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

Figures

• 

  Fig. 1. Experimental design. a During effort-based choice trials participants were presented with a series of choices between a risky option that could result in either a large level of exertion or no exertion with equal probability (Flip), or exerting a lower level of effort with certainty (Sure). The effort amounts were presented on a 0 to 100 scale, which participants were trained on during an association phase prior to choice. An effort level of zero corresponded to no exertion and 100 to 80% of a participant’s maximum exertion. To study the effects of fatigue on effort-based decision-making, blocks of exertion trials were interspersed with blocks of effort-based choice trials. During exertion trials, participants exerted grip force on a hand dynamometer to fill a horizontal bar, where the full bar corresponded to an effort level of 80. b Experiment schedule. The experiment was divided into baseline and fatigue choice phases, which were both scanned with fMRI. The baseline choice phase consisted of 170 effort-based choices presented in a randomized order. Following the baseline choice phase, participants performed the fatigue choice phase of the experiment, in which they underwent…

• 

  Fig. 2. Behavioral results (n = 20). a Exertion trial performance during the fatigue choice phase. Performance is represented as the mean number of repetitions until the exertion block termination criterion was met, across participants. The initial exertion block is indicated in the shaded gray region. Participants’ ability to successfully achieve the performance criteria diminished across exertion blocks, and participants consistently performed the same number of exertion trial repetitions after the second exertion block. Error bars indicate SEM. Paired t-test (two-tailed): ***p = 0.0002. b Mean exertion during the fatigue choice phase. Mean exertion is presented in units of effort. For the initial exertion block (shaded gray), the mean exertion for the first and last five trials is shown. For subsequent exertion blocks the mean exertion over the full exertion block is shown. Mean exertion decreased between the initial and final trials in the first exertion block and mean exertion never recovered to its initial level during the fatigue choice phase. Error bars indicate SEM. Paired t-test (two-tailed): ***p = 0.003. c The function used to model the subjective cost of effort. This…

• 

  Fig. 3. Neural signatures of chosen effort value (n = 20). a Overall effort value encoding. Whole brain results thresholded at voxelwise p < 0.005 (red) and p < 0.001 (yellow). Activity in bilateral insula (rIns: peak = [34, 20, 2]; lIns: peak = [−32, 24, 2]; small volume corrected p < 0.05 in a priori ROI) reflects the difference in chosen and unchosen effort value at the time of choice, across both the baseline and fatigue choice phases. Activity was also observed in ACC (MNI coordinate: peak = [−8, 24, 42]), although it does not survive small volume correction in our a priori ACC ROI. b Effort value encoding in rIns increases with fatigue. Activity in rIns (peak = [34, 18, 2]; small volume corrected p < 0.05 in a priori ROI) reflects the difference of chosen and unchosen effort value between the baseline and fatigue choice phases. c Effects in rIns (5 mm sphere centered at [34, 18, 2]) for chosen and unchosen effort value in the baseline and fatigue choice phases are shown. This plot was not used for statistical inference (which was performed in the SPM framework) and is shown solely to illustrate the pattern of the BOLD signal. Error bars indicate SEM.

• 

  Fig. 4. Neural signatures of exertion-induced fatigue (n = 20). a Exertion-induced changes in brain activity. Whole brain results thresholded at voxelwise p < 0.005 (red) and p < 0.001 (yellow). Activity in primary motor cortex (peak = [−38, −30, 52]) and PM (peak = [−36, −18, 54]) was reduced in the fatigue condition, relative to baseline, at the time of choice. These activations are significant small-volume corrected p < 0.05 within our a priori premotor ROI. b Illustration of the PM (5 mm sphere centered at [−36, −18, 54]) effects reported in (a). This plot was not used for statistical inference, which was performed in the SPM framework. Error bars indicate SEM. c Bold signal in PM is positively modulated by fatigue-induced increases in ρfatigue relative to ρbaseline. Between-participant regression analysis entering the change in effort subjectivity ρfatigue–ρbaseline as a covariate for the decreasing activity between the fatigue and baseline choice phases in PM (peak = [−32, −14, 58]; small volume corrected p < 0.05 in a priori ROI). d Relationship between activity in PM (5 mm sphere centered at [−32, −14, 58]) and the degree of fatigue-induced change in effort subjectivity…

• 

  Fig. 5. Experimental Design of Control Experiment 2. a At the beginning and end of each exertion block, participants were queried on their level of agreement (“Strongly disagree”, “Somewhat disagree”, “Unsure”, “Somewhat agree”, “Strongly agree”) with the statement “I feel fatigued”. Participants were free to select anywhere on the scale (not just the indicated points), and there was no time constraint for them to select their rating. b Experiment schedule. The control experiment was divided into baseline and modified fatigue choice phases. The baseline choice phase consisted of the same 170 effort-based choices used in the main experiment. Following the baseline choice phase, participants performed the modified fatigue choice phase, in which they alternated between exertion trials and choice trials. The modified fatigue choice phase began with participants alternating between low effort exertions (10U; light gray) and blocks of choice trials (white). Following this sequence of exertion/choice, participants alternated between high effort exertions (60U; dark gray) and blocks of choice trials. Participants were then given a two minute period of rest in which they did not make any…

• 

  Fig. 6. Results from Control Experiment 2 (Control Experiment 2; n = 17). a Exertion trial performance during the modified fatigue choice phase. Performance is represented as the mean repetitions required to successfully perform five exertions (successful trials need not be consecutive); higher values indicate more exertion trials were required in order to successfully exert at the target effort five times. Error bars indicate SEM. b Mean exertion during the modified fatigue choice phase. Mean exertion is presented in units of effort. Error bars indicate SEM. c Self-reported measures of participants’ state of fatigue over the course of the modified fatigued choice phase. The plot shows mean self-reported ratings of fatigue at the beginning and end of the 17 exertion blocks. Self-reported fatigue persisted when the exertion task returned to low effort trials after a two minute rest. Error bars indicate SEM. d Mean frequency of the power spectrum of EMGs between first and second low effort sections of the modified fatigue choice phase. The plot shows mean frequency measures from EMGs for all successful exertions averaged within the first and second low effort (10U) sections. Mean frequency…

• 

  Fig. 7. Comparison of choice preferences during high and low exertion efforts, controlling for the number of exertion repetitions (participants from the Main Experiment and Control Experiment 2; n = 37). a We compared behavior during the first 25 fatiguing exertions between the main experimental group (first 3 exertion and choice blocks) and the control experimental group (first 5 exertion and choice blocks). In the main experiment participants exerted at a level of 80U, and in the control experiment participants exerted 10U. Participants did not perform a significantly different number of exertion trials between these experimental epochs (Paired t-test (two-tailed): t34 = 0.08, p = 0.94). Error bars indicate SEM. b Participants exerted significantly more effort in the main experimental group compared to the control experimental group (Unpaired t-test (two-tailed) comparing the mean effort produced between groups: t34 = 38.83, p < 0.001). Error bars indicate SEM. ***p = 9.44 × 10−30. c Comparison of choice similarity metrics between the main fatigue experiment and control experiment, averaged across the selected epoch. Relative to the baseline choice phase, positive values indicate more…