A new approach to differentiate states of mind wandering: Effects of working memory capacity

Cognition 179 (2018) 202–212

Contents lists available at ScienceDirect

Cognition journal homepage: www.elsevier.com/locate/cognit

Original Articles

A new approach to differentiate states of mind wandering: Effects of working memory capacity

T

⁎

Matthew J. Voss, Meera Zukosky , Ranxiao Frances Wang University of Illinois at Urbana-Champaign, United States

A R T I C LE I N FO

A B S T R A C T

Keywords: Mind wandering Meditation Working memory Memory and attention

Although widely studied, the process of how mind wandering occurs and is subsequently sustained still remains unclear. Moreover, the traditional concept of mind wandering tendency/frequency based on the self- or probecaught methods alone is incoherent and problematic. We developed a new approach to characterize the mind wandering process by combining the self-caught and probe-caught methods to estimate the time of focus and time of mind wandering separately, and examined their relationship to working memory capacity. Participants performed an OSPAN task and subsequently a basic Mindfulness Meditation Task (focus on breath). During the meditation task, participants indicated when they became aware that they were mind wandering (self-caught method), or were asked if they were mind wandering when probed (probe-caught method). Results showed that time of focus but not time of mind wandering increased with greater working memory capacity. This suggests that individuals with higher working memory capacity were able to focus on the current task longer, but had little effect on the ability to monitor and terminate mind wandering once it occurred. The theoretical and methodological implications of this new approach are discussed.

1. Introduction While performing a task, internal thoughts often grab our attention, preventing us from focusing on the current task at hand. These internally driven thoughts are termed mind wandering, and encompass thoughts that are unrelated to the current task at hand (Christoff, Irving, Fox, Spreng, & Andrews- Hanna, 2016; Randall, Oswald, & Beier, 2014; Smallwood & Schooler, 2006, 2015). Although the specific terms used and the exact definition of mind wandering varies across the literature, it is agreed upon that mind wandering can be described as a shift of attention away from the primary task at hand to internal thoughts (Christoff et al., 2016; Randall et al., 2014; Smallwood & Schooler, 2006, 2015). Various factors have been found to affect mind wandering. Task difficulty, clinical symptomology of ADHD and Dysphoria, as well as an individual’s mood have all been shown to affect the tendency to mind wander (Feng, D’Mello, & Graesser, 2013; Phillips, Mills, D’Mello, & Risko, 2016; Poerio, Totterdell, & Miles, 2013; Seli, Smallwood, Cheyne, & Smilek, 2015; Smallwood, O’Connor, Sudbery, & Obonsawin, 2007). For example, in regards to task demands, Smallwood, Ruby, and Singer (2013) examined task unrelated thoughts in relationship to task demands and found that task unrelated thoughts under a non-demanding condition of a simple choice reaction time task were

⁎

associated with reduced delay discounting, but were not in a more attention demanding working memory task. Although these findings have given us insights into the nature of mind wandering, the current methodology of measuring the tendency to mind wander is ambiguous. Here we explain the problems that arise with the current mind wandering methodology, and develop a novel, more informative way to examine mind wandering. Research on mind wandering has employed various methods to estimate people’s tendency to mind wander. One of the main ways to measure mind wandering is with thought sampling, i.e., asking participants directly through self report whether their mind was focused on the task at hand, or whether they were mind wandering (Bastian & Sackur, 2013; Christoff, Gordon, Smallwood, Smith, & Schooler, 2009; Hasenkamp, Wilson-Mendenhall, Duncan, & Barsalou, 2012; Seli, Carriere, Levene, & Smilek, 2013; Smallwood & Schooler, 2006, 2015). For example, participants may be interrupted while performing a task to check for mind wandering (dichotomous probe-caught methods, more fine-tuned probe-caught methods such as utilizing likert scales or confidence ratings, and self-caught methods) (Christoff et al., 2009; Hasenkamp et al., 2012; Mrazek, Smallwood, & Schooler, 2012; Seli, Jonker, Cheyne, Cortes, & Smilek, 2015; Smallwood & Schooler, 2006; McVay, Kane, and Kwapil, 2009, etc.). Post-hoc self reports, where individuals are not interrupted during the task, have also been utilized

Corresponding author at: Department of Psychology, University of Illinois at Urbana-Champaign, 603 E. Daniel St., Champaign, IL 61820, United States. E-mail address: [email protected] (M. Zukosky).

https://doi.org/10.1016/j.cognition.2018.05.013 Received 17 May 2017; Received in revised form 18 May 2018; Accepted 18 May 2018 0010-0277/ © 2018 Published by Elsevier B.V.

Cognition 179 (2018) 202–212

M.J. Voss et al.

report to another self report of an individual catching him/herself mind wandering. From the self-caught method alone, one can get a measurement of the entire focus-mind wandering episode duration (denoted as Tself), as illustrated in Fig. 1a. However, this measurement includes both the duration of time when an individual is focusing on the current task at hand, which can be defined as the duration from the start of the task to the moment one begins mind wandering (Tfocus), and the duration of time an individual is actually mind wandering, which can be defined as the duration from the start of mind wandering to the moment that one catches him/herself and terminates the mind wandering process (TMW). Since it is unknown when the time of focus stops and when the time of mind wandering begins with this method, one cannot distinguish the focus state duration (Tfocus) from the mind wandering state duration (TMW) using the self-caught method alone. As a result, people with the same self-caught mind wandering duration (or same self reported mind wandering frequency) can have drastically different durations that they actually spend mind wandering. For example, as illustrated in Fig. 1b, one participant could have a short duration of being in the focus state and a longer duration of being in the mind wandering state, while another participant might have a long duration of focusing on the task at hand and a short duration of being in the mind wandering state, even though both participants catch themselves mind wandering at the same frequency (and thus have the same focus-mind wandering episode duration). This difference is indistinguishable using only the self-caught method, and thus the actual duration of time an individual spent in the mind wandering state remains unclear. Thus, the self-caught frequency of mind wandering is not a satisfactory measure of how much people mind wander. The probe-caught method measures the relative proportion of time people are in a focus state and a mind wandering state. That is, if the probes are sampled randomly or uniformly across the focus-mind wandering episode, then the chance a probe happens to land at a time while the participant is mind wandering is determined by the overall proportion of time s/he spends in mind wandering (p(MW)). If an individual spends 60% of the time in an actual mind wandering state, then the chance a random probe occurs at a time s/he is mind wandering would be 0.6. Although this method can estimate the proportion of time an individual is in a focus state and a mind wandering state, it cannot determine the duration an individual spends in a focus state and in a mind wandering state in individual episodes. As illustrated in Fig. 1c, individuals with the same probe-caught mind wandering rate can have drastically different durations of focus and mind wandering in individual episodes, which reflects their different ability to maintain attention on the current task and their ability to detect mind wandering. Participant 1 (P1) is able to focus on the current task for 10 s before mind wandering is initiated, and remains in a mind wandering state for 30 s before catching mind wandering and returning to the task, while P2 is able to stay on focus for 5 s before entering a mind wandering state which lasts for 15 s. If we randomly probe these two participants, they will both be caught mind wandering 75% of the time, however both the duration of focus (Tfocus) and the duration of mind wandering (TMW) are longer for P1 compared to P2. Thus, the probe-caught rate of mind wandering is also not a satisfactory measure of the underlying mechanism of mind wandering. These methodological limitations can lead to confusion and misinterpretations in evaluating people’s tendencies to mind wander. For example, Fig. 1d shows two individuals and their mind wandering time course. The first participant (P1) has an average of 15 s in the focus state, and 5 s in the mind wandering state, while the second participant (P2) has 10 s in the focus state, and 30 s in the mind wandering state. Therefore P1 has an overall focus-mind wandering episode duration of 20 s (or will self-catch her/himself mind wandering 3 times per minute), while P2 has an overall focus-mind wandering episode duration of 40 s (or self-caught mind wandering rate of 1.5 times per minute). Thus, according to the self-caught method, P1 would be conventionally interpreted as having a higher tendency to mind wander

to estimate the occurrence of mind wandering (Helton & Warm, 2008; Smallwood et al., 2012; Smallwood, Fitzgerald, Miles, & Phillips, 2009). Among them, the self-caught and dichotomous probe-caught methods administered during a task are two of the most common thought sampling methods used to measure mind wandering, and are the primary focus of the present study. In the self-caught method, participants are asked to report whenever they realize they are mind wandering (Braboszcz & Delorme, 2010; Bastian & Sackur, 2013; Hasenkamp et al., 2012; Smallwood & Schooler, 2006, 2015). For example, participants are asked to perform a task and to report, usually with a button press, whenever they notice that their focus is off the current task. Here the number of times an individual catches her/himself mind wandering can be measured, as well as the duration of time between reports. The frequency an individual catches her/himself mind wandering or the duration of time between reports 1is interpreted as an indication of the likelihood or tendency for an individual to mind wander. That is, a longer duration between reports, or consequently a lower number of mind wandering instances reported (lower frequency of self-caught reports of mind wandering), is usually interpreted as a lower tendency or likelihood to mind wander. In contrast, a higher number of mind wandering instances reported (higher frequency of self-caught reports of mind wandering), or a shorter duration between reports, is usually interpreted as a higher tendency or likelihood to mind wander (Braboszcz & Delorme, 2010; Bastian & Sackur, 2013; Hasenkamp et al., 2012; Smallwood & Schooler, 2006, 2015). In the probe-caught method, participants are interrupted (usually with a tone or a screen display) at different time intervals while performing a task (Bastian & Sackur, 2013; Christoff et al., 2009; Seli et al., 2013; Smallwood & Schooler, 2015). At the time of the probe, participants are asked to report whether their focus was on the task at hand, or if they were mind wandering. The proportion of time an individual was mind wandering can be calculated by dividing the probes labeled as being in the mind wandering state by the total number of probes given. The probe-caught method thus can provide a measure of the overall proportion of time an individual was in a mind wandering state based on the mind wandering probe rate (Bastian & Sackur, 2013; Seli et al., 2013; Smallwood & Schooler, 2006). With the probe-caught method, the proportion or percentage of time reported to be in a mind wandering state is usually interpreted as an indication of the likelihood or tendency for an individual to mind wander. That is, a higher proportion or percentage of probes caught in the mind wandering state is thought to indicate a higher tendency or likelihood to mind wander. In contrast, a lower proportion or percentage of mind wandering probes is thought to indicate a lower tendency or likelihood to mind wander. Although intuitively sound, a closer examination of the self- and probe-caught method reveals issues with the traditional interpretations of mind wandering, and suggests that the concept of mind wandering frequency/tendency itself is incoherent. As discussed by Smallwood (2013), the inability to covertly detect the onset of mind wandering poses difficulties in studying mind wandering. For example, as illustrated in Fig. 1a, a focus-mind wandering episode can be defined as the time when an individual starts focusing (or refocusing) on a task, to when an individual realizes s/he is mind wandering and terminates the mind wandering. A focus-mind wandering episode therefore consists of a period when an individual is focused on the current task s/he is performing, and a period when s/he is actually mind wandering. This time course within a focus-mind wandering episode illuminates potential problems with the current use of the self- and probe-caught methods to measure the tendency of mind wandering. For example, the self-caught method provides a measure of the frequency of self-caught reports or the duration of time from one self

1 Frequency of reports and the duration between reports are inversely related; therefore they are essentially the same measurement in different forms.

203

Cognition 179 (2018) 202–212

M.J. Voss et al.

Fig. 1. A theoretical analysis of the traditional interpretations of mind wandering tendency based on the self- and probe-caught methods. (a) The different states of a focus-mind wandering episode. (b) Problems of the self-caught method as a measure of mind wandering tendency. Participant 1 (P1) and participant 2 (P2), both have focus-mind wandering episodes of the same duration, but the duration in the mind wandering state of P1 is longer and the duration in the focus state is shorter, than for P2. These two scenarios cannot be distinguished using only the self-caught method. (c) Problems of the probe-caught method as a measure of mind wandering tendency. The ratio between the duration in a focus state and the duration in a mind wandering state is the same for both participants, but the actual durations of being in the focus state and mind wandering state in a given episode are different. These two scenarios cannot be distinguished using only the probe-caught method alone. (d) Conflicts between the selfcaught method and the probe-caught method. Here the probe-caught and the self-caught method used alone lead to contradictory interpretations when evaluating an individual’s tendency to mind wander.

The present study took a new approach to characterize the mind wandering process by estimating the time course of the two main states of the focus-mind wandering episode separately, following the principles of the Process-Occurrence Framework by Smallwood (2013). This theory emphasizes the importance of distinguishing between processes that lead to the occurrence of an experience from those supporting the continuation of the experience once it is initiated. According to this framework, the first state Tfocus can be taken as a measure of one’s ability to maintain attention on the current task and resist the occurrence/initiation of mind wandering, while the second state TMW can be taken as a measure of processes that keep people in the mind wandering state after it occurs. We used this new approach to examine one of the main factors that has been shown to affect mind wandering, namely working memory capacity. The effect of working memory capacity on mind wandering has been an important issue on theories of mind wandering. There is debate

compared to P2, as P1′s self-reported mind wandering frequency is higher than P2′s. In contrast, using only the probe-caught method, one would reach the opposite conclusion. When probed randomly, P1 would be caught in the focus state 75% of the time and in the mind wandering state 25% of the time, while P2 would be caught in the focus state 25% of the time and in a mind wandering state 75% of the time. According to the probecaught method, P1 would be traditionally interpreted as having a lower tendency to mind wander compared to P2, as P1 is in a focus state proportionally more and a mind wandering state proportionally less than P2. Therefore for the same participants, the self-caught method and probe-caught method give exactly the opposite conclusion on who mind wanders more. This contradiction of the interpretations of more vs less tendency to mind wander based on the two methods suggests an intrinsic ambiguity of the traditional concept of mind wandering tendency itself.

204

Cognition 179 (2018) 202–212

M.J. Voss et al.

between different states of mind wandering and the corresponding underlying processes, both at the theoretical and at the experimental level. The failure to separately measure how working memory capacity affects different states of mind wandering can cause confusion in the interpretation of the data. The goal of this study was to develop a new approach to estimating time of focus (Tfocus) and time of mind wandering (TMW) within a self-caught focus-mind wandering episode separately in relationship to working memory capacity, to examine the role of executive control on the ability to focus on the task at hand, and that on the ability to maintain and/or detect mind wandering separately. Experiment 1 is a pilot study following the traditional methods to estimate some basic parameters and provide an initial measurement of the relationship between working memory capacity and the raw selfcaught or probe-caught mind wandering rate in a mindfulness of breath meditation task with our experimental setup and procedure. This served as the basis for Experiment 2, where the two methods were combined within the same participants to estimate the two states of the focusmind wandering episode.

in the literature on whether mind wandering consumes executive control resources or whether mind wandering rather results from a failure of executive control (Levinson, Smallwood, & Davidson, 2012; McVay & Kane, 2010; Randall et al., 2014; Smallwood & Schooler, 2006, 2015). The Resource Hypothesis states that mind wandering instances take up executive control resources (Smallwood & Schooler, 2006). In this view, both the current task at hand and mind wandering instances are competing for the same executive control resources. Here the same executive control resources intended to keep an individual focused on the current task at hand, are diverted from the current task and used to entertain internal mind wandering instances (Smallwood & Schooler, 2006). Thus, the resource hypothesis emphasizes that maintaining a mind wandering instance requires executive resources. As such, when more executive control resources are available, more mind wandering should be present (Antrobus, 1968; Levinson et al., 2012; Rummel & Boywitt, 2014; Smallwood & Schooler, 2006; Teasdale et al., 1995; Thomson, Besner, & Smilek, 2013). In contrast to the Resource Hypothesis, McVay and Kane (2010) proposed the Executive Control Hypothesis, stating that instead of consuming the executive control resources needed to focus on the current task at hand, mind wandering is actually a result of a failure in our executive control (McVay & Kane, 2010). Thus mind wandering is a result from a failure in executive control to block thoughts about current concerns (the default mode) during task performance (McVay & Kane, 2010). As such, individuals with lower executive control resources will be worse at blocking out this default mode, which will lead to more mind wandering during task performance (Buckner, AndrewsHanna, & Schacter, 2008; Christoff et al., 2016; Christoff et al., 2009; Hasenkamp et al., 2012; Mason et al., 2007; McVay & Kane, 2009, 2010, 2012). In summary, these two theories predict different behaviors in the relationship between executive control, as measured by working memory capacity, and mind wandering. If mind wandering instances do in fact consume executive control resources (Resource Hypothesis), individuals who have higher working memory capacity should have more available executive control resources to divert to mind wandering. Thus it is thought that individuals with higher working memory capacity should report being in a mind wandering state more compared to individuals who have lower working memory capacity (Levinson et al., 2012; Rummel & Boywitt, 2014; Teasdale et al., 1995; Thomson et al., 2013; Smallwood & Schooler, 2006). If, on the other hand, mind wandering instances are a result of a failure in executive control (Executive Control Hypothesis), individuals who have higher working memory capacity should have better executive control capacities to block out mind wandering instances. Thus it is thought that individuals should report being in a mind wandering state less compared to individuals who have lower working memory capacity (McVay & Kane, 2010, 2012). The Process-Occurrence Framework by Smallwood (2013), on the other hand, sought to reconcile this debate between the Executive Control Hypothesis and the Resource Hypothesis. The Process Occurrence Framework stresses the importance of separating different elements of mind wandering, such as mind wandering’s initiation and the maintenance of mind wandering (Smallwood, 2013). For example, as Smallwood (2013) explains, executive control could influence mind wandering directly (inhibiting mind wandering from being initiated) or indirectly (maintenance and continuation of mind wandering). The Executive Control Hypothesis and the Resource Hypothesis thus are not in direct conflict, as these different hypotheses can illuminate the role of executive control in different aspects of mind wandering. However, the lack of empirical methods to indicate when mind wandering begins, makes it difficult to separate the initiation of mind wandering (for example when a mind wandering state begins) and the maintenance of mind wandering once it has been initiated (for example maintaining mind wandering until it is recognized and terminated) (Smallwood, 2013). As a result, most previous work did not distinguish

2. Experiment 1a Experiment 1 was conducted to estimate the temporal parameters of self-caught mind wandering and examine the relationship between working memory capacity and mind wandering using the self-caught method alone within a non-demanding task, i.e., a Mindfulness Breath Meditation Task. 2.1. Methods 2.1.1. Participants Twenty undergraduate students aged 18–22 from the University of Illinois participated in experiment 1a. All participants had less than a total of 10 h of meditation practice experience during their lives and were novice meditators, to avoid practice effects of meditation, as it has been shown that meditation practice can improve focused attention and reduce mind wandering (Braboszcz, Hahusseau, & Delorme, 2010; Brewer et al., 2011; Hasenkamp et al., 2012; Lutz, Slagter, Dunne, & Davidson, 2008). 2.1.2. Working memory task Individual’s working memory score was measured using the Automated OSPAN task. This OSPAN task is a measure of working memory capacity (Unsworth, Heitz, Schrock, & Engle, 2005). During this OSPAN task, participants first mentally solved a basic math problem (eg. (2 ∗ 4) + 2 = ?). Once participants finished solving the math problem, a number would be presented on the screen. The participant would then respond with a mouse click, whether the number presented on the screen was true or false to the solution the participant mentally calculated. After they made their selection, participants would see a letter of the alphabet flash on the screen. After 3–7 trials, participants were prompted to report what letters they saw in the order they saw them. The entire OSPAN task took approximately 20 min where participants saw a total of 75 math problems and letters. The absolute OSPAN score is calculated as the total number of letters correctly recalled (for more information on the OSPAN task please see Unsworth et al., 2005). The OSPAN test used for the current study was created by Millisecond Software from the following website (1http://www. millisecond.com/download/library/OSPAN/). 2.1.3. Mindfulness breath meditation task The main task for the current study was a Mindfulness Breath Meditation Task. During the Mindfulness Breath Meditation Task, participants were instructed to breathe normally and to focus on their breath (participants were instructed to focus on the sensation of their in breath and out breath from their nostrils). The Mindfulness Breath 205

Cognition 179 (2018) 202–212

M.J. Voss et al.

again). The Tself ranged from 13.33 s to 132.54 s (M = 58.12 s, SD = 34.56 s). Fig. 2a shows the linear regression on the mean Tself as a function of the participants’ OSPAN score. The results did not show a significant correlation between working memory capacity score and the average focus-mind wandering episode duration (r = −.15, t (18) = −0.66, p = .52), suggesting that the mean focus-mind wandering episode duration (Tfocus + TMW) was not affected by the OSPAN score. These results replicated those of previous studies that show that selfcaught task unrelated thoughts did not correlate with working memory capacity in a breath counting task (Levinson et al., 2012). As stated in the introduction, due to the self-caught method’s inclusion of different components within a focus-mind wandering episode (Tself and TMW), it is unclear what these results alone mean in relationship to working memory and mind wandering. To develop a new measurement to assess different states of the focus-mind wandering episode, the probe method is also needed. As such, Experiment 1b was conducted to measure the basic parameters of probe-caught mind wandering and to examine the relationship between working memory capacity and mind wandering using a modified probe-caught method.

Meditation Task has been chosen as the task for this current study as it is thought to be a task with a high propensity for mind wandering and thus ideal to study the natural fluctuation between states of focus and states of mind wandering (Braboszcz & Delorme, 2010; Hasenkamp et al., 2012; Levinson et al., 2012). 2.2. Procedure Participants were first given the OSPAN working memory test on a computer. After completion of the OSPAN working memory test, participants were instructed on how to practice mindfulness meditation. Participants sat in a fully lit room to avoid drowsiness, but were instructed to at least partially close their eyes to avoid distractions. Participants were told to focus only on the sensation of breath and not to count breaths in their head. During this task, participants were instructed to press a spacebar when they noticed their focus was not on their breath and their attention was instead focused on other thoughts (any thought that is not focused on the breath, for example a past event, a future plan, etc.). Anytime participants noticed a task- unrelated thought, they were instructed to recognize the task unrelated-thought, push it out of their mind, and continue to focus on their breath. Once participants felt comfortable in doing this, they were instructed to press the space bar on the computer when they recognized a thought, then return to their breath. The computer ran a program on Eprime Psychology Software 2.0, which tracked and recorded every button press during the meditation session. The participants completed two 10-minute meditation sessions with a 5 min break in between where they were allowed to use the restroom and have a drink of water. The total experiment took no longer than 50 min.

3. Experiment 1b This experiment followed the traditional probe-caught method to examine whether working memory capacity affects the rate of mind wandering reports, with a methodological modification on the distribution of probes that addressed an important issue in the traditional methods.

2.3. Results and discussion

3.1. Methods

The OSPAN working memory scores ranged from 23 to 69 (M = 48.85, SD = 13.50). The self-caught mind wandering time (Tself) was calculated as the mean interval between the start of a meditation session to the first key press indicating the end of one focus-mind wandering episode, or between one key press and the next (i.e., from the end of the previous mind wandering event after which participants returned to meditation, until they caught themselves mind wandering

3.1.1. Participants Thirty-four undergraduate students aged 18–22 from the University of Illinois participated in Experiment 1b. Similar to Experiment 1a, in Experiment 1b all participants had less than a total of 10 h of meditation practice experience during their lives and were novice meditators.

Fig. 2. Results of Experiments 1a and 1b. (a) The scatter plot of Tself as a function of the OSPAN scores in Experiment 1a showing the linear regression line. (b) The scatter plot of p(MW) as a function of the OSPAN scores in Experiment 1b showing the linear regression line. 206

Cognition 179 (2018) 202–212

M.J. Voss et al.

reported they were mind wandering decreased. Experiments 1a and 1b measured some basic parameters of the selfand probe-caught mind wandering using our modified procedure, and provided an initial test of the effect of working memory capacity on the two traditional measures of mind wandering tendency (Tself and p (MW)). However, to better characterize the mind wandering process, we need to estimate the two main states of mind wandering (Tfocus and TMW) separately. Experiment 2 developed a new method to calculate the mean duration in the mind wandering state and duration in the focus state for each participant by combining the self-caught and probecaught method within the same participant.

3.2. Procedure The design and procedure was the same as Experiment 1a except the following. After completion of the OSPAN working memory test and the instruction on how to practice mindfulness meditation, a series of two meditation sessions were performed with a break in between. During each session, participants were instructed to practice the mindfulness breath meditation, but this time when they heard a tone they were to respond with where their focus was at the time of the tone. If their focus was solely on the sensation of their breath, they were instructed to respond with a V on the keyboard for yes they were focused on their breath. If they found themselves mind wandering at the time of the tone, they were instructed to respond with an N on the keyboard for no they were not focused on their breath and were having other thoughts. Once the participants began their mindfulness meditation, a computer program was run using E-prime Psychology Software 2.0. This program was set to run three consecutive cycles of 8 different trials each (with no division or gap between cycles). Each trial was set to play a tone at one of these time intervals from the previous response: 5 s, 10 s, 15 s, 20 s, 25 s, 30 s, 35 s, and 40 s. All 8 trials were randomized within each of the three cycles. Once the participants heard the tone, they had ten seconds to respond if their focus was on or off their breath. If the participants did not respond, the program would move to the next trial and the trial was marked as a “miss.” Once the participants responded, the next time interval would begin until the participants were given all three cycles of eight trials each for a total of 24 probes during the entire meditation session. Once the session was over, the participant was given a five minute break. Upon conclusion of the break, the second session of mindfulness breath meditation began. Each session lasted around 10.5 min. A key modification of the current procedure from previous studies is the timing of the probe. Based on the theoretical discussion in the Introduction, the probes need to be distributed randomly or uniformly across the focus-mind wandering episode for a probe-caught measure to accurately reflect the true proportion of time one is in a focus state versus in a mind wandering state. That is, if more probes are distributed in the earlier portion of the focus-mind wandering episode, then the probe-caught mind wandering rate can potentially be under-estimated. In contrast, if more probes are distributed in the later portion of the focus-mind wandering episode, then the measured proportion of mind wandering responses would be an over-estimation of the time people spend mind wandering. Since the all-trial mean Tself measured in Experiment 1a was about 40 s, the probes in Experiment 1b were set to the eight different time intervals evenly distributed across an average focus-mind wandering episode to provide a uniform sample of the mind wandering event and minimize any potential bias created by probing shorter or longer intervals.

4. Experiment 2 In this experiment, each participant performed both the self-caught and the probe-caught mind wandering sessions. The duration of focus (Tfocus) and duration of mind wandering (TMW) were calculated for each participant, and the relationship between their OSPAN scores and these two duration estimates were tested separately to examine whether working memory capacity affects the two aspects of mind wandering, i.e., people’s ability to remain on task and resist the occurrence of mind wandering (indicated by Tfocus), and the continuation/detection of mind wandering once it occurs (indicated by TMW). 4.1. Methods The methods were the same as in Experiments 1a and 1b except the following. 111 participants participated in the current study. Participants ranged in age from 18 years of age to 22 years of age and were recruited from the University of Illinois at Urbana-Champaign. Like Experiment 1, only participants who have had less than a total of 10 h of meditation experience were included in the data analysis, to avoid practice effects of meditation, as it has been shown that meditation practice can improve focused attention and reduce mind wandering (Braboszcz et al., 2010; Brewer et al., 2011; Hasenkamp et al., 2012; Lutz et al., 2008). Each participant completed the OSPAN task and four Mindfulness Breath Mediation Sessions in the following order. 1. Self-caught session 1: The participant performed the Mindfulness Breath Meditation Task and focus-mind wandering episodes were measured using the self-caught method. This session allowed us to calculate the average focus-mind wandering episode duration. 2. Probe-caught session 1: The participant performed the Mindfulness Breath Meditation Task and mind wandering instances were measured using the probe-caught method, where the probe intervals were based on the average focus-mind wandering episode duration calculated from the first self-caught session (further explained below). 3. Probe-caught session 2: If participants’ self-caught average time was less than 80 s, participants performed a second probe-caught session. This session was exactly the same as probe-caught session 1. If participants’ self-caught average time was longer than 80 s, participants only performed one probe-caught session, due to experimental time limitations (further explained below). 4. Self-caught session 2: The participant performed the Mindfulness Breath Meditation Task and mind wandering episodes were measured using the self-caught method. This session allowed us to calculate the average focus-mind wandering episode duration, and was exactly the same as the self-caught session 1. The session was conducted to ensure that the timing measure between a participant’s first self-caught session and the last self-caught session remained consistent over time (further explained below).

3.3. Results and discussion Five participants failed to respond within the 10-second response period in some of the trials and were not included in the data analysis.2 For the remaining 29 participants, the OSPAN working memory scores ranged from 4 to 75 (M = 51.10, SD = 18.28). Fig. 2b showed the linear regression on the proportion of mind wandering responses (p(MW)) as a function of participants’ OSPAN scores to determine if there was any relation between their working memory capacity and the proportion of times they were mind wandering. The p(MW) ranged from 0 to .77 (M = .39, SD = .18). The OSPAN working memory score was able to predict p(MW), (r = −.41, t (27) = −2.37, p = .03). These data showed that as the participants' working memory capacity increased, the proportion of times they

Overall, the entire experiment lasted approximately 2 h, where participants were allowed to take a break in between sessions.

2 The results remained the same when these participants were included in the data (r = −.38, t(32) = −2.35, p = .02).

207

Cognition 179 (2018) 202–212

M.J. Voss et al.

The self-caught session was the same as Experiment 1a except that it lasted 20 min (similar to Hasenkamp et al., 2012). The probe-caught session was the same as Experiment 1b, except that the timing of the probes were based on the average self-caught time duration of the participant’s own individual focus-mind wandering episode, calculated from the first self-caught session. Upon hearing a probe, participants were asked to press the keyboard button “m” if their focus was on their breath, and press a keyboard button “z” if their focus was off their breath. These button presses were counterbalanced between participants to account for possible response biases, where half of the participants were instructed to press “m” if their focus was on their breath and “z” if there focus was off their breath, and half of the participants were instructed to press “m” if their focus was off their breath, and “z” if their focus was on their breath. Focus off their breath was defined as any thought that is not on the breath or breath sensation, such as future events, past events, personal events, and events that are happening around the world. The probe-caught session contained 4 cycles (each cycle containing 8 probes), and was run using a program created in C+ +. This session time varied depending on the average timing from the self-caught session. The placement timing of the probes for the probe-caught session was based on the average duration of the self-caught session. After the 20 min self-caught session, the average duration of a focus-mind wandering episode was calculated by finding the average duration in seconds between button presses during the self-caught session (measuring the start of focusing on breath, to when a participant realized their mind had wandered). We then used this average to calculate the presentation timing of the probes for the following probe-caught sessions. To determine when to present the probes for each participant, we divided the mean duration from the self-caught session by 8, and added three seconds to each time to ensure a minimal time that would be meaningful to probe and allow time for participants to enter the meditation state. For example, if a participant has an average of 40 s for the self-caught session, the 8 probe delays for a cycle would be 8 s, 13 s, 18 s, 23 s, 28 s, 33 s, 38 s, and 43 s. These delays were measured from participants’ response to the previous probe, e.g., an 8 s probe would be presented 8 s after the participant responded to the previous probe. The order of the probes was randomized for each cycle.

Tfocus = Tself ∗ (1−p (MW))

(1)

TMW = Tself ∗p (MW)

(2)

Tself is the average time of a focus-mind wandering episode calculated from the first and second self-caught sessions and ranged from 10.27 s to 124.90 s (M = 43.08 s, SD = 27.36 s). p(MW) is the proportion of time a participant said that s/he was mind wandering based on the probe caught session alone, and ranged from .23 to .83 (M = .47, SD = .13). Both Tfocus and TMW were calculated separately for each participant individually. The logic for Eqs. (1) and (2) is straightforward. As illustrated above in Fig. 1a, Tself is a measure of the average duration of an individual’s focus-mind wandering episodes in seconds (when they start focusing on the task to when they catch themselves mind wandering), which is the sum of the time of focus and the time of mind wandering. That is,

Tself = Tfocus + TMW

(3)

Moreover, when sampled randomly or uniformly, the chance a probe falls in the period of focus state (p(focus)) relative to the chance of the probe falling in the period of mind wandering (p(MW) = 1 − p (focus)) is determined by the relative duration of time of focus and time of mind wandering in a focus-mind wandering episode. That is,

p (focus): p (MW) = Tfocus: TMW

(4)

Solving Eqs. (3) and (4), we will get Eqs. (1) and (2). To test whether a participant’s first self-caught session and the last self-caught session were generally consistent over time, the first selfcaught session and the last self-caught session for each participant were compared. Of the 98 participants who had data from the first selfcaught session (ranging from 6.53 s to 353.94 s (M = 50.12 s, SD = 45.50 s)) and the second self-caught session (ranging from 6.91 s to 304.46 s (M = 55.27 s, SD = 51.20 s)), there was a strong significant correlation between the average duration of the self-caught session one and the self-caught session two, r(96) = .78, p < .001. Because there were a relatively large number of participants excluded from analysis, mostly due to incomplete data, the average OSPAN score of included participants was compared with the average OSPAN score of excluded participants to determine whether we have differentially excluded certain participants. A two-sample t-test showed that there was not a significant difference in OSPAN scores for included participants (M = 49.66, SD = 14.20) and OSPAN scores for excluded participants (M = 49.33, SD = 15.16), t(1 0 1) = .11, p = .91, d = 0.02, 95% CI [−5.55, 6.19].

4.2. Results and discussion Overall forty-seven participants were excluded from data analysis. Twelve participants (10.8%) were excluded from data analysis because they had meditated for more than 10 h, eight participants (7.2%) were excluded from data analysis because the OSPAN score was not properly collected, twenty-two participants (19.8%) were excluded because of incomplete/inaccurate data3 (none or less than 32 data points for the probe session, only one self-caught session, etc.), and five participants (4.5%) were excluded because there was a large discrepancy between their self-caught 1 and self-caught 2 data (more than one minute).4 Overall, data from sixty-four participants were analyzed for the current study.

4.2.2. Linear regression analysis As illustrated in Fig. 3a, there was not a significant relationship between the average duration of self-caught focus-mind wandering episodes calculated from reports using only the self-caught sessions (Tself) and OSPAN score, r = .16, t(62) = 1.26, p = 0.21 with an R2 of .03, F(1, 62) = 1.60. This result replicated that of Experiment 1a. As illustrated in Fig. 3b, there was a marginally significant relationship between proportion of mind wandering probes and OSPAN score, r = −.24, t(62) = −1.96, p = .054 with an R2 of .06, F(1, 62) = 3.86. This result was similar to that of Experiment 1b. Of most importance to the current study, to analyze the relationship between working memory capacity and different states within a focusmind wandering episode, two linear regressions were performed separately: time of focus (Tfocus) as a function of OSPAN score, and time of mind wandering (TMW) as a function of OSPAN score. As illustrated in Fig. 3c ((Tfocus) as a function of OSPAN score), there was a significant positive relationship between mean time within the focus state (focusing on their breath) and OSPAN score, r = .27, t(62) = 2.20, p = .03 with an R2 of .07, F(1, 62) = 4.83. As OSPAN scores increased, so did the time within the focus state. As illustrated in Fig. 3d (time in a mind wandering state (TMW) as a function of OSPAN score), there was not a significant relationship

4.2.1. Calculating time of focus and time of mind wandering The duration for the time of focusing on a task (Tfocus) and the time of mind wandering (TMW) were calculated by combining both the selfcaught sessions and the probe caught sessions using the following formulas: 3 The regression results remained the same when these participants were included (for Tself and OSPAN: r = .17, t(80) = 1.52, p = .13; for p(MW) and OSPAN: r = −.20, t (80) = −1.85, p = .07; for Tfocus and OSPAN: r = .25, t(80) = 2.27, p = .03; for TMW and OSPAN: r = .07, t(80) = .66, p = .51). 4 Discrepancies larger than 1 min were more than 2 standard deviations from the mean and were in the long tail of the distribution.

208

Cognition 179 (2018) 202–212

M.J. Voss et al.

Fig. 3. Results of Experiment 2. (a) The scatter plot along with the linear regression line of duration of a full focus-mind wandering episode (including both Tfocus and TMW), calculated from the self-caught sessions alone (Tself), as a function of OSPAN score. (b) Scatter plot along with the linear regression line of the proportion of time in the mind wandering state (mind wandering proportion or p(MW)) overall, calculated from the probe-caught session alone, as a function of OSPAN score. (c) The scatter plot along with the linear regression line, of the mean time in the focus state (Tfocus) during a focus-mind wandering episode, calculated from both the selfcaught and probe-caught sessions, as a function of OSPAN score. (d) The scatter plot along with the linear regression line, of the mean time in the mind wandering state (TMW) during a focus-mind wandering episode, calculated from both the self-caught and probe-caught sessions, as a function of OSPAN score.

memory capacity. Two traditional measures of mind wandering tendency (mean duration/frequency of self-caught mind wandering and the proportion of mind wandering responses when probed) were estimated as a function of working memory capacity (OSPAN score). In both experiments, there was no significant correlation between working memory capacity and the mean duration of the self-caught focus-mind wandering episode. This result replicated that of previous research (Levinson et al., 2012). However, there was a negative correlation between working memory capacity and proportion of mind wandering responses in a probe-caught method (significant for Experiment 1b, marginally significant for Experiment 2). This finding was in contradiction to that of Levinson et al., 2012, but broadly consistent with other work in the literature (McVay & Kane, 2009; Randall et al., 2014). More importantly, Experiment 2 was conducted to estimate the mean time of focus (Tfocus) and mean time of mind wandering (TMW) separately using the raw self-caught duration and the proportion of mind wandering measured in the same participants. The results showed a significant positive relationship between working memory capacity and time of focus (Tfocus). These results suggest that individuals with higher working memory capacities were better able to block out mind

between time of mind wandering and OSPAN score, r = .01, t (62) = 0.05, p = .96 with an R2 of .00, F(1, 62) = 0.003. Overall, the results of mean time of focus (Tfocus) and mean time of mind wandering (TMW) were of most importance for the second experiment. Results showed that Tfocus increased with greater working memory capacity, suggesting that individuals with higher working memory capacity were able to focus on the current task longer. Results did not show a significant relationship between TMW and working memory capacity, suggesting that the ability to sustain the mind wandering state and/or detect the mind wandering state was not affected by working memory capacity. 5. General discussion The aim of the current experiments was to develop a more informative method to measure mind wandering, by calculating the mean time focused on the current task at hand (measuring the ability to stay on task and resist mind wandering (Tfocus)) and mean time in a mind wandering state (measuring processes for the continuation of mind wandering and/or ability to catch mind wandering (TMW)), and how these new measures relate to executive control as measured by working 209

Cognition 179 (2018) 202–212

M.J. Voss et al.

et al., 2012; Randall et al., 2014; Smallwood & Andrews-Hanna, 2013; Smallwood et al., 2013) could be partly due to issues related to the inappropriate probe sampling method, in addition to other moderating factors such as task demand. Allowing some variations of the delay to make “random” or “unpredictable” probes (e.g., instead of probing exactly every 3 min, one may probe every 2–4 min) may alleviate the issue to some degree, however it’s still an inappropriate sample unless the variation window matches the focus-mind wandering episode duration, which becomes equivalent to our method. Increasing the interval between two probes may also alleviate the issue. Due to variations in Tself, when two consecutive probes are separated by more episodes in between, they become more independent of each other and the sampling may approximate a random sample. However, lengthening the interval between probes may substantially increase the experimental time/cost, and the traditional method is still theoretically inappropriate. Thus, although our sampling method is still not ideal (see more discussion on the limitations below), it’s theoretically more justified and improved than the typical method used in previous research, and should be adopted in mind wandering research using on-line probecaught methods instead of the traditional one. Although the current study divided a focus-mind wandering episode into only two main states, the same approach can be used to study mind wandering events in a more fine grain scale by using more fine-tuned questions in the probe tests. For example, instead of asking participants to make a two-alternative forced choice between “on-task” vs “mind wandering,” one can include intermediate states such as partially ontask/partially mind wandering (Mrazek et al., 2012). Using the finer scale probes, one may divide the states of a focus-mind wandering episode into more detailed sub-components, such as including a substate of gradually entering the focus state at the beginning, a sub-state of already being aware of mind wandering but having not terminated the task-unrelated thoughts yet, and so on. Our method can in principle be adopted to estimate the durations of these finer grain subcomponents of the focus-mind wandering episode separately. In addition, our method may also be used to examine the effects of other factors, such as to differentiate spontaneous mind wandering vs. intentional mind wandering (Seli, Smallwood et al., 2015; Seli, Risko, & Smilek, 2016; etc.), since the time course of the focus-mind wandering episode is likely to be different for spontaneous and intentional mind wandering.

wandering to focus on the current task at hand longer. Moreover, working memory capacity did not have a significant effect on the time of mind wandering (TMW) in a focus-mind wandering episode, suggesting that once mind wandering occurs, it will last a comparable amount of time regardless of the working memory capacity. 5.1. Implications on the methodology of mind wandering research These findings have important implications both on theories of mind wandering and on the methodologies used to study mind wandering. There are three main implications regarding measurements of mind wandering. First, a careful examination of the time course of a focus-mind wandering episode revealed that the traditional concept of “tendency to mind wander” is not self-coherent. That is, measurements of how often people mind wander based on the self-caught method and the probe-caught method can be intrinsically contradictory and do not have a consistent interpretation. Therefore despite the intuitive appeal, researchers studying mind wandering should restrain from characterizing mind wandering by the vague notion of “how often people mind wander”. Instead, characteristics of mind wandering may be better described by measuring the durations of its sub-components separately, such as Tfocus and TMW. Second, we developed a new method to estimate the duration of the two states of a focus-mind wandering episode, which was a key distinction proposed in the Process-Occurrence Framework (Smallwood, 2013). However, there has been no empirical method to measure them separately to date. Our method relies on the combination of the traditional self-caught and the probe-caught methods, where the self-caught method provides a measure of the length of the entire episode, while the probe-caught method provides a measure of the relative proportion of each of the components. The duration of each component can thus be calculated, providing the first empirical methodology to examine the time course of the sub-components of the focus-mind wandering episode separately. Third, we developed a new probe sampling scheme in the timing of probe placement. The probe placement in our probe caught session was distributed uniformly within one focus-mind wandering episode cycle. This method of probe distribution was adopted because it was a necessary requirement for our experimental logic, i.e., for Equation #4 to hold. This new method corrected a common error in the probe placements in previous research using the probe-caught method. The most common probing scheme used in the literature was to distribute the probes uniformly across an experimental session. For example, the participants may be probed every certain number of trials, or every certain amount of time during an experiment. Although this method appears intuitive and reasonable, it is actually not appropriate and most likely biased in general. For example, if during an experimental session, a probe is presented every ∼3 min, this is not a uniform distribution of probes across the focus-mind wandering episode, but a fixed delay probe. Because the task (and the focus-mind wandering episode) restarts after each probe, probing the participants every 3 min means the probe always occurs 3 min after the start of each episode. If a participant tends to remain on task for ∼2 min, followed by ∼3 min of mind wandering before she catches herself, then the probe will almost always catch her during the mind wandering state, and overestimate the true proportion of time she’s in a mind wandering state. Similarly, if a participant tends to have ∼4 min of task focus followed by ∼1 min of mind wandering, then the probe is most likely going to catch him during the focus state and again mis-represent the likelihood he’s in a mind wandering state. As a result, probes evenly distributed across an experimental session generally will not provide an appropriate assessment of the proportion of time a given participant is in the mind wandering state, and consequently could affect the estimation of its relation to other factors. For example, the apparent contradictory findings on the relationship between WMC and mind wandering in previous research (e.g., Levinson

5.2. Implications on theories of mind wandering The potentially different effects of working memory capacity on Tfocus versus TMW in a focus-mind wandering episode suggest that theories of mind wandering need to address these different states separately. As outlined in the Introduction, there is debate as to whether mind wandering instances consume executive control resources (Resource Hypothesis), or if mind wandering is a result of a failure in executive control to block out mind wandering instances (Executive Control Hypothesis). Each theory predicts different patterns of how mind wandering should behave in relationship to executive control, as measured by working memory capacity. For example, the Resource Hypothesis predicts that individuals with higher working memory capacity should report being in a mind wandering state more compared to individuals who have lower working memory capacity (Levinson et al., 2012; Rummel & Boywitt, 2014; Teasdale et al., 1995; Thomson et al., 2013; Smallwood & Schooler, 2006), while the Executive Control Hypothesis predicts that individuals with higher working memory capacity should report being in a mind wandering state less compared to individuals who have lower working memory capacity (McVay & Kane, 2010, 2012). Both theories rely heavily on the idea of how much mind wandering one should have. However, neither theory in its current form is based on a clear definition of what “more” mind wandering means. Our 210

Cognition 179 (2018) 202–212

M.J. Voss et al.

5.3. Limitations and future directions

analyses of the focus-mind wandering time course suggest that the concept of how much one mind wanders is intrinsically ambiguous. In fact, the intuitive idea of the “tendency to mind wander” can have at least three totally different meanings. First, the tendency to mind wander can be defined as the likelihood that mind wandering will occur after one has started focusing on a task for a certain amount of time. The mean duration one can stay on the current task before switching to mind wandering (i.e., Tfocus) provides an index for this likelihood. That is, the shorter the Tfocus, the higher one’s tendency/likelihood to mind wander. Second, the tendency to mind wander can be defined as the likelihood one will stay in a mind wandering state once it occurs. The mean duration of mind wandering in each episode (i.e., TMW) measures this type of tendency. That means the longer the TMW, the higher one’s tendency to mind wander. Third, the tendency to mind wander can also be defined as the total amount of time one ends up spending in mind wandering. The proportion of mind wandering in a probe-caught method provides a measure of this tendency. Among these three different definitions of mind wandering tendency, the proportion of mind wandering is the only type that has been empirically examined in the literature. However, the theoretical meaning of this type of tendency is the most vague and most difficult to interpret. As discussed in the Introduction, the proportion of time one spends on mind wandering is determined jointly by two independent quantities Tfocus and TMW, and does not have a clear correspondence to the underlying mechanism or processes. Therefore this measure is theoretically vague unless such a correspondence can be established. Because these theories of mind wandering and existing experimental work are all based on the ambiguous concept of “mind wandering tendency,” the experimental findings are difficult to interpret with regards to whether they support or reject these theories. As such, these theories themselves are difficult to test empirically until the conceptual vagueness of the above mentioned theories are clarified. The Process-Occurrence Framework, on the other hand, provides a theoretical account for the two states of the focus-mind wandering episode separately. According to the Process-Occurrence Framework (Smallwood, 2013), executive control could influence mind wandering directly by increasing focus on the task and thus inhibiting the initiation of mind wandering from occurring. Smallwood (2013) argued that attentional control would help participants stay on task, so that “under tasks demanding continual external focus attentional control can limit the occurrence of self-generated thought by ensuring the continuity of a task-relevant train of thought” (p. 529). Our finding that time of focus (Tfocus) positively correlates with working memory capacity is consistent with this prediction. Although working memory capacity affects one’s ability to maintain task focus, it does not show a reciprocal relationship to the duration of the mind-wandering state. There are at least two possible explanations for the independence of TMW from the working memory capacity, depending on what determines the termination of the mind wandering state. If the termination of a mind wandering state is due to the failure to sustain the internal thought processes (e.g., due to lack of resources), then these results suggest that processes that sustain one on the current task are different from those that buffer the internal train of thought. Alternatively, it is also possible that the termination of a mind wandering state is primarily due to a separate, self-monitoring system that determines one’s meta-awareness of the mind wandering event. Thus, a mind wandering event may be terminated even if there are sufficient resources to maintain its continuation. Under this scenario, our findings would suggest that the self-monitoring/meta-awareness system is not dependent on working memory capacity. Which of these hypotheses is true awaits future research. Regardless of the cause of the termination of a mind wandering event, our findings provide empirical evidence that theories of mind-wandering need to account for multiple component processes, as emphasized in the Process-Occurrence Framework (Smallwood, 2013).

Despite the improvements in our methodology, there are still some limitations in the method we adopted in this study. The current study probed individuals in a uniform manner based on a fixed number derived from the average of the first self-caught session. However, the actual duration of the focus-mind wandering episode varied from one episode to the next, therefore the uniform probe distribution is only an approximation and may not fully reflect the true proportion of Tfocus and TMW in the focus-mind wandering episode. Ideally, the probe distribution should match the individual episode durations. However, due to the unpredictability of the individual episode durations, it’s not possible to match all of them in practice. Whatever size of the window within which we choose to distribute the probes, we can only correctly estimate the proportion of mind wandering for focus-mind wandering episodes of that specific length, and estimation of all other events will have some errors. For example, if we distribute probes within a range longer than the mean episode duration, we can match focus-mind wandering episodes that happen to be at that particular length, however as a result it will not properly estimate those episodes equal to or shorter than the average duration. Therefore as a compromise we tailored the probe sampling distribution to the mean episode duration to maximize the number of episodes we can match properly. Whether there are alternative ways to determine the probe distribution to further improve the sampling accuracy remains a challenge for future research, for example, by varying the sampling window location (e.g., instead of sampling from a fixed range within [3 s, Tself + 3 s], one may sample from [x, Tself + x] with varying x), or by varying the sampling window size according to the variance of Tself, instead of using a fixed value of mean Tself. Whether these alternative methods improve the overall accuracy of the estimation awaits further investigation. Another consequence of our method is that by uniformly distributing the probes within a focus-mind wandering cycle, some of the probe delays were quite short. The frequent probe presentation could potentially interfere with typical task focus, especially for individuals who have a short focus-mind wandering episode. One way to ameliorate this limitation in future research is to move the time window for the uniform distribution of the probes past the average focus-mind wandering episode to increase the delay between each probing, especially for people with short Tself. In addition, the probe presentation was based on the average duration of the self-caught mind wandering measure for each individual. This means that each individual was probed at different times depending on their self-caught mind wandering measure. Whether the different sampling times can change individuals’ behavior is a topic for future research. It is also important to stress that the relationship between working memory capacity and the durations of the two states of a focus-mind wandering episode found here is specific to the mindfulness of breath task implemented. It has been well established that people’s mind wandering can be affected by various factors, including the demand of the current task (Levinson et al., 2012; Randall et al., 2014; Smallwood & Andrews-Hanna, 2013; Smallwood et al., 2013). Because the main task for the current study was a mindfulness breath meditation task, which is a low demanding task, the relationships found here may only apply to low demanding tasks such as mindfulness meditation, and may not generalize to other tasks, specifically higher demand tasks that require more cognitive control, or tasks where motivation to perform the task is higher. In those cases the relationship might even show the reserve pattern. For example, the moderating effect of task demand has been proposed to account for different results in the effects of working memory capacity on mind wandering in previous research. It was suggested that in lower demanding tasks, individuals who have higher working memory capacity might mind wander because of an excess of executive resources, but in tasks that require more attentional demands to perform, mind wandering can be a result of a failure of executive control, showing different patterns of results in regards to task difficulty 211

Cognition 179 (2018) 202–212

M.J. Voss et al.

(Levinson et al., 2012; Randall et al., 2014; Smallwood & AndrewsHanna, 2013; Smallwood et al., 2013). Further research should address the differences between these two states of mind wandering in relation to task difficulty. Our method relies on an implicit assumption that the only way to return to focus is through the self-awareness mechanism (self monitoring/meta-awareness system), where the participants realize that they are mind wandering, then voluntarily terminate the mind wandering process and return to the basic task, unless they are interrupted by an external probe. This is an assumption implied (but not required) in previous theories of mind wandering (e.g., Schooler et al., 2011; Smallwood, 2013). If individuals can return to focus without self awareness, and have multiple switches between the focus and mind wandering states during a single self-caught focus-mind wandering episode, then the current method may not measure these switches, and the estimations we obtained (Tfocus and TMW) do not indicate the mean duration of each individual focus (or mind wandering) state, but instead indicate the total duration of the multiple focus (or mind wandering) states within a self-caught focus-mind wandering episode. Whether there can be multiple switches without self-awareness between focus and mind wandering states within a single self-caught focus-mind wandering episode awaits future research. In summary, the current study developed a new methodology to estimate the timing of different states within a focus-mind wandering episode and measured how they are affected by working memory capacity. The results showed that in a low demanding task, people with higher working memory capacity tend to remain on the current task longer. However, once mind wandering occurred, it lasted for a comparable duration regardless of the working memory capacity. These findings suggest that existing theories of mind wandering and experimental work based on a vague, ambiguous concept of mind wandering tendency need revision to address different states and processes involved in mind wandering.

Christoff, K., Irving, Z. C., Fox, K. C. R., Spreng, R. N., & Andrews- Hanna, J. R. (2016). Mind-wandering as spontaneous thought: A dynamic framework. Nature Reviews Neuroscience, 17, 718–731. Feng, S., D’Mello, S., & Graesser, A. C. (2013). Mind wandering while reading easy and difficult texts. Psychonomic Bulletin & Review, 3, 586–592. Hasenkamp, W., Wilson-Mendenhall, C. D., Duncan, E., & Barsalou, L. W. (2012). Mind wandering and attention during focused meditation: A fine-grained temporal analysis of fluctuating cognitive states. Neuroimage, 59(1), 750–760. Helton, W. S., & Warm, J. S. (2008). Signal salience and the mindlessness theory of vigilance. Acta Psychologica, 129, 18–25. 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. Lutz, A., Slagter, H. A., Dunne, J. D., & Davidson, R. J. (2008). Attention regulation and monitoring in meditation. Trends Cognitive Science, 12(4), 163–169. Mason, M. F., Norton, M. I., Van Horn, J. D., Wegner, D. M., Grafton, S. T., & Macrae, C. N. (2007). Wandering minds: the default network and stimulus-independent thought. Science, 315, 393–395. McVay, J. C., & Kane, M. J. (2009). Conducting the train of thought: Working memory capacity, goal neglect, and mind wandering in an executive-control task. Journal of Experimental Psychology, Learning, Memory, and Cognition, 35(1), 196–204. 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. McVay, J. C., & Kane, M. J. (2012). Drifting from Slow to “d’oh!” Working memory capacity and mind wandering predict extreme reaction times and executive-control errors. Journal of Experimental Psychology, Learning, Memory, and Cognition, 38(3), 525–549. McVay, J. C., Kane, M. J., & Kwapil, T. R. (2009). Tracking the train of thought from the laboratory into everyday life: An experience-sampling study of mind wandering across controlled and ecological contexts. Psychonomic Bulletin & Review, 16(5), 857–863. Mrazek, M. D., Smallwood, J., & Schooler, J. W. (2012). Mindfulness and mind-wandering: Finding convergence through opposing constructs. Emotion, 12, 442–448. Phillips, N., Mills, C., D’Mello, S. K., & Risko, E. (2016). On the influence of re-reading on mind wandering. The Quarterly Journal of Experimental Psychology, 69(12), 2338–2357. Poerio, G. L., Totterdell, P., & Miles, E. (2013). Mind-wandering and negative mood: Does one thing really lead to another? Consciousness and Cognition, 22, 1412–1421. 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. 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. 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. Seli, P., Carriere, J. S., Levene, M., & Smilek, D. (2013). How few and far between? Examining the effects of probe rate on self-reported mind wandering. Frontiers in Psychology, 4, 430. Seli, P., Risko, E. F., & Smilek, D. (2016). On the necessity of distinguishing between unintentional and intentional mind wandering. Psychological Science, 27, 685–691. Seli, P., Jonker, T. R., Cheyne, J. A., Cortes, K., & Smilek, D. (2015). Can research participants comment authoritatively on the validity of their self-reports of mind wandering and task engagement? Journal of Experimental Psychology, 41, 703–709. Seli, P., Smallwood, J., Cheyne, J. A., & Smilek, D. (2015). On the relation of mind wandering and ADHD symptomatology. Psychonomic Bulletin & Review, 22, 629–636. Smallwood, J. (2013). Distinguishing how from why the mind wanders: A process-occurrence framework for self generated thought. Psychological Bulletin, 139, 519–535. 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. Smallwood, J., Brown, K. S., Baird, B., Mrazek, M. D., Franklin, M. S., & Schooler, J. W. (2012). Insulation for daydreams: A role for tonic norepinephrine in the facilitation of internally guided thought. PLoS ONE, 7, e33706. Smallwood, J. M., Fitzgerald, A., Miles, L. K., & Phillips, L. H. (2009). Shifting moods, wandering minds: Negative moods lead the mind to wander. Emotion, 9, 271–276. Smallwood, J., O’Connor, R. C., Sudbery, M. V., & Obonsawin, M. (2007). Mind-wandering and dysphoria. Cognition Emotion, 21, 816–842. Smallwood, J., Ruby, F. J., & Singer, T. (2013). Letting go of the present: Mind-wandering is associated with reduced delay discounting. Consciousness and Cognition, 22(1), 1–7. Smallwood, J., & Schooler, J. W. (2006). The restless mind. Psychological Bulletin, 132(6), 946–958. Smallwood, J., & Schooler, J. W. (2015). The science of mind wandering: Empirically navigating the stream of consciousness. Annual Review of Psychology, 66(1). Teasdale, J. D., Dritschell, B. H., Taylor, M. J., Proctor, L., Lloyd, C. A., Nimmo-Smith, I., et al. (1995). Stimulus-independent-thought depends upon central executive resources. Memory and Cognition, 28, 551–559. Thomson, D. R., Besner, D., & Smilek, D. (2013). In pursuit of off-task thought: Mind wandering-performance trade-offs while reading aloud and color naming. Frontiers in Psychology, 4, 360. Unsworth, N., Heitz, R. P., Schrock, J. C., & Engle, R. W. (2005). An automated version of the operation span task. Behavior Research Methods, 37, 498–505.

Acknowledgement We would like to especially thank Mike Kane, Paul Seli, Jonathan Smallwood, Dan Simons, and David Irwin for discussion of the findings, and Sophia Dumlao, Dashiell Koester, Jackin Feng Sheh, Heidi Soon, and Faith Tan with help with data collection. This project was funded in part by the National Science Foundation grant #1519407. Comments should be sent to Meera Zukosky (email: [email protected]). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cognition.2018.05.013. References Antrobus, J. S. (1968). Information theory and stimulus-independent thought. British Journal of Psychology, 59, 423–430. Bastian, M., & Sackur, J. (2013). Mind wandering at the fingertips: Automatic parsing of subjective states based on response time variability. Frontiers in Psychology, 4, 573. Braboszcz, C., & Delorme, A. (2010). Lost in thoughts: Neural markers of low alertness during mind wandering. Neuroimage, 54, 3040–3047. Braboszcz, C., Hahusseau, S., & Delorme, A. (2010). Meditation and neuroscience: From basic research to clinical practice. In R. Carlstedt (Ed.), Integrative clinical psychology, psychiatry and behavioral medicine: Perspectives, practices and research (pp. 1910–1929). New York, NY: Springer Publishing. Brewer, J., Worhunsky, P., Gray, J., Tang, Y., Weber, J., & Kober, H. (2011). Meditation experience is associated with differences in default mode network activity and connectivity. Proceedings of the National Academy of Sciences, 20254–20259. Buckner, R. L., Andrews-Hanna, J. R., & Schacter, D. L. (2008). The brain’s default network: Anatomy, function, and relevance to disease. Annals of the New York Academy of Sciences, 1124, 1–38. Christoff, K., Gordon, A., Smallwood, J., Smith, R., & Schooler, J. (2009). Experience sampling during fMRI reveals default network and executive system contributions to mind wandering. Proceedings of the National Academy of Sciences, 8719–8724.

212