The study was conducted in accordance with the Cochrane Handbook for Systematic Reviews of Interventions and the PRISMA Statement for Systematic Evaluation (27). Four electronic literature databases were searched to identify included studies: Cochrane, EMBASE, PubMed and Web of Science. Search terms included combinations of subject and free words for interventions, outcome indicators, and study subjects (see Table 1).

The following inclusion criteria were used to select studies: (1) overweight or obese adults aged ≥18 years; (2) comparison of the HIIT with the MICT; (3) reported measures of affect, pleasure, and intention and (4) overweight or obese who participated in HIIT and MICT. All the studies are randomized controlled trials.

In this study, 2 evaluators (YL and JSZ) independently assessed the quality of the included literature by using Cochrane Collaboration’s tool for assessing risk of bias. The quality of the literature was systematically evaluated in the following 7 areas: (1) description of the randomization method; (2) concealment of the allocation scheme; (3) double-blind principle; (4) blinding principle for outcome evaluation; (5) data completeness; (6) selective reporting of outcome results; and (7) assessment of the presence of other biases (Figure 1).

Data extraction and literature quality assessment were conducted independently by 2 evaluators (YL and JSZ) who extracted data from the included studies into an electronic data extraction form. When there is disagreement, a third reviewer will reevaluate it. Extracted data included literature general study information, study participant information (number, age, BMI), intervention characteristics (intensity, duration), and data on outcome indicators. When situations existed where data were unavailable, but graphical displays were available, we extracted data using a freely available web-based data extraction tool (Engauge Digitizer version 12.1) (28–31).

Data extraction and literature quality assessment were compared between two evaluators, with any disagreements being resolved by a third evaluator.

A visual analysis of funnel plot and the Egger’s and Begg’s test were performed to assess the publication bias across studies. At the same time, we used trim-and-fill process to assess the publication bias. This method involves examining the correlation between effect sizes and standard errors of effect sizes to determine if there is a significant association between study effect size and study precision.

Effect sizes were determined by calculating the standardized mean difference (SMD). Heterogeneity was tested using I². I² for 0–50% is low heterogeneity and 50–100% is high heterogeneity (32). Due to the heterogeneity of the Meta-analysis, a random effects model was chosen to integrate the combined data across the text. Review Manager 5.4.1 was used to perform three statistical analyses with confidence intervals (CI) of 95% for the three outcomes (Physical Activity Enjoyment Scale, Feeling Scale and Felt Arousal Scale) of this study.

The initial review yielded 73 studies, and after excluding 10 duplicates, 11 studies that were not relevant to the topic, 21 studies that did not compared HIIT and MICT, 5 review studies, 2studies that had the same data and could not be analyzed based on the titles and abstracts, 2 other language studies, 1 chronic study, 2paediatric studies and 3 other studies, there were still 16 studies that needed to be evaluated in full text. Figure 2 illustrates the PRISMA process for study selection.

A total of 537 participants met the inclusion criteria in 16 studies, including 213 participants in HIIT, 183 participants in MICT, 84 participants in alternating HIIT and MICT, and 57 participants in other forms of intervention (self-selected intensity exercise, very high-intensity interval exercise, repetitive sprint training, and blank control). All participants were between the ages of 18–70 years, had an intervention duration of 1–16 weeks, and had a BMI of 25 or higher. Five studies crossed over two exercises with the same group of participants. Eleven studies had participants under the age of 40 years, three between the ages of 40–60 years, and two over the age of 60 years.

Nine studies described the use of randomized grouping methods, such as the random number table method or computer generated randomization (21, 29, 31, 33–38), they were therefore assessed as having a low risk of bias. Seven studies were assessed as having an unclear risk of bias due to the lack of description of the randomization method (28, 30, 39–43).

Seven studies described a method of allocation concealment in which sealed opaque envelopes were employed or allocated by a third people (31, 33, 36–40), thus were assessed as having a low risk of bias. The remaining 10 studies did not describe the allocation hiding method, thus indicating an unclear risk of bias (21, 28–30, 34, 35, 41–43).

One study showed that participants were not blinded (21), it was therefore assessed as having a high risk of performance bias. Three studies have a low risk of bias due to the description of the double-blind method (34, 40, 43). The rest 12 studies did not describe the use of blinding of participates and personnel (28–31, 33, 35–39, 41, 42), thus indicating an unclear risk of bias.

In four studies, detection bias was judged to be low risk because they provided detailed information on the assessment of blinded outcomes (21, 28, 35, 41). The other 12 studies were judged to have uncertain risk because there was no description of blind outcome assessment (29–31, 33, 34, 36–40, 42, 43).

All 16 studies, reported complete outcome data, thus were assessed as having a low risk of bias.

Two studies did not report all pre-specified primary outcome (30, 34), thus were judged to be high reporting risk. Two studies had insufficient information to determine whether there was a risk of selective reporting of results (39, 41), thus were judged to be unclear reporting risk bias. The other 12 studies were judged to have low risk because all needed outcomes have been reported (21, 28, 29, 31, 33, 35–38, 40, 42, 43).

Meta-analysis showed high heterogeneity of PACES and FS, and exclusion of the studies failed to reduce the level of heterogeneity, indicating a stable overall outcome. However, after excluding two studies (31, 34) the heterogeneity of PACES was significantly reduced to a relatively low level (I² = 46%). All two studies had an intervention duration of 12 weeks, three times per week, to further investigate the possibility of high heterogeneity, we performed subgroup analyses for age, form of exercise, and duration of exercise, but the significance of the results did not change significantly.

We used funnel plot and trim and fill process to assess publication bias. Based on visual observations, we found that the funnel plot for PACES and FS, there is no significant asymmetry (Figures 3, 4), which suggests that the included studies may not be significantly affected by publication bias, so the results of the meta-analysis may be relatively reliable. Further quantitative tests showed that there was no significant publication bias across studies (Begg’s test, p = 0.100; Egger’s test, p = 0.152, Supplementary Figure 1; The estimated effect size is 1.160 with a 95% confidence interval of 0.147 to 2.173, Supplementary Figure 2. Begg’s test, p = 0.368; Egger’s test, p = 0.867, Supplementary Figure 3; The estimated effect size is −1.362 with a 95% confidence interval of −3.787 to 1.063, Supplementary Figure 4). The analysis revealed no studies to be supplemented, indicating that there was no significant publication bias in the current dataset, or that the effect of publication bias was not large enough to be corrected by trim and fill process.

Study characteristics are summarized in Table 2.

While affect is a conscious response to an emotional direction (positive, neutral, or negative), enjoyment is a more specific feeling (48). Traditional concept suggests that high-intensity exercise above the ventilation threshold can cause unpleasant feelings about exercise (49). Although HIIT is a form of high-intensity exercise, it is characterized by brief, repeated intervals of rest or low-intensity exercise (50). There are multiple “recovery” cycles, which can lead to a psychological “rebound effect” in exercisers. Jung suggests that during recovery intervals there may be a “rebound effect” whereby participants may feel a more positive emotional response during the recovery period, as the intervals may contribute to a repeated boost in confidence during a single workout, allowing participants to know that they are approaching a “comfort zone” for recovery, and thus mobilizing their own positive emotions (51), Participants’ confidence and mobilization of positive emotions can thus be continuously enhanced. However, the “rebound effect” seems to occur only on the post-exercise scale, because participants have more time to cushion and recover after exercise, and the use of the scale during exercise may not be a good explanation for this effect, because participants will still feel tired during short intervals of exercise. Therefore, the overall effect of HIIT on the PACES scale measured after exercise will be better than MICT.

In addition, exercise preference is one of the factors that influence people’s choice of different types of exercise. Hedonistic theories of behavior suggest that people are intrinsically predisposed to behaviors that bring them pleasure and stay away from those that bring them displeasure (52). It has been suggested that HIIT may be preferred for achieving personal health goals better than MICT, which may cause participants to become frustrated and give up more easily because it may take more time (53).

It is worth noting that it involves both acute and chronic studies on enjoyment response. The result of subgroup analysis of acute and chronic studies shows that HIIT may contribute to obtaining psychological responses that are equal to or more positive than MICT in short period. The reason why it is feasible to include both acute and chronic is because we are looking at an overall effect. Acute effects may disappear over time, or adaptive changes may occur over a long period of time. By combining acute and chronic studies, we can reveal more complex time-dependent effects. The overall effect shows beneficial effects of HIIT on enjoyment indicating that overweight or obese people may more willing to try HIIT due to hedonistic theories of behavior.

Results show that the difference in the rates of outcome events between the two groups was not statistically significant. It might because HIIT is performed using multiple sets of stimulus/recovery combinations, and variations in the ratio of stimulus to recovery time will affect the emotional experience of participants, which will likely have an impact on affective responses (43). In the experimental design of Oliveira et al. (29), the affective responses to HIIT were lower than those of MICT in a pattern of 1 min of exercise with 1 min of active recovery as a group (29). However, in the experimental design of Ram et al. the affective responses to HIIT were higher than those of MICT in a pattern of 10 * 1 min intervals at 90% peak heart rate with 1 min active recovery intervals at a low workload (15% WRpeak) (21), suggesting that by rationalizing the ratio of stimulation to recovery time in high-intensity exercise and improving the affective responses resulting from influencing it, it might be a good prescription for the overweight and obese groups.

Two studies reported data on the assessment of Arousal during exercise using Felt Arousal Scale (47). Due to the small number of literatures and the different measurement times of FAS scale in the two studies, the average score was calculated for analysis. The results showed that both HIIT and MICT could bring positive emotional activation responses to participants, but there was no significant difference between groups (SMD = −0.29; 95% CI = −0.86 ~ 0.28; p > 0.05). Due to the limited amount of literature, it may not be possible to fully understand the research status and development trend of this field. In addition, the lack of sufficient information may also limit the in-depth discussion of certain specific issues. However, studies have shown that emotional activation depends on the intensity of the activity (e.g., light or moderate intensity) or the different moments of the session (e.g., warm up, cool down) (54). Thus, the results should be treated with caution.

In conclusion, this review showed that HIIT can bring better pleasure response than MICT in overweight or obese people, but there is no significant difference in emotional response. It is possible that the inconsistency between this conclusion and the results of existing studies may also be due to differences in the interventions, such as the relative intensity, duration, and total number of work sessions completed (34, 39); and heterogeneity in the backgrounds of the participants, such as age, activity level, and obesity (29, 35, 42). Since the study population of the present meta-analysis was exclusively obese and overweight patients, whereas previous meta-analyses have had a much broader study population (22, 23), BMI may be one of the factors influencing the discrepancy between the results of the present study and those of previous studies. It has been shown that obese women experience lower levels of pleasure and energy compared to non-obese women, which may partially explain their significantly lower levels of participation in physical activity (55), and therefore this may make a smaller difference in the emotional responses of the obese group to the two types of exercise.