Male (n = 48) subjects between the ages of 18 and 41 years old and currently serving in the U.S. military through Active Duty, Reserve, National Guard, or Reserve Officer Training Corps (ROTC) were eligible to participate in this study. Eligible subjects underwent service-specific physical fitness tests within the last year, reported a high level of comfort with shooting an M4/M16 weapon, had no current or recent (within last 3 months) injury that would prevent participation in sport or prevent deployment, and were not working shiftwork or taking medications known to affect sleep or cognitive performance. Individuals with current alcohol use disorder, a history of bipolar, psychotic, seizure, or neurological disorder were excluded. Individuals with a prior diagnosis of traumatic brain injury with current chronic post-concussive symptoms and rehabilitative treatment for traumatic brain injury, or suspected traumatic brain injury in the previous 6 months based upon the medical review of post-concussive symptoms were also excluded. Furthermore, individuals at high risk of obstructive sleep apnea without treatment were excluded. Total cohort participant characteristics are presented in Table 1. The study received approval from the Institutional Review Board at the University of Pittsburgh (STUDY19090271) and the U.S. Army Medical Research and Development Command’s Human Research Protection Office (HRPO). After completing a comprehensive telephone screening interview, eligible subjects were scheduled for an in-person consent process. Once subjects provided written, informed consent, a urine drug screening and breathalyzer test were conducted to confirm eligibility.

Data presented herein are part of a larger study (U.S. Department of Defense award #W81XWH-17-2-0070; Beckner et al., 2021a; Conkright et al., 2021a,b). This prospective cohort study was designed to characterize the impact of 48 h of SMOS, involving sleep and caloric restriction and physical work, on military tactical adaptive decision making. Eligible subjects completed a 96-h protocol that occurred over five consecutive days and nights. The SMOS protocol has been previously described in detail (Beckner et al., 2021a; Conkright et al., 2021a). Briefly, subjects completed 1 day of familiarization testing (D0) followed by an adaptation night of uninterrupted sleep from 2,300 to 0700, that included sleep apnea testing. Night 0 was intended to familiarize subjects to the sleep environment and setup. After 1 day of baseline testing (D1), subjects were exposed to 48-h of SMOS (D2 and D3). During SMOS, subject’s sleep opportunity was reduced to 50% of the familiarization and baseline night sleep opportunities (0100–0300 and 0500–0700). Additionally, subjects received only 50% of their individual caloric needs, based on energy expenditure calculations using air displacement plethysmography (BodPod Body Composition System, Life Measurement Instruments, Concord, CA). On the evening of D3, subjects were allowed uninterrupted sleep from 2,300 to 0700. Following the night of recovery sleep, subjects completed the final day of testing (D4) and were dismissed at 1930. All meals were provided by the laboratory study team and consisted of a standard breakfast and “meals, ready to eat” (MRE). Water intake was tracked and provided by the study team ad libitum. Caffeine was prohibited during the study.

The SPEAR task is a computer-based assessment of adaptive decision making during realistic military scenarios (Haufler et al., 2018). An abbreviated version of the task was utilized in the present study, though a detailed description of the entire SPEAR task has been previously reported (Haufler et al., 2018). Briefly, the task begins with instructions followed by a strategic context, mission statement, and commander’s intent, presented for 30–45 s each. Eighteen trials are then presented to closely approximate tactical challenges associated with the block’s strategic guidance, mission statement, and commander’s intent. Each trial begins with a fixation cross-presented for 3 s to alert the participant that the trial is beginning. This is followed by a scenario description (30 s), a video (30 s), a response prompt (reporting to higher headquarters, adjacent units, or subordinates; 10 s), and the response period (105 s) in which the subject types their plan of action in response to the challenge presented in the trial. The video presents a situation that is either consistent or inconsistent with the initial scenario description in order to introduce uncertainty and heighten adaptability. Consistent/inconsistent trials and direction of the response (provided in the prompt) is balanced within each block and presented in a pseudo-randomized order.

An abbreviated practice SPEAR task was administered on day 0 during familiarization testing in order to control for learning effects. Subjects completed all 18 trials on day 1 (baseline) and in a different order on day 3 (peak stress) within the same testing environment. Subject responses were subsequently graded by three trained, independent graders based on a scoring rubric that consisted of an eight-category scale in which each subcategory represented a different aspect of an adaptable response (Haufler et al., 2018). Subjects could earn one point for each category, with a range of possible scores per trial between zero and eight. If a score deviated by more than 3 points between graders, discussion, reconciliation, and rationale for an agreed upon trial score were written and used as the final grade. The responses were scored using the rubric which contained eight different aspects of an adaptable response (Haufler et al., 2018). The eight aspects are shown in Table 2. The results of the inter-rater reliability assessment for the 18 prompts measured on Day 1 and Day 3 are listed in Appendix A. Gwet’s AC estimates ranged from 0.95 to 0.98, almost perfect, according to the benchmarks for strength of agreement recommended by Landis and Koch (1977).

The psychological test battery occurred following informed consent, prior to the familiarization day, and consisted of 3 questionnaires: the Connor-Davison Resilience Scale (CD-RISC), the NEO Personality Inventory Five Factor Inventory (NEO), and the Deployment Risk and Resilience Inventory-II (DRRI-II). The CD-RISC is a validated 25-item self-report measure that uses a 5-point Likert scale ranging from 0 to 4 to measure resilience, with higher scores indicative of higher resilience (Connor and Davidson, 2003). All 25 items are summed to derive the total CD-RISC score. The CD-RISC has demonstrated strong internal consistency among military cohorts and has been associated with retention during basic military training (Xie et al., 2016; Bezdjian et al., 2017; Beckner et al., 2021a). The NEO is a 240-item self-report that measures responses using a five-point Likert scale, ranging from 1 to 5: “strongly disagree,” “disagree,” “neutral,” “agree,” or “strongly agree.” NEO assesses five personality traits that are known to modulate reactivity to stressful experiences: extraversion, agreeableness, conscientiousness, neuroticism, and openness to experience, with higher scores within each of the five traits representing greater propensity toward that specific personality characteristic (Costa, 1992; Ashton, 2013). The Combat Experiences scale is one of the 17 scales within the DRRI-II, a self-report instrument that assesses different factors and domains that contribute to risk and resilience following military deployment. This scale measures exposure to combat-related circumstances (i.e., firing a weapon or being fired on, being attacked or witnessing an attack, encountering explosive devices etc.) with scores ranging from 17 to 102, with a higher score denoting more combat-related experiences (Vogt et al., 2013).

Aerobic fitness was assessed using the Bruce protocol treadmill test (Froelicher et al., 1975). On the familiarization day, subjects completed the test on a Woodway treadmill to determine relative peak oxygen consumption (mL·kg⁻¹·min⁻¹) as measured using a metabolic cart (Parvo TrueOne® 2,400; Salt Lake City, UT).

Neurocognitive assessments pertaining to this analysis included measures of risk propensity, emotion recognition, vigilant attention, short-term memory, and language-based logical reasoning. The Balloon Analog Risk Task (BART) quantifies risk propensity and impulsivity (Lejuez et al., 2002). The Emotion Recognition Test (ERT) assesses emotional identification through facial expressions (Basner et al., 2015). The Psychomotor Vigilance Test (PVT) assesses vigilant attention (Basner et al., 2015). An abbreviated (3-min), validated version of the PVT was used in this study (Basner et al., 2011). The BART, ERT, and PVT are from the Cognition test battery, a computerized task designed to assess cognitive function in high-performing adults and deemed most relevant to the present study (Basner et al., 2015). Subjects completed these tests each morning. Each afternoon after completion of a physical testing battery, subjects completed a neurocognitive test battery consisting of two tests: Match to Sample (MATCH) to assess short-term spatial memory (Shurtleff et al., 1994; Lieberman et al., 2002, 2014), and Grammatical Reasoning (GRAM) was adapted from the Baddeley Grammatical Reasoning Test (Baddeley, 1968) to evaluate language-based logical reasoning (Lieberman et al., 2005, 2014). Speed and accuracy scores were calculated for the BART, ERT, and PVT as previously described (Moore et al., 2017; Basner et al., 2020; Beckner et al., 2021a). GRAM and MATCH accuracy were calculated as the number of trials answered correctly throughout the assessment, per total number of trials completed with 24 total trials for MATCH and 32 for GRAM (Lieberman et al., 2002, 2014). Speed was calculated as reaction time to correct responses in seconds (Lieberman et al., 2002, 2014).

The Profile of Mood States Questionnaire (POMS) was used to assess subjects’ mood each day at approximately the same time (0830). POMS is a validated psychological test consisting of a 65-item survey, wherein subjects current mood state according to 6 specific domains is categorized on a 5-point Likert scale (Curran et al., 1995). The six domains include: tension-anxiety, depression-dejection, anger-hostility, vigor-activity, fatigue-inertia, and confusion-bewilderment (McNair et al., 1971; Lieberman et al., 2005). Subjects responses were subsequently summed to determine a total mood disturbance (TMD) score, TMD is the sum of the negative scales plus the “non-positive” remainder of the positive Vigor-Activity scale such as that: Total Mood Disturbance = (anger and hostility + confusion and bewilderment + depression and dejection + fatigue and inertia + tension and anxiety) + (maximum vigor and activity score – actual vigor and activity score; McNair et al., 1971; Curran et al., 1995). Lower scores indicated a more positive mood state and conversely, higher scores indicating a more negative mood state (Conkright et al., 2021a).

Subjects were fasted, and their blood was drawn from an upper extremity vein each morning (~08:00 h) via a 21- or 23-gauge needle (Becton, Dickinson and Company, Franklin Lakes, NJ). Using standard venipuncture procedures, blood was collected into appropriate collection tubes, EDTA for plasma (10 mL), SST for serum (10 mL), and P100 for protein preserved plasma (2 mL; Becton, Dickinson and Company, Franklin Lakes, NJ). SST tubes were allowed to clot at room temperature for 30 min prior to centrifugation, EDTA and P100 tubes were centrifuged immediately after collection. All tubes centrifuged at 1500 ×g for 15 min at 4°C (Beckner et al., 2021a). Blood supernatant was aliquoted and stored at −80°C for future analysis. Enzyme-linked immunoassays were conducted for each biomarker, using plasma samples from EDTA collection tubes for IGF-I (APLCO, Salem, United States), and BDNF MILLIPLEX Magnetic Bead Panel 3 (EMD Millipore, Burlington, Massachusetts). plasma obtained from P100 tubes for NPY analysis (R&D Systems, Minneapolis, MN, United States), and serum samples for cortisol (Alpco, Salem, United States). Sensitivity for each assay was as follows: IGF-I: 0.09 ng/ML; BDNF: 10 pg./mL, cortisol: 0.4 μg/dL. Sensitivity information was not available for NPY. All samples were measured in duplicate with intra-assay coefficients of variation of 10% or less.

A total of three trained raters participated in evaluating each subject completed SPEAR trials. Performance was evaluated on an ordinal scale (0–8) with 8 representing the best possible score, per trial. In order to account for high percent agreement, the Gwet’s AC coefficient was calculated (Landis and Koch, 1977; Gwet, 2008; Gwet, 2014). Disagreement among raters was weighted using ordinal weights. Statistical analysis for the inter-rater reliability was conducted using Stata/SE 16 (StataCorp LLC; College Station, TX).

Differences in SPEAR score from baseline to peak stress (D3 minus D1) were calculated to assess change in military tactical adaptive decision making that may occur following 48-h SMOS. The sample was divided into low adaptors (SPEAR total block score decrease from baseline to peak stress; N = 20) and high adaptors (SPEAR total block score increase from baseline to peak stress; N = 28). Subjects whose scores remained the same were excluded from analysis (SPEAR total block score change of 0 from baseline to peak stress; N = 3). Independent samples t tests were performed to analyze differences in baseline measures (i.e., CD-RISC, NEO, DRRI-II, aerobic fitness) between groups. Two-way mixed-measures ANOVAs were performed to analyze the interaction of group (low or high adaptors) and day (D1, baseline or D3, peak stress) on each domain (i.e., neurocognitive, mood state, neuroendocrine). If the two-way interaction was not significant, main effects of group and of day were reported and interpreted. If assumptions for ANOVAs were not met, logarithmic (LN), square root (SQRT), and reciprocal (REC) data transformations were conducted. Successful data transformations were performed for the following variables: LN basal BDNF, REC basal cortisol, REC IGF-I, SQRT years of service, SQRT ERT speed, SQRT PVT speed, LN BART speed. If assumptions for normality were not met (i.e., Shapiro–Wilk p < 0.05), nonparametric tests were performed. All analyses were performed using IBM SPSS Statistics for Mac, version 26 (IBM Corp., Armonk, NY). Alpha was set at 0.05 a priori for all analyses and Bonferroni-adjusted for multiple pairwise comparisons. Effect size via partial eta-squared(ηp2) was calculated from ANOVA results.

Overall, military tactical decision-making adaptability remained stable throughout the operational stress scenario, from D1 (69.0 ± 18.3) to D3 (68.1 ± 21.1). Low adaptors (n = 20) exhibited an average score of 69 ± 17 on D1 and 56 ± 21 on D3, a statistically significant mean decrease of −12.50 ± 12.56 (p < 0.001, ηp2 = 0.530). High adaptors (n = 28) also scored an average of 69 ± 19, but then improved on D3 with an average score of 76 ± 17, a statistically significant mean increase of 7.32 ± 6.27, p < 0.001, ηp2 = 0.530.

High adaptors reported significantly higher scores of self-reported resilience at baseline compared to low adaptors (87.7 ± 9.7 vs. 80.2 ± 11.2, p = 0.020, ηp2 = 0.119). Low adaptors reported greater scores in neuroticism compared to high adaptors (12.4 ± 6.9 vs. 24.5 ± 8.0; p < 0.001, ηp2 = 0.401) and conversely, high adaptors were more extroverted (32.4 ± 6.7 vs. 18.2 ± 6.7, p < 0.001, ηp2 = 0.534) and conscientious (38.3 ± 5.7 vs. 30.4 ± 4.7, p < 0.001, ηp2 = 0.362) than low adaptors (Figure 1). Groups did not differ in openness or agreeableness.

Groups significantly differed at baseline aerobic fitness testing such that high adaptors had a significantly higher aerobic capacity compared with low adaptors (48.4 ± 8.2 vs. 30.4 ± 4. 7 mL/kg/min, respectively; p < 0.001, ηp2 = 0.627).

There was a main effect of day for multiple neurocognitive tests including those testing for language-based reasoning, vigilant attention, and emotion recognition (Table 4). GRAM speed was statistically significantly different on D1 (M = 3.6, SD = 0.9) compared to D3 (M = 2.8, SD = 1.2), p < 0.001, ηp2 = 0.328, a mean decrease of −0.11.

ERT accuracy decreased from D1 (M = 71.6, SD = 11.0) compared to D3 (M = 66.3, SD = 16.0), p = 0.026, ηp2 = 0.103, a mean decrease of −0.052. PVT slowness increased from D1 (M = 2.4, SD = 0.1) compared to D3 (M = 2.5, SD = 0.1), p = < 0.001, ηp2 = 0.311, a mean increase of 0.03. PVT accuracy increased from D1 (M = 0.9, SD = 0.1) compared to D3 (M = 0.8, SD = 0.2), p < 0.001, ηp2 = 0.308, a mean increase of 0.03. There were no main effects of day or group for risk propensity (BART), or short-term memory (MATCH; Table 5).

Tension, fatigue, confusion and total mood disturbance all significantly increased from D1 to D3 (all p-values < 0.001), while vigor decreased (p < 0.001). Depression and anger did not significantly change over the course of the stressor (Figure 2). There were no main effects of group for POMS.

Natural logarithmic transformations corrected for non-normality in resting IGF-I showed significantly different concentrations on D1 (M = 36.6, SD = 13.1) compared to D3 (M = 33.6, SD = 11.2), p = 0.026, ηp2 = 0.110, a mean decrease of-2.94. Reciprocal transformations corrected for non-normality in resting NPY and cortisol but did not yield significant differences by day or group. Natural log transformations corrected for non-normality in resting BDNF, and similarly did not yield significant differences (Table 6).

High adaptors exhibited significantly less neuroticism, and greater conscientiousness and extroversion than low adaptors. This finding is supported by prior literature examining the role of personality in military leadership and in group effectiveness, wherein low neuroticism and high conscientiousness best predicted high leadership effectiveness in the Australian Military Officer Training School (McCormack and Mellor, 2002). Conscientiousness, including such personal characteristics as order, achievement striving and dependability (Callister et al., 1999; Halfhill et al., 2005), positively aligns with the traits necessary to be successful in the simultaneous perception, cognition, and action aspects demanded in the adaptive decision making task (Haufler et al., 2018). Further, the high extroversion displayed in high adaptors may have caused these subjects to be more engaged in the group dynamics of the study protocol, theoretically acting as a buffer from deficits during SMOS. This theory aligns with the widely studied benefits of social coherence (McCraty, 2017; Salas et al., 2017). In contrast, neuroticism, including such personal characteristics as impulsivity and vulnerability has been inversely associated with success in military cohorts (Callister et al., 1999; Paunonen et al., 2006; Lee, 2010) and positively correlates with mental health problems in a large cohort of Chinese medical school students (Peng et al., 2012).

High adaptors scored significantly higher on the CD-RISC test compared to low adaptors. The CD-RISC, with higher scores indicative of higher resilience, assesses the ability to adapt to challenging circumstances (Connor and Davidson, 2003), a trait paramount to warfighters. As described by Haufler et al. (2018) military service members must consistently respond to multifaceted and competing demands within the modern warfare operational setting. Further, they must be able to acutely distinguish relevant from irrelevant information, consider their options, and adapt their trajectory in a timely fashion often under conditions of severe stress (Haufler et al., 2018). Following 6-months of U.S. Air Force basic training, the CD-RISC scale successfully predicted early unsuitability attrition (e.g., mental health or behavioral adjustment difficulties), revealing those who remained active exhibited higher CD-RISC scores, than did the separated trainees (Bezdjian et al., 2017). Further, resiliency—as measured by the CD-RISC scale—was found to be a prominent moderating factor between negative life events and mental health issues (Peng et al., 2012).

High adaptors demonstrated significantly greater aerobic capacity than did low adaptors. It is well recognized that a high level of physical fitness, specifically aerobic fitness, is necessary for success in military duties, including deployment and occupation (Kyröläinen et al., 2018). High aerobic capacity has obvious physical performance benefits, including elevated mission-specific performance, reduction of injury risk (Taylor et al., 2008; Lisman et al., 2013), and enhanced recovery from high-intensity exercise (Hoffman, 1997) with limited recovery time (Ojanen et al., 2018; Conkright et al., 2021a). Further, higher aerobic fitness has been shown to improve cognition (e.g., reasoning, planning, and problem solving; Stern et al., 2019) and is positively associated with working memory, (Bunce and Murden, 2006) greater task accuracy, and improved reaction time, and sleep quality (Taheri and Valayi, 2018; Lefferts et al., 2019). In a large community based clinical trial, greater aerobic capacity was universally found to be highly associated with greater executive function, a relationship that became more pronounced with age (Stern et al., 2019). Similarly, within the same experimental protocol, Beckner et al. (2021a) found that greater aerobic fitness and greater resiliency were protective in buffering the impact of operational stress on vigilant attention. High adaptors demonstrated high aerobic capacity prior to undergoing SMOS, a key mediator to offset the inevitable detriment in physical capacity sustained during stress.

After sustaining 48-h of SMOS, high and low adaptors experienced similar changes in neurocognitive function. The speed at which the correct response was given in the language-based reasoning battery improved similarly across low and high adaptors while vigilant attention speed declined. This finding is consistent with prior literature demonstrating impaired reaction time and vigilance following simulated stress (Lieberman et al., 2005; Vrijkotte et al., 2016; Beckner et al., 2021a). Vigilant attention contributes to the readiness to respond to relevant stimuli, a profoundly important function during the volatility of warfare. Prior literature supports the notion that the processes underlying social and emotional cognition suffer detriment under stressful conditions, but there have been mixed results pertinent to the effect of stress on facially-oriented emotional recognition. Some studies have demonstrated the positive, or even compensatory reaction to acute stress resulting in greater attention to emotional cues, which is thought to be synchronized with the evolutionary framework suggesting that the ability to detect and assess potential threats under adverse conditions is essential to human survival (Barel and Cohen, 2018). In contrast, others have found no such enhancement (Li et al., 2014). An integral part of the adaptive decision making, the basis on which the groups were stratified, is the necessity for Warfighters to effectively interact and engage with locals from communities that do not share their personal, religious, or cultural beliefs and customs. As Oden et al. (2015) describes, these differences induce uncertainty into an already chaotic and intense operational environment, during which an abrupt, unchecked response can catalyze a violent, sometimes unnecessary, escalation. Response time during the grammatical reasoning test was not impaired over the course of the stressor, as both high and low adaptors improved from baseline to peak stress. This is in contrast to prior findings as others have observed between a 7%–15% decline in grammatical reasoning number correct after 53–60 h of sleep deprivation (Lieberman et al., 2005) and deterioration in performance in even elite Special Forces soldiers, following acute stress (Kallinen and Ojanen, 2021).

During a 5-day protocol including 48-h of SMOS, high and low adaptors experienced similar changes in mood state and neuroendocrine biomarkers. High and low adaptors experienced similar changes in negative mood states including significant increases in tension, fatigue, confusion, and total mood disturbance while suffering a decline in vigor. When Harris et al. assessed the cognitive function and psychological state prior to and 1-week post a Navy Survival, Evasion, Resistance, and Escape School—a period of physical stress, sleep and caloric deprivation and psychological distress—similar deteriorations in vigor were found, as well as increased fatigue and confusion (Harris et al., 2005). Similarly, in another study examining negative mood states in male Marines after completion of high-altitude military training, Marines’ experienced significant changes in fatigue and vigor post training, with levels that did not significantly differ from adult male psychiatric outpatients (Bardwell et al., 2005). IGF-I has consistently shown sensitivity to military operational stress (Nindl et al., 2003; Henning et al., 2011; Vikmoen et al., 2020; Beckner et al., 2021b) and was the only neuroendocrine biomarker to decline by −8.3% during SMOS, while BDNF, cortisol and NPY remained stable. Caloric restriction—particularly energy and protein restriction—have been shown to exert strong influences on IGF-I (Nindl and Pierce, 2010). In the present study, subjects energy and protein declined by 49 and 48%, respectively on restricted (D2, D3) vs. unrestricted days (D0, D1, D4).

The present study has several limitations that should be considered. While this study simulated stressors that typically occur in military settings, namely physical work, sleep and caloric restriction, there were no real-life consequences for poor performance. Despite explicit instruction to perform best effort across all domains, non-compliance and limited effort may be a potential confounding factor. The present study lacked a control group and thus, despite attempting to control for learning effects by way of multiple test exposures, limited definitive discrimination. Future studies should consider control group inclusion as to further eliminate potential extraneous variables. Furthermore, men and women respond to stress differently (Chaplin et al., 2008; van den Bos et al., 2009), such that these results cannot be extrapolated to female subjects. Though the impact of each individual stressor cannot be ascertained, the present study provides an ecologically valid assessment of the impact of SMOS on tactical decision-making and how it relates to psychological factors, physical fitness, mood, and neurocognitive performance.