Don’t Tell Me What to Think: Comparing Self- and Other-Generated Distraction Methods for Controlling Intrusive Thinking

Abstract

Cognitive control is central to the phenomenon of intrusive thinking in obsessive-compulsive and related disorders. The current study tested how attempts at cognitive control are impacted by self- vs. other-generated distractor thoughts. Participants (N=1913) were randomly assigned to suppress or monitor an intrusive thought and also randomly assigned to: a) self-generate a distractor, b) receive a distractor, or c) receive no distractor guidance. Participants reported subsequent thought recurrences, perceived success and effort keeping the thought out of mind, and positive and negative affect during a one-minute thinking period and a one-minute monitoring period. During the first thinking period only, self-generated distractors resulted in greater perceived control (p<.001; during monitoring instructions only) relative to no guidance, and less effort (ps<.001) relative to both other conditions. Interestingly, self-generated distractors led to longer duration of recurrences relative to both other conditions (ps≤.007). Finally, there were no distractor differences in trajectories of positive and negative affect (ps>.10). These findings suggest that the source of distractors may inform when attempts to control intrusive thinking will be helpful versus harmful.

Introduction

Cognitive control, which is the ability to coordinate mental processes and behaviors corresponding to goals (Miller & Cohen, 2001), is increasingly recognized as relevant to the development and maintenance of emotion dysregulation and psychopathology (Hoorelbeke, Koster, Demeyer, Loeys, & Vanderhasselt, 2013; Mcteague, Goodkind, & Etkin, 2016; Ochsner & Gross, 2005). The importance of cognitive control has been highlighted in leading transdiagnostic frameworks for examining mechanisms underlying psychopathology, such as The National Institute of Mental Health’s Research Domain Criteria (RDoC; Cuthbert & Insel, 2013; Insel et al., 2010).

Cognitive control differences are particularly important in the context of obsessive-compulsive and related disorders (OCRDs; Salkovskis, 1989), in which maladaptive attempts to control intrusive thinking (e.g., thought suppression; Marcks & Woods, 2007; Smári & Hólmsteinsson, 2001) may cause and/or maintain the recurrent intrusive thinking at the heart of these disorders. For example, the tendency to use thought suppression may be related to higher symptom levels for OCD and other disorders (Belloch, Morillo, & Garcia-Soriano, 2009; Morillo, Belloch, & García-Soriano, 2007), and suppression attempts might contribute to the development of obsessive thinking (Clark & Purdon, 2016). An improved understanding of the methods used to try to control intrusive thinking may shed light on how cognitive control differences put individuals at risk for developing OCRDs.

Importantly, a more comprehensive evaluation of cognitive control attempts may also help clarify the impact of efforts to manage intrusive thinking across multiple cognitive and affective domains. Previous research has focused heavily on the frequency of intrusive thinking as a primary outcome, but this is a limited view of the many ways control attempts may be linked to OCRDs. Unhelpful cognitive control attempts may also impact the duration of intrusive thinking, perceived effort and success at controlling intrusive thinking, and positive and negative affect (Purdon, 2001; Purdon, Rowa, & Antony, 2005). Critically, these additional outcomes may also play a role in the development or maintenance of OCRDs and the many forms of psychopathology associated with intrusive thinking (Lambert, Hu, Magee, Beadel, & Teachman, 2014). For this reason, in the present study we assessed a variety of cognitive and affective outcomes associated with attempted cognitive control of intrusive thinking.

Cognitive Control and Thought Suppression

Much of the research connecting cognitive control and OCRDs has centered on attempts to suppress unwanted thinking, which implicate the cognitive control functions of suppression and monitoring. In the context of thought suppression (i.e., the process of attempting to put thoughts out of mind), suppression reflects a person’s effortful attempts to combat recurrences, often by initiating a search for distractors. In contrast, monitoring corresponds to the automatic, goal-driven process of identifying recurrences of intrusive thinking. Critically, suppression and monitoring of thought recurrences have been linked to increased activity in the dorsolateral prefrontal cortex and anterior cingulate cortex, respectively (Mitchell et al., 2007), neural regions central to current neurocognitive models of cognitive control (MacDonald, Cohen, Stenger, & Carter, 2000; Shenhav, Botvinick, & Cohen, 2013). The cognitive control subconstructs of suppression and monitoring are also delineated in ironic process theory (Wegner, 1997; Wegner, 1994), a seminal theory outlining the difficulties of thought suppression. First, attempted control of intrusive thinking involves a monitoring process, which automatically searches for new recurrences of the intrusive thought. Second, a simultaneous operating process attempts suppression through an effortful, intentional search for alternatives to the intrusive thought (Wegner, 1997; Wegner, 1994). The automatic search for the target word increases accessibility of the target word, resulting in a paradoxical rebound of unwanted thinking for experimentally-induced (Abramowitz, Tolin, & Street, 2001; Magee, Harden, & Teachman, 2012) as well as naturally-occurring (Salkovskis & Campbell, 1994) intrusive thinking. An important area of research, therefore, is to determine whether certain methods of responding to intrusive thinking are associated with differential OCRD-relevant cognitive and affective outcomes.

Type and Source of Distractor May Matter

Early literature on thought suppression raised the possibility that the way one distracts from intrusive thinking during suppression may lead to different thought recurrence outcomes. In particular, focused distraction, which involves using a single distractor repeatedly, appears to be relatively more effective than other methods at reducing the frequency of intrusive thinking (Wegner & Erber, 1992; Wegner, Schneider, Carter, & White, 1987). In contrast, unfocused distraction, which involves a wandering search through a series of distractors after each recurrence of intrusive thinking, has limited effectiveness (Wegner, 2011; Wegner & Schneider, 1989). Unfocused distraction is thought to be the default suppression method for many individuals (Wegner & Schneider, 1989), perhaps accounting for the relative ineffectiveness of many thought suppression attempts. These distraction findings are intriguing but are based on a small number of studies, often using thought suppression designs that have been criticized for features such as inadequate control groups and focusing only on frequency of intrusive thinking (Rassin, 2005). In the present study, we examine the impact of distraction methods on multiple outcomes related to attempts at cognitive control of intrusive thinking. Specifically, we compared the impact of instructing participants to use one of three cognitive control methods: focused distraction using a self-selected distractor, focused distraction using an other-generated distractor, or no particular guidance (i.e., no distraction instructions).

In line with the previous evidence, we expected the ‘best’ outcomes (defined in this case as less frequency, shorter duration, and higher positive and lower negative affect) would result from focused distraction using an other-generated distractor. This prediction follows from Najmi and Wegner's (2008) suggestion that distractors associated with an intrusive thought ironically prime that intrusive thought during the control attempt, making disengagement attempts more challenging (see Wenzlaff, Wegner, & Roper, 1988). Because using self-generated distractors may unintentionally prime multiple connections to the intrusive thought as a person searches for an effective distractor, these additional connections would be expected to reduce the strategy’s effectiveness relative to an other-generated distractor, which would presumably prime fewer personal connections to the intrusive thought (Hooper, Saunders, & McHugh, 2010).

Further, we examined the impact of different distraction methods when participants were randomly assigned to either monitor an intrusive thought (so any choice to attempt suppression was self-generated) or to suppress the thought (so the experimenter directs the person to initiate control attempts). As with self-generated distractors, it seems plausible that self-generated control attempts may cue more connections for the participant, on average, than when control attempts are externally initiated. This possibility is based on research of action authorship, or the ascribed agent of a particular action. Specifically, attributions of action authorship may stem from an automatic authorship processing system connected to the self, which can make self-attributions more accessible (Dijksterhuis, Preston, Wegner, & Aarts, 2008). When attention is focused on the self, individuals are more likely to take responsibility for events that do not have a clear cause (Dijksterhus et al., 2008; Duval & Wicklund, 1973). Thus, when control attempts are self-generated, a stronger sense of self-authorship may occur, and in turn, heighten automatic connections between self-concept and the to-be-suppressed thought, which has been demonstrated to result in poorer suppression outcomes (Renaud & McConnell, 2002). Taking this evidence together, the greater associations between the self-generated control efforts and the to-be-controlled intrusive thought would lead to the prediction that self-generated control attempts that occur during monitoring would be less effective and result in greater initial thought recurrence than experimenter-directed control attempts that occur under suppression instructions.

Cognitive and Affective Outcomes Following Attempts at Cognitive Control

Thought recurrence is a standard outcome to assess following cognitive control attempts, like thought suppression, but even this outcome raises interesting complexities. Much of the literature has treated the recurrence of intrusive thinking as a unitary construct, but recent research has pointed toward the notion that the frequency and duration of intrusive thinking recurrence do not always align, and may have unique ties to cognitive control. For example, Lambert and colleagues (2014) found that, for participants assigned initially to either a thought monitoring condition or a thought suppression condition, frequency and duration diverge: frequency of thought recurrences declined over time across both conditions, whereas for those in the suppression condition, duration remained stable and then increased, and for those in the monitoring condition, duration increased, then decreased, and then remained stable. Further corroborating the notion that duration and frequency are dissociated, Gorlin, Lambert, and Teachman (2016) assessed effort to control thoughts, and instructed participants to monitor or suppress thoughts. Efforts to control thoughts had opposite relationships with frequency vs. duration; greater effort predicted greater frequency but shorter duration of thought recurrences across both conditions. One interpretation of these studies’ findings is that frequency might reflect more automatic processes (i.e., the monitoring process), whereas duration might reflect more conscious, effortful processing (i.e., the operating/suppression process). Together, these findings highlight the importance of studying multiple outcomes tied to thought suppression, and suggest that frequency and duration outcomes might differ as a function of the distraction method used.

Variability in distraction method might also impact other important outcomes of suppression attempts via the way the distraction method influences subjective evaluations of cognitive control. Individuals vary in their judgments and attributions when intrusive thinking inevitably recurs during thought suppression attempts (e.g., Magee & Teachman, 2007, Purdon, 2004; Tolin, Abramowitz, Hamlin, Foa, & Syndoi, 2002). Past studies have found that participants who interpret intrusive thinking as indicative of problematic personality qualities or mental states experience increased anxiety and negative mood after unsuccessful thought suppression attempts (Purdon, 2001; Purdon et al., 2005). Further, appraisals that are self-blaming or ascribe significance to recurrences of intrusive thinking during thought suppression are associated with heightened negative affect (Corcoran & Woody, 2009) and account for the relation between obsessive-compulsive symptoms and elevated fear and guilt during control attempts (Magee & Teachman, 2007), whereas external attributions (i.e., “the thought returned because the task was silly”) do not.

Although speculative, attributions about thought recurrence may be influenced by whether individuals self-generate or are supplied a distractor. If individuals self-generate the distractor, they may feel a heightened sense of responsibility over future thought recurrences (e.g., Dijksterhus et al., 2008; Duval & Wicklund, 1973), and when the intrusive thinking recurs, exaggerate perceptions of failure and perceive themselves as having lower success controlling their intrusive thinking. Because of this heightened sense of responsibility, they may also experience poorer affect, and perceive and/or exert greater effort in trying to control their intrusive thinking. If instead they are supplied a distractor, they might attribute recurrences more externally (e.g., to the experimenter), bypassing the heightened sense of responsibility and consequent impact on perceived success at controlling intrusive thinking, perceived effort expended attempting control, and affect. In sum, individuals’ attributions for thought recurrences might be impacted by type of distraction method used, and these attributions might be associated with differences in a range of OCRD-relevant cognitive and affective outcomes.

As noted above, we tested the impact of the distraction methods across both monitoring and suppression instructions. Following similar logic as described above for attributions related to source of distractors, it is possible that individuals may perceive heightened responsibility for self-initiated (i.e., while following monitoring instructions) vs. experimenter-instructed suppression attempts. However, contrary to this notion, there is previous data suggesting minimal differences in attributions across suppression vs. monitoring instructions (Magee & Teachman, 2007). Nonetheless, including both suppression and monitoring instructions allowed us to test these competing possibilities.

Overview and Hypotheses

The present study tested whether OCRD-relevant outcomes—specifically, the frequency and duration of thought recurrences, perceived success of control attempts, reported suppression effort, and positive and negative affect—were impacted by self-generated distractors vs. other-generated distractors vs. receiving no distractor instructions, and by suppression vs. monitoring instructions. Participants were asked to write down an intrusive thought, specifically that a friend lost his/her wallet, and were then randomly assigned to either a suppression condition, in which they actively tried not to think about their friend losing his/her wallet, or a monitoring condition, in which all attempts to control this thought were self-initiated. They were then asked to select their own distractor thought, assigned to think about a red jeep, or given no specific distractor guidance (control condition). Recurrence was measured by assessing the frequency and duration of participants’ key presses during the suppression or monitoring thinking period, while perceptions of successful control, effort and affect were assessed with self-report questions following the thinking period.

We hypothesized that frequency and duration would be higher when using self-generated vs. other-generated distractors, based on theory and evidence that the activation of mental associations to the intrusive thought might undermine self-selected distractors to a greater extent than other-generated distractors (Ju & Lien, 2016; Najmi & Wegner, 2008, 2009). For perceived success controlling intrusive thinking, we expected that participants with self-generated distractors would perceive lower success controlling an intrusive thought because the generation of the distractors may lead them to have heightened perceptions of responsibility over control of the thought. We also hypothesized that self-generated distractors would lead to higher negative affect and lower positive affect vs. other-generated distractors based on evidence that negative attributions about intrusive thinking contribute to heightened thinking and negative affect (e.g., Magee & Teachman, 2007; Purdon, 2001), although we did not measure attributions in the current study. Finally, while the main focus of study was on distractors, we expected that during an initial thinking period, monitoring instructions would result in greater recurrence, and potentially lower perceived success, higher negative affect, and lower positive affect than suppression instructions due to heightened responsibility for self-initiated control attempts. We did not expect this pattern to persist over time given the well-established pattern of monitoring instructions being associated with greater initial recurrence than suppression instructions but lower longer-term recurrence (Abramowitz et al., 2001; Magee et al., 2012). We did not have specific hypotheses about the interaction between the distractor conditions and suppression vs. monitoring instructions.

Method

Participants were 1913 U.S. citizens (age M=42.66, SD=16.29, range=18–82, 66.6% women, 79.0% white) recruited using online data collection. Most participants accessed the study through the publicly available Project Implicit website (http://implicit.harvard.edu/). Because these data were collected to examine questions concerning age differences (a separate focus than the current study), the selection rules were weighted so that participants would be relatively evenly distributed across the adult lifespan (i.e., similar number of participants in their 20s as in their 30s, etc.). In addition to direct access from Project Implicit, 15 participants were recruited through a Google advertisement, and 10 through Amazon’s Mechanical Turk. Table 1 shows the sample’s demographic and symptom characteristics according to distractor conditions and suppression vs. monitoring instructions.

Table 1.

Descriptive Statistics for Sample Demographics and Symptoms by Period 1 Distraction Condition and Suppression vs. Monitoring Instructions

	Thought Suppression			Monitoring
	Self-Gen.	Other- Gen.	None	Self-Gen.	Other- Gen.	None	Total
	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)	M (SD) or n (%)
Age	42.84 (15.92)	42.08 (16.34)	42.00 (16.58)	43.23 (16.17)	44.13 (16.58)	41.75 (16.10)	42.66 (16.29)
Sex (Female)	216 (70)	224 (67)	204 (66)	194 (64)	212 (66)	224 (68)	1274 (67)
Race
White	259 (84)	265 (80)	242 (78)	236 (78)	251 (80)	258 (79)	1511 (80)
Other	50 (16)	67 (20)	68 (22)	65 (22)	64 (20)	67 (21)	381 (20)
Ethnicity	10 (4)	27 (9)	22 (8)	21 (8)	25 (8)	23 (8)	128 (8)
Hispanic
Non-Hisp	265 (96)	263 (91)	266 (92)	238 (92)	270 (92)	274 (92)	1576 (92)
Anxiety	1.21 (1.03)	1.31 (1.14)	1.23 (1.10)	1.36 (1.17)	1.32 (1.10)	1.20 (1.06)	1.27 (1.10)
Total	310	333	311	306	321	332	1913

Note: All category comparisons above were non-significant at ps>.10. “Non-Hisp,” to Non-Hispanic, ‘Self-Gen.’ refers to self-generated distractors, ‘Other-Gen.’ refers to other-generated distractors, ‘None’ refers to no distractor guidance, and ‘Anxiety’ refers trait anxiety. Sample sizes for some variables may not add up to the overall sample size due to some participants choosing not to answer that demographic question.

Sample Attrition

Of the 3062 participants consenting to the study, 1913 (62%) provided an answer for every analyzed variable. The completers were significantly younger (t₍₃₀₆₀₎=2.18, p=.03, d=.08), and more likely to report female sex (χ²_{(1, N=3053)}=21.97, p<.001) and white race (χ²_{(1, N=3028)}=8.46, p=.004) than the non-completers. Completion rates did not differ by ethnicity (Wald χ²_{(1, N=2674)}=1.17, p=.28).

Materials

Positive and Negative Affect

Positive and negative affect were assessed using the 20-item Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988), which has good psychometric properties, and demonstrates validity and reliability in adult lifespan samples (Crawford & Henry, 2004; in the current study average Cronbach’s αs were .92 and .90 for the positive and negative subscales, respectively, averaged across three measurements).

Thought Stimulus and Thinking Instructions

During computerized instructions, participants were instructed to choose a close friend and told that they would think about this friend during the task. They were then presented with the sentence “I hope [name of friend] loses [her/his] wallet,” and asked to type this sentence while inserting the appropriate words. This thought is negative and unpleasant, similar to real-life intrusive thinking (Rachman, 1997), and is analogous to the ‘friend in a car accident’ thought commonly used in thought suppression studies (e.g., Rachman, Shafran, Mitchell, Trant, & Teachman, 1996).

Next, participants completed a 20-second orientation to the thought reporting methodology (referred to as Orientation). Participants were instructed to keep the wallet thought in mind continuously, and to press and hold the space bar whenever the thought was in mind. This practice period was followed by the first test thinking period (referred to as Period 1) in which participants received instructions combining elements from two independent randomizations (see Figure 1). In one randomization, participants received either thought monitoring or suppression instructions. In the monitoring instructions, participants were told: “You can think about anything. It can be the thought you focused on in the last thinking period “[lost wallet thought shown]” or it can be anything else. This period lasts 1 minute and the screen will again be yellow.” In the suppression instructions, participants were told “Your task now is to not think about the thought, “[lost wallet thought shown].” This period lasts 1 minute and the screen will again be yellow. You should try to avoid the thought for the entire time.”

Figure 1.

A visual depiction of participant flow through the study randomization points and thinking periods.

In the second randomization, also tied to Period 1, participants were provided with one of three types of distractor instructions: a specific distractor generated by the participant (self-generated distractor), a distractor thought suggested by the experimenter (other-generated distractor), or given no further guidance about how to conduct the thinking period. For the self-generated distractor, participants were told: “If you would like a new specific thought to think about, try selecting one of your own thoughts right now.” For the other-generated distractor, participants were told “If you would like a new specific thought to think about, try thinking of a red jeep.” Participants were then queried whether they had selected a thought and asked to type out the thought if they were willing. The instructions concluded with a reminder of the thinking and thought tracking instructions.

In a second test thinking period (referred to as Period 2), all participants were given identical instructions to monitor the wallet thought, with no further specifications. In both Periods 1 and 2, the screen was yellow and blank except for a small white 60-second countdown timer on the right of the screen. This timer was included to assure participants that the program had not stalled. Note that because all participants monitored the wallet thought during Period 2, references to ‘suppression instructions’ indicate a sequence of suppression then monitoring, whereas references to ‘monitoring instructions’ denote a sequence of monitoring then monitoring again.

Recurrence, Perceived Success Controlling Intrusive Thinking, and Effort

During Periods 1 and 2, participants recorded the recurrence of intrusive thinking by pressing and holding down the spacebar. This procedure yielded both frequency (number of key presses during a period) and duration (percentage of the period that the spacebar was pressed, rounded to the nearest percent) of thought recurrence. After each period, participants rated items measuring their perceived success controlling the intrusive thought and their effort keeping the intrusive thought out of mind. For perceived success, participants rated how successful they were at keeping the thought out of mind, how frequently the thought entered into their awareness (reverse scored), and how much of the time the thought was in their awareness (reverse scored). For effort, participants reported how motivated they were to keep the thought out of mind, how hard they tried to keep the thought out of mind, and how much energy they put into keeping the thought out of mind. All items were rated on a 5-point scale and responses to each set of three questions were averaged to create a score for that variable. Average Cronbach’s αs were .83 for effort and .85 for perceived success across Periods 1 and 2.

Trait Anxiety

A single item was used to characterize participants’ trait anxiety: “To what extent would you describe yourself as an anxious or ‘on edge’ person; that is, someone who has difficulty with fear, anxiety, panic or worry?” The item was rated on a 0 (“Not at all anxious”) to 4 (“Extremely anxious”) scale.

Procedure

All instructions were presented via the Internet. The consent page informed participants that the study would examine how thoughts and emotions can influence one another and that unpleasant thoughts would be involved. After consenting, participants first completed a baseline assessment of affect. They then completed the thought suppression paradigm consisting of three thinking periods: the 20-second Orientation and the 1-minute Periods 1 and 2. After Period 1, all participants rated how much effort they put into NOT thinking about the thought, how successful they were at keeping it out of mind, and rated their affect again. Following Period 2, participants again rated how much effort they expended, how successful they were at keeping the thought from mind, and a final affect measure. Participants then completed the trait anxiety item. All participants were then debriefed, and this information was emailed to anyone who did not finish the study.

Analytic Plan

For the outcomes concerning cognitive control of intrusive thinking that only had two measurements available (i.e., frequency, duration, perceived success, and effort), we used generalized estimating equations (GEE; Liang & Zeger, 1986). This approach permitted the modeling of correlated, repeated measurements across Periods 1 and 2 while flexibly accounting for different distributions of the dependent variables. For effect sizes, GEE produces an exponentiated beta value [Exp(b)] for continuous outcomes that is analogous to the way an odds ratio (OR) acts as an effect size for dichotomous outcomes. The Exp(b) value is interpreted as the multiplicative effect that a one-unit increase in the independent variable has on the outcome, after accounting for the other model variables. To analyze positive and negative affect, we used generalized multilevel models, because this approach used the three time points available to estimate both individual trajectories in affect (i.e., level 1), as well as the hypothesized, between-subject moderators of these trajectories: distractor instructions and suppression vs. monitoring instructions (i.e., level 2). All participants included in the current analyses provided complete data. Given the large sample size and the number of statistical tests performed, we set the alpha level at p<.01 for all statistical tests.

Results

Sample Characteristics

We first conducted a series of tests to examine possible demographic and symptom differences as a function of instructions (see Table 1). The suppression and monitoring instructions did not differ by age (t₍₁₉₁₁₎=.97, p=.33, d=.04), race (χ²_{(1, N=1892)}=.56, p=.46), ethnicity (χ²_{(1, N=1704)}=.87, p=.35), sex (χ²_{(1, N=1908)}=.56, p=.46), baseline negative affect (t₍₁₉₁₁₎=.63, p=.53, d=.03), baseline positive affect (t₍₁₉₁₁₎=.11, p=.91, d=.01), or trait anxiety (t₍₁₉₀₁₎=.73, p=.47, d=.03). Similarly, the three distractor conditions did not differ by age (F_(2,1910)=1.14, p=.32, η_p²=.001), race (χ²_{(2, N =1892)}=1.13, p=.57), ethnicity (χ²_{(2, N =1704)}=3.86, p=.14), sex (χ²_{(2, N=1908)}=.01, p=.99), baseline negative affect (F_(2,1910)=2.37, p=.09, η_p²=.002), baseline positive affect (F_(2,1910)=3.41, p=.03, η_p²=.004), or trait anxiety (F_(2,1900)=1.37, p=.25, η_p²=.001). Tables 2 and 3 display the means and standard deviations for dependent variables, as well as their intercorrelations.

Table 2.

Means and Standard Deviations for Dependent Variables by Period 1 Distraction Condition and Suppression vs. Monitoring Instructions

	Thought Suppression									Monitoring
	Self-Gen.			Other-Gen.			None			Self-Gen.			Other-Gen.			None			Total
	Base.	Per. 1	Per. 2	Base.	Per. 1	Per. 2	Base.	Per. 1	Per. 2	Base.	Per. 1	Per. 2	Base.	Per. 1	Per. 2:	Base.	Per. 1	Per. 2	Base.	Per. 1	Per. 2
	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)
Freq .	-	3.58 (3.35)	3.35 (9.50)	-	4.07 (6.41)	2.79 (2.73)	-	4.62 (7.01)	2.79 (3.07)	-	3.67 (5.16)	3.02 (2.69)	-	3.71 (3.53)	3.06 (4.19)	-	3.98 (3.78)	3.09 (3.61)	-	3.94 (5.08)	3.02 (4.88)
Dur.	-	31.79 (36.90)	29.60 (37.41)	-	30.47 (36.37)	24.11 (34.10)	-	23.30 (31.16)	20.15 (31.99)	-	48.84 (39.46)	36.90 (39.69)	-	39.64 (37.60)	31.90 (36.36)	-	35.66 (35.39)	23.82 (31.21)	-	34.89 (37.02)	27.66 (35.59)
Succ .	-	3.33 (.95)	3.58 (1.07)	-	3.38 (.94)	3.68 (.94)	-	3.31 (.91)	3.63 (.98)	-	3.38 (1.16)	3.49 (1.10)	-	3.20 (1.09)	3.38 (1.12)	-	3.01 (.99)	3.51 (1.03)	-	3.27 (1.02)	3.54 (1.05)
Eff.	-	3.02 (1.04)	2.45 (1.12)	-	3.33 (.98)	2.65 (1.12)	-	3.34 (.96)	2.43 (1.07)	-	2.27 (101)	2.28 (1.10)	-	(.98)	2.36 (1.04)	-	2.64 (1.03)	2.41 (1.08)	-	2.85 (1.08)	2.43 (1.09)
PA	29.32 (7.87)	27.90 (8.90)	26.74 (9.89)	30.48 (7.80)	29.49 (8.91)	28.48 (9.72)	29.20 (7.81)	28.12 (9.16)	26.83 (10.05	29.87 (7.72)	28.27 (8.90)	26.50 (9.99)	30.14 (8.04)	27.69 (9.26)	26.43 (10.07)	29.19 (7.72)	27.61 (9.10)	26.55 (10.08)	29.70 (7.85)	28.19 (9.05)	26.93 (9.98)
NA	13.61 (5.30)	13.97 (5.92)	13.22 (5.92)	14.83 (6.17)	14.94 (6.37)	14.06 (6.55)	13.98 (5.58)	14.01 (5.72)	12.87 (4.85)	14.55 (5.87)	14.92 (6.64)	14.25 (6.23)	13.97 (5.42)	14.74 (5.97)	14.06 (6.13)	13.52 (4.62)	14.36 (5.94)	13.55 (6.05)	14.08 (5.53)	14.49 (6.11)	13.67 (6.00)

Note: Freq. refers to frequency, Dur. refers to duration, Succ. refers to perceived success, Eff. refers to effort, PA refers to positive affect, NA refers to negative affect, ‘Self-Gen.’ refers to self-generated distractors, ‘Other-Gen.’ refers to other-generated distractors, ‘None’ refers to no distractor guidance, Base. refers to baseline, Per. 1 refers to the thinking period 1, and Per. 2 refers to thinking period 2.

Table 3.

Correlations among Dependent Variables

Frequency		Duration	Success	Effort	Positive	Negative
					Affect	Affect
Frequency	-	-	-	-	-	-
Duration	.36*	-	-	-	-	-
Success	−.47*	−.26*	-		-	-
Effort	.18*	−.05	−.23*	-	-	-
Positive Affect	−.10*	.02	.04	.18*	-	-
Negative Affect	.20*	.12*	−.20*	.08*	−.14*	-

Note:

^*Indicates significance at p < .01 or less. Each dependent variable was averaged across measurement points prior to computing correlations. Correlations show Spearman rank-order correlation coefficients, which were used instead of Pearson coefficients due to several variables (frequency, duration, and negative affect) following non-normal distributions.

Thought Suppression Manipulation Check and Use of Distractors

The suppression vs. monitoring instructions manipulation was effective, in that participants who received suppression instructions reported greater suppression effort than did those who received monitoring instructions during the randomization period (Period 1; t₍₁₉₁₁₎=16.38, p<.001, d=.75).

We also checked the extent to which individuals reported purposefully focusing on a different specific thought or image than the to-be-suppressed thought, although this retrospective rating encompassed both thinking periods and was not specific to the self- and other-generated distractors. Overall, participants reported use between ‘some of the time’ and ‘most of the time’ (M=2.38, SD=1.28). Participants who received suppression instructions during the randomization period reported greater use of a specific thought or image than did those who received monitoring instructions (F_(1,1907)=338.13, p=.003, η_p²=.99). Although the main effect of distractor did not reach significance (F_(2,1907)=22.31, p=.04, η_p²=.96), to explore potential distractor condition differences in use we conducted follow-up tests examining two distractor conditions at a time. Self-generated distractor instructions were associated with non-significantly higher use of specific thoughts/images relative to those given no specific distractor guidance (control condition, t₍₁₂₅₇₎=2.29, p=.02, d=.13). Self-generated distractor instructions were not associated with higher use of specific thoughts/images than other-generated distractor instructions (t_(1251.83)=1.09, p=.28, d=.06). Other-generated distractor instructions were not associated with higher use of specific thoughts/images relative to the control condition (t₍₁₂₉₅₎=1.26, p=.21, d=.07).

Impact of Distractor and Suppression/Monitor Instructions on Control of Intrusive Thinking

To examine our main research questions, we first used GEE to examine the effects of the distractor condition and suppression vs. monitoring instructions on four dependent variables following attempted cognitive control over intrusive thinking during the two test thinking periods: frequency and duration of actual thought recurrences, perceived success controlling intrusive thinking, and effort. Each GEE analysis included terms for distractor (self-generated distractor, other-generated distractor, and control condition), suppression vs. monitoring instructions, thinking period (Period 1 vs. Period 2), and the interactions among these variables (note, the 20-second Orientation was not included). Full model results can be seen in Table 4.

Table 4.

Generalized Estimating Equations for Measures of Attempted Control Over Intrusive Thinking

	Frequency				Duration				Perceived Success				Effort
	b	(SE)	Wald χ2	Exp (b)	b	(SE)	Wald χ2	Exp (b)	b	(SE)	Wald χ2	Exp (b)	b	(SE)	Wald χ2	Exp (b)
Intercept	1.38	(.05)	2274.16*	3.98	3.57	(.05)	20282.82*	35.66	3.01	(.05)	28699.68*	20.19	2.64	(.06)	16328.20*	14.04
Period	−.25	(.06)	47.87*	.78	−.40	(.06)	70.41*	.67	.50	(.06)	127.94*	1.65	−.23	(.05)	316.43*	.79
Distr.	-	-	.91	-	-	-	37.38*	-	-	-	2.84	-	-	-	20.74*	-
-Self-Gen.	−.08	(.09)	.74	.92	.31	(.07)	19.46	1.37	.38	(.09)	19.20	1.46	−.37	(.08)	21.35	.69
-Other-Gen.	−.07	(.07)	.90	.93	.11	(.08)	1.95	1.11	.20	(.08)	5.93	1.22	−.14	(.08)	3.01	.87
Sup./Mon.	.15	(.10)	.16	1.16	−.43	(.09)	38.50*	.65	.30	(.07)	15.58*	1.36	.69	(.08)	124.11*	2.00
Distr. X Sup./Mon.	-	-	.10	-	-	-	.23	-	-	-	5.63	-	-	-	4.13	-
-Self-Gen.	−.17	(.14)	1.56	.84	−.004	(.12)	.001	1.00	−.36	(.11)	9.91	.70	.06	(.11)	.27	1.06
-Other-Gen.	−.06	(.14)	.17	.94	.16	(.13)	1.68	1.18	−.13	(.11)	1.32	.88	.13	(.11)	1.48	1.14
Distr. X Per.	-	-	6.30*	-	-	-	2.13	-	-	-	16.77*	-	-	-	23.84*	-
-Self-Gen.	.06	(.10)	.31	1.06	.12	(.08)	2.54	1.13	−.39	(.09)	17.95	.68	.24	(.08)	9.37	1.27
-Other-Gen.	.06	(.09)	.45	1.06	.19	(.08)	5.56	1.20	−.33	(.09)	14.85	.72	.08	(.08)	1.18	1.09
Sup./Mon. X Per.	−.25	(.09)	1.76	.78	.26	(.11)	7.84*	1.30	−.18	(.08)	.39	.84	−.67	(.08)	156.63*	.51
Sup./Mon. X Distr. X Per.	-	-	3.50	-	-	-	5.21	-	-	-	9.76*	-	-	-	1.41	-
-Self-Gen.	.38	(.20)	3.49	1.46	−.05	(.14)	.13	.95	.33	(.12)	7.22	1.38	.09	(.12)	.58	1.09
-Other-Gen.	.07	(.14)	.24	1.07	−.28	(.14)	4.04	.76	.30	(.11)	6.89	1.35	.14	(.12)	1.37	1.14

Note: Bold values with asterisks indicate the significance of main or interaction effects at p<.05 or less. ‘Distr.’ refers to distraction instructions, ‘Sup./Mon.’ refers to suppression vs. monitoring instructions, ‘Per.’ refers to thinking period (Period 1 vs. Period 2), ‘Self-Gen.’ refers to self-generated distractors, and ‘Other-Gen.’ refers to other-generated distractors. For distractor-related effects, the Wald χ² and significance values are first reported provided for the overall effects. Beneath the overall effects, parameters are listed for the individual model effects comparing self-selected and provided distractors with no distractor guidance.

Recurrence: Frequency

There was a main effect of thinking period (b=−.25, SE=.06, Wald χ²=47.87, Exp(b)=.78, p<.001), with participants indicating greater frequency of thought recurrences during Period 1. Neither the distraction condition by thinking period interaction (Wald χ²=6.30, p=.04), nor any other main or interaction effect reached significance (all ps>.10), indicating that the effect of thinking period held across suppression vs. monitoring instructions and distraction conditions.

Recurrence: Duration

For duration of recurrences, there was a main effect of distractor (Wald χ²=37.38, p<.001). Pairwise comparisons indicated that participants in the self-generated distractor condition reported longer duration than either the other-generated distractor (p=.007) or the control (p<.001) conditions. There were also main effects of thinking period (b=−.40, SE=.06, Wald χ²=70.41, Exp(b)=.67, p<.001) and suppression vs. monitoring instructions (b=−.43, SE=.09, Wald χ²=38.50, Exp(b)=.65, p<.001), with these main effects being qualified by a suppression vs. monitoring instruction by thinking period interaction (b=.26, SE=.11, Wald χ²=7.84, Exp(b)=1.30, p=.005). Follow-up Generalized Linear Models (GLMs) examining each period individually indicated that for Period 1, individuals told to suppress (vs. monitor) reported shorter duration of the intrusive thought (b=−.43, SE=.08, Wald χ²=64.42, Exp(b)=.65, p<.001). During Period 2, individuals in the suppression instruction condition again reported shorter duration, although the relationship was weaker than during Period 1 (b=−.17, SE=.08, Wald χ²=22.78, Exp(b)=.85, p<.001). No other main or interaction effects reached significance (all ps>.05), indicating that the distractor effects held across thinking periods, as well as the suppression vs. monitoring randomization instructions.

Perceived Success Controlling Intrusive Thinking

For perceived success at keeping the intrusive thought out of mind, there were main effects of thinking period (b=.50, SE=.06, Wald χ²=127.94, Exp(b)=1.65, p<.001), and suppression vs. monitoring instructions (b=.30, SE=.07, Wald χ²=15.58, Exp(b)=1.36, p<.001), as well as an interaction between distractor condition and thinking period (Wald χ²=16.77, p<.001). However, all of these effects were qualified by a three-way distractor by suppression vs. monitoring instructions by thinking period interaction (Wald χ²=9.76, p=.008). To break down this interaction, we conducted follow-up GEEs examining suppression vs. monitoring instructions separately. For individuals randomly assigned to suppress during Period 1, there was no distractor by thinking period interaction (Wald χ²=.67, p=.71). For those randomly assigned to monitor during Period 1, there was a significant distractor by thinking period interaction (Wald χ²=23.38, p<.001). Follow-up GLMs examining one thinking period at a time revealed a significant effect of distractor condition for Period 1 (Wald χ²=19.26, p<.001) but not Period 2 (Wald χ²=2.73, p=.25). Pairwise comparisons were then conducted to evaluate distractor differences among participants assigned to the monitoring instruction during Period 1. Results indicated that self-generated distractors led to greater perceived success than the no distraction guidance control condition (p<.001) and non-significantly greater perceived success controlling the intrusive thought compared to other-generated distractors (p=.04). Other-generated distractors resulted in non-significantly greater perceived control than the no distraction guidance control condition (p<.02). No other main or interaction effects reached significance (all ps>.10).

Effort

There were main effects of distractor condition (Wald χ²=20.74, p<.001), thinking period (b=−.23, SE=.05, Wald χ²=316.43, Exp(b)=.79, p<.001), and suppression vs. monitoring instructions (b=.69, SE=.08, Wald χ²=124.11, Exp(b)=2.00, p<.001) on reported effort keeping the intrusive thought out of mind. Each of these main effects was qualified by an interaction. First, there was a distractor condition by thinking period interaction (Wald χ²=23.84, p<.001). Follow-up GLMs examining one thinking period at a time revealed a significant effect of distractor during Period 1 (Wald χ²=41.52, p<.001), but not during Period 2 (Wald χ²=5.42, p=.07). Pairwise comparisons indicated that participants in the self-generated distractor condition reported less effort expended during Period 1 than either the other-generated distractor (p<.001) or the control (p<.001) conditions, which did not differ from one another (p=.21). There was also a suppression vs. monitoring instruction by thinking period interaction (b=−.67, SE=.08, Wald χ²=156.63, Exp(b)=.51, p<.001). Follow-up GLMs examining one thinking period at a time revealed that suppression (vs. monitoring) instructions were associated with significantly greater effort during Period 1 (b=.69, SE=.08, Wald χ²=276.20, Exp(b)=2.00, p<.001), but the relationship between suppression instructions and greater effort weakened during Period 2 (b=.02, SE=.09, Wald χ²=10.76, Exp(b)=1.02, p=.001). No other main or interaction effects reached significance (all ps>.10).

Impact of Distractor and Suppression/Monitor Instructions on Affect

Next, we used multilevel modeling to examine distractor condition and suppression vs. monitoring instructions as independent variables associated with the two affective dependent variables. Both the positive affect and negative affect models explained individual variation in the linear trajectory of affect over time. Each analysis included terms for the main and interacting effects of time, the three distractor conditions (self-generated distractor, other-generated distractor, and control), and suppression vs. monitoring instructions. We initially included the three-way interaction between time, distractor condition, and suppression vs. monitoring instructions, but dropped it for parsimony because it did not near significance in any analysis and we did not have a priori hypotheses for this interaction. Full model results can be seen in Table 5.

Table 5.

Multilevel Models Examining Positive and Negative Affect

	Positive Affect			Negative Affect
	Estimate	(SE)	95% CI	Estimate	(SE)	95% CI
Fixed Effects
Intercept	3.26*	(.02)	3.23, 3.30	2.58*	(.02)	2.55, 2.61
Time	−.07*	(.006)	−.09, −.06	−.01*	(.006)	−.03, −.002
Time^2	-	-	-	−.01*	(.002)	−.02, −.01
Distr.	-	-	-	Sig.*	-	-
-Self-Gen.	.02	(.02)	−.03, .07	.05*	(.02)	.000, .09
-Other-Gen.	.02	(.02)	−.02, .07	.03	(.02)	−.02, .07
Sup./Mon.	.02	(.02)	−.02, .07	−.01	(.02)	−.06, .04
Dist. X Sup./Mon.	-	-	-	Sig.*	-	-
-Self-Gen.	−.03	(.03)	−.09, .03	−.05	(.03)	−.12, .01
-Other-Gen.	.02	(.03)	−.04, .08	.03	(.03)	−.04, .09
Dist. X Time	-	-	-	-	-	-
-Self-Gen.	−.008	(.008)	−.02, .007	.007	(.008)	−.009, .02
-Other-Gen.	−.004	(.008)	−.02, .01	.006	(.008)	−.008, .02
Sup./Mon. X Time	.02*	(.006)	.008, .03	−.02*	(.006)	−.03, −.009
Variance Components
Intercept	.10*	(.003)	.09, .10	.08*	(.003)	.08, .09
Slope	.01*	(.001)	.010, .012	.005*	(.001)	.004, .006
Residual	.01*	(.0004)	.014, .016	.18*	(.005)	.17, .19

Note: Bold values with an asterisk indicate significance at p<.05 or less. ‘Time’ refers to linear time, ‘Time^2’ to quadratic time, ‘Distr.’ refers to distraction instructions, ‘Self-Gen.’ refers to self-generated distractors, ‘Other-Gen.’ refers to other-generated distractors, and ‘Sup./Mon.’ refers to suppression vs. monitoring instructions. For distractor-related effects, ‘Sig.’ indicates that the overall effect was significant at p<.05. Beneath overall effects, parameters are listed for the individual model effects comparing self-selected and provided distractors with no distractor guidance. Results are presented in log units because they were examined with generalized multilevel models that used a log link function. Figure 1. A visual depiction of participant flow through the study randomization points and thinking periods.

For positive affect, the fixed linear term indicated that, on average, positive affect decreased across the time points (estimate=−.07, SE=.006, F_(1,5729)=490.79, p<.001). Neither distractor condition nor suppression vs. monitoring instructions were associated with higher intercepts for positive affect (ps>.10). For slopes, there was an interaction with linear time for suppression vs. monitoring instructions (estimate=.02, SE=.006, F_(1,5729)=10.96, p=.001). The positive coefficient indicated that suppression instructions were associated with a less steep decline in positive affect over time than monitoring instructions. No other main effects or interactions were significant (ps>.10).

For negative affect, the fixed linear term indicated that, interestingly, on average, negative affect also decreased across the three time points (estimate=−.01, SE=.006, F_(1,5728)=42.18, p<.001). However, the significant fixed quadratic term, which was centered upon the middle measurement, qualified this pattern (estimate=−.01, SE=.002, F_(1,5728)=92.10, p<.001). The negative coefficient, given the coding scheme and the linear trend, indicated a rise then fall of negative affect over time. Taken together, these patterns indicate an increase then decline for negative affect, with overall levels demonstrating a linear decrease from the start to the end of the study. There was no significant association between intercepts for negative affect and distractor condition, suppression vs. monitoring instructions, or their interaction (ps≥.04). For the key question of negative affect change, similar to positive affect, there was an interaction with linear time for suppression vs. monitoring instructions (estimate=−.02, SE=.006, F_(1,5728)=12.08, p<.001). This negative coefficient indicated that suppression instructions were associated with a less steep decline in negative affect. No other main effects or interactions were significant (ps>.05).

Together, suppression instructions were associated with a less steep decline for both positive and negative affect, suggesting a pattern in which randomization to suppression (vs. monitoring) during Period 1 preserved positive affect while hindering the decline of negative affect across thinking periods.

Discussion

This study tested the impact of self- vs. other-generated distractors (vs. a no distraction guidance control condition) and suppression vs. monitoring instructions on a variety of outcomes relevant to the cognitive control of intrusive thinking. Across several cognitive control outcomes (perceived success controlling intrusive thinking and perceived effort), there were advantages to personally selecting a distractor, particularly during an initial thinking period (i.e., Period 1) and for individuals instructed to monitor (vs. suppress) an intrusive thought. Despite these advantages, individuals assigned to self-generate a distractor also experienced longer duration of recurrences relative to the other two distractor conditions. This latter finding held across suppression vs. monitoring instructions and both thinking periods. Interestingly, the distractor conditions did not translate into differences in the trajectories of positive and negative affect, despite impacting several cognitive control outcomes. Together, results suggest that the source of a distractor differentially impacts a broad range of outcomes tied to the recurrence of intrusive thinking, and highlight the importance of assessing a variety of outcomes to investigate how efforts to control intrusive thinking may be helpful or harmful.

In interpreting these complicated findings, there are several notable patterns that emerge. First, contrary to expectations, the self-generated distractor generally appeared to be the most effective cognitive control method as reflected by greater perceived success controlling the intrusive thought and lower perceived effort, at least under some control attempt contexts. Although speculative, these unexpected findings could be explained in part by the characteristics of self-generated distractors. While we expected the other-generated ‘red jeep’ distractor would prime fewer existing mental associations with the intrusive thought than self-generated distractors, it is possible that self-generated distractors were more self-relevant and, consequently, more absorbing, making it easier to do focused (vs. unfocused) distraction. Relatedly, participants may have chosen distractors based upon their prior experiences managing intrusive thinking (Charles, 2010), favoring distractors with characteristics that had fared well during past control attempts. It would be interesting in future work to see how OCRD or other clinical symptoms influence the type of distractor that is self-selected. In controlled settings, individuals with OCD and other types of psychopathology surprisingly perform equivalently or better than non-clinical samples in terms of recurrence (Magee et al., 2012), suggesting that the selection of distractors among individuals with psychopathology would be beneficial to examine.

In terms of how self-generated distractors may provide benefits, perhaps they require less cognitive load to access and maintain in awareness, as compared to a distractor that is not relevant to the individual (i.e., a red jeep). If so, given cognitive resources are required to successfully avoid intrusive thinking (Barrett, Tugade, & Engle, 2004; Najmi, Riemann, & Wegner, 2009), using a self-generated distractor could have been a more effective cognitive control method insofar as it was less cognitively demanding. This interpretation is consistent with findings that cognitive control methods are less effective when people are engaged in mentally taxing tasks. For example, people under cognitive load show more Stroop task interference (Wegner & Erber, 1992) and greater recurrence of intrusive thinking during thought suppression attempts (Arndt, Greenberg, Solomon, Pyszczynski, & Simon, 1997; Newman, Duff, & Baumeister, 1997; Page, Locke, & Trio, 2005). Finally, self-generated distractors could have had greater ecological validity than other-generated distractors in the study. If other-generated distractors were perceived as being artificial or unlike the experience of everyday distractors, this experimental limitation could also be a limitation for clients receiving therapist-suggested distractors in clinical settings.

Notwithstanding this possible rationale, the advantages for the self-generated distractor were not universally apparent. For instance, self-generating a distractor led to greater perceived success controlling the intrusive thought during Period 1 only for participants assigned to monitor (vs. those assigned to suppress). We speculate that this may have to do with the attributions made by participants about their intrusive thinking, although we did not directly assess attributions. In formulating our initial hypotheses, we focused on the idea that self-generated distractors might lead to a heightened sense of responsibility and subsequent negative attributions when suppression attempts inevitably are not completely successful. However, it is possible that participants focused more on attributions for perceived successes (vs. failures). Thus, if someone generated their own distractor and they initiated the control attempts while monitoring, they may have been more likely to attribute successful suppression attempts to something about themselves. Consequently, they may have been more likely to perceive a greater sense of success in controlling the intrusive thought. In contrast, one may not have the same degree of ownership (and success attributions) if someone else told them to control their thought (i.e., if they were assigned to suppress), and/or if they did not generate their own distractor thought (i.e., if they received an other-generated distractor or received no guidance about control methods). Because we did not directly assess attributions, we could not determine whether attributions or other mechanisms accounted for the pattern of results. Again, examining this effect in OCRDs and other clinical areas while assessing possible mechanisms such as attributions would be interesting in future work.

Further, it is interesting that the benefits for the self-generated distractor were most pronounced during Period 1 (vs. Period 2), a time when the perceived amount of effort expended during control attempts was higher relative to Period 2. Thus, it could be that the corresponding differences as a function of distractor source might only matter during more intensive cognitive control efforts. This idea is consistent with previous evidence pointing to the initial thinking period as being particularly sensitive to variation in the intensity of cognitive control efforts; namely, both adding cognitive load and requiring longer suppression periods are associated with greater thought recurrence frequency, but only during the initial thinking period of control attempts (Abramowitz et al., 2001; Magee et al., 2012). Nevertheless, it is important to highlight that we anticipated that the effects of the initial distractor instructions would continue to carry through to Period 2, allowing us to empirically test whether the effects of distractor instructions extended beyond Period 1. However, following other thought suppression studies that include a Period 2 monitoring condition as an unstructured “control” (Abramowitz et al., 2001), distractor instructions were not reinstated prior to Period 2. Thus, it is also possible that individuals did not continue adhering to the instructions they were randomly assigned to during the first thinking period, thereby minimizing the impact of distractor instructions during Period 2.

Another striking pattern to emerge in the present study was the divergence between frequency vs. duration outcomes. Rather than conferring an advantage, individuals assigned to self-generate a distractor actually experienced longer durations of thought recurrences, on average, relative to the other two distractor conditions. This finding is consistent with prior work suggesting that greater self-reported suppression effort predicts shorter duration and greater frequency of intrusive thinking (Gorlin et al., 2016). Compatible with this finding, in the present study, the self-generated distractor condition was associated with relatively less suppression effort and longer duration of intrusive thinking, compared to either one or both of the other distraction conditions. As articulated previously, self-generating a distractor might have been relatively less cognitively demanding for participants, resulting in more available cognitive resources and less perceived effort expended controlling one’s thoughts. Expending less effort to suppress one’s thoughts may function as a double-edged sword, however (see Gorlin et al., 2016). On the one hand, by expending less effort to suppress one’s thoughts, an individual may be less motivated to engage in the types of conscious, effortful processes needed to override intrusive thinking once it has started, resulting in longer duration of intrusive thinking. On the other hand, expending less effort to suppress one’s intrusive thinking may also lead to less frequent activation of that same thinking. Future research employing a longitudinal design will be valuable to test the proposed time sequence and relations between suppression effort, duration, and frequency of intrusions.

Another interpretation of the duration results is that duration may index not only difficulties with disengaging from intrusive thinking, but also individuals’ subjective goals for control while processing their intrusive thinking. Although longer duration of intrusive thinking is typically considered a “maladaptive” outcome of cognitive control efforts, there are certain “adaptive” response methods that in the short term will promote longer durations of intrusive thinking. For instance, it is possible that individuals could be calmly meditating and processing an intrusive thought vs. doggedly trying to get the intrusive thought out of their mind and failing. Thus, although self-generated distractors are associated with longer duration of intrusive thinking, this is not necessarily problematic. Indeed, pointing to a less “alarming” picture of duration, self-generated distractors were also associated with lower perceived effort and greater perceived success controlling intrusive thinking during Period 1.

Finally, it should be noted that despite distractor condition differences in recurrence, perceived success, and effort controlling the intrusive thinking, the distractor conditions did not translate into differences in affect. Instead, only suppression vs. monitoring instructions led to distinct affect trajectories, as suppression instructions resulted in a less steep decline for both positive and negative affect compared to monitoring instructions. It may be that for affect, heightened efforts at control matter more than the manner in which one attempts control (i.e., the distractor conditions). Perhaps this occurs because of the links between suppression instructions and effort; that is, individuals following suppression (vs. monitoring) instructions reported greater suppression effort. In turn, perhaps this greater effort slowed the normative habituation process because the active suppression effort kept the success or failure of the control attempt salient for a longer period, reducing the decline for positive and negative affect.

Limitations and Conclusion

Although this study focused upon self- vs. other-generated distractors, there are a variety of other cognitive control methods that are used with intrusive thinking. Future research that samples a wide variety of cognitive control methods and assesses their consumption of cognitive resources will help to tease apart how cognitive control methods achieve their outcomes. In this study, participants self-reported the frequency and duration of intrusive thinking. It is likely that participants’ ability to attend to mental events (and report these outcomes) is itself related to elements of cognitive control (Schooler et al., 2011). Comparisons of self-report with methodologies such as experience sampling may further elucidate the role of meta-awareness in both the experience and measurement of attempts to control intrusive thinking. Further, it is important for future empirical work to more directly examine the similarities and differences between conceptualizations of cognitive control in RDoC vs. in thought suppression. The typical use of different paradigms in these investigations makes it difficult to draw direct comparisons. This study also used online data collection among a non-clinical sample. While online data collection facilitated a large sample size and greater participant anonymity while engaging with intrusive thinking, we were unable to control participants’ environments to the same degree as in laboratory settings. In this study, our confidence in implementing the particular paradigm was bolstered by having previously used it both online and offline, finding that the online version replicated laboratory-based findings (Magee, Smyth, & Teachman, 2014). Nonetheless, examination of cognitive control methods among clinical samples and across lab-based and everyday settings will allow investigation of the robustness and generalizability of the effects.

These limitations aside, the current findings demonstrate that distraction methods matter for exerting cognitive control over intrusive thinking, particularly during the initial, more intensive parts of control attempts. For several important markers of cognitive control, it appears that if one wants a distractor done well, the job is best done oneself.

Highlights.

• Compared self- vs. other-generated distractors for controlling intrusive thoughts.

• Self-generated distractors led to greater perceived successful control of thoughts.

• Self-generated distractors led to less effort controlling intrusive thoughts.

• Self-generated distractors led to longer intrusive thought duration.