11614_Working Memory and Mindfulness in an RCT of ABBT and AR

University of Massachusetts Boston
University of Massachusetts Boston
ScholarWorks at UMass Boston
ScholarWorks at UMass Boston
Graduate Masters Theses
Doctoral Dissertations and Masters Theses
5-2019
Working Memory and Mindfulness in an RCT of ABBT and AR
Working Memory and Mindfulness in an RCT of ABBT and AR
Anna M. Hall
University of Massachusetts Boston
Follow this and additional works at: https://scholarworks.umb.edu/masters_theses
Part of the Clinical Psychology Commons
Recommended Citation
Recommended Citation
Hall, Anna M., “Working Memory and Mindfulness in an RCT of ABBT and AR” (2019). Graduate Masters
Theses. 547.
https://scholarworks.umb.edu/masters_theses/547
This Open Access Thesis is brought to you for free and open access by the Doctoral Dissertations and Masters
Theses at ScholarWorks at UMass Boston. It has been accepted for inclusion in Graduate Masters Theses by an
authorized administrator of ScholarWorks at UMass Boston. For more information, please contact
library.uasc@umb.edu.
WORKING MEMORY AND MINDFULNESS IN AN RCT OF ABBT AND AR

A Thesis Presented
by
ANNA M. HALL

Submitted to the Office of Graduate Studies,
University of Massachusetts Boston,
in partial fulfillment of the requirements for the degree of

MASTER OF ARTS
May 2019
Clinical Psychology Program

© 2019 by Anna M. Hall
All rights reserved

WORKING MEMORY AND MINDFULNESS IN AN RCT OF ABBT AND AR

A Thesis Presented

by

ANNA M. HALL

Approved as to style and content by:

___________________________________________________
Lizabeth Roemer, Professor
Chairperson of Committee

___________________________________________________
Sarah Hayes-Skelton, Associate Professor
Member

___________________________________________________
Alice Carter, Professor
Member

________________________________________________
David W. Pantalone, Program Director
Clinical Psychology Program

________________________________________________
Lizabeth Roemer, Chairperson
Psychology Department

iv

ABSTRACT

WORKING MEMORY AND MINDFULNESS IN AN RCT OF ABBT AND AR

May 2019

Anna M. Hall, B.A., Skidmore College
M.A., University of Massachusetts Boston

Directed by Professor Lizabeth Roemer

Working memory capacity (WMC) can be degraded by anxiety, stress, and worry,
but can also be protected by mindfulness interventions (Jha et al., 2010). The current
study was the first to investigate the relations between WMC, anxiety, and mindfulness
within two interventions for Generalized Anxiety Disorder (GAD) that promote
mindfulness: Acceptance Based Behavioral Therapy (ABBT) and Applied Relaxation
(AR). In this exploratory study, we analyzed a subset of participants from a RCT of
ABBT and AR who had completed the Operation Span Task (OSPAN; n = 21). First, we
found that pre- to post-treatment measures of WMC (e.g., OSPAN scores) did not
v

significantly increase due to time or condition, nor was there a significant interaction
effect, although the interaction was associated with a medium effect size: for the
between-group variable of treatment condition, F(1,19) = .40, p = .54, η 2 = .02; for the
repeated measure of time, F(1,19) = .14, p = .71, η 2 = .007; and for the interaction,
F(1,19) = .97, p = .34, η 2 = .05. Second, we found that increases in WMC were not
significantly related to reductions in anxiety; however, medium effect sizes correlating
WMC to several anxiety measures (i.e., GAD CS, r = -.38, HAM A, r = -.35, and DASS
Anxiety, r = -.32) are notable. Third, we found no significant relations and small effect
sizes between changes in mindfulness and changes in WMC, r’s = .05 to -.19. Fourth,
contrasting with findings in previous literature, a medium non-significant negative
correlation, r = -.32, suggested that practicing therapy skills (as operationalized currently)
might be related to less improvement in WMC. Important limitations include the small
sample and absence of repeated measures of WMC over the course of treatment, which
preclude analyses of temporal precedence of changes needed to determine directionality
of relations. Research with larger sample sizes is needed to further explore the relations
between WMC and mindfulness in anxiety treatments, as well as more thorough
assessment of practice to determine its role in therapeutic change.

vi

ACKNOWLEDGMENTS

To my mentor, Liz Roemer, thank you for your support, mindful guidance, and
generosity of kindness. You are truly a full-service mentor and your commitment to
mentoring shows. To my husband, Jensen Ying, thank you for walking along side me:
encouraging, comforting, cheering, playing, praying, abiding. Thank you for being with
me and for me. To my family, thank you for nurturing me, sacrificing for me, and the
immeasurable ways that you love me. To my friends, thank you checking in on me,
listening to me, and making me laugh. To my lab mates, thank you for your always kind
words and sage advice. And to my thesis committee, thank you for your time, energy, and
thoughtful feedback. I appreciate you all.

vii

TABLE OF CONTENTS

ACKNOWLEDGMENTS ……………………………………………………………………………………. vi
LIST OF TABLES
…………………………………………………………………………………………….. viii
LIST OF FIGURES …………………………………………………………………………………………….. ix

CHAPTER

Page

1. SPECIFIC AIMS AND HYPOTHESES ……………………………………………………………….1

2. BACKGROUND AND SIGNIFICANCE
……………………………………………………………..4
Working Memory Capacity
……………………………………………………………………….4
Working Memory Capacity & Anxiety
……………………………………………………..10
Working Memory Capacity & Mindfulness……………………………………………….17
Acceptance Based Behavioral Therapy and Applied Relaxation
…………………..22
The Current Study ………………………………………………………………………………….25

3. METHOD ……………………………………………………………………………………………………….27
Sample
………………………………………………………………………………………………….27
Measures and Tasks ……………………………………………………………………………….30
Primary Outcome Measures ………………………………………………………………30
Additional Measures
…………………………………………………………………………32
Automated Operation Span Task
………………………………………………………..33
Procedure
………………………………………………………………………………………………36

4. RESULTS ……………………………………………………………………………………………………….40
Sample Normality and Equivalence
………………………………………………………….40
Hypothesis 1
………………………………………………………………………………………….41
Hypothesis 2
………………………………………………………………………………………….43
Hypothesis 3
………………………………………………………………………………………….44
Hypothesis 4
………………………………………………………………………………………….46

5. DISCUSSION ………………………………………………………………………………………………….47

REFERENCES ……………………………………………………………………………………………………55

viii

LIST OF TABLES

Table

Page

1. Demographic Characteristics of the Sample
…………………………….
29

2. Pre-Treatment Outcome Measures …………………………………………
36

3. Means and Standard Deviations of WMC Scores at

Pre-Treatment and Post-Treatment ………………………………….
42

4. Correlations of Residualized Gain Scores for WMC and

Anxiety Measures in Full Sample and Split by Treatment

Condition……………………………………………………………………..
44

5. Correlations of Residualized Gain Scores for WMC and

Mindfulness Subscales in Full Sample and Split by

Treatment Condition ……………………………………………………..
45

ix

LIST OF FIGURES

Figure

Page

1. Participant flow through the study
………………………………………….
39

2. Means of WMC Scores by Condition at Pre-Treatment

and Post-Treatment ……………………………………………………….
42

1

CHAPTER 1
SPECIFIC AIMS AND HYPOTHESES

Acceptance-Based Behavioral Therapy (ABBT) and Applied Relaxation (AR)
have been shown to be effective treatments for anxiety, particularly Generalized Anxiety
Disorder (GAD). A randomized controlled trial of these interventions indicated that 63.3
to 80.0% of clients receiving ABBT and 60.6 to 78.8% of clients receiving AR exhibited
significant clinical improvement across five calculations of change at posttreatment and
follow-up (Hayes-Skelton, Roemer, & Orsillo, 2013). While these interventions differ in
practice, it appears that both influence mindfulness. ABBT uses mindfulness to promote
awareness of a client’s internal experience (e.g., a worry cycle), as well as to alter the
client’s relationship with this experience through decentering (i.e., recognizing that the
thoughts may not be all-encompassing truth), acceptance, and self-compassion. Rather
than being held captive by distressing internal experiences, ABBT uses mindfulness to
empower clients to make intentional choices about how they would like to respond, such
as through valued action. While not explicitly teaching mindfulness, AR appears to also
implicitly cultivate mindfulness (Hayes-Skelton, Usmani, Lee, Roemer, & Orsillo, 2012).
AR teaches clients to notice muscle tension and early cues of anxiety, which appears to
encourage clients to increase awareness of their internal experience as well as altering
their relationship with their internal experience to be more decentered, accepting, and
2

self-compassionate. Both ABBT and AR, explicitly or implicitly, incorporate
mindfulness as a tool to treat anxiety.
However, more research is needed to explore potential underlying mechanisms of
mindfulness as a tool to treat anxiety. For example, a cognitive factor called working
memory capacity (WMC) has been shown to be related to both mindfulness and anxiety
(e.g., Jha et al., 2010; Sorg & Whitney, 1992). Briefly, WMC can be defined as
maintaining information as initially encoded despite distraction or interference. WMC is
mutable and the literature indicates that anxiety, worry, and stress degrade WMC. Yet,
mindfulness has been shown to both treat anxiety and enhance or protect WMC (Jha et
al., 2010). It is possible that practicing mindfulness skills in ABBT or AR may also
practice or strengthen WMC because clients are needing to maintain particular thoughts
amidst interference (e.g., strong emotion). Thus, more investigation is needed to elucidate
the relationships between anxiety, mindfulness, and working memory capacity.
The overarching goal of this study was to understand the role of working memory
in mindfulness and the therapeutic effects of Acceptance-Based Behavioral Therapy
(ABBT) and Applied Relaxation (AR) as interventions for GAD. Specifically, this study
addressed the following four aims:
We examined whether WMC improves following treatment in general and if the
improvement differs by treatment. Supported by previous research (e.g., Jha et al., 2010),
we hypothesize that post-treatment measures of WMC will significantly increase
from pre-treatment levels; however, this improvement will not differ by condition.
3

Secondly, we explored whether changes in working memory were correlated with
decreases in general anxiety symptoms. Specifically, we investigated whether increases
in WMC were related to reductions in anxiety.
Thirdly, we explored whether changes in mindfulness scores were related to
changes in WMC. We hypothesized that changes in mindfulness scores would be
related to changes in WMC regardless of condition.
Lastly, we explored whether percentage of time spent engaging in mindfulness
(for the ABBT condition) or noticing early cues of anxiety (for the AR condition) was
associated with changes in WMC within each condition. We hypothesized that time
spent practicing these skills would be significantly related to changes in WMC, and
that this relationship would not differ by condition.

4

CHAPTER 2
BACKGROUND AND SIGNIFICANCE

In this review, I investigate the state of the literature on working memory, its
relationship to anxiety, its relationship to mindfulness, and the relevance of working
memory in anxiety treatments. First, I define working memory capacity and how it has
been measured. Then I discuss the mutability of WMC, especially in relation to stress,
anxiety, and worry, including a potential bi-directional relationship between WMC and
anxiety. Next, I review potential anxiety interventions in which WMC may be relevant or
enhanced, such as Cognitive Behavioral Therapy (CBT) and mindfulness. Finally, I
review how ABBT and AR specifically, as interventions that incorporate explicit or
implicit mindfulness, treat anxiety and may enhance WMC.
Working Memory Capacity

Considering the frequent meta-cognitive thinking and skills that require cognitive
resources in therapy, working memory capacity may be a relevant construct to
investigate. First, we discuss working memory capacity (WMC). Consider an instance of
someone driving while talking on the phone via Bluetooth. The person on the other line
gives the driver a phone number to memorize immediately before the driver attempts to
change lanes in heavy traffic. Changing lanes serves as a significant distractor, making it
5

difficult for the driver to maintain the phone number in memory (e.g., through rehearsal).
The necessity for the driver to organize information and resist distraction illustrates
working memory. More specifically, working memory capacity (WMC) is the active
maintenance of information during simultaneous distraction, interference, and/or
processing for a short period of time (Conway et al., 2005; Kane & Engle, 2002).
Similarly, Jha and colleagues define WMC as the “capacity to selectively maintain and
manipulate goal-relevant information without getting distracted by irrelevant information
over short intervals” (Jha et al., 2010, p.55).

What does active maintenance of information entail? Active maintenance can
include domain-general executive attention and domain-specific storage and skills
(Conway et al., 2005). The use of domain-general executive attention or domain-specific
skills may differ based on the context, individual variables, or a combination of the two.
For example, playing chess can illustrate both types of maintenance. An amateur chess
player may rely mostly on domain-general executive attention to plan a few moves ahead
while also attending to the status of the board. On the other hand, an expert chess player
may rely more on domain-specific skills (e.g., learned strategies) with comparably less
reliance on domain-general executive attention.

Domain-general executive attention is not only characterized by a non-specific
domain, but also the maintenance of memory representations in the face of interference.
Interference creates opportunities for error, necessitating the active maintenance of
correct information. In the driving example, changing lanes serves as the interference
while the driver attempts to maintain the phone number in memory. When interference is
6

not present, then these memory representations could be drawn from short-term or long-
term memory, demonstrating how the interference component of domain-general
executive attention is so critical to the construct of WMC (Kane & Engle, 2002). Another
example of therapeutically relevant interference could be a situation that triggers strong
emotion or anxiety. An individual may be distracted by their strong emotions and less
able to maintain a plan of behaviors or responses they would want to carry out. Specific
strategies, such as mindfulness discussed below, may strengthen an individual’s ability to
resist the impact of distraction (e.g., strong emotion) as well as enhance working
memory.

Given the importance of interference and domain-general executive attention,
measures of WMC should capture these constructs. The literature indicates that working
memory span tasks (or complex span tasks) may best include these components by
interleaving the to-be-remembered target stimuli (e.g., digits or words) with an
interference or processing task (e.g., comprehending sentences or solving equations;
Conway et al., 2005; Daneman & Carpenter, 1980; Redick et al., 2012; Turner & Engle,
1989). For example, counting span, operation span, and reading span are widely used
complex span tasks (CSTs) with demonstrated reliability and validity (Conway et al.,
2005). Automated, computerized CSTs are also increasingly being used in research.
Because interference of rehearsal is an important feature of CSTs, automated CSTs can
more easily control the amount of time between stimulus presentations and prevent the
rehearsal that lends itself to short term memory capacity (STMC). To prepare participants
for this kind of task, automated CSTs follow the same basic structure of practice
7

conditions: 1) storage only task, 2) processing only task, and 3) storage and processing
tasks interleaved (Redick et al., 2012). For example, in the automated operation span task
(OSPAN), participants practice: 1) recalling random letter strings only, 2) solving basic
arithmetic problems only, and 3) recalling random letter strings interleaved with solving
arithmetic (Unsworth, Heitz, Schrock, & Engle, 2005). Automated CSTs are also useful
because they’ve shown little to no gender effects (Redick et al., 2012), despite research
with other WMC measures claiming male advantages driven by advantage in g (Lynn &
Irwing, 2008). Additionally, automated CSTs are useful in research because they have
been shown to have high test-retest reliability, high internal consistency, convergent and
discriminant construct validity, and criterion-related validity (Redick et al., 2012).

In addition to complex span tasks, researchers also use simple span tasks and
dynamic span tasks to measure working memory. However, some argue that simple span
tasks (e.g., digit span) measure short-term memory or brief storage and rehearsal (e.g.,
remembering a string of numbers without interference) and thus do not adequately
include components of maintaining information during interference or distraction
(Conway et al., 2005; Redick et al., 2012). In WMC and anxiety literature, some
researchers have also recently included dynamic span tasks, like the N-Back task (Moran,
2016; Shackman et al., 2006; Vytal, Cornwell, Arkin, & Grillon, 2012). In the N-Back
task, participants are given a series of items (e.g., letters), attempting to maintain the most
recent “n” items and identify when an item matches that of “n” items ago. For example, a
stream of letters includes: T L H C H O C Q L C K L H C Q T R R K C H R. If the “n”
is 3, the participant needs to identify the letters (marked in bold here) that match the letter
8

3 items before it. Some researchers argue that dynamic span tasks and complex span
tasks are not interchangeable measures of WMC (Redick & Lindsey, 2013). In a meta-
analysis, Redick and Lindsey found low correlations between the N-Back and both the
complex span (r = .20) and simple span (r = .25). They argue that dynamic tasks and
complex span tasks may measure different underlying processes. Further, Shipstead and
colleagues found that performance on dynamic span tasks was related to storage, while
performance on complex span tasks was related to that of attentional control (Shipstead,
Lindsey, Marshall, & Engle, 2014). Thus, it appears that simple span and dynamic span
tasks may not be appropriate tasks to measure working memory capacity.

Further, because the distinguishing feature of WMC is the domain-general
executive attention, some researchers argue that findings should be consistent across
perceptual domains (e.g., verbal or spatial). Research using complex span tasks supports
the claim that they capture the domain-general executive attention component of WMC
regardless of presentation modality. A latent-variable study compared the domain-general
or domain-specific qualities of working memory and short-term memory tasks (Kane et
al., 2004). They investigated several measures, including verbal WMC, visuospatial
WMC, verbal short-term memory capacity (STMC), visuospatial STMC, verbal and
spatial reasoning, and general fluid (Gf) intelligence. Confirmatory factor analyses and
structural equation models demonstrated that WMC tasks indeed reflected domain-
general qualities, strongly predicted Gf, and weakly predicted domain-specific reasoning.
On the other hand, STMC tasks reflected domain-specific qualities, weakly predicted Gf,
and strongly predicted domain-specific reasoning. Thus, a distinction between verbal or
9

spatial WMC tasks may not be necessary because individual differences on these tasks
should be driven by domain-general executive attention.
Considering these individual differences, Conway and colleagues (2005) argue
that variation in WMC illustrate both stable, normally distributed individual variation as
well as mutable, state-dependent variation. As an example of its mutability and relevance
to mental health, the literature suggests that stress and anxiety reduce WMC (as measured
by complex span tasks). Klein and Boals, for example, explored the relationship between
life stress, state anxiety, and WMC (Klein & Boals, 2001). Participants answered a
survey of various life events, their positive or negative impacts, and how recently they
occurred. Participants also completed the operation span task as well as measures of state
anxiety and self-report intrusive and avoidant thinking. They found that participants with
more life event stress occurring recently exhibited lower WMC. They also found that
negative life events were related to more intrusive thoughts. The authors argue that
cognitive representations of negative life events compete for cognitive resources, which
diminish WMC. Examining a similar relationship in another domain, Schmader and
Johns investigated the relationship between stress due to gender stereotype threat and
WMC (Schmader & Johns, 2003). In this study, gender was defined as men and women,
with no acknowledgment or assessment of nonbinary gender identity. To prime
stereotype threat, a male researcher described a working memory test as a reliable
measure of “quantitative capacity,” which may highlight “underlying gender differences
in quantitative capacity” to participants. A manipulation check asking participants to rate
their concern that their math ability would be judged based on their gender also indicated
10

that both men and women expressed concern. They found that women exhibited lower
OSPAN scores than men in the stereotype condition, which led the authors to argue that
the negative stereotype, in which women perform worse on math, interfered with their
WMC. Finally, the authors found that the working memory deficit mediated the effect of
stereotype threat on women’s math performance. Thus, it appears there is a relationship
between stress and WMC in which stress may reduce WMC.

Working Memory Capacity & Anxiety

Similar to the relationship with stress, the literature indicates a relationship
between WMC and anxiety. Briefly, researchers differ in how this relation between
WMC and anxiety should be studied, such as the emphasis on domain specificity (e.g.,
phonological vs. spatial) versus emphasis on domain generality as well as the direction of
causality (Moran, 2016). However, there appears to be agreement in two ways. First,
“anxiety” is broken down into worry and arousal (e.g., Andrews & Borkovec, 1988;
Heller & Nitschke, 1998; Hope & Izard, 1996). “Worry” is characterized by verbal
rumination about future negative events, and is a primary symptom of generalized
anxiety disorder (GAD). “Arousal” is characterized by physiological symptoms (e.g.,
dizziness, sweating, increased pulse) and hypervigilance, and is a primary symptom of
panic (Watson et al., 1995). The second area of agreement is that the relation between
anxiety (both worry and arousal) and working memory capacity may involve interference
or competition with task-relevant resources, similar to the relationship with stress as
discussed above (Moran, 2016). For example, some claim that anxiety causes deficits of
11

cognition by competing with attention, phonological resources, or storage of memory
representations (e.g., Eysenck & Calvo, 1992; Eysenck, Derakshan, Santos, & Calvo,
2007; Robinson, Krimsky, & Grillon, 2013; Shackman et al., 2006). Others claim that
pre-existing cognitive deficits predispose individuals to anxiety, suggesting a possible bi-
directional relationship between WMC and anxiety (e.g., Mathews & MacLeod, 2005;
Ouimet, Gawronski, & Dozois, 2009). Even further, some claim that particular domains
of anxiety influence particular domains of WMC – specifically that arousal obstructs
spatial processes and worry obstructs phonological processes (Shackman et al., 2006).
A meta-analysis of 177 samples (N = 22,061 individuals) integrated some of these
competing theories (Moran, 2016). In general, they found a moderate but robust negative
association between self-report measures of anxiety (both worry and arousal) and
measures of working memory capacity (g = -.334, p < 10-29). They included varied types of anxiety presentations and working memory tasks. Further, they found this association to be true across complex span (e.g., OSPAN; g = -.342, k = 30, N = 3,196, p = .000001), simple span (e.g., digit span; g = -.318, k = 127, N = 17,547, p < 10-17), and dynamic span tasks (e.g., N-Back; g = -.437, k = 20, N = 1,318, p <.001), with largely comparable effect sizes, despite literature stating that simple span and dynamic span tasks likely capture different underlying constructs than those of complex span tasks. The authors also note that the results indicate that both domains of anxiety (i.e., worry and arousal) were associated with deficits in both domains of working memory measures (i.e., verbal and visuospatial). However, effects were more pronounced in measures of domain-general executive attention than domain-specific measures. 12 Despite the findings across different types of working memory tasks, the relative dearth of studies using complex span tasks is an important weakness of the state of literature on anxiety and WMC. Simple span measures are most frequently used in research on anxiety, perhaps because they are already included in most psychological evaluations (e.g., the Wechsler Intelligence Scale). As mentioned earlier, simple span tasks do not capture interference or domain-general executive attention, but rather, assess domain-specific short-term storage. Also as mentioned earlier, research indicates that dynamic tasks do not measure the same underlying construct as complex span tasks. Thus, anxiety research using simple span and dynamic span tasks must be interpreted carefully. As supported by Moran’s meta-analysis, anxiety research that does use complex span measures of WMC generally supports a relationship between the two. Studies inducing anxiety have found that it reduces WMC. For example, Sorg and Whitney conducted a study in which individuals of high or low trait anxiety were exposed to 10 minutes of competitive video games, simulating a stressful environment, and then completed both simple and complex span tasks (Sorg & Whitney, 1992). They found no differences on the simple span task; however, higher trait-anxiety individuals in the stress condition performed worse on the complex span tasks than those of low trait-anxiety. Additionally, high trait-anxiety individuals performed better than those of low trait- anxiety in the non-stress condition. Therefore, the authors argue that WMC deficits arise when individuals predisposed to anxiety experience stress. Further, Shi, Gao, and Zhou, explored a similar relationship with test anxiety (Shi, Gao, & Zhou, 2014). Their sample 13 included 53 Chinese undergraduates with high test anxiety and 58 with low test anxiety. They administered a measure of state anxiety and a modified complex Reading Span task intended to induce test anxiety, in which participants were required to remember letters serially presented, interleaved with directions to identify pseudo-words in sentence stimuli. In this case, the sentence stimuli were either neutral facts or related to test anxiety (e.g., I feel my heart beating very fast during important tests). The authors termed performance on test anxiety stimuli, emotional WMC. They found that individuals with high test anxiety performed worse on emotional WMC than neutral WMC, indicating that anxiety interfered with performance and reduced WMC. Thus, research supports the mutability of WMC, in which anxiety reduces it. In addition to the negative effects of anxiety on WMC, researchers have found that pre-existing high WMC may bolster or protect against the effects of anxiety, indicating a bidirectional relationship. For example, Johnson and Gronlund found this bidirectional relationship (2009). They conducted a study in which fifty undergraduate students completed the OSPAN, trait measures of anxiety, and a dual-task (combined short-term memory task and a tone-discrimination task) designed to induce performance anxiety. Notably, the authors did not actually measure state anxiety induced by this task. They found an interaction of trait anxiety and WMC, such that worse performance on the dual-task in those low in WMC indicated vulnerability to disruption or interference by anxiety, whereas those high in WMC performed better on the dual-task and were protected from the effect of anxiety. The authors note implications of the role of WMC in individuals predisposed to anxiety when completing anxiety-inducing tasks, such as 14 standardized testing. Thus, the bidirectional relationship is evident in that anxiety reduced WMC, but pre-condition high WMC attenuated the effects of anxiety. Similar to the buffering effect found by Johnson and Gronlund, another group investigated whether high WMC could attenuate the negative impact of trait anxiety on attentional control (Wright, Dobson, & Sears, 2014). High trait anxiety individuals performed worse than low trait anxiety individuals on attentional control (i.e., antisaccade task); however, when individuals were both high in trait anxiety and WMC, they performed better on a task of attentional control. Even further, high WMC and high trait anxiety individuals performed similarly to those with high WMC and low trait anxiety. Thus, the authors argue that higher WMC buffered against the effects of high trait anxiety on performance of this attentional control task. Interestingly, they found this interaction with the complex reading span task, but not the operation span task. The authors argue that OSPAN scores may have been affected by participants’ anxiety toward this task. Taken together, these findings suggest that high WMC may preserve subsequent performance, suggesting that a treatment that enhances WMC may in turn minimize the disruption of anxiety, leading to an iterative process of improvement. Researchers have also explored the relationship between WMC and worry. Bredemeier and Berenbaum explored the relationship between WMC and worry in GAD (Bredemeier & Berenbaum, 2013). A sample of 198 college students completed self- report measures of worry, a diagnostic interview for GAD, two N-Back tasks, and the OSPAN task. They found that 2-back scores (i.e., participants had to maintain 2 items of an N-Back task) had a negative relationship with worry and GAD symptoms. However, 15 OSPAN scores, which modestly correlated with 2-back scores (r = .18), were not significantly related. Of note, only 6 participants qualified for GAD diagnosis, and the self-report measures of worry appear to be within a normative level. It is possible that the association between worry and the N-Back, but lack of association with the OSPAN, is due to low clinical severity in the sample. In another study on worry, Sari, Koster, and Derkashan investigated whether active worrying could impair WMC (Sari, Koster, & Derakshan, 2017). A sample of 64 undergraduate students completed self-report measures of anxiety and worry, a pre-condition and post-condition change detection task (not a complex span task), and were assigned to complete a worry or non-worry control condition. The authors found that level of self-reported worry mediated the relationship between condition and changes in WMC, in which the indirect effect of worry on WMC changes was significant but the direct effect of condition on WMC was not significant. They found a similar relationship between state anxiety and WMC. However, again, research using different measures of WMC must be interpreted carefully. Thus, it appears that worry and anxiety may reduce WMC. The literature also lacks substantial research on the role of WMC in anxiety treatments. In fact, no studies were found that explored the effect of CBT interventions on WMC. Although a different construct than working memory, Mohlman suggests that the elderly may have lower response rates to psychotherapy treatments for anxiety and depression: they argue that as executive functioning (EF) diminishes with age, the elderly may have fewer resources to reason or regulate emotion (Mohlman, 2005). In the only study found investigating the effect of CBT on executive function, Mohlman and Gorman