Inhibitory control mediates the effect of high intensity interval exercise on food choice

Exercise is associated with changes in food consumption and cognitive function. The aim of this study was to examine the immediate effects of acute exercise on appetite, food choices, and cognitive processes, and the mediating role of cognitive functioning, namely inhibitory control, working memory, cognitive flexibility and decision making. We compared the effects of high-intensity interval exercise (HIIE) to a resting condition on appetite and food choices, using visual analogue rating scales and a computerised portion selection task. Mediation analysis was performed with exercise/rest condition as a predictor variable and cognitive measures were entered as mediating variables and food choice measures as outcomes. Young women with low activity levels, aged between 18 and 35 years with a body mass index (BMI) between 18 and 25 kg/m 2 , were recruited. Participants (n = 30) demonstrated improved performance on a Stroop task following HIIE compared to the rest session, indicating enhanced attentional inhibition. Accuracy on an N-back task was significantly higher after HIIE, indicating an improvement in working memory and response times on the N-back task were shorter after HIIE, suggesting increased processing speed. Delay discounting for food (but not money) was reduced after HIEE but there were no significant effects on go/no-go task performance. On the trail-making task (a measure of cognitive flexibility), the time difference between trail B and A was significantly lower after HIIE, compared to rest. HIIE reduced rated enjoyment and ideal portion size selection for high energy dense foods. The relationship between exercise and food choices was mediated by inhibition as assessed by the Stoop task. These results suggest that HIIE leads to cognitive benefits and a reduced preference for high-calorie foods and that an enhancement of attentional inhibition may underlie this relationship.


Background
With the advancement of modern technology, there has been a sharp increase in the amount of people following a sedentary lifestyle and an unhealthy eating pattern (Vaynman & Gomez-Pinilla, 2006;Kao et al., 2017).Nowadays, people are less physically active, with 23-32% of the worldwide adult population being insufficiently physically active (World Health Organization, 2020).This lack of physical activity is associated with significant public health concerns and an increased risk of developing obesity and associated health conditions including cardiovascular diseases, type 2 diabetes, certain types of cancer, mental health conditions, and neurodegenerative disorders (Avgerinos et al., 2019;Beydoun et al., 2008;Di Angelantonio et al., 2016).In this context, it is important to explore the potential benefits of exercise on weight management and appetite suppression.
Acute bouts of exercise have been reported to have an immediate effect to reduce hunger (Blundell et al., 2003;King et al., 1994) and food intake post-exercise (Westerterp-Plantenga, et al., 1997).Various exercise modes, including running, cycling, resistance training, and rope-skipping, are known to suppress appetite (Lambourne & Tomporowski, 2010;King et al., 2010King et al., , 2013;;Broom et al., 2009;Becker et al., 2012;Kawano et al., 2013).This exercise induced suppression of appetite may explain, at least in part, why exercise-induced energy deficits do not appear to be fully compensated for by additional food intake on the day of exercising (Schubert et al., 2013) and why exercise can contribute to weight loss (Bellicha, et al., 2021;Ueda et al., 2009).
There is evidence that the short-term anorectic effect of exercise is related to physiological processes, including changes in the secretion of gut hormones, gastric emptying, muscle metabolism and blood flow redistribution (King et al., 2010;Stensel, 2010;Ueda et al., 2009).Another mechanism through which exercise might affect appetite and food intake is by improvement in high level cognitive processes, which underpin appetite control, such as working memory, attention, inhibitory control, and decision making (Higgs, 2016;Higgs & Spetter, 2018;Lowe et al., 2019).Significant improvements in cognitive functioning have been reported after a single bout of exercise (Chang et al., 2012;Ishihara et al., 2021;Moreau & Chou, 2019).Specifically, it has been reported that acute exercise improves performance on the Stroop task (Barella et al., 2010;Byun et al., 2014;Tsukamoto et al., 2016;Yanagisawa et al., 2010).This task requires inhibitory control, as participants must supress saying the name of a colour word, in order to correctly state the ink colour in which the word is printed (i.e.saying red when the word BLUE is printed in red).Similar improvements in inhibitory control have been reported for performance on the go-no-go task which assesses the ability to withhold a pre-potent motor response (Bailey et al., 2021).Acute bouts of exercise have also been found to improve other cognitive processes including working memory (McMorris & Hale, 2015), attention (Fernandes et al., 2019) and cognitive flexibility (Hwang et al., 2016), which may contribute to the observed reduction in food intake post-exercise by enhanced cognitive control of eating and improved ability to adhere to dietary goals.
Only two studies have investigated the effects of acute bouts of moderate/vigorous exercise in lean participants, on both cognitive performance and appetite-related measures at the same time and examined the association between these measures.Lowe et al., (2014b) found that performance on a Stroop task was better after low-moderate intensity (30-50% heart rate reserve) exercise on a cycle ergometer, compared with performance after minimal exercise (without increasing heart rate).The authors also reported that in the moderate intensity exercise condition a larger effect of exercise on Stroop performance was associated with a greater consumption of less palatable/healthier (control) foods but had no effect on appetising snack foods.In another study by the same group, moderate exercise (target heart rate of 50%) was reported to improve performance on a Stroop task and reduce consumption of highly palatable foods (Lowe et al., 2016).Moreover, the improvement in Stroop performance after exercise was a significant predictor of the mean difference in appetising snack food consumption between exercise conditions.These data suggest that the exercise-induced improvement in inhibitory control may have caused the reduction in palatable food consumption, perhaps by increasing the ability to resist consuming these foods due to improved control over food-related decision making (Lowe et al., 2019).However, further work is required to substantiate this conclusion, given the limited number of studies as well as the contradictory findings in terms of the effects of exercise on dietary choices.Whether exercise decreases the intake of highly palatable and unhealthy foods by an underlying change in inhibitory function, or an increase in the motivation to eat more healthier food options, is yet to be uncovered.
Here, we extend the work of Lowe and colleagues by examining the effect of an acute bout of exercise on food choices and by formally analysing whether exercise-induced benefits in cognition mediate this effect.We also make some methodological refinements by including multiple facets of executive functioning.Given that previous studies found robust effects of acute exercise on Stroop performance, we included this task in our cognitive battery as a measure of attentional inhibition (the ability to resist interference from distracting stimuli; Tiego et al., 2018).We also included the go/no-go task as a measure of response inhibition (the ability to suppress a prepotent motor response) (Tiego et al., 2018) as the results of a meta-analysis of 35 studies showed that greater inhibitory control was associated with reduced food intake on both objective measures of food consumption and self-reported measures (McGreen et al., 2023).In addition, we included tasks assessing other potential cognitive mediators: working memory (the N-back task) and cognitive flexibility (the trail-making task).A measure of delay discounting (which assesses preference for immediate versus delayed rewards) was included, given that exercise may reduce the rewarding value of food during decision-making (McNeil et al., 2015).
The intensity and duration of exercise may also impact appetite.High-intensity interval exercise (HIIE) and longer exercise durations tend to have a more pronounced suppressive effect on appetite compared to moderate-intensity exercise (Coats et al., 2003;Deighton et al., 2013;Martins et al., 2015;Rossi et al., 2022;Sugama et al., 2015).Previous studies have reported increased metabolic signals, insulin sensitivity, muscle building and fat loss, after HIIE in healthy individuals (Helgerud et al., 2007;Hood et al., 2011) and in people with type 2 diabetes (Hollekim-Strand et al., 2014).Improvements in different aspects of executive function were enhanced with a longer duration of HIIE, compared to continuous moderate exercise (Tsukamoto et al., 2016).Besides having larger effects on cognitive and metabolic processes, HIIE procedures are perceived as more enjoyable than moderate-intensity continuous exercise (Bartlett et al., 2011;Lambrick et al., 2016).Furthermore, HIIE is often favoured for its time efficiency; as it involves shorter bursts of intense exercise, individuals can achieve similar or even greater benefits in a shorter amount of time.
Because higher intensity exercise might be more effective than moderate exercise in altering cognition and food intake (Hu et al., 2023;Tsukamoto et al., 2016) we examined the effects of a bout of HIIE versus a rest condition.To assess appetite and food-related decision making, we used visual-analogue rating scales and to assess food choices, and we used the computerised portion-selection task of Brunstrom and Rogers (2009), which is likely to be a more sensitive measure of the effect of exercise on food choices compared to intake at buffet meals in a laboratory setting (Farah et al., 2012).Sex differences in response to exercise have been reported as well as differential responses depending on fitness levels and body mass index (Douglas et al., 2017;Farah et al., 2012;Kawano et al., 2013;Ueda et al., 2009).Evidence suggests lean women respond to exercise with more robust compensatory alterations in appetite-regulatory hormones (post-exercise hyperinsulinemia) than men, which results in a more prominent suppression of appetite (Farah et al., 2012;Marliss et al., 2000).Therefore, to reduce overall heterogeneity and to maximise statistical power, we recruited only young, lean, sedentary women.
We hypothesised that a single bout of HIIE would reduce rated appetite and selected portion size and that this effect would be mediated by an effect of exercise on inhibitory control consistent with some previous findings (Lowe et al. 2016(Lowe et al. , 2019)).Given the potential effects of other facets of executive functioning on eating behaviour, we further hypothesised that working memory, cognitive flexibility and delayed discounting might also be mediating factors.

Participants
Young women with low activity levels (no more than 75 min of vigorous or 150 min of moderate-intensity physical activity per week; World Health Organization, 2020) aged between 18 and 35 years with a body mass index (BMI) between 18 and 25 kg/m 2 , were recruited via posters and social media.They were required to be non-smoking, have no dietary restrictions, nor any significant recent weight loss and not have a current metabolic/psychiatric condition nor be taking any medications (as assessed by an in-/exclusion criteria questionnaire).The recruitment criteria were in place to ensure a homogenous sample (young lean sedentary women) and to maximise statistical power.Based on a composite effect size (0.6 Dz) from a pilot study of cognitive and food portion selection measures (n = 10), a power analysis (G*power 3.1.9.7; Faul et al., 2009) indicated that a minimum sample size of 24 participants was required to obtain 80% power (at alpha = 0.05).
Forty-five healthy lean women completed an initial screening session.A total of thirty participants passed screening and completed both test sessions.The participants were paid £50 to compensate for their time.Written consent was given prior to participation and ethical approval was provided by the Research Ethics Committee of the University of Birmingham.

Study design
Participants completed two test sessions (rest and HIIE) according to a cross-over within-subjects design.The order of conditions was counterbalanced with a minimum of one week/maximum of 3 weeks washout (to avoid any carry-over effects).

Screening session
Participants completed an in-/exclusion criteria questionnaire, that included questions on current dietary restriction and recent weight loss (Do you have any dietary restrictions (e.g.dieting/vegan/vegetarian)? Have you lost more than 5 kg in the last 3 months?), the General Health Questionnaire (GHQ-12; Goldberg, 1972), and Dutch Eating Behaviour Questionnaire (DEBQ;van Strien et al., 1986), their BMI was calculated, and habitual physical activity was assessed using the International Physical Activity Questionnaire (IPAQ; Craig et al., 2003).Participants also completed a VO₂ max test on a cycle ergometer.They started with a warm-up of 5 min, cycling at 75 Watts, after which the workload increased at a rate of 50 Watts per 2.5 min, or 25 Watts if the respiratory exchange ratio (RER) was higher than 1.The participants were asked to maintain a cadence of 60 rpm.Breath-by-breath pulmonary gas-exchange data were collected and averaged every 5 s (SentrySuite; Vyntus system).The peak VO 2 was determined as the highest 5 s mean value attained prior to exhaustion.The maximum exhaustion was determined when three of the following criteria were obtained: 1) a plateau in the VO₂, despite increasing workload, 2) an RER above 1.10, 3) a heart rate above 90% of the age-predicted maximum, and 4) pedal rate dropped below 60 rpm (Tsukamoto et al., 2016).The results of VO₂ max test were used to determine the exercise intensity (Wattage) for the HIIE protocol.

Test day procedure
Each study session had a duration of about 3 h and 15 min and took place in the morning and at the same time of the day.Before the start of the test sessions, participants were asked to fast for 12 h and to avoid drinking any alcohol or caffeine containing beverages for 24 h previously.They were also asked to avoid engaging in any structured physical activity (e.g., playing sports or training) for at least 7 days.On arrival, they were asked about their last meal, amount of sleep, and menstrual cycle.To standardise appetite, breakfast bars were consumed, based on approximately 20% of their basal metabolic rate (calculation based on Oxford equations for BMR; Henry, 2005) and participants then completed a set of questionnaires (see Supplementary Table 1 for details): DEBQ, Barratt impulsivity scale (BIS-11;Patton et al., 1995), Beck Depression Inventory (BDI; Beck et al., 1987), Intuitive Eating Scale-2 (IES-2; Tylka & Kroon Van Diest, 2013).Participants then either had a rest period, where they were allowed to read magazines (any pages with food images were removed), or they started the HIIE procedure.The cognitive tasks started about 20 min after the exercise/rest session and were presented in random order.Throughout the day, participants rated their mood and appetite using visual analogue scales (VAS; Blundell and Hill, 1986) and completed the computer-based portion selection task.Three blood samples were taken at three different time-points (see Fig. for an overview of the testday).Upon analysis (cv, sample fit to the standard curve) it was clear that the blood data were unreliable.It is impossible to say if this was due to researcher error, sample handling issues, or assay error.The present study is not weakened by the lack of presentation of these data as they do not impact on the primary outcome.

Exercise/rest conditions
The HIIE protocol, based on the study of Tsukamoto et al. (2016), started off with a warm-up of 5 min carried out at 40% of peak VO₂.After the warm-up, the HIIE protocol started with 5 min of moderate intensity exercise at 60% of peak VO₂, followed by three high intensity bouts of min at 90% of peak VO₂, with 3 min of active recovery at 60% of peak VO₂.The total protocol lasted 31 min (Tsukamoto et al., 2016).Intensity was lowered during the high-intensity bouts to an intensity above 70% of peak VO₂ if participants deemed it too difficult to keep up due to physical pain.During the rest condition, participants sat quietly for min.

Cognitive measures 2.6.1. Stroop task
The Stroop task is a measure of attentional inhibition.Stimuli consisted of either five coloured X's (neutral condition; 72 trials), a congruent colour word (12 trials), or an incongruent condition (60 trials) (Lowe et al., 2016).Participants indicated if the font colour matched the stimulus by a mouse-click and simultaneously said the font colour out loud.The stimuli remained on the screen until a response was given (maximum 2000 ms), followed by a fixation-cross of 1000 ms.Accuracy and RT were recorded, and Stroop interference scores were calculated: the RT of correct incongruent trials minus the RT of correct neutral trials.

Go/no-go task
The go/no-go paradigm used in this study is adapted from Price et al. (2016) and assessed response inhibition to food and neutral stimuli.Participants were instructed to respond to instructed 'go' stimuli by pressing the spacebar and withhold any response to 'no-go' stimuli (total of 280 trials).Stimuli were presented for 200 ms, followed by a blank screen for 500 ms and a fixation cross for 500 ms.In the neutral condition (80 trials), images of sports equipment were presented as the 'no-go' stimuli and toiletries as the 'go' stimuli.In the food condition (80 trials), food images were presented as 'no-go' stimuli, and neutral office supplies as 'go' stimuli.Two conditions were added with reversed instructions (2 × 60 trials), to decrease the chance of any possible practice effects.The presentation order of the conditions was counterbalanced.The measure of interest was the number of commission errors (responses incorrectly made to 'no-go' trials).

N-back task
The N-back task is a continuous performance task assessing working memory (Kirchner, 1958).A sequence of blue circles was presented on a 3 × 3 grid.Circles appeared on the screen for 500 ms and were followed by a fixation cross for 1500 ms.Participants had to identify whether the current location of the blue circle matched location of the circle that appeared N trials ago (either 2-or 3-trials back to vary task difficulty).They completed 70 trials of each N-back condition (2-back and 3-back), with a short break in between.Working memory performance, reaction time (RT) and accuracy, were compared between conditions.

Trail making task
The trail making task is a measure of cognitive flexibility (Lezak, 2004;Corrigan et al., 1987).Participants had to first connect consecutively numbered circles (trail A) and then connect an alternating sequence of numbered and lettered circles (trail B).The time difference between trail B and A was calculated as a measure of flexibility.This task was administered on paper and timings were measured by a stopwatch.

Delay discounting task
The delay discounting task assesses preference for immediate versus delayed rewards (Koffarnus et al., 2013;Reynolds, 2006).In the monetary version there were nine delays ranging from one day to one year.For example, participants were asked if they preferred '£50 now' or '£100 in one week'.A similar paradigm was used for food items, in which participants had to choose between a smaller amount of food now, or a larger amount later (between 1 h and one day).Participants could choose from 9 different palatable food items (Cadbury chocolate eggs, sprinkled donuts, chocolate bars, ice-cream, sausages, chocolate cake, Twix chocolate bars, and pizza).Difference in the total area-under-the-curve (AUC) was compared between conditions.To calculate the AUC (Myerson et al., 2001)

Computer-based portion selection task
A modified version of the Standard Computerised Portion Selection Task was used, adapted from Brunstrom and Rogers (2009) and Wilkinson et al. (2012).A set of 3 tasks assessed ideal portion size (1), enjoyment (2), and familiarity (3).In each task, 8 food items with low energy densities (LED; on average 104 kcal/100 g, ranging between 52 and 180 kcal per 100 g; e.g.tropical fruit salad, bananas, eggs, potatoes, Caesar salad, fish & tomato salad, and couscous salad) and 8 food items with high energy densities (HED; on average 501 kcal/100 g, ranging between 344 and 595 kcal; e.g.M&M's, Cadbury milk chocolates, Pringles, cashew nuts, KitKats, Haribo Starmix candy, ready salted crisps) were presented.Participants indicated their preferred portion size by using the arrow keys on the keyboard.The first picture presented a 20-kcal portion and for the subsequent pictures, the portion was increased by 20 kcal.Nine participants were familiar with all food items except for the fish and tomato salad.All other participants reported familiarity with all foods.

Data analysis
Statistical analysis was conducted using SPSS version 29 (IBM Corp, Armonk, NY).Outliers in the data were identified as larger than three times the standard deviations above the mean.Outlier data were handled by Winsorizing, a widely used method for transforming outliers, by setting the extreme values to the value at the 95th percentile (Kotz et al., 2005).Task specific data were removed if participants scored less than the chance finding of 50% accuracy on any cognitive task (except the N-back task, where a cut-off point of 30% was used; Jaeggi et al., 2010).
Effects of exercise on the cognitive and food choice outcomes were analysed using repeated-measures ANOVA or paired t-test with Bonferroni corrected post-hoc tests where appropriate.Partial eta-squared was reported for effect sizes.Mediation analysis was performed using MEMORE for SPSS (Montoya & Hayes, 2017;IBM Statistics).We tested the significance of this indirect effect using bootstrapping procedures.Unstandardised indirect effects were computed for each of 5000 bootstrapped samples.The 95% confidence interval was computed by determining the indirect effects at the 2.5th and 97.5th percentiles.Alpha level was set at 0.05 as the significance criterion.

Participants
Participants demonstrated sedentary behaviour; they spent on average 7.5 ± 2.4 (Mean +_ SD; standard deviation) hours a day sitting, they walked an average of 23.8 ± 19.4 min per day (in the last week) and scored a mean VO 2 max of 30.7 ± 5.7 (see Supplemental Table 1 for details of participants characteristics).Participants took part during random times of their menstrual cycle.The luteal and follicular phases were evenly distributed over both conditions (Supplementary Fig. 1), with no significant differences found between HIIE and rest conditions (X 2 (3, N = 60) = 1.57, p = 0.667).During the exercise sessions, participants had an average energy expenditure of 195 ± 23 kcal.During the HIIE procedure, 15% of the participants could not fully cope with the high-intensity bouts, and the high-intensity bouts were lowered by 10-15 Watts.
On the trail making task, the time difference between trail B and A was significantly lower after HIIE, compared to rest (t (29) = 2.440, p = 0.021 see Table 1).

Mediation analysis
Exercise/rest condition was entered as a predictor variable and the cognitive measures were entered separately as mediating variables and food choice measures as outcomes.We restricted the analyses to the HED outcome at T3, given that T3 is the only food choice measure occurring after cognitive functioning was assessed.There was a significant indirect effect of exercise on the ideal portion size through the Stroop interference score (β = − 27.82, BootSE = 19.01,CI = − 74.0 to − 1.23).This result indicates that participants had lower preferred ideal portion sizes after exercise, compared to rest, through the process of inhibition (lower Stroop interference scores).The direct effect of exercise condition on portion sizes was not significant (β = − 33.51, t = 1.16, p = 0.256, CI = − 92.74 to 25.71).
No significant mediating effects were found for any other cognitive measure.Furthermore, there was no significant mediation for enjoyment ratings, through any cognitive measure.See Fig. 3 and Table 2 for  summary of the mediation results.

Discussion
We found an increase in attentional inhibitory control (reduced Stroop interference) following HIIE, as well as an improvement in working memory, cognitive flexibility and delayed discounting for food (but not money).HIIE also reduced rated enjoyment of and ideal portion size selection for high energy dense foods.The effect of exercise on ideal portion size selection was mediated by increased attentional inhibitory control (as assessed by the Stroop task).
Acute exercise has been previously reported to affect food decision making, characterised by diminished hunger and selection of smaller ideal portion sizes (Farah et al., 2012).Similarly, we found that exercise reduced ideal portion size and liking of high calorie food items.However, we did not observe an effect of exercise on appetite ratings, which differs from the findings of Farah et al. (2012).The reason for this discrepancy is unclear but there are several differences between the studies including testing in the fasted versus fed state and the type of exercise (moderate intensity treadmill running versus HIIE cycling).An exercise induced reduction in preference for high fat vs. low fat foods in the fed state and in the absence of any changes in appetite ratings has been reported previously (McNeil et al., 2015).Thus, it is possible that exercise reduced ideal portion size for high calorie foods due to a reduced preference for high energy dense foods that is unrelated to perceived levels of hunger/fullness.
To investigate the processes underlying the effects of exercise on food choices we examined the potential mediating effect of several cognitive processes that are known to underpin appetite control (Higgs & Spetter, 2018).As reported previously (Lowe et al, 2014), we identified a mediating role for enhanced inhibitory control (manifested as reduced Stroop interference).Reduced inhibitory control is associated with overeating (Guerrieri et al., 2007a,b;Batterink et al., 2010) and heightened responsiveness to food cues (Jasinska et al., 2012), whereas high levels of inhibitory control have been associated with reduced consumption of highly palatable foods (Hall, 2012).Therefore, one potential reason why exercise was associated with lower portion selection of high energy dense foods is that participants were better able to  inhibit the desire for these foods, instead acting more in line with healthy eating goals (Hofmann et al., 2009).Given that we identified attentional inhibitory control as a specific mediator it may be participants were able to ignore food-related cues that would usually trigger appetitive responses (Davidson et al., 2019).In the context of the present study, an exercise-induced enhancement of attentional inhibitory control specifically may have reduced high (but not low) calorie portion size selection due to a reduced influence of appetitive food cues on decisions about how much would be desirable to consume alongside enhanced dietary self-regulation (McGreen et al., 2023).
A plausible neurobiological mechanism linking exercise with inhibitory control and portion selection is an exercise-induced increased in blood flow/neurotransmitter release in the dorsolateral prefrontal cortex (DLPFC; Yanagisawa et al., 2010).Activity in the DLPFC has been associated with a range of cognitive tasks involving inhibitory control (Langenecker & Nielson, 2003;Sylvester et al., 2003) and to food choice/dietary control.For example, experimental attenuation of DLPFC excitability using repetitive transcranial magnetic stimulation has been found to decrease palatable snack intake (Lowe et al., 2018) and DLPFC activity is implicated in successful self-initiated dietary regulation (Hare et al., 2009).The release of serotonin in prefrontal brain regions is linked to both inhibitory control (Cools et al., 2008) and appetite (Dourish, 1995), and acute exercise has been reported to increase plasma serotonin and improve reaction times on a Stroop task (Zimmer et al., 2016).Moreover, individuals with the largest enhancements in response inhibition showed the greatest increases in serotonin concentrations (Zimmer et al., 2016).These data suggest that increased prefrontal serotonin activity could underlie the effect of exercise on both inhibitory control and food choices.
Although exercise improved other aspects of cognitive performance (working memory, cognitive flexibility, delay discounting) these processes were not significant mediators, suggesting that attentional inhibitory control specifically mediates the effect of exercise on portion size selection.Further work is required to examine whether exerciseinduced improvements in working memory, cognitive flexibility, and delay discounting contribute to aspects of food-related decision making not assessed in this study.For example, a positive correlation between working memory and fruit/vegetable intake has been reported (Allom & Mullan, 2014).Hence, future studies could test whether exercise also may promote healthier food choices via improvements in working memory.
Comparing the results of the present study with the broader literature on the effects of exercise on cognition we note that a systematic review of 24 studies (Ai et al., 2021) on the effects of HIIE on a range of cognitive processes found the largest positive effects on inhibition, but none on cognitive flexibility or working memory.However, as there were limited studies assessing cognitive flexibility and working memory, the results should be considered with caution.Here, we found positive effects of HIIE on Stroop interference, working memory, cognitive flexibility and delay discounting suggesting a general effect of exercise to improve higher cognitive functions.We did not observe an effect on go-no-go responding.The Stroop and go/no-go tasks assess different aspects of inhibitory control and so the results might suggest a specific effect of exercise on attentional (Stroop) rather than response (go/no-go) inhibition.However, the total number of errors committed for the go-no task was very low suggesting that floor effects may also explain the lack of effect of exercise on response inhibition.
Strengths of the present work include that it was well powered and included a range of cognitive and food-related measures.However, the results should be considered in the light of some limitations.We restricted our recruitment to young, lean, sedentary women and so further work is required to assess whether the results would generalise to men and to other age groups with different levels of fitness and body weight.We also note as a limitation that we did not control for menstrual cycle stage.However, we found that luteal and follicular phases were evenly distributed over both conditions with no significant differences found between HIIE and rest conditions suggesting that this was not a systematic confound.Further work is also required to investigate whether similar effects would be seen after exercise of more moderate intensity and for different modes of exercise, but based on the suggestion that enhanced activity in DLPFC may underlie the effects we would speculate that any mode or intensity of activity that has this outcome would be effective.We assessed hypothetical portion size selection using a computerised task that has been found to predict actual portion selection (Wilkinson et al., 2012) but further work could also examine whether the results hold when participants make actual portion size selections/food choices.
The finding that an acute bout of exercise improves inhibitory control and that this results in choice of smaller portions of high calorie foods suggests that there may be undercompensation for calories expended during exercise over time, which could be beneficial for individuals who are seeking to achieve a negative energy balance.In other words, in addition to positive effects on brain health, exercise induced improvement in cognition may also explain why exercise-induced energy deficits do not appear to be fully compensated for by additional food intake on the day of exercising.However, the results of studies of the longer-term effects of exercise on food choices suggest that delayed partial compensation can occur, although this is highly variable across individuals and could be affected by changes in body weight (Blundell et al., 2015;Whybrow et al., 2008).Greater understanding of the mechanisms underlying the effects of acute and chronic physical activity on appetite and how this is influenced by individual characteristics is required to be able to predict when and for whom exercise might result in weight loss and the potential role of cognition.

Conclusions*
The results suggest that acute bouts of HIIE reduce preference for high-calorie foods in young, sedentary lean women and offer cognitive benefits.Improvement in attentional inhibition after exercise may contribute to a greater ability to resist impulsive food choices in the short term.Building on our findings future studies might focus on the chronic effects of regular exercise on cognitive performance and everyday dietary decisions.
, delays and indifference points are first normalized, expressed as a proportion of the maximum value.The AUC is then computed by summing the results of a specific equation for each pair of successive delays and indifference points: x2 − x1 [(y1 + y2)/2].Here, x1 and x2 represent successive delays, and y1 and y2 denote the indifference points associated with those delays.The AUC can range from 1 (indicating no discounting) to 0 (reflecting maximum discounting).Larger AUC values indicate less discounting by delay, implying lower impulsivity or, conversely, greater self-control.

Fig. 1 .
Fig. 1.Timeline of the test days.An overview of the test days in minutes.

Fig. 2 .
Fig. 2. Results of the computer-based portion selection task.Ideal portion sizes (A) and enjoyment (B) for HED and LED food items are presented at three different timepoints (T1: after breakfast and before the HIIE/rest session; T2: immediately after the HIIE/rest session; T3: 1.5 h after the HIIE/rest session), for both conditions (rest and HIIE).* indicates p-value <0.05.

Table 1
a Results of the trail making task, delay discounting task, Stroop task, N-back task, and go/no-go task.Mean ± standard deviation.* indicates p-value <0.05.

Table 2
Mediation analysis of exercise condition, cognition, and food choice measures (ideal portion sizes and enjoyment ratings).Separate analyses were done for the trail making task, delay discounting task, Stroop task, N-back task, and go/no-go task.Significant results are presented in bold (p-value <0.05).