19 adult participants (aged from 21 to 26 years old), with 14 male and 5 females, participated in this study. The sample size is larger than the minimum expectation of 15 for human factors validation testing (Food and Drug Administration’s FDA, 2016). The exclusion criterion was prior traumatic injuries in the upper back-neck area. The reason for this criterion was that traumatic injuries often modify the muscular structure of the damaged tissue, and this may induce asymmetric neuromuscular loading and activation (Ball and Scurr, 2011). The exact degree of internal asymmetry is difficult to measure and/or model which can lead to systematic errors in the data. The demographic and basic anthropometry data of the participants were collected prior to the sitting tests (Table 1).

The experiment was conducted on a typical cushioned business aircraft seat, as shown in Figure 1. The seat is equipped with an extendable headrest along the longitudinal direction for its height adjustments. The seat’s cushion surface has little geometric curvature with no additional neck support provided. The seat pan has a tilting angle of 15° from the horizontal direction. The backrest can be inclined, and such flexibility makes the seat capable of simulating various preferred backrest conditions of different in-seat activities, e.g., preferred backrest inclination of 56° for sleeping (Smulders et al., 2016). The backrest inclination angle was tracked by a digital inclinometer (accuracy: ±0.2°). The bed shown in Figure 1 was used for muscle initialization before the sitting test and for conducting the muscle’s maximum voluntary contraction, which is elaborated in Sec 2.3.2.

The muscle activation was measured using surface electromyography (sEMG) sensors (TEA CAPTIV T-Sens EMG with sampling rate of 2048 Hz), whose connected electrode pad is adhesively attached to the designated body area, as shown in Figure 2. A webcam was positioned laterally to the seat to capture the movement and details during the test, i.e., posture adjustments, change of trunk position for different backrest inclination setup, sitting down and standing up, etc. The video tracking of motion details throughout the experiment helps to exclude useless sEMG data during post-experiment data processing and analysis.

Before starting the test, the participant was first briefed about the test procedure and asked to read and sign the consent form. Then, the basic demographic information (age and gender) and anthropometry (stature, weight) of the participant were collected. Next, the sEMG sensors were attached to the corresponding areas for SCM and UT, as explained in Sec 2.3.2.

Afterwards, the participant was asked to lay flat on the bed with pillow (Figure 1) simulating the normal sleeping condition (full relaxation) for 3 min for body initialization before the sitting test, inspired by (Smulders et al., 2019) Then, the participant was asked to sit on the seat with the head naturally leaning against the headrest and the full back in contact with backrest. This requirement is to ensure all participants sit with the same constraint for better test consistency regardless of their personal sitting habits. After the sitting was stabilized, the condition remained static and undisturbed for a short period of time of 1 min to produce regions of consistency in the sensor data. The relatively short stable sitting duration was applied in the test to avoid potential tediousness developed due to multiple trials of each test, which could potentially affect the perceived comfort. The participant was then asked to provide the comfort rating of each muscle area, referring to Sec 2.3.3. The mentioned sitting test procedures were repeated for the backrest inclination of 30°, 40° and 50°, with three trials for each condition. MVC test was conducted at the end of the test for the normalization of interested sEMG data during sitting.

The conducted experiment focused on two muscle groups and was repeated for each subject a total of three times, and the average sEMG and comfort rating values of the trials was used for the analysis. For the muscles’ sEMG data, the irrelevant and unstable data regions were identified and excluded based on the webcam recording, e.g., standing up from the bed and walking to the seat, posture change for inclination adjustment, and wiggling for comfort position.

The sEMG signal was recorded and processed with CAPTIV 2.4.0, the companion software of the sensor system. The root mean square (RMS) value of the recorded rectified raw data was calculated. Then, the signals of SCM of UT during the sitting test with 30°, 40° and 50° backrest inclination were normalized based on the MVC test maximum value, and the muscle activation was expressed as %MVC.

With the obtained data of two muscles’ activations and their corresponding comfort ratings, statistical analysis was performed to identify the association of observed values and influence of human and seat conditions. Figure 4 illustrates the overview of data collection and analysis. Correlation analysis was performed between muscle activation and comfort rating to understand the role of muscle activation in perceived seating comfort in the neck region. In addition, One-way ANOVA were used to test the statistical significance in the means of data groups of different backrest inclinations.

The measured activation and comfort rating of the studied muscles are plotted for each backrest inclination, as shown in Figure 5. Figures 5A,B are the distributions of the subjective levels of comfort reported during experimental trials for the SCM and UT muscle groups, respectively. Each bar represents the average perceived level of comfort on a scale of 1–5, reported by the participant across three trials for each seat recline angle. A high degree of variability of SCM and UT area comfort level across participants is observed. The standard deviation of each inclination condition is displayed in each chart.

Figures 5C,D illustrates the distribution of the average measured sEMG activation percentage across the three experimental trials performed for each recline angle for the SCM and UT muscle groups, respectively. It is evident that the inter-subject variation for the measured sEMG activation percentage is high (standard deviations shown in charts), and there is little change in the measured activation percentage across different backrest inclinations for the same participant.

The data can be generally visualized by plotting the total average of each evaluated comfort parameter against the seat reclination grouping to explore the effect of seat reclination on the studied parameters of comfort and muscle activation (Figure 6; Figure 7).

The following observations can be made about the effects of seat reclination on the evaluated comfort parameters. First, Figure 6 shows that subjects reported a steady increase in the perceived levels of comfort in the SCM region as the seat became more reclined; however, no clear trend can be observed in the perceived level of comfort reported by subjects in the UT region as the seat became more reclined. In Figure 7, relatively constant muscle activations are found for both SCM and UT at different backrest inclination angles. Second, higher comfort ratings were generally reported in SCM compared to that of UT muscle area, suggesting that passengers felt more uncomfortable in the posterior side of the neck when seated, given the test condition of the neck being unsupported. The muscle activation measurement reveals reasonable corresponding phenomenon that UT activation is significantly higher than SCM. The less relaxed muscle may induce discomfort and generate fatigue for long-duration sitting. Due to the unsupported condition of the neck, the neck’s body weight introduces more forces on the headrest, creating excessive flexion torque in the joints in the neck region. This would require the posterior muscles to strain more to keep a balanced body position. The results also imply the efficacy of using neck support to improve upper-body seating comfort.

The collected data of muscle activation (sEMG data) and comfort rating are grouped based on the muscle and the seat’s backrest inclination, respectively, for processing and analysis. To analyze the effect of backrest’s inclination on the evaluated dependent measures (the muscle activation and comfort rating) for each of the muscle groups studied, one-way analysis of variance (ANOVA) study is performed on the experimental data. The one-way ANOVA test helps uncover significant differences in the means of the dependent variable across the three tested inclination conditions (30°, 40° and 50°). From Table 2, the results of the one-way ANOVA study indicate that the null hypothesis is retained for all of the analysis configurations (F < F_crit,p > 0.05). Based on the analysis of 19 participants’ experimental data, backrest inclinations do not exhibit statistically significant differences in subjective comfort rating or muscle activation measurement, which indicates that different backrest inclination statistically shows nonsignificant influence on the participants seating experience in the neck region.

In addition to investigating the impact of the backrest inclination on neck comfort, the measured muscle activation was also tested for correlation with the perceived neck comfort. Such a correlation study based on the intersubject data helps to understand how subjective comfort can be reflected with the objective measure of muscle reaction (EMG signal). The correlation test results are provided in Table 3, consisting of Pearson’s correlation coefficient (R) and the corresponding p-value. From the results, all significant correlations between the measured muscle’s EMG and comfort rating were found in UT except for 40° backrest inclination. This may be due to the limited sample size (n = 19); thus, a nonsignificant correlation is identified. The last column of the table, with the highest correlation factor (medium correlation with R = 0.533), shows the correlation across all test inclinations, which can be interpreted as the indication of overall muscle activation-comfort association for various seat conditions.

Interestingly, the significant correlations found in UT have positive coefficients, while the positive correlation means that an occupant’s higher UT muscle activation corresponds to its higher perceived comfort. This is contrary to the common understanding that, for the same muscle group, a more relaxed condition always leads to better comfort feelings. To comprehend the observed phenomenon, it should be first understood that the strained muscle in the posterior neck region is induced by the supporting load from the headrest pointing towards the sitter’s frontal direction from a seating biomechanical point of view (Zhong et al., 2021), as mentioned in Sec 3.2. Thus, this observed tendency imply that a higher headrest supporting load would result in better comfort in the UT area. It is worth mentioning that such an inference does not conflict with that stated in Sec 3.2, that the less relaxed muscle (UT) may induce discomfort (Figure 6; Figure 7). The statement remains true when comparing different muscle groups (UT and SCM).

Nevertheless, it should be noted that the experiment designed in this study only focuses on short-term sitting. Further investigation on the monotonic trend of muscle activation and comfort can be conducted with a longer period of sitting to assess comfort and muscle fatigue in prolonged sitting. The finding from this paper shows that for a short period, a higher strained (for the same muscle) condition could generate better comfort, and this may explain why the sitter does not always feel as comfortable with a constant posture and tends to move and change posture to relieve discomfort during sitting (Jackson et al., 2009; Tan et al., 2009; Søndergaard et al., 2010; Cascioli et al., 2016; Hiemstra-van Mastrigt et al., 2017), regardless how relaxed the initial condition was. In addition, while the studied backrest inclination range (30°-50°) covers most seated activities, higher backrest angle can be considered as literature shows the comfortable sleeping angle for the backrest can reach 55.5° Smulders et al . , 2016 .