Abstract
The vigilance decrement, the decline in behavioral adaptability seen after sustaining attention over time, has been a monumental problem in the applied sciences for the past 70 years. Trait selfcontrol is a measure of the general capability to regulate one’s own cognitive resources. In the present study, participants performed a 24-minute vigilance task that was either synchronous (temporally regular) or asynchronous (temporally irregular). We empirically investigated the effects of self-control, watch period, and event asynchrony on performance and neural resource expenditure through functional near-infrared spectroscopy (fNIRS). Here we demonstrate that trait self-control indexes an operator’s capability to prevent making false alarms. Additionally, we elaborate on the functional role of self-control in optimizing right parietal resource usage and subsequently, resilience against the vigilance decrement. Finally, we discuss theoretical and practical implications for Neuroergonomics practitioners.
Keywords
Introduction
Vigilance involves extended attention directed at a wide variety of real-world tasks such as monitoring in airtraffic control, driving, studying, and surgery (Davies & Parasuraman, 1982). The concept has been studied for decades in applied experimental psychology. Vigilance performance is most often explained in terms of cognitive resource theories of performance in which the decline in performance seen in vigilance tasks is attributable to a loss of cognitive resources available for performance. However, the reason why vigilance performance declines over time is poorly understood. One explanation that has recently been explored is that trait self-control, an individual’s ability to change the default way they would otherwise think, feel, or behave, might influence their performance (Baumeister, Heatherton, & Tice, 1994). In an arduous scenario such as a vigilance task, the ability to persevere while performing effectively differs between individuals, as some are able to successfully maintain attention over time while others fail. In Baumeister’s selfcontrol or self-regulatory strength model (Muraven & Baumeister, 2000), it is suggested that exercising self-control consumes self-control strength which is dependent upon available resources. If these resources are drained too much, there are fewer resources available for subsequent tasks that require self-control. Thus, we sought to explore whether vigilance performance might be dependent on an individual’s intrinsic self-control strength. Previous studies have shown that performing acts of self-control such as controlling one’s emotions (Muraven, Tice, & Baumeister, 1998), or resisting tempting foods like cookies (Baumeister et al., 1998) leads to poorer performance on a subsequent test of self-control. When it comes to vigilance, there is some evidence that suggests that dietary restriction leads to poorer performance on vigilance tasks (e.g., Green, Elliman, & Rogers, 1995; Green, Rogers, Elliman, & Gatenby 1994). Thus, it could be the case that high trait self-control might make the use of cognitive resources more efficient during task performance.
The purpose of the current study is to examine the relation between trait self-control and vigilance when the task is temporally uncertain. Additionally, we used functional Near-Infrared Spectroscopy (fNIRS) to examine the neural underpinnings of performance. Previous studies examining the relation between self-control and performance have revealed mixed results. For example, Satterfield and colleagues (2019) investigated the relation between state selfcontrol, trait self-control, and vigilance performance. In that study, the authors attempted to deplete self-control using a typing task that had been previously demonstrated to successfully deplete state self-control in previous studies (Muraven et al., 2006). Moreover, trait self-control was assessed using the Tangney, Baumeister, & Boone SelfControl Scale (Tangney et al., 2004). The results of that study revealed no relation between self-control and vigilance, even when self-control was presumably depleted in one of the conditions.
Self-Regulation in Vigilance and the Brain
An examination of individuals with high self-control and the associated brain mechanisms tells a different story than the Satterfield et al. (2019) study. Becker and colleagues (2015) examined the relation between self-control and vigilance but also included a neurophysiological measure called transcranial doppler sonography (TCD). TCD is an ultrasound technique used to examine cerebral blood flow velocity in the middle cerebral artery. There is an abundance of evidence that suggests that the TCD measure can successfully index the utilization and allocation of cognitive resources during vigilance (see Shaw et al., 2019 for a review). While the results of that study did not show differences in performance between high and low self-control observers (although there was a marginally significant trend favoring the high self-control group), results relating to the neurophysiological measure showed that while there was a decline in blood flow velocity in the low self-control group, there was no such decrement in the high self-control group. This points to the likelihood that high self-control individuals have superior allocation strategies during vigilance. Perhaps one explanation for the lack of a performance effect when exploring the self-control-vigilance relation is that the vigilance contexts previously explored did not require self-control. For example, in both the Satterfield and the Becker studies, the task used was a short 12-min vigilance task that was temporarily regular and somewhat predictable. Perhaps having high self-control will better serve an observer in situations where the onset of events is irregular, or asynchronous. In a previous study exploring the event asynchrony effect, Shaw and colleagues (2012) found that individuals exhibited the trademark behavioral decrement over time for both synchronous and asynchronous events. However, cerebral blood flow velocity (CBFV) displayed effects of event schedule and time on task depending on hemispheric laterality. Hemovelocity declined more steeply in the right hemisphere as compared to the left hemisphere. Intriguingly, the effects of event schedules were solely dependent on interactions with both time period and hemisphere. Here the left hemisphere indicated asynchronyspecific declines in performance, while the right hemisphere exhibited declines in both synchronicity groups. The authors suggest that bilaterality of activation in the brain during vigilance increases with higher expenditure of cognitive resources, a finding consistent with other research (Harwood et al., 2017; Shaw et al., 2016).
Here, we use fNIRS to re-examine the self-control and vigilance relation, but specifically in contexts where event presentations are either synchronous (temporally regular) or asynchronous (temporally irregular). Utilizing fNIRS affords superior spatial resolution to previous vigilance studies using TCD, because near-infrared light’s absorbance in the cortex allows for examination of functional subregions of activation anywhere in the superficial cortex layers (i.e., right orbitofrontal cortex or right superior parietal lobule). This is advantageous over TCD, which is only able to determine activation via the left- versus right-hemispheres. Additionally, because fNIRS is fully portable, we are now able to map functional networks in the brain in highly ecological contexts such as operational environments (Curtin & Ayaz, 2018). This allows for a novel conception of cognitive resources via supply (oxygenated hemoglobin, HbO2) and demand (deoxygenated hemoglobin, HbR) dynamics, which may be particularly useful for Human Factors researchers studying sustained attention.
It is hypothesized that self-control will modulate vigilance performance, but specifically in contexts where event presentations are asynchronous. More specifically, we predict superior performance of high self-control individuals in the asynchronous condition as compared to the synchronous condition. Moreover, we predict that high self-control individuals will show increased neural efficiency (Becker et al., 2015) consistent with previous evidence suggesting that self-control influences the allocation of information processing resources. Taken together, these findings will have implications for resource theories of vigilance and the selection issue that has long been a concern of vigilance researchers (e.g., Finomore et al., 2009; Matthews et al. 2011; Shaw et al., 2010).
Methodology
Participants and Design
Twenty-nine college-age (M = 20.38 years, SD = 3.21) participants (16 female, 13 male) were recruited from a large, mid-Atlantic university in the United States. All participants were compensated for their time and ethically treated according to the guidelines of the American Psychological Association.
A 2×4 mixed-factorial design was used, where participants were randomly assigned to the synchronous (temporally regular) or asynchronous (temporally irregular) event schedule condition and time was discretized into four watch periods as a repeated-measures factor.
Procedure
Participants first completed a set of surveys pertaining to trait self-control (Tangney, Baumeister, & Boone, 2004) and pre-task stress states (Helton, 2004). After a baseline period of 10 minutes (for rest and neuroimaging purposes), they were randomly assigned to a computerized vigilance task consisting of either asynchronous or synchronous event schedules. The asynchronous condition involved randomly jittered inter-stimulus-intervals (ISIs) (.6 to 3 seconds between events); in contrast, the synchronous condition involved consistent ISIs (2 seconds between each event). Both conditions were programmed to have an average ISI of 2 seconds. A 2mm by 9 mm horizontal white rectangle served as the stimulus against a grey background, and the stimulus duration delineated whether a signal was critical (125 ms) or neutral (250 ms). Participants were instructed to press the spacebar if a critical signal was present and abstain from responding otherwise. After a 2-minute forced-choice practice task, they engaged in the 24-minute experiment (four blocks of 6 minutes each), followed by the same stress survey. The procedure and stimuli were taken directly from Shaw et al. (2012), but with the task instead being reduced from 40 to 24 minutes in length.
FNIRS Procedure
Functional near-infrared spectroscopy (fNIRS) measurements of relative HbO2 and HbR concentrations (850nm and 760 nm; 10.2 Hz) were recorded during the experiment using two daisy-chained NIRx NIRSport 2 imaging systems (NIRx Medical Technology, Berlin). Participants’ head circumferences were measured using a measuring tape and were then fitted with a cap size (54, 56, 58, or 60 cm). The montage consisted of 46 optodes total, covering both the prefrontal cortex (8 sources, 7 detectors, 8 short-distance detectors) and parietal cortex (same as previous). The montage layout was determined a priori for maximal coverage of our desired Brodmann areas using fNIRS Optodes’ Location Decider (fOLD), and then created in NIRSite. Optodes were configured according to the international 10-20 system and used a 3cm long sourcedetector separation (short-distance channels = 8mm).
During fNIRS acquisition, automatic triggers were scripted in PsychoPy for the onsets of 32 critical signals (20 seconds each) across the 24-minute vigilance task. Eight critical signals appeared in each watch period (signal probability = 6%, event rate = 30/minute). Due to the theoretical problem of measuring uninterrupted high eventrate vigilance with the slow hemodynamic response, we used a slow event-related design consisting of only critical signals versus participants’ baseline hemodynamic response functions (HRFs). This was done in order to generate “rest” periods for the HRFs, as critical signals were highly infrequent compared to the neutral stimuli.
In the group-level analyses, the same exact regression form of the ANOVA design was used in order to model the effects of self-control on neural responses in addition to watch period and temporal uncertainty. All brain-behavior correlations and neural depictions consist of standardized beta coefficients that survived multiple comparisons adjustments.
Results
Behavioral Analyses
First, ANOVAs were conducted using the aforementioned design (2×4 mixed design) on all relevant behavioral variables (CDs, FAs). Then, ANCOVAs (with self-control) were conducted to see if trait self-control accounted for any of these effects.
Correct detections
Results indicated a significant main effect of watch period on correct detections, F(3, 27) = 6.38, p <.05, ƞp2 =.15. This suggests that during the course of the task, correct detections decreased from the first watch period (M = 82%, SE = 4.3%) up until the third (M = 58.2%, SE = 4.3%) and fourth watch periods (M = 60%, SE = 4.3%). However, no main effect of synchronicity nor interaction was present. Self-control was not a significant covariate with correct detections in the ANCOVA model, F(1, 27) =.89, p >.05.
False alarms
Results showed a significant main effect of watch period on false alarms, F(3, 27) = 2.90, p <.05, ƞp2 =.08. This suggests that during the course of the task, false alarm rates decreased from the first watch period (M = 15%, SE = 1.8%) up until the third (M = 7.3%, SE = 1.8%) and fourth watch periods (M = 7.6%, SE = 1.8%). Additionally, a main effect was present for synchronicity, F(1, 27) = 36.46, p <.05, ƞp2 =.25. This demonstrated that false alarm rates were significantly higher in the asynchronous (M = 15%, SE = 1.3%) than in the synchronous (M = 4.3%, SE = 1.2%) condition (see Figure 1). No interaction between watch period and synchronicity was present. Self-control was a significant covariate with false alarms in the ANCOVA model, F(1, 27) = 5.31, p <.05, ƞp2 =.05. This suggests that self-control does account for differences in false alarms, and that this happens only in the asynchronous task (r = -.32, p =.02), but not in the synchronous task (r =.18, p =.21) (see Figure 2). What this suggests is that higher self-control individuals commit less false alarms than their low self-control counterparts, specifically in the more temporally uncertain task.
Main effects of watch period and synchronicity on false alarm rates.
Correlation between self-control and FAs.
Neuroimaging Analyses
Mixed effects regressions (random intercept for participant, random slopes for watch periods) were conducted to investigate whether a two-way interaction between trait self-control and watch period was present.
Trait Self-Control
Results for neural supply (HbO2) did not survive multiple comparison-adjusted tests. However, results did indicate a moderation by self-control from watch period 1 to watch period 2 deoxygenation at left orbitofrontal cortex (lOFC) for those high in self-control, t(11195, 27) = 3.01, q = 0.02. Furthermore, results also showed a moderation by self-control from watch period 1 to watch period 2 deoxygenation at right inferior parietal lobule (rIPL) for those low in self-control, t(11195, 27) = -4.12, q = 0.01. Lastly, results indicated a moderation by self-control from watch period 1 to watch period 3 deoxygenation at rIPL for those high in self-control, t(11195, 27) = 3.86, q = 0.01. Altogether, these findings suggest that low self-control individuals were using their right parietal cortex more than their high selfcontrol peers (see Figure 3).
Moderation of watch period by trait self-control (red = high self-control) on neural deoxygenation. From watch period 1 to watch period 3, high self-control deactivated right inferior parietal lobule (rIPL; shown in the brain map). Only tests surviving FDR multiple comparisons adjustments at the q < 0.05 level are depicted.
Brain-Performance Correlations
There was a correlation between self-control and right superior lobule (rSPL) deoxygenation in the asynchronous task (r =.56, p =.04) but not the synchronous task (r =.14, p =.62), for which high self-control individuals had less activation. What this suggests is that high self-control individuals were using less resources in their right superior parietal lobule as compared to their low self-control counterparts in the temporally uncertain task. Furthermore, there was a correlation between right superior parietal lobule deoxygenation and FAs in the synchronous task (r = -.60, p =.02) but not the asynchronous task (r = -.13, p =.66). This demonstrates that individuals with less activation were better able to resist committing false alarms during the period of performance decrement (see Figure 4).
Correlation between right superior parietal lobule deoxygenation and FAs. From watch period 1 to watch period 3, greater activation of rSPL (shown in the brain map) was associated with greater decrement in perceptual sensitivity (+FA = more decrement) during the synchronous task.
Discussion and Implications
The present results are intriguing, given the murky role of self-regulation in operators’ capabilities to manipulate their sustained attention in demanding tasks. It is noteworthy that the behavioral benefits of high self-control (less false alarms) were restricted to the asynchronous task, suggesting that self-control plays a role in sustained attention only when the demands are great enough to require it.
From the first period to the greatest period of decrement, individuals high in self-control had increased right inferior parietal lobule deoxygenation, meaning that selfcontrol lowers resource usage in the right parietal cortex. More specifically, what these findings might illustrate is that the capability to regulate oneself promotes the efficiency of resource utilization at “the right place and the right time.” This is further evidenced by the finding that decreased expenditure of resources in rSPL was associated with better avoidance of false positives, whereas increased activation was instead associated with committing more false positives. Here, high self-control individuals attain more operational and perceptual benefits than their low self-control peers across differentially dynamic sustained attention tasks.
Theoretical and practical implications of this study include expanding the body of work on cognitive resources as well as identifying a role for trait self-control in the maintenance of sustained attention. The present study is the first to demonstrate performance benefits of trait self-control within the context of sustained attention. However, it remains to be seen if general dispositions towards regulating life habits also transfer to the direct ability to regulate one’s neural activation – and whether this would further facilitate sustained attentional performance.
This study has a few limitations. The first is that a newer, multidimensional self-control scale (Nilsen et al., 2020) that correlates very well with Tangney and colleagues’ (2004) original self-control scale has been developed to parcellate out self-control into excitatory and inhibitory factors. Thus, the present study cannot indicate whether the benefits of self-control here were possibly driven by individuals’ strengths in excitatory versus inhibitory selfcontrol, or both aspects. Given that the vigilance task used is an activation task, it is crucial to understand whether greater neural resource efficiency is primarily associated with a traitlevel adeptness in inhibiting oneself (inhibitory self-control) as opposed to engaging oneself (excitatory self-control). The former would suggest that high self-control individuals might be inhibiting themselves more during an activation task – meaning they are able to expend the ideal amount of attention as compared to their low self-control counterparts (whom are instead wasting too much attention). Additionally, the present study did not investigate a broader range of demographics than college-age participants, whereby there may be differences in how self-control is mobilized to task performance in the broader population. Future studies should investigate how the frontoparietal networks are modified in practical, real-life environments – such as in driving, aviation, and combat or medical procedures. If we can learn to better regulate ourselves in daily life, then perhaps we can more effectively pay attention in highly demanding scenarios.
