URGENT 28hr DEADLINE Psychology Lab Report

Contents lists available at ScienceDirect

Consciousness and Cognition

journal homepage: www.elsevier.com/locate/concog

Full Length Article

Multiple routes to mind wandering: Predicting mind wandering with resource theories☆

Jason G. Randalla,⁎, Margaret E. Beierb, Anton J. Villadoc

aUniversity at Albany, State University of New York, Department of Psychology, Social Science 387, 1400 Washington Ave., Albany, NY 12222, United States b Rice University, Department of Psychology, 6100 Main St., MS-25, Houston, TX 77005, United States c PracticeCraft, LLC, 1210 Green Knoll Dr., Sugar Land, TX 77579, United States

A R T I C L E I N F O

Keywords: Mind wandering Resource theory Working memory Attention Task complexity

A B S T R A C T

Three experiments examine individual (attentional capacity) and task-related characteristics leading to mind wandering, and the effect of mind wandering on task performance. Drawing on resource theories, we tested interactive nonlinear effects of these predictors, manipulating task demand using math tests of varying difficulty (Exp 1: N=143, three levels between-subjects; Exp 2: N=59, three levels within-subjects; Exp 3: N=133, four levels within-subjects). Results confirmed that mind wandering was most frequent during extreme task demand levels, although the effect varied somewhat between experiments. Additionally, results from Experiment 3 and an integrated analysis demonstrated that people with relatively higher attentional capacity were less likely to mind wander as task demand increased. Moreover, mind wandering was more detri- mental to performance as task demand increased across all experiments. Our findings build on past research by demonstrating the importance of accounting for interactions and nonlinear ef- fects of task demand and attentional capacity in mind wandering research.

1. Introduction

Peoples’ minds wander for different, and seemingly opposite reasons. You might find your mind wandering while listening to a podcast about a complex, novel issue such as glacial geomorphology; or while listening to an explanation of a simple, familiar topic, such as tooth brushing. Both situations could incite mind wandering, despite varying markedly in attentional demand. This apparent inconsistency would appear to limit the degree to which theory can predict the occurrence of the mind wandering phenomenon—a limitation that has convoluted research on the predictors of mind wandering, and subsequently, its outcomes. In other words, if mind wandering is frequent in both high-demand and low-demand tasks, then predicting when and for what reason mind wandering is likely to occur for a given task becomes a complicated issue. In order to reconcile this apparent inconsistency, the three experiments reported here draw upon resource theories of information processing to examine nonlinear, interactive effects of task demand and individual abilities on mind wandering and performance. These studies advance theory by integrating mind wandering with resource theories, and provide experimental evidence for key task and individual factors that predict the occurrence and consequences of mind wandering.

https://doi.org/10.1016/j.concog.2018.11.006 Received 8 August 2018; Received in revised form 13 November 2018; Accepted 16 November 2018

☆ An earlier version of this article was presented at the Society for Industrial and Organizational Psychology Annual Conference in Anaheim, California, April 2016.

⁎ Corresponding author. E-mail addresses: [email protected] (J.G. Randall), [email protected] (M.E. Beier), [email protected] (A.J. Villado).

Consciousness and Cognition 67 (2019) 26–43

Available online 28 November 2018 1053-8100/ © 2018 Elsevier Inc. All rights reserved.

T

1.1. Mind wandering

Although mind wandering—defined as a shift in attention away from task-related thoughts toward the processing of other, task- unrelated thoughts (Smallwood & Schooler, 2006)—is a frequent universal experience (Killingsworth & Gilbert, 2010), relatively little is known about the conditions that lead to mind wandering. Also referred to as task-unrelated thought, or off-task thought, mind wandering is usually operationalized in experimental research as a frequency of thoughts directed away from a primary task at hand (e.g., thinking about an upcoming vacation while typing an email).

Different theories explaining the occurrence, or onset, of mind wandering (Smallwood, 2013) suggest the phenomenon occurs: (a) because personal concerns outweigh the importance of the current task (Klinger, Gregoire, & Barta, 1973), (b) as a failure to maintain executive control in the face of internal distractions (McVay & Kane, 2010), or (c) from improper self-monitoring (i.e., a lack of meta- awareness; Schooler et al., 2011). Multiple theoretical perspectives for why the mind wanders underscore the viewpoint that there may be multiple predictors of the mind-wandering experience. In this study we separate predictor characteristics that originate from the person and those that originate from the task to examine the influence of individual and task characteristics on mind wandering. Individual predictors of mind wandering studied to date focus primarily on attentional capacity (Kane & McVay, 2012), and task features include factors such as complexity or attentional demand (Giambra, 1995; Kane et al., 2007; Randall, Oswald, & Beier, 2014). Without an interactive approach to incorporate individual and task characteristics, however, empirical results appear mud- dled. For example, some studies demonstrate increased mind wandering in low-demand relative to high-demand tasks (Forster & Lavie, 2009; Giambra, 1995) and others show the opposite effect (Kane et al., 2007, 2017). Similarly, attentional capacity, usually operationalized as working memory capacity (WMC), is sometimes positively related to mind wandering (Levinson, Smallwood, & Davidson, 2012; Rummel & Boywitt, 2014), at times negatively related (Kane et al., 2016; McVay & Kane, 2012a, 2012b; Robison & Unsworth, 2015, 2017, 2018; Unsworth & McMillan, 2013, 2014; Unsworth & Robison, 2017), and still other times unrelated (Robison & Unsworth, 2015; Smeekens & Kane, 2016). Such mixed findings suggest that mind wandering may depend on contextual factors—that simple main effects may over-simplify richer relationships.

A more nuanced perspective, and one that has only recently been explicitly incorporated into mind wandering theory (Marcusson- Clavertz, Cardeña, & Terhune, 2016; Randall et al., 2014; Robison & Unsworth, 2017; Rummel & Boywitt, 2014; Smallwood & Andrews-Hanna, 2013; Thomson, Besner, & Smilek, 2015; Xu & Metcalfe, 2016), is that the predictors and performance consequences of mind wandering may depend on a number of conditional factors, including individual characteristics and task demands. Ac- counting for the moderating influences of task demand and attentional capacity may allow researchers to reconcile existing incon- sistent empirical results regarding the prediction of mind wandering and its consequences. The lack of a theoretical framework that incorporates both task and individual features to reconcile the disparate results in mind wandering research may continue to impede progress in its scientific study. Indeed, as Robison and Unsworth (2017) note, “future research is necessary to enhance our under- standing of the dynamic nature of WMC’s relationship with mind-wandering in a context-dependent manner” (p. 53). The current paper extends theoretical perspectives and integrative empirical evidence on the task and individual contingencies of the mind wandering phenomenon (Kane & McVay, 2012; Rummel & Boywitt, 2014; Smallwood & Andrews-Hanna, 2013; Xu & Metcalfe, 2016) by integrating the theoretical frameworks of mind wandering and resource theories (Kanfer & Ackerman, 1989; Norman & Bobrow, 1975) in order to account for such interactive (i.e., moderated), nonlinear (i.e., dynamic) relationships.

1.2. Resource theories

Resource theories combine consideration of person variables (attentional capacity, motivation) with task characteristics to un- derstand the attention necessary for skilled performance (Beier & Oswald, 2012; Kanfer & Ackerman, 1989; Norman & Bobrow, 1975). The resource allocation framework identifies the individual differences and motivational processes at play in skilled per- formance. According to this framework, individuals engage self-regulatory strategies to allocate attentional resources to different thoughts and activities, and these strategies are influenced by task complexity (Kanfer & Ackerman, 1989). Tasks at the extreme levels of demand—both high and low—are resource insensitive, meaning that exerting additional effort or focus to the task will have little to no effect on performance. For example, simple or well-learned tasks (e.g., tooth brushing) require little attentional capacity; as such, exerting additional effort should have no demonstrable effect on performance. Likewise, during extremely challenging tasks that exceed one’s capabilities (e.g., solving a glacial geomorphology problem), exerting extra effort or focus may fail to translate to performance benefits when the actor knows little about the topic. Only for resource sensitive tasks, those for which task demands are within the person’s range of capabilities, will variation in effort and attention produce performance differences.

1.3. Integrating mind wandering and resource theories

Resource theories, therefore, suggest two key determinants of mind wandering and its effect on task performance: task demand and individual attentional capacity. These variables are interrelated because the extent to which a task is demanding will depend on the performer’s attentional capacity (Ackerman, 1988). People with limited attentional resources, for instance, may be overwhelmed with moderate- to high-demand tasks (that are resource insensitive for them), and subsequently will be more likely to engage in mind wandering while performing such tasks. Yet, for these same tasks, people with relatively more attentional resources might be able to perform well with a reasonable amount of effort (i.e., the tasks will be resource sensitive for them), making mind wandering less likely. Low-demand tasks that are so easy they do not necessitate attention to perform well would also be resource insensitive for most people, and would thus promote more mind wandering, particularly when attentional capacity is higher (Beier & Oswald, 2012;

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

27

Kanfer & Ackerman, 1989; Norman & Bobrow, 1975; Rummel & Boywitt, 2014). As a notable example, in a recent study, Beck and Schmidt (2018) found that a task’s resource sensitivity affected the resource allocation process, with individuals investing less effort on resource insensitive tasks than on resource sensitive tasks. However, this study differed from the current study’s approach in key ways, as the researchers manipulated task resource sensitivity using rewards and time scarcity, and measured resource regulation as time on task rather than mind wandering (Beck & Schmidt, 2018). In summary, resource theories suggest that mind wandering – its antecedents and outcomes – depends on how resource (in)sensitive the focal task is. Resource sensitivity, in turn, depends on the attentional demands of the task and the attentional capacity of the individual (Beier & Oswald, 2012).

Fig. 1 presents our theoretical expectations regarding the nonlinear trend of mind wandering occurrence with task demand on the x-axis and mind wandering frequency on the y-axis. As highlighted previously, high- and low-demand tasks should be more resource insensitive, meaning that attention to the task should yield less of an effect on task performance, so individuals will be more likely to engage in mind wandering during task extremes (i.e., high and low levels of task demand; Beier & Oswald, 2012; Kanfer & Ackerman, 1989). Conversely, mind wandering would be least likely during resource sensitive tasks of moderate demand level, where increases in attention and effort may yield more demonstrable increases in task performance and where decreases in attention would lead to performance decrements. The expected u-shaped (concave) quadratic pattern is displayed in Panel A in Fig. 1. However, resource theories also incorporate a moderating effect of individual capability, to acknowledge that tasks that may be demanding or difficult for one person may not be so for another. Thus, Panel B in Fig. 1 displays three different concave quadratic trends that represent the varying levels of resource sensitivity for three people who vary in attentional capacity. The solid black line mirrors the theoretically predicted trend from Panel A, representing the mind wandering trend for an individual with a moderate level of attentional capacity. The dashed line in Fig. 1, Panel B represents an individual with relatively low attentional capacity. This individual’s level of resource sensitivity is shifted further to the left, meaning that he or she is relatively less likely to mind wander during a task of low-demand and more likely to disengage during high-demand tasks. By contrast, the dotted line in Fig. 1, Panel B represents an individual with high attentional capacity. This individual is more capable of staying engaged in more difficult tasks, as evidenced by the resource sensitive region shifting further to the right, resulting in less mind wandering during difficult tasks and more mind wandering during easier tasks. These figures present the theoretical propositions under investigation in this paper: that is, whether evaluating nonlinear relationships (e.g., the predicted concave quadratic effects) and interactive effects (e.g., varying trends moderated by attentional capacity) will improve prediction of mind wandering occurrence and outcomes, and subsequently clarify mixed results in the extant research.

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Moderate

M in

d W

an de

rin g

Fr eq

ue nc

y

Task Demand

Panel A

0

1

2

3

4

5

6

7

8

Very Low Low Moderate High Very High

M in

d W

an de

rin g

Fr eq

ue nc

y

Task Demand

Panel B Moderate CapacityLow Capacity High Capacity

Fig. 1. Predicted concave mind wandering trends as a function of task demand alone in Panel A, and as a function of the interactive effects of task demand and attentional capacity in Panel B. The three trends in Panel B represent mind wandering frequency predictions for individuals with low, moderate, and high levels of attentional capacity.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

28

Recent work in the mind wandering literature finds some support for resource theory predictions. First, a recent meta-analysis found that people with lower attentional capacity tended to mind wander more than those with higher capacity, especially during high-demand tasks (Randall et al., 2014). One of the primary studies included in the meta-analysis, McVay and Kane (2012b), demonstrates this effect, with working memory negatively associated with mind wandering in a more difficult task requiring in- hibitory control, but not predicting mind wandering in an easier version of the task. Moreover, mind wandering during high-demand tasks was more strongly associated with performance decrements than mind wandering during less demanding tasks. However, due to the coarseness of meta-analytic methods, Randall et al. (2014) could not fully examine resource theory predictions regarding the changing relationships between these variables—an endeavor requiring the measurement of attentional capacity and experimental manipulation of task demand at multiple levels in order to evaluate nonlinear trends. Specifically, in order to more accurately investigate the nuanced relationships between mind wandering predictors and outcomes, the necessary empirical evidence requires both (a) interactive and (b) nonlinear effect estimation. Rummel and Boywitt (2014) found support for the idea that individuals with higher attentional capacity (working memory capacity) were more likely to flexibly adapt their available resources to the demands of the task—mind wandering more during an easy task than a difficult one. However, Rummel and Boywitt (2014) only used two levels of task demand, so they were unable to evaluate nonlinear quadratic trends to capture the full range of resource sensitivity (low- demand: resource insensitive, moderate-demand: resource sensitive, high-demand: resource insensitive).

Xu and Metcalfe (2016) drew on the regional of proximal learning model (Metcalfe, 2009; Metcalfe & Kornell, 2003), which states that attentional focus will be best when people engage in tasks that are neither too easy nor hard for them. They found that mind wandering was less frequent when the content studied (Spanish vocabulary) was within the individual’s zone of proximal learning based on expertise in Spanish. That is, more knowledgeable people mind wandered more during easier tasks and less during harder tasks, relative to less knowledgeable people. Notably, working memory capacity was not examined in the Xu and Metcalfe (2016) study, rather Spanish knowledge was used to understand how resource sensitive a task would be for an individual learner. This is appropriate given that prior knowledge provides cognitive resources in the form of knowledge structures into which people integrate new learning (Beier & Ackerman, 2005). However, because knowledge was the focal cognitive resource in the Xu and Metcalfe (2016) study, questions remain about the interaction of attentional capacity as measured by WMC and task demands on mind wandering frequency.

Despite some evidence supporting resource theories in the prediction of mind wandering, conflicting results regarding the as- sociation between attentional capacity, mind wandering, and task performance persist. For example, Robison and Unsworth (2017) recently found a significant negative correlation between working memory capacity and mind wandering even in very low-demand reaction time tasks. In order to answer the call for further research on the context-dependencies involved in mind wandering as- sociations (Robison & Unsworth, 2017), we aim to clarify mixed results and to extend and validate initial findings on new samples and tasks (i.e., mathematics as opposed to memory, Rummel & Boywitt, 2014, and language Xu & Metcalfe, 2016). To do so, we draw on the resource allocation framework (Kanfer & Ackerman, 1989) that integrates both individual and task characteristics in the same theoretical perspective and that has already been found to account for moderating effects on mind wandering at the meta-analytic level (Randall et al., 2014).

In the following set of experiments, we rely on resource theories to help reconcile and synthesize competing perspectives in mind wandering research about task demands and individual cognitive abilities (e.g., WMC) in order to predict who is most likely to mind wander under what circumstances, and the consequences mind wandering has for task performance. We draw on integrated resource theories (Kanfer & Ackerman, 1989; Norman & Bobrow, 1975) to acknowledge the context dependencies on the relationships be- tween individual capabilities, mind wandering, and task performance. Additionally, we build on the work of Xu and Metcalfe (2016), by evaluating non-linear curves in mind wandering frequency across task demand levels, and the work of Rummel and Boywitt (2014), by using WMC as the moderator that affects these non-linear trends. Specifically, we predict that mind wandering is most likely to occur in resource insensitive tasks (i.e., very easy or very difficult math tasks), and this determination depends in part on task demand and individual resource capacity. Therefore, in the following three experiments, we hypothesize that:

Hypothesis 1:. Mind wandering will demonstrate a concave relationship across task demand levels such that mind wandering will be more frequent for low- and high-demand tasks than moderate-demand tasks.

Hypothesis 2:. Attentional capacity interacts with task demand to predict mind wandering. Specifically, the negative relationship between attentional capacity and mind wandering will be stronger (more negative) for high-demand tasks than for low-demand tasks.

Moreover, the performance consequences of mind wandering during tasks that are more resource insensitive will be different for tasks of high and low difficulty levels. Mind wandering is a shift in attention from a primary task to other thoughts (Smallwood & Schooler, 2006). Therefore, we expect detrimental consequences on performance of a primary task to the extent that the task requires attention to successfully perform and the mind wanders (Randall et al., 2014; Robison & Unsworth, 2017). Note, however, that this prediction does not preclude benefits for performance on a secondary task that is the subject of the mind wandering; e.g., a personal task, such as successfully planning a dinner for later that evening during a staff meeting (Mooneyham & Schooler, 2013; Randall et al., 2014). Furthermore, this prediction would not apply to tasks that do not require constant attention (2015; Thomson, Smilek, & Besner, 2014). Indeed, in accordance with resource theory’s recognition of the interactive effects of task demand and individual capability (Kanfer & Ackerman, 1989), we expect that a tasks’ resource sensitivity will moderate the extent to which mind wandering is more or less detrimental to primary task performance (Rummel & Boywitt, 2014). Support for the idea that mind wandering is more detrimental during high-demand vs. low-demand tasks also comes from research with different tasks, including simple memory and perceptual tasks (e.g., Thomson et al., 2014), and even more complex cognitive tasks such as reading (e.g., Feng, D'Mello, & Graesser,

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

29

2013). We aim to replicate these findings by manipulating task demand within-study on a high-level cognitive task—mathema- tics—in accordance with a resource theory perspective. Specifically, it is expected that mind wandering will be more detrimental to performance in high demand versus low demand tasks.

Hypothesis 3:. Mind wandering will result in worse performance during a high-demand task than during a low-demand task. That is, the negative relationship between mind wandering and task performance will be stronger (more negative) for high-demand tasks than for low-demand tasks.

We test these hypotheses in a series of three experiments in the performance domain of a math task, manipulating levels of task demand by administering math tests at varying levels of difficulty: from single digit addition to complex trigonometry. Individual differences in WMC are obtained as an indicator of attentional capacity. Finally, mind wandering is assessed via self-report following each math task. There are two common methods of self-reported mind wandering assessment: (a) experience-sampling methods (often called thought probes) that interrupt task performance intermittently to ask for current attentional direction, and (b) post-hoc scales asking participants to reflect on thought direction after task completion. Historically, thought probes are more frequently used than post-hoc scales (70% vs. 30%, respectively, of the samples included in a recent meta-analysis of mind wandering; Randall et al., 2014). Each self-report method has its advantages and disadvantages; however, meta-analysis found no evidence that measurement method moderated the effect of mind wandering on task performance, suggesting both methods can successfully capture self-reported mind wandering (Randall et al., 2014). This is further supported by primary studies that find convergence in neurological measures of the brain in its mind wandering state and online or post-hoc self-report scales of mind wandering (Barron, Riby, Greer, & Smallwood, 2011; Christoff, Gordon, Smallwood, Smith, & Schooler, 2009). Thus, in order to minimize the effects that thought probes might have on attention and performance in the moment, and due to the relatively short 15-min performance episodes, we elected to use a post- hoc scale to assess mind wandering as opposed to thought probes. More information about the selected scale is provided in the Measures section. Protocol and materials for Experiments 1–3 were approved by university institutional review board and all data analyses were performed with SAS® software, Version 9.4.

2. Experiment 1

2.1. Experiment 1 method

2.1.1. Participants One hundred forty-two undergraduate participants were recruited from a psychology subject pool at a university in the

Southwestern U.S. and awarded research credit for study participation (47.89% female; M age= 19.59). Subjects were randomly assigned to one of three between-subjects conditions: Low-difficulty math test (n=48, 50.0% female, M age= 19.88), moderate- difficulty math test (n=47, 47.8% female; M age=19.48), and high-difficulty math test (n=48, 45.83% female, M age= 19.42). An a priori power analysis for a linear multiple regression with three predictors (task demand, working memory, and the interaction term) demonstrated a sample size of 127 would be adequate to test our hypotheses (power: 0.85; α=0.05; small-to-medium effect size= 0.15 consistent with meta-analytic estimates, Randall et al., 2014).

2.2. Procedure

We experimentally manipulated task demand between subjects using math items of low, moderate, and high difficulty levels. All participants completed a computerized working memory measure and one version of a paper-pencil math test depending on condition assignment. Following completion of the 15-min math tests, participants completed a post hoc mind wandering scale.

2.2.1. Measures 2.2.1.1. Mind wandering. The selected post-hoc mind wandering scale, the Attention Regulation Scale (Randall & Beier, 2017), was developed using exploratory and factor analysis of items from the most common assessments of on- and off-task thought (Kanfer & Ackerman, 1989; Kanfer, Ackerman, Murtha, Dugdale, & Nelson, 1994; Sarason, Sarason, Keefe, Hayes, & Shearin, 1986; see Randall & Beier, 2017 for further details on scale development). Eight items assessed the frequency of various off-task thoughts or mind wandering during the math test (e.g., “I thought about other activities (for example, assignments, work),” “I let my mind wander while doing the task”) in Likert-format (1=Never to 5= Very Often; α=0.82–0.90). The Attention Regulation scale is available by contacting the first author, and reliability estimates are provided in Tables 1–3.

2.2.1.2. Task demand. Task demand was operationalized as math test difficulty level. The low-demand math test was 155 fifth-grade level math questions (e.g., 3

6 + 7

6 = , What is the value of 72?). The moderate-demand math test was 42 SAT quantitative questions

rated as “easy” or “moderately easy” on a scale provided by test publishers (e.g., “If x =−2 and y=−3, what is the value of x2

(x− y)?”). The high-demand math test was 23 of the most difficult sample items from the GRE quantitative section (e.g., “What is the largest integer x such that (37x)(54x) divides evenly into 4547?”). Performance was scored as the percentage of correct answers out of the total number of questions attempted per test. Reliability estimates (split-half reliabilities with a Spearman Brown correction) are presented in Table 1.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

30

2.2.1.3. Attentional capacity. Attentional capacity was operationalized as individual WMC, assessed using the automated Reading Span (RSPAN) task (Unsworth, Redick, Heitz, Broadway, & Engle, 2009). This task requires subjects to read a sentence (10–15 words long) and judge whether it makes sense (e.g., “The grades for our finals will classroom the outside posted be door.”), while remembering a set of unrelated letters presented after each sentence. After 3–7 trials, subjects recall the memorized letters in order. Participants were assigned partial-credit representing the proportion of correctly recalled letters in each trial, which was their RSPAN score.

Table 1 Descriptive statistics and correlations between study variables in Experiment 1.

M SD Reliability estimate Correlation with WMC Correlation with mind wandering

1. WMC 43.56 15.22 – – – 2. Mind Wandering – Low 1.66 0.73 0.88 −0.17 – 3. Mind Wandering – Moderate 1.51 0.53 0.88 0.05 – 4. Mind Wandering – High 2.06 0.95 0.90 −0.12 – 5. Performance – Low 0.95 0.08 0.99 0.16 0.08 6. Performance – Moderate 0.88 0.08 0.78 0.09 0.03 7. Performance – High 0.57 0.32 0.63 0.15 −0.18

Note. Total N=143; Low-demand task condition n=48; Moderate-demand task condition n=47; High-demand task condition n=48. WMC=Working memory capacity. Low, Moderate, & High refer to task demand levels. Performance scores represent proportions for the math tests. Estimates were calculated using Cronbach’s Alpha for the mind wandering scores, and using Spearman Brown-corrected split-half reliabilities for the performance math test scores. *p < .10.

Table 2 Descriptive statistics and correlations between study variables in Experiment 2.

M SD 1 2 3 4 5 6 7

1. WMC 43.44 15.51 – 2. Mind Wandering – Low 1.97 0.89 −0.36** 0.87 3. Mind Wandering – Moderate 2.11 0.91 −0.33** 0.41*** 0.88 4. Mind Wandering – High 2.40 0.99 −0.15 0.28** 0.57*** 0.85 5. Performance – Low 0.95 0.06 0.15 0.06 0.00 −0.05 0.99 6. Performance – Moderate 0.86 0.10 0.21 −0.15 −0.42*** −0.23* 0.39** 0.85 7. Performance – High 0.57 0.27 −0.13 −0.09 −0.22* −0.24* 0.13 0.28** 0.31

Note. N=59. WMC=Working memory capacity. Low, Moderate, & High refer to task demand levels. Performance scores represent proportions for the math tests. Reliability estimates are presented in italics along the diagonal. Estimates were calculated using Cronbach’s Alpha for the mind wandering scores, and using Spearman Brown-corrected split-half reliabilities for the performance math test scores. * p < .10. ** p < .05. *** p < .001.

Table 3 Descriptive statistics and correlations between study variables in Experiment 3.

M SD 1 2 3 4 5 6 7 8 9

1. WMC 42.74 17.32 – 2. Mind Wandering – Very Low 2.68 1.14 0.12 0.91 3. Mind Wandering – Low 2.15 1.03 0.07 0.26** 0.91 4. Mind Wandering – Moderate 2.06 0.93 0.00 0.35*** 0.43*** 0.90 5. Mind Wandering – High 2.10 0.96 −0.10 0.25** 0.53*** 0.48*** 0.92 6. Performance – Very Low 1.00 0.00 0.08 0.09 0.16* 0.11 0.15* 0.99 7. Performance – Low 0.95 0.05 0.11 0.24** −0.27** −0.18** −0.30*** 0.08 0.99 8. Performance – Moderate 0.87 0.13 0.12 0.07 −0.23** −0.21** −0.25** 0.13 0.67*** 0.87 9. Performance – High 0.59 0.27 0.08 0.08 −0.19** −0.18** −0.25** 0.11 0.34*** 0.46*** 0.53

Note. N=133. WMC=Working Memory Capacity. Very Low, Low, Moderate, & High refer to task demand levels. Performance scores are pro- portions for the math tests. Reliability estimates are presented in italics along the diagonal. Estimates were calculated using Cronbach’s Alpha for the mind wandering scores, and using Spearman Brown-corrected split-half reliabilities for the performance math test scores. * p < .10. ** p < .05. *** p < .001.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

31

2.3. Experiment 1 results

Descriptive statistics, reliability estimates, and inter-correlations for all study variables are included in Table 1. To evaluate Hypothesis 1—that mind wandering would demonstrate a concave pattern across task demand levels—we conducted an ANOVA with mind wandering as the dependent variable and task demand/condition (low, moderate, high) as the between-subjects independent variable, with a planned contrast to evaluate the predicted quadratic curvilinear relationship. Fig. 2 shows average mind wandering for participants in the different conditions across demand levels. The omnibus ANOVA was statistically significant (F[2, 140]= 6.74, p= .002, η2= 0.09), suggesting that average mind wandering levels differed between task demand levels. In support of Hypothesis 1, the predicted concave quadratic contrast comparing mind wandering during the low- and high-difficulty tasks to that of the mod- erate-difficulty task, was also significant (F[2, 140]=6.60, p= .011, η2= 0.05) in the expected direction (see Fig. 2).

Hypothesis 2 predicted that attentional capacity would interact with task demand to predict mind wandering, such that those with relatively higher attentional capacity would be more likely to mind wander during a low-demand task and less likely to mind wander during a high-demand task. To test this hypothesis, we ran a moderated multiple regression with grand centered means of working memory (RSPAN), task demand, and the interaction of these two variables as independent variables, and mind wandering as the dependent variable. The overall regression model was not significant, R2= 0.03, F(3, 139)= 2.56, p= .058. Working memory did not significantly predict mind wandering, B=−0.01, SE B=0.01, t(139)=−1.05, p= .297, although task demand did, B=0.20, SE B=0.08, t(139)= 2.59, p= .011, suggesting individuals in the higher task demand condition engaged in more mind wandering than those in low-demand task condition. However, the interaction of these two predictors was not significant, B < 0.01, SE B=0.01, t(139)= 0.08, p= .935, indicating no support for our expectation that mind wandering could be predicted from the interaction between WMC and task demand. Hypothesis 2 was not supported.

Hypothesis 3 predicted that mind wandering would be associated with worse task performance during the high-demand task relative to the low-demand task. To test this hypothesis, we again ran a moderated multiple regression using grand mean-centered scores for mind wandering, task demand, and the interaction of these two variables as independent variables, and task performance as the dependent variable. The overall regression model was significant, R2= 0.38, F(3, 139)= 30.36, p < .001. Mind wandering did not significantly predict task performance, B=−0.03, SE B=0.02, t(139)=−1.44, p= .151. However, task demand did predict performance such that performance was worse for participants in high-demand conditions, B=−0.17, SE B=0.02, t (139)=−8.41, p < .001. Additionally, the predicted interaction was negative and significant at the p < .10 level, B=−0.04, SE B=0.02, t(139)=−1.67, p= .096, providing some support for the idea that as task demand increased, higher levels of mind wandering predicted lower math task performance relative to the negative effect of mind wandering on lower-demand task per- formance (see Fig. 3). Although the effect was not large, we considered this finding in partial support of Hypothesis 3.

2.4. Experiment 1 discussion

The results from Experiment 1 supported our prediction of a concave trend in mind wandering across task demand levels and the previous findings of Xu and Metcalfe (2016), with mind wandering being more frequent in the low-difficulty and high-difficulty math tasks compared to the moderate-difficulty task. The results did not, however, support our hypothesis that mind wandering could be predicted by the interaction between WMC and task demand. In contrast to meta-analytic associations (Randall et al., 2014), we failed to find a significant negative relationship between WMC and mind wandering overall. Thus, the findings from Experiment 1 failed to support theoretical predictions that mind wandering frequency depends on the interaction of individuals’ attentional ca- pacity and task demand level.

The data also supported our prediction that the relationship between mind wandering and performance depends on task demands.

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Low Moderate High

M in

d W

an de

ri ng

F re

qu en

cy

Task Demand

Fig. 2. Mind wandering frequency across three levels of task demand in Experiment 1. Error bars represent standard error of the mean. Low-demand n=48, Moderate-demand n=47, High-demand n=48.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

32

In a moderated multiple regression test, mind wandering interacted with task demands, suggesting that the degree of decrease in task performance as mind wandering increased depended on task demand level. Specifically, as hypothesized, mind wandering was more harmful in the high-demand condition than the low-demand condition, suggesting that the consequences of mind wandering on task performance are more harmful at higher levels of task demand, although the effect was significant only at the p < .10 level.

Despite support for two of our hypotheses, limitations of Experiment 1 include the between-subjects nature of the experiment introducing challenges for statistical power. Specifically, comparisons regarding interactive effects were evaluated between-subjects instead of within-subjects, making it difficult to determine the extent to which relationships between individual differences in WMC and mind wandering change as a function of multiple task difficulty levels. Additionally, participants completed only one level of the math task, resulting in only 15min of task performance. This short time-on-task may have affected resource requirements, making it easier to pay attention, which has been shown in previous research to influence the relationships between WMC, mind wandering, and task performance (Randall et al., 2014). Thus, we designed a second experiment addressing these concerns by manipulating task difficulty within-subjects—resulting in a longer time-on-task, and a more comprehensive evaluation of the same hypotheses with more statistical power, which is key for evaluating interactive effects and nonlinear trends (Aguinis, 1995).

3. Experiment 2

3.1. Experiment 2 method

3.1.1. Participants Sixty-three undergraduate participants were recruited from a psychology subject pool at a university in the Southwestern U.S. and

awarded research credit for study participation. Prior to data analysis, four of the participants were removed from the dataset: two due to a computer error that failed to record responses, and two for not following directions on the WMC measure. Consistent with typical scoring methods for the RSPAN task (Conway et al., 2005), participants who made frequent errors (less than 85% accuracy) in the secondary task (judging whether a simple sentence makes sense) were considered to be not following instructions and were not included in the analyses. The remaining sample consisted of 59 participants (54.2% female;M age=19.3) and comprises the subjects included in the analyses. Power analyses for a repeated measures ANOVA, within-between interaction for three groups and three measurements demonstrated a sample size of 54 would be adequate to test our hypotheses (power: 0.95; effect size= 0.25 is con- sistent with meta-analytic effect sizes; Randall et al., 2014).

3.1.2. Procedure Participants were randomly assigned to one of three conditions that counterbalanced the presentation order of task demand level

within-subjects (e.g., low-high-moderate difficulty, moderate-low-high difficulty, and high-moderate-low difficulty). All participants completed a computerized WMC measure and each of three versions of a paper-pencil math test (low, moderate, and high demand levels). Following each math test, participants completed mind wandering scales. Beyond the change in experimental design from between-subjects to within-subjects, necessitating conditions to counterbalance task demand order, the tasks and assessments used in Experiment 2 were identical to those in Experiment 1.

3.2. Experiment 2 results

Descriptive statistics, reliability estimates, and inter-correlations for all study variables are included in Table 2. There was no effect of condition on mind wandering overall, F(2, 56)= 1.33, p= .272, η2= 0.05, suggesting the counterbalanced presentation order of test difficulty did not produce different rates of mind wandering overall. To evaluate Hypothesis 1—that mind wandering would demonstrate a concave pattern across task demand levels—we examined linear and quadratic curvilinear relationships of mind wandering in ANOVA with task demand (low, moderate, high) as the within-subjects independent variable. Fig. 4 shows means and trends for mind wandering across demand levels. The omnibus ANOVA was statistically significant (F[2, 116]= 5.63, p= .005,

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

Low Task Demand High Task Demand Ta

sk P

er fo

rm an

ce

Low Mind Wandering

High Mind Wandering

Fig. 3. The interactive effects of mind wandering and task demand on task performance in Experiment 1, showing the differential effects of high and low levels of mind wandering (+1 SD, −1 SD, respectively) on performance at high and low levels of task demand.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

33

η2= 0.09), as was the linear contrast (F[1, 58]= 8.44, p= .005, η2= 0.13), showing a linear increase in mind wandering as task demand increased. However, the quadratic contrast was not significant, (F[1, 58]= 0.74, p= .395, η2= 0.01), showing that, al- though mind wandering increased for the high-demand task, it did not also increase for the low-demand task as anticipated. Therefore, Hypothesis 1 was not supported.

Hypothesis 2 predicted that attentional capacity would interact with task demand to predict mind wandering, such that those with relatively higher attentional capacity would be more likely to mind wander during a low-demand task and less likely to mind wander during a high-demand task. Because Experiment 2 used both between (working memory) and within (task difficulty) variables, we used latent growth curve analysis (LGCA) to test Hypothesis 2, which allowed us to examine both intercept (level of mind wandering on the low-demand task) and slope (the extent to which mind wandering changes across task demand levels), and to evaluate potential changes in these relationships as a function of participants’ WMC (the covariate). This was a single-level, single-group SEM approach to LGCA that modeled intercept and slope as covarying latent factors (i.e., random effects) similar to how a Confirmatory Factor Analysis (CFA) is modeled, with observed variables that define latent intercept and slope (growth) factors. There were two steps to evaluating this hypothesis with LGCA. First, mind wandering scores (the DV) for the three levels of task demand (low, moderate, high) were modeled as manifest indicators in the unconditional model, with intercept (level of mind wandering on the low- demand task) and slope (change in mind wandering across levels of task demand) treated as latent variables. Second, we followed the analysis of the unconditional model with a conditional model to which we added WMC as a covariate to assess its moderating effect on mind wandering across task demand levels. Analysis of the conditional model permitted us to test whether the amount of mind wandering people experience in low-demand tasks and the change in this level of mind wandering as task demand increases, depends on WMC (Hypothesis 2).

Maximum likelihood unstandardized parameter estimates of the unconditional latent growth analysis revealed an intercept of 1.95 (SE=0.10), representing the average mind wandering score for the low-demand task, and a positive slope of 0.22 (SE= 0.07), indicating an increase in 0.22 units of mind wandering for each level of increasing task demand. Hypothesis 2 was tested with a second conditional model that added WMC to the LGCA in order to evaluate whether attentional capacity influences either the overall level of mind wandering (i.e., the intercept) or the extent to which mind wandering changes across demand levels (i.e., the slope). The results of this analysis showed that incorporating WMC into the latent growth model significantly reduced the estimated in- tercept of mind wandering (−0.34, SE= 0.10, p < .001), suggesting that individuals with higher WMC engaged in less mind wandering during the low-demand task. However, the increase in mind wandering for increasing levels of task demand was un- affected by WMC, as the slope for the WMC covariate (0.09, SE=0.07, p= .206) was not significant. In other words, WMC did not alter the positive slope of task demand on mind wandering as expected. Hypothesis 2 was not supported.

Hypothesis 3 predicted that mind wandering would be associated with worse task performance during the high-demand task relative to the low-demand task. Unlike Hypothesis 2, all variables were within-subjects. Thus, we ran a moderated multiple re- gression with grand mean-centered mind wandering, task demand, and the interaction of these two variables as independent vari- ables, and task performance as the dependent variable. The overall regression model was significant, R2= 0.48, F(3, 173)= 54.17, p < .001. Mind wandering significantly negatively predicted task performance, B=−0.04, SE B=0.01, t(173)=−2.65, p= .001, as did task demand, B=−0.18, SE B=0.02, t(173)=−11.39, p < .001, meaning that overall performance was worse for people who mind wandered more frequently, and for more demanding tasks. More importantly, the interaction of the two predictors was significant, B=−0.04, SE B=0.02, t(173)=−2.46, p= .015, indicating support for our expectation that the relationship between mind wandering and performance depends on task demand. The negative direction of the moderation suggests the negative re- lationship between mind wandering and performance increased in strength (i.e., became more negative) across demand level, as hypothesized. Fig. 5 plots the interaction, demonstrating that the difference between high and low levels of mind wandering (+1 SD, −1 SD, respectively) is negligible at low levels of task demand, but that people who engage in more frequent mind wandering during high-demand tasks perform significantly worse than those who mind wander less. Hypothesis 3 was supported.

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Low Medium High

M in

d W

an de

ri ng

F re

qu en

cy

Task Demand

Fig. 4. Mind wandering frequency across three levels of task demand in Experiment 2. Error bars represent standard error of the mean. N=59.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

34

3.3. Experiment 2 discussion

The results from Experiment 2 did not support our prediction of a concave trend in mind wandering across task demand levels as observed in Experiment 1. Instead, the effect was linear, with people reporting the most mind wandering in the high-demand task and the least during the low-demand task. Mind wandering levels were also highest in the high-difficulty task in Experiment 1, but the mirror increase on the low-difficulty end of the demand spectrum was not observed in Experiment 2. One possibility for this finding is that resources were still sufficiently engaged in controlled processing during the low-demand math task in order to prevent disen- gagement. In other words, this low-demand task may not have been easy enough to trigger mind wandering. Latent growth curve models evaluating the changing relationship between individuals’ attentional capacity and mind wandering across task demand levels did not reveal the hypothesized effects. Although individuals with higher attentional capacity mind wandered less frequently overall, the difference in mind wandering frequency based on task demand level was not dependent on an individual’s attentional capacity.

Consistent with Experiment 1, the results did support our prediction that the relationship between mind wandering and per- formance depends on task demand, with this relationship decreasing as demand increased. This supports the idea that the perfor- mance consequences of mind wandering during low-demand tasks are minimal, whereas the performance consequences of mind wandering during more demanding tasks are more pronounced and harmful (see Fig. 5). This underscores the importance of in- corporating task demand into a discussion of the performance consequences of mind wandering.

One limitation of Experiment 2 that may have accounted for our mixed results is that the demand of the low-demand math test may not have been resource-insensitive enough to adequately test the hypothesized effects. Because mind wandering was not more frequent in the low demand task than the medium demand task (Fig. 4), it may have been that the low demand task (i.e., fifth grade math such as adding and subtracting fractions) required attentional capacity for most subjects to complete and as such did not permit mind wandering (i.e., it was resource sensitive for most people). Thus, we designed a third experiment adding a fourth, very-low level of task demand to re-examine the same hypotheses.

4. Experiment 3

4.1. Experiment 3 method

4.1.1. Participants One hundred thirty-four participants were recruited from the same undergraduate student subject pool as Experiments 1 and 2.

Due to a computer recording error, data from one participant was not recorded; the remaining 133 participants were used in all analyses (51% women;M age=19.4). The increase in task demand levels in Experiment 3 and our desire to increase statistical power to detect effects in the latent growth curve analysis (see Hertzog, Lindenberger, Ghisletta, & von Oertzen, 2006) necessitated a larger sample than the previous experiments.

4.1.2. Procedure Similar to Experiment 2, a within-subjects design was used in Experiment 3. Participants were randomly assigned to one of four

conditions, representing different presentation orders of task demand within-subjects, including the three demand levels from Experiments 1 and 2, with the addition of a very-low demand (i.e., extremely easy) condition (very low-moderate-high-low; low-high- moderate-very low; moderate-very low-low-high; high-low-very low-moderate). Beyond this, the tasks and assessments used in Experiment 3 were identical to those in the previous experiments.

4.1.3. Measures 4.1.3.1. Task performance. The very-low demand math test comprised 928 single digit addition problems with two response options (e.g., “1+2 =”). Performance scores on this test, as with the others, represented the percentage of correct responses divided by the

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

Low Task Demand High Task Demand Ta

sk P

er fo

rm an

ce

Low Mind Wandering

High Mind Wandering

Fig. 5. The interactive effects of mind wandering and task demand on task performance in Experiment 2, showing the differential effects of high and low levels of mind wandering (+1 SD, −1 SD, respectively) on performance at high and low levels of task demand.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

35

total number of items attempted.

4.2. Experiment 3 results

Descriptive statistics, reliability estimates, and intercorrelations for all study variables are presented in Table 3. There was no effect of condition on mind wandering overall, F(3, 129)= 0.01, p= .999, η2 < 0.01, suggesting the counterbalanced presentation order of test difficulty did not produce different rates of mind wandering overall. To evaluate Hypothesis 1, that mind wandering would be more frequent among the very low and high demand tasks than during the low and moderate demand tasks, we used the same procedure as in Experiment 2: a repeated-measures ANOVA with task demand as the within-subjects independent variable and mind wandering as the dependent variable. The omnibus ANOVA was statistically significant (F[3, 396]=17.31, p < .001, η2= 0.12), and both the linear (F[1, 132]= 29.28, p < .001, η2= 0.18) and quadratic (F[1, 132]= 19.33, p < .001, η2= 0.13) contrasts were significant. An examination of the mind wandering trend across task demand levels revealed the hypothesized concave relationship was partially displayed. The increase in mind wandering from low-demand to very low-demand was not mirrored by a similar increase in mind wandering on the high end of the task demand spectrum (see Fig. 6), providing partial support for Hypothesis 1.

We used the same LGCA approach as in Experiment 2 to test Hypothesis 2—that attentional capacity moderates the influence of task demand on mind wandering. Maximum likelihood unstandardized parameter estimates of the unconditional model (task demand predicting mind wandering without WMC) revealed an intercept of 2.43 (SE=0.08) on the very-low demand task and decreasing slope of −0.14 (SE= 0.03) across all participants for each increasing level of task demand. Thus, the inclusion of the very low- demand task in Experiment 3 produced a higher intercept than in Experiment 2, signaling higher mind wandering levels, and a negative rather than a positive slope, demonstrating that mind wandering decreased 0.14 units across increasing levels of task demand. As in Experiment 2 we ran a conditional model to predict mind wandering (DV) with WMC as a covariate (IV 2) potentially interacting with task demand levels (IV 1). Maximum likelihood unstandardized parameter estimates of the conditional LGCA re- vealed the inclusion of WMC significantly increased the estimated intercept (0.14, SE=0.08, p= .079), and significantly reduced the negative slope of mind wandering across increasing task demand levels (−0.08, SE=0.03, p= .012). Therefore, consistent with Hypothesis 2, higher WMC scores were associated with higher mind wandering scores during the very low-demand task. Additionally, as task demand increased, mind wandering scores decreased, and the magnitude of that decrease was influenced by WMC. Speci- fically, as predicted, individuals with higher WMC engaged in less mind wandering as task demand increased, relative to those with lower WMC, providing support for Hypothesis 2. Fig. 7 displays the nature of this interaction—that individuals with relatively higher levels of attentional capacity (1 SD above mean WMC) are more likely to mind wander during very low-demand tasks than they are during high-demand tasks. In contrast, those with lower levels of attentional capacity (1 SD below mean WMC) are less likely to mind wander during a very low-difficulty task, but are more likely to mind wander during a high-difficulty task.

To test Hypothesis 3, that mind wandering would be associated with worse task performance during the high-demand versus the very low-demand task, we again used moderated multiple regression with grand mean-centered mind wandering, task demand, and their interaction as independent variables, and task performance as the dependent variable. The overall model was significant R2= 0.47, F(3, 527)= 159.92, p < .001. There was not a main effect of mind wandering on task performance, B < 0.01, SE B=0.02, t(527)= 0.22, p= .823, but there was a main effect of task demand level, B=−0.10, SE B=0.01, t(527)=−6.73, p < .001, with lower performance scores for higher-demand tests. Furthermore, the interaction of mind wandering and task demand level was significant and negative, B=−0.02, SE B=0.01, t(527)=−2.58, p= .010, supporting our prediction that the re- lationship between mind wandering and task performance is moderated by task demand, such that the negative relationship becomes stronger as demand increases. Fig. 8 plots the interaction, demonstrating that the difference between high and low levels of mind wandering (+1 SD,−1 SD from mean mind wandering scores, respectively) is less pronounced at low levels of task demand, but that the harmful effects of more frequent mind wandering are exaggerated at higher levels of task demand. Therefore, Hypothesis 3 was

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Very Low Low Moderate High

M in

d W

an de

ri ng

F re

qu en

cy

Task Demand Level

Fig. 6. Mind wandering frequency across four levels of task demand in Experiment 3. Error bars represent standard error of the mean. N=133.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

36

supported.

4.3. Experiment 3 discussion

The results of Experiment 3 largely replicate and extend key findings from Experiments 1 and 2, providing additional support for our predictions and complementing the findings of previous work (Rummel & Boywitt, 2014; Xu & Metcalfe, 2016). As in Experi- ments 1 and 2, the level of mind wandering was dependent on task demand, and similar to Experiment 1, there was evidence of a curvilinear effect. However, the expected concave relationship was not as clearly displayed as in Experiment 1—perhaps not sur- prisingly, given the notorious difficulty in detecting and predicting curvilinear effects in psychological science (Carter et al., 2014; Pierce & Aguinis, 2013). In Experiment 3, mind wandering appeared to be most frequent in the very low-demand task; a simple arithmetic task that college students (indeed most people) should be able to complete without much attentional effort. This is inconsistent with Experiment 2 where mind wandering frequency was highest in the high-demand task; although Experiment 2 did not include as low demand of a task as Experiment 3.

Interactive effects of attentional capacity and mind wandering across the four task demand levels, tested using latent growth curve models, supported our theoretical predictions. First, higher levels of mind wandering were observed in the very low-demand task for individuals with more attentional resources relative to those with fewer resources. Additionally, as hypothesized, those with relatively higher levels of attentional resources reduced their mind wandering more as task demand increased compared to people with relatively fewer attentional resources, who were more likely to mind wander during the most difficult task. This demonstrates that people with ample attentional resources may be more likely to mind wander when their resources exceed task demands, but less likely to disengage during more demanding tasks. This is different from the trend for individuals with lower attentional capacity who were less likely to mind wander during a very low-demand task when their resources were still engaged, but more likely to disengage at high levels of task demand that exceeded their resource capacity. These findings support predictions derived from resource theories that mind wandering depends on the match between individual attentional resources and task demand.

Similar to Experiments 1 and 2, the performance consequences of mind wandering differed on the low and high ends of the task demand continuum where tasks are more resource insensitive. Specifically, as predicted, mind wandering harmed performance more

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Very Low Low Moderate High

M in

d W

an de

ri ng

F re

qu en

cy

Task Demand

-1 SD WMC

+1 SD WMC

Fig. 7. Latent growth curve analysis displaying the interaction of task demand and working memory capacity (WMC) in the prediction of mind wandering frequency.

-0.5

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

Low Task Demand High Task Demand

Ta sk

P er

fo rm

an ce

Low Mind Wandering

High Mind Wandering

Fig. 8. The interactive effects of mind wandering and task demand on task performance in Experiment 3, showing the differential effects of high and low levels of mind wandering (+1 SD, −1 SD, respectively) on performance at high and low levels of task demand.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

37

during a high-demand task than a very low-demand task (see Fig. 8). Although this finding may seem intuitive, our study provides consistent empirical support for the importance of investigating mind wandering in the context of task demand and person-ability levels.

5. Combined analysis of hypotheses across experiments

To draw conclusions based on all three experiments, we examined our predictions using a combined dataset as did Xu and Metcalfe (2016). Drawing conclusions based on all of the data is possible because the same measures were used in all experiments and participants were sampled from the same student population. We recognize that each study had unique design features: most notably, Study 1 was between-subjects and 2 and 3 were within-subjects, also Study 3 had four levels of task difficulty instead of three. However, a combined dataset with participants from all three experiments allows us to re-evaluate the hypotheses across all par- ticipants in the same between-subjects manner as in Experiment 1. In an attempt to account for the violation of the assumptions of homogeneity of variance and data independence required by this integrative analysis (i.e., treating within-subjects effects as between- subjects), we statistically examined the effect of experiment and homogeneity of variances in each analysis. Specifically, we evaluated the degree to which statistical assumptions of the independence of data had been violated by (a) testing whether variables and relationships demonstrated study effects and (b) using Levene’s test of the homogeneity of variance.1 These integrative analyses are intended to be more suggestive than definitive in order to display general trends across the three studies.

In evaluation of the predicted concave mind wandering trend across task demand levels (Hypothesis 1), we conducted a one-way ANOVA with mind wandering as the dependent variable and task demand (very low, low, moderate, and high) as the independent variable. As in Experiment 1, planned contrasts compared the hypothesized concave quadratic relationship of mind wandering across demand levels with a linear trend (increasing or decreasing). The results from this exploratory analysis confirmed our theoretical predictions (see Fig. 1, Panel A) that mind wandering would be most frequent in more resource insensitive tasks (very low-demand task MW: Mean=2.68, SD=1.38; high-demand task MW: Mean= 2.17, SD=0.97), and less frequent during more resource sen- sitive tasks (low-demand task MW: Mean=2.01, SD=0.96; moderate-demand task MW: Mean= 1.96, SD=0.89). Additionally, the contrast modeling the expected concave quadratic effect (F[3, 848]=40.25, p < .001, η2= 0.05) was stronger than the linear trend (F[3, 848]=21.79, p < .001, η2= 0.03). Hypothesis 1 was supported.

To evaluate Hypothesis 2, that attentional capacity interacts with attentional demand to predict mind wandering, we ran the same moderated multiple regression as run in Experiment 1 because not all data was collected within-subjects. Thus, the dependent variable, mind wandering was regressed onto three independent variables: task demand, attentional capacity (WMC), and the in- teraction of task demand and attentional capacity (task demand*WMC). The overall regression model was significant, R2= 0.02, F(3, 848)= 5.77, p < .001. Working memory did not significantly predict mind wandering, B < −0.01, SE B < 0.01, t(848)=−1.37, p= .170, although task demand did, B=−0.11, SE B=0.03, t(848)=−3.42, p < .001, suggesting individuals engaged in more mind wandering when completing lower-demand tasks than higher-demand tasks. Additionally, the predicted negative interaction of these two variables was significant, B < 0.01, SE B < 0.01, t(848)=−1.97, p= .049, indicating support for our expectation that mind wandering was less likely for individuals with greater attentional capacity at higher levels of task demand than for low-capacity individuals who, in turn, were less likely to mind wander during less demanding tasks. Hypothesis 2 was supported.

Finally, Hypothesis 3, that mind wandering is more detrimental to performance during more demanding tasks than during less demanding tasks, was also evaluated with moderated multiple regression. The dependent variable was math task performance and the independent variables were mind wandering, task demand level (very low, low, moderate, and high), and the interaction of these two variables (task demand*mind wandering). The overall regression model was significant, R2= 0.46, F(3, 847)= 237.76, p < .001. Mind wandering significantly negatively predicted task performance, B=−0.04, SE B=0.01, t(847)=−6.85, p < .001, indicating worse performance for people who engaged in more mind wandering. Task demand also negatively predicted task performance, B=−0.14, SE B=0.01, t(847)=−25.34, p < .001, suggesting individuals performed worse on the more de- manding tasks. Additionally, the predicted negative interaction of these two variables was significant, B=−0.02, SE B=0.01, t (847)=−3.14, p= .002, providing support for our expectation that task performance was harmed by mind wandering during high- demand tasks more than during low-demand tasks. Hypothesis 3 was supported. Altogether, despite the limitations of this combined analysis of the data from our three experiments in a manner similar to Xu and Metcalfe (2016) integrated analysis, the results suggest full support for a resource theory-based perspective of mind wandering and its context-dependent relationship with predictors (WMC) and consequences (task performance).

1 A mixed ANOVA predicting mind wandering demonstrated significant main effects of task demand: F(3,842)=18.16, p < .001, η2= 0.06, and study: F(2,842)= 8.95, p < .001, η2= 0.02, but no interactive effect of the two: F(4,842)=2.26, p= .061, η2= 0.01. So although participants in the three experiments had different levels of mind wandering overall, task-level differences did not depend on which experiment they participated in. Levene’s test suggested no violation of the homogeneity assumption for mind wandering scores in the low-demand: F(2,237)=2.85, p= .060, η2= 0.02, or high-demand: F(2,237)=0.06, p= .945, η2 < 0.01 tasks, but did for the moderate-demand task: F(2,236)= 4.37, p= .014, η2= 0.04. A one-way ANOVA revealed no significant difference in RSPAN scores between subjects in the three experiments, F(2,849)=0.21, p= .809, η2 < 0.01, but there was a small, significant difference between samples when testing for hetereogeneity of variance: F(2,849)=4.23, p= .015, η2= 0.01. Finally, a mixed ANOVA predicting task performance revealed no interactive effect of difficulty by study: F(4,841)=0.19, p= .945, η2 < 0.01, and Levene’s tests showed no evidence of heterogeneous variances: low F(2,236)=0.46, p= .635, η2 < 0.01, moderate F (2,236)= 1.50, p= .224, η2= 0.01, and high F(2,237)=2.13, p= .122, η2= 0.02, suggesting consistency in task performance between samples.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

38

6. General discussion

The three experiments and integrated analysis reported here demonstrated key evidence for resource theory predictions that individual attentional resources and task demands are important interactive determinants of mind wandering and its performance consequences. Across all experiments, task demand influenced the level of mind wandering observed, with mind wandering more common during resource insensitive tasks (i.e., very low or very high demand tasks). This is consistent with predictions that at- tentional effort is more important during the performance of tasks where focused attention may be expected to lead to changes in task performance—what is referred to as a resource sensitive task (Kanfer & Ackerman, 1989; Norman & Bobrow, 1975)—than during a more resource insensitive task where the demand is sufficiently far above or below an individuals’ capability as to mitigate the effect of attention on performance. Using the example from earlier, paying extra attention during tooth brushing or while listening to an expert explain glacial geomorphology might not yield much of a demonstrable effect on dental health or one’s ability to understand a novel, complex topic, resulting in more mind wandering under such circumstances. Consequently, mind wandering should be more frequent during resource insensitive tasks of very high and very low attentional demand, than during those of moderate demand- level.

In Experiment 1 the predicted quadratic trend was evident, with mind wandering more frequent in low-demand and high-demand tasks than during a moderately-demanding task. This finding was replicated only in part across the two subsequent experiments. In Experiment 2 mind wandering was most frequent in a high-demand task (with no evidence of a nonlinear trend). In Experiment 3 mind wandering exhibited a quadratic trend across task demand levels, but was most frequent in a very low-demand task, showing no mirror increase on the high-demand end. However, the predicted quadratic trend was evident in the integrated analysis with data from all three experiments. This difficulty in replicating the exact quadratic trend in each experiment could be related to the relative versus absolute effect of task difficulty on mind wandering. Specifically, it may be that the introduction of the very low-demand task in Experiment 3 affected perceptions of task difficulty for higher-demand tasks such that difficult tasks seemed achievable and thus people were less likely to mind wander during task execution. In other words, it may be that the introduction of the very low difficulty task made the high difficulty task more resource sensitive for most people. This is consistent with recent findings from Beck and Schmidt (2018), who found that high self-efficacy levels prompt a reduction of resource allocation (time devoted to a task) in order to prevent overinvesting in resource insensitive tasks. Self-efficacy is related to mastery experiences (Bandura, 1977) and easily completing very low-demand math items without great effort to focus attention (i.e., regulate resources) may have raised participant self-efficacy for more difficult items and, by contrast, encouraged participants to stay more focused during the high-demand task where their resources were in more demand. Further research should examine the effects of relative difficulty and self-efficacy on mind wandering, as well as other forms of resource allocation (e.g., time on task; Beck & Schmidt, 2018) to address such possibilities.

Alternatively, the inconsistencies between studies when examining mind wandering rates could be due in part to task presentation order. Despite our findings that the counterbalanced presentation order for different conditions did not alter mind wandering rates overall, order may yet serve an additional interactive role in influencing mind wandering rates. For example, research suggests task difficulty may prompt more proactive attentional control in response to more demanding tasks, which may reduce mind wandering for participants who are first introduced to a difficult task before completing an easier one (Engle & Kane, 2004; McVay & Kane, 2010). Alternatively, performing two resource-insensitive tasks prior to resource-sensitive tasks (e.g., high and very low-demand tasks first, followed by a moderate-demand task), may exacerbate tendencies to engage in mind wandering as it could deplete resources more quickly, especially because mind wandering becomes more frequent and more harmful as time-on-task increases (Kane et al., 2007; Randall et al., 2014). Nonetheless, presentation order was not a variable of interest in the current study, and thus we did not include all possible permutations of order presentation in order to adequately examine such possibilities. Therefore, future research can examine task order as an additional interactive effect beyond cognitive resources and task demand to help explain why different ends of the task resource insensitivity continuum (very low-demand vs. very high-demand) may lead to more or less mind wandering.

Also in support of resource theory, we found evidence of an interaction between individual attentional capacity and task demand: people with more attentional resources tended to mind wander relatively less as task demands increased. This key finding, consistent with Rummel and Boywitt (2014) earlier work, suggests that task resource sensitivity depends on both the task and the person—as not all tasks will be equally demanding for all people. This finding also suggests that the skills that people develop over the lifespan may affect mind wandering. Tasks that were once so demanding that they precipitated mind wandering (i.e., outside of one’s zone of proximal development, Vygotsky, 1978; or the region of proximal learning, Xu & Metcalfe, 2016) could become increasingly resource sensitive and achievable as people develop a repertoire of skills and expertise. For instance, solving geomorphology problems might exceed the average individuals’ capability to comprehend, consequently causing the mind to wander, but the same might not be expected for someone familiar with the topic. Although we did not find the same interaction between attentional capacity and task demand in all experiments, our results from Experiment 3—which had the most levels of task demand and most statistical power to detect the hypothesized effects—as well as the integrated analysis, point to the general trend of resource insensitivity when task demand is well within, or well beyond, a person’s attentional capacity.

These results combine to inform our perspective on why it matters that task demand and individual differences interactively predict mind wandering frequency. In all three experiments we found that the performance consequences of mind wandering during low-demand tasks are less detrimental than those associated with mind wandering during high-demand tasks. These results reinforce past work showing mind wandering demonstrates stronger negative associations with more difficult as opposed to easier tasks (e.g., Feng et al., 2013; Randall et al., 2014; Thomson et al., 2014) and extends these findings to a new task domain – mathematics. This consistent finding underscores the point that mind wandering may not be universally harmful in terms of its effect on task

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

39

performance, or at least that there are situations—namely, when primary task performance is more difficult and demands more attention—that disengaging will be more harmful. This finding is consistent with the idea that mind wandering driven by low task demand may reflect a beneficial or adaptive response to boredom with benign consequences, whereas mind wandering driven by high task demand could reflect disengagement and frustration, yielding adverse performance consequences (McVay & Kane, 2012b; Randall et al., 2014; Rummel & Boywitt, 2014).

The possible link between mind wandering, emotional experiences, and motivation is something that has received limited re- search attention until recently (Killingsworth & Gilbert, 2010; Randall, 2015; Robison & Unsworth, 2018; Seli, Cheyne, Xu, Purdon, & Smilek, 2015; Unsworth & McMillan, 2013), and thus necessitates more research to understand the key role of task demand. However, as introduced previously, resource theories account for both cognitive abilities and motivation as sources influencing the self-regulation process, therefore motivation may serve as an additional determinant of available resources to deter mind wandering in addition to WMC. Indeed, this idea has found recent support in the mind wandering literature (Seli et al., 2015; Seli, Maillet, Smilek, Oakman, & Schacter, 2017; Seli, Schacter, Risko, & Smilek, 2017; Unsworth & McMillan, 2013), and is consistent with such integrative theory as the resource allocation framework (Kanfer & Ackerman, 1989) that incorporates both ability- and motivation- based determinants of skilled performance. Some of the interest in motivation and self-regulation may be connected to mind wan- dering researchers’ attempts to account for what is referred to ask “task-related interference,” or interfering evaluative thoughts about one’s performance on a task (Smallwood, Obonsawin, & Reid, 2003) that have been shown to harm performance (McVay & Kane, 2012a). Such thoughts may be more likely in high-demand, but not low-demand resource insensitive tasks, as individuals are more likely to experience emotion control failures when overwhelmed by hard tasks rather than when bored by easy tasks (Kanfer & Heggestad, 1997), which could help explain the reasons for mind wandering during different types of resource insensitive tasks. Thus, motivation, emotions, and other individual differences (e.g., daydreaming style; Marcusson-Clavertz et al., 2016), deserve future research attention as additional interactive predictors of mind wandering.

Our findings further highlight the challenge in identifying the complex interactive and nonlinear relationships posited by resource theory, particularly as related to the moderating effect of attentional capacity on the relationship between task demand and mind wandering. Scholars have frequently emphasized the difficulty in detecting and predicting curvilinear effects in psychological science (Carter et al., 2014; Pierce & Aguinis, 2013), especially interactive nonlinear effects, such as those under investigation here. This point is further complicated by our decision to manipulate task demand ourselves. Resource theories acknowledge that what is considered low-demand or easy for one individual may in fact be more demanding or difficult for another (Ackerman, 1988). Indeed, we predicted that would be the case, therefore we examined interactive effects of person and task characteristics on mind wandering frequency. However, it is possible that our selected experimental manipulations of task demand did not adequately test the limits of peoples’ attention, as evidenced by the relatively low means for mind wandering. Perhaps a longer duration in task engagement such as that experienced in a regular work day would result in increased mind-wandering independent of task difficulty (Giambra, 1995; Randall et al., 2014). This possibility, in part, was addressed by changing the between-subjects nature of Experiment 1 to a within- subjects design in Experiments 2 and 3, requiring more time-on-task overall for these participants. We did not directly examine task duration in this study, however, so we cannot make any conclusions about its effect on mind wandering, although recent evidence suggests such study design considerations (e.g., between vs. within-subjects) can also affect mind wandering (Forrin, Risko, & Smilek, 2018). It is also likely that the relationship between attentional capacity and the predictors and outcomes examined in this study was attenuated given that the sample used was restricted in range of abilities (enrolled in a selective university). Future research might account for these possibilities by manipulating task duration and using a sample with a wider range of ability.

7. Implications & conclusions

These findings help enrich existing theories of mind wandering (Kane & McVay, 2012; Smallwood & Schooler, 2006; Smallwood, 2013) by further integrating this phenomenon with resource-based theories of information processing (Beier & Oswald, 2012; Kanfer & Ackerman, 1989; Norman & Bobrow, 1975; Randall et al., 2014) and answering the call for additional research on the context dependencies of mind wandering (Robison & Unsworth, 2017). Specifically, mind wandering theory and research should acknowl- edge that any prediction regarding either the expected frequency or expected benefit/harm of mind wandering will be improved by accounting for interactive relationships between task and individual characteristics. Our study extends recent work (Randall et al., 2014; Rummel & Boywitt, 2014; Xu & Metcalfe, 2016) demonstrating that both individual attentional capacity and task demand influence mind wandering by demonstrating the nuanced, nonlinear, and interactive effects between these key variables. Overall conclusions from the current set of experiments support resource theory predictions that (a) mind wandering is more frequent in both very-low and very-high demand (i.e., resource insensitive) tasks, (b) the determination of resource sensitivity depends in part on attentional capacity (WMC), and subsequently, (c) the effect of mind wandering on task performance depends on the demand of the task. These findings build upon previous mind wandering research identifying task demand and individual resources as important determinants of mind wandering (e.g., Feng et al., 2013; Forster & Lavie, 2009; Kane et al., 2007, 2017; Konishi, Brown, Battaglini, & Smallwood, 2017; Marcusson-Clavertz et al., 2016; Robison & Unsworth, 2017; Rummel & Boywitt, 2014; Smallwood & Andrews- Hanna, 2013; Thomson et al., 2014; Xu & Metcalfe, 2016) in an integrative, theory-driven fashion. For example, in accordance with resource theories, we demonstrated that sufficient treatment of the role of task demand should also account for individual differences in attentional capacity (e.g., WMC), and that both individual and task characteristics interact to predict both mind wandering frequency as well as its performance consequences. This will help move theory along and will also address what, on the surface, appear to be inconsistent results regarding, for example, the simple correlation between attentional capacity (WMC) and mind wandering (e.g., Baird, Smallwood, & Schooler, 2011; Levinson et al., 2012; McVay & Kane, 2012a, 2012b; Robison & Unsworth,

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

40

2017) without a consideration of the more nuanced view of attentional capacity as a moderator of the relationship between task demand and mind wandering.

Our results provide several more implications that are relevant in both research and applied settings. First is that the reasons for mind wandering as well as the performance consequences for doing so are more harmless during low-demand tasks as opposed to high-demand tasks. Mind wandering can thus be considered productive in many instances when performance of a primary task does not necessitate as much effortful processing. Second, it would be short-sighted to claim individuals with lower levels of cognitive resources will always mind wander more than those with more resources. Instead, rather than considering mind wandering as a main effect, we highlight the moderating influences of person (WMC) and task characteristics (difficulty) on mind wandering rates and task performance, respectively. Individuals may engage in more mind wandering at the extremes of task demand—during either very high or very low-difficulty tasks that are less sensitive to changes in effort and attention (i.e., more resource insensitive tasks). Moreover, the extent to which tasks will vary in resource sensitivity will depend on the attentional capacity of the actor, pointing to the dynamic nature of these interactions. That is, as people develop skills and expertise over the lifespan the likelihood of mind wandering should likewise be affected.

The conclusion that task- and person-characteristics are important determinants of mind wandering and its effect on task per- formance has implications for researchers who tend toward a one-size-fits-all approach to inducing mind wandering in laboratory settings (e.g., Dixon & Li, 2013; Forster & Lavie, 2009). Moreover, it implies that in applied settings, mind wandering interventions should address individual and task characteristics in order to be effective. For example, optimizing task demands to align with individuals’ attentional capacity may help reduce unwanted mind wandering (such as would be expected when people are engaged in autotelic activity; Csikszentmihalyi, 1990). For difficult tasks this might take the form of skill acquisition through practice (Ackerman, 1988), whereas for simple tasks increasing expectations (e.g., imposing a time limit) might be effective. In employment and educational domains, design latitude allowing individuals to structure their work/study and switch between tasks varying in demand to avoid boredom or frustration might help in these efforts. By identifying multiple routes to mind wandering, the results of this study suggest a combination of attentional resources and task considerations will improve the prediction and potentially the management of mind wandering.

8. Data statement

Research data from the three studies and other study materials, including the Attention Regulation Scale (Randall & Beier, 2017) and information about scale development are all available from the first author upon request.

9. Declarations of interest

None.

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.concog.2018.11.006.

References

Ackerman, P. L. (1988). Determinants of individual differences during skill acquisition: Cognitive abilities and information processing. Journal of Experimental Psychology: General, 117, 288–318. https://doi.org/10.1037/0096-3445.117.3.288.

Aguinis, H. (1995). Statistical power problems with moderated multiple regression in management research. Journal of Management, 21, 1141–1158. https://doi.org/ 10.1016/0149-2063(95)90026-8.

Baird, B., Smallwood, J., & Schooler, J. W. (2011). Back to the future: Autobiographical planning and the functionality of mind wandering. Consciousness and Cognition, 20, 1604–1611. https://doi.org/10.1016/j.concog.2011.08.007.

Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191–215. https://doi.org/10.1037/0033-295X.84.2.191. Beck, J. W., & Schmidt, A. M. (2018). Negative relationships between self-efficacy and performance can be adaptive: The mediating role of resource allocation. Journal

of Management, 44, 555–588. Beier, M. E., & Ackerman, P. L. (2005). Age, ability, and the role of prior knowledge on the acquisition of new domain knowledge: Promising results in a real-world

learning environment. Psychology and Aging, 20, 341–355. https://doi.org/10.1037/0882-7974.20.2.341. Beier, M. E., & Oswald, F. L. (2012). Is cognitive ability a liability? A critique and future research agenda on skilled performance. Journal of Experimental Psychology:

Applied, 18, 331–345. https://doi.org/10.1037/a0030869. Carter, N. T., Dalal, D. K., Boyce, A. S., O’Connell, M. S., Kung, M., & Delgado, K. M. (2014). Uncovering curvilinear relationships between conscientiousness and job

performance: How theoretically appropriate measurement makes an empirical difference. Journal of Applied Psychology, 99, 564–586. https://doi.org/10.1037/ a0034688.

Christoff, K., Gordon, A. M., Smallwood, J., Smith, R., & Schooler, J. W. (2009). Experience sampling during fMRI reveals default network and executive system contributions to mind wandering. Proceedings of the National Academy of Sciences of the United States of America, 106, 8719–8724. https://doi.org/10.1073/pnas. 0900234106.

Conway, A. R. A., Kane, M. J., Bunting, M. F., Hambrick, D. Z., Wilhelm, O., & Engle, R. W. (2005). Working memory span tasks: A methodological review and user’s guide. Psychonomic Bulletin and Review, 12, 769–786. https://doi.org/10.3758/BF03196772.

Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York: Harper & Row. Dixon, P., & Li, H. (2013). Mind wandering in text comprehension under dual-task conditions. Frontiers in Psychology, 4, 682. https://doi.org/10.3389/fpsyg.2013.

00682. Engle, R. W., & Kane, M. J. (2004). Executive attention, working memory capacity, and a two-factor theory of cognitive control. In B. Ross (Ed.). The psychology of

learning and motivation (pp. 145–199). New York, NY: Academic Press. Feng, S., D'Mello, S., & Graesser, A. C. (2013). Mind wandering while reading easy and difficult texts. Psychonomic Bulletin & Review, 20, 586–592.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

41

Forrin, N. D., Risko, E. F., & Smilek, D. (2018). In the eye of the beholder: Evaluative context modeulates mind-wandering. Acta Psychologica, 185, 172–179. Forster, S., & Lavie, N. (2009). Harnessing the wandering mind: The role of perceptual load. Cognition, 111, 345–355. https://doi.org/10.1016/j.cognition.2009.02.

006. Giambra, L. M. (1995). A laboratory method for investigating influences on switching attention to task-unrelated imagery and thought. Consciousness and Cognition, 4,

1–21. https://doi.org/10.1006/ccog.1995.1001. Hertzog, C., Lindenberger, U., Ghisletta, P., & von Oertzen, T. (2006). On the power of multivariate latent growth curve models to detect correlated change.

Psychological Methods, 11, 244–252. https://doi.org/10.1037/1082-989X.11.3.244. Kane, M. J., Brown, L. H., McVay, J. C., Silvia, P. J., Myin-Germeys, I., & Kwapil, T. R. (2007). For whom the mind-wanders, and when: An experience-sampling study

of working memory and executive control in daily life. Psychological Science, 18, 614–621. https://doi.org/10.1111/j.1467- 9280.2007.01948.x. Kane, M. J., Gross, G. M., Chun, C. A., Smeekens, B. A., Meier, M. E., Silvia, P. J., & Kwapil, T. R. (2017). For whom the mind wanders, and when, varies across

laboratory and daily-life settings. Psychological Science, 28, 1271–1289. https://doi.org/10.1177/0956797617706086. Kane, M. J., & McVay, J. C. (2012). What mind wandering reveals about executive-control abilities and failures. Current Directions in Psychological Science, 21, 348–354.

https://doi.org/10.1177/0963721412454875. Kane, M. J., Meier, M. E., Smeekens, B. A., Gross, G. M., Chun, C. A., Silvia, P. J., & Kwapil, T. R. (2016). Individual differences in executive control of attention,

memory, and thought, and their associations with schizotypy. Journal of Experimental Psychology: General, 145, 1017–1048. Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive abilities: An integrative/aptitude-treatment interaction approach to skill acquisition. Journal of Applied

Psychology, 74, 657–690. https://doi.org/10.1037/0021-9010.74.4.657. Kanfer, R., Ackerman, P. L., Murtha, T., Dugdale, B., & Nelson, L. (1994). Goal setting, conditions of practice, and task performance: A resource allocation perspective.

Journal of Applied Psychology, 79, 826–835. https://doi.org/10.1037/0021-9010.79.6.826. Kanfer, R., & Heggestad, E. D. (1997). Motivational traits and skills: A person-centered approach to work motivation. In L. L. Cummings, & B. M. Staw (Vol. Eds.),

Research in organizational behavior: Vol. 19, (pp. 1–56). Greenwich, CT: JAI Press. Killingsworth, M. A., & Gilbert, D. T. (2010). A wandering mind is an unhappy mind. Science, 330, 932. https://doi.org/10.1126/science.1192439. Klinger, E., Gregoire, K. C., & Barta, S. G. (1973). Physiological correlates of mental activity: Eye movements, alpha, and heart rate during imagining, suppression,

concentration, search, and choice. Psychophysiology, 10, 471–477. https://doi.org/10.1111/j.1469-8986.1973.tb00534.x. Konishi, M., Brown, K., Battaglini, L., & Smallwood, J. (2017). When attention wanders: Pupillometric signatures of fluctuations in external attention. Cognition, 168,

16–26. https://doi.org/10.1016/j.cognition.2017.06.006. Levinson, D. B., Smallwood, J., & Davidson, R. J. (2012). The persistence of thought: Evidence for a role of working memory in the maintenance of task-unrelated

thinking. Psychological Science, 23, 375–380. https://doi.org/10.1177/0956797611431465. Marcusson-Clavertz, D., Cardeña, E., & Terhune, D. B. (2016). Daydreaming style moderates the relation between working memory and mind wandering: Integrating

two hypotheses. Journal of Experimental Psychology: Learning, Memory, and Cognition, 42, 451–464. McVay, J. C., & Kane, M. J. (2010). Does mind wandering reflect executive function or executive failure? Comment on Smallwood and Schooler (2006) and Watkins

(2008). Psychological Bulletin, 136, 188–197. https://doi.org/10.1037/a0018298. McVay, J. C., & Kane, M. J. (2012b). Why does working memory capacity predict variation in reading comprehension? On the influence of mind-wandering and

executive attention. Journal of Experimental Psychology: General, 141, 302–320. https://doi.org/10.1037/a0025250. McVay, J. C., & Kane, M. J. (2012a). Drifting from slow to “D’oh”: Working memory capacity and mind-wandering predict extreme reaction times and executive

control error. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38, 525–549. https://doi.org/10.1037/a0025896. Metcalfe, J. (2009). Metacognitive judgments and control of study. Current Directions in Psychological Science, 18, 159–163. https://doi.org/10.1111/j.1467-8721.

2009.01628.x. Metcalfe, J., & Kornell, N. (2003). The dynamics of learning and allocation of study time to a region of proximal learning. Journal of Experimental Psychology General,

132, 530–542. https://doi.org/10.1037/0096-3445.132.4.530. Mooneyham, B. W., & Schooler, J. W. (2013). The costs and benefits of mind-wandering: A review. Canadian Journal of Experimental Psychology, 67, 11–18. https://doi.

org/10.1037/a0031569. Norman, D. A., & Bobrow, D. B. (1975). On data-limited and resource-limited processes. Cognitive Psychology, 7, 44–64. https://doi.org/10.1016/0010-0285(75)

90004-3. Pierce, J. R., & Aguinis, H. (2013). The too-much-of-a-good-thing effect in management. Journal of Management, 39, 313–338. https://doi.org/10.1177/

0149206311410060. Randall, J. G. (2015). Mind wandering and self-directed learning: Testing the efficacy of self-regulation interventions to reduce mind wandering and enhance online training

(Doctoral dissertation). Houston, TX: Rice University. Randall, J. G., & Beier, M. E. (2017, May). Mind wandering: Theoretical and empirical establishment of a nomological network. In Poster presented at the Association

for Psychological Science Annual Convention, Boston, MA. Randall, J. G., Oswald, F. L., & Beier, M. E. (2014). Mind-wandering, cognition, and performance: A theory-driven meta-analysis of attention regulation. Psychological

Bulletin, 140, 1411–1431. https://doi.org/10.1037/a0037428. Robison, M. K., & Unsworth, N. (2015). Working memory capacity offers resistance to mind-wandering and external distraction in a context-dependent manner.

Applied Cognitive Psychology, 29, 680–690. Robison, M. K., & Unsworth, N. (2017). Working memory capacity and mind-wandering during low-demand cognitive tasks. Consciousness and Cognition, 52, 47–54. Robison, M. K., & Unsworth, N. (2018). Cognitive and contextual predictors of spontaneous and deliberate mind-wandering. Journal of Experimental Psychology:

Learning, Memory, and Cognition, 44, 85–98. Rummel, J., & Boywitt, C. D. (2014). Controlling the stream of thought: Working memory capacity predicts adjustment of mind-wandering to situational demands.

Psychonomic Bulletin & Review, 21, 1309–1315. Sarason, I. G., Sarason, B. R., Keefe, D. E., Hayes, B. E., & Shearin, E. N. (1986). Cognitive interference: Situational determinants and traitlike characteristics. Journal of

Personality and Social Psychology, 51, 215–226. https://doi.org/10.1037/0022-3514.51.1.215. Schooler, J. W., Smallwood, J., Christoff, K., Handy, T. C., Reichle, E. D., & Sayette, M. A. (2011). Meta-awareness, perceptual decoupling and the wandering mind.

Trends in Cognitive Sciences, 15, 319–326. https://doi.org/10.1016/j.tics.2011.05.006. Seli, P., Cheyne, J. A., Xu, M., Purdon, C., & Smilek, D. (2015). Motivation, intentionality, and mind wandering: Implications for assessments of task-unrelated thought.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 41, 1417–1425. Seli, P., Maillet, D., Smilek, D., Oakman, J. M., & Schacter, D. L. (2017). Cognitive aging and the distinction between intentional and unintentional mind wandering.

Psychology and Aging, 32, 315–324. Seli, P., Schacter, D. L., Risko, E. F., & Smilek, D. (2017). Increasing participant motivation reduces rates of intentional and unintentional mind wandering.

Psychological Research, 1–13. Smallwood, J. (2013). Distinguishing how from why the mind wanders: A process-occurrence framework for self-generated mental activity. Psychological Bulletin, 139,

519–535. https://doi.org/10.1037/a0030010. Smallwood, J., & Andrews-Hanna, J. (2013). Not all minds that wander are lost: The importance of a balanced perspective on the mind-wandering state. Frontiers in

Psychology, 4, 441. https://doi.org/10.3389/fpsyg.2013.00441. Smallwood, J., Obonsawin, M., & Reid, H. (2003). The effects of block duration and task demands on the experience of task unrelated thought. Imagination, Cognition

and Personality, 22, 13–31. Smallwood, J., & Schooler, J. W. (2006). The restless mind. Psychological Bulletin, 132, 946–958. https://doi.org/10.1037/0033-2909.132.6.946. Smeekens, B. A., & Kane, M. J. (2016). Working memory capacity, mind wandering, and creative cognition: An individual-differences investigation into the benefits of

controlled versus spontaneous thought. Psychology of Aesthetics, Creativity, and the Arts, 10, 389–415. Thomson, D. R., Besner, D., & Smilek, D. (2015). A resource-control account of sustained attention: Evidence from mind-wandering and vigilance paradigms.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

42

Perspectives on Psychological Science, 10, 82–96. Thomson, D. R., Smilek, D., & Besner, D. (2014). On the asymmetric effects of mind wandering on levels of processing at encoding and retrieval. Psychonomic Bulletin &

Review, 21, 728–733. Unsworth, N., & McMillan, B. D. (2013). Mind wandering and reading comprehension: Examining the roles of working memory capacity, interest, motivation, and

topic experience. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39, 832–842. https://doi.org/10.1037/a0029669. Unsworth, N., & McMillan, B. D. (2014). Similarities and differences between mind-wandering and external distraction: A latent variable analysis of lapses of attention

and their relation to cognitive abilities. Acta Psychologica, 150, 14–25. Unsworth, N., Redick, T. S., Heitz, R. P., Broadway, J., & Engle, R. W. (2009). Complex working memory span tasks and higher-order cognition: A latent variable

analysis of the relationship between processing and storage. Memory, 17, 635–654. Unsworth, N., & Robison, M. K. (2017). The importance of arousal for variation in working memory capacity and attention control: A latent variable pupillometry

study. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43, 1962–1987. Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Boston, MA: Harvard University Press. Xu, J., & Metcalfe, J. (2016). Studying in the region of proximal learning reduces mind wandering. Memory & Cognition, 44, 681–695.

J.G. Randall et al. Consciousness and Cognition 67 (2019) 26–43

43

How-pervasive-is-mind-wandering--really--_2018_Consciousness-and-Cognition.pdf

• How pervasive is mind wandering, really?,
• Introduction
• Materials and method
• Participants
• Experience sampling
• Rates of mind wandering
• Results
• Discussion
• Supplementary material
• References