The emergence of SARS-CoV-2 suddenly and profoundly altered the daily lives, workplace operations, and internal experiences of many people. The novel coronavirus disease 2019 (COVID-19) caused more than 500 million infections and 6.2 million deaths worldwide. Vaccination effort has ramped up, but the virus and its many variants continue to pose a threat to a large swath of the population. As the race between widespread vaccination and new variants continues, researchers are just beginning to understand the effects of the pandemic’s ever-changing landscape on human behavior. With increasing concerns and shifting priorities, many feel preoccupied with thoughts about the physical and financial threats the virus and the resulting economic upheaval may pose. A poll conducted in July 2020 found that 66% of Americans were worried about themselves or a family member being infected (Washington Post–ABC News, July 17, 2020). The elevated concerns may draw attention away from important stimuli in daily tasks, such as driving, schooling, and workplace responsibilities. Might preoccupation with the physical and financial threats of COVID-19 interfere with performance on tasks requiring attention?
To assess the potential impact of COVID-related concerns on cognitive performance, in this study we measured the correlation between levels of concern about COVID-19 and performance on several different cognitive tasks. Our goals are to (i) determine whether there is a connection between these heightened concerns and performance on tasks relying on attention, and (ii) whether some components of attention are more susceptible to interference from current concerns than others. In turn, the findings will shed light more broadly on the relationship between emotion and cognition, two research fields that have seen increasing connections in recent years (Pessoa, 2008).
Some evidence suggests that internal states, such as ongoing worries and fear, interfere with attention. Studies on mind wandering showed that people frequently engage in task-unrelated thinking in everyday activities (Killingsworth & Gilbert, 2010). These task-unrelated thoughts (TUTs) are often grounded in participants’ current concerns (McVay & Kane, 2010; Poerio et al., 2013; Smallwood & Schooler, 2015). These findings suggest that in the face of heightened concerns, the intrusion of task-irrelevant thoughts can draw attention away from ongoing tasks. In other words, severe concerns about COVID-19 may lead people to become more distractable. As yet, it is unclear whether COVID-related concerns manifest as intrusive task-unrelated thoughts or associated deficits in attention, nor whether, as with mind wandering, the relationship between attention and concerns would be affected by factors such as task difficulty or an individual’s working memory capacity (McVay & Kane, 2012; Seli et al., 2018; Smallwood & Schooler, 2015). The present study addresses whether people with greater COVID-related concerns are more prone to mind wandering than those with low concern levels, and if so, whether these concerns induce general task impairment, or whether the impairment is specific to tasks that implicate a particular attentional function.
Other studies have found that emotions, such as fear, modulate attention. Compared with neutral stimuli, fearful stimuli tend to capture attention more and are less susceptible to the attentional blink (Anderson & Phelps, 2001; Compton, 2003; Ohman et al., 2001). Fear may also narrow the focus of attention to fear-inducing stimuli, as in the “weapon-focus” effect (Steblay, 1992). However, COVID-related concerns may give rise to general anxiety, rather than specific fears. The relationship between anxiety and attention is complex (Robinson, Krimsky, et al., 2013; Robinson, Vytal, et al., 2013). For example, although patients with an anxiety disorder may have deficits in brain regions regulating cognitive control, they do not always perform more poorly on cognitive control tasks (Eysenck et al., 2007; Lagarde et al., 2010). Experimentally induced anxiety, such as the threat of receiving an electric shock in an experiment, does not consistently impair cognitive control (Choi et al., 2012; Robinson Krimsky, & Grillon, 2013; Robinson, Vytal, et al., 2013). Thus, while studies on acute emotions suggest that internal states may affect attention, studies on prolonged emotional states like anxiety suggest that individuals may suppress the cognitive effects of sustained environmental stressors during performance of cognitive tasks.
However, studies on the effects of poverty and resource scarcity argue that severe concerns do influence cognitive performance. Mani et al. (2013) showed that poverty is associated with impairments in cognitive control. In one experiment, they asked shoppers at a mall in the USA to think about a car repair. Some were told the repairs would cost $1,500, whereas others were told it would cost $150. They then completed the Raven’s Progressive Matrices and a spatial compatibility task. Performance was comparable between high- and low- income participants when contemplating the more affordable car repair, but performance was impaired when the low-income (but not high-income) participants contemplated the expensive car repair. In another experiment, seasonal sugarcane farmers from India were tested in a numerical Stroop task either before harvest, when the farmers had limited financial resources, or after harvest, when the farmers had more financial resources. Performance was worse before than after harvest. These findings suggest that severe financial concerns compete for attention with ongoing tasks.
The Mani et al. (2013) study has received some criticism regarding its methods, controls, and conclusions (Dang et al., 2015). It nonetheless sparked an interest in the effects of resource scarcity on cognitive performance. Other studies have found behavioral changes in people who are scarce in other resources, such as time (Cannon et al., 2019; Roux et al., 2015). These findings, in the context of the broader literature linking attention and emotion, raise the possibility that as access to financial or health resources declines during the COVID-19 pandemic, cognitive control may also suffer, whether due to depleted resources or increased anxiety surrounding those resources.
Two shortcomings of these prior studies may account for the mixed findings regarding the impairment of or robustness of cognitive performance in the face of anxiety or concerns. First, attention is a multifaceted construct that relies on multiple brain systems (Pashler, 1997). For example, whereas cognitive control primarily engages the prefrontal cortex (Badre, 2008; Braver, 2012), sustained attention relies on a broad neural network including the frontal, parietal, temporal, and cerebellar cortices (Esterman et al., 2013; Rosenberg et al., 2016). Yet previous studies of anxiety and attention have largely used a single index of attention. It is unclear, then, how widespread the effects are or whether there is a systematic relationship between the effects of anxiety and components of attention at various levels of cognitive processing. Secondly, measuring only “latent” concerns, without taking into account “active” concerns in the form of task-unrelated thought, results in an incomplete picture of the economics of scarcity and the effects of concerns on attention. The contents of TUTs frequently reflect an individual’s current goals (Klinger, 1971, 2013), which vary over time and across tasks (Rummel et al., 2021; Zanesco, 2020). TUTs are also sensitive to future-related concerns (Stawarczyk et al., 2013). However, people may be able to exert cognitive control over long-term concerns, rendering them “latent” rather than active. Do scarcity and the anxiety it causes consistently influence attention, or do their effects depend on the concerns being activated and emerging into conscious thought?
A comprehensive understanding of the effects of pandemic-related concerns on attention therefore requires the inclusion of multiple cognitive tasks that tap into distinct components of attention, as well as measures of both latent and active concerns. Regarding the selection of cognitive tasks, several studies have sought to identify distinctive attentional constructs that can be measured by psychophysics (Fan et al., 2002; Skogsberg et al., 2015; Treviño et al., 2021), converging on a few findings. Fan et al. (2002) measured individual differences using the Attention Network Test (ANT), a timed reaction time (RT) task that combined spatial cues with flanker interference. It provided a measure of an executive control component in the difference between congruent and incongruent flanker trials, as well as measures of orienting and alerting components in the difference between spatial cue conditions. Using hierarchical-cluster analysis across 11 different attention tasks, Skogsberg et al. (2015) identified spatiotemporal attention, global attention, transient attention, and sustained attention as the four components of visual attention. Each of the 11 tasks held a place within a 2-dimensional space that differed along the global to spatiotemporal dimension on one axis and the sustained to transient dimension on the other. More recently, Treviño et al. (2021) used six experimental tasks, including the scene CPT as a measure of sustained attention and five selective attention tasks (multiple object tracking, spatial configuration visual search, visual working memory, approximate number sense, and flanker interference). They identified five components of attention based on exploratory factor analysis: capacity, search, digit span, arithmetic, and sustained.
Although these studies differed in terms of the tasks used and components identified, there were some consistencies. First, sustained attention featured in each set of components, represented as alerting in Fan et al. (2002), and sustained attention in Skogsberg et al. (2015) and Treviño et al. (2021). Second, spatial orienting consistently appeared as a component, termed orienting in Fan et al. (2002), spatiotemporal attention in Skogsberg et al. (2015), and search in Treviño et al. (2021). Third, two of the three included a component related to executive function or cognitive control. Fan et al. (2002) identified an executive control component, and although Treviño et al. (2021) did not identify cognitive control as a specific component, their arithmetic component is related to cognitive control. Most importantly, all three studies divided the general cognitive function of attention into multiple separable components.
Of the different components identified by these studies, some are more externally driven (e.g., search in Treviño et al., 2021 or spatiotemporal attention in Skogsberg et al., 2015), while others are more internally oriented (e.g., executive control in Fan et al., 2002 or arithmetic in Treviño et al, 2021). The different components of attention may vary along a scale from external, or perceptual, to internal, or central, attention (Chun et al., 2011). Tasks whose performance relies on external, visual stimuli are more perceptual in nature, whereas tasks whose performance relies on internal regulation of rules and responses are more closely tied to central attentional functions.
Regarding the evaluation of both latent and active concerns, a recent sustained attention study (Jun et al., 2021) provided evidence of the importance of this distinction. Jun et al. (2021) was among the first to directly evaluate the effects of concerns about COVID-19 on attention. The study was conducted in the first year of the pandemic before vaccines became widely available. Young adults from Europe and the USA first rated their levels of health-related and financial concerns surrounding the COVID-19 pandemic. They then completed a demanding continuous performance task (CPT), withholding response to infrequent mountain images presented among a stream of city images. Participants also self-reported the proportion of the time that their mind had wandered from the task. Despite expressing a wide range of COVID-related concerns in the pre-task questionnaire, participants with severe concerns did just as well on the scene CPT as participants with low concerns. However, those experiencing a higher rate of mind wandering during the scene CPT did more poorly. Jun et al. (2021) suggest that young adults are largely successful in preventing their pre-task COVID concerns from intruding into the scene CPT. TUTs that did occur during the task, however, became a source of distraction that interfered with performance. This finding underscores the need to consider two types of concerns: concerns that people can regulate and minimize during a task (“latent concerns”) and concerns that intrude into a task (“active concerns”). Whereas latent concerns can be measured using a pre-task questionnaire, active concerns manifest as active TUTs during the task. Because Jun et al. (2021) tested a single attentional construct—sustained attention—it is unclear whether effects of current concerns on attention are broad, or whether they are limited to certain attentional functions.
In this preregistered study, we go beyond the single-task assessment of Jun et al. (2021) in pursuit of two main goals: (1) to test the effects of reported COVID-related concerns on multiple components of attention, and (2) to investigate whether task-unrelated thoughts primarily influence externally or internally driven components of attention. We collected data from young adults during the COVID-19 pandemic from October 2020 to February 2021, before vaccines for COVID-19 became widely accessible (https://osf.io/5y9bt/?view_only=615432100a084e2faf535942d1015073). We tested 234 participants in four attention tasks that probe different components of attention: visual search, visual working memory, sustained attention (using a scene CPT), and task switching. Immediately before the attention tasks, participants completed a questionnaire assessing their level of health concerns and financial concerns related to COVID-19 (Jun et al., 2021). They also completed a questionnaire using the State-Trait Anxiety Inventory (STAI-6; Spielberger, 1983; Tluczek et al., 2009) to measure anxiety. These questionnaires are taken as a measure of latent concerns. After each attention task, participants reported the proportion of time spent on TUT during the task. The TUT reports are taken as a proxy for active concerns. We test the prediction that elevated active concerns, reflected in the TUT measure, are associated with impaired attention, especially for central attention components, rather than perceptual attention components. Here, we provide an overview of the four attention tasks (Fig. 1).
Visual search
The visual search task varied in two factors: nature of search (feature or conjunction search) and the number of items (i.e., set size of 8 or 16). This task loaded onto the search component of attention in Treviño et al. (2021) and is closely related to the orienting and spatiotemporal components identified in Fan et al. (2002) and Skogsberg et al. (2015), respectively. Participants searched for a letter T among distractor letters and reported whether the T was black or white. In the feature search condition, the distractors were Os, yielding a large, featural difference between the target and the distractors. In the conjunction search condition, also known as the spatial configuration search condition, the distractors were Ls, yielding a small, spatial configurational difference between the target and distractors (Wolfe, 1998). The slope of the linear function relating RT to set size indexes search efficiency. Feature search produces shallow search slopes, owing to the availability of preattentive feature information guiding search quickly to the target (Treisman, 1996; Wolfe, 2021). But conjunction or spatial configuration search cannot rely on the same preattentive information and therefore yields steep search slopes that provide an index of search efficiency.
Visual working memory
Many people categorize visual working memory as an example of attention directed internally (Chun et al., 2011), requiring some cognitive control. To assess visual working memory, we adapted a standard color visual working memory task (Xie et al., 2020) using the change detection procedure. This task loaded onto the capacity component in Treviño et al. (2021). Participants encoded to memory a display of 5 color patches that were presented for 0.5 s. After a blank delay interval of 1.5 s, a test display of 5 colors was presented in the same locations, including four colors that were the same as before and one that had changed. Participants were asked to click on the color that had changed. Because the task requires a comparison between the test array and an internally maintained memory representation, our preregistered predictions were based on the assumption that the visual working memory task relies primarily on central attention. However, the change detection task does not require manipulation of information in visual working memory (Pailian & Alvarez, 2020); as such, it may be more robust to current concerns than other versions that require information manipulation.
Scene CPT
As a continuous performance task, the scene CPT presented participants with a continuous stream of city and mountain images (800 ms/scene) for two minutes (Esterman et al., 2013; Jun et al., 2021). It taps into the sustained attention component identified in previous studies. In our study, participants pressed the spacebar to cities that occurred on 90% of the trials and withheld response to mountains that occurred on 10% of the trials. The mountains and cities were not visually salient, nor did the two differ from one another in salient ways, reducing bottom-up signals that could be used for detection. Although CPT tasks are commonly used to index sustained attention, performance relies on multiple mechanisms, including perceptual processing, rhythmic responses to the onset of the stimuli (Hawkins et al., 2019), and inhibitory control of responses (Jun & Lee, 2021).
Task-switching
As a measure of cognitive control, we adapted a version of the task-switching paradigm used by Mani et al. (2013). This task involves both congruent and incongruent trials, similar to the flanker aspect of Fan et al. (2002) that was used to assess executive function. The task also requires memory of and application of task rules, which likely taps into similar mechanisms to those categorized in Treviño et al. (2021)’s arithmetic component. In this task, a geometric shape appeared in one of four locations around the center (top, bottom, left, or right). The participants’ task was to report its location using arrow keys on their keyboard. The response rule was either compatible–pressing the arrow corresponding to the location of the shape, or incompatible–pressing the arrow opposite to the location of the shape, which incorporated a Stroop-like congruency effect. In task-stay blocks, the same rule was used for a block of trials. In task-switch blocks, the two rules were randomly intermixed within a block of trials: outline shapes required the incompatible mapping, but solid shapes required the compatible mapping. Following Mani et al. (2013), we used accuracy in the task-switch blocks as an index of performance. Because the complexity of this task-switching paradigm largely originates from a change in response rules, this task relies primarily on central attention.
Together, visual search, visual working memory, scene CPT, and task switching capture a variety of attentional mechanisms (Treviño et al., 2021), allowing us to assess the potentially different effects of current concerns on different aspects of attention.