Exercise characteristics
During the 60 min of Football, average heart rate was 151 ± 16 beats.min− 1, maximum heart rate was 186 ± 13 beats.min− 1 and relative exercise intensity was 75 ± 8% (Table 2). Furthermore, average (t (32) = − 2.8, p = .009) and maximum heart rate (t (32) = 2.7, p = .010), as well as relative exercise intensity (t (32) = − 4.1, p < .001), were lower in the high-fit group compared to low-fit (Table 2). Average heart rate across the whole 4 h exercise trial (105 ± 13 beats.min− 1) was higher than during the whole 4 h resting trial (84 ± 11 beats.min− 1, t (35) = 11.8, p < .001).
Table 2 Average and maximum heart rate, relative exercise intensity and GPS characteristics for the group overall, as well as the high- and low-fitness splits, during the 60 min Football session. Data are mean ± SD During the Football session, the average total distance covered was 2788 ± 432 m. Of this, 2129 ± 278 m was covered at low speed (< 9 km·h− 1), 477 ± 120 m at moderate speed (9–13 km·h− 1) and 177 ± 84 m at high speed (> 13 km·h− 1). All GPS variables during the 60 min football session were similar between the high- and low-fit groups (Total Distance; p = .959. Low Speed Distance; p = .542. Moderate Speed Distance; p = .710. High Speed Distance; p = .305, Table 2).
Cognitive function
Data for both cognitive tests, across all time points, for the exercise and control trials are displayed in Table 3. Given that there were no differences at baseline between the exercise and resting trials (all p > .05) and for ease of interpretation, the figures are displayed as change from baseline. An overview of the results of the statistical analyses is displayed in supplementary Table 1.
Table 3 Cognitive function data across the exercise and control trials, for the high- and low-fitness groups, as well as the group overall. Data are mean ± SD Stroop test
Response times
Congruent level
Overall response times were quicker in the high-fit group compared to the low-fit group (main effect of fitness: high-fit; 719 ± 134 ms, low-fit; 794 ± 164 ms; F(1, 5304) = 130.2, p < .001). Overall response times were similar between the exercise and resting trial (main effect of trial; p = .363) and became quicker across the course of the day (main effect of time; F(3, 5304) = 18.5, p < .001). The pattern of change between the exercise and resting trial was similar (trial by time interaction: p = .373). The pattern of change across the exercise and resting trials was, however, different between the high- and low-fit groups (trial by time by fitness interaction; F(3, 5304) = 4.8, p = .002; Fig. 1). A separate ANOVA revealed a difference in the pattern of change for the low-fit participants (trial by time interaction; F(3, 2619) = 4.10, p = .007, Fig. 1a). Specifically, response times were quicker 45 min following seated rest compared to 45 min post-exercise (p = .045). However, in the high-fit participants, response times were similar across the morning between the exercise and resting trials (trial by time interaction; p = .231; Fig. 1b).
Incongruent Level
Overall response times were quicker in the high-fit group compared to low-fit (main effect of fitness; high-fit; 960 ± 209 ms, low-fit; 1084 ± 243 ms; F(1, 10,668) = 317.1, p < .001). Overall response times were similar between the exercise and resting trial (main effect of trial; p = .994) and became quicker across the course of the day (main effect of time; F(3. 10,668) = 22.4, p < .001). The pattern of change was similar between the exercise and resting trial (trial by time interaction: p = .204), as was the pattern of change between the high- and low-fit groups (trial by time by fitness interaction; p = .099).
Accuracy
Congruent Level
Overall accuracy was similar between the high- and low-fit groups (main effect of fitness; p = .316). Accuracy was also similar between the exercise and resting trial (main effect of trial; p = .324) and across the day (main effect of time; p = .409). The pattern of change across the day was similar between the exercise and resting trial (trial by time interaction; p = .428), as was the pattern of change between the high- and low-fit groups (trial by time by fitness interaction; p = .425). Incongruent Level
Overall accuracy was similar between the high- and low-fit groups (main effect of fitness; p = .317). Accuracy was also similar between the exercise and resting trial (main effect of trial; p = .317) and across the day (main effect of time; p = .410). The pattern of change across the day was similar between the exercise and resting trial (trial by time interaction; p = .410), as was the pattern of change between the high- and low-fit group (trial by time by fitness interaction; p = .413).
Sternberg paradigm
Response times
One-item
Overall response times were quicker in the high-fit group compared to their low-fit counterparts (main effect of fitness; high-fit; 496 ± 91 ms, low-fit; 529 ± 124 ms, F(1, 4372) = 44.1, p < .001). However, response times were similar between the exercise and resting trials (main effect of trial; p = .639) but became quicker across the course of the day (main effect of time; F(3, 4372) = 11.8, p < .001). The pattern of change in response times was different between the exercise and resting trial (trial by time interaction: F(3, 4372) = 9.2, p < .001) and furthermore, the pattern of change between the exercise and resting trial was different between the high- and low-fit participants (trial by time by fitness interaction; F(3, 4372) = 4.2, p = .006). A separate ANOVA revealed a difference in the pattern of change for high-fit participants (trial by time interaction; F(3, 2225) = 3.0, p = .030, Fig. 2d). Specifically, response times were quicker 45 min post-exercise, when compared to 45 min seated rest (p = .022). A separate ANOVA also revealed a difference in the pattern of change for low-fit participants (trial by time interaction; F(3, 2147) = 9.7, p < .001, Fig. 2a). Specifically, low-fit participants were quicker 90 min following seated rest, when compared to 90 min post-exercise (p = .020).
Three-item
Overall response times were quicker in the high-fit group compared to low-fit (main effect of fitness; high-fit: 628 ± 115 ms, low-fit: 703 ± 160 ms, F(1, 8725) = 184.4, p < .001). However, response times were similar between the exercise and resting trials (main effect of trial; p = .327) but became quicker across the course of the day (main effect of time; F(3, 8725) = 7.5, p < .001). The pattern of change was different between the exercise and resting trials (trial by time interaction: F(3, 8725) = 2.7, p = .042), as was the pattern of change across the day between the high- and low-fit groups (trial by time by fitness interaction; F(3, 8725) = 3.9, p = .009). A separate ANOVA revealed a similar pattern of change between the exercise and resting trial for the low-fit participants (trial by time interaction; p = .390, Fig. 2b), yet there was a difference in the pattern of change for the high-fit participants (trial by time interaction; F(3, 4388) = 6.5, p = < .001, Fig. 2e). Specifically, response times were quicker 90 min following seated rest, when compared to 90 min post-exercise (p < .001).
Five-item
Overall response times were also quicker in the high-fit group compared to low-fit on the five item level of the Sternberg paradigm (main effect of fitness; high-fit: 761 ± 151 ms, low-fit: 834 ± 207 ms, F(1, 8236) = 99.8, p < .001). Overall response times were quicker in the control trial compared to the exercise trial (main effect of trial; exercise: 803 ± 168 ms, control: 791 ± 200 ms, F(1, 8236) = 4.0, p = .046) and became quicker over the course of the day (main effect of time; F(3, 8236) = 27.1, p < .001). The pattern of change was different between the exercise and resting trial (trial by time interaction: F(3, 8236) = 5.6, p < .001) and furthermore, the pattern of change across the day was different between the high- and low-fit groups (trial by time by fitness interaction; F(3, 8236) = 4.6, p = .003). A separate ANOVA revealed a difference in the pattern of change for the low-fit participants (trial by time interaction; F(3, 4041) = 3.4, p = .018, Fig. 2c). Specifically, response times were slower immediately post-exercise compared to immediately after seated rest (p = .012) and slower 90 min post-exercise compared to 90 min following seated rest (p = .033). A separate ANOVA also revealed a difference in the pattern of change for the high-fit participants (trial by time interaction; F(3, 4195) = 7.7, p < .001, Fig. 2f). Specifically, response times were quicker 45 min following seated rest compared to 45 min post-exercise (p = .003).
Accuracy
One item
Overall accuracy was similar between the high- and low-fit groups (main effect of fitness; p = .314), similar between the exercise and resting control trials (main effect of trial; p = .314). and similar across the course of the day (main effect of time; p = .398). The pattern of change across the day was similar between the exercise and resting trial (trial by time interaction; p = .396) as was the pattern of change between the high- and low-fit participants (trial by time by fitness interaction; p = .399).
Three-item
Overall accuracy was similar between the high- and low-fit groups (main effect of fitness; p = .315), similar between the exercise and resting trials (main effect of trial; p = .317) and similar across the course of the day (main effect of time; p = .398). The pattern of change across the day was similar between the exercise and resting trial (trial by time interaction; p = .394), as was the pattern of change between the high- and low-fit groups (trial by time by fitness interaction; p = .390).
Five-item
Overall accuracy was similar between the high- and low-fit groups (main effect of fitness; p = .321), similar between the exercise and resting trials (main effect of trial; p = .316) and similar across the course of the day (main effect of time; p = .404). The pattern of change across the day was similar between the exercise and resting trial (trial by time interaction; p = .400), as was the pattern of change between the high- and low-fit groups (trial by time by fitness interaction; p = .412).
Brain derived Neurotrophic factor
Serum BDNF concentration was similar between the high- and low-fit groups (main effect of fitness; high-fit: 27.1 ± 6.8 ng·ml− 1, low-fit: 29.1 ± 7.8 ng·ml− 1, p = .210). Serum BDNF concentrations were also similar between the exercise and resting trial (main effect of trial; p = .082) and were also similar across the course of the day (main effect of time; p = .085). The pattern of change was similar between the exercise and resting trial (trial by time interaction; p = .167), as was the pattern of change between the high- and low-fit groups (trial by time by fitness interaction; p = .704, Table 4).
Table 4 Serum BDNF concentrations (ng·ml− 1) across the course of the resting and exercise trials, for the high- and low-fitness groups, as well as the group overall. Data are mean ± SD