Approximately 50–75% of chronic post-stroke survivors have mild-to-moderate cognitive dysfunction (Desmond, Moroney) (1). Executive functions are higher order of cognitive functions that consist of set-shifting, initiation, monitoring one's behavior, and self-regulating functions (2). Basic cognitive functions could be modulated and organized by executive functions to achieve goal-oriented behaviors (3). The research conducted by Laakso et al. (2) showed that ischemic stroke patients aged 55–85 years, who participated in the Helsinki (Finland) Stroke Aging Memory Study (SAM), had worse performance in executive functions assessments than healthy individuals, including the measures of set-shifting, initiation, response inhibition, and strategy formation. Executive function was considered to be the main factor that influenced the clinical outcome and the overall functional status in patients with stroke after rehabilitation (4). This subsequently contributes to limitations in performing daily activities and participating in social activities (2, 5). Therefore, the precise evaluation of executive function capacity in patients with stroke is essential.

The Montreal Cognitive Assessment (MoCA) is commonly adopted as a brief screening test of general cognitive function. Three subtests of the MoCA, which are the Trail Making B task, a phonemic fluency task, and a two-item verbal abstraction task, could be applied to assess executive functions (6). Executive dysfunctions are indicated if the patient could not complete the subtests. There are others specifically designed assessments to evaluate executive functions such as the Stroop test or the Color Trail Test (CTT) (2). A previous study showed that a large sample of stroke or brain tumor patients with lesions in the frontal region performed worse in the Stroop test than healthy individuals. This suggested an inhibition impairment in stroke patients who had unilateral frontal cortex lesions (7). However, the MoCA, Stroop test, and CTT do not consider the patients' capacity in different executive control processes, e.g., after a cue or target stimulus. Braver (2012) proposed the theory of dual mechanisms of control (DMC), which comprised of proactive and reactive controls. Proactive control is an early selection of goal-relevant information that is processed in a sustained manner of the working memory before the occurrence of the cognitively demanding events (8). It plays an important role in orienting the behavior before the event occurs in our daily life. On the contrary, reactive control refers to the immediate resolution of the current conflict or interference (8), especially when the conflict is without any preparation. The DMC model has been employed to explore the performance of executive function in a different sample population. For instance, published literature reported that elderly people and children tended to employ reactive control, while young adults were more reliant on proactive control in the response-compatibility task (9) and in the AX version of Continuous Performance Task (AX-CPT)(10). Thus, the DMC theory provides a better way to understand the flexibility of behavior regulation in complex situations in our daily life (8). However, most of the studies published to date explored executive dysfunctions without investigating the proactive and reactive controls in patients with chronic stroke.

Proactive control impairment is presented as the inability to employ the information conveyed by the cue to prepare for a response to resolve an upcoming conflict, which required working memory and sustain attention (8, 11). Execution of a functional movement requires an extent of motor anticipation (12), which was comparable to that in the proactive control of the DMC model (13). The results of a study that utilized an electroencephalogram to assess motor anticipation in patients with stroke indicated that motor anticipation impairment and hand motor function were moderately associated with motor planning (14). In Delphine et al.'s study, eight stroke patients were required to grasp and hold a brisk loading under predictive and reactive conditions. The onset time of grip force after the impact was significantly later in the paretic hand of stroke patients than in controls during both predictive and reactive conditions (15). The findings of these previous studies suggested that stroke patients have deficit in proactive control in a simple motor reaction task. This deficit may be related to the prominent impairments of motor task initiation and generation with anticipation loss (16). It remains unclear in terms of the behavioral regulation that involves proactive and reactive controls in patients with chronic stroke when executing a complex task that demands executive function.

This cross-sectional study aimed to explore the performance of proactive and reactive controls of executive functions in post-stroke survivors by the task-switching paradigm and AX-CPT. As impairment of working memory, sustained attention, and motor anticipation was previously reported in stroke patients (16–18), the hypothesis of the present study was that proactive control was more affected than reactive control when performing a complex functional task in patients with stroke. The findings of the present study extend the knowledge pool on executive dysfunction in patients with stroke and would be of interest to a wide audience.

The present study investigated the performance of proactive and reactive controls of executive functions in post-stroke survivors by the dual cognitive control model. The main findings indicated that post-stroke participants require longer response times in the BX condition of the AX-CPT, but did not exhibit a reduction in the proactive and reactive control of switching compared with controls during the task-switching paradigm. These novel findings supported the hypothesis that proactive control was more affected than reactive control when performing a complex functional task.

In the BX condition, participants needed to maintain the “B” cue in the working memory, so that adequate preparation, which could enhance the reaction time for the probe (20). This study found that the post-stroke participants required a longer response time in the BX condition of the AX-CPT, suggesting that proactive control was impaired in post-stroke survivors. This finding was supported by two other previous studies that explored the proactive and reactive controls without involving executive functions in stroke patients (15, 24). One study showed impairment in the feed-forward adjustments of the multi-finger synergies prior to quick action in patients after a single cortical stroke with mild motor impairments (24). Dispa et al. found that the onset time of grip force after the impact was significantly later in the paretic hand of stroke patients than in controls in both predictive and reactive conditions (15). These findings suggested that stroke patients had deficits in proactive control in a motor reaction task. In the present study, the worse performance in proactive control of AX-CPT observed in the stroke group maybe resulted from the impairment of working memory, sustained attention, and anticipation loss. The execution of a complex task places a high demand on cognitive resources, including working memory, anticipation, and sustained attention during the proactive control period. The potential reasons for the worse performance in the proactive control in the stroke group are as follows. First, proactive control in AX-CPT was reported to be reliant upon the anticipation and prevention of interference before it occurred in the working memory (8, 20). Several studies have indicated working memory deficits in patients with stroke (18, 25). In the study by van Geldorp et al. (18), the participants were required to remember which cue was presented after a short temporal delay in a computerized delayed-match-to-sample task which included spatial, object, or binding (spatial + object) cues. Compared to healthy controls, the stroke participants showed working memory deficits in the conditions of spatial and object cues. Second, the impaired proactive control might be related to the anticipation loss in stroke patients. A study conducted by Roussel and Martinaud (16) indicated that in post-stroke participants, the dysexecutive profile was characterized by prominent impairments of motor initiation and generation with anticipation loss in the behavioral domain. In addition, several studies reported that proactive control is a sustained top-down attentional control (11, 13, 26). Manard et al. found that impaired proactive control of elderly people was related to a reduction in sustained neural activity at the frontal cortex (26). The sustained neural activity was related to sustained attention (11, 13). Patients with stroke were reported to have sustained attention deficits (17, 27), which could reduce the ability to expect and prepare for an upcoming stimulus (28). Therefore, deficits in sustained attention are likely to be a contributing factor to the impaired proactive control of stroke patients. However, previous studies indicated that motivation, fatigue, and mind wandering might have a negative effect on sustained attention for elderly people (29). Proactive control generally requires more effort, subsequently leads to more fatigue, which is most likely associated with the tendency to favor reactive over proactive control. Even though the participants in this study rested between any two tasks, it could not be ascertained that whether motivation, fatigue, and mind wondering might affect the behavior performance of proactive control. A future study that investigates the effect of these factors on the behavioral performance of proactive control would be beneficial to substantiate the findings of the present study.

According to the results of the task-switching paradigm, the stroke group had lower accuracy rates and efficiency scores in both repeat and switch trials than the control group. The two cohorts, however, did not show the between-trial difference in the 100 and 50% validities, which may suggest participants in the stroke group did not perform worse than healthy participants in the proactive and reactive controls of set-shifting function. These findings were consistent with the previous study, which demonstrated patients with stroke had different switch costs of accuracy rates compared with the control group in alternating switch tasks, but not in the cued task-switching (30). The findings in Pohl et al.'s study suggested that task switching in alternating switch tasks with a high demand of endogenous proactive control was impaired in adults after stroke, whereas task switching in the cued task-switching requiring exogenous proactive control was not impaired. The first reason was likely due to the higher difficulty level under endogenous proactive control than under exogenous proactive control. The second reason may be that the key neural structures that are responsible for the proactive and reactive control in task switching are different. Proactive control involves the frontal and/or parietal cortices (13, 20), whereas a wider network of brain regions are responsible for the reactive control in task switching, which involved the dorsolateral–ventrolateral plane of the prefrontal cortex and anterior cingulate cortex (ACC) (31, 32). However, both the present study and Pohl et al.'s study did not consider the different lesion locations between the two cohorts. This might potentially contribute to the lack of observed difference between the proactive and reactive controls of switching function in the stroke group.

The efficiency scores in the task-switching paradigm showed no significant between-trial difference in the stroke group, whereas the efficiency scores in the repeat trial of the control group were significantly higher than those in the switch trial regardless of 100 or 50% validity. These findings suggested that post-stroke survivors were less susceptible to proactive interference, which may be due to their lower working memory function. Allport and colleagues assumed that switch costs arose from proactive interference (33). Higher susceptibility of proactive interference in task switching was associated with higher working memory capacity (34, 35). Several studies had indicated working memory deficits in post-stroke survivors (18, 36), which was likely to be a factor that contributed to lower susceptibility to proactive interference in post-stroke survivors than healthy individuals. Impaired proactive control in the AX-CPT of stroke patients also supported deficits in the working memory since proactive control in the AX-CPT was related to working memory (20).

A larger interference effect in CTT and higher word interference of completion time in Stroop test were observed in stroke participants compared with healthy controls. The interference effect in CTT assessed the shifting function (24), while the word interference of completion time in the Stroop test taps on the conflict inhibition during the naming color task (37). The results of the CTT and Stroop test suggested that the general executive function was impaired in stroke participants. These findings were consistent with a previous study (2), which showed that stroke patients performed poorly in the Trail Making test and Stroop test. The results of the correlation and regression analysis between the Stroop test and the performance in BX condition of AX-CPT of the two groups further supported the notion that the proactive control for working memory function was impaired in post-stroke survivors. The process of interference control in the Stroop test included the top-down control process (38), which was also required during proactive control in AX-CPT (26). These two previous studies reported that the prefrontal cortex was the key node activated in the top-down control process in the Stroop test and proactive control. Thus, the response time in the BX condition related to proactive control was a significant factor related to the performance in the Stroop test.

There are several limitations in the present study that limit the interpretation of the data. First, the sample size was not power calculated and was likely to contain type II error. This might contribute to the lack of between-group differences in the task-switching paradigm. Second, the effects of lesion locations or lesion size on the proactive and reactive controls were not considered in the present study. Future study should consider recruiting sample group that has homogeneous lesion site and size. Third, the data related to the lesion such as side, temporal proximity, and the extent of the lesion were not taken into consideration in the present study. Future studies should consider the lesion data to further assess the impact on behavioral data. Forth, the sample population of the present study was heterogeneous which limits the interpretation of the findings. Fifth, two-way/three-way repeated-measures ANOVA applied to non-parametric data might lead to Type-I error. Sixth, the present study only employed a simple algorithm to explore the between-group difference in the speed-accuracy trade-off. Future studies should employ the diffusion drift models or other methodologies for in-depth exploring the between-group difference in the behavioral responses. In addition to behavioral outcomes, future research could adopt electrophysiology and brain-imaging techniques to investigate the underlying neural mechanisms (39). Last but not least, other research results showed a significant gender difference in proactive and reactive controls (40). Thus, there might be gender differences in proactive and reactive control performance of stroke patients. Further studies are required to clarify the gender effect on the performance of proactive and reactive controls in post-stroke survivors.

Proactive control is the preparatory process before the target stimulus, whereas reactive control is an imperative process of interference resolution after the target stimulus. Post-stroke survivors showed a deficit in the performance of proactive control that is related to working memory. However, post-stroke patients seemed not to show a significant deficit in the reactive control of executive functions. Interventions that focus on proactive control may result in improved clinical outcomes.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Institutional Ethical Committee for Clinical Research and Animal Trials of the First Affiliated Hospital of Sun Yat-sen University (Approval No. [2020]073). The patients/participants provided their written informed consent to participate in this study.

All authors meet the four primary ICMJE criteria for authorship. In addition, all authors have been actively involved in the study in different capacities: QY and WL: designed the study and conducted all stages of the study including data collection, analysis, interpretation, and drafting of the manuscript. XH, ZH, BZ, and ZLi: participated in the design of the research protocol, data analysis, and drafting of the manuscript. TZ, ZH, MD, and ZLia: participated in data collection, analysis, and interpretation of the data. WL and ZLia: revised the manuscript, interpreted the data, and managed the trial. All authors have read and approved the final manuscript.

This research project was supported by the National Natural Science Foundation of China (Grant Numbers: 81971224 and 82002375); Science and Technology Program of Guangzhou (Grant Number: 201803010083).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.