Optimizing Muscle Activation in Cadets Using Electromyography Biofeedback During Anti-G Training
INTRODUCTION: Effective execution of the anti-G straining maneuver (AGSM) is essential for pilots to maintain consciousness under high gravitational forces (+Gz). This study evaluated whether electromyographic (EMG) biofeedback enhances muscle activation patterns during AGSM training in novice cadets.
METHODS: There were 58 Brazilian Air Force cadets (age: 25 ± 1 yr) who performed two AGSM sessions involving sustained submaximal isometric contractions of the gastrocnemius, vastus medialis, and rectus abdominis muscles, synchronized with rhythmic breathing every 3 s. Subjects completed trials under counterbalanced visual EMG feedback conditions (real-time visualization vs. no visualization) and were randomly assigned to verbal feedback conditions (instructor guidance vs. no guidance). EMG signals were recorded at 1500 Hz and normalized to each subject’s peak amplitude during the AGSM trials.
RESULTS: Muscle-specific responses to feedback were observed. Verbal feedback enhanced gastrocnemius activation but reduced vastus medialis activation. Combined visual and verbal feedback produced the highest activation in the rectus abdominis. Visual feedback alone had minimal effect across all muscles. Despite submaximal instructions, brief peak activations were sufficient for normalization.
DISCUSSION: EMG biofeedback facilitated motor learning of AGSM by selectively improving activation in targeted muscles. However, effects varied by muscle group, suggesting the need for tailored instructional strategies. Although task-based normalization offers ecological validity, it may limit comparisons with MVC-based protocols. Incorporating EMG biofeedback may enhance AGSM training, particularly in novice populations or settings without centrifuge access.
Massaferri R, Calvo APC, Coutinho ABB, Guimarães TT, Farinatti P. Optimizing muscle activation in cadets using electromyography biofeedback during anti-G training. Aerosp Med Hum Perform. 2025; 96(11):985–992.
Effective training in the anti-G straining maneuver (AGSM) is essential for pilots and crew operating high-performance aircraft. This training enables them to withstand increased gravitational forces (+Gz), reducing the risk of +Gz-induced loss of consciousness (G-LOC)—a condition that occurs when excessive acceleration prevents sufficient oxygen from reaching brain tissues. Under high-G conditions, blood is rapidly displaced toward the lower extremities due to the hydrostatic force of acceleration, resulting in a substantial drop in cerebral perfusion pressure. This hemodynamic shift compromises oxygen delivery to the brain despite adequate systemic oxygenation. Simultaneously, the heart struggles to maintain sufficient blood flow against the gravitational gradient, further reducing systemic arterial pressure. If not counteracted, this cascade can lead to visual disturbances and, ultimately, loss of consciousness.1,2
The AGSM is a physiological countermeasure designed to prevent this hemodynamic collapse. It combines sustained isometric contractions of the lower body—primarily the calf, thigh, and abdominal muscles—with cyclical forced exhalations against a closed glottis (typically every 3 s). The muscle strain increases peripheral vascular resistance, reducing venous pooling in the lower limbs, while the respiratory component elevates intrathoracic pressure, enhancing cardiac output and arterial pressure. Together, these mechanisms help preserve cerebral perfusion and prevent G-LOC during high +Gz exposure.3,4
Electromyography (EMG) offers a promising approach for quantifying muscle activation during AGSM, providing an objective measure of muscular activation. Implementing EMG-based AGSM training—without the use of a centrifuge—presents a cost-effective alternative to traditional ground-based AGSM training methods focused on technique improvement. It is important to clarify that this approach is not intended to replace centrifuge-based training but rather to complement it, particularly by offering enhanced feedback for neuromuscular control and technique refinement. Previous studies have demonstrated EMG’s utility in assessing muscle fatigue during simulated air combat maneuvers3 and in evaluating AGSM techniques.5 Additionally, research suggests that lumbar support may enhance AGSM effectiveness by influencing muscle activity, as measured by EMG.6 Notably, electromyographic activity in the gastrocnemius muscle has been identified as a potential early indicator of G-LOC7,8 and a strong correlation between muscular contraction and +Gz tolerance has been established.9
When AGSM training is applied to inexperienced individuals, such as cadets, the pedagogical focus should prioritize motor control development over maximal performance. Motor learning literature emphasizes that the early stages of skill acquisition benefit from strategies aimed at improving neuromuscular coordination and sustained muscle engagement, rather than maximum contraction efforts.10,11 This approach is particularly relevant for AGSM, a complex task requiring simultaneous isometric contractions and respiratory maneuvers.12,13
This study aimed to evaluate EMG as a training tool for improving AGSM technique. We hypothesized that the provision of verbal feedback, visual feedback, or both would produce significant differences in EMG-measured activation of key muscle groups (medial gastrocnemius, vastus medialis, and rectus abdominis) during AGSM execution among novice cadets. Specifically, we expected that feedback conditions would lead to higher or more stable muscle activation compared to conditions without feedback. By integrating EMG visual biofeedback into AGSM training, this study sought to enhance motor learning and optimize technique. Specifically, we compared the impact of AGSM training with EMG biofeedback to training without it, focusing on differences in muscle activation levels.
METHODS
Subjects
There were 63 male Brazilian Air Force (FAB) cadets, all in their third year of training—the stage at which flight instruction begins at the FAB Academy—initially recruited for this study. There were 5 subjects excluded from the final analysis due to incomplete EMG data or technical problems during data acquisition, resulting in a final sample of 58 cadets (25 ± 1 yr; 75.8 ± 4.2 kg; 178.1 ± 3.8 cm) (Fig. 1). All cadets had previously completed physiological training at the Aerospace Medicine Institute and volunteered for participation. Eligibility criteria required subjects to be medically cleared through the institution’s annual health assessment, while individuals with musculoskeletal injuries that could compromise performance were excluded. This study was approved by the Research Ethics Committee of the Hospital de Força Aérea de São Paulo (Approval Number 4.349.491), and all subjects provided written informed consent in accordance with the Declaration of Helsinki.
Citation: Aerospace Medicine and Human Performance 96, 11; 10.3357/AMHP.6714.2025
Procedure
To assess the impact of different feedback types on EMG activity in the thigh, calf, and abdominal muscles during AGSM training, subjects completed three 30-s sessions, with the first serving as a familiarization trial and the remaining two used for analysis, following a mixed-design approach (between- and within-subject factors) (Fig. 2). All AGSM trials were performed in a controlled laboratory setting, without exposure to actual G-forces or centrifuge-based simulation. Subjects remained seated while executing the simulated AGSM synchronized with rhythmic breathing.
Citation: Aerospace Medicine and Human Performance 96, 11; 10.3357/AMHP.6714.2025
It is important to highlight that this AGSM training represented the first practical experience for all subjects. The protocol was designed with an instructional focus, prioritizing the acquisition of the motor skill necessary to properly activate and sustain the contraction of the target muscle groups, rather than maximizing force output.
Visual feedback involved real-time monitoring of EMG signals displayed on a monitor positioned approximately 1.5 m in front of the subject at eye-level during the AGSM trials. The EMG amplitude envelope (smoothed rectified signal) was shown in microvolts, allowing subjects to visually track their muscle contraction levels and adjust them in real time to maintain stable and continuous activation, as instructed. In contrast, verbal feedback consisted of instructor-provided guidance immediately after each trial. The instructor displayed the subject’s EMG signal on the same monitor and provided individualized feedback, highlighting strengths and areas for improvement with the goal of enhancing performance in the subsequent attempt.
Subjects performed the AGSM under two within-subject conditions in a counterbalanced order: one with real-time EMG visualization and one without visual feedback. Additionally, they were randomly assigned to one of two between-group conditions: receiving verbal feedback from an experienced AGSM instructor or not receiving any instructor input. This design allowed for a comprehensive evaluation of the effects of visual and verbal feedback on muscle activation during AGSM training. Fig. 2 summarizes the experimental design.
Each AGSM training session lasted 30 s, during which subjects performed sustained isometric contractions of the thigh, calf, and abdominal muscles, synchronized with forced breathing every 3 s. A standardized instructional cue was provided to all subjects before the trials: “You need to contract the muscles of your thighs, calves, and abdomen at a submaximal and continuous level throughout the entire maneuver. Simultaneously, you must perform breathing cycles every 3 s, consisting of a forced exhalation, against a closed glottis, followed by a quick inhalation, without releasing the muscle tension at any point. Because the session lasts 30 s, avoid very intense contractions to prevent premature fatigue and ensure you can sustain the maneuver for the full period.” The breathing sequence involved an initial inhalation to approximately 70% of perceived lung capacity, followed by a rapid exhalation and subsequent inhalation every 3 s to maintain lung volume throughout the session. The primary objectives of the AGSM were to: 1) generate and sustain muscle tension; and 2) increase intrathoracic pressure. This physiological mechanism mitigates blood pooling in the lower extremities under high +Gz forces, thereby preserving cardiac output and central perfusion, ensuring pilots remain conscious and maintain optimal operational performance.
EMG data were collected from the gastrocnemius, vastus medialis, and rectus abdominis muscles using a Noraxon DTS system (Noraxon USA Inc., Scottsdale, AZ, USA) at a sampling rate of 1500 Hz, following the SENIAM guidelines.14 The experimental procedure comprised two 30-s AGSM sessions, separated by a 1-min rest period. EMG signals were digitally processed using a 4th-order Butterworth band-pass filter (10–500 Hz) with a 60-Hz notch filter and harmonics to minimize noise.
Muscle activation was analyzed using the windowed normalized area under the curve (nEMG) method, employing 1-s intervals with 50% overlap. EMG amplitude was normalized to the peak activation recorded during each subject’s AGSM trials (self-normalization method, nSELF). For each muscle, EMG amplitude was normalized to the highest value observed during either of the two 30-s AGSM trials—whichever exhibited the greatest voltage. These peaks typically occurred within the initial 5–10 s of the first trial, when subjects were more attentive and less fatigued. Due to the novelty of the task and the subjects’ limited prior experience with AGSM, fluctuations in neuromuscular control were common, occasionally resulting in brief peaks in EMG amplitude suitable for normalization.
This approach was adopted considering the intraindividual comparison design and the complexity of the AGSM, which involves simultaneous contractions of multiple agonist and antagonist muscle groups coordinated with breathing patterns. Performing isolated MVIC tests for each muscle was deemed impractical and not representative of the specific motor demands of AGSM, particularly for novice subjects.
EMG data were normalized using a self-referenced approach, in which each subject’s highest EMG amplitude recorded during the AGSM trials was used as the reference value. This method, while less common than maximal voluntary isometric contraction (MVIC) normalization, was selected for its feasibility and ecological validity within the intrasubject design of this study. Although Slungaard et al.15 successfully applied MVC-based normalization in aerospace research, our approach is supported by EMG methodological literature recommending task-specific normalization for complex motor tasks.16,17 This processing strategy offers high sensitivity in detecting activation patterns during AGSM and is particularly useful for identifying reductions in muscle activity that may precede pilot G-LOC.7 All EMG normalization was performed during postprocessing. An illustrative example of the processed EMG signal from a single subject is shown in Fig. 3 to demonstrate the typical temporal pattern and peak selection used for normalization.
Citation: Aerospace Medicine and Human Performance 96, 11; 10.3357/AMHP.6714.2025
Statistical Analysis
Sample normality was assessed using skewness and kurtosis. A two-way repeated measures ANOVA (2 × 2) was conducted separately for each muscle to analyze changes in normalized EMG (nEMG) during AGSM exercises, with verbal feedback and real-time visual feedback as factors. When significant interactions or main effects were detected, simple main effects analyses were performed as follow-up tests. Effect sizes were reported using partial eta-squared (η2p) and classified as small (0.01–0.04), medium (0.06–0.14), or large (≥0.14). Additionally, η2p values of 0.01–0.039 were interpreted as instructional effects, while values ≥0.039 indicated desirable learning effects. Statistical analyses were conducted using JASP software (Version 0.18.1; University of Amsterdam, Netherlands), with significance set at α = 0.05. Post hoc power analysis (1-β > 0.8) was performed using G*Power (Version 3.1.9.7; Heinrich-Heine-Universität Düsseldorf, Germany).
RESULTS
Fig. 4A presents data of medial gastrocnemius activation. For the calf muscle, no significant main effect of visual feedback on muscle activation was observed (F(1, 1730) = 0.613, P = 0.434, η2p = 0.000354), indicating that real-time EMG visualization alone did not influence activation levels. In contrast, the main effect of verbal feedback was significant (F(1, 1730) = 19.393, P < 0.001, η2p = 0.011), suggesting that instructor guidance played a role in modulating muscle activity. Additionally, a significant interaction between visual and verbal feedback was found (F(1, 1730) = 8.335, P = 0.004, η2p = 0.005), indicating that the presence of verbal feedback influenced the effect of visual feedback conditions differently. Post hoc analyses revealed that, within the verbal feedback group, the “without visual” condition resulted in significantly greater muscle activation compared to all other conditions (P ≤ 0.05).
Citation: Aerospace Medicine and Human Performance 96, 11; 10.3357/AMHP.6714.2025
Fig. 4B presents the results for the vastus medialis activation. The main effect of visual feedback (F(1, 1730) = 1.220, P = 0.270) and the interaction between visual and verbal feedback (F(1, 1730) = 0.511, P = 0.475) were not statistically significant. However, the main effect of verbal feedback was statistically significant (F(1, 1730) = 17.761, P < 0.001, η2p = 0.006), indicating that instructor guidance independently reduced muscle activation, while visual feedback alone or combined had no additional significant effects.
Fig. 4C depicts data obtained for the rectus abdominis muscles. The main effect of visual feedback was significant (F(1, 1730) = 4.696, P = 0.030, η2p = 0.003), suggesting that real-time EMG visualization contributed to increased muscle activation. The main effect of verbal feedback was also significant (F(1, 1730) = 18.019, P < 0.001, η2p = 0.006). Additionally, a significant interaction was observed (F(1, 1730) = 4.548, P = 0.033, η2p = 0.003), indicating that the effectiveness of visual feedback was influenced by the presence of verbal guidance. Post hoc comparisons revealed that conditions involving both visual and verbal feedback led to significantly higher activation compared to conditions without any feedback (P ≤ 0.05).
DISCUSSION
The present study investigated the impact of visual and verbal feedback on EMG activation during the AGSM among cadet aviators. Results indicated muscle-specific responses, with significant interactions showing higher activation in the medial gastrocnemius and rectus abdominis muscles under certain feedback conditions. When analyzed separately, verbal feedback had a greater effect size (η2p = 0.011) compared to visual feedback (η2p ≤ 0.003). Specifically, in the medial gastrocnemius, verbal feedback alone without visual feedback led to the greatest muscle activation, exceeding all other conditions. In contrast, instructor guidance (verbal feedback) independently led to a significant reduction in vastus medialis activation (η2p = 0.006), while visual feedback alone or combined with verbal guidance showed no additional significant effects.
These findings suggest that verbal feedback consistently influenced muscle activation, whereas visual feedback alone had a limited effect. Interestingly, the condition without visual feedback combined with verbal instructions generated higher gastrocnemius activation than all other conditions, indicating a possible interference or attentional distraction caused by simultaneous visual EMG monitoring. Interaction effects observed in the calf and abdominal muscles suggest that verbal feedback enhances the effectiveness of visual feedback, while the vastus medialis activation decreased specifically due to instructor guidance. Our results emphasize the importance of biofeedback in modulating muscle activation differently across muscle groups, particularly highlighting its effectiveness in the gastrocnemius and abdominal muscles, and its distinct, inhibitory influence on vastus medialis activation.
The significant effect of verbal feedback on gastrocnemius activation suggests that verbal guidance enhances motor learning and neuromuscular control specifically for this muscle during AGSM. However, this effect was not consistent across all muscles, as verbal feedback was associated with a reduction in vastus medialis activation, suggesting that the influence of feedback was muscle-specific rather than universal. The unexpected finding that visual feedback diminished the beneficial effect of verbal cues in the gastrocnemius muscle might be explained by cognitive overload or divided attention when visual EMG feedback is presented concurrently with instructor guidance. This result reinforces previous studies that have identified the gastrocnemius as an important muscle involved in AGSM execution and as a potential early indicator of G-LOC vulnerability in other contexts.8 It also suggests caution when implementing combined feedback modalities, highlighting the need for further research on optimal feedback strategies. This concurs with research suggesting that EMG biofeedback can serve as a foundation for developing predictive algorithms for pilot safety in high-G conditions.7
Similarly, the rectus abdominis showed enhanced activation when both feedback modalities were combined, highlighting its role in sustaining intrathoracic and intra-abdominal pressure—essential for maintaining arterial pressure and cerebral perfusion during AGSM. This finding suggests that feedback-driven training may improve AGSM efficacy by reinforcing the respiratory and pressure-generation components critical to G-protection.5 The abdominal muscles play a fundamental role in AGSM execution, contributing to core stabilization and intra-abdominal pressure generation, both of which are essential for sustaining high-G conditions. Our findings agree with previous studies emphasizing lumbar support and core engagement in enhancing AGSM performance,6 reinforcing the premise that multichannel feedback systems optimize motor learning and neuromuscular coordination.18
It is important to clarify that the relatively low mean normalized EMG values observed—ranging between approximately 0.1–0.2—do not reflect poor performance but rather the specific instructional strategy adopted. Subjects were explicitly instructed to perform submaximal yet continuous contractions throughout the 30-s AGSM trials, prioritizing muscle activation control and breathing coordination over maximal effort. This approach aligns with motor learning principles for complex isometric tasks,10,11 particularly when training novice populations, where the focus is on motor control and fatigue management rather than maximal neuromuscular output. Moreover, this study used a submaximal version of the AGSM to maintain sustained contractions over 30 s. Future research should investigate the effectiveness of EMG biofeedback during full-intensity AGSM, where real-time muscle activation data may more clearly identify underperforming muscle groups and optimize training.
While task-specific peak normalization ensures ecological validity and practicality during complex maneuvers like AGSM, it may introduce variability due to differences in subject engagement and motor control. In contrast, MVC normalization offers greater standardization across individuals and studies but lacks task specificity in highly coordinated motor tasks. These differences should be considered when interpreting our findings. An illustrative example of a processed EMG trace from a single subject is presented in Fig. 3 to highlight the temporal variability and support the normalization strategy adopted. Although self-normalization to peak EMG during task execution provides practical feasibility and task-specific comparability, this method is less common than normalization to maximal voluntary contraction (MVC normalization; %MVC) and may hinder cross-study comparisons. Future studies should incorporate MVC-based normalization protocols when experimental conditions allow, to improve generalizability and interpretation.
In contrast, instructor guidance (verbal feedback) significantly reduced vastus medialis activation, while visual feedback alone or combined showed no additional significant effect. This finding may reflect a neuromuscular redistribution strategy resulting from verbal instructions directing attention toward activating less intuitive muscle groups. Previous studies indicate that quadriceps activation during sustained acceleration is relatively stable,19 and according to motor control literature,20 verbal cues emphasizing specific muscle groups can lead to selective recruitment patterns, thereby reducing activation in muscles that are naturally recruited with less effort. Further research should explore whether alternative biofeedback strategies, such as prolonged training or higher-intensity contractions, could enhance quadriceps recruitment during the AGSM maneuver.
Our findings corroborate previous research demonstrating the benefits of EMG biofeedback in neuromuscular training.18,21 Studies on motor learning suggest that real-time feedback enhances skill acquisition, particularly when combined with verbal instruction.22 However, it is important to note that the benefits of EMG biofeedback observed in this study were muscle-specific and context-dependent. While it enhanced activation in key muscles like the gastrocnemius and rectus abdominis, it produced no improvement—or even a reduction—in vastus medialis activity. This suggests that EMG biofeedback can be an effective tool when targeted toward muscle groups that benefit from enhanced activation during AGSM, but it may require adapted strategies for muscles where feedback could inadvertently lead to reduced engagement. The results also support the notion that multichannel feedback systems optimize training outcomes by addressing both conscious and subconscious motor control mechanisms.23
Additionally, the interpretation of effect sizes (η2p) in this study was framed according to the educational approach proposed by Hattie.11 Effect sizes between 0.01–0.039 represent “instructional effects,” meaning moderate but relevant contributions to skill acquisition, while values above 0.039 are considered “desired learning effects,” indicating stronger, meaningful improvements. This framework provides an educationally grounded interpretation of how different feedback modalities influenced AGSM learning in this sample of novice participants.
The integration of EMG biofeedback into AGSM training presents a valuable opportunity to improve pilot performance. By tailoring feedback strategies to specific muscle groups, instructors can refine training protocols to enhance muscle activation patterns critical for high-G tolerance.9 Such approaches align with previous findings highlighting the effectiveness of physiological training in Brazilian Air Force cadets, reinforcing the operational importance of optimized AGSM techniques.24 Consistent use of EMG biofeedback technology allows instructors to swiftly detect and rectify technical deficiencies, enhancing muscle awareness and control, which are essential factors for optimal performance in high-stress, high-G environments. In consequence, they would be able to promote more efficient technique development, optimizing AGSM performance, and reducing the risk of G-LOC.
Despite its contributions, this study presents certain limitations. First, the absence of direct centrifuge testing precludes definitive conclusions regarding the impact of feedback-driven muscle activation on actual G-tolerance. Second, as the sample consisted exclusively of cadet aviators, the findings may not generalize to experienced pilots or broader populations exposed to high-G environments. Third, the long-term retention and durability of biofeedback training effects remain unknown, requiring further investigation into whether the observed neuromuscular adaptations persist under operational stressors or over extended periods. Additionally, the EMG analysis was limited to the gastrocnemius, rectus abdominis, and vastus medialis. While the hamstrings and gluteal muscles are biomechanically relevant to AGSM, they were excluded due to high susceptibility to movement artifacts in the seated position, which compromised signal reliability. Future studies should consider strategies to include these muscle groups. Finally, conducting the study in a controlled, simulated environment may not fully replicate the physiological and psychological challenges encountered in real-world high-G conditions. Future research should include centrifuge testing to directly assess how biofeedback-enhanced muscle activation influences actual G-force endurance. Additionally, further studies are needed to explore the effectiveness of alternative biofeedback strategies to enhance quadriceps activation during AGSM. Finally, the selective influence of verbal feedback observed in the current study suggests that future AGSM training protocols should strategically emphasize or de-emphasize specific muscle groups to optimize performance outcomes.
Our findings suggest that combining verbal and visual feedback enhances muscle activation in key muscle groups involved in AGSM, particularly in the gastrocnemius and abdominal regions. However, quadriceps activation significantly decreased in response to verbal feedback, indicating that instructional strategies must account for muscle-specific recruitment patterns influenced by verbal cues. Although the study has limitations, particularly the absence of direct G-tolerance assessment, the results strongly support the integration of EMG biofeedback into AGSM training programs. Future research should further explore the direct effects of feedback-enhanced muscle activation on G-force endurance and its adaptability in real-world high-G conditions.
Subjects flow diagram. Subjects were randomized into two groups: with or without verbal feedback. Final analyses included 30 subjects in the verbal-feedback group and 28 in the no-verbal-feedback group.
Experimental design. Subjects were randomized into two groups (with or without verbal feedback) and performed two anti-G straining maneuver (AGSM) training sessions, with or without real-time electromyography (EMG) visual feedback, in a counterbalanced order.
Example of a processed electromyography (EMG) signal from a single subject during anti-G straining maneuver (AGSM). The trace illustrates the temporal pattern of activation and highlights the peak amplitude used for normalization. This example supports the rationale for the task-specific normalization strategy. All data were collected in a controlled laboratory setting, without exposure to G-forces. AUC = area under the curve.
EMG activation during the anti-G straining maneuver (AGSM) in different feedback groups (verbal or without verbal) and conditions (visual or without visual): A) medial gastrocnemius muscle, B) rectus femoris muscle, and C) rectus abdominis muscle. **: significant difference between groups (verbal vs. without verbal) (P ≤ 0.05); *: significant difference vs. visual in verbal group (P ≤ 0.05).
Contributor Notes
