Computational cognitive modeling of the temporal dynamics of fatigue from sleep loss

Hybrid framework of fatigue: connecting motivational control and computational moderators to gamma oscillations

Introduction

There is a need to develop a comprehensive account of time-on-task fatigue effects on performance (i.e., the vigilance decrement) to increase predictive accuracy. We address this need by integrating three independent accounts into a novel hybrid framework. This framework unites (1) a motivational system balancing goal and comfort drives as described by an influential cognitive-energetic theory with (2) accumulating microlapses from a recent computational model of fatigue, and (3) frontal gamma oscillations indexing fluctuations in motivational control. Moreover, the hybrid framework formally links brief lapses (occurring over milliseconds) to the dynamics of the motivational system at a temporal scale not otherwise described in the fatigue literature.

Methods

EEG and behavioral data was collected from a brief vigilance task. High frequency gamma oscillations were assayed, indexing effortful controlled processes with motivation as a latent factor. Binned and single-trial gamma power was evaluated for changes in real- and lagged-time and correlated with behavior. Functional connectivity analyses assessed the directionality of gamma power in frontal-parietal communication across time-on-task. As a high-resolution representation of latent motivation, gamma power was scaled by fatigue moderators in two computational models. Microlapses modulated transitions from an effortful controlled state to a minimal-effort default state. The hybrid models were compared to a computational microlapse-only model for goodness-of-fit with simulated data.

Results

Findings suggested real-time high gamma power exhibited properties consistent with effortful motivational control. However, gamma power failed to correlate with increases in response times over time, indicating electrophysiology and behavior relations are insufficient in capturing the full range of fatigue effects. Directional connectivity affirmed the dominance of frontal gamma activity in controlled processes in the frontal-parietal network. Parameterizing high frontal gamma power, as an index of fluctuating relative motivational control, produced results that are as accurate or superior to a previous microlapse-only computational model.

Discussion

The hybrid framework views fatigue as a function of a energetical motivational system, managing the trade-space between controlled processes and competing wellbeing needs. Two gamma computational models provided compelling and parsimonious support for this framework, which can potentially be applied to fatigue intervention technologies and related effectiveness measures.

Multiscale entropy in a 10-minute vigilance task

Research has shown multiscale entropy, brain signal behavior across time scales, to reliably increase at lower time scales with time-on-task fatigue. However, multiscale entropy has not been examined in short vigilance tasks (i.e., ≤ 10 min). Addressing this gap, we examine multiscale entropy during a 10-minute Psychomotor Vigilance Test (PVT). Thirty-four participants provided neural data while completing the PVT. We compared the first 2 min of the task to the 7th and 8th minutes to avoid end-spurt effects. Results suggested increased multiscale entropy at lower time scales later compared to earlier in the task, suggesting multiscale entropy is a strong marker of time-on-task fatigue onset during short vigils. Separate analyses for Fast and Slow performers reveal differential entropy patterns, particularly over visual cortices. Here, observed brain-behavior linkage between entropy and reaction time for slow performers suggests that entropy assays over sensory cortices might have predictive value for fatigue onset or shifts from on- to off-task states.

Neural superstatistics for Bayesian estimation of dynamic cognitive models

Mathematical models of cognition are often memoryless and ignore potential fluctuations of their parameters. However, human cognition is inherently dynamic. Thus, we propose to augment mechanistic cognitive models with a temporal dimension and estimate the resulting dynamics from a superstatistics perspective. Such a model entails a hierarchy between a low-level observation model and a high-level transition model. The observation model describes the local behavior of a system, and the transition model specifies how the parameters of the observation model evolve over time. To overcome the estimation challenges resulting from the complexity of superstatistical models, we develop and validate a simulation-based deep learning method for Bayesian inference, which can recover both time-varying and time-invariant parameters. We first benchmark our method against two existing frameworks capable of estimating time-varying parameters. We then apply our method to fit a dynamic version of the diffusion decision model to long time series of human response times data. Our results show that the deep learning approach is very efficient in capturing the temporal dynamics of the model. Furthermore, we show that the erroneous assumption of static or homogeneous parameters will hide important temporal information.

Predicting and mitigating fatigue effects due to sleep deprivation: A review

The deleterious effects of insufficient sleep have been well-established in the literature and can lead to a wide range of adverse health outcomes. Some of the most replicated findings demonstrate significant declines in cognitive functions such as vigilance and executive attention, psychomotor and cognitive speed, and working memory. Consequently, these decrements often lead individuals who are in a fatigued state to engage in substandard performance on everyday tasks. In the interest of curtailing these effects, prior work has attempted to identify mechanisms that predict fatigue onset and develop techniques to mitigate its negative consequences. Nonetheless, these results are often confounded by variables such as an individual’s resistance to fatigue, sleep history, and unclear distinctions about whether certain performance decrements are present due to fatigue or due to other confounding factors. Similar areas of research have provided approaches to produce models for the prediction of cognitive performance decrements due to fatigue through the use of multi-modal recording and analysis of fatigue-related responses. Namely, gathering and combining response information from multiple sources (i.e., physiological and behavioral) at multiple timescales may provide a more comprehensive representation of what constitutes fatigue onset in the individual. Therefore, the purpose of this review is to discuss the relevant literature on the topic of fatigue-related performance effects with a special emphasis on a variety of physiological and behavioral response variables that have shown to be sensitive to changes in fatigue. Furthermore, an increasing reliance on sleep loss, meant to assist in meeting the demands of modern society, has led to an upsurge in the relevance of identifying dependable countermeasures for fatigued states. As such, we will also review methods for the mitigation of performance effects due to fatigue and discuss their usefulness in regulating these effects. In sum, this review aims to inspire future work that will create opportunities to detect fatigue and mitigate its effects prior to the onset of cognitive impairments.

New insights into the cognitive effects of sleep deprivation by decomposition of a cognitive throughput task

STUDY OBJECTIVES

A cognitive throughput task known as the Digit Symbol Substitution Test (DSST) (or Symbol Digit Modalities Test) has been used as an assay of general cognitive slowing during sleep deprivation. Here, the effects of total sleep deprivation (TSD) on specific cognitive processes involved in DSST performance, including visual search, spatial memory, paired-associate learning, and motor response, were investigated through targeted task manipulations.

METHODS

A total of 12 DSST variants, designed to manipulate the use of specific cognitive processes, were implemented in two laboratory-based TSD studies with N = 59 and N = 26 subjects, respectively. In each study, the Psychomotor Vigilance Test (PVT) was administered alongside the DSST variants.

RESULTS

TSD reduced cognitive throughput on all DSST variants, with response time distributions exhibiting rightward skewing. All DSST variants showed practice effects, which were however minimized by inclusion of a pause between trials. Importantly, TSD-induced impairment on the DSST variants was not uniform, with a principal component analysis revealing three factors. Diffusion model decomposition of cognitive processes revealed that inter-individual differences during TSD on a two-alternative forced choice DSST variant were different from those on the PVT.

CONCLUSIONS

While reduced cognitive throughput has been interpreted to reflect general cognitive slowing, such TSD-induced impairment appears to reflect cognitive instability, like on the PVT, rather than general slowing. Further, comparisons between task variants revealed not one, but three distinct underlying processes impacted by sleep deprivation. Moreover, the practice effect on the task was found to be independent of the TSD effect and minimized by a task pacing manipulation.

Motorcycling performance and sleepiness during an extended ride on a dynamic simulator: relationship with stress biomarkers

OBJECTIVE

Powered two-wheelers (PTW) make up a large proportion of fatal accidents. The aim of this study was to investigate the effects of time-of-day and total sleep deprivation (SD) on simulated motorcycling performance during extended riding sessions (60 min), while evaluating stress mechanisms.

APPROACH

A total of 16 healthy males participated in four simulated motorcycling sessions at 07:00, 11:00, 15:00 and 19:00, including city (8 min), country (2 min) and highway pathways (40 min), after a normal night of sleep and after total SD (30 h), in a randomized counterbalanced order. The recorded motorcycle parameters included: variation of lateral position, number of inappropriate line crossings (ILC), falls, riding errors, speed and speed limit violations. Subject parameters included the number of microsleeps in each pathway, the number of lapses during the 3-min psychomotor vigilance task (PVT-Brief version), and the Karolinska sleepiness scale (KSS) score. Saliva samples were used to assess cortisol (sC), α-amylase (sAA), and chromogranin-A (sCgA). ANOVAs and Pearson's correlation analysis were performed between these variables.

MAIN RESULTS

Most parameters were influenced by an interaction effect between 'Motorcycling pathways' × 'SD' (speed (p < 0.05), legal speed violations (p < 0.01), variation of lateral position (p < 0.001), falls (p < 0.001), EEG-microsleeps (p < 005)). An interaction effect between 'SD' × 'Time-of-day' influenced the number of ILCs (p < 0.01), sC (p < 0.05) and sCgA (p < 0.05) levels. SD affected KSS scores (p < 0.001) and PVT lapses (p < 0.05). The highest disturbances were associated with highway motorcycling simulation.

SIGNIFICANCE

Sleepiness due to circadian or SD and fatigue effects significantly affect riding and increase the risks involved with PTWs. The activation of both stress systems seems not sufficient to alleviate these deleterious effects.

Sleep deprivation, vigilant attention, and brain function: a review

Vigilant attention is a major component of a wide range of cognitive performance tasks. Vigilant attention is impaired by sleep deprivation and restored after rest breaks and (more enduringly) after sleep. The temporal dynamics of vigilant attention deficits across hours and days are driven by physiologic, sleep regulatory processes-a sleep homeostatic process and a circadian process. There is also evidence of a slower, allostatic process, which modulates the sleep homeostatic setpoint across days and weeks and is responsible for cumulative deficits in vigilant attention across consecutive days of sleep restriction. There are large inter-individual differences in vulnerability to sleep loss, and these inter-individual differences constitute a pronounced human phenotype. However, this phenotype is multi-dimensional; vulnerability in terms of vigilant attention impairment can be dissociated from vulnerability in terms of other cognitive processes such as attentional control. The vigilance decrement, or time-on-task effect-a decline in performance across the duration of a vigilant attention task-is characterized by progressively increasing response variability, which is exacerbated by sleep loss. This variability, while crucial to understanding the impact of sleep deprivation on performance in safety-critical tasks, is not well explained by top-down regulatory mechanisms, such as the homeostatic and circadian processes. A bottom-up, neuronal pathway-dependent mechanism involving use-dependent, local sleep may be the main driver of response variability. This bottom-up mechanism may also explain the dissociation between cognitive processes with regard to trait vulnerability to sleep loss.

Differential and interacting effects of age and sleep restriction on daytime sleepiness and vigilance in adolescence: a longitudinal study

Study Objectives

There is contradictory evidence on whether sleep need decreases across adolescence. We investigated this question longitudinally with a dose-response design to test the effects of varied sleep durations on daytime sleepiness and on vigilance and to test whether these relations change with age across early and mid-adolescence.

Methods

Data from 76 participants who completed at least 2 years of the 3-year study are included in this report. Annually, participants ranging in age from 9.8 to 16.2 years completed three different time in bed (TIB) schedules each consisting of four consecutive nights of 7, 8.5, or 10 hours. Daytime sleepiness (multiple sleep latency test [MSLT]) and vigilance (psychomotor vigilance test [PVT]) were measured on the day following the fourth night of each TIB schedule.

Results

Electroencephalogram (EEG)-measured sleep durations changed linearly with TIB. MSLT-measured daytime sleepiness decreased with longer TIB and increased with age. The TIB and age effects interacted such that the TIB effect decreased with age. PVT performance improved with longer TIB and improved with age, but the benefit that increased TIB conferred on PVT performance did not change with age.

Conclusions

These results seem paradoxical because daytime sleepiness increased but vigilance improved with age. The significant age effect on the relation between TIB and sleepiness compared to the lack of an age effect on the relation between TIB and vigilance performance suggests different rates of maturation in underlying brain systems. We interpret these findings in relation to our model of adolescent brain development driven by synaptic elimination.

Further Evidence That Sleep Deprivation Effects and the Vigilance Decrement Are Functionally Equivalent: Comment on Altmann (2018)

Veksler and Gunzelmann (2018) argue that the vigilance decrement and the deleterious effects of sleep loss reflect functionally equivalent degradations in cognitive processing and performance. Our account is implemented in a cognitive architecture, where these factors produce breakdowns in goal-directed cognitive processing that we refer to as microlapses. Altmann (2018) raises a number of challenges to microlapses as a unified account of these deficits. Under scrutiny, however, the challenges do little to discredit the theory or conclusions in the original paper. In our response, we address the most serious challenges. In so doing, we provide additional support for the theory and mechanisms, and we highlight opportunities for extending their explanatory breadth.

Associations of Shift Work and Its Duration with Work-Related Injury among Electronics Factory Workers in South Korea

This study aimed to explore the association between shift work and work-related injuries. We collected data on workers from an electronics factory. This cross-sectional study included 13,610 subjects, who were assessed based on a self-reported questionnaire about their shift work experiences, work-related injuries, and other covariates. Multiple logistic regression models were used to evaluate the associations between shift work and work-related injuries and were estimated using the odds ratio. We found that the current and past shift workers, compared to non-shift workers, were associated with a 2.7- and 1.7-fold higher risk of work-related injury. There was a dose-response relationship between shift work duration and work-related injury among current female shift workers. Shift work increased the risk of work-related injuries, and the impact could be different depending on gender.