweating and stomach discomfort. It has a questionnaire format consisting of 27 items, with nine items for each subscale. Each of them is classified on a four-point Likert-type scale, which yields scores from nine to 36 in each subscale. A higher score related to cognitive and somatic anxiety indicates a higher level of anxiety (Martens et al., 1990). Arruda et al. (2017) study suggests that cognitive anxiety increases as a function of the level of the opponent, causing more psychological stress. Furthermore, it highlights that cognitive anxiety increases exponentially depending on the rival, more than somatic anxiety. This information could offer an approximation to the concept of cognitive load, dissociating the somatic state from the cognitive one.
Measurement Period: After the session/competition.

POMS
The POMS (Profile of Mood States) questionnaire was proposed by McNair, Losr & Droppleman (1971) with the aim of evaluating the mental fatigue related to physical effort. The questionnaire is classified on a Likert scale and it contains 65 items that provide measures of six specific mood states: tension, depression, anger, vigor, fatigue and confusion. These factors can be combined to create a compound measurement of mood state by adding the five negative factors and subtracting the positive factor of vigor. Further, a 100-point baseline score is added to prevent negative scores (Raglin, Morgan & O'Connor, 1991).
It has been proved that POMS is a reliable and valid questionnaire to measure affective features, mood state and emotions (Lin, Hsiao & Wang, 2014). That is why Mashiko et al. (2004) use it as a test of mood state in order to evaluate mental fatigue. In their study they could observe significant changes between different game positions in rugby games, obtaining important differences between the relation of mental and physical fatigue.
Measurement Period: After the session/competition.

Cognitive RPE
The RPE (rating of perceived exertion) is a valid method for quantification of the effort expended in an athletic training activity during a wide variety of exercise types (Foster et al., 2001). Depending on the question the athlete is asked, distinctive subjective variables can be measured. These differential ratings of perceived exertion (dRPE) can provide additional information to that obtained by a single measurement (Gil-Rey, Lezaun & Los Arcos, 2015). In this review, we focus on cognitive RPE (RPE-T in the classification of dRPE), which is the one that answers the question "How much mental effort and decision-making has this task required?" To quantify it, two scales are used, CR-100, also known as "centiMax", (Borg & Borg, 2001), scored from 0 to 100 and CR-10 (Borg, Hassmen. & Langerstorm, 1985), varying from no exertion (0) to maximum effort, such as a competition (10). The important difference between them is that the CR-10 uses whole numbers corresponding to verbal anchors and the CR-100 does not (Fanchini et al., 2016).
In Farrow, Pyne & Gabbett (2008) study it was shown that cognitive RPE (CR-10) was sensitive to tasks that involved increased decision-making in game-like situations. Having followed this line of research and found results that agree with the mentioned study, Gabbett, Jenkins & Abernethy (2010) justify the use of cognitive RPE as valid for control of cognitive load, but since it is a subjective measurement it would require more objective methods to be compared and measured. In order to relate the subjective data obtained to some objective data, McLaren et al. (2016b) finds strong associations between RPE-T multiplied by the session time (sRPE-T) and the tasks focused on speed and specific abilities, the latter being open tasks focused on the game. This RPE-T increases significantly when the game is against the top teams of the league (Barrett et al., 2018). Therefore, we can say that according to the results of these four studies, cognitive RPE is sensitive to open tasks that seek to replicate the competitive game, creating an uncertain environment.

SIATE/VSC
SIATE (Sistema Integral de Analisis de Tareas de Entrenamiento, from its Spanish initials) and VSC (Valoración Subjetiva de la Carga, from its Spanish initials) are ecologic systems for quantifying the load of a basketball training session (Vallés, Fernández-Ozcorta & Suero, 2017;Reina et al., 2019). They are intended to provide a holistic view of training, taking into account more than just internal and external load. Ibáñez, Feu & Cañadas (2016), in order to quantify training using direct observations, designed a methodological system to register and subsequently analyze different factors that are relevant to the athletic process. For this purpose, they created an observational survey known as Integrated System of Training Task Analysis (SIATE). This survey records six variables: degree of opposition, density of task, percentage of simultaneous participants, competitive load, space of play and cognitive involvement. The variables are scored from one (minimum load) to five (maximum load). The sum of the scores provides a measurement of the total load of the task.
Coque (2009) suggests a tool that, similarly to the previous one, aims to evaluate the training load using a direct observation. The analysis system is the Subjective Evaluation of the technical-tactical training load (VSC), in which six variables are recorded: degree of obstacle, density of the task, percentage of simultaneous executions, competitive load, field of play and cognitive involvement. The variables are scored from one (minimum load) to four (maximum load). The sum of the scores provides a measurement of the total load of the task.
Both of them can be used to calculate the total load of the task, which would be the total of the assessed values, and the load weighted for the length of the task, which would be calculated by multiplying the total load by the useful time of the task in minutes. The latter shows more precision in the real load of the task. Reina et al. (2019) shows that the SIATE organic system gives us the same information as that recorded by inertial devices or HR monitors. Also, Vallés, Fernández-Ozcorta & Suero (2017) show that VSC has a strong correlation with RPE. Studies that analyze correlations between individual variables to see the percentage of the variability as a function of the cognitive load have not been found.
Measurement Period: After the session/competition.

PHYSIOLOGICAL INDICATORS
Heart rate variability (HRV) Heart Rate Variability is defined as the temporal variation of the heart rate during a specified time period (Capdevila & Niñerola, 2006). In the simplest way, HRV has been analysed within the time domain, but more complex evaluations include an analysis within the frequency domain and nonlinear methods (García Manso, 2013).
The Autonomic Nervous System (ANS) is responsible for the regulation of the HRV through parasympathetic and sympathetic modulation (Bricout, DeChenaud & Favre, 2010), the balance of which is disrupted after changes in the training load (Pichot, Busso & Roche, 2002). This is the reason why many studies define HRV analysis as a useful, non-invasive method to evaluate the function of the ANS (Bellenger et al., 2016;Bosquet, Merkari & Arvisais, 2008;Hynynen et al., 2006;Parrado et al., 2010). The findings in the Thayer et al. (2009) study suggest an important relationship between cognitive performance and HRV, reaffirming the relevance of this information to measure the effect of cognitive load.
Even though the literature in control situations suggests that HRV is very sensitive to demands on the cognitive load (Luque-Casado et al., 2016), no significant differences have been found in HRV comparing HRV control situations and mental overload (Gantois et al., 2019).
Measurement period: Before the session.

Reaction time or response time
Reaction time is calculated as the time from the beginning of the stimulus until the corresponding response of the participant (Gabbett, Kelly & Sheppard, 2009). This indicator depends mainly on, but may be affected by age, gender or duration of the stimulus (Der & Deary, 2006), on cognitive processes, which means that mental fatigue could be inhibitory (Huijgen et al., 2015) and could increase the response time. It should be noted, however, that beyond a certain duration and/or intensity of exercise, muscle fatigue induces an increase in reaction time which may be due to a decrease in cognitive performance. fatigue induces an increase in reaction time that may be due to a decrease in cognitive performance (Chmura, Nazar & Kaciuba-Uscilko, 1994).
In Scanlan et al. (2013) response time is identified as the only variable that predicts the time of reactive agility in the phased model. This serves us to determine the influence of physical and cognitive factors on the development of reactive agility in basketball players. Gantois et al. (2019) found a significant increase in reaction time in relation to increased mental fatigue induced by the Stroop task. This increased reaction time impairs the decision making of soccer players and thus their performance. Coinciding with these results, Moreira et al. (2018) saw an increase in reaction time caused by mental fatigue in a Stroop task. Reaction time underwent a significant decrease during the test, except for the last 5 min in which it was maintained Measurement period: During a task previous to training/competition.

Decision-making time
Making a successful decision depends on the ability of the player to identify, select and then execute the correct action in response to the postural signs of opponents or teammates, recognizing significant patterns in the game and determining the situational probabilities (Roca et al., 2013;Williams et al., 2011). Decision-making time is determined as the time interval between the first identifiable contact of the stimulus-player and the first identifiable contact that initiates the participant's response (Gabbett, Kelly & Sheppard, 2009). This cognitive ability to make fast and precise decisions is fundamental for success in football (Smith et al., 2016), an observation that can be applied to the rest of the team sports.
In Scanlan et al. (2013) decision-making time is identified as a variable which is highly associated with agility time. This cognitive measurement, together with reaction time, was the one that influenced the basketball players the most. Gantois et al. (2019) showed that induced mental fatigue damaged the decision-making process, provoking a decrease in performance. These findings agree with those of Smith et al. (2016), determining that mental fatigue affected the decision-making precision and time of the football players.
Measurement period: During the task.

CONCLUSIONS Relation with internal and external load
Cognitive factors interact with physiological and mechanical factors which are present during training and competition. For this reason, distinguishing their connections can be a significant element in the research of load management in sport (Soligard et al., 2016).
In this context, the loads that the athletes assume involve stress and provoke changes in their physical and psychological well-being. Furthermore, understanding the interaction between these loads, perceptive well-being and readiness for training or competition will provide us with significant individual training prescriptions (Gabbett et al., 2017). The IL measured through RPE is significantly connected to variations in cognitive load imposed by the training exercises, as evaluated by cognitive indicators such as the NASA-TLX questionnaire (Camacho et al., 2020) or the VAS MF scale (Badin et al., 2016;Sansone et al., 2020;Veness et al., 2017). This increase in cognitive fatigue has a direct impact on the internal load, provoking a modulation of the endocrine response attenuating the concentration of testosterone and alpha-amylase, markers of the activity of the mesolimbic pathways and the sympathetic nervous system, respectively (Moreira et al., 2018). There were no studies showing significant correlations between the internal load taken from the HR and the cognitive load taken from the NASA-TLX questionnaire (Badin et al., 2016), except for the study conducted by Farrow, Pyne & Gabbett (2008) which showed increases in the RPE-Cog and the HR during open exercises.
Moreover, the EL recorded through tracking systems shows that players under high cognitive load conditions (VAS MF) register higher values for total distance, average speed and run time at a moderate speed, demonstrating higher intensity efforts (Coutinho et al., 2017;Kunrath et al., 2020). This condition of mental fatigue damages aspects of the tactical behavior of the players, causing a compensatory increase in the physical performance (Kunrath et al., 2020). Position variables, taken from GPS analysis, are also affected in high cognitive load conditions, provoking a decrease of the longitudinal synchronization (coordination of longitudinal movements of the players during the game), of the stretch index (represented by the average distances of each player from the geometric gravity centers of the team) and of the distance between dyads (distance between a pair of players that share the same environment and intend to reach the same team objective) (Coutinho et al., 2018). This deterioration of physical performance (physical variables) and elapsed synchronization time (position variables) should alert us that cognitive load should be considered a variable that can be controlled with the objective of improving collective behavior.
The models of training load control that take into account cognitive load such as the CSE or VSC can offer the same information as more difficult methods (Reina et al., 2019;Vallés, Fernández-Ozcorta & Suero, 2017). However, we must take into account that they are both subjective recording systems.

Practical applications
To optimize performance and exert more control over training processes and/or competition in team sports, it is necessary to have load control strategies which include cognitive load in the monitoring cycles of the athlete. However, there is limited literature related to this concept.
NASA-TLX, VAS MF or Cognitive-RPE analysis strategies are considered valid to measure the impact of the cognitive load in team sports. Other control methods like VFC, reaction time or decision-making time need scientific evidence. In this sense, there is a clear lack of studies which use objective tools to measure the cognitive load.
The evaluation of external, internal and cognitive demands in an isolated way would be potentially problematic, because we would only obtain information on the provided stimulus, the processed stimulus or the players responses without examining the inherent connections. Even though it is accepted that external, internal and cognitive demands are separate constructs, they must be analyzed and interpreted in the same context, and they must be considered as a whole in order to optimize performance and prevent injuries. Processing systems will produce an alteration in the stimulus that generates physiological responses, so the manipulation of one of these three constructs will provoke changes in the others. With this objective, future studies that delve into this paradigm will be justified.

ADDITIONAL INFORMATION AND DECLARATIONS Funding
This study was funded by the Ministry of Science and Innovation (Ministerio de Ciencia e Innovación) of the Spanish government (Grant Number: PID2019-107473RB-C21). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.