Effects of Perceived COVID-19 Exposure and Action-Outcome Predictability on the Motivation to Invest Cognitive Effort.

Abstract: Everyday life situations characterized by poor controllability because of restrictions and uncertainty about action outcomes may attenuate motivational states and executive control. This article explores the interaction of a prior experience with COVID-19 and the susceptibility to respond to a challenging situation with low action-outcome predictability. We assessed cognitive effort readiness as the willingness to invest in cognitively demanding tasks. Individuals with a COVID-19 history exhibited a more pronounced reduction in cognitive effort readiness after experiencing experimentally induced action-outcome unpredictability compared to controls. These results suggest a generalization of perceived loss of action-outcome control among individuals with a COVID-19 history. These findings contribute to conceptualizing and assessing the long-term consequences of pandemic-induced emotional and motivational problems.

Zusammenfassung: Alltägliche Lebenssituationen, die durch eine geringe Kontrollierbarkeit aufgrund von Restriktionen und Unsicherheiten bezüglich der Handlungsergebnisse gekennzeichnet sind, können zu einer Beeinträchtigung motivationaler Zustände sowie der exekutiven Kontrolle beitragen. Hier untersuchten wir die Wechselwirkung einer subjektiv erlebten COVID-19-Infektion und der Disposition, auf eine herausfordernde Situation mit geringer Vorhersagbarkeit der Handlungsergebnisse zu reagieren. Dazu wurde die kognitive Anstrengungsbereitschaft als das potentielle Engagement in anspruchsvolle Arbeitsgedächtnisaufgaben zu investieren erfasst. Personen mit einer COVID-19-Vorerfahrung zeigten im Vergleich zu Kontrollpersonen eine stärkere Reduktion der kognitiven Anstrengungsbereitschaft, nachdem sie eine experimentell induzierte Situation mit geringer Vorhersagbarkeit der Handlungsergebnisse erlebt hatten. Diese Ergebnisse deuten auf eine möglicherweise verstärkte Generalisierung wahrgenommener Kontrollverluste bei Personen mit COVID-19-Vorerfahrungen hin. Über die direkten Auswirkungen von COVID-19 hinaus, tragen diese Beobachtungen zu einer besseren Konzeptualisierung und Erfassung von langfristigen Pandemie-Folgen, insbesondere emotionaler und motivationaler Beeinträchtigungen, bei.

Keywords: cognitive effort; Expected Value of Control Theory; COVID-19; cognitive control; learned helplessness; kognitive Anstrengungsbereitschaft; kognitive Kontrolle; erlernte Hilflosigkeit

The Coronavirus disease 2019 (COVID-19) pandemic instigated unparalleled and unforeseen transformations across almost every aspect of life, prompting a redefinition of social norms and rules. Previously socially rewarded behaviors, such as engaging in social gatherings, underwent a shift, becoming viewed as potentially harmful and irresponsible. Adapting to the novel obligations and legal regulations thus necessitated a redefinition of behavioral strategies and suppression of previously learned goal-directed behaviors. The constraints imposed by pandemic restrictions posed challenges to established individual habits and norms, introduced uncertainties, and, in some cases, caused a loss of control across various behavioral domains ([2]; [45]).

Concurrently, extensive media coverage concerning the potential long-term health implications of COVID-19 created ambiguity and concerns that persist today. Specifically, potential cognitive consequences and persistent exhaustion, commonly called "brain fog," attracted considerable attention and stimulated concerns about long-term negative effects, even following mild infections ([65]). One overlapping symptom between the Chronic Fatigue Syndrome (CFS)/Myalgic Encephalomyelitis (ME) and possible long-term consequences of COVID-19 was characterized as enduring fatigue unmitigated by rest and unrelated to cognitive or physical activities ([16]). In addition, CFS and depressive disorders may mutually intensify each other. Anhedonia – a general lack of motivation and willingness to exert effort for goal-oriented behavior – emerges as a prominent symptom in both conditions ([64]). This holds considerable importance, as in the postpandemic period, emotional and motivational changes endure, including reduced overall well-being, depressive symptoms, and persistent anxiety disorders. Many of these changes may be attributed to pandemic-related restrictions, such as reduced social interactions and less regulated daily routines ([40]).

Individuals may experience persistent symptoms, such as fatigue, concentration difficulties, affective complaints, and pain even after a mild infection ([11]; [67]). In particular, fatigue remains a predominant long-term consequence following COVID-19, with reduced motivational functioning identified as a core symptom and the most frequent manifestation ([50]; [57]). Although the link between viral infections, such as COVID-19, and motivational alterations remains underexplored, existing evidence suggests that inflammatory processes triggered by viral infections may affect the reward system and consequently influence adaptive behavior ([20]). Over the long term, contextual factors of the COVID-19 pandemic, including increased anticipation of negative health outcomes and stress caused by the perceptions of uncertainty and low action-outcome predictability, may independently contribute to motivational problems.

This study investigated the effects of an experimentally designed situation with low outcome predictability to assess the participants' motivation to exert cognitive effort. It compared individuals who had previously experienced a COVID-19 infection with controls who professed no awareness of having been exposed to the virus. Thus, the focus of this study extends beyond the direct effects of COVID-19 to encompass stress arising from perceived cognitive uncontrollability during the pandemic and potential long-term health consequences (for an overview, see [76]). The subsequent sections elaborate on the cognitive sequelae of COVID-19, the broader psychological effects related to poor outcome predictability, and new behavioral methods to explore motivational functioning.

Following the pandemic, general mental health scores, reflecting anxiety, stress, and affective complaints, remained elevated in many individuals, irrespective of their COVID-19 history ([19]; [81]). Academic performances declined compared to prepandemic levels, a trend attributed to various contextual factors associated with the pandemic ([46]). Even in mild or asymptomatic COVID-19 cases, cognitive deficits may emerge, often without the individuals' awareness but detectable through neuropsychological screenings ([1]; [79]). A comprehensive longitudinal study encompassing various cognitive domains such as executive functioning, long- and short-term memory, and logical reasoning ([70]) revealed significant alterations following mild to severe COVID-19 infections. In addition, it also discovered that emotional changes and increased anxiety persisted for 6–9 months. Notably, the severity of infection did not uniformly correlate with long-term deficits. For instance, executive functioning exhibited independent reductions, with a remarkable lack of awareness ([70]). These findings suggest that undetected executive dysfunction may occur even in individuals who experienced a mild course of COVID-19.

Research exploring the link between viral infections and alterations in motivation is limited. Yet, the evidence suggests that inflammatory processes, possibly triggered by the infection, may compromise the reward system and, consequently, adaptive behavior ([20]). Irrespective of an actual COVID-19 infection, pandemic-related factors, such as increased anticipation of negative health outcomes, may stimulate subjective deficit awareness and increase stress. Indeed, affective symptoms during and after the COVID-19 pandemic surpass prepandemic prevalence ([43]). Affective disorders, in turn, may influence cognitive processes, including associative learning, cognitive control, and motivational functioning, consequently affecting effortful behavior ([22]; [33]). Particularly reduced associative learning and cognitive control may compromise adaptive responsiveness to environmental stimuli and influence behavior such as effort expenditure ([68]).

The relationship between willingness to expend effort and well-being appears to be nonlinear and influenced by additional factors. For example, persistent engagement in a task without considering the costs and benefits of the goal can lead to increased stress and may result in depressive symptoms ([42]; [78]). Conversely, early disengagement may lead to suboptimal performance ([31]), potentially leading to failure and depressive symptoms.

The COVID-19 pandemic poses new challenges for individuals when evaluating the costs and benefits of their efforts. These challenges were marked by increased unpredictability and a feeling of reduced control over how their actions might be associated with specific outcome contingencies. The motivational intensity theory ([6], [7]) suggested that effort willingness relies on factors such as demand, value, and attainability, with a tipping point when a goal seems unattainable or its value does not justify the required effort. Negative mood and affect can additionally increase the perception of task difficulty and reduce the willingness to exert effort ([22]). Decreased willingness to expend effort has been observed in circumstances similar to the COVID-19 pandemic, such as in situations in which responses were not predictive of the outcome. For example, in an animal study, [59]) demonstrated that uncontrollable shocks resulted in learned helplessness. The subjective perception of diminished action-outcome control during the COVID-19 pandemic might have contributed to reduced motivation and fatigue, partly attributable to the experience of uncontrollability generalizing to other domains of life ([10]).

Cognitive and behavioral theories have provided insights into reduced effort willingness during the COVID-19 pandemic, suggesting that learned processes and anticipated negative health consequences may contribute to diminished motivation. These theories are compatible with neuroscientific findings (for an overview, see [30]). The expected value of control (EVC) theory highlights different brain areas involved in assessing the subjective net value of information related to value, cost, and utility. These structures include the insula, ventral prefrontal cortex (PFC), amygdala, midbrain, and striatum (for an overview, see [63]). The dACC plays a critical role by receiving various afferent inputs, including information on executive intrinsic costs from the lateral PFC and inputs from the amygdala. It then assesses the costs and benefits, forwarding the outcome of this evaluation to the lateral PFC. This feedback mechanism to the lateral PFC facilitates the incorporation of executive costs into the overall net value.

Therefore, anticipated costs, including cognitive effort, benefits, and attainability, are incorporated into the decision about effort engagement. Alterations in the weighting of these variables – possibly induced by increased cognitive effort, reduced outcome value, and unpredictability – are likely to change the outcomes of willingness to invest effort in cost-benefit analyses.

In addition, the COVID-19 infection itself might trigger alterations in the reward system, with a potential connection between inflammation and motivational symptoms ([12]; [77]). Supporting evidence comes from autoimmune diseases such as systemic lupus erythematosus, multiple sclerosis, type 1 diabetes, celiac disease, and rheumatoid arthritis, which often include motivational impairments ([82]). The Research Domain Criteria (RDoC) advise that affective and motivational impairments should not solely describe symptoms but also consider their underlying neuronal impairments ([27]). In line with RDoC, [48]) outlined several pathways by which COVID-19 infections might lead to cognitive, emotional, and motivational changes, including a pathological inflammatory response that potentially induces alterations in the reward system, for instance, in dopamine homeostasis ([66]).

Dopamine plays a central role in assessing motivational processes and cognitive control, encompassing the evaluation of effort as worthwhile, instrumental learning, hedonia, and decision-making. Dopaminergic projections establish connections between the ventral tegmental area (VTA), the ventral striatum, and the frontal regions of the neocortex. The VTA maintains a tonic baseline transmission level, which is modulated by phasic changes resulting from positive (phasic increase) or negative (phasic decrease) events or event expectations ([14]; [66]; [75]; for an overview, see, e. g., [30]). Even minor alterations of these interactions caused by COVID-19 or pandemic-related stress could lead to profound changes in motivation and behavior.

The limited research on the impact of COVID-19 on motivation may stem from inherent problems in assessing motivation. While various questionnaires and semistructured interviews encompass emotional, motivational, and executive aspects, self-assessment remains critical in participants with reduced self-awareness. Furthermore, even in individuals with intact self-awareness, one cannot rule out systematic bias because of socially desired expectations. Recognizing that questionnaires or open-end interviews may not fully capture effort and willingness to expend effort; adopting specific behavioral tasks or psychophysiological measures becomes imperative ([30]). Relying on a singular measure to assess motivation or fatigue may not be appropriate. Instead, employing a multifaceted approach with measurements across various levels of behavior appears more adept to comprehensively capturing the complex nature of motivation. The Cognitive Effort Discounting Paradigm (COG-ED) ([73]) employs a forced-choice approach, requiring participants to decide between engaging in a low-cognitive effort task with a smaller monetary reward or a high-cognitive effort task with a larger monetary reward. This method has proved to possess strong reliability ([38]), in addition to demonstrating ecological validity and the ability to discriminate effectively ([18]). This paradigm is a favorable approach to motivation as one can evaluate the subjective value of cognitive effort (SV) without executing the chosen tasks and therefore control for confounding factors.

Taken together, prior experiences with COVID-19 infections may result not only in varying levels of enduring cognitive impairment as described by the long-/post-COVID-19 syndrome, but it may manifest as motivational problems and fatigue in vulnerable individuals who even experience only a mild initial course of the disease. This study explored the potential degree of motivational problems in individuals who experienced COVID-19 without overt cognitive symptoms and the relevance of stress caused by experienced unpredictability, resembling some of the situations of unpredictability during the pandemic. It specifically investigated the effects of perceived COVID-19 exposure and susceptibility to low action-outcome predictability on the willingness to expend cognitive effort. By examining the interaction between the prior COVID-19 experience and experimentally induced stress caused by unpredictability on cognitive effort, we aim to elucidate the subtle cognitive and motivational alterations following COVID-19 infections. We hypothesize that poor action-outcome predictability diminishes the willingness to exert cognitive effort, rendering it less adaptive in individuals reporting a previous COVID-19 infection.

We recruited 48 participants (age range: 18–34 years, mean age: 23.2 ± 3.5 years; 17 males) from the University of Marburg for inclusion in the present study. Exclusion criteria encompassed noncompensable visual impairments, auditory deficits, psychiatric conditions, any overt symptoms following COVID-19, and any documented chronic inflammatory conditions. Prior to their engagement, the participants provided written informed consent and underwent a comprehensive debriefing procedure upon completion of the study. Compensation for participation involved either academic credit points or monetary remuneration. Furthermore, a modest financial incentive was contingent on the performance in the effort discounting task. We excluded 7 participants because of language barriers, attrition at the second assessment, or PCS syndrome. We categorized the remaining participants into two groups: those with an experienced history of previous COVID-19 infection (n = 23, female = 15, mean age: 24.2 ± 3.7 years) and controls without a previous COVID-19 infection (n = 18, female =11, mean age: 21.9 ± 2.32 years).

Participants were seated approximately 125 cm from a 20-inch monitor (BenQ, 1680×1050 pixels) that presented experimental stimuli and questionnaires. A dim-gray screen with white fonts (Calibri, size 40) displayed all verbal instructions, and behavioral responses were recorded using a QWERTZ keyboard (Cherry KC 1000). We delivered the auditory stimuli through headphones (EPOS Sennheiser ADAPT 165 II). The participants had to keep the headphones on throughout the entire experiment. The experiment was programmed utilizing PsychoPy software (version v2021.2.3, [49]). We administered the questionnaires via SoSci Survey ([35]).

The procedure consisted of two sessions conducted within a 3-day period, with both sessions scheduled at the same time of day. Each session included identical behavioral tasks. The initial session, lasting 75 min, was extended to approximately 100 min during the second session because of the inclusion of supplementary questionnaires. The sessions started with the individual calibration of the auditory feedback volume, followed by a WM task with varying levels of difficulty. We assessed perceived workload between these tasks using the Subjective Workload Assessment Technique (SWAT, [52]). Following the WM task, we conducted a baseline measurement of effort expenditure as part of the COG-ED paradigm ([73]) as well as an assessment of emotional well-being using the Positive and Negative Affect Schedule (PANAS, [8]). Subsequently, participants engaged in a deductive reasoning task, with contingent or randomized outcome feedback. We implemented stratification randomization, considering gender and feedback application order; feedback type was counterbalanced. We then reassessed the PANAS and the COG-ED paradigm. Postsessions, the participants completed screen questionnaires. The participants were explicitly encouraged to invest maximal effort in the behavioral task and informed of potential additional compensation based on their decisions in the effort investment task. This study adhered to the principles outlined in the latest iteration of the Helsinki Declaration ([80]) and received approval from the Ethics Committee of the Faculty of Psychology at the Philipps University Marburg.

COVID-19 exposure and long-term symptoms were evaluated using a comprehensive questionnaire-based assessment. We assessed chronic fatigue symptoms specifically associated with COVID-19 using a screening questionnaire from the Charité Fatigue Center ([53]). This questionnaire adhered to the core symptoms outlined in the Long-COVID-19 Guidelines established by the German Association of Scientific Medical Societies ([29]), except for the time criterion of symptoms. The participants who received a positive score on any screening question underwent further assessments based on the German Long-COVID-19 Guidelines, encompassing the following domains: (1) significant social or occupational impairment related to postinfection weakness (Multidimensional Fatigue Inventory [MFI]; [37]); (2) perception of illness or physical exhaustion (Post-Exertional Malaise [PEM], [17]); (3) nonrestorative sleep patterns (Pittsburgh Sleep Quality Index [PSQI]; [4]); (4) cognitive disturbances (Cognitive Failures Questionnaire [CFQ], [9]; [54]); (5) orthostatic intolerance (Orthostatic Grading Scale [OGS], [55]). We included participants only if their responses to the initial screening questions and the subsequent questionnaires revealed no indications of potential COVID-19 impairments, independent of time criteria.

We induced acquired uncontrollability experimentally by implementing high or low action-outcome predictability in a deductive reasoning task. Drawing on the Informational Helplessness Training (IHT) paradigm ([58]), this task involved the presentation of figures, each comprised of distinct properties varying in form, size, pattern, line positions, and arrow direction. Across each block, it randomly presents a set of 12 uniquely composed figures (for a detailed description of the task, see Figure 1). The participants had to identify a specific expression for each set. Following the observation of each figure, participants were presented with a dichotomous choice, requiring them to indicate whether they believed the specified expression was present. In the high action-outcome predictability condition, participants received contingent auditory feedback that potentially facilitated instrumental learning and the identification of the target expression. Correct responses led to short-tone feedback, incorrect responses to a long-tone feedback. Conversely, auditory feedback was pseudorandomized in the low-predictability condition, preventing potential instrumental learning. Each figure was displayed for 3 s in black on a white background at the center of the screen (size [0.8,0.8]). The participants had 5 seconds to respond to the dichotomous choices via mouse click. Failure to respond prompted a long tone, indicating incorrect response behavior. At the end of each block, participants had to identify the target expression from a multiple-choice answer set of all possible features, with feedback provided through a tone.

Graph: Figure 1 Induction of action-outcome unpredictability through aversive randomized feedback in a deductive reasoning task. A block consisted of 12 different figures that varied across five dimensions, each characterized by two expressions: (1) form: circle or rectangle; (2) line position: above or below the center; (3) direction of arrows: left or right; (4) pattern: striped or blank; (5) size of form: large or small. Each figure was demonstrated for 3 s; thereafter, participants had 5 s to indicate whether they believed the target expression was present within the figure. After their responses, the participants received auditory feedback. Depending on the condition, this (randomized) auditory feedback would be predictable, i. e., congruent to the response, or unpredictable. This figure indicates an exemplary answer, with potential feedback sounds for both predictability scenarios displayed below. A microphone icon accompanied by one sound wave denotes a short sound, indicating a correct response; three sound waves followed by a long sound represent an incorrect response. In the predictable, contingent feedback scenario, the target feature – in this example, an arrow pointing to the left – can be identified through instrumental learning. Conversely, in the unpredictable, randomized feedback scenario, no identification of the target feature is possible.

This auditory feedback was generated through Audacity 3.3.3 ([71]), employing a nonsinusoidal sawtooth sound wave (amplitude: 0.8 and frequency: 2500Hz). The sound volume underwent individual adjustment using the affective slider, a scale ranging from 0 to 9, measuring valence and arousal ([5]), wherein 0 indicated a strong dislike and no arousal. We adjusted the sound volumes until valence ratings fell within the range of 0–2 and arousal ratings within 7–9, while ensuring the sound intensity did not exceed 65 dB.

In the predictable, contingency feedback condition, accurate responses elicited a 0.5 s ± 0.25 s duration tone, while incorrect answers prompted a tone lasting 3 s ± 0.5 s. A thorough approach was necessary to ensure comparable sound stress exposure levels across the different conditions of predictability: The unpredictable condition employed two methods of pseudorandomized tone generation based on the order of noncontingent and contingent feedback. The participants, starting with the noncontingent feedback, experienced a 80/20 short to long tone distribution during dichotomous choices, which was reversed during the final multiple-choice response of each block. If the total tone exposure in the noncontingency condition failed to match the exposure in the contingent situations for a given block, we slightly adjusted the duration of the tones subsequently to maintain a consistent sound exposure. If the order was reversed, we once again presented responses recorded from the earlier contingent condition in a randomized order.

The reasoning task comprised 12 blocks in both sessions. We administered practice blocks to familiarize the participants with the task, closing with the accurate identification of a target expression. Subsequent to the predictability intervention, the participants engaged in the PANAS and SWAT to evaluate changes in emotional experience and workload. Additionally, we included questions regarding task predictability and controllability as as outlined by [24]).

We implemented a WM task to elicit varying levels of cognitive effort investment, utilizing adapted versions of N-Back tasks (1–4 Back) ([28]). In the traditional N-Back task, participants respond to a stream of letters, indicating whether the current letter corresponds to the letter N steps before. Here, we replaced letters with geometrical figures characterized by variations in shape (rectangle, polygon, circle, triangle), size (large, small, medium), and line pattern (horizontal, ascending, descending, or none). Each figure was displayed for 1.5 s at the center of the screen (size [0.5, 0.5]), with an interstimulus interval of 3.5 s containing a fixation cross (size [0.05, 0.05]). Responses were recorded using the k- or s-keys on the keyboard.

Figures and fixation crosses of each difficulty level appeared in distinct colors. The practice runs, organized in an ascending order of difficulty, concluded after 10 correct responses. The participants received on-screen instructions before each trial, real-time on-screen feedback displaying accuracy and reaction time (RT) of their responses, and the option to repeat practice runs for each level if desired. The order of the experimental N-Back tasks was randomized. Each block ended after the participant had correctly identified 35 geometrical figures or a total of 50 figures had been displayed. Feedback regarding the percentage of accurate responses was provided at the end of each difficulty block.

We evaluated effort investment by identifying the indifference point, employing the COG-ED Paradigm ([74]). The indifference point represents the point at which participants exhibit equal preference between two choices, perceiving the options as equally favorable. The participants engaged in a sequence of choices, selecting between more or less demanding levels of the above N-Back WM task using the left or right key. The lower effort opportunity consistently featured the 1-Back task, while the higher effort opportunity could encompass any other task difficulty. The participants had 5 seconds to respond to each offer before the subsequent choice presentation. If participants did not respond, the offer reappeared in the subsequent sequence of choices. Higher-effort opportunities consistently yielded a fixed monetary reward, either 2 € or 4 €. Simultaneously, the lower effort investment was initially set at half the reward of the higher effort option (1 € or 2 €). Based on the choice, the amount offered for the less-demanding task was titrated until the indifference point was reached (following the methodology of [73]). We presented the offers in a pseudorandomized manner across eight staircases, with each of the four levels of difficulty encompassing two staircases for the high (4 €) and low (2 €) monetary reward of the difficult task. Each staircase contained six stairs that could not be revisited until reaching the indifference point. The collected data encompassed all decisions made by the participants.

We conducted the statistical analyses and data visualization using Python 3.8 ([51]) using the following libraries: scipy ([69]), statsmodel (Seabold & Perktold, 2010), seaborn ([72]), and matplotlib ([26]). We set the significance level at p ≤.05. Effect sizes, indicated as partial eta squared (η2), were utilized and interpreted according to [13]). If sphericity was violated, we corrected the p-values according to Greenhouse-Geisser.

We divided the Indifference Point values by the base offer for the demanding task, summed, and divided them by the total number of addends, resulting in a global subjective value (SV) for each participant. We computed the difference scores for SVs by subtracting the baseline from the SV after the feedback conditions. We applied a 2×2 repeated measures analysis of variance (rmANOVA), followed by posthoc multiple comparisons using correction to control for the familywise error rate.

We analyzed the perceived predictability and controllability ratings using a multivariate analysis of variance (MANOVA). Separate analyses for the ratings of task-related stress, auditory-related stress, perceived controllability, perceived goal attainability, and perceived capabilities were conducted using rmANOVA, with multiple comparisons for follow-up analyses.

Difference scores were calculated for both positive and negative sum scores of the PANAS by subtracting the baseline value of each session. We then applied MANOVA, rmANOVA, and multiple comparisons for follow-up analyses.

We obtained reaction times and accuracy of different N-Back levels. We then converted hits, omission, and commission rates into the d'prime parameter of the signal detection theory ([23]). We conducted differential analyses in WM performance between difficulty levels and the impact of prior COVID-19 using a 2×4 factorial rmANOVA. We examined subjectively experienced workload, analyzed by the SWAT sum score, by 2×4 factorial rmANOVA, including the factors COVID-19 and task demand.

This study investigated the influence of perceived COVID-19 exposure and susceptibility to action-outcome unpredictability elicited by unsolvable reasoning tasks on the participants' willingness to invest effort in WM tasks. Table 1 documents the main effects and descriptive data.

Graph

Table 1 Statistical parameters categorized by subjectively experienced COVID-19 exposure and outcome predictability

The first hypothesis posited that poor predictability significantly reduces the willingness to exert executive effort, which was supported by a one-sided rmANOVA. The results indicated a significant mean difference of –0.12 (p =.04) in the SV – the measure reflecting a generally diminished propensity to invest effort under unpredictable conditions. The second hypothesis explored the potential influence of a prior experience of COVID-19 on the willingness to invest effort. The rmANOVA did not reveal statistically significant main effects or mean differences (–0.02, p = ns), suggesting that COVID-19 per se did not independently affect effort investment. Our third hypothesis investigated the interaction between a prior COVID-19 exposure and a present experience of unpredictability on SV. The interaction demonstrated a trend with a small effect size (F(1,40) = 1.12, p =.11, η2 = 0.03). The participants, having perceived a previous COVID-19 infection, displayed a more pronounced decrease in SV to invest executive effort compared to controls following exposure to unpredictable feedback (cf. Figure 2).

Graph: Figure 2 Changes in willingness to invest effort measured by the subjective value (SV) after experiencing predictable or unpredictable outcomes. Negative scores indicate a decrease, and positive values indicate an increase in SV compared to baseline scores. The participants who had experienced a prior COVID-19 infection were more susceptible to reacting toward predictable or unpredictable outcomes compared to controls. Each point on the plots corresponds to the baseline-corrected mean rating score for the respective condition, with error bars reflecting 2 standard deviations (SD).

To analyze the subjective experience of predictability and stress, we conducted a MANOVA, including prior COVID-19 exposure as a grouping factor. The MANOVA included the variables of Task-Related Stress, Auditory-Related Stress, Perceived Capabilities, Goal Attainability, and Task Predictability. The MANOVA revealed no overall effect of prior COVID-19 exposure. However, we did observe a significant main effect for predictability on stress ratings.

We computed subsequent separate univariate ANOVAs for the variables 1. Task-Related Stress Rating, 2. Auditory-Related Stress Rating, 3. Perceived Task Controllability, 4. Perceived Goal Attainability, and 5. Perceived Capabilities, considering Action-Outcome Predictability, COVID-19 Exposure, and their interaction as factors. Our analyses revealed a significant main effect of COVID-19 Exposure on Perceived Task Controllability, and Auditory-Related Stress Rating with medium effect sizes. Posthoc comparisons supported these results for perceived Task Controllability (t(82) = 1.86, p =.06, mean difference = 0.73) and for Auditory-Related Stress Ratings (t(82) = 2.19, p =.03, mean difference = 0.8). For the ratings of Task-Induced Stress, Perceived Capabilities, and Goal Attainability, we found only small effects of a prior COVID-19 exposure, and, contrary to expectation, prior COVID-19 exposure was associated with decreased stress in response to the auditory feedback and with slightly increased Perceived Task Controllability.

We observed a significant main effect of Predictability on Task-Related Stress, Auditory-Related Stress, Perceived Capabilities, Goal Attainability, and Task Controllability. A posthoc comparison demonstrated a significant increase in Task-Related Stress (p =.05, mean difference = 0.73) in response to unpredictable feedback. However, a posthoc analysis of Auditory-Related Stress did not reveal a significant increase (p = ns, mean difference = 0.45). As expected, the ratings of Task Predictability (p =.06, mean difference = –0.07), Goal Attainability (p <.001, mean difference = –1.73), and Perceived Capabilities (p =.03, mean difference = –0.88) decreased after unpredictable feedback.

The interaction effects between COVID-19 Exposure and Predictability were not significant for Perceived Task-Related Stress (F(1, 40) = 0.32, p =.90, η2 = 0.02), Perceived Capabilities (F(1, 40) = 0.77, p =.38, η2 = 0.01), Task Controllability (F(1, 40) = 0.25, p =.67, η2 < 0.01), or Goal Attainability (F(1, 40) = 0.17, p =.69, η2 < 0.01). However, we did observe a close to moderate interaction effect for Perceived Auditory-Related Stress (F(1, 40) = 2.12, p =.152, η2 = 0.05), suggesting that participants who experienced a prior COVID-19 exposure were more susceptible to changes in the predictability situations in terms of auditory related stress.

The MANOVA results indicated no significant effects of prior COVID-19 exposure on affective ratings. However, predictability exerted a highly significant effect on well-being without a notable interaction effect. Subsequent separate univariate ANOVAs were conducted for ratings of negative and positive well-being, incorporating COVID-19 exposure, predictability, and their interaction. Predictability significantly influenced both negative and positive ratings, showing considerable effect sizes. Negative affect increased following the experience of unpredictable feedback (mean difference = 3.16, p <.001), while positive affect ratings decreased (mean difference = –1.83, p <.001). COVID-19 exposure exhibited no significant main effects on positive or negative ratings. Interaction effects were not significant for positive (F(1,39) = 4.85, p =.19, η2 < 0.04) as well as negative affective ratings (F(1,39) = 2.36, p =.64, η2 < 0.01). However, as Figure 3 illustrates, participants with a COVID-19 history tended to exhibit a more pronounced change in positive affect between the predictability conditions: The control group was more stable in their positive affective ratings across predictability conditions. Alterations of negative affect were comparable across groups and predictability conditions.

Graph: Figure 3 Baseline-corrected changes in positive and negative affective ratings as indicated by PANAS differential score following exposure to predictable or unpredictable outcomes across participants with and without prior COVID-19 exposure. Negative scores indicate a decrease in positive or negative affective value compared to the baseline assessment, while positive scores indicate an increase. Each point on the plots corresponds to the differential mean rating scores for the respective condition, with error bars reflecting two standard errors.

To analyze reaction time and accuracy rates, we computed a rmANOVA, including the factors of COVID-19 Exposure and Task Demands. We found no significant main effect of COVID-19 Exposure on reaction times. However, task demands had a notable effect on reaction times. Multiple comparisons revealed statistically significant increases in reaction times at all task difficulty levels, suggesting that reaction times increased with escalating task difficulty (1-Back vs. 2-Back: t(41) = -8.410, p <.001, Hedges' g = –0.91; 1-Back vs. 3-Back: t(41) = –12.19, p <.001, Hedges' g = –1.43; 1-Back vs. 4-Back: t(41) = –10.44, p <.001, Hedges' g = –1.54). While there was no significant interaction effect between COVID-19 Exposure and Task Difficulty, there was a trend toward significance (F(3,120) = 2.13, p =.095, η2 = 0.05), which suggests that participants with prior COVID-19 exposure exhibited faster reaction times in less demanding N-Back tasks, but not in highly demanding tasks as compared to controls.

For accuracy rates, we detected a trend toward a main effect of COVID-19, suggesting that individuals with COVID-19 exposure showed a lower rate of errors. We confirmed this trend by posthoc comparison (t(298) = 1.71, p =.08). Increasing Task Demands significantly influenced accuracy rates, according to our expectation: Posthoc comparisons revealed a deterioration of task performance with increasing difficulty (1-Back vs. 2-Back: t(41) = 9.37, p <.001, Hedges' g = 1.66; 1-Back vs. 3-Back: t(41) = 16.87, p <.001, Hedges' g = 3.48; 1-Back vs. 4-Back: t(41) = 17.74, p <.001, Hedges' g = 18.81). We found no significant interaction effect between prior COVID-19 Exposure and Task Demands for accuracy rates (F(3,120) = 0.09, p =.95, η2 < 0.01).

Prior COVID-19 exposure yielded no significant main effect on the subjective perceived executive effort ratings of the WM tasks. Corresponding with behavioral data, participants subjectively perceived greater effort parallel to the increased N-Back task demands (1-Back vs. 2-Back: t(41) = 3.67, p <.001, Hedges' g = 0.67; 1-Back vs. 3-Back: t(41) = 9.05, p <.001, Hedges' g = 1.11; 1-Back vs. 4-Back: t(41) = 9.44, p <.001, Hedges' g = 1.67). The interaction between COVID-19 and task demands did not show differential effects on subjective effort ratings (F(3, 120) = 0.58, p =.62, η2 = 0.01).

The current study elucidates the interaction between potential pandemic-related alterations in behavioral control and the participant's disposition to be discouraged by experienced failure in an executive challenging task, shedding light on cognitive resilience mechanisms during COVID-19 recovery. Such a disposition may give rise to feelings of uncertainty, reduced experienced control, and perceived mental problems, some of which have been linked to the cognitive and motivational sequelae observed in individuals with a history of COVID-19 infections ([34]). A scenario simulating high and low action-outcome predictability – mirroring unpredictability as experienced during the pandemic – was implemented to assess the willingness to engage in cognitively demanding executive tasks. Our findings suggest that individuals with prior exposure to COVID-19 may show a propensity for alterations in cognitive effort investment. Controls without a history of COVID-19 exhibited minimal changes in their cognitive effort expenditure in response to experienced unpredictability, suggesting a different state of resilience than individuals with prior COVID-19 exposure.

In contrast, COVID-19-exposed individuals demonstrated more pronounced fluctuations in their willingness to exert cognitive effort, with an increase or decrease depending on the perceived predictability of the preceding feedback. Both groups exhibited a decline in positive affect in response to both motivational conditions. However, participants with a history of COVID-19 displayed an increased susceptibility to context, experiencing a more significant decrease in positive affect following the encounter with unpredictability. The increased perception of stress and reduced predictability ratings corresponded to the affect ratings and the willingness to invest effort between the two groups. These findings highlight the importance of considering the interaction between prior COVID-19 exposure and susceptibility to motivational controllability and the need for a tailored cognitive intervention that addresses these domains during the COVID-19 recovery phase.

Aversive uncontrollability is known to induce passivity and behaviors such as depressive symptoms, including reduced effort. The physiological mechanisms underlying this response appear to be associated with serotonergic activities ([39]). Accordingly, passivity is considered a default state that goal-directed, controlled behaviors can overcome. This finding converges with the EVC theory, which posits that a response is elicited when the outcome is perceived as valuable, and that the action is believed to achieve the desired outcome ([62]). Moreover, these neurocognitive processes interact with emotional and motivational processes at a neurobiological level ([47]). The anterior cingulate cortex (ACC) is critical in this cost-benefit analysis ([61]). The amygdala transmits affective information to the dorsal ACC, which, in turn, forwards it to the prefrontal cortex (PFC), initiating cognitive control processes.

Our findings support these theories, suggesting that perceived exposure to COVID-19 and experiences of unpredictability may influence cognitive control and goal-oriented behavior, such as decisions about effort investments. This highlights the interplay between cognitive and emotional processes in the decision about effort investment. Specifically, the evaluation of benefits and likelihoods appears to be influenced differently following experiences of unpredictability, in particular, in individuals who perceived a prior health threat such as COVID-19. An increasing number of reports suggest that even a mild COVID-19 infection may impair executive functioning and cognitive control ([21]; [41]). The precise biological mechanisms underlying these deficits are not yet fully understood; one proposed explanation suggested that stress may arise from anticipating long-term consequences and alter prefrontal functioning, affecting cognitive control and goal-directed behavior ([21]).

Dysfunctional goal-directed behavior may, therefore, affect the willingness to exert effort, particularly when unpredictability generalizes to other unrelated goal stimuli. This generalization may occur when goal stimuli are subjectively perceived as similar ([10]). Our study induced unpredictability through randomized feedback in a reasoning task, while assessing effort expenditure with a WM task. The participants are likely to have perceived the goals of these tasks as similar. If exposure to COVID-19 or the stress response to an infection diminishes goal-directed behavior, individuals with prior COVID-19 exposure might be more susceptible to generalization processes. Furthermore, goal stimuli typically follow a hierarchical structure, and failure to control a goal at a lower level may generalize to higher levels (Bodezz et al., 2022). Individuals with a COVID-19 infection might have interpreted this experience as a failure to safeguard themselves against a significant health threat, which may explain increased stress and an overgeneralization of goal stimuli. In terms of the cost-benefit analysis guiding this study, we propose that perceived unpredictability in the abstract reasoning task may have generalized to the WM task and thus reduced the perceived likelihood of goal attainability, consequently reducing the willingness to invest effort.

Positive and negative affect can bias the perception of task demands and influence effort allocation. Implicit negative affect and mood can amplify perceived task demands, consequently reducing the likelihood of effort investment ([22]). Here, the decrease in positive affect may additionally have influenced the cost-benefit analysis underlying effort expenditure. Given the interconnected nature of emotional and cognitive processes, it is plausible that affect and generalization processes might have mutually facilitated each other.

SV negatively correlated with increased WM demands, suggesting that WM allocation is costly ([73]). Therefore, individuals with reduced WM capacity need to invest greater effort than those with greater capacity. When comparing WM performance between groups, we found a slightly increased WM performance among participants with prior COVID-19 exposure, which suggests that WM performance does not explain the observed group differences in SV. Furthermore, the design of the COG-ED paradigm allowed us to rule out other confounding factors, such as differences in outcome value, likelihood, or delay discounting, as it maintains incentives across demand levels, outcome likelihood, and delay discounting at constant levels ([73]).

The adaptive reduction of effort in response to challenging and uncontrollable life events is economically plausible. However, the generalization of this decrease to other goal stimuli may give rise to dysfunctional goal-directed behaviors. Researchers have utilized this phenomenon extensively to explain motivational deficits in conditions such as depression and fatigue disorders. It also applies to various diseases where physiological changes affect the neuronal systems underlying goal-directed behavior. In particular, neurological diseases associated with dopaminergic degeneration, such as Parkinson's disease ([36]), have been recognized to explain deficits in goal-directed behavior and effort investment.

This research also stimulates novel approaches for assessment and treatment. Conditions such as fatigue and depression typically addressed by cognitive-behavioral therapy ([25]) may also benefit from interventions targeting cognitive control processes. A study comparing pre- and postpandemic neuropsychological assessments suggested that decreased executive functioning before the pandemic predicted subsequent anxiety ([44]). Individuals who relied on reactive control strategies were particularly vulnerable. Therefore, interventions enhancing executive functioning, such as proactive control strategies, might also complement treatments for fatigue symptoms. Furthermore, neuropsychological evaluation of executive functioning could help identify individuals susceptible to fatigue and motivational dysfunction. Detecting potential signs before the onset of fatigue could be critical. In our study, no participant exhibited a severe initial course of the disease or reported any subjective long-term consequences. Nevertheless, even minor changes, as observed in the present sample, may gain societal relevance if the size of the affected group is sufficiently large – as is the case in a pandemic.

This study was based on the global SV, which is commonly employed in similar research ([60]). However, this did not differentiate well between efforts in tasks of varying difficulty levels. For instance, the propensity to exert effort in tasks of medium difficulty may be particularly influenced by inflammatory processes ([32]). Therefore, future analyses should consider different types (e. g., cognitive, attentional, motor, etc.) and levels of task demands in greater detail. If cognitive control and impaired goal-directed mechanisms result in a diminished SV, alternative executive tasks that more directly tap into cognitive control processes, such as the Cued Set Shifting Task ([15]), might offer additional insights into the mechanisms of executive effort expenditure.

The current sample was relatively homogeneous, rendering generalizations to the broader population more difficult. The participants demonstrated well-developed goal-directed skills and a predisposition toward engaging in cognitive effort. The inclusion of participants with different backgrounds and ages could increase the generalizability. Moreover, future studies should critically reassess the internal validity of our COVID-19 between-subjects factor. This factor was not amenable to experimental manipulation and instead relied on subjective reporting. Therefore, one could interpret its causality conversely, that is, as a variable that depends on achievement motivation. Our control participants claim not to have suffered from COVID-19, potentially ignoring health signs or avoiding Coronavirus testing. The participants with high achievement motivation might display increased hardiness and low health concerns that one might interpret as a regulation strategy to maintain high achievement goals ([3]). Therefore, elevated achievement motivation in controls not reporting a prior infection might be responsible for a high willingness to invest effort even in the face of frustration and a willingness to ignore health concerns and Coronavirus testing in the service of successful goal attainment.

In summary, our finding provided insights into the complex interaction of perceived action-outcome predictability, resulting from stress and necessary behavioral adaptations in the context of a prior COVID-19 infection and experimentally designed unpredictability on cognitive effort readiness. When stimulated by an unpredictable task, effort expenditure was more diminished in participants with a history of COVID-19. These findings imply that certain amotivational symptoms may arise from subtle alterations in involuntary cost-benefit evaluations and subsequent dysfunctional goal-directed behaviors in challenging situations.

We thank Alina Bourquin for her assistance in data collection and for her insightful contributions.

Conflicts of Interests: The authors declare that no conflicts of interest exist.