Effects of fatigue on interception decisions in soccer

Abstract

In competitive soccer, match-induced fatigue is a common phenomenon that may negatively influence performance. Focusing on decision-making, the current study experimentally tested the impact of fatigue on players’ interception decisions and, in doing so, also took into account potential effects on physical capacity and motivation. Using a counterbalanced pretest–posttest design, 30 well-trained amateur soccer players were tested in a fatigued state vs. a non-fatigued control state (i.e. after performing a simulated soccer match and after watching soccer on television). Before and after both protocols, players performed a sprint test and an interception test in which they were instructed to either intercept or not intercept passes of different speeds. Dependent variables included physical capacity (i.e. maximal sprint capacity), motivation to intercept, decisions to intercept, and success rate of interceptions. Results showed that motivation to intercept but not sprint capacity was negatively affected by fatigue. Despite changes in motivation, fatigue did not significantly affect the number of interception attempts or the success rate of interceptions. In conclusion, findings suggest that match-induced fatigue reduces players’ self-reported motivation to engage in effortful actions (i.e. interception attempts) but does not necessarily affect objective (maximal) physical capacity and decision-making.

Keywords:

• physical exertion
• task execution
• motivation
• physical capacity
• football

Fatigue is a psycho-physiological state that may arise from prolonged task execution. During soccer matches, prolonged high-intensity intermittent exercise is likely to lead to fatigue and may negatively influence performance. Previous research in soccer indicates that running performance (Mohr, Krustrup, & Bangsbo, 2005) and the execution of technical skills (Rampinini et al., 2008; Rampinini, Impellizzeri, Castagna, Coutts, & Wisloff, 2009) are negatively affected by fatigue. The current study investigated whether fatigue may also influence players’ decision-making (i.e. action selection).

Although decision-making in sports has been extensively investigated (see e.g. Bar-Eli, Plessner, & Raab, 2011), it remains unclear whether and how fatigue influences soccer-specific decision-making. Previous research did show that fatigued players still recognise the best action in a given tactical situation (e.g. McMorris & Graydon, 1996a, 1996b; McMorris et al., 1999), but it remains unclear to what extent players would also choose to execute these actions when they are fatigued (Iodice et al., 2017). Taking players’ physical capacity and motivation to execute effortful actions into account, the current study aimed to provide more insight into the effects of fatigue on the soccer-specific decision-making.

Following Enoka and Duchateau (2016), fatigue comprises both (i) a decline in performance that relates to a reduced capacity of involved muscles and the central nervous system to produce work and (ii) changes in sensations associated with performance that reduce individuals’ willingness to spend effort (e.g. feelings of fatigue, increased perception of effort). Regarding the first aspect, it may be conceived that fatigue negatively influences soccer players’ physical capacity (i.e. their physiological capacity to perform soccer-specific actions). Indeed, several studies have shown that fatiguing exercise may cause energy storages in muscles to become depleted (Mohr et al., 2005; Reilly, Drust, & Clarke, 2008) and may reduce the number of muscle fibres that can be recruited (Bangsbo, 1994). Although it is unclear whether such physiological changes directly influence players’ physical capacity, fatigue was experimentally shown to reduce muscle strength (Rahnama, Reilly, Lees, & Graham-Smith, 2003) and maximal sprint capacity (Rampinini et al., 2011).

Regarding the second aspect of Enoka and Duchateau’s conceptualisation of fatigue, it may be conceived that also changes in sensations that are associated with performance (e.g. feelings of fatigue and increased perception of effort) may negatively influence performance. Although this has not widely been investigated in soccer, empirical evidence suggests that anticipated effort is perceived to be higher with fatigue (Iodice et al., 2017; Marcora & Staiano, 2010; Wright, 2008). This increases perceived costs and may cause individuals to be less motivated to engage or exert further effort in to-be-performed actions (Hockey, 2013; Kurzban, Duckworth, Kable, & Myers, 2013; van der Linden, Frese, & Meijman, 2003).

Based on the above, changes in physical capacity and motivation to invest effort may both be regarded as potential mediators of the relation between fatigue and decision-making in soccer. First, physical capacity determines the feasibility of actions (Araujo, Davids, & Hristovski, 2006; Smits, Pepping, & Hettinga, 2014). When fatigue reduces players’ physical capacity, specific actions may become less feasible, which is likely to affect players’ decision-making (Kahneman & Tversky, 1979; McMorris & Graydon, 1997; Renfree, Martin, Micklewright, & Gibson, 2014). For example, if a fatigued player is no longer able to run fast enough to prevent a ball from crossing the sideline, the player may decide not to run at all. Second, when fatigue increases anticipated effort, this increases the subjective cost of to-be-performed actions and reduces motivation to engage in these actions (Hockey, 2013; Iodice et al., 2017; Noakes, 2012; Smith, Marcora, & Coutts, 2015; Tucker, 2009). Consequently, players may choose to (a) invest less effort in task execution or (b) select actions that require less effort. For example, fatigued players may choose to reduce their running speed during an interception attempt or decide to hold their position rather than attempting to run and intercept an opponent’s pass.

Acknowledging that fatigue-induced changes in physical capacity and effort perception affect the feasibility and subjective costs of to-be-performed actions, it seems logical that testing the effects of fatigue on decision-making in soccer should require players to actually perform selected actions. Previous studies, however, have often failed to do so. In the 1990s, a series of experiments by McMorris and colleagues generally showed that high levels of exercise (which are likely to induce fatigue) resulted in no effect on the quality of decision-making in soccer (McMorris & Graydon, 1996a, 1996b; McMorris et al., 1999). In these studies, players were shown static tactical game-play situations and had to recognise and verbally indicate the proper action for that situation. Crucially, these experiments showed that after fatiguing exercise, soccer players were still able to recognise the best action in a given tactical situation. However, because players did not have to execute the indicated actions, potential consequences of reduced physical capacity (decreasing the feasibility of an action) and reduced motivation to invest effort (following from an increase in anticipated costs) were not incorporated, and it remains unclear to what extent fatigue influences players’ decisions to actually engage in these actions.

Against this background, the current study aimed to investigate the effects of fatigue on decision-making in soccer by using an experimental decision-making task that required actual task execution. To increase ecological validity, fatigue was manipulated by using a well-validated soccer match simulation (Russell, Rees, Benton, & Kingsley, 2011). Decision-making was tested sport-specifically by means of an interception task, which required players to intercept (high effort) or not intercept (low effort) passes that were projected with different speeds. Whilst not testing a full mediation model, underlying insight in the effects of fatigue on players’ physical capacity and motivation to intercept was captured by means of a functional sprint test and by asking players to report their motivation to intercept before each interception trial. Based on Enoka and Duchateau (2016), we hypothesised that fatigue would negatively affect players’ sprint capacity (Hypothesis 1) and their motivation to intercept (Hypothesis 2). Finally, following our argumentation around the feasibility (Araujo et al., 2006; Smits et al., 2014) and subjective cost (Hockey, 2013; Iodice et al., 2017; Noakes, 2012; Smith et al., 2015; Tucker, 2009) of to-be-performed actions, we hypothesised that if either of these effects would occur, players’ decision-making would also be affected. In the context of the current interception task, this may be reflected by a reduction in the number of interception attempts (i.e. decisions to intercept; Hypothesis 3a) and/or the success rate of interceptions (i.e. quality of their decisions; Hypothesis 3b).

Methods

Participants

Thirty well-trained and experienced competitive amateur soccer players participated in our experiment. To keep a coherent sample of comparable level, only male soccer players were included. They were on average 20.3 ± 3.3 years, weighted 72.8 ± 8.1 kg, and had a BMI of 22.4 ± 2.0 kg/m². All participants trained at least twice a week and played competition on an interregional level. All participants were informed about the purpose of the study (i.e. the effect of fatigue on decision-making) and signed an informed consent statement. Ethical approval was granted by the Ethics Committee Faculty of Social Sciences of the Radboud University (ECSW2014-2411-264).

Experimental design and setup

To isolate effects of fatigue from potential effects of repeated assessment (and the passing of time), each player was measured on two separate days in either a control condition or a fatigue condition, the order of which was counterbalanced between participants (within-subject design). In both conditions, players performed a pretest and a posttest (see Figure 1), which both consisted of an interception test and a sprint test. Before measurements were started, players were made familiar with the tests and performed a 15-min warming up.

Figure 1. Overview of the measurements in this study.

The experiment was performed outdoors on a FIFA Quality artificial-turf soccer pitch that is used for official club competitions and took place on players’ regular training evenings on weekdays between 18.00 and 23.00 h. Each player was tested individually, and in all cases, the measurements were conducted by the same experimenters. Measurements lasted approximately 2.5 h per condition, with 48 h of scheduled rest in between conditions.

Fatigue protocols

In the fatigue condition, players performed a high-intensity intermittent running exercise. This exercise was based on a previously validated soccer match simulation that resembles activities during actual soccer matches (Russell, Rees, et al., 2011). In previous studies, the type and intensity of these activities resulted in elevated blood lactate concentrations (between 6 and 11 mmol L⁻¹), reductions in blood glucose, and a rate of perceived exertion of about 17 (i.e. “very hard”) (Russell, Benton, & Kingsley, 2011; Russell, Rees, et al., 2011). The fatigue protocol consisted of 2 × 47 min of exercise with 13 min rest during halftime. Both halves were divided into five blocks. These blocks consisted of 7 min of exercise divided into four cycles that each consisted of 3 × 20 m walking, 1 × 20 m sprinting (100% of maximal running intensity), 3 × 16 m jogging (60%), 3 × 20 m jogging (60%), 1 × 20 m backwards jogging (60%), 1 × 16 m striding (85%), 1 × 20 m striding (85%), 1 × 20 m moving sideward (60%), and short passive recovery. In addition to these cycles, each block (except for the last block of each half) contained 1.5 min of passing/shooting and 1.5 min of passive recovery.

In the control condition, players watched a soccer match for 2 × 47 min. To prevent large changes in players’ emotions, a relatively unimportant soccer match of a foreign competition from a previous season was shown. After watching this match, players performed a 13-min warming up before commencing the posttest.

Interception test

In line with previous literature on decision-making in sports, participants performed a forced decision-making task (e.g. Raab & Johnson, 2004; Savelsbergh, Williams, Van der Kamp, & Ward, 2002), in which they were required to either intercept (i.e. high effort option) or not intercept (i.e. low effort option) passes that were provided at different speeds. This interception task was specifically designed for the purpose of the current experiment, included a clear moment of decision-making (i.e. to intercept or not intercept), and reflected the potentially large implications that interception decisions have during actual soccer matches (e.g. setting up a counter attack or – in case of failure – giving your direct opponent an open field). An overview of the set-up of the interception task is shown in Figure 2.

Figure 2. Experimental set-up for the interception task.

The interception test consisted of 15 trials. A ball release system (G1 ballshooter, Lobta Sports) was used to shoot straight ground passes with different pre-set velocities, which varied around a ball speed that an average experienced player – running 5.8 m s⁻¹ – was just able to intercept (based on a pilot study with players of a similar level to our sample). The selected velocities (i.e. 10.76, 11.39, 12.01, 12.64, and 13.26 m s⁻¹; order randomised between trials) successfully manipulated the feasibility of interceptions.¹ On each trial, players were instructed to intercept the pass if they perceived that interception was possible before the ball reached cone A (Figure 2). When players perceived that they were able to intercept a specific pass, they had to sprint towards the ball, receive the ball before it reached cone A, and take it through an optical gate (which was placed 1.5 m behind the line of interception) to successfully complete their interception attempt. In case players perceived that they were not able to intercept a specific pass, they had to sprint to cone B (Figure 2) and take a defensive position in order to catch an imaginary winger (who would receive the pass) on his way to the goal. Players were not allowed to change their decisions on the fly and could only go one way (i.e. either to cone A or to cone B). After attempting to intercept or taking a defensive position, players had 10 s to slowly jog back to their starting position for their next trial.

Sprint test

Before and after each interception test, players performed two maximal sprints to assess their maximal sprint capacity. To maintain functional equivalence with the interception test, the same experimental setup was used and sprints included ball handling. Upon a starting signal, players sprinted as fast as they could towards a stationary ball that was placed next to cone A (i.e. at 15.5 m from their starting position) and – as in the interception test – took this ball with them through the optical gate (Figure 2) to complete their sprint. To make sure that players gave maximal effort (i.e. maximal sprint capacity), they were motivated by means of verbal encouragement. In addition, it was emphasised that rankings based on players’ sprint times would be provided to the coaches.

Measurements

An overview of the measurements is provided in Figure 1.

Manipulation checks

The intensity of the control and fatigue protocol was checked by measuring players’ session rate of perceived exertion (sRPE, 0–10) (Foster et al., 2001) and their average heart rate (Polar RS800). Furthermore, following the approach of Oosterholt, Maes, Van der Linden, Verbraak, and Kompier (2014), players’ rated their subjective fatigue levels on a 1–10 scale (“How fatigued do you currently feel?”, “not at all”–“very much”) immediately before and after performing the pre-test and posttest.

Sprint capacity

Sprint times during each sprint test were calculated using a pressure mat (FR security type KM103 PM3 NO, accuracy: 1 ms) that measured players’ start of the sprint and an optical gate (Omron type F3ET2018600, response time: 7.6 ms) that measured the finish of the sprint. This equipment was operated on a laptop using Python (version 2.7). For analyses, the fastest sprints before and after each interception test were retained (Chelly et al., 2010; Gelen, 2010; Rampinini et al., 2011) and subsequently averaged to get a representative impression of each players’ maximal sprint capacity during the test.

Motivation to intercept

Motivation to intercept was measured before every interception trial by means of a single-item question (“How motivated are you to intercept the next pass”, 1–10 scale: “not at all”–“very much”). Single-items measurements have shown to be a legitimate representation of one-dimensional and unambiguous constructs such as players’ feelings and experiences (Bowling, 2005; Fisher, Matthews, & Gibbons, 2016; Robins, Hendin, & Trzesniewski, 2001; van Hooff, Geurts, Kompier, & Taris, 2007). For analyses, average motivation scores (across the 15 trials) were calculated for each test (i.e. pre-test and posttest in both the control and fatigue condition). The intraclass correlations of the 15 motivation scores were fair to good and ranged between .55 and .63 (Cicchetti, 1994).

Interception decisions

The percentage of trials on which players chose to intercept (i.e. % interception attempts) and the success rate of these interceptions (i.e. % successful interceptions) were calculated for each interception test. In addition, to check whether players chose less effortful alternatives, players rated their invested effort in relation to either action on a trial-to-trial basis, by answering a single-item question (“How much effort did you invest in the previous action”, 1–10 scale: “not at all”–“very much”). For each interception test, average scores were calculated for both types of actions.

Analyses

To confirm that the fatigue protocol induced higher levels of fatigue than the control protocol, RPE scores and average heart rates of these protocols were compared using paired t-tests. Furthermore, subjective fatigue ratings were compared between the control condition and the fatigue condition using paired t-tests.

To provide insight in the effects of fatigue on sprint capacity, motivation to intercept, interception attempts, success rate of interceptions, and effort invested (Hypotheses 1–3), 2 × 2 (condition × test; i.e. control/fatigue × pretest/posttest) repeated measures (RM) ANOVAs were used. In these analyses, the condition × test interaction represented the effect of fatigue. If significant effects were found, paired t-tests were conducted to further specify these effects. Potential effects of order (i.e. control condition first vs. fatigue condition first) were tested a priori using independent t-tests, which indicated that the order of conditions (counterbalanced; see “Experimental design and setup”) did not significantly affect the influence of the control and fatigue protocol on any of the dependent variables (all p's > .05).

For all analyses, effect sizes (${\eta }_p^2$ for RM ANOVAs and Cohen’s d for paired t-tests) were calculated. All analyses were performed in SPSS 23.0 and p < .05 was considered statistically significant.

Results

Manipulation check

The sRPE scores were substantially higher for the fatigue protocol than for the control protocol (6.5 ± 1.8 vs. 0.8 ± 0.8, t(29) = 15.446, p < .001, d = 4.31) and indicated that, on average, players experienced the fatigue protocol as being close to “very heavy”. Furthermore, average heart rates were significantly higher during the fatigue protocol than during the control protocol (146.1 ± 10.0 vs. 90.2 ± 8.8 beats per minute, t(25) = 29.872, p < .001, d = 5.93). In line with these findings, subjective fatigue ratings at the posttest (i.e. after the control or fatigue protocol) – but not at the pre-test (i.e. before the control or fatigue protocol) – were significantly higher in the fatigue condition than in the control condition. This was the case before the posttest (7.4 ± 1.4 vs. 2.9 ± 2.2, t(29) = 11.319, p < .001, d = 2.44) and also after the posttest (8.6 ± 1.2 vs. 7.2 ± 1.3, t(29) = 5.114, p < .001, d = 1.09). Taken together, these results indicate that the fatigue protocol was more demanding and induced much higher fatigue levels than the control protocol.

Sprint capacity

The RM ANOVA for sprint capacity showed no significant main effect of condition $\left(F\right(1,20)=1.115,p=.30,{\eta }_p^2=.053)$, no significant main effect of test $\left(F\right(1,20)=0.038,$ $p=.848,{\eta }_p^2=.002)$, and no significant interaction $\left(F\right(1,20)=0.713,p=.41,{\eta }_p^2=.034)$ (see Table 1).

Table 1. Means (SD) of outcomes during pretest and posttest for control and fatigue condition.

	Control condition		Fatigue condition		Condition × Test interaction
	Pretest	Posttest	Pretest	Posttest	p-value	${\eta }_p^2$
Sprint times (s)^a	2.71 (0.18)	2.68 (0.19)	2.71 (0.16)	2.73 (0.17)	.41	.034
Motivation to intercept (1–10)	8.1 (0.9)	7.9 (1.1)	8.2 (1.1)	7.3 (1.7)	.03	.148
Interception attempts (%)	62.8% (17.5)	62.1% (19.8)	61.6% (22.2)	55.2% (22.6)	.18	.061
Success rate interceptions^b (%)	66.3% (19.3)	68.5% (26.2)	65.4% (26.6)	71.2% (29.0)	.55	.013
Effort invested in attempting to intercept (1–10)^b	7.7 (1.0)	7.7 (1.1)	7.9 (1.0)	8.0 (1.1)	.52	.015
Effort invested in taking defensive position (1–10)^c	6.0 (2.1)	5.9 (2.2)	6.4 (2.1)	6.2 (2.2)	.81	.003

^aDue to technical difficulties, sprint data could only be analysed for 21 participants.

^bN = 29; one player made no interception attempts during at least one test.

^cN = 25; five players took no defensive positions during at least one test.

Motivation to intercept

The RM ANOVA for motivation to intercept showed no significant main effect of condition $\left(F\right(1,29)=1.413,p=.24,{\eta }_p^2=.046)$, a significant main effect of test $\left(F\right(1,29)=15.937,$ $p<.001,{\eta }_p^2=.355)$, and a significant interaction $\left(F\right(1,29)=5.020,$ $p=.03,{\eta }_p^2=.148)$. Paired t-tests following up on this interaction showed that motivation to intercept decreased significantly from pretest to posttest in the fatigue condition (t(29) = 3.572, p = .001, d = .64) but not in the control condition (t(29) = 1.397, p = .173, d = .26). In line with these effects, the difference in motivation to intercept between the fatigue condition and the control condition approached significance on the posttest (t(29) = 1.973, p = .058, d = .42), while no difference was observed on the pretest (t(29) = 0.465, p = .646, d = .10).

Interception decisions

The RM ANOVA for interception attempts showed no significant main effect of condition $\left(F\right(1,29)=0.75,p=.39,{\eta }_p^2=.03)$, a significant main effect of test $\left(F\right(1,29)=5.178,$ $p=.03,{\eta }_p^2=.152)$, and no significant interaction $\left(F\right(1,29)=1.895,p=.18,{\eta }_p^2=.061)$. The RM ANOVA for the success rate of interceptions showed no significant main effect of condition $\left(F\right(1,28)=0.02,p=.89,{\eta }_p^2=.001)$, no significant main effect of test $\left(F\right(1,28)=1.090,$ $p=.305,{\eta }_p^2=.037)$, and no significant interaction $\left(F\right(1,28)=0.359,p=.55,{\eta }_p^2=.013)$.

Finally, the analysis of effort ratings confirmed that effort invested was consistently higher for interception attempts than for taking a defensive position (see Table 1, all p's < .001, d's = 0.95 to 1.05). RM ANOVA’s on either type of action revealed no significant main effects of condition or test and no significant interactions (all p's > .05).

Discussion

The current study aimed to provide insight in the effects of fatigue on interception decisions in soccer. Conceptually, physical capacity and motivation are important factors in soccer decision-making, as they reflect both the feasibility and anticipated costs of to-be-performed actions (Kahneman & Tversky, 1979; Marcora & Staiano, 2010; Noakes, 2012). It was hypothesised that if fatigue would negatively affect players’ maximal sprint capacity (Hypothesis 1) and motivation to intercept (Hypothesis 2), this would be accompanied by a reduction in the number of interception attempts (Hypothesis 3a) and/or reduced interception success (Hypothesis 3b). Results showed that motivation to intercept but not sprint capacity was negatively affected by fatigue. Despite significant changes in motivation, fatigue did not have a significant effect on the number of interception attempts and the success rate of interceptions.

Against Hypothesis 1, results of the current study showed that fatigue did not have a significant impact on players’ maximal sprint capacity. This finding stands in contrast with earlier studies that found slower sprint times after fatigue (Little & Williams, 2007; Rampinini et al., 2011). It should be noted, however, that in these studies players either had to perform a substantially larger number of sprints (Little & Williams, 2007) or were not explicitly motivated to do their best (Rampinini et al., 2011). In the current study, on the other hand, players performed only two sprints before and after each test (of which the fastest sprint was retained for analysis (Chelly et al., 2010; Gelen, 2010; Rampinini et al., 2011)) and motivation was increased by means of verbal encouragement and listing of sprint times (see Methods section). In this regard, the current null findings concur well with studies that show that – when individuals are sufficiently motivated – performance decreases under fatigue may be temporarily counteracted (Marcora and Staiano (2010); also see Hockey (2013); Hopstaken, van der Linden, Bakker, and Kompier (2015); Hopstaken, van der Linden, Bakker, Kompier, and Leung (2016)). Apparently, when required, fatigued soccer players are still able to perform a limited number of maximal sprints at similar speeds as in a non-fatigued state.

In contrast to players’ maximal sprint capacity, players’ motivation to intercept was significantly affected by fatigue. That is, after fatiguing soccer-specific exercise, but not after following the control protocol, players’ motivation to intercept was significantly reduced, thereby supporting Hypothesis 2. The observed decrease in motivation is in line with previous studies on mental fatigue (Boksem & Tops, 2008; Hockey, 2013) and reflects that fatigue reduces individuals’ willingness to spend effort (Hockey, 2013; Iodice et al., 2017; Noakes, 2012; Smith et al., 2015; Tucker, 2009); either to protect body homeostasis (Noakes, 2012) or to motivate the individual towards engaging in more rewarding (i.e. less costly) alternatives (Kurzban et al., 2013).

With players showing reduced motivation to intercept (and considering that intercepting a pass was generally regarded as more effortful than taking a defensive position), we hypothesised that fatigue would significantly impact players’ decision-making, such that it would lead them to engage in fewer interception attempts (Hypothesis 3a) and/or that it would negatively affect the quality of their task execution (i.e. success rate of interceptions; Hypothesis 3b). Against our hypotheses, but in line with the studies of McMorris and colleagues (McMorris & Graydon, 1996a, 1996b; McMorris et al., 1999), neither the number of interception attempts nor the success rate of interceptions was significantly affected by fatigue. A plausible explanation for this finding is that although the observed drop in motivation was significant, it was quite small in absolute terms and players’ motivation to intercept remained relatively high, also when they were fatigued (i.e. scoring an average of 7.3 on a 10-point scale; see Table 1). In other words – and given that players’ maximal sprint capacity was not significantly affected by fatigue – the subjective costs of to-be-performed interception attempts may not have increased enough to trigger a behavioural change (cf. Iodice et al., 2017).

In considering these findings, it is important to note that fatigue levels in the current study were representative of an average 90-min soccer match (Russell, Benton, et al., 2011; Russell, Rees, et al., 2011). In this respect, our observation that fatigue did not significantly impact players’ decisions to intercept (nor did it significantly impact the success rate of interceptions) holds clear practical value. A question that remains, however, is whether higher levels of fatigue (e.g. as players might experience during 30 min of extra time) would have had a significant impact on decision-making. Or, alternatively, whether the same level of fatigue would have led to significant changes in decision-making if players were required to choose between alternatives that are more extreme in terms of their associated costs (i.e. very high effort vs. very low effort (Iodice et al., 2017)). Future research, which directly assesses relations between fatigue, motivation, and decision-making, across a wider range of fatigue levels and action possibilities, is needed to investigate this further.

Although the current study contributed to decision-making research in soccer by showing that fatigue reduces motivation to spend effort, but does not necessarily impact actual engagement in effortful soccer actions, it also holds a few limitations. First, to increase ecological validity, the current decision-making test was performed on an outside soccer pitch. Although our counterbalanced design (fatigue vs. control protocol) prevented systematic differences from occurring, a general lack of control over environmental circumstances (e.g. wind, humidity), may have introduced variation in ball speed or players’ decision-making which potentially lowered the power to statistically detect actual differences. To optimise power, the current study adopted a within-subject design and included twice the number of participants as in previous studies that tested effects of the same fatigue manipulation on soccer skill execution (i.e. n = 30 vs. n = 15; cf. Russell, Benton, et al. (2011)). Note, however, that due to technical difficulties, statistical power for our analyses of sprint capacity was slightly lower than for our other dependent variables (n = 21). Finally, to increase experimental control, we used a forced-choice decision-making test (cf. Raab & Johnson, 2004; Savelsbergh et al., 2002) in which players had to choose between two options (intercept or not intercept). In real matches, however, players are likely to have a larger range of options available and can constantly correct their decisions. Future research is needed to address these potential limitations.

In conclusion, the current study suggests that soccer-specific fatigue decreases players’ motivation to engage in effortful actions but does not significantly influence their decisions to actually engage in these actions. Furthermore, and in support of motivational perspectives on the fatigue-performance relationship (e.g. Marcora & Staiano, 2010), the current study showed that when it is required, fatigued soccer players may still be able to perform a limited number of maximal sprints at similar speeds as in a non-fatigued state. Based on the current paradigm, future research may provide more insight in how physical capacity and motivation influence soccer-specific decision-making; either by inducing more extreme levels of fatigue or by testing choice-alternatives that differ more extremely in terms of their associated costs (i.e. very high effort vs. very low effort).

Acknowledgements

The authors want to thank Sven van As, Bart Oosterholt, Melanie Knufinke, and Jonas Dora for their assistance with the data collection. Finally, we want to express our gratitude to sc NEC, sv Orion, and sv CDW for their participation in this study.

Notes

1 Average speeds from release to place to intercept (20 m) was 6.16 m s⁻¹ (SD = 0.93) for slowest velocity, 7.50 m s⁻¹ (SD = 1.53) for the middle velocity, and 8.96 m s⁻¹ (SD = 1.47) for the fastest velocity. Using Repeated Measures ANOVAs, the different ball velocities were shown to affect the interception attempts $\left(F\right(2.730,79.175)=37.119,p<.001,{\eta }_p^2=.561)$ and success rate of interceptions $\left(F\right(2.470,71.642)=36.125,p<.001,{\eta }_p^2=.555)$, indicating that interception attempts and success rate of interceptions were higher for lower ball velocities. The test–retest reliability coefficient (pretest vs. posttest) for interception attempts in the non-fatigue condition was .731 (p < .001).