Neural mechanisms underlying the effects of physical fatigue on effort-based choice

Our decisions are profoundly influenced by our fatigue levels. After strenuous activity, we may avoid further exertion. Prior research identified brain regions (anterior cingulate cortex (ACC), anterior insula, ventromedial prefrontal cortex (vmPFC)) involved in effort-based decision-making, primarily in rested states. However, the impact of physical fatigue on these processes remains unclear. Previous studies showed decreased activity in motor and somatosensory cortices after fatiguing exertion, potentially reflecting reduced motor pathway recruitment. Inconsistencies between action consequences and sensory/motor outputs have been suggested as a fatigue source, implicating somatosensory regions and the posterior insula in generating fatigue feelings. While the posterior insula encodes proprioceptive signals and the anterior insula encodes effort value, the interaction of fatigue and prospective effort valuation requires further investigation. Prior studies either examined neural signals during exertion/rest without assessing prospective effort valuation, or focused on effort valuation without considering fatigue's influence. This study aimed to address these gaps and investigate how fatigue impacts subjective effort value and associated neural mechanisms. The hypotheses were: (1) behaviorally, fatigue increases subjective effort valuation, leading to decreased risk preferences for prospective effort; and (2) neurally, decisions about effort involve value signals encoded in the ACC and insula, with the insula being sensitive to fatigue-induced effort value changes. This forms a neurobehavioral model of how fatigue influences effort decisions.

Existing literature indicates a network of brain regions, including the ACC, bilateral anterior insula, and vmPFC, are crucial in computing the value of effortful options and making effort-based decisions. However, most studies have focused on rested states and not considered how physical fatigue might alter these valuations. Studies have reported neural correlates of physical fatigue in motor regions, with decreased activity in motor and somatosensory cortex after exertion. Transcranial magnetic stimulation (TMS) studies corroborate these findings, showing decreased motor cortical excitability post-exertion. The theory that fatigue stems from inconsistencies between action beliefs and sensory/motor feedback suggests the involvement of regions processing proprioceptive, exteroceptive, and interoceptive signals (somatosensory regions and posterior insula). While studies suggest the insula plays a mediating role between interoceptive and exteroceptive feelings and effort values, the interplay of these signals in prospective effort valuation under fatigue has not been adequately investigated. Previous studies, while including fatigue parameters, did not experimentally manipulate fatigue, hindering a conclusive understanding of its effects on effort valuation and brain activity. Studies on motor control have also explored internal models of effort value and their influence on movement decisions but lacked the neuroimaging component to understand brain mechanisms involved in effort value encoding. No studies have directly examined the impact of exertion-induced changes in bodily state on prospective effort valuation and decisions, leading to a limited mechanistic understanding of fatigue's influence on effort-based choice.

The study used fMRI to record brain activity during uncertain choices involving prospective physical effort, before and after bouts of exertion. The experimental design involved a baseline choice phase (170 trials) where participants chose between a sure option (lower effort) and a risky option (high effort or no effort). Following this, participants performed fatiguing exertion trials until exhaustion. This was followed by alternating blocks of effort choice and exertion trials to maintain a fatigued state. Ten trials were randomly selected for actual execution post-experiment. Participants used their left hand for choices and right hand for exertion, allowing lateral dissociation of motor signals. The effort choice paradigm, employing risky choices, allowed assessment of subjective effort valuation. The model used a power function (V(x) = (-x)^γ) to represent subjective effort cost, with 'p' representing sensitivity to effort cost changes and 'r' representing choice stochasticity. A control experiment (9 participants) involved two phases of effort choices without exertion to rule out a mere exposure effect. A second control experiment (17 participants) involved varying exertion levels (10U and 60U), interspersed with choice trials, self-reported fatigue ratings, and EMG recording to assess muscle fatigue using power spectral density analysis of the EMG signal. fMRI data were preprocessed (slice timing correction, motion correction, spatial transformation, smoothing) using SPM12. A general linear model (GLM) was used to analyze fMRI data, including regressors for choice trials (baseline and fatigue phases, chosen and unchosen options), exertion trials, and missed trials. Parametric modulators for effort values were included. Regions of interest (ROIs) based on previous literature were analyzed for activity related to effort value and fatigue. Small volume correction was applied to the ROIs for statistical inference.

Before exertion, most participants showed risk aversion for effort (pbaseline > 1). Repeated exertion resulted in a significant decrease in exertion repetitions and mean exertion levels, indicating fatigue. Fatigue significantly increased risk aversion for effort (pfatigue > pbaseline), reflecting an increased subjective cost of effort. This increase in effort cost was correlated with the rate of exertion decay during the initial exertion block. A control experiment showed that the behavioral changes were fatigue-specific, not due to mere exposure. fMRI analysis showed that the BOLD signal in the bilateral insula and ACC was modulated by the difference between chosen and unchosen effort values across both baseline and fatigue phases. The right anterior insula (rIns) showed increased sensitivity to chosen effort value in the fatigue phase but not baseline. This suggests that rIns activity reflects fatigue-induced changes in effort valuation. There was no significant correlation between rIns activity and fatigue-induced changes in risk preferences, indicating the activity primarily reflects effort valuation. Regions of left premotor (PM) and primary motor cortices showed significant activity decreases in the fatigue phase compared to baseline, consistent with previous studies of physical fatigue. Participants exhibiting larger fatigue-induced increases in subjective effort parameters showed less change in PM activity. Moderation analysis suggested that PM deactivation moderated the relationship between fatigue-induced reductions in motor performance and changes in subjective effort valuation. Control Experiment 2 confirmed that self-reported fatigue and EMG-measured muscle fatigue increased with higher exertion levels, with these changes persisting even after rest. Subjective effort parameters significantly increased between low and high exertion sections, and remained elevated after rest, consistent with sustained fatigue. A final analysis comparing high effort (80U) and low effort (10U) groups demonstrated that changes in choice preferences were specific to high-intensity exertion-induced fatigue and not simply a result of repeated exertion trials.

The study demonstrates that fatigue increases the subjective cost of effort, influencing effort-based decisions. The rIns plays a crucial role in representing fatigue-induced effort value changes, possibly by integrating proprioceptive and interoceptive information to assess the motor system's functional state. The findings suggest a mechanism where motor cortical state (PM) modulates effort value computation in the rIns. The lack of fatigue-induced PM deactivation among those with the highest increase in subjective effort value supports the theory that fatigue arises from miscalibration of motor cortical state, where an individual's motor system does not adequately adjust its drive in response to exertion, creating a discrepancy between perceived and actual capacity. The study also highlights the importance of considering both central and peripheral mechanisms in future research on fatigue. Although the current study focused on central fatigue mechanisms through neuroimaging, it is noted that peripheral mechanisms related to muscle fatigue undoubtedly contributed to effort valuations. The intermixing of choice and exertion trials in this study, in contrast to studies that separated decision-making and exertion, may have led to different neural findings from previous research, implicating ACC and insula in effort valuation in this specific experimental context.

This study provides strong evidence for the interplay between motor cortical state, interoceptive processing, and effort valuation during fatigue. The findings suggest a neurobiological account of how bodily state influences effort-based decisions, indicating that a miscalibration of motor cortical state, specifically in the PM, underlies the increased subjective cost of effort in a fatigued state. Future research should investigate the interaction of central and peripheral fatigue mechanisms and further refine the understanding of interoceptive and proprioceptive signals' contributions to effort valuation.

The study primarily focused on central fatigue mechanisms, and peripheral mechanisms were not explicitly investigated. The number of trials in the varying exertion sections of the control experiment was relatively small compared to the main experiment, potentially affecting the precision of parameter estimates. The use of a priori ROIs might limit the detection of additional brain areas involved. The current study did not explicitly measure participants’ interoceptive and proprioceptive sensations during decision making which could affect the interpretation of results.