Introduction

Anxiety is an emotional state1 characterised by excessive and persistent worry with no clear direction to the worry2. Common symptoms also include irritability, somatic tension, and insomnia3. Anxiety is one of the most common psychological problems among college students, and studies have shown that the detection rate of anxiety symptoms among college students has risen from 16.6 to 34.8% over the past 10 years4, and has shown an increasing trend in recent years5. The detection rate of anxiety among Chinese college students was 24.9%6. Personal emotional balance is challenged in the face of gradually increasing academic pressure7. College students are gradually becoming a high prevalence of anxiety, and there is an urgent need to explore effective ways to alleviate anxiety in order to protect their normal life, learning and personality development.

Anxiety symptoms are modulated by a variety of neurobiological systems, in which the behavioural activation system (BAS) and the behavioral inhibition system (BIS) play key roles8. the BAS is sensitive to rewarding, non-punishing, and punishment-avoiding stimuli, as convergence to incentive signals, which is responsible for positive emotional experiences. The BIS is sensitive to punishment, non-rewarding, and novelty stimuli and manifests itself as withdrawal to avoidance of error signals. It as a behavioral inhibition system of the individual’s behavioral responses, and is responsible for negative emotional experiences9,10. Thus, abnormal sensitivity to BIS and BAS can lead to behavioural and psychopathological problems in humans, anxiety being one of them11. BIS was found to be a stable predictor of anxiety risk12, while BAS sensitivity negatively predicted anxiety to a lesser extent13. Previous studies have shown that personality traits (BIS and BAS) explained all of the genetic variance in depressive and anxiety symptoms14. In terms of neuroanatomical features, it has been shown that females displayed a negative correlation between BIS sensitivity and regional gray matter volume (rGMV) in the parahippocampal gyrus, as well as positive correlations between BAS sensitivity and rGMV in the ventromedial prefrontal cortex and inferior parietal lobule, whereas males showed the opposite pattern. These findings suggest that the brain regions associated with processing of negative emotions and reward-related information may contribute to sex-related differences in rGMV correlates of BIS and BAS, respectively15.

Physical activity can improve the behavioural activation system and anxiety symptoms16, and has the advantages of high compliance, low side effects and stable effects17,18. The underlying mechanism may be that physical activity increases cerebral blood perfusion19, upregulates neurotrophic factor levels20,21, improves the function of the hypothalamic-pituitary-adrenal axis22, and inhibits the secretion of proinflammatory cytokines23, which effectively stimulates the central nervous system, improves the BIS/BAS system24, and improves the ability of emotional homeostasis7, which in turn alleviates negative emotions, such as anxiety24,25.Evidence shows that moderate to high intensity physical activity is positively associated with BAS26, and low intensity physical activity is positively associated with BIS27. And people with high BAS are more actively driven to participate in physical activity and have more enjoyable emotional responses to exercise28. Regular exercisers have better self-control, which can increase relative left frontal cortex activity, especially at moderate intensities, and contribute to individual BAS29. On the other hand, exercise reduces individual BIS by improving individual cardiovascular capacity and tolerance, especially moderate and high intensity exercise30.

Previous studies have clarified the strong relationship between exercise and anxiety symptoms31, and it is also clear that exercise enhances the behavioural inhibition system30, and that the behavioral inhibition and activation system is an influence on anxiety symptoms13. However, previous researchers have not verified the existence of the action path of the behavioural inhibition and activation system mediating the involvement of exercise in anxiety symptoms in the college student population, not to mention clarifying which of the four sub-dimensions of reward responsiveness, drive, fun seeking and behavioural inhibition mediate the action path, and the effect of the action needs to be further clarified. Based on the foundation of previous studies and research gaps, the present study proposed the following three hypotheses: (a) There are differences in physical activity, behavioural inhibition system and activation system among college students with different anxiety symptom scores. (b) Physical activity of college students is significantly negatively correlated with anxiety and behavioural inhibition system, and significantly positively correlated with behavioural activation-reward responsiveness, drive, and fun seeking. (c) Physical activity, behavioural inhibition and activation were all predictive of anxiety symptoms in college students, and behavioural inhibition and activation mediated the effect between physical activity and anxiety.

The present study intends to adopt a cross-sectional design to test the research hypotheses and use structural equation modelling to explore the role of behavioral inhibition and activation systems of college students in mediating exercise participation and anxiety symptoms. It will provide a reference for university workers and researchers to promote the physical and mental health of university students.

Subjects and methodology

Participants

Participants in this study were current university students, and all participants volunteered to take the test and signed an informed consent form. The informed consent was obtained from all participants and/or their legal guardians. The study was approved by the university ethics committee (SISUGJ2024015), see the attachment for details. The research have been performed in accordance with the Declaration of Helsinki. We confirm that all research was performed in accordance with relevant guidelines/regulations.

The inclusion and exclusion criteria were as follows: (1) aged 18–24 years old; (2) right-handedness; (3) no history of psychiatric illness and no use of psychotropic medication; (4) no chronic physical illness, no external brain damage, and no visual abnormalities.

Sample size calculation and sampling process: College students were recruited from seven colleges and universities in Songjiang University City, Shanghai, using class-based random cluster sampling. The sample size was calculated according to the following formula: “ n ≥ µα/2 * p * (1-p) / δ2 * deff ”32, where the test level α was set at 0.05, the sampling tolerance δ was set at 0.03, and the correction factor for cluster sampling was set at 3. The previous data showed that the detection rate of anxiety symptoms among male university students was p 24.215%, and the detection rate of anxiety symptoms in female college students p is 25.146%6. The final calculation is that the sample size of male college students needs 2350 people, and the sample size of female college students needs 2411 people, totalling 4761 people. Based on the actual situation, assuming a balanced ratio of men and women in the college student population in Songjiang University City, and assuming a 5% sample attrition rate, 5011 college students were planned to be recruited. Sampling was done on a class basis, and a total of 125 classes were selected to participate in the study. A total of 5125 college students volunteered to participate in this study through the preaching and recruitment by teachers and researchers, to whom questionnaires were distributed. The sample of this study is regionally representative and can reflect the physical and mental health status of college students in Shanghai to a greater extent.

Questionnaire survey process and quality control:

  1. (1)

    Before the test: the researchers were centrally trained to ensure that each person was familiar with the inclusion and exclusion criteria of the research subjects, only those who met the inclusion criteria and did not meet the exclusion criteria were included, and the research subjects were given the number according to the order of inclusion; the researchers grasped the standardised operation methods of each screening scale, and retained all the real data in a timely manner. Before the participants filled in the form, the researchers read out the instructions, explained the entries, clarified that the data obtained were only used for scientific research, emphasised the truthfulness, independence and voluntary responses, and informed the participants of their right to withdraw halfway through the test.

  2. (2)

    During the test: The researcher prompted the participants to answer the questions as required.

  3. (3)

    After the test: The researchers checked the missing items and those that were against common sense, and ensured that the information was complete, accurate and truthful by means of filling in and re-filling. All samples containing missing values were excluded. Questionnaires with a filling time shorter than 3 min, questionnaires with regular responses and questionnaires that do not meet the inclusion and exclusion criteria are eliminated to ensure the effective response rate.

The specific process is shown in Fig. 1.

Fig. 1
figure 1

Flowchart of the test process with the inclusion and exclusion process.

Full size image

Test instruments

General information questionnaire

It includes the basic information situation of the study subjects such as age, gender, height, weight, grade, smoking and drinking habits, household location, family status and interpersonal relationship.

Physical activity rating scale-3 (PARS-3)

The PARS-3 is suitable for the Physical Activity Survey of Chinese Youth Groups33 and has good reliability and validity34. The scale mainly investigates physical activity in the last month, and it includes intensity, duration, frequency and volume. Where physical activity was graded by intensity, duration and frequency. The scoring scale was as follows: intensity and frequency of exercise were divided into 5 grades ranging from 1 to 5. Exercise duration is also divided into 5 grades, 0–4 points. Exercise = intensity * duration * frequency. Low-intensity physical activity 19 points and below (including 19 points), moderate-intensity physical activity 20–42 points, and high-intensity physical activity 43 points and above (including 43 points); the maximum total score is not more than 100 points, and the minimum total score is not less than 0 points. The Cronbach’s α coefficient for this questionnaire was 0.8533. The questionnaire has good reliability.

Behavioral inhibition/activation system scale (BIS/ BAS)

The Behavioural Inhibition/Activation System Scale (BIS/ BAS) was developed by Carver and White in 199435, and was revised by Li Yanzhang et al.36 in China into four levels, including behavioral inhibition and behavioural activation (reward responsiveness, drive, and fun seeking) dimensions, with a total of 18 question items. The Likert 4-point scale was used, with each item ranging from 1 to 4 on a scale from “Completely disagree” to “Completely agree”. This questionnaire has good reliability36, the Cronbach’s α coefficient for this scale in this study was 0.912, KMO = 0.937.

Generalized anxiety disorder 7 item scale (GAD-7)

Anxiety symptoms were assessed using the Generalised Anxiety Disorder 7 Item Scale (GAD-7), which describes the most important diagnostic criteria for generalised anxiety disorder in the DSM-IV-TR. It is the most widely used measure of anxiety in clinical practice and research37,38, and national and international studies have shown the GAD-7 to have high reliability and validity for anxiety screening39,42.The GAD-7 asks people how often they have experienced the seven core symptoms of GAD in the past two weeks, with response options of “Not at all”, “Several days”, “Over half the days” and “Nearly everyday” (on a scale of 0–3). The seven core symptoms include nervousness (GAD1), uncontrollable worrying (GAD2), worry too much (GAD3), trouble relaxing (GAD4), restlessness (GAD5), irritability (GAD6), feeling afraid (GAD7). The total score is the main statistical indicator, the higher the score, the more severe the anxiety. A total score of 0–4 is classified as no anxiety or no clinically significant anxiety: 5–9 is classified as mild anxiety; 10–14 is classified as moderate anxiety; 15–21 is classified as severe anxiety. This questionnaire has good reliability37, the Cronbach’s α coefficient for this scale in this study was 0.940, KMO = 0.935.

Quality control

  1. (1)

    Quality control of research personnel

The researchers were centrally trained to ensure that each of them was familiar with the inclusion and exclusion criteria of the study subjects, mastered the standardised operation methods of the various screening scales, and retained all authentic data in a timely manner.

  1. (2)

    Quality control of subject enrolment

The researchers screened the subjects strictly according to the inclusion exclusion criteria. Only those who met the inclusion criteria and did not meet the exclusion criteria were included.

  1. (3)

    Quality control of questionnaire completion

The questionnaire in the recruitment of this study used electronic questionnaires throughout, and it took about 15 min to complete the whole content. All questions were mandatory, so as to avoid students’ omission. The raw data exported from the e-questionnaire platform were checked and cleaned by the researchers to eliminate invalid questionnaires and ensure the effective response rate.

Data analysis

Measurement data were expressed as mean ± standard deviation, and the results were retained with 3 decimals. T-test, one-way ANOVA and LSD post hoc multiple test were used for comparison between groups; count data were expressed as n (%), and chi-square analysis was used for comparison. Pearson correlation analysis and linear regression analysis were used to explore the relationship between physical activity level, executive function, and anxiety; Harman’s one-way test was used to test the effect of common method bias, and structural equation modelling was established to examine the role of each subcomponent of executive function in the relationship between physical activity level and anxiety (all the variables were standardised prior to modelling), and the model evaluation indexes were chosen as RMR ( root mean square residual), RMSEA (root mean square error of approximation), GFI (goodness of fit index), NFI (normed fit), and NFI (normed fit index), CFI (comparative fit index)43. Path analysis parameters were estimated using a non-parametric percentage bootstrap method (without strict requirements on the distribution of the variables), with the number of samples set at 5000, and the mediation effect was defined as statistically significant by the Bias-Corrected 95% Confidence Interval (Bias-Corrected 95% CI) of the mediated path product not crossing 0, with the Percentile 95% CI as a secondary indicator. All statistical inferences were made using two-tailed tests with the test level α set at 0.05. T-tests, one-way ANOVA, LSD post-hoc multiple testing, chi-square analysis, Pearson correlation analysis, Harman one-way test, and multiple linear regression analyses were performed using SPSS Statistics 29.0, and structural equation modelling, path analysis and testing were performed using Amos 23.0.

Results

Differences in demographic characteristics and physical activity, behavioural inhibition and activation systems among university students with different anxiety symptom scores

As shown in Table 1, the age of the subjects was (19.410 ± 1.261) years, BMI was (22.004 ± 4.113) kg/m2, 47.33% were male, 21.77% were in the habit of drinking alcohol, 5.92% were in the habit of smoking, and 9.25% were in the single parent family. The population was divided into two groups and the scores of non-anxious and anxious population were 0.95 ± 1.367 and 7.82 ± 2.962 respectively. Between-group comparisons by independent t-tests and chi-square tests revealed that there were no significant differences between the non-anxious and anxious populations in terms of BMI, smoking, and single parenthood (all p > 0.05); there were differences in age and whether or not alcohol was consumed between the two groups; and there was a significant difference in gender between the groups (p < 0.001), which provided a basis for subsequent grouping. In addition, 4.29% of college students had poor family status, 17.24% had difficulties with coursework, and 15.65% never participated in social activities. Family status, coursework pressure, social activities and anxiety symptoms were all significantly associated (all p < 0.001), suggesting that this group of people is susceptible to anxiety symptoms.

Table 1 Demographic characteristics of university students with different anxiety levels.
Full size table

As shown in Table 2, the anxiety group had lower scores in reward responsiveness, drive, and fun seeking; higher scores in behavioural inhibition; and lower intensity, shorter time, lower frequency, and smaller total exercise. Independent t-tests showed that the differences between groups were greater than the differences within groups (p < 0.001).

Table 2 Comparison of anxiety scores, behavioral Inhibition and activation, and physical activity between groups of college students with different anxiety levels.
Full size table

One-way ANOVA showed that the drive, fun seeking and behavioural inhibition scores were significantly higher in the male non-anxious group than in the female non-anxious group; whereas the reward responsiveness and behavioural inhibition scores were significantly lower in the male anxious group than in the female anxious group; and the exercise volume was higher in the males than in the females, as detailed in Table 3. In particular, anxiety, exercise volume, and intensity had large effect sizes (η2 > 0.14). Exercise duration had a medium effect size (η2 > 0.06).

Table 3 Comparison of anxiety scores, behavioural Inhibition and activation, and physical activity scores of male and female colloge students at different levels of anxiety.
Full size table

Relationships between physical activity and anxiety, behavioural Inhibition and activation systems in college students

Pearson’s correlation coefficients were used to examine the relationship between physical activity level, anxiety and behavioural inhibition and activation system of college students. Anxiety is positively correlated with BIS and negatively correlated with BAS, with small differences between male and female (Fig. 2).The correlations between physical activity and the behavioural inhibition system varied across gender groups, although the correlations between physical activity frequency and the behavioral inhibition system were slightly higher for male students than for female students. Physical activity was significantly negatively correlated with anxiety, weakly correlated with the behavioural inhibition system, and significantly positively correlated with behavioural activation-reward responsiveness, drive, and fun seeking, which provided the basis for the construction of the subsequent mediation model.

Fig. 2
figure 2

Heat map of correlation coefficients between physical activity and anxiety, behavioural inhibition and activation in college students.

Full size image

Predictive effects of physical activity level, behavioural Inhibition and activation on anxiety

To examine the predictive effects of physical activity, behavioural inhibition and activation together on anxiety and to explore the feasibility of the structural relationship model, three models were constructed for multiple linear regression analysis using stepwise regression.

Our previous calculations indicated that family status, coursework stress, and social activities were important factors influencing anxiety symptoms, so we controlled for these factors as covariates.

Model 1 (Table 4) was analysed by multiple linear regression with physical activity (exercise intensity, duration, frequency) as the independent variable and anxiety as the dependent variable. The results showed that in the total group, exercise intensity (Beta = − 0.073, p < 0.001) and exercise duration (Beta = − 0.094, p < 0.001) were predictive of anxiety in college students, and the regression coefficient for exercise frequency (Beta = − 0.027, p = 0.070) was not significant. The male group performed consistently with the total group, with more exercise intensity (Beta = − 0.032, p = 0.143) regression coefficients not significant in the female group. Higher exercise intensity and longer exercise duration were associated with lower anxiety levels in males, and the strongest associations were found between exercise intensity and anxiety levels in males (Beta = − 0.088, p = 0.001); and exercise duration was most strongly associated with anxiety levels in females (Beta = − 0.084, p < 0.001). After controlling for confounders, the relationship between physical activity hours and anxiety symptoms in boys was no longer significant, and the relationship between the remaining independent variables and anxiety symptoms still held.

Model 2 (Table 5) was subjected to multiple linear regression analysis with behavioural inhibition and activation (reward responsiveness, drive, fun seeking, and behavioral inhibition) as independent variables and anxiety as dependent variable. The results showed that in the total group, reward (Beta = − 0.167, p < 0.001) and drive (Beta = − 0.236, p < 0.001) were negatively predictive of college students’ anxiety; behavioural inhibition (Beta = 0.264, p < 0.001) was positively predictive; and fun seeking (Beta = − 0.061, p = 0.004) regression coefficients were not significant. The gender groups performed consistently with the total group, with reward responsiveness having the strongest association with anxiety levels in males (Beta = − 0.255, p < 0.001); and behavioural inhibition having the strongest association with anxiety levels in females (Beta = 0.291, p < 0.001). After controlling for confounders, the relationship between the independent variables and anxiety symptoms remained valid.

Model 3 (Table 6) was subjected to multiple linear regression analysis with physical activity level (exercise intensity, duration, frequency), behavioural inhibition and activation (reward responsiveness, drive, fun seeking, behavioural inhibition) as independent variables and anxiety as dependent variable. The total group results showed that the regression model passed the test of significance (F(4, 5011) = 112.039, p < 0.001, R2 = 0.135), and that exercise intensity (Beta = − 0.048, p = 0.002), exercise duration (Beta = − 0.066, p < 0.001), reward responsiveness (Beta = − 0.165, p < 0.001 ), drive (Beta = − 0.217, p < 0.001), fun seeking (Beta = − 0.063, p = 0.003), and behavioural inhibition (Beta = 0.251, p < 0.001) were all predictive of anxiety in college students, whereas regression coefficients for exercise frequency (Beta = − 0.020, p = 0.155) were not significant. After controlling for confounders, the length of physical activity was no longer significant, the frequency of physical activity remained significant for male students, and the intensity of physical activity was no longer significant for female students.

The p-values of all three models were significant. And the VIF values of all independent variables in the model were less than 5, indicating that the results of this study can basically exclude the influence of multiple covariance, as shown in Table 4. Multiple linear regression analyses showed that physical activity, behavioural inhibition system and activation all had a predictive effect on college students’ anxiety symptoms. However, the predictive effects of exercise frequency and fun seeking were not obvious. Exercise intensity did not have a predictive effect on female college students’ anxiety symptoms. Exercise intensity and exercise duration were strongly associated with anxiety levels of male and female college students, respectively.

Table 4 Multiple linear regression analysis of physical activity on anxiety symptoms.
Full size table
Table 5 Multiple linear regression analysis of behavioural Inhibition and activation on anxiety symptoms.
Full size table
Table 6 Multiple linear regression analysis of physical activity, behavioural Inhibition and activation on anxiety symptoms.
Full size table

Analysis of the mediating effects of behavioural Inhibition and activation on anxiety

Firstly, Harman’s one-way test was used to test for common method bias. The results showed that the analysis of the overall data yielded five factors with characteristic roots greater than 1, and the variance explained rate of the first common factor was 29.122%; the analysis of the data of male college students yielded four factors with characteristic roots greater than 1, and the variance explained rate of the first common factor was 33.791%; and the analysis of female college students yielded five factors with characteristic roots greater than 1, and the variance explained rate of the first common factor was 25.185%, all these three values were less than 40% of the critical value, which indicated that there is no serious common method bias in this study.

The interrelationships of physical activity, behavioural inhibition and activation, and anxiety symptoms among college students were used as the basis to test the research hypotheses. Model 1 was established with physical activity as the independent variable, behavioural inhibition and activation as the mediator variable, and anxiety as the dependent variable, and then 2 sub-models were established with the data of male and female college student groups respectively. Paths with non-significant coefficients were eliminated one by one and recalculated until all path coefficients passed the Bootstrap significance test. Repeating this process, the results of the path analysis are shown in Fig. 3, and the results of the mediation effect test are shown in Table 5.

Referring to the reference criteria of CMIN/df < 5, RMR < 0.05, RMSEA < 0.08 and GFI, NFI and CFI values > 0.935,43. Model 1 has CMIN/df = 6.865, indicating a good model fit, and the 2 sub-models have CMIN/df < 5, which is a good model fit. The rest of the goodness-of-fit metrics of the 3 models are fully compliant with the criteria, indicating that the structural equation models are reasonable and reliable.

Three measures had factor loadings greater than 1.00 on physical activity; four measures had factor loadings between 0.81 and 1.03 on behavioural inhibition and activation; and seven measures had factor loadings between 0.85 and 1.07 on behavioral inhibition and activation. All loadings reached significance at the α = 0.001 level. This indicates that these measures can better characterise the various potential qualities of the study’s predictors.

Direct effects of physical activity on anxiety

Physical activity had a direct effect on reducing anxiety by 84.37% (effect size − 0.286, 95% bootstrap CI:=[ -0.366,-0.213]), as shown in Table 5. Physical activity had a significant predictive effect on anxiety for both sexes and was more predictive for the female cohort. In terms of factor loading values for each measure, exercise duration (M:1.81; F:2.71) was more important than exercise intensity (M:1.74; F:2.07), and frequency (M:1.00; F:1.00) in the manifestation of anxiolytic symptoms. Details are in Fig. 3b, c. Specifically, it was shown to be more effective in alleviating the four dimensions of uncontrollable worrying (1.07), worry too much (1.03), trouble relaxing (1.00), and nervousness (1.00), with less difference in values between the male and female groups. Details are in Fig. 3a. Physical activity was more effective in alleviating anxiety symptoms of male college students in the dimensions of feeling afraid (M:0.96; F:0.88), restlessness (M:0.94; F:0.82). Details are in Fig. 3b, c.

Mediating effects of behavioural Inhibition and activation system

The mediating effect of behavioural inhibition and activation between physical activity and anxiety was significant (effect value − 0.053, contribution to effect 15.63%, 95% bootstrap CI=[ -0.079,-0.030]), as shown in Table 5. Behavioural inhibition and activation mediated in both sexes and anxiety symptoms were more controlled by behavioural inhibition and activation in the female group (29.92%) than in the male group (15.76%) were more controlled by behavioural inhibition versus activation. In terms of factor loading values for each measure, drive (M:1.01; F:1.04) was better than behavioural inhibition (M:0.73; F:0.65), fun seeking (M:0.98; F:1.02), and reward responsiveness (M:1.00; F:1.00) in the performance of anxiolytic symptoms, as shown in Fig. 3b, c.

Fig. 3
figure 3

Schematic diagram of the path analysis of physical activity levels affecting anxiety through the behavioural inhibition and activation system.

Full size image
Table 7 Bootstrap test results for mediation effects.
Full size table

Discussion

This study not only examined the predictive roles of gender and physical activity on BIS/BAS, and validated the direct effects of PA on anxiety, but also revealed the mediating role of BIS/BAS, providing new evidence for understanding how PA affects anxiety symptoms through neuropsychological mechanisms. This study had a large sample size and strict process control. The results mainly found the explanation of behavioural inhibition and activation and physical activity on anxiety of college students of different genders. It is expected to provide theoretical support and guiding significance for the study of college students’ psychological state or the construction of physical education programmes in universities.

The present study found that the anxiety symptomatic population had lower intensity, shorter duration, lower frequency, and smaller volume in physical activities; the anxiety symptomatic population showed lower scores in reward responsiveness, drive, and fun seeking, and higher scores in behavioural inhibition, which was consistent with previous studies44,45, and verified hypothesis a. In addition, the effect sizes of family status, coursework pressure, and social activities were larger in the comparison of anxious versus non-anxious college students. This study further found that there was a significant difference in gender between the non-anxious and anxious groups, with females having a higher prevalence of anxiety symptoms, which also verified the previous studies46,47. Physical activity is effective in stimulating the central nervous system48,49, increasing self-efficacy, and enhancing emotional regulation in situations of stress and anxiety7,50,51. Therefore the difference in scores in the non-anxiety group may be related to the fact that men exercise more than women52.

The present study found that physical activity of college students was significantly negatively correlated with anxiety, which is consistent with the results of a previous study53. And the correlation coefficients between anxiety and the amount, intensity, and duration of exercise were greater. College students’ physical activity was weakly correlated with the behavioral inhibition system, and significantly positively correlated with behavioral activation-reward responsiveness, drive, and fun seeking, consistent with the findings of Firoozjah et al.54, which to some extent verified hypothesis b. And the correlation coefficients between anxiety and BAS-reward responsiveness, drive, and fun seeking were greater compared to BIS. A large number of scholars have explored the intrinsic mechanisms of exercise to improve mood from physiological and psychological mechanisms. And the results of the present study provide a new perspective for the field that the behavioural inhibition and activation system (mainly the behavioural activation system) may be the potential pathway of action for the improvement of anxiety symptoms after physical activity.

The present study supported that physical activity, behavioural inhibition and activation were all predictive of college students’ anxiety symptoms, which was consistent with previous studies44,48,55,56, and verified hypothesis c. The present study further found that, in the physical activity dimension, exercise frequency and fun seeking were not significant predictors of college students’ anxiety symptoms. And exercise intensity was not predictive of anxiety symptoms in female college students. Exercise intensity and exercise duration were strongly associated with anxiety levels of male and female college students, respectively. In the behavioural inhibition and activation dimensions, this study further found gender differences in BIS/BAS reactivity. By making gender comparisons in the anxiety symptom group: Charles et al. concluded that females scored significantly higher than males on BIS responsiveness57,58,59. Functional magnetic resonance imaging (fMRI) studies have shown that women have stronger connections between the prefrontal cortex (PFC) and the amygdala, which may make them more sensitive to negative emotions60. This may explain why BIS is a stronger predictor of anxiety in women. Carver and White found that females scored higher than males on BAS reward responsiveness in a sample of college students57. By making gender comparisons in the non-anxiety symptom group: Verbeken et al. found that males scored slightly higher than females on the BAS drive61. The study further builds on previous research by finding that the male non-anxiety group had similarly significantly higher scores on fun seeking and behavioural inhibition system than the female non-anxiety group.

The important finding of the present study is that physical activity can indirectly affect anxiety symptoms by acting on the behavioural inhibition and activation system with gender differences. Guyer et al. identified areas of brain activity associated with reward and punishment including the ventral striatum, hippocampus, hypothalamus, and insula, which are associated with emotional processing62. This explains the mediating role of behavioural inhibition and activation systems in anxiety symptoms in both sexes63. From a neuroanatomical point of view, females have higher BIS sensitivity and higher FC (Functional connectivity) between the seed areas of the ventral medial prefrontal cortex and the posterior parietal areas compared to males; this FC mediates the effect of gender on BIS sensitivity60. As a result, anxiety symptoms are more controlled by the behavioural inhibition and activation system in the female population than in the male population64. In addition, consistent with previous work65 the present study also found that exercise time was more important in the expression of anxiolytic symptoms. This was shown to be more effective in alleviating the four dimensions of uncontrollable worrying (GAD2), worry too much (GAD3), trouble relaxing (GAD4), and nervousness (GAD1). Feeling afraid (GAD7), restlessness (GAD5) anxiety symptom dimensions had a greater effect on anxiety symptoms among male college students.

Through a large sample and strictly controlled experimental design, this study verified the alleviating effect of physical activity on anxiety, and revealed the gender differences in the BIS/BAS system and its mechanism of action in anxiety symptoms. The results of the study can directly provide a scientific basis for anxiety intervention for college students and have the following practical implications. ① Promote individualised exercise intervention and optimise the design of physical education curriculum in colleges. For males, high intensity and prolonged exercise (e.g., aerobic training of more than 60 min) was emphasized to enhance fun seeking and drive, and to increase the activity of the behavioural activation system. For females, the emphasis is on moderate-intensity, long-duration exercise (e.g., yoga, jogging) to reduce behavioural inhibition and improve anxiety relief. ② Precision intervention for anxiety symptoms. For individuals whose anxiety is accompanied by nervousness, meditative exercises such as yoga and tai chi are recommended to reduce BIS sensitivity. For individuals with worry too much, rhythmic exercises, such as jogging and swimming, are recommended to enhance BAS drive and improve emotional stability. By accurately matching the type of anxiety symptoms with exercise modalities, it provides a new research direction for further exploring the neural mechanisms of anxiety. This not only contributes to the mental health management of college students, but also provides new theoretical support for sports intervention in anxiety disorders.

Limitations and prospects

  1. (1)

    This study used the cross-sectional design, which limits causal inferences. It is recommended that more longitudinal studies be conducted in the future to further explore the time series of changes in the causal relationship between physical activity, behavioural inhibition and activation, and anxiety, and the magnitude of the related effect sizes;

  2. (2)

    Physical activity in this study was measured using a self-report questionnaire, which may have some bias and it is recommended that in the future, we combine accelerometers, heart rate bands, and other measurement tools to collect information from subjects and report physical activity more accurately;

  3. (3)

    The demographic diversity in this study is limited, as participant data were collected exclusively from the University Town in Songjiang District, Shanghai, which may affect the generalizability of the results.

  4. (4)

    This study used the GAD-7 to measure anxiety symptoms. Some scholars have proposed a variety of anxiety symptoms such as state anxiety, trait anxiety, and social anxiety, which can be further categorised and discussed in the future according to the source and nature of anxiety.

Conclusions and recommendations

Physical activity has a direct effect on anxiety, and the anxiolytic effects of physical activity can be mediated through the behavioural inhibition and activation system, and the mediating effect is highest in the female college student population. It is recommended that psychoeducation in colleges should be combined with exercise to encourage students to engage in more physical activities, increase sensitivity to behaviourally activated coping strategies and reward messages, and alleviate anxiety symptoms. Attention should also be paid to gender differences and individualised exercise programmes should be adopted, especially for female students who are at higher risk of anxiety. In addition, in the long term, designing and conducting more engaging physical education programmes is necessary for higher education institutions in order to promote the formation of stable exercise habits among students. In the future, we hope to build on this study to explore more specific exercise improvement options for anxious college students.