Shorter High-Intensity Cycling Intervals Reduce Performance and Perceived Fatigability at Work-Matched but Not Task Failure : Medicine & Science in Sports & Exercise

Journal Logo


Shorter High-Intensity Cycling Intervals Reduce Performance and Perceived Fatigability at Work-Matched but Not Task Failure


Author Information
Medicine & Science in Sports & Exercise 55(4):p 690-699, April 2023. | DOI: 10.1249/MSS.0000000000003097


High-intensity interval training (HIIT) is a popular exercise paradigm adopted in several research, clinical, and applied settings (1–4). HIIT consists of a series of constant load work bouts conducted within the severe-intensity domain and interspersed by recovery phases, which differs from sprint interval training, typically repeated “all-out” work bouts (5). The HIIT exercise paradigm permits the completion of a greater amount of work within the severe-intensity domain as compared with conventional constant-load exercise (5,6). The enhanced work capacity within the severe-intensity domain through HIIT is the result of an attenuated metabolic disturbance resulting from the succession of work–recovery intervals (6,7). As such, the durations of work and recovery phases become critical modulatory factors of the metabolic and perceptual perturbations elicited by HIIT. For example, in isolated knee-extension exercise, shorter work intervals during HIIT were associated with reduced oxygen consumption and pH fluctuations compared with longer intervals performed at the same workload (7). The level of metabolite perturbations (such as Pi and H+) is linked to muscle force-generating capacity (8–10). Accordingly, it is plausible that the reduced metabolic perturbations during shorter work interval HIIT would alleviate psychophysiological responses characterizing performance fatigability and enhance work capacity. Whereas fatigue is generally defined as a symptom of tiredness or weakness resulting from mental or physical exertion (11,12), performance fatigability has been defined as a decline in an objective measure of performance such as time to task failure and maximal voluntary or evoked force production capacity (11,13). From a neuromuscular perspective, performance fatigability is traditionally attributed to central (i.e., reduced central motor drive) and/or peripheral (i.e., compromised muscle action potential propagation and contractile function) mechanisms (10,11).

Consequential to a reduction in local metabolic perturbations, shorter work interval HIIT might also reduce feedforward and feedback sensory discharges associated with perceptual responses such as effort, pain, and dyspnea. More specifically, shorter work interval HIIT might reduce activation of mechanosensitive and metabosensitive type III/IV sensory afferents, which could abate perceived pain development (14) and require a reduced ventilatory load that is associated with dyspnea (15). In addition, mitigated disruptions in muscle contractile function during shorter HIIT might reduce the necessity of additional muscle unit recruitment (16) and consequently decrease associated corollary discharge and perceived effort (17). Overall, alleviation of feedforward and feedback sensory responses during shorter work interval HIIT can increase self-efficacy and control mastery during the exercise and reduce the sensation of fatigue at a work-matched level compared with longer work interval HIIT (18). Integration of these responses will postpone the attainment of the sensory tolerance limit and increase work capacity during shorter versus longer work interval HIIT.

To the best of our knowledge, only one study has explored the effect of different work interval durations, while matching work-to-rest ratios, on performance fatigability responses to HIIT exercise. This study, albeit using sprint interval exercise, demonstrated that shorter work intervals were associated with greater reductions in muscle contractile function (6). However, the exercise protocols in this study were not performed until task failure, and a multifactorial assessment of physiological and perceptual responses was not performed. It is plausible that for the same power output, shorter HIIT would permit a longer overall exercise duration to task failure. Furthermore, according to the concept of sensory tolerance limit (for a review, see (17)) and recent evidence on the effect of exercise intensity–duration (19), it could be hypothesized that a unique and differential integration of performance and perceived fatigability would characterize the limit of exercise tolerance in the shorter versus longer interval HIIT protocols.

The aims of the present study were twofold. First, we aimed to characterize neuromuscular, perceptual, and cardiorespiratory contributions to performance fatigability at a work-matched time point during work-to-rest matched HIIT protocols differing in work interval duration (3:3 vs 1:1 min work/rest) and during constant load exercise to task failure. We hypothesized that, at work-matched time points, shorter HIIT intervals would result in less psychophysiological stress responses. Second, we aimed to explore the same measures at task failure. Accordingly, we hypothesized that shorter HIIT intervals would permit a greater exercise duration but that, nevertheless, the extent of neuromuscular and perceptual alterations would be similar at task failure no matter the exercise protocol/paradigm.



Twelve healthy, recreationally active participants (n = 6 women, 26.1 ± 5.3 yr, 172.0 ± 10.3 cm, 71.2 ± 11.4 kg) were recruited for this study. The sample size required was estimated using G*Power software (version, with data from a previous investigation that explored neuromuscular fatigability development during two work-matched high-intensity intermittent exercises (6). A sample of 12 participants was estimated to achieve statistically significant difference in neuromuscular responses, for an expected effect size of 0.8 and power of 0.8 with an α level of 0.05. All participants completed the Physical Activity Readiness Questionnaire (20) and were free from any musculoskeletal, neurological, cardiovascular, and/or cardiorespiratory disorders. Participants provided written informed consent and were instructed to avoid vigorous exercise 24 h in advance of experimental sessions and to avoid caffeine and/or alcohol within 12 h of a testing period. Hormonal changes during the menstrual cycle might affect neuromuscular functions and exercise performance (21). However, coincidentally all women who participated in this study were regularly menstruating, monophasic oral contraceptive users (tested during 3 wk of the active pill phase) or had a hormonal intrauterine device. The protocols for this study were approved by the University of Calgary Conjoint Health Research Ethics Board (REB21-0629) and conducted based on the Declaration of Helsinki (without registration).

Experimental Protocol

Participants reported to the laboratory to complete the following four experimental sessions (i) one ramp-incremental test, (ii) one HIIT protocol consisting of a 1-min work-phase and a 1-min rest phase (HIIT1min), (iii) one HIIT protocol consisting of a 3-min work-phase and a 3-min rest phase (HIIT3min), and (iv) a constant-load cycling trial (CL). The sessions were performed in a randomized sequence and separated by at least 48 h. During experimental sessions 2 to 4, participants were required to complete cycling at 90% of peak power output (PPO) accomplished during the ramp incremental test. All cycling trials and neuromuscular assessments were completed on a custom-made, semirecumbent, electronically braked cycling ergometer (22).

During their first visit, participants were familiarized with the neuromuscular assessment procedure, the perceptual response scales (rating of perceived effort, general fatigue, leg pain, and dyspnea), and the cycling ergometer setup. Next, participants completed a 20 W·min−1 ramp-incremental test to volitional exhaustion to measure PPO and peak rate of oxygen consumption (V̇O2peak) (22). In the second, third and fourth laboratory visits, participants completed the HIIT (i.e., HIIT1min, HIIT3min) and CL trials in random order. For HIIT1min and HIIT3min sessions, the rest phase consisted of passive recovery on the semirecumbent cycle ergometer. Throughout all experimental sessions, participants were instructed to cycle at their preferred cadence, which was self-selected within the range of 80 to 90 rpm. Task failure was determined when participants were unable to maintain 60 rpm for 10 s despite strong verbal encouragement (23).


Force production

The participant’s dominant leg (the preferred leg used to kick a ball (24)) was used for neuromuscular evaluation. The horizontal force output, which was evaluated when the knee angle was at 90°, was recorded through a pedal mounted on the ergometer crankshaft that was measured by a wireless PowerForce analysis system (Model PF1.0.0; Radlabor GmbH, Freiburg, Germany). Force was sampled at 500 Hz and recorded using Imago Record, version 8.50 (Radlabor, GmbH). The signal was then transferred to a PowerLab system (16/35; ADInstruments, Bella Vista, NSW, Australia) and then, using a National Instruments 16-bit A/D card (NI PCI-6229; National Instruments, Austin, TX), displayed on a computer monitor positioned directly in front of the cycle ergometer.

Neuromuscular assessment

During all cycling trials and neuromuscular assessments, EMG signals were measured using self-adhesive Ag/AgCl surface electrodes (Kendall MediTrace; Covidien LLC, Mansfield, MA) that were placed on the muscle belly of vastus lateralis, and rectus femoris. To decrease signal impedance, the area for each electrode was first shaved and then cleaned with an alcohol swab. EMG signals were recorded at a sampling rate of 2000 Hz using PowerLab (16/30-ML880/P; ADInstruments) and an octal bioamplifier (ML 138, ADInstruments; common mode rejection ratio, 85 dB; gain, 500; low-pass Bessel filter of fourth-order and high-pass filter of first-order) with a bandpass width (5–500 Hz) and then analyzed using LabChart 8 software (ADInstruments).

Peripheral nerve stimulation was delivered using an electrical stimulator (DS7A; Digitimer, Welwyn Garden City, United Kingdom), with the cathode electrode (Kendall MediTrace) placed on the femoral nerve within the femoral triangle and the anode electrode (Durastick Plus; DJO, Global, Vista, CA) placed on the gluteal fold. The femoral nerve stimulations with 10-mA increments were delivered until a maximal amplitude of single-twitch force and muscle compound action potential (Mmax) was observed (25). A supramaximal intensity (130% of the amperage used to evoke a maximal single-twitch force) was used for stimulations throughout the remainder of the session. The mean ± SD values of the supramaximal stimulus for the CL, HIIT3min, and HIIT1min were 129.2 ± 26.5, 132.6 ± 34.4, and 129.4 ± 27.0 mA, respectively.

The neuromuscular assessment was conducted on an innovative semirecumbent cycle ergometer that demonstrated a high degree of relative (intraclass correlation coefficient >0.90) and absolute (coefficient of variation <5.2%) reliability (22). This assessment consisted of a high-frequency doublet (100Db) stimulation (PNS) superimposed on the maximal voluntary contraction (MVC) followed by three stimuli: a 100-Db, a low-frequency doublet (10Db), and a single twitch, which were evoked every 3 s after the MVC (Fig. 1) (23). Neuromuscular assessments were performed at baseline (two baseline evaluations were then averaged for a mean baseline value) and immediately (within 1–2 s) of task failure for all conditions. In addition, neuromuscular assessments were performed every 6 min of work (every 6 and 2 HIIT intervals for HIIT1min and HIIT3min, respectively; Fig. 1).

Schematic figure of the experimental protocol. The three cycling protocols at 90% of PPO to task failure included (i) HIIT protocol consisting of a 1-min on-phase and a 1-min rest phase (HIIT1min), (ii) HIIT protocol consisting of a 3-min on-phase and a 3-min rest phase (HIIT3min), and (iii) CL cycling trial (A). The neuromuscular assessment included an MVC, a 100Db superimposed on the MVC and three stimuli including 100 Db, 10Db, and a single twitch evoked every 3 s after the MVC (B).

Cardiorespiratory and metabolic measures

Cardiorespiratory data (oxygen uptake (V̇O2), minute ventilation (E), breathing frequency) were collected on a breath-by-breath basis throughout all cycling conditions using a metabolic cart (Quark CPET; COSMED, Rome, Italy). Heart rate (HR) data were collected continuously using radiotelemetry (Garmin International, Schaffhausen, Switzerland). Blood lactate concentration ([Lac]b) was measured at baseline, after every 6 min of work (immediately after each neuromuscular assessment), and at task failure using a capillary sample collected via a finger pinprick. The blood sample was immediately analyzed using a portable lactate analyzer (SensLab Gmb, Lepzig, Germany).

Perceptual measures

A multifactor perceptual assessment evaluating effort, pain, fatigue, and dyspnea was conducted at baseline (measured in the final 10 s of the 20-W warm-up protocol), every 3 min of work, at task failure, and in each condition (every three intervals in HIIT1min; every interval in HIIT3min; after 3 min in CL). At each stage, the order in which the perceptual factors were evaluated remained constant and began with effort, followed by pain, fatigue, and dyspnea, respectively. Perceived Effort was defined to participants as how hard and strenuous the exercise task was and assessed using the 6–20 Borg scale (6 represented no exertion, and 20 represented extremely hard exertion) (26), with the verbal cue “how much is your perceived effort?” (27). This measure reflects the corollary discharge of central motor output driving exercising muscles (28). Fatigue (Fatigue) was defined to participants as how exhausted and/or unable they were to continue physical exercise and was quantified using a 0–10 fatigue scale (0 represented no fatigue, and 10 represented total fatigue and exhaustion) with the verbal cue “how fatigued do you feel?” (29). This measure reflects an integration of physiological, neurobiological, and motivational alterations during exercise (18,29). Leg pain (Pain) was assessed using 0–10 visual analog scale (0 represented no leg pain, and 10 represented worst possible exercise-induced leg pain) (30), and participants were instructed to rate the sensations of aching and burning within their legs with the verbal cue “how much leg pain do you feel?” (30). Pain reflects the activation of group III/IV nociceptive afferents that are triggered by intramuscular mechanical (i.e., pressure) and metabolic stimuli during exercise (31). Breathlessness (Dyspnea) was quantified using 0–10 dyspnea scale (0 represented no breathlessness, and 10 represented worst possible exercise induced breathlessness) (32) whereby participants were asked to rate how breathless they felt during the task with the verbal cue “how breathless do you feel?” (32). Dyspnea is associated with the integration of stimuli from the respiratory centers within the brainstem, corollary discharge of central motor drive, and afferent feedback from respiratory muscles (33). This multifactor perceptual assessment provides an integrative perspective about the psychophysiological constructs that determine exercise tolerance during tasks (19).

Data Analysis

Neuromuscular analysis

Force developed during an MVC and the maximal root mean square (RMS) EMG (of the vastus lateralis) were defined as their respective maximal values during the 500 ms before superimposed PNS. Maximal RMS during the MVC was normalized to the single-twitch maximal M-wave (Mmax) peak to peak amplitude (RMS·Mmax−1). Voluntary activation (VA) was calculated with the modified interpolated twitch technique formula


where FB is the maximum MVC force before superimposed PNS, D is the difference between FB and 100Db stimulus, Fmax is the MVC force, and FPt is the force produced from the potentiated 100Db twitch stimulus. Low-frequency fatigue (LFF) was defined as 10Db stimulus normalized to 100Db stimulus (Db10:100). To determine an estimated work-matched value for each participant, neuromuscular and perceptual metrics during the three conditions were interpolated as a function of time to task failure of the shortest session (34). For all participants, the shortest session was CL. Because perceptual and neuromuscular assessments were performed every 3 and 6 min, respectively, a linear interpolation between the two nearest time point around task failure for the CL condition was used to calculate work-matched values for HIIT1min and HIIT3min conditions. Values corresponding to 0% (baseline) and 100% of time to task failure in the shortest session (work-matched) were averaged and analyzed with raw time to task failure values.

Cardiorespiratory and metabolic analysis

The cardiorespiratory variables were first cleaned by removing data points laying more than 3 SD from the mean of a monoexponential function. Cardiorespiratory variables were then 1 s interpolated. At the time points of interest, cardiorespiratory and metabolic data were averaged into 10-s bins. To compare cardiorespiratory measures at a time point not influenced by the work-to-rest ratio, the 10-s average bin associated with 60 s of exercise for cardiorespiratory variables (Time60) was implemented as this time point provided common cardiorespiratory and metabolic characteristics between conditions. Next, the peak values for cardiorespiratory variables were defined as the highest 10-s average value and were collected in HIIT1min and HIIT3min conditions during the final fully completed interval (HIITfinal) to account for scenarios where participants reached task failure before the end of the interval and cardiorespiratory variables might not have fully developed, and during CL, peak values were collected at the highest 10-s rolling average throughout the entire task (also abbreviated to HIITfinal for comparison across conditions). For the analysis of [Lac]b, to provide an estimate of [Lac]b at a work-matched value, within each subject, [Lac]b was linearly interpolated as a function of time of the shortest session (CL). Values corresponding with 0% (baseline) and 100% of time to task failure in the shortest session (work-matched) were analyzed with raw task failure values (as previously done by Aboodarda et al. (34)).

Statistical Analysis

Descriptive analysis is presented as mean ± SD. Statistical analyses were conducted using GraphPad Prism version 8.0.0 (GraphPad Software V9.4, San Diego, CA). Shapiro–Wilk and Mauchly tests were implemented to assess data normality and sphericity for dependent variables, with a Greenhouse–Geisser correction applied when sphericity was violated. One-way repeated-measures ANOVA with Tukey post hoc analysis were used to compare work accomplished in each condition (HIIT1min, HIIT3min, CL). Two-way repeated-measures ANOVA with Tukey post hoc analysis was used to assess (i) performance fatigability, perceptual, and [Lac]b for the three experimental conditions (i.e., HIIT1min, HIIT3min, and CL) at three time points (baseline, work-matched, and task failure), and (ii) cardiorespiratory outcomes for the three conditions (i.e., HIIT1min, HIIT3min, and CL) at two time points (Time60, HIITfinal). A Partial eta squared (ηp2) was implemented for each ANOVA comparison (i.e., small, <0.02; medium, 0.02–0.26; large, >0.26) (35). A two-way repeated-measures ordinal regression with cumulative link mixed model and Tukey post hoc analysis were used to assess perceptual measures for three experimental conditions and three time points (baseline, work-matched, and task failure); this was implemented to adjust for violations in data normality. Effect size was interpreted using Hedges’ g for paired t-tests (i.e., small, <0.2; medium, <0.8; large, >0.8) (36).


Task duration outcomes

The work done in the three cycling tasks (HIIT1min, HIIT3min, CL) was different (F2,22 = 25.50, P < 0.001, ηp2 = 0.700), whereby greater work was accomplished in HIIT1min (8.46 ± 5.12 kJ) than HIIT3min (1.93 ± 0.71 kJ; P < 0.001, g = 1.78) and CL (1.08 ± 0.36 kJ); P < 0.001, g = 2.03; Fig. 2A). There was no difference between HIIT3min and CL conditions (P = 0.733). In other words, the time to task failure was longer in HIIT1min (50.8 ± 34.1 min) than HIIT3min (7.9 ± 2.1 min) and CL (4.4 ± 1.0 min).

Neuromuscular responses at baseline, work-matched, and task failure for the three experimental conditions. The total amount of work completed in each condition (A), MVC (B), vastus lateralis RMS normalized to muscle compound action potential (RMS·M max −1) (C), VA (D), single-twitch force (E), and Db10:100 (the ratio representing LFF) (F). Experimental conditions included HIIT protocol consisting of a 1-min on-phase and a 1-min rest phase (HIIT1min), HIIT protocol consisting of a 3-min on-phase and a 3-min rest phase (HIIT3min), and CL. aAll conditions different from baseline value. bHIIT1min different from work-matched value. cHIIT3min different from work-matched value. dHIIT1min different from baseline value. *Different between all conditions. $HIIT1min different from HIIT3min and CL. $HIIT1min different from CL.

Neuromuscular function underpinning performance fatigability

Neuromuscular measures at baseline, work-matched, and task failure time points for HIIT1min, HIIT3min, and CL conditions are shown in Figure 2. No differences were found for any measure at baseline. There was an interaction effect for MVC force (F4,44 = 6.819, P < 0.001, ηp2 = 0.383), where at work-matched time point, a greater reduction in MVC force was evident after HIIT3min (P = 0.001, g = 0.19) and CL (P < 0.001, g = 0.46) compared with HIIT1min (Fig. 2B). No differences were found at task failure (P > 0.481).

There was an interaction effect for twitch force (F4,44 = 7.224, P < 0.001, ηp2 = 0.396; see Supplemental Table 1, Supplemental Digital Content, Summary of main and interaction ANOVA effects for neuromuscular fatigue,, whereby at the work-matched time point, HIIT1min produced less reduction in force in comparison to both HIIT3min (P < 0.001, g = 0.76) and CL (P < 0.001, g = 2.09), and HIIT3min engendered less reduction than CL (P = 0.004, g = 0.71). At task failure, a lesser reduction in twitch force developed in HIIT1min compared with CL (P = 0.021, g = 0.54; Fig. 2E).

With regard to Db10:100 (i.e., the index of LFF), a significant interaction effect (F4,44 = 12.52, P < 0.001, ηp2 = 0.532) demonstrated that at the work-matched time point, HIIT1min resulted in less decline of Db10:100 compared with both HIIT3min (P = 0.004, g = 0.55) and CL (P < 0.001, g = 1.44), and HIIT3min elicited less decline Db10:100 than CL (P < 0.001, g = 0.71). No differences between conditions were found for Db10:100 at task failure (P > 0.160; Fig. 2F).

An interaction effect for VA (F1.342,14.76 = 5.217, P = 0.029, ηp2 = 0.322) was present, demonstrating that, although no differences between conditions were found at the work-matched time point (P > 0.540), at task failure, VA for HIIT1min was lower than both HIIT3min (P = 0.043, g = 1.06) and CL (P = 0.010, g = 0.94) conditions (Fig. 2D).

With respect to RMS·Mmax−1 for the vastus lateralis muscle, a significant main effect of time (F2,22 = 8.365, P = 0.0020, ηp2 = 0.432) was present without condition (P = 0.188) or interaction (P = 0.184) effects. The RMS·Mmax−1 was higher at the work-matched time point (P = 0.002, g = 42.59) and task failure (P = 0.020, g = 38.75) compared with baseline (Fig. 2C).

Perceptual responses

Perceptual responses across time for HIIT1min, HIIT3min, and CL conditions are presented in Figure 3. There was a significant interaction effect for Perceived Effort (χ2(4) = 72.93, P < 0.001), Pain (χ2(4) = 32.52, P < 0.001), Dyspnea (χ2(4) = 99.16, P < 0.001), and Fatigue (χ2(4) = 48.76, P < 0.001).

Perceptual responses at baseline, work-matched, and task failure for each condition. Perceived Effort (A), Pain (B), Dyspnea (C), and Fatigue (D) were collected during HIIT protocol consisting of a 1-min on-phase and a 1-min rest phase (HIIT1min), HIIT protocol consisting of a 3-min on-phase and a 3-min rest phase (HIIT3min), and CL. aAll conditions different from baseline value. bHIIT1min and HIIT3min different from work-matched values. *Different between all conditions. #HIIT1min different from HIIT3min.

At work-matched, compared with CL, HIIT3min resulted in lower Perceived Effort (P < 0.001, g = 1.56), Pain (P < 0.001, g = 1.14), Dyspnea (P < 0.001, g = 2.44), and Fatigue (P < 0.001, g = 1.69). In addition compared with CL, HIIT1min elicited lower Perceived Effort (P < 0.001, g = 5.15), Pain (P < 0.001, g = 2.65), Dyspnea (P < 0.001, g = 6.38), and Fatigue (P < 0.001, g = 3.55). Compared with HIIT3min at work-matched, HIIT1min resulted in lower Perceived Effort (P < 0.001, g = 1.71), Pain (P < 0.001, g = 1.33), Dyspnea (P < 0.001, g = 2.34), and Fatigue (P < 0.001, g = 1.29; Fig. 3).

At task failure, Fatigue was greater in the HIIT1min condition compared with HIIT3min (P = 0.039, g = 0.89). No differences were present task failure between conditions for Perceived Effort, Pain, and Dyspnea (P > 0.676; Fig. 3).

Cardiorespiratory and metabolic responses

The two-way ANOVA showed an interaction effect for [Lac]b (F4,44 = 16.58, P < 0.001, ηp2 = 0.601; see Supplemental Table 2, Supplemental Digital Content, Summary of main and interaction ANOVA effects for cardiorespiratory responses, At work-matched, [Lac]b was significantly higher in CL (13.58 ± 3.13 mmol·L−1) compared with HIIT3min (10.23 ± 2.78 mmol·L−1; P < 0.001, g = 1.13) and HIIT1min (7.52 ± 2.05 mmol·L−1; P < 0.001, g = 2.29). Furthermore, at work-matched, HIIT3min elicited higher [Lac]b than in HIIT1min (P < 0.001, g = 1.11). At task failure, [Lac]b was significantly higher in HIIT3min (13.79 ± 3.77 mmol·L−1) and CL (13.58 ± 3.13 mmol·L−1) compared with HIIT1min (10.1 ± 5.12 mmol·L−1; P < 0.001, g = 0.82; P < 0.001, g = 0.82). No differences occurred at task failure between HIIT3min and CL conditions (P = 0.928).

The cardiorespiratory responses are presented in Table 1. The group V̇O2peak from the ramp incremental protocol was 3.35 ± 0.42 L·min−1. With regard to V̇O2, an interaction effect (F2,22 = 15.03, P < 0.001, ηp2 = 0.577) demonstrated that no differences existed between conditions at Time60. However, at HIITfinal, V̇O2 was lesser in HIIT1min (81% ± 12% V̇O2peak) compared with both HIIT3min (94% ± 10% V̇O2peak; P < 0.001, g = 0.97) and CL (92% ± 11% V̇O2peak; P < 0.001, g = 0.73).

TABLE 1 - Group mean and SD (n = 12) for cardiorespiratory responses recorded at 60 s of exercise (Time60s) and the peak value during the last completed HIIT protocols.
Condition Time V̇O2 (L·min−1) E (L·min−1) HR (bpm) Breathing Frequency (breaths per minute)
HIIT1min Time60s 2.32 ± 0.33 76 ± 18 145 ± 18 39 ± 6
HIITfinal 2.71 ± 0.48 a 115 ± 25 a 176 ± 12 65 ± 11
HIIT3min Time60s 2.34 ± 0.37 76 ± 20 147 ± 15 36 ± 8
HIITfinal 3.16 ± 0.45 a,b 152 ± 34 a,b 180 ± 12 65 ± 8
CL Time60s 2.31 ± 0.36 74 ± 16 149 ± 15 36 ± 5
HIITfinal 3.11 ± 0.61 a,b 153 ± 38 a,b 179 ± 10 65 ± 11
Boldface indicates a main effect of time from Time60s to HIITfinal.
aAll conditions different from baseline value.
bHIIT1min different from CL.
HIIT1min, HIIT protocol consisting of a 1-min on-phase and a 1-min rest phase; HIIT3min, HIIT protocol consisting of a 3-min on-phase and a 3-min rest phase.

There was an interaction effect for E (F2,22 = 18.93, P < 0.001, ηp2 = 0.632) that presented no differences between conditions at Time60 (P > 0.941). At HIITfinal, E was lesser in HIIT1min than in HIIT3min (P < 0.001, g = 1.24) and CL (P < 0.001, g = 1.16).

With respect to HR, a main effect of time (F1,11 = 8.36, P < 0.001, ηp2 = 0.432) was present without condition (P = 0.112) or interaction (P = 0.696) effects, demonstrating that at HIITfinal, HR is higher than at Time60 (P < 0.001, g = 2.30).

A main effect of time for breathing frequency (F1,11 = 136.1, P < 0.001, ηp2 = 0.925) was present without condition (P = 0.132) or interaction (P = 0.386) effects. Breathing frequency at Time60 was lower than at HIITfinal (P < 0.001, g = 3.37).


This study explored the integrated neuromuscular, perceptual, and cardiorespiratory mechanisms determining exercise tolerance during work-to-rest ratio matched HIIT protocols with differing interval durations. The major findings are as follows: (i) HIIT1min resulted in a greater amount of work being performed compared with both HIIT3min and CL (Fig. 2A); (ii) at the work-matched time point, MVC, twitch force, and LFF demonstrated lesser declines during HIIT1min compared with HIIT3min and CL, and during HIIT3min compared with CL; (iii) at task failure, twitch force declined less in HIIT1min compared with CL, and VA reduced from baseline in HIIT1min compared with HIIT3min and CL, but no other differences in neuromuscular measures occurred at this time point (Fig. 2); and (iv) at the work-matched time point, Pain, Perceived Effort, Fatigue, and Dyspnea exhibited less increase in HIIT1min compared with HIIT3min and CL, and less increase in HIIT3min compared with CL, but these measures were not different between conditions at task failure; except for HIIT1min at the end of which perceived Fatigue was greater than the other conditions (Fig. 3). Together, these findings indicate that mitigated performance and perceived fatigability during shorter work intervals (HIIT1min) permits exercise be tolerated to a greater extent (HIIT3min, CL). Thus, considering such distinct neuromuscular and perceptual responses at both work-matched and task failure time points, the mechanisms of performance and perceived fatigability determining exercise tolerance during short HIIT intervals are similar to those generally observed during heavy- rather than severe-intensity exercise (19,37). Such uncoupling needs to be taken into consideration to improve the characterization of the acute responses and chronic adaptations to HIIT.

Physiological responses at work-matched level

By assessing neuromuscular responses with a minimum time delay, our data demonstrated a greater MVC decline in HIIT3min, and CL conditions compared with HIIT1min at a work-matched level (Fig. 2B). Previous literature indicates that during severe-intensity exercise, performance fatigability is primarily attributed to peripheral neuromuscular factors (38). Our study supports this notion when matching the amount of work completed. Indeed, although all conditions caused reductions in Db10:100 (Fig. 2F) and in single-twitch force (Fig. 2E), this occurred to a greater extent in CL and HIIT3min as compared with HIIT1min, whereas no accompanying decline in VA was evident in any condition at the work-matched level (Fig. 2D). It has been suggested that the accumulation of metabolites such as Pi can impair muscle contractile function during high-intensity exercise (39). Blood lactate concentrations, which are a surrogate measure of metabolite accumulation, support this notion and indicate that metabolic stress was greater in HIIT3min and CL compared with HIIT1min. Therefore, in the absence of compromised muscle membrane excitability (i.e., no decrease in M-wave during the exercise), alterations in contractile machinery, secondary to increased concentration of Pi, are plausible explanations for the greater decline in twitch force and Db10:100 at work-matched stage in HIIT3min and CL compared with HIIT1min (10,40,41). Another factor in MVC decline is the contribution of central neuromuscular perturbations, which has been attributed to decreases in the central motor drive during maximal contraction. Prior investigations have proposed that decreases in VA might be influenced by exercise duration (42–44); however, our study indicates that at a work-matched level, no differences in VA are noted (Fig. 2D). This result suggests that at the work-matched time point, exercise duration might be insufficient to produce central drive attenuation.

When considering perceptual responses, shorter work interval HIIT in our study mitigated the development of Perceived Effort, Pain, Dyspnea, and Fatigue in comparison to CL and HIIT3min (Fig. 3). It can be postulated that reduced metabolic perturbations during shorter HIIT might have (i) alleviated motor unit firing rate and associated corollary discharge determining perceived effort during shorter HIIT (17,45); (ii) mitigated type III/IV afferent fibers stimulation attenuating perceived pain and exercise pressor reflex, resulting in reduced cardiorespiratory stress and perceived dyspnea (46); and (iii) reduced sensory input could limit the extent to which second order sensations such as Fatigue develop during HIIT1min compared with HIIT3min and CL (18). However, these responses are unique to a scenario where the amount of work completed is matched between conditions and responses at task failure (discussed hereinafter) are distinct.

Physiological and perceptual responses at task failure

In the present study, HIIT1min resulted in a fourfold greater amount of work completed compared with HIIT3min and CL (Fig. 2A). Despite this considerable increase in total work, similar levels of MVC force decline were observed at task failure (Fig. 2B). Fiorenza et al. (6) investigated neuromuscular responses to longer and shorter HIIT protocols but did not elaborate on these factors at task failure. Although speculative, it has been surmised that performance fatigability at task failure would be largely determined by peripheral neuromuscular mechanisms (38). For the longer work duration protocols (HIIT3min and CL), our results support this notion, whereby at the absence of reduced VA (Fig. 2D), the entirety of MVC decline at task failure could be attributed to peripheral neuromuscular mechanisms (twitch force; Fig. 2E). However, for the shorter work duration (HIIT1min), our findings argue that a combination of both central and peripheral mechanisms contributed to MVC decline as less decline in twitch force was accompanied by a significant decrease in VA compared with HIIT3min and CL. This intriguing result indicates that despite performing HIIT protocols with the same work-to-rest ratio, shorter work and rest intervals can facilitate longer overall exercise duration. Furthermore, this HIIT protocol may modulate metabolic perturbations in a manner that is reflective of the heavy- rather than severe-intensity domain (7), and in doing so, it diminishes modulations of peripheral neuromuscular function compared with longer work interval HIIT protocols.

Perceptual responses at task failure indicate that perceived Fatigue was greater in HIIT1min (Fig. 3D), whereas no differences were observed for Perceived Effort, Pain, or Dyspnea between conditions (Fig. 3). The production of maximal or near maximal Perceived Effort in all conditions suggests that this factor might act as a holistic attribute determining exercise tolerance during HIIT and high-intensity constant load exercise alike. However, considering that Pain, Dyspnea, and Fatigue did not reach near maximal values in any condition, it can be postulated that no single perceptual metric determines task failure in HIIT and that the combined advancement of these variables along with cardiorespiratory and neuromuscular factors could contribute toward the eventual attainment of the sensory tolerance limit (17) and eventually characterization of task failure. In addition, it can be hypothesized that increased exercise duration might be responsible for enhanced perceived Fatigue in HIIT1min, and that the development of perceived Fatigue facilitates task disengagement. Furthermore, our study demonstrated that, although no differences in dyspnea occurred between conditions at task failure, E was instead lower in HIIT1min. This illustrates that the perception of dyspnea can be uncoupled from the absolute ventilatory demand. An alteration in the pattern of respiration could indicate that respiratory muscle fatigue occurred in exercise with longer task duration (i.e., HIIT1min) and exacerbated the sensation of breathlessness (47,48).

Implications for “exercise intensity” and prescription of HIIT

Davies et al. (7) have elegantly demonstrated that, for the same work rate, the metabolic fluctuations are considerably reduced if HIIT intervals are shorter. By integrating measures of performance and perceived fatigability, herein we substantiate these findings by showing that the reduced metabolic perturbations result in less decline of twitch force, lower perceptual strain, and extended exercise tolerance (~4-fold) compared with longer HIIT intervals and constant-load exercise. As anticipated by Davies et al. (7), these findings demonstrate a substantial dissociation between the external load and the mechanisms of performance and perceived fatigability responsible for exercise intolerance. Indeed, despite performing a severe-intensity work rate, task failure during HIIT1min did not culminate with the attainment of V̇O2peak and the magnitude of the corresponding reduction in twitch force was smaller compared with HIIT3min and CL exercise. These physiological profiles are very much in line with profiles generally reported for heavy- rather than severe-intensity exercise (7), highlighting that the amplitudes of metabolic stress, rather than the relative work rate, are the main determinant of the mechanisms and progression of performance and perceived fatigability.

Considering that the metabolic fluctuations and the attributes of performance fatigability are strictly linked to the metabolic signaling for adaptations (10), the prescription of HIIT should take into consideration the dynamics discussed herein. For instance, given their lower metabolic fluctuations, shorter HIIT intervals may not be as effective as their longer counterparts for driving skeletal muscle adaptations (49). On the other hand, the better-preserved metabolic stability during such protocols makes them more suitable for unconditioned and/or clinical populations, whose premature achievement of intolerable levels of perceived exertion often precludes them from accumulating large volumes of exercise training (50).

Methodological considerations

Previous studies have demonstrated that performance fatigability might be influenced by sex hormonal changes typically fluctuating across the menstrual cycle (21). Coincidentally, all the women who participated in this study were regularly menstruating monophasic oral contraceptive users (tested during 3 wk of active pill phase) or had a hormonal intrauterine device. Accordingly, this study did not test eumenorrheic women or oral contraceptive users during the nonactive pill phase. Therefore, although the results can be widely applied, they cannot be generalized to eumenorrheic females.


This study demonstrates that the greater exercise tolerance during shorter versus longer work interval HIIT can be attributed to attenuated neuromuscular and perceptual alterations occurring when reducing the work interval duration. Crucially, some of the neuromuscular (e.g., twitch force and VA) and perceptual indices of performance and perceived fatigability were different between HIIT1min, and HIIT3min at task failure. This finding demonstrates that the neuromuscular and perceptual determinants of exercise capacity might differ with respect to the duration of the work interval during HIIT, and mechanisms underpinning exercise tolerance during shorter intervals seem more aligned to those usually allotted to heavy- rather than severe-intensity exercise. Collectively, these findings have important implications to further our understanding of the physiological responses and adaptations to HIIT.

We would like to thank the participants of this study. The results of this study are presented clearly, honestly and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine. This project was funded by the Natural Sciences and Engineering Research Council of Canada (RGPIN-439 2020-07075).

The authors declare no conflicts of interest.


1. Laursen PB, Shing CM, Peake JM, Coombes JS, Jenkins DG. Influence of high-intensity interval training on adaptations in well-trained cyclists. J Strength Cond Res. 2005;19(3):527–33.
2. Laursen PB, Jenkins DG. The scientific basis for high-intensity interval training: optimising training programmes and maximising performance in highly trained endurance athletes. Sports Med. 2002;32(1):53–73.
3. Ross LM, Porter RR, Durstine JL. High-intensity interval training (HIIT) for patients with chronic diseases. J Sport Health Sci. 2016;5(2):139–44.
4. Foster C, Farland CV, Guidotti F, et al. The effects of high intensity interval training vs steady state training on aerobic and anaerobic capacity. J Sports Sci Med. 2015;14(4):747–55.
5. Tschakert G, Hofmann P. High-intensity intermittent exercise: methodological and physiological aspects. Int J Sports Physiol Perform. 2013;8(6):600–10.
6. Fiorenza M, Hostrup M, Gunnarsson TP, et al. Neuromuscular fatigue and metabolism during high-intensity intermittent exercise. Med Sci Sports Exerc. 2019;51(8):1642–52.
7. Davies MJ, Benson AP, Cannon DT, et al. Dissociating external power from intramuscular exercise intensity during intermittent bilateral knee-extension in humans. J Physiol. 2017;595(21):6673–86.
8. Broxterman RM, Craig JC, Smith JR, et al. Influence of blood flow occlusion on the development of peripheral and central fatigue during small muscle mass handgrip exercise. J Physiol. 2015;593(17):4043–54.
9. Blain GM, Mangum TS, Sidhu SK, et al. Group III/IV muscle afferents limit the intramuscular metabolic perturbation during whole body exercise in humans. J Physiol. 2016;594(18):5303–15.
10. Hureau TJ, Broxterman RM, Weavil JC, Lewis MT, Layec G, Amann M. On the role of skeletal muscle acidosis and inorganic phosphates as determinants of central and peripheral fatigue: a 31P-MRS study. J Physiol. 2022;600(13):3069–81.
11. Enoka RM, Duchateau J. Translating fatigue to human performance. Med Sci Sports Exerc. 2016;48(11):2228–38.
12. Twomey R, Aboodarda SJ, Kruger R, Culos-Reed SN, Temesi J, Millet GY. Neuromuscular fatigue during exercise: methodological considerations, etiology and potential role in chronic fatigue. Neurophysiol Clin. 2017;47(2):95–110.
13. Kluger BM, Krupp LB, Enoka RM. Fatigue and fatigability in neurologic illnesses: proposal for a unified taxonomy. Neurology. 2013;80(4):409–16.
14. Silva JCG, Domingos-Gomes JR, Freitas EDS, et al. Physiological and perceptual responses to aerobic exercise with and without blood flow restriction. J Strength Cond Res. 2021;35(9):2479–85.
15. Spranger MD, Krishnan AC, Levy PD, O’Leary DS, Smith SA. Blood flow restriction training and the exercise pressor reflex: a call for concern. Am J Physiol Heart Circ Physiol. 2015;309(9):H1440–52.
16. Pageaux B, Angius L, Hopker JG, Lepers R, Marcora SM. Central alterations of neuromuscular function and feedback from group III–IV muscle afferents following exhaustive high-intensity one-leg dynamic exercise. Am J Physiol Regul Integr Comp Physiol. 2015;308(12):R1008–20.
17. Hureau TJ, Romer LM, Amann M. The ‘sensory tolerance limit’: a hypothetical construct determining exercise performance? Eur J Sport Sci. 2018;18(1):13–24.
18. Greenhouse-Tucknott A, Butterworth JB, Wrightson JG, Harrison NA, Dekerle J. Toward the unity of pathological and exertional fatigue: a predictive processing model. Cogn Affect Behav Neurosci. 2022;22(2):215–28.
19. Iannetta D, Zhang J, Murias JM, Aboodarda SJ. Neuromuscular and perceptual mechanisms of fatigue accompanying task failure in response to moderate-, heavy-, severe-, and extreme-intensity cycling. J Appl Physiol (1985). 2022;133(2):323–34.
20. Warburton DER, Jamnik VK, Bredin SSD, Gledhill N. The Physical Activity Readiness Questionnaire for Everyone (PAR-Q+): English North America version. Health Fit J Can. 2011;4(2):18–20.
21. Ansdell P, Brownstein CG, Škarabot J, et al. Menstrual cycle-associated modulations in neuromuscular function and fatigability of the knee extensors in eumenorrheic women. J Appl Physiol (1985). 2019;126(6):1701–12.
22. Doyle-Baker D, Temesi J, Medysky ME, Holash RJ, Millet GY. An innovative ergometer to measure neuromuscular fatigue immediately after cycling. Med Sci Sports Exerc. 2018;50(2):375–87.
23. Zhang J, Iannetta D, Alzeeby M, MacInnis MJ, Aboodarda SJ. Exercising muscle mass influences neuromuscular, cardiorespiratory, and perceptual responses during and following ramp-incremental cycling to task failure. Am J Physiol Regul Integr Comp Physiol. 2021;321(2):R238–49.
24. van Melick N, Meddeler BM, Hoogeboom TJ, Nijhuis-van der Sanden MWG, van Cingel REH. How to determine leg dominance: the agreement between self-reported and observed performance in healthy adults. PLoS One. 2017;12(12):e0189876.
25. Aboodarda SJ, Mira J, Floreani M, et al. Effects of endurance cycling training on neuromuscular fatigue in healthy active men. Part II: corticospinal excitability and voluntary activation. Eur J Appl Physiol. 2018;118(11):2295–305.
26. Borg GAV. Psychophysical bases of perceived exertion. Med Sci Sports Exerc. 1982;14(5):377–81.
27. Mancora SM. Effort: Perception of. In: Goldstein EB, editor. Encyclopedia of Perception of Effort. Thousand Oaks (LA): Sage Publications Inc.; 2010. pp. 380–3.
28. de Morree HM, Marcora SM. Psychobiology of perceived effort during physical tasks. In: Handbook of Biobehavioral Approaches to Self-Regulation. New York, NY: Springer; 2015. pp. 255–70.
29. Micklewright D, St Clair Gibson A, Gladwell V, Al Salman A. Development and validity of the rating-of-fatigue scale. Sports Med. 2017;47(11):2375–93.
30. McCormack HM, Horne DJ, Sheather S. Clinical applications of visual analogue scales: a critical review. Psychol Med. 1988;18(4):1007–19.
31. Pollak KA, Swenson JD, Vanhaitsma TA, et al. Exogenously applied muscle metabolites synergistically evoke sensations of muscle fatigue and pain in human subjects. Exp Physiol. 2014;99(2):368–80.
32. Borg E, Borg G, Larsson K, Letzter M, Sundblad BM. An index for breathlessness and leg fatigue. Scand J Med Sci Sports. 2010;20(4):644–50.
33. Laviolette L, Laveneziana P; ERS Research Seminar Faculty. Dyspnoea: a multidimensional and multidisciplinary approach. Eur Respir J. 2014;43(6):1750–62.
34. Aboodarda SJ, Iannetta D, Emami N, Varesco G, Murias JM, Millet GY. Effects of pre-induced fatigue vs. concurrent pain on exercise tolerance, neuromuscular performance and corticospinal responses of locomotor muscles. J Physiol. 2020;598(2):285–302.
35. Bakeman R. Recommended effect size statistics for repeated measures designs. Behav Res Methods. 2005;37(3):379–84.
36. Cohen J. Statistical Power Analysis for the Behavioural Sciences. Hillsdale (NJ): Lawrence Erlbaum Associates; 1988. p. 567.
37. Brownstein CG, Metra M, Sabater Pastor F, Faricier R, Millet GY. Disparate mechanisms of fatigability in response to prolonged running versus cycling of matched intensity and duration. Med Sci Sports Exerc. 2022;54(5):872–82.
38. O’Leary TJ, Collett J, Howells K, Morris MG. Endurance capacity and neuromuscular fatigue following high- vs moderate-intensity endurance training: a randomized trial. Scand J Med Sci Sports. 2017;27(12):1648–61.
39. O’Connell JM, Weir JM, MacIntosh BR. Blood lactate accumulation decreases during the slow component of oxygen uptake without a decrease in muscular efficiency. Pflugers Arch. 2017;469(10):1257–65.
40. Cheng AJ, Place N, Westerblad H. Molecular basis for exercise-induced fatigue: the importance of strictly controlled cellular Ca2+ handling. Cold Spring Harb Perspect Med. 2018;8(2):a029710.
41. Sundberg CW, Fitts RH. Bioenergetic basis of skeletal muscle fatigue. Curr Opin Physiol. 2019;10:118–27.
42. Krüger RL, Aboodarda SJ, Jaimes LM, MacIntosh BR, Samozino P, Millet GY. Fatigue and recovery measured with dynamic properties versus isometric force: effects of exercise intensity. J Exp Biol. 2019;222(Pt 9):jeb197483.
43. Krüger RL, Aboodarda SJ, Jaimes LM, Samozino P, Millet GY. Cycling performed on an innovative ergometer at different intensities–durations in men: neuromuscular fatigue and recovery kinetics. Appl Physiol Nutr Metab. 2019;44(12):1320–8.
44. Froyd C, Beltrami FG, Millet GY, Noakes TD. Central regulation and neuromuscular fatigue during exercise of different durations. Med Sci Sports Exerc. 2016;48(6):1024–32.
45. Williamson JW, McColl R, Mathews D, Mitchell JH, Raven PB, Morgan WP. Hypnotic manipulation of effort sense during dynamic exercise: cardiovascular responses and brain activation. J Appl Physiol (1985). 2001;90(4):1392–9.
46. White M, Bruce R. The role of muscle mechano and metaboreflexes in the control of ventilation: breathless with (over) excitement? Exp Physiol. 2020;105(12):2250–3.
47. Aldrich TK. Respiratory muscle fatigue. Clin Chest Med. 1988;9(2):225–36.
48. Romer LM, Polkey MI. Exercise-induced respiratory muscle fatigue: implications for performance. J Appl Physiol (1985). 2008;104(3):879–88.
49. Fiorenza M, Gunnarsson TP, Hostrup M, et al. Metabolic stress–dependent regulation of the mitochondrial biogenic molecular response to high-intensity exercise in human skeletal muscle. J Physiol. 2018;596(14):2823–40.
50. White LJ, Dressendorfer RH. Exercise and multiple sclerosis. Sports Med. 2004;34(15):1077–100.


Supplemental Digital Content

Copyright © 2022 by the American College of Sports Medicine