2022-2023 Doctoral Awardee: Fareshte Erani

Testing an Effort-Reward Imbalance Framework for Cognitive Fatigue in Individuals with Multiple Sclerosis

Fareshte Erani, Drexel University

Project Abstract 

Fatigue is experienced in up to 95% of persons with multiple sclerosis (PwMS). In over half of PwMS, the cognitive component of fatigue is reported to be the most distressing aspect of their fatigue, negatively affecting their ability to engage in work and quality of life. Previous research has examined the influence of reward and effort separately on cognitive fatigue; however, these processes have not been explicitly integrated to test whether the interaction between effort and reward leads to cognitive fatigue. To address this gap, we aim to evaluate whether an effort-reward imbalance predicts cognitive fatigue in PwMS. We recruited 20 individuals, 18 to 65 years old, with a pre-existing diagnosis of MS and 20 cognitively healthy controls. In a within-subjects design, participants provided responses on psychosocial, neuropsychological, and subjective fatigue measures. A DTI scan and fMRI data were collected during a computerized switching task with independent effort and reward manipulations. If cognitive fatigue results from combinations of dysfunctional effort and reward processes, evidenced by both behavior and pathophysiology, we could develop a model to predict cognitive fatigue trajectory and guide cognitive fatigue assessment and interventions in MS and other neurological syndromes.

Introduction/Background 

Fatigue is one of the most common symptoms in neurological and psychiatric syndromes^1,2. However, its nature has made it one of the most controversial and least understood phenomena despite its prevalence. The primary goal of this study is to test a novel theoretical approach to understanding fatigue, beginning with cognitive (mental) fatigue. In order to inform assessment and treatments for fatigue, we must better understand and clarify the mechanisms underlying this symptom. Cognitive neuroscience and behavioral economics offer an explanation of cognitive fatigue through an effort-reward imbalance framework. Drawing from models that describe the costs and benefits of effortful cognition³ and its neuroanatomical basis in fronto-striatal circuitry⁴, the effort-reward imbalance framework suggests that cognitive fatigue may be the brain’s way of signaling to itself that the effort required for a task no longer merits the rewards received⁵. In this framework, cognitive fatigue functions as a signal to abandon behavior when effort is high, and rewards are comparably low. Despite evidence suggesting that the effort-reward framework could be true^6–10, studies testing the framework examine effort and reward on cognitive fatigue separately. However, effort, reward, and the role of the fronto-striatal circuit in predicting cognitive fatigue have not been explicitly integrated and tested together.

To address this gap, my dissertation tests the effort-reward imbalance framework by manipulating effort and reward in the same paradigm to explain cognitive fatigue. This study focused on persons with MS (PwMS), given the high prevalence of fatigue experienced in this population and the disease’s association with anatomical damage to the brain’s white matter. We hypothesized that cognitive fatigue results from a mismatch between effort and reward processing. We specifically predicted that increased effort and decreased reward levels would be associated with greater cognitive fatigue when performing a task. If cognitive fatigue results from an effort-reward imbalance in fronto-striatal circuitry, then we expect it to be a function of both anatomical integrity and the ability to functionally recruit the circuit. Thus, the integrity and functional recruitment of fronto-striatal circuitry that regulates effortful cognition and reward processing will contribute to cognitive fatigue.

To summarize, our framework makes specific predictions for clinical populations in which fatigue is a prevalent symptom. For example, if PwMS have dysfunction in both effort-related and reward systems, they would cognitively fatigue the fastest and to the greatest degree, providing one explanation for differences in cognitive fatigue between clinical populations who experience fatigue and healthy populations. Thus, this framework fundamentally suggests that cognitive fatigue in neurological conditions such as MS could be an amplified version of that which exists in healthy populations.

Methods

The study used a within-subjects design to test the effort-reward framework by integrating effort and reward in the same paradigm. Our paradigm was a computerized switching-task¹¹ that we administered in-scanner that manipulated both levels of effort and reward. Data collection occurred during one in-person session lasting approximately 2.5-3 hours. We collected data on demographic, psychosocial, neuropsychological measures, and self-reported fatigue and task performance.

Participants. Based on an a priori power analysis that revealed at least 36 subjects were required to detect small-sized differences, we aimed to enroll at least 40 subjects (20 per group) for the study. Individuals over the age of 65 were not included in this study to mitigate the chance of an undiagnosed second disorder (i.e., progressive neurodegenerative diseases). Individuals diagnosed with MS with a relapsing-remitting (RR) subtype were recruited to maintain a uniform phenotype for this initial study. RRMS also represents a larger percentage of the MS sample with less nuanced behavioral and brain variance which is common in other MS subtypes. To further limit the impact of physical limitations, inclusion criteria also included individuals who can ambulate at least 500 feet without assistance (EDSS<6). Participants with MS were at least 4 weeks past their most recent exacerbation and any use of steroids, benzodiazepines, stimulants, or neuroleptics. Exclusion criteria included: 1) history of neurological disease other than MS, 2) severe psychiatric diagnosis, 3) substance abuse diagnosis, 4) recent (within 4-week) symptom exacerbation, 5) history of anemia, 6) history of known structural brain abnormality, 7) acute medical illness, and 8) any contraindications to MRI. The healthy control group was matched to the MS group on age, sex, and education level. Inclusion criteria included English fluency. Exclusion criteria for the control group was the same as the MS group with an additional exclusion for a history of MS.

Measures. Fatigue: Fatigue measures included the Visual Analog Scale for Fatigue (VAS-F)¹² to measure cognitive fatigue during the task.

Psychosocial and Neuropsychological Measures: Psychosocial measures included demographic and medical history, depressive symptoms, state and trait sleepiness, and reward sensitivity and behavioral inhibition. We administered a brief neuropsychological battery which included measures of verbal learning and memory, set-shifting, processing speed, semantic fluency, inhibition and switching, motor speed, and MS-related disability.

Computerized Switching Task: We used a computerized cognitive switching-task adapted from a validated dual-task paradigm¹¹. The task rules engage in switches between two mathematical rules. The task manipulated effort by requiring two different levels of effort during high-demand (switching between two rules) and low-demand (following one rule only) conditions¹¹. The task also included two levels of reward (high-reward, low-reward) presented at the end of each block. Together the task had four blocks and allowed us to test the framework by looking at the separate effects of effort and reward and their interaction.

Task Performance: Task performance was measured byresponse time (RT) and accuracy during the computerized switching task.

Neuroimaging: fMRI data was collected during the switching task to collect information about fronto-striatal cerebral activation. A standard anatomical scan, T2 scan to estimate lesion load, and a DTI scan were also collected to obtain information about fronto-striatal tract integrity.

Procedures. After individuals indicated their interest in the study, inclusion/exclusion criteria were delineated. If the participant met criteria, a study session was scheduled. Participants provided responses to the demographic surveys, neuropsychological battery, and psychosocial measures. In the scanner, they underwent T1-weighted structural image, resting-state fMRI, and then the block-related fMRI where they performed two runs (four blocks of the switching task in a randomized order), and completed the VAS-F items before the start of the switching-task and after each block to capture cognitive state fatigue. After the task, FLAIR, DTI-HARDI, and coplanar T2 scans were collected. MRI imaging parameters were consistent with those used in other protocols in our lab selected for high signal-to-noise ratios.

Results

We enrolled 20 individuals, 18 to 65 years old, with a pre-existing diagnosis of MS and 20 cognitively healthy controls. Data analyses is currently underway. Several covariates were collected in both groups to control for potential confounding variables: age, sex, education, depression, sleepiness, neuropsychological measures, and subjective trait fatigue. An independent samples t-test will be used to determine if the MS and control groups significantly differ on any of the continuous measures. Any variables with a statistically significant difference between groups will be included as covariates in our models. The hypotheses will be tested using mixed effects linear regression models within the R statistical software lme4 package¹³.

Our hypotheses will be tested using two mixed effects linear regression models. The dependent variable in both models will be cognitive fatigue (VAS-F scores). In the first model, effort- and reward-level, and their interaction will be the independent variables. In the second model, lesion load, cerebral activation within our fronto-striatal ROIs, and their interaction will be the independent variables. We will use a stepwise regression in each model to test whether each main effect and interaction contributes significant variance, set at p=0.05. We will also test group membership and interaction between group, fronto-striatal lesion load, and cerebral BOLD activation within the ROIs. Exploratory analyses will create models with RT and accuracy as dependent variables, with the same structure of independent variables described above.

 We expect that if the tradeoff between effort and reward is affected in PwMS, cognitive fatigue will increase relative to the age- and sex-matched controls. Specifically, increased effort or decreased reward levels would be associated with greater subjective cognitive fatigue when performing a cognitive task. Additionally, if cognitive fatigue results from an effort-reward imbalance in fronto-striatal circuitry, then we expect it to be a function of both anatomical integrity and ability to recruit the circuit. Crucially, the interaction between effort and reward will predict cognitive fatigue above and beyond either function alone.

Discussion

A better understanding of the causes of cognitive fatigue will enhance our ability to detect and develop precise treatments for those at greatest risk for cognitive fatigue. If the framework is supported, it will directly lead to new studies on the link between disease pathophysiology and cognitive fatigue as a function of burdens to fronto-striatal effort and reward processing. Additionally, the behavioral task could be used as an assessment tool that examines both sensitivities to reward and cognitive control to predict both task performance and cognitive fatigue. Moreover, utilizing the framework allows for a precise target for interventions and treatments. For example, the framework and neural correlates could inform mechanism-driven therapies, including non-invasive brain stimulation (NIBS). The framework would not only build tasks and assessments that relate to mechanisms in the brain but also provide information to then target and modulate with NIBS or other mechanisms.

Impact Statement

Fatigue is an example of a symptom that, despite its prevalence, remains poorly understood and is often overlooked in clinical settings because of its complex and multifaceted nature. The current study has made progress on a 100-year problem by combining novel methods from neuroscience, neuropsychology, and behavioral economics to test a framework that could explain cognitive fatigue. The current research could lead to reliable, valid, and efficient computerized tasks for routine clinical care that measure dysfunction leading to cognitive fatigue, and our neuroimaging analyses could point to personalized neural targets useful for behavioral, pharmaceuticals, and brain stimulation treatments. My long-term goal is to continue to focus on what underlies difficult-to-manage symptoms so we can target those mechanisms and facilitate symptom relief for patients.

References

1. Merkelbach, S. et al. Fatigue, sleepiness, and physical activity in patients with multiple sclerosis. J. Neurol. 258, 74–79 (2011).
2. Sadaghiani, S. & D’Esposito, M. Functional Characterization of the Cingulo-Opercular Network in the Maintenance of Tonic Alertness. Cereb. Cortex 25, 2763–2773 (2015).
3. Boksem, M. A. S. & Tops, M. Mental fatigue: Costs and benefits. Brain Res. Rev. 59, 125–139 (2008).
4. Chaudhuri, A. & Behan, P. O. Fatigue in neurological disorders. The Lancet 363, 978–988 (2004).
5. Wylie, G. R., Genova, H. M., DeLuca, J. & Dobryakova, E. The relationship between outcome prediction and cognitive fatigue: A convergence of paradigms. Cogn. Affect. Behav. Neurosci. 17, 838–849 (2017).
6. Dobryakova, E., DeLuca, J., Genova, H. M. & Wylie, G. R. Neural Correlates of Cognitive Fatigue: Cortico-Striatal Circuitry and Effort–Reward Imbalance. J. Int. Neuropsychol. Soc. 19, 849–853 (2013).
7. Dobryakova, E. et al. Fronto-striatal network activation leads to less fatigue in multiple sclerosis. Mult. Scler. J. 24, 1174–1182 (2018).
8. DeLuca, J., Genova, H. M., Hillary, F. G. & Wylie, G. Neural correlates of cognitive fatigue in multiple sclerosis using functional MRI. J. Neurol. Sci. 270, 28–39 (2008).
9. Dobryakova, E. et al. Reward presentation reduces on-task fatigue in traumatic brain injury. Cortex 126, 16–25 (2020).
10. Kohl, A. D., Wylie, G. R., Genova, H. M., Hillary, F. G. & Deluca, J. The neural correlates of cognitive fatigue in traumatic brain injury using functional MRI. Brain Inj. 23, 420–432 (2009).
11. Botvinick, M. M., Huffstetler, S. & McGuire, J. T. Effort discounting in human nucleus accumbens. Cogn. Affect. Behav. Neurosci. 9, 16–27 (2009).
12. Shahid, A., Wilkinson, K., Marcu, S. & Shapiro, C. M. Visual Analogue Scale to Evaluate Fatigue Severity (VAS-F). in STOP, THAT and One Hundred Other Sleep Scales (eds. Shahid, A., Wilkinson, K., Marcu, S. & Shapiro, C. M.) 399–402 (Springer New York, 2011). doi:10.1007/978-1-4419-9893-4_100.
13. Bates, D., Mächler, M., Bolker, B. & Walker, S. Package Lme4: Linear Mixed-Effects Models Using Eigen and S4. R package version vol. 67 (2014).