|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
INTERFERENCE EFFECTS OF ANXIETY AND AFFECTIVE PROCESSING ON
WORKING MEMORY: BEHAVIORAL AND ELECTROPHYSIOLOGICAL
DAVID ANDREW STIGGE-KAUFMAN
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
DAVID ANDREW STIGGE-KAUFMAN
I thank my advisor, William M. Perlstein, for his excellent guidance and support
during this project. I also thank the other members of the Clinical-Cognitive
Neuroscience Lab for their assistance. Lastly, I thank my wife and family for their love
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ................................................................................................. iii
LIST OF TABLES ................................ .. .......... .................................... vi
LIST OF FIGURES ............................. .. .......... .................................. vii
A B S T R A C T ........................................................................................................ ........... ix
1 IN TR O D U C T IO N ........ .. ......................................... ..........................................1.
P hobias and F ear .......................................... .... ... ....... ....... ..................... 1
Attentional and Perceptual Components of Emotional Processing ............................. 2
Emotional Connections with Working Memory............... .....................................5...
E v ent-R elated P otentials ...........................8.......... .. ....................... ..................... 8
Neural Correlates of Cognitive and Affective Picture Processing .......................... 10
Summary and Rationale for the Current Study............... .....................................12
P re d ictio n s .............................................................................................................. . 14
W M T ask P perform ance...................................................................... ............... 14
ERP Reflections of Task-Irrelevant Interference Stimulus Processing ............ 15
2 M E T H O D ................................................................................................................. .. 16
P a rtic ip a n ts ................................................................................................................ 1 6
M materials and P procedure .................................................................... ............... 18
E E G A acquisition and R eduction............................................................ ................ 21
E E G D ata A acquisition ......................................... ........................ ................ 2 1
E E G D ata R education ................................................................... ................ 22
Statistical Analyses .................. .. ............ ............................... 23
W M P perform ance D ata ........................................................................................24
P picture R response D ata ......................................... ........................ ................ 24
E R P D ata ............................................................................................................. 2 4
3 R E S U L T S ................................................................................................................. .. 2 6
T ask P perform an ce .................................................. ............................................. 2 6
E effects of W M L oad .............................................. ......................... ............... 2 6
Effects of Interference on WM Task Accuracy................................................ 27
Effects of Interference on WM Task Reaction Time ....................................28
P picture R response D ata ..................................................................... ... ......... 3 1
Interfering Picture D election D ata.................................................. ................ 31
View ing Tim es During Picture Rating ........................................... ................ 32
Manipulation Check: Picture Rating Data............... .....................................33
V alence ratings .................................................................. ............... 33
A rou sal rating s ........................................................ ................ ... . ........ 34
Event-R elated Potential (ERP) D ata...................................................... ................ 36
E effects of W M L oad .............................................. ......................... ............... 36
Effects of Picture Interference.................. .................................................. 40
4 D IS C U S SIO N ................................................................................................... 4 8
Interference Effects on WM Task Performance ...................................................... 48
Picture R response D ata ................... ............................................... 50
ERP Reflections of Interference-Stimulus Processing .........................................51
WM Load Effects on Interference-Related Activity ....................51
Valence-Related Effects Interference Activity ..............................................52
Emerging Patterns................. .... ..... ............. ...................... 53
Alternative Explanations and Possible Limitations...............................................54
F u tu re D direction s ........................................................................................................ 57
S u m m a ry .................................................................................................................. .. 5 8
L IST O F R EFE R E N C E S .... ....................................................................... ................ 59
BIO GR APH ICAL SK ETCH .................................................................... ................ 65
LIST OF TABLES
2-1. Demographic and emotional functioning data for all participants ............................. 18
2-2. Intercorelations between measures of anxiety and depression .................................. 18
3-1. ANOVA statistics from the WM task accuracy data.............................................. 28
3-2. Valence effects in the WM task accuracy data....................................................... 28
3-3. ANOVA statistics from the probe reaction time data...........................................29
3-4. Valence effects in the probe reaction time data....................................................29
3-5. ANOVA statistics from the ERP components......................................................42
3-6. V alence effects in the ERP com ponents............................................... ................ 43
LIST OF FIGURES
1-1. Extraction of the ERP waveform from ongoing EEG..........................................9...
2-1. Overview of each trial, showing the time course of each cue, picture
interference, and probe .......................................... .. ....................... ............... 20
2-2. Sensor layout of the 64-channel geodesic sensor net..........................................22
3-1. WM task performance for each WM load during no-interference trials.............. 27
3-2. Error rates by interference category, and fear group...........................................29
3-3. Probe reaction time by interference category, and fear group ...............................31
3-4. Mean picture detection reaction times by valence and fear group ........................32
3-5. Mean RTs for picture viewing during the rating procedure at the end of the
ex p e rim e n t............................................................................................................... 3 3
3-6. Subjective ratings for picture valence ................................................. ................ 34
3-7. Subjective ratings for picture arousal ................................................. ................ 35
3-8. Grand-averaged ERPs for all scalp sites during interference picture processing in
the low (blue) and high (red) WM load conditions.............................................37
3-9. Spherical-spline interpolated scalp voltage maps representing the differences in
neural processing between the high and low WM loads for the early and late
L P P s an d slow w av e ................................................................................................. 3 8
3-10. Grand-averaged ERPs for site #34 during interference picture processing in the
low (blue) and high (red) W M load conditions...................................... ............... 38
3-11. Mean ERP amplitudes for low and high WM load conditions for the early and
late L PP and slow w ave.. .. ............................................................... ................ 39
3-12. Grand-averaged ERPs for all scalp sites during interference picture processing in
the neutral (blue), pleasant (black), unpleasant (green), and threat (red)
c o n d itio n s. ............................................................................................................... 4 1
3-13. Spherical-spline interpolated scalp voltage maps representing the differences in
neural processing between affective and neutral pictures for the early and late
LPPs and slow w aves of high load trials............................................. ................ 42
3-14. Grand-averaged ERPs for site #34 during interference picture processing in the
neutral (blue), pleasant (black), unpleasant (green), and evolutionary threat (red)
conditions at high WM loads, collapsed across groups....................43
3-15. Mean ERP amplitudes for the early and LPP and slow wave for controls ..............44
3-16. Mean ERP amplitudes for the early and LPP and slow wave for high-fear
p articip an ts. .............................................................................................................. 4 5
3-17. Grand-averaged ERPs for controls and high-fear participants during threat and
unpleasant interference picture processing in the at high WM load ................ 45
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
INTERFERENCE EFFECTS OF ANXIETY AND AFFECTIVE PROCESSING ON
WORKING MEMORY: BEHAVIORAL AND ELECTROPHYSIOLOGICAL
David Andrew Stigge-Kaufman
Chair: William M. Perlstein
Major Department: Clinical and Health Psychology
Emotionally salient information is often given preferential processing in attention
and perception. While this offers many benefits, it can become problematic in disorders
of anxiety when threat-related processing is excessive and dysfunctional. Affective
processing has been shown to modulate higher-level cognitive functions like working
memory, although the nature of these effects is poorly understood for both anxious and
healthy individuals alike. This study examined whether affective interference impairs
representations in working memory (WM) and whether this impairment is greater in
response to interfering emotional stimuli that have intrinsic clinical or evolutionary
relevance. High-fear participants and low-fear controls performed a novel delayed
matching-to-sample (DMS) task involving high and low WM "loads" and the
presentation of interference pictures of varying valence (pleasant, neutral, unpleasant, and
clinically/evolutionarily-relevant threat) during the delay. Error rates, reaction time (RT),
and high-density event-related potentials (ERP) were acquired while participants
performed the task. Performance data from WM task revealed that WM performance is
disrupted by affective interference and that arousal and anxiety play key roles in that
interference. ERP data suggest that increased WM load and the processing of task-
irrelevant affective interference compromise neural resources of attention. Deficits in the
neural processes involved in sustained attention appear to be associated with enhanced
threat-based interference of WM in high-fear participants, which may have important
implications for broader behavioral symptoms associated with anxiety. In the end, these
findings offer new insights into the complex interplay between competing cognitive and
affective demands on attention, and suggest ways that anxiety disrupts higher-level
cognitive processes involved in WM.
Phobias and Fear
Recent epidemiological studies estimate that phobias are the single most common
mental disorder in the United States, affecting approximately 5 to 10 percent of the
population (Sadock & Sadock, 2003). Distress associated with phobias often results in a
restricted lifestyle and can lead to many other psychiatric complications, including other
anxiety disorders, major depressive disorder, and substance-related disorders (American
Psychiatric Association [APA], 2000). Although phobias are extremely common, many
persons with phobias do not seek help to overcome their phobias or are misdiagnosed
when given medical or psychiatric attention (Sadock & Sadock, 2003). Specific phobics
have been shown to exhibit cognitive biases for threat-related pictures and words (Kindt
& Brosschot, 1997), which likely play a role in the maintenance of fear. Additionally,
individuals with subclinical levels of anxiety exhibit notable deficits in the cognitive
processing of threatening information (Fox, Russo, Bowles, & Dutton, 2001; Ohman &
Soares, 1998). Animals that pose evolutionary threat (e.g., snakes, spiders) are common
objects of intense fear, reported by as many as 38% of females and 12% of males
sampled (Ohman & Mineka, 2003).
Given their high prevalence, the functional consequences of phobias and high
levels of fear are worthy of considerable attention. Previous investigations have
examined the perceptual and attentional processing of emotional information, while other
lines of research have begun to focus on connections between emotional processing and
higher levels of cognition, including working memory. The combined study of affective
and cognitive processes is complex, but previous findings now lay the foundation for a
better understanding of the neural processes involved in affective influences on cognition,
which makes it possible to more carefully study the functional impact of fear and other
emotional factors on higher cognitive processes like working memory.
Attentional and Perceptual Components of Emotional Processing
Examinations of emotions from an evolutionary perspective are quick to highlight
the importance of allocating preferential perceptual processing to certain environmental
stimuli (Dolan, 2002). For millions of years, species' survival has depended on the
ability to efficiently process the wide variety of sensory cues that exist in the natural
world (Ledoux, 1996). Emotional stimuli that occur naturally in the environment (such
as snakes and spiders) are detected faster than non-emotional stimuli, suggesting that
evolutionarily relevant threatening stimuli are effective in capturing visual attention
quickly and consistently (Ohman, Flykt, & Esteves, 2001). Enhanced attention has even
been shown to spread to non-emotional stimuli that merely appear within spatial
proximity to emotional cues (Williams, Watts, MacLeod, & Matthews, 1997).
Participants of spatial orienting tasks respond faster to non-emotional targets appearing
on the same side or same location as an emotional cue (e.g., positive or negative faces),
while a slower response results for targets appearing on the opposite side or different
location (Armony & Dolan, 2002), suggesting preferential processing for emotionally
Despite the obvious benefits that accompany enhanced perceptual processing of
emotional stimuli, excessive levels of attention may give rise to the emotional
dysfunction present in certain psychological disorders. It has been suggested that at least
some anxiety disorders are caused by automatic biases of attention that cause excessive
engagement to threat-related stimuli (Ledoux, 1996; Williams et al., 1997). These claims
have been fueled in part by results from tasks that measure the reaction time of
effectively modulated attentional processing. In studies involving the "dot-probe"
paradigm, for example, anxious participants detected a target dot more quickly after it
appeared in the location previously occupied by a threat-related word (MacLeod,
Matthews & Tata, 1986; Fox, 1993). Other studies employing the emotional Stroop
paradigm have found that color naming was slower on anxiety-related words compared to
neutral words (MacLeod, 1991, Vrana, Roodman, & Beckham, 1995), suggesting that an
interference effect can be facilitated by emotional processing. However, not all
researchers attribute anxiety to hypervigilence of the attentional system.
Fox and colleagues (2001) demonstrated that threatening words and faces may not
serve to attract attention, but instead disrupt the disengagement of attention by using a
modified version of the exogenous cueing paradigm (Posner, Inhoff, Friedrich, & Cohen,
1987). They concluded that threatening cues increase the amount of time that attention is
allocated to a stimulus by making it difficult to disengage attention from that stimulus.
While initial disengagement may constitute an early stage of the attentional bias for
threatening stimuli, research in spider phobics has suggested that high levels of anxiety
can also cause rapid disengagement at later processing stages (Hermans, Vansteenwegen,
& Eelen, 1999). Although the debate over the mechanism underlying these findings has
not been resolved, it seems likely that emotionally-motivated attention is mediated by a
complex interplay between multiple, overlapping neural systems that mediate different
kinds of psychological processing (Compton & Banich, 2003).
Capturing the attentional system does not appear to be the only way that
emotional stimuli influence perception. In situations where attention has been
systematically limited, emotional stimuli continue to show preferential processing.
Visual backward masking paradigms tap the automatic regulation of attention by
presenting an threatening target stimulus briefly (30 milliseconds) that is then masked by
an immediately following second stimulus. Even when the emotional target (e.g.,
pictures of snakes or spiders) is not consciously perceived, physiological processing (as
measured by skin conductance responses) reveals differential autonomic processing in
individuals who are fearful of those stimuli (Ohman & Soares, 1994). Furthermore, other
studies have found similar effects in non-fearful subjects who were aversively
conditioned to fear the masked stimuli (Ohman & Soares, 1993). From these findings, it
appears that certain threat-provoking stimuli are selectively processed by autonomic and
central nervous systems even in the absence of conscious awareness.
A brain structure that plays a key role in automatic processing of threat is the
amygdala, a subcortical structure of the limbic system located deep in the anterior medial
temporal lobe (Armony & LeDoux, 1997). Neuroimaging studies consistently find the
amygdala to be active during the evaluation of fearful stimuli, such as fearful facial
expressions (Morris et al., 1996) and affective pictures (Lane et al., 1997). Patients with
amygdala lesions show significant deficits in recognition of fearful facial expressions
(Adolphs et al., 2005). Anatomical analyses reveal that the amygdala receives inputs
from the cortex as well as direct sensory inputs from the thalamus, suggesting that the
amygdala can process emotional information independently of conscious processing
streams (Ledoux, 1996). Rapid processing of emotional information is highly adaptive
when the brain needs to evaluate threatening information; however, this automaticity
comes at the cost of more in-depth cortical processing of stimuli (Armony & LeDoux,
1997). With its anatomical connections and its role in evaluating fearful stimuli, the
amygdala appears to serve the evolutionary role of the brain's "threat detector" (Ohman
& Mineka, 2001).
All in all, there is strong evidence that humans give preferential attention to
emotional stimuli. This bias in attention allows for rapid adaptation to environmental
cues that promotes species survival. However, humans possess higher levels of cognitive
control that allow for more flexible adaptation of behavior to the contextual demands of
our environments (Miller & Cohen, 2001). Are these higher levels of cognitive
functioning also affected by threat-provoking cues that are present in the environment, or
is affective modulation of cognition limited only to the basic levels of attention and
Emotional Connections with Working Memory
The ability to internally store and manipulate information provides additional
survival benefits beyond those afforded by perception and attention alone. These benefits
have been provided by a fundamental set of cognitive processes known as working
memory. Working memory (WM) involves the active maintenance and manipulation of
selective symbolic information while inhibiting other information, utilizing two
components: short-term storage and executive processes (Baddeley, 1986). Through its
dependence on processing symbolic representations, WM serves to free an organism from
a dependence on the presence of environmental cues and is critical for behavioral
flexibility, internal monitoring, and guidance of contextually appropriate action (Miller &
Cohen, 2001). In experimental tasks that measure WM, participants are required to
actively maintain representations or manipulate information over the course of a delay.
Anatomical analyses in primates have concluded that WM is mediated by a network of
brain structures that includes the prefrontal cortex (e.g., Goldman-Rakic, 1987).
Functional neuroimaging research in humans has shown that multiple frontal regions are
active during short-term WM storage (including Broca's area and motor areas), while the
dorsolateral prefrontal cortex (dlPFC) is involved in active maintenance of
representations (Cohen, et al., 1997; Fuster, 1997; Smith & Jonides, 1999). Recent
findings suggest that the dlPFC aids in the maintenance of information by directing
attention toward internal representations of sensory and motor information that are stored
in more posterior processing centers of the brain (Curtis & D'Esposito, 2003).
Understanding the neurochemical functioning of the PFC provides clues as to
how WM may be integrated with affective processing. Research has shown that normal
PFC activity is dependent on dopamine (DA), a catecholanime neurotransmitter that is
modulated by the amygdala during times of stress (Arnsten, 1998). Both stress and
pleasant affect can trigger the release of DA, which also increases the activity of the
amygdala, in turn triggering more production of DA in the PFC (Ashby, Isen, & Turken,
1999). Unfortunately, when Dl receptors receive too much DA in the PFC, cognitive
dysfunction such as poor attention, impaired response inhibition, and disruptions in WM
often result (Arnsten, 1998, Goldman-Rakic, 1996). Paradoxically, mild levels of
positive affect have been shown to enhance performance on WM tasks (Ashby et al.,
1999), while anxiety can serve to impair both visuospatial (Lavric, Rippon, & Gray,
2003) and verbal WM (Ikeda, Iwanaga, and Seiwa, 1996). The end result of this appears
to be that "both too little and too much dopamine Dl stimulation are detrimental to
prefrontal function" (Goldman-Rakic, 1996, p. 13478), which may serve to take the
dlPFC offlinee" (Arnsten, 1998).
If the dlPFC is taken offlinee," subcortical structures like the amygdala may have
the opportunity to guide behavior with more automatic and reflexive affective processing
and responding. Recent evidence has found that task-irrelevant pictures that are
emotionally arousing cause greater interference effects in cognitive tasks (e.g., solving
math problems) than emotional pictures that represent evolutionary threat or general
negativity in low-fear individuals (Schimmack, 2005). This suggests that emotional
arousal may be the key ingredient for interference of cognitive functioning; however,
little is known about the underlying neural processes that are involved with affective
interference on higher cognitive functions. It is also not clear if these effects would
generalize to high-fear individuals, who show more susceptibility for threat-evoked
decreases in dlPFC activity than controls (Carlsson et al., 2004).
Further clues about the dissociations between affective and cognitive processing
have been studied using fMRI in attention- and WM-related tasks. Recent research has
found that the middle frontal gyrus (i.e., dorsolateral region) of the PFC is actively
involved in the processing of cognitive targets but deactivated during the detection of
novel, non-target emotional stimuli (Yamasaki, LaBar, & McCarthy, 2002). Perlstein,
Elbert, and Stenger (2002) found that WM-related dlPFC activity was influenced by the
emotional characteristics of task-relevant stimuli, but only when brought "on-line" by
task demands that required active maintenance. Furthermore, they found that unpleasant
affective content reduced WM-related brain activation in the dlPFC relative to neutral
and pleasant content. Other researchers have sought to examine the effects of emotional
states, finding that pleasant states enhance verbal WM and decrease spatial WM
performance, while unpleasant states enhance spatial WM and decrease verbal WM
performance (Gray, Braver, & Raichle, 2002). Overall, the lateral PFC was highly
correlated with these behavioral changes, suggesting that this region of the brain is
important for emotional-cognitive integration. This research has yielded many insights
into the interactions between cognition and emotion, but a key disadvantage of fMRI
research is that it offers less sensitivity to rapid changes in the brain compared to direct
measures of neural activity, such as electrophysiological techniques like event-related
Before examining the contributions of event-related potential (ERP) methodology
to the study of emotion, one must understand the basic assumptions behind ERP methods.
One assumption is that the distribution of electrical activity across the scalp is reflective
of the activities of underlying neural structures. A second assumption is that this neural
activity corresponds with specific cognitive states and processes. As an extension of
these assumptions, electrical potentials then represent information regarding cognitive
states and processes (Kutas & Dale, 1995).
The electrical activity of the brain can be measured non-invasively across the scalp
using electrodes. The electroencephalogram (EEG) is the record of the volume-
conducted electrical activity of the brain. EEGs can be used to observe ongoing electrical
brain activity. Alternatively, electrical activity can be averaged in association with the
presentation of specific events of interest. Initially, the event-related response associated
with the presentation of a stimulus is embedded in the ongoing EEG activity. Extracting
an ERP waveform associated with a specific stimulus is accomplished by averaging
multiple samples of the EEG that are time-locked to repeated occurrences of the stimulus
(see Figure 1-1). The benefit of averaging is that the ERPs should remain somewhat
consistent from trial to trial, while the ongoing background EEG is random and will be
(a) Stimulus 1 Stimulus 2... -
EEG On nVV I", -W,--
_--- ---------- -- ---- -------
Stimuls -------- --V^-- -- g
Stimulus N A2 V
L-- -- Time (ms)
Figure 1-1. Extraction of the ERP waveform from ongoing EEG. (a) Stimuli (1.. .N) are
presented while the EEG is being recorded, but the specific response to each
stimulus is too small to be seen in the much larger EEG. (b) To isolate the
ERP from the ongoing EEG, the EEG segments following each stimulus are
extracted and averaged together to create the averaged ERP waveform. Taken
from Luck, Woodman, & Vogel (2000).
ERPs are highly sensitive to changes in neural activity on the level of
milliseconds (ms), making them the "gold standard" among noninvasive imaging
methods in terms of temporal resolution (Fabiani, Gratton, & Coles, 2000). ERP
waveforms usually consist of discrete voltage deflections that can be positive or negative,
which are often followed by longer lasting potentials that are called "slow-waves."
Specific components of ERP waveforms are usually named in accordance with their
polarity (positive or negative) and latency (in ms). A common example is "P300," which
refers to an ERP component with a positive peak that has latency of approximately 300
ms. Sometimes it is more appropriate for descriptors to use broader latency terminology,
referring to deflections as "early" (100-200 ms) or "late" potentials (300-600 ms). Other
descriptors may incorporate scalp location at which the component is maximal, such as
"late centro-parietal potentials."
The neural activity associated with ERP activity is attributed primarily to post-
synaptic potentials in pyramidal neurons of the cortex (Williamson & Kaufman, 1990).
Neurons and the extracellular space that surrounds them are filled with charged ions like
sodium (Na ) and potassium (K ). While neurons are at rest, Na+ ions are more plentiful
in the extracellular space than in the neurons, but this changes when action potentials
cause the post-synaptic membrane to depolarize. When dendrites of a neuron depolarize,
Na+ ions flow into the cell, making the extracellular space more negative (known as the
current sink). These Na+ ions that enter the cell repel other positive ions (like K ) away
from the dendrites, creating a current that sends them toward the cell body. The
accumulation of positively charged ions in the cell body (known now as the current
source) repels positively charged ions away from the surrounding region in the
extracellular space, sending those extracellular positive ions back toward the current sink.
This process creates a dipolar extracellular current that accumulates with large
populations of neurons, giving rise to detectable scalp potentials (Coles & Rugg, 1995).
Neural Correlates of Cognitive and Affective Picture Processing
Since phobia and specific fear often have strong visual components, much can be
learned about their neural correlates by examining the neural processes underlying
affective picture processing. Past investigations have demonstrated that a variety of ERP
components have larger amplitudes to emotionally salient or arousing pictures (pleasant
or unpleasant) compared to neutral pictures. This finding has been consistently
associated with enhanced amplitudes of late positive potentials (including P300) and slow
wave components (Cuthbert, Schupp, Bradley, Birbaumer, & Lang, 2000; Johnson,
Miller, & Burleson, 1986). Some researchers have claimed that these effects in picture
viewing paradigms have theoretical links to motivated attention, in which attentional
resources are aroused and directed by motivationally relevant stimuli (Lang, Bradley, &
Research has shown that affective pictures elicit larger amplitudes for late
positive potentials (LPPs) that typically begin 300-400 ms after picture onset (Cuthbert et
al., 2000). These LPPs show greater amplitude differences when the pictures are more
arousing and emotionally salient, even when the pictures are only presented briefly
(Schupp et al., 2004). Visual processing of pictures associated with threat has evoked
larger deflections in early LPPs (Baas, Kenemans, Bocker, & Verbaten, 2001). Early
components of LPPs have been interpreted as a response of capturing attention,
recognition, and stimulus evaluation (Donchin & Coles, 1988). As a result, ERPs can be
used to investigate the amount of resources available for attentionally-dependent
processing. Schupp and colleagues (1997) found that the attentional resources engaged
while viewing affective pictures can serve to detract from the attentional resources that
are available for other forms of processing, leading to a diminished probe startle ERP to a
burst of noise. From this finding, it appears that the resources of attention that are
available for a given task can be limited by other forms of attentional processing, even
across different sensory domains. A provocative question that remains is how the
attentional demands of higher-order cognitive processes, like WM, could potentially limit
the attentional processing available for processing affective content.
LPPs are not the only ERP components that are sensitive to affective stimuli.
Enhanced ERP positivity from affective pictures may last up to 5 seconds after the
picture onset in the form of slow waves (Cuthbert et al., 2000). Ruchkin and colleagues
(1988) have proposed that positive slow waves vary with amount of memory storage
required for processing. Positive slow potentials have been interpreted as reflecting
increased sustained attention (Cuthbert et al., 2000) and possible roles in memory storage
(Donchin & Coles, 1988). Given their apparent functional significance and capacity for
lengthy durations, positive slow waves are likely to have meaningful effects subsequent
objects of attention or motivation.
Summary and Rationale for the Current Study
Previous research has paved the way for a more comprehensive understanding of
the complex interplay between cognitive and affective processing systems of the brain
and the additional role that anxiety plays in these interactions. Emotional stimuli receive
preferential processing on the perceptual and attentional level, and can even be processed
automatically without conscious awareness. Emotional stimuli activate the amygdala,
which modulates PFC concentrations of dopamine and may serve to disrupt prefrontally-
mediated WM functioning in times of heightened emotional arousal. The dlPFC is
involved in active maintenance of representations in WM, yet recent studies have also
shown that its activity can be modulated by affective or emotional context, as well.
Additionally, threat has been shown to modulate WM performance in both visuospatial
and verbal domains.
ERP methods can be used to detect rapid neural changes in the affective
processing in order to map out a time-course of brain responses that are difficult to assess
using other techniques. The method of ERP is well suited for measuring changes in
neural processing that reflect changes in attentional processing, yet little research has
used this technique to study cognitive-emotional interactions in this regard. Certain
forms of affective processing (e.g., picture viewing) can serve to modulate the amount of
attentional resources available for other forms of emotional behavior (e.g., startle), as
evidenced by LPPs. However, much still needs to be learned about the role that anxiety
and other forms of emotion play in modulating neural and cognitive behavior.
The current study examined the potential role of task-irrelevant emotionally
evocative interference stimulation on PFC-mediated WM processes. The primary
objective was to determine the effects of task-irrelevant affective (pleasant and
unpleasant) stimuli presented during the delay period of a task requiring active
maintenance of stimulus representation in WM. Key questions regarding this issue
include: 1) does affective interference "capture" attention and, thereby, lead to a
degradation of the strength of task-relevant stimulus representations in WM, resulting in
impaired performance; 2) does the emotional valence or arousal of these interfering
stimuli differ in the extent to which they draw resources away from the active
maintenance of task-relevant representations; 3) do evolutionarily-relevant threat stimuli
(i.e., snakes, spiders) disproportionately capture attention and impair WM performance;
and 4) does active maintenance of task-relevant representations in WM draw resources
away from the ability to process task-irrelevant interfering stimuli? A secondary goal
was to determine if individuals with high fear of these evolutionarily-relevant threat
stimuli (e.g., specific phobics/sub-clinical phobics) evidence a disproportionate
attentional capture by these stimuli and a consequent increase in WM impairment.
To address these aims, high- and controls performed high and low "load"
conditions of a novel visual delayed matching-to-sample (DMS) task in which task-
irrelevant pictorial interfering stimuli of different valence categories (pleasant, neutral,
unpleasant, evolutionary threat) were presented during the delay or retention period.
Task performance was measured by error rate and reaction time (RT); high-density event-
related potentials (ERPs) indexed the extent to which task-irrelevant interfering stimuli
WM Task Performance
It is predicted that error rates and probe RTs during the WM task will be
modulated by the emotional valence of the interfering pictures. Emotionally arousing
interference pictures (pleasant, unpleasant, evolutionary threat) are expected to cause
greater task performance decrements than neutral pictures. Furthermore, it is predicted
that high-fear participants will show a disproportionate decrement in task performance to
the fear specific (i.e., evolutionary threat) interference stimuli than controls, and that this
dissociation will be most evident at the high WM load condition when active
maintenance of task-relevant stimulus representations is greatest. Finally, it is not
anticipated that high-fear subjects will show deficits in "baseline" WM task performance,
as there is little evidence to suggest that under non-arousing conditions they have
impaired dlPFC/WM function.
ERP Reflections of Task-Irrelevant Interference Stimulus Processing
If, as predicted, the active maintenance of task-relevant representations in WM
draws resources away from the ability to process task-irrelevant interfering stimuli, it is
expected that there will be load-dependent effects on the LPP components and slow
waves evoked during the presentation of interference pictures. That is, the LPP
components will be smaller in amplitude during the high- than low-load task conditions.
Furthermore, it is predicted that the LPPs and slow wave amplitudes evoked by the
interfering pictures will be greater in amplitude for emotionally-arousing than neutral
interference stimuli, indicative of a greater capture-of-attention effect by emotionally-
arousing interference stimuli. Finally, it is also predicted that high-fear participants will
show greater LPP and slow wave amplitudes to the evolutionary threat interference
compared to interference of other valence categories and compared to controls.
Thirty individuals (16 female) between the ages of 19 and 45 participated in the
study in exchange for course credit or financial compensation. All participants provided
written informed consent in accordance with procedures of the University of Florida
Health Science Center Institutional Review Board. All participants were given the
"Specific Phobia" section of the Anxiety Disorders Interview Schedule for the DSM-IV
(ADIS-IV; Brown et al., 1994), along with other relevant sections necessary to assess
comorbid psychopathology. The ADIS-IV provides diagnostic information pertaining to
all anxiety disorder categories and the full range of mood, substance-related, and
Based on the results of the ADIS-IV, 14 of the participants were judged to have
high levels of fear to either snakes or spiders, ranging from "moderate" to "very severe."
Six of the high fear participants also met diagnostic criteria for other anxiety disorders,
including Panic Disorder without Agoraphobia, Generalized Anxiety Disorder, and
Social Phobia. Gender makeup was not significantly different between the high-fear and
control groups [X2(1, N= 30) = 1.27, p > .25], although 64% of the high-fear participants
were female, compared to 44% of the controls. Age and educational background were
also equated across fear group, as shown in Table 2-1. All of the participants completed
the interview and WM task, yet one of the controls had to be excluded from the ERP
analysis due to technical problems. This subject is included in the sample statistics for
demographic information, emotional assessment, and WM task performance (N= 30),
while the ERP data does not include this subject (N= 29).
Although the ADIS-IV provided the diagnostic information necessary to form the
two participant groups, several other assessments were used to quantify each participant's
level of emotional functioning and affective health. The Snake Anxiety Questionnaire
and Spider Questionnaire (SNAQ and SPQ; Klorman, Hastings, Weerts, Melamed, &
Lang, 1974) further assessed each participant's fear of snakes or spiders. The State-Trait
Anxiety Inventory (STAI; Spielberger, Gorsuch, & Lushene, 1970) was also used to
provide broader information regarding the participants' general levels of anxiety as
manifest in temporary states of distress and more long-term personality traits. Finally the
Beck Depression Inventory Second Edition, (BDI-II; Beck, Steer, & Brown, 1996) was
used to assess for elevated levels of depressive symptoms. Participants were excluded if
they reported depressive symptoms sufficient for a current diagnosis of major depressive
disorder, previous neurological disease, traumatic brain injury (TBI), or current
psychotropic medication use. A total of seven participants were excluded from
participation due to excessive symptoms of depression or substance abuse. As shown in
Table 2-1, participants in the high fear group exhibited significantly higher levels of
specific fear, state anxiety, trait anxiety, and depressive symptoms compared to the
controls. Levels of specific fear, state anxiety, trait anxiety, and depressive symptoms
were highly correlated with each other across all subjects, as shown in Table 2-2. Mean
BDI scores, while different between groups, were below clinical cut-off levels considered
to reflect depression (e.g., 14 for mild depression; Beck et al., 1996).
Table 2-1. Demographic and emotional functioning data for all participants
High Fear (n=14) Low Fear (n=16) t-statistic
Age 25.3 (7.5) 25.0(5.8) .117ns
Education 15.6 (1.7) 15.7 (2.1) -.165 n
SNAQ/SPQ 10.6 (2.6) 2.7(2.9) 7.93**
STAI-State 37.7(10.1) 29.1 (11.4) 2.17*
STAI-Trait 42.9 (11.3) 29.7 (6.7) 3.96**
BDI 10.2 (5.1) 5.0 (5.4) 2.70*
p> .85. *p < .05. p < .001
Table 2-2. Intercorelations between measures of anxiety and depression
STAI-State STAI-Trait BDI
SNAQ/SPQ .391 .536* .538"
STAI-State --- .572 .539*
STAI-Trait --- --- .719***
p < .05. **p<.01. *p<.001. df= 28.
Materials and Procedure
Participants performed a visual delayed match to sample (DMS) task, using
stimuli modified from Low, Rockstroh, Harsch, Berg, & Cohen (2000). Participants
were seated in front of a computer screen (approximately one meter from screen to nose)
and a keyboard. In this task, geometric shapes were presented as cue-probe pairs with
pictorial interference occurring during the delay period (Figure 2-1). Participants gave
forced-choice responses indicating if they saw the probe stimuli in the previously
presented cue. Each participant completed a total of 400 trials. Each trial began with the
computer presentation of a visual cue, which remained on the screen for 2500 ms. The
cues consisted of two difficulty levels, including one diamond (low load) and three
diamonds (high load). The high and low load trials were presented randomly with equal
probability. The diamonds varied by size and orientation, and participants were
instructed to remember each diamond in the cue by its individual combination of size and
A task-irrelevant interfering picture was presented in 80% of the trials. This
interfering picture was presented 1000 ms following cue offset and remained on the
screen for 750 ms, during which the participants were instructed to press a button to
ensure that the picture was detected and encoded. The pictures were taken from the
International Affective Picture System (IAPS; Lang, Bradley, & Cuthbert, 1998), and
consist of four distinct valences, or emotional categories: pleasant, neutral, unpleasant,
and evolutionary threat (snakes or spiders). Each of the four valence categories consisted
of 25 pictures that were presented multiple times in a pseudo-randomized fashion over
the course of the experiment.1 Because of limited numbers of IAPS pictures that consist
of snakes or spiders, this category of pictures were supplemented with several non-IAPS
pictures. IAPS pictures have standardized ratings for pleasantness (valence) and arousal
on a 1 to 9 scale (9 equals most pleasant and most arousing). The ratings for the pleasant,
neutral, and unpleasant pictures used in this experiment can be seen in Table 2-3. With
regard to pleasantness, pleasant pictures were rated significantly higher than neutral, t(48)
= 18.8, p < .001, which were rated significantly higher than unpleasant, t(48) = 22.0, p <
.001. With regard to arousal, pleasant and unpleasant pictures did not differ, t(24) = .22,
p > .80, but these pictures were rated significantly higher than neutral, ts(24) > 10.2, ps <
.001. In 20% of the trials, participants viewed a continuous fixation point in place of the
interfering picture in order to create a "no interference" condition.
1 IAPS identification numbers for pleasant pictures are 1440, 1460, 1463, 1604, 1710, 2040, 2070, 2071,
4599, 4608, 4623, 4680, 5470, 5621, 5623, 7230, 7260, 7330, 8030, 8370, 8380, 8470, 8501, 8510; for
neutral pictures: 2191, 2235, 2372, 2383, 2393, 2394, 2515, 2745.1, 2840, 2850, 2980, 5390, 5740, 7036,
7100, 7140, 7180, 7217, 7234, 7285, 7491, 7496, 7595, 7710, 8311; for unpleasant pictures: 2095, 2205,
2375.1, 2683, 2800, 6213, 6550, 6570, 6831, 6834, 6838, 8485, 9050, 9220, 9280, 9342, 9400, 9410, 9470,
9471, 9520, 9561, 9830, 9910, 9921; for evolutionary threat pictures: 1019, 1022, 1050, 1051, 1052, 1070,
1080, 1090, 1101, 1110, 1111, 1112, 1113, 1114, 1120, 1201, 1220.
Table 2-3. Mean ratings of each group of IAPS pictures (from Lang et al., 1998)
Mean Ratings (SD)
Pleasant Neutral Unpleasant
Valence 7.59 (.40) 5.25 (.47) 2.44 (.43)
Arousal 5.51 (.92) 3.29 (.58) 5.56 (.88)
A memory probe (single diamond in a box) was presented on the screen 2750 ms
following the offset of the original cue. The probe remained on the screen for 1500 ms,
during which the participants were asked to determine if the diamond shown was an exact
match in both size and orientation to one of the previously shown diamonds for that trial.
Participants pressed "m" to indicate that the probe diamond was a match to the cue and
"n" to indicate that the probe diamond was not a match. 1000 ms following the offset of
the probe, the next trial began with a new cue. Participants were instructed to look at a
fixation point (+) during the time intervals between the cue, picture, and probe.
1000 ms Response
200ms 1000 ms----
Figure 2-1. Overview of each trial, showing the time course of each cue, picture
interference, and probe. Since the probe diamond in this example matches
one of the cue diamonds exactly in its size and orientation, the correct
response for the trial shown here is "m".
After completing the experimental task, participants then rated each picture that
they viewed during the course of the experiment in order to provide a manipulation check
on effects of valence and arousal. These ratings were computerized, using the Self-
Assessment Manikin (SAM) rating system developed by Lang (1980). Participants were
instructed to view each picture for as long as they needed in order to be able to rate the
pictures, and they self-terminated their viewing of each picture with a button press.
Participants then rated each picture separately for valence and arousal, using the same 1
to 9 scale used in the standardized IAPS ratings (9 equals most pleasant and most
EEG Acquisition and Reduction
EEG Data Acquisition
EEG was recorded from 64 scalp sites using a 64-channel geodesic sensor net
(Figure 2-1) and amplified at 20K using an Electrical Geodesics Incorporated (EGI)
amplifier system (nominal bandpass .10 100Hz). Electrode placements enabled
recording vertical and horizontal eye movements reflected in electro-oculographic (EOG)
activity: one placed above and below each eye and centered around the pupil to record
vertical eye movements; the others placed at the outer canthus of each eye for recording
horizontal eye movements. EEG was referenced to Cz and was digitized continuously at
250 Hz with a 16-bit analog-to-digital converter. A right posterior electrode served as
common ground. The impedance of all electrodes was maintained below 50 kQ,
consistent with procedures suggested by the manufacturer.
14 7 1
15 8 3 61
23 16 9 5B 57 59
Left ear 21 53 Right ear
I aft ar 24 18 43 52 Right ear
25 30 50
26 29 42 51
27 28 (34) 46 49
31 32 45 48
55 36 44 Com
Figure 2-2. Sensor layout of the 64-channel geodesic sensor net. Electrode #34
(outlined) was used for measurement of picture-related ERPs.
EEG Data Reduction
Due to the volume-conducting nature of the brain, no single scalp site can be
considered an "inactive" reference site (Tucker, Liotti, Potts, Russell, & Posner, 1994);
therefore, data were mathematically re-referenced against an average reference (Bertrand,
Perrin, & Pernier, 1989). In this procedure, the activity of each electrode site is reflected
as the difference between itself and the average of all the other recording sites. Editing of
the EEG for movement, electromyographic muscle artifact, electro-ocular eye movement,
and blink artifacts was performed by computer algorithm in Brain Electrical Source
Analysis software (BESA version 5.0; Scherg, 1990). EEG during which voltage
exceeded 150 [V and with point-to-point transitions exceeded 125 [V were excluded
Individual-subject event-related potentials (ERPs) were extracted and averaged
together from the ongoing EEG recording in discrete temporal windows that coincided
with the onset of each stimulus (refer back to Figure 1-1). ERP averages from each
subject were divided into three categories: cue activity, interfering picture activity, and
probe activity. Stimulus-locked epochs were extracted with a duration of 100ms prior to
stimulus presentation (constituting a 100ms pre-stimulus baseline period) and 1000ms
post-stimulus presentation. Collapsing across certain conditions was necessary because
of insufficient numbers of responses to conduct specific analyses. All averaged ERP
epochs were baseline corrected using a 100 ms window prior to stimulus onset and
digitally filtered at 15 Hz low-pass and a .5 Hz high-pass. Although ERPs were acquired
for stimulus-related activity for the cues, probes, and interfering pictures, only the ERPs
for interfering pictures will be discussed in this study.
WM task performance, picture responses, and ERP amplitudes were analyzed
using repeated measures analyses of variance (ANOVAs). Because of the a priori
hypotheses regarding group differences on these measures, planned contrasts were used
to decompose interaction effects within the ANOVAs that were performed. When
applicable, these contrasts utilized orthogonal comparisons of different levels of affective
picture categories: 1) affective vs. neutral, 2) pleasant vs. negative (unpleasant and
threat), and 3) unpleasant vs. threat. In addition to these planned contrasts, follow-up
contrasts were employed and, where appropriate, adjusted for multiple comparisons using
the modified Bonferroni method (Keppel, 1982). For ANOVAs where there were more
than two levels of a within-subject factor, the Huynh-Feldt epsilon adjustment (Huynh &
Feldt, 1976) was used; uncorrected degrees of freedom and corrected p-values are
reported. In addition, two-tailed Pearson product-moment correlations were performed to
determine associations between continuous variables of interest.
WM Performance Data
For each participant, error rates and probe RTs were calculated for each WM load
and interfering picture valence category. For all RT measurements, median RTs were
initially calculated for each participant in order to better accommodate outliers (Ratcliff,
1993), and then means of these individual RTs were compared in subsequent inferential
analyses. Error rates and probe RTs for the no-interference condition were then
examined for an effect of WM load, using a 2-Group x 2 Load repeated measures
ANOVA. After detecting a significant effect of WM load on task performance, error
rates and probe RTs were analyzed by subjecting them both to a 2-Group x 4-Valence
Picture Response Data
A 2-Group x 4-Valence repeated measures ANOVA was performed on the RTs of
the interfering picture detection responses, as well as the mean viewing times during the
picture rating procedure. Post-experimental picture ratings were subjected to 2-Group x
4-Valence repeated measures ANOVAs separately for ratings of pleasantness and
Analysis of ERP waveforms focused on activity reflecting processing of the
interfering pictures. Statistical analyses of ERP waveforms assessed the mean voltages
over specified temporal windows (epochs) of individual subject ERPs extracted from
individual electrode sites for correct trials. Scoring windows and electrode positions for
each condition of interest were determined by examination of grand-averaged ERP
waveforms and the grand mean global power of the voltage obtained over all electrode
sites. Picture-related ERP activity was quantified at electrode site #34 (Figure 2-2) and
was examined over three different time periods believed to represent different aspects of
visual processing. The first time period was 296 ms to 356 ms, which corresponds to the
early portion of late positive potentials (LPP) associated with affective picture
processing, while the second time period was 452 ms to 512 ms, corresponding to the late
portion of LPP (Schupp et. al., 2003). The third time period was 660 ms to 760 ms,
corresponding to the slow wave component of visual processing (Keil, et al., 2002).
Thus, three ERP components associated with processing interfering pictures were
examined: early LPP, late LPP, and slow wave. A 2-Group x 2 Load repeated measures
ANOVA was performed on the amplitudes of each ERP component to determine the
overall effects of WM load on ERPs, collapsing across different picture types to
determine the effects of WM load on ERPs to the interfering pictures. Subsequently,
amplitudes of each ERP component were analyzed in 2-Group x 4-Valence repeated
measures ANOVAs to determine the effects of interfering picture valence on ERPs to the
Initial analyses were performed to examine the possibility of a speed/accuracy
trade-off by correlating error rates on the WM task and reaction times to the probe.
These analyses revealed that probe reaction time (RT) and error-rate were not
significantly correlated, r(27) = .145, p > .40, indicating that speed-accuracy trade-off
does not account for the behavioral findings reported below.
Effects of WM Load
An important aspect of the present research was to demonstrate that the WM load
manipulation was indeed associated with alterations in behavioral performance. To
examine effects of WM load in the absence of interfering pictures, 2-Group x 2 Load
repeated measures ANOVAs were performed on error rates and probe RTs for the no-
interference trials. As expected, participants as a whole exhibited more errors, F(1,28) =
106.26, p < .001, r2 = .791, and longer probe RT, F(1,28) = 181.46, p < .001, r2 = .866,
in high WM load trials compared to low load trials, as shown in Figure 3-1. Fear group
did not exert a significant main effect on error rate, F(1,28) = .01, p > .90, qr2 < .001, or
probe RT, F(1,28) = 3.44, p > .07, qr2 = .109. Furthermore, examination of Group x
Load interactions revealed that high-fear participants did not significantly differ from
controls in their load-related WM performance with regard to error rates, F(1,28) = .13, p
> .70, qr2 = .005, or probe RTs, F(1,28)=2.92,p > .10, r2 = .094, for no-interference
Effects of WM Load on Accuracy Effects of WM Load on Probe RT
E 800- 0 Low Load
a 0.2 Ea High Load
o E 600
Controls High-Fear Controls High-Fear
Figure 3-1. WM task performance for each WM load during no-interference trials.
Error bars represent standard errors.
These findings suggest that high WM load trials required more active
maintenance of the stimulus representations (three diamonds) in WM than low load trials
(one diamond), and that high levels of fear did not impair WM performance. Since the
pictures presented during the delay were predicted to interfere with the active
maintenance of WM representations, it is likely that the interference will have its
maximal effect on the high WM load trials. As a result, subsequent analyses will focus
on high WM load trials only.
Effects of Interference on WM Task Accuracy
A 2-Group x 4-Valence repeated measures ANOVA was performed on the error
rates for high WM load trials to determine the effects of the interfering pictures. Planned
orthogonal contrasts were used to test a priori hypotheses about the effects of interfering
picture valence. Statistics for the ANOVA main effects and interactions can be found in
Table 3-1, while results from the contrasts can be found in Table 3-2.
Table 3-1. ANOVA statistics from the WM task accuracy data
Effect F-Ratio rj2
Group (G)a 1.25 .043
Valence (V)b 3.19* .102
G x Vb 1.60 .054
adf= 1,28. bdf= 3,84. *p < .05.
Table 3-2. Valence effects in the WM task accuracy data
Valence Valence x Group
(Across Group) (Across Load)
Valence Contrast" F-Ratio qr2 F-Ratio r|2
Affective vs. Neutral 9.49 .253 1.56 .053
Negative vs. Pleasant .53 .018 3.11t .100
Threat vs. Unpleasant .02 .001 .28 .010
adf= 1,28. tp< .10. *p <.01.
Means and standard errors of the error rates as a function of fear group and
interference category can be seen in Figure 3-2. A main effect of picture valence was
present for the error rates of the high WM load interference trials. Planned contrasts
revealed that participants as a whole made more errors in each of the three affective
picture trials compared to neutral. A trend-level Group x Valence interaction emerged,
with high-fear participants making more errors than controls during negative (unpleasant
and threat) picture trials relative to pleasant. Follow-up group-wise contrasts using
modified Bonferroni-corrected comparisons at each valence level (critical p = .0188)
revealed that controls exhibited more errors during pleasant picture trials than neutral (p
< .0188). Error rates for high-fear participants did not significantly differ among the
interfering picture valences or in comparison with those of the controls.
Effects of Interference on WM Task Reaction Time
A 2-Group x 4-Valence repeated measures ANOVA was performed on the RTs
for high WM load trials to determine the effects of the interfering pictures. Planned
orthogonal contrasts were used to test a priori hypotheses about the effects of interfering
picture valence. Statistics for the ANOVA main effects and interactions can be found in
Table 3-3, while results from the contrasts can be found in Table 3-4.
Effects of Interference on Accuracy
Neutral Pleas. Unpleas. Threat
Figure 3-2. Error rates by interference category, and fear group. Error bars represent
Table 3-3. ANOVA statistics from the probe reaction time data
Effect F-Ratio q2
Group (G)a 1.58 .053
Valence (V)b 3.76* .118
G x Vb 1.78 .060
adf= 1,28. bdf= 3,84. p < .05.
Table 3-4. Valence effects in the probe reaction time data
Valence Valence x Group
(Across Group) (Across Load)
Valence Contrasta F-Ratio q2 F-Ratio q2
Affective vs. Neutral .76 .026 5.60* .167
Negative vs. Pleasant 8.27** .228 .17 .006
Threat vs. Unpleasant 1.32 .045 .02 .001
adf= 1,28. *p< .05. **p<.01.
A main effect of picture valence was present for the probe RTs of the high WM
load interference trials, and planned contrasts revealed that participants took longer to
respond to the probe during negative (unpleasant and threat) picture trials compared to
pleasant. Means and standard errors of the reaction times to the probes as a function of
fear group and interference category can be seen in Figure 3-3. Follow-up contrasts
clarified that this negative picture effect was actually due to the threat pictures, as
interfering threat picture trials resulting in significantly longer probe RTs than pleasant,
(p < .0188), while unpleasant picture trials did not (p = .05). Follow-up group-wise
contrasts using modified Bonferroni-corrected comparisons revealed that high-fear
participants took longer to respond to the probe during threat picture trials than neutral (p
< .0188), while controls failed to show this pattern (p > .200). Instead, controls exhibited
faster probe RTs during pleasant picture trials compared to neutral (p < .0188). Faster
probe RTs for pleasant picture trials in controls were not significantly correlated with
increased error rates, r(27) = .013, p > .95, suggesting that they did not exhibit a
speed/accuracy tradeoff for pleasant picture trials. Although high-fear participants took
longer to respond in every category of interfering picture trial, their probe RTs did not
significantly differ from controls.
To summarize the WM task performance data, WM load had very strong effects
on both accuracy and RT, with participants making more errors and taking longer to
respond to the probe during the high WM load trials. Participants as a whole made more
errors during trials involving affective picture interference compared to those involving
neutral pictures. Controls made more errors and took less time to respond to the probe
during pleasant picture trials compared to neutral. High-fear participants exhibited a
trend for making more errors in the negatively-valenced (unpleasant and threat) picture
trials compared to pleasant and exhibited significantly longer probe RTs to threat picture
trials compared to neutral, showing that they demonstrated the expected pattern of WM
impairment from the threat picture interference.
Effects of Interference on Probe RT
Neutral Pleas. Unpleas. Threat
Probe reaction time by interference category, and fear group. Error bars
represent standard errors.
Picture Response Data
Interfering Picture Detection Data
Timed responses were measured as participants indicated that they detected the
interfering pictures during the WM task. While the primary goal of requiring participants
to button-press to the presentation of interfering pictures was to increase the likelihood
that they did, indeed, actually view the pictures, RTs to the presentation of interfering
pictures could potentially shed light on aspects of picture processing. Thus, a 2-Group x
4-Valence repeated measures ANOVA was performed on the mean picture detection
RTs. Means and standard errors of the reaction times to the pictures as a function of fear
group and valence can be seen in Figure 3-4. A significant main effect of picture valence
was seen in the interference picture detection, F(3,81) = 13.07, p < .001, r2 = .326.
Participants were significantly faster in responding to interfering threat pictures compared
to neutral, pleasant, and unpleasant pictures (ps < .0188). A main effect of fear group
was also seen in these RTs, F(1,27) = 5.25, p < .05, r2 = .163, with high-fear participants
taking significantly longer to detect the interfering pictures as a whole than controls.
Mean Picture Detection RT
) 540 T
rE 0 Control
Neutral Pleas. Unpleas. Threat
Figure 3-4. Mean picture detection reaction times by valence and fear group. Error bars
represent standard errors.
Viewing Times During Picture Rating
Timed responses to pictures were measured as participants viewed the self-
terminating pictures prior to making their subjective ratings at the end of the experiment.
Means and standard errors of the picture viewing time as a function of fear group and
valence can be seen in Figure 3-5. A significant main effect of valence emerged, F(3,84)
= 7.48, p < .01, r2 = .211, as participants as a whole chose to view unpleasant pictures
longer than pleasant (p < .0188) and threat (p < .0188) pictures. Although high-fear
participants took more time than controls to view the pictures as a whole before rating
them, this difference was not significant, F(1,28) = 1.72, p = .20, r2 = .058.
Mean Picture Viewing Time
50 Ce o Controlr
4600 H High-Fear
Neutral Pleas. Unpleas. Threat
Figure 3-5. Mean RTs for picture viewing during the rating procedure at the
end of the experiment. Error bars represent standard errors.
Manipulation Check: Picture Rating Data
An important premise of the current research with respect to modulation of WM
by emotionally interfering stimuli, and examination of evolutionarily-relevant threat
stimuli in low- and high-fear participants, is that the interfering stimuli were, indeed,
perceived as both emotionally-arousing and as producing the intended emotional valence.
These "manipulation checks" are examined below.
Significant main effects were found for participants' ratings of the pictures'
pleasantness, F(3,84) = 174.94, p < .001, r2 = .862, which can be seen in Figure 3-6. As
expected, participants rated pleasant pictures as more "pleasant" compared to ratings of
neutral, unpleasant, and threat pictures (ps < .0188). Also as expected, participants rated
unpleasant pictures as more "unpleasant" than neutral and pleasant pictures (ps < .0188).
However, this main effect of valence was qualified by a Group x Valence interaction,
F(3,84) = 5.86, p < .01, r2 = .173, as high-fear participants rated threat pictures as more
"unpleasant" than controls (p < .0188). Controls rated threat pictures as less "unpleasant"
than the unpleasant pictures (p < .0188), while high-fear participants rated threat and
unpleasant pictures as equally "unpleasant" (p > .95).
Picture Valence Ratings
.* | 8 OControl
n 7 High-Fear
Neutral Pleas. Unpleas. Threat
Figure 3-6. Subjective ratings for picture valence. Error bars represent standard
errors. p < .001
Significant main effects were found for participants' ratings of the pictures' level
of arousal, F(3,84) = 19.83,p < .001, q2 = .415, which can be seen in Figure 3-7.
Participants rated the pleasant, unpleasant, and threat pictures as more arousing than the
neutral (ps < .05). However, this main effect of arousal was qualified by a Group x
Valence interaction, F(3,84) = 8.98, p < .001, q2 = .243, as controls rated pleasant
pictures as more arousing than high-fear participants (p < .0188), while high-fear
participants rated threat pictures as significantly more arousing than controls (p < .0188).
Additionally, high-fear participants rated threat pictures as more arousing than neutral
pictures (p < .0188), while controls did not (p > .35), but instead rated pleasant pictures
as more arousing than neutral (p < .0188). There was no significant difference between
high-fear participants' arousal ratings of threat pictures and controls' arousing pictures of
pleasant pictures, t(28) = 1.138, p > .26, although Levene's test was significant, F(1,28)
= 11.05, p < .01, indicating that these two sets of ratings did not show equal variances
and this result should be interpreted with caution.
Picture Arousal Ratings
9 n Control
Neutral Pleas. Unpleas. Threat
Figure 3-7. Subjective ratings for picture arousal. Error bars represent standard
errors. *p < .05, **p < .001
To summarize the picture response data, participants detected interfering threat
pictures during the WM task faster than the other categories, and high-fear participants
took longer than controls to signify their detection of interfering pictures. Picture SAM
ratings, which provide a manipulation check on participant's judgments of valence and
arousal, demonstrated that the interfering pictures indeed were effective in producing the
intended arousal and valence effects. Participants' ratings of valence and arousal
differentiated neutral, pleasant, and unpleasant pictures in ways that were generally
expected. High-fear participants rated threat pictures as significantly less pleasant than
the neutral pictures, giving the threat pictures more arousing ratings than controls.
Controls surprisingly rated pleasant pictures as significantly more arousing than high-fear
participants. These arousal ratings show interesting correspondence with the task
performance data, as high-fear participants made exhibited errors and longer probe RTs
during threat picture trials, while controls exhibited more errors and shorter RTs during
pleasant picture trials. Finally, during SAM ratings, participants viewed unpleasant
pictures the longest prior to making their ratings.
Event-Related Potential (ERP) Data
Effects of WM Load
ERP responses to interfering pictures were acquired from all scalp electrodes, and
converted into grand averaged waveforms of picture interference by load (Figure 3-8).
Electrode #34 showed clearly defined LPP and slow wave components. Spherical-spline
interpolated scalp voltage maps showed that a centro-parietal region, encompassing
electrode #34, was sensitive to WM load-related effects in the early and late LPP epochs
(Figure 3-9). Thus, for simplicity, and because a primary aim of the research was to
examine affective interference effects on active maintenance in WM, electrode #34 was
used in all of the WM-related ERP analyses that follow.
Differential load-related activity can clearly be seen over the centro-parietal region of
the scalp in the earlier stages of picture processing (LPP), yet this effect is absent in the
slow wave (see Figure 3-9). Figure 3-10 illustrates the grand-averaged ERP waveforms
for the low and high WM load trials. 2-Group x 2 Load repeated measures ANOVAs
were performed to determine the overall effects of WM load on ERPs in the early LPP,
late LPP, and slow wave epochs from the interfering pictures, collapsing across different
picture types. As shown in Figure 3-11, voltage amplitudes measured from electrode #34
were found to be more positive during low WM load trials compared to the high load for
the LPP time windows, yet no load effect was seen in the slow wave, F(1,27) = 1.484, p
> .20, r2 = .052. The WM load effects on LPP were stronger for the early, F(1,27) =
28.367, p < .001, r2 = .512, than late, F(1,27) = 7.239, p < .05, qr2 = .211, time window.
Controls and high-fear participants were equally affected by these WM load effects, as no
significant Group x Load interaction emerged, Fs(1,27) < .423, ps > .50, qr2 < .015.
Z_ Front 3
- Low Load
both 3cor-vec avr,
0r .. 0 1000
Back Time [ms]
Grand-averaged ERPs for all scalp sites during interference picture
processing in the low (blue) and high (red) WM load conditions.
_.,: ,, ,, !
, ,, .. ".
;*:.. ^ : ;1,,*J
326.0 ms reference free
EEG -Voltage 0.20 pVf step
1 f 0.20 pV/step
710.0 ms + reference free
EEG Voltage T IITiIITM 0.20 IpV step
Spherical-spline interpolated scalp voltage maps representing the differences
in neural processing between the high and low WM loads for the early and
late LPPs and slow wave. Voltage difference maps were calculated by
subtracting high load activity from low load, and were calculated at 326ms
(early LPP), 482ms (late LPP), and 710ms (slow wave) after interference
picture onset. Positive voltage differences are indicated in red, showing
strongest load-related differences in the early LPP and the late LPP.
0 250 500 750
Grand-averaged ERPs for site #34 during interference picture processing
in the low (blue) and high (red) WM load conditions.
Effects of WM Load on ERP Amplitudes
U Low Load
E High Load
-1 Early LPP Late LPP Slow Wave
Figure 3-11. Mean ERP amplitudes for low and high WM load conditions for the early
and late LPP and slow wave. Error bars represent standard errors.
Figure 3-11 clearly shows that early stages (LPP) of interference picture
processing show greater effect of WM load than later stages (slow wave), which fits the
scalp distribution maps of the load-related differences (Figure 3-9). In order to compare
the effect of WM load on these different time-locked ERP components, a 2-Group x 2-
Load x 3-Epoch repeated measures ANOVA was performed. Significant main effects of
load, F(1,27) = 12.49, p = .001, r2 = .316, and epoch, F(1.64, 44.13) = 33.45, p < .001,
2 = .546, were found, but these were qualified by a significant Load x Epoch interaction,
F(1.95, 52.54) = 11.05,p <.001, r2= .290. Planned contrasts revealed that the effect of
WM load on slow wave amplitudes was significantly smaller than it was for the two LPP
epochs, F(1,27) = 13.42, p .001, r2= .332. Of the two LPP time windows, the early LPP
was modulated more by WM load than the late LPP, F(1,27) = 8.46, p < .01, q2= .239.
These load effects did not differ between controls and high-fear participants, as there was
no main effect of group or group interactions with regard to load or epoch effects, Fs <
.80,ps > .45, 2<.030.
Overall, the effects of WM load on the ERP components of interest support the
prediction that fewer attentional resources are available for processing interfering picture
stimuli during the delay period of the high WM load condition relative to the low load.
That is, less neural activity is engaged in the capture of attention for the pictures while
participants actively maintain greater stimulus representations during the high load
conditions. Since the pictures and WM representations are both competing for similar
reservoirs of neural processing, this finding suggests that more neural resources are
allocated to the active maintenance of high load representations (three diamonds)
compared to the low load (one diamond). As the high WM load engages more neural
resources, the opportunity is optimal for interference effects caused by the pictures. As a
result, only ERP components taken from high WM load trials will be examined in the
analyses that follow.
Effects of Picture Interference
ERP responses to interfering pictures were acquired from all scalp electrodes, and
converted into grand averaged waveforms of picture interference for each picture
category (Figure 3-12). Spherical-spline interpolated scalp voltage maps showed that a
central region (including electrode #34) was differentially sensitive to affective pictures
relative to neutral in the early LPP epoch; however, this region of highest differential
sensitivity shifted more centro-frontally during the late LPP and slow wave time windows
(Figure 3-13). In determining the electrode site to use for evaluating the effects of picture
interference, consideration was given to prior research involving affective pictures and
the earlier findings of this study relating to WM-related effects. Since previous research
has found centro-parietal regions to show maximal positive potentials in response to
affective pictures and this region also showed sensitivity to load-related effects in this
study, electrode #34 was used for the picture-related analyses that follow.
44 Front 3
1 d Neutral
1:4 1: -Pleasant
jA -. --'-\ Threat
1 3 j d '
1:16 '---- 43.J- .
4both inor-i c aE r -2
-,----,-i---0 -, "----
processing in the neutral (blue), pleasant (black), unpleasant (green), and
both line -with the WM task performance data, 2-Group x 4-Valence repeated
,5 0---_4-__ 6 t~, ~
,. .. 0|. ,1.0.00.
Back- Time [ms]
Figure 3-12. Grand-averaged ERPs for all scalp sites during interference picture
processing in the neutral (blue), pleasant (black), unpleasant (green), and
threat (red) conditions.
In line with the WM task performance data, 2-Group x 4-Valence repeated
measures ANOVAs were performed on mean amplitudes taken from electrode #34 for
the early LPP, late LPP, and slow wave time periods in order to determine the effects of
the interfering pictures at the high WM load. Planned orthogonal contrasts were used to
test a priori hypotheses regarding effects of interfering picture valence. Statistics for the
ANOVA main effects and interactions can be found in Table 3-5, while results from the
contrasts can be found in Table 3-6.
i- L "
..'j_'., "' i'"' W'
326.0 ms reference free
EEG- Voflage 0.11l V I step
482.0 ms + reference free
EEG Voltage f -- 0.11 pV step
710.0 ms + reference free
EEG Voltage 0.11 V f step
Spherical-spline interpolated scalp voltage maps representing the
differences in neural processing between affective and neutral pictures for
the early and late LPPs and slow waves of high load trials. Voltage
difference maps were calculated by subtracting neutral from affective
picture activity, and were calculated at 326ms (early LPP), 482ms (late
LPP), and 710ms (slow wave) after picture onset. Positive voltage
differences are indicated in red, showing strongest affective-related
differences in the early LPP and the late LPP.
Table 3-5. ANOVA statistics from the ERP components
Early LPP Late LPP Slow Wave
Effect F-Ratio r 2 F-Ratio r2 F-Ratio r2
Group (G)a .025 .001 .194 .007 .24 .009
Valence (V)b 6.70*** .199 6.67*** .198 12.64** .319
G x Vb .173 .006 .619 .022 1.62 .056
adf= 1,27. bdf= 3,81. *p <.001.
Table 3-6. Valence effects in the ERP components
Affective vs. Neutral
Negative vs. Pleasant
Threat vs. Unpleasant
Affective vs. Neutral
Negative vs. Pleasant
Threat vs. Unpleasant
Affective vs. Neutral
Negative vs. Pleasant
Threat vs. Unpleasant
adf= 1,28. p < .05. 'p < .01. j'p < .001.
Valence x Group
0 250 500 750 1000
Figure 3-14. Grand-averaged ERPs for site #34 during interference picture processing
in the neutral (blue), pleasant (black), unpleasant (green), and evolutionary
threat (red) conditions at high WM loads, collapsed across groups.
Figure 3-14 illustrates the grand-averaged ERP waveforms as a function of
interfering picture valence category during high WM load trials. As can be seen, valence
category of the interfering pictures modulated the amplitudes of all ERP epochs. Planned
contrasts revealed that participants as a whole exhibited greater positive deflections in
both LPP time windows for each of the three affective picture types compared to neutral.
-------------------------------------- ------- ------- ---
Both LPPs and slow waves had elevated positive amplitudes for negative (unpleasant and
threat) compared to pleasant interfering pictures. Additionally, slow waves were more
positive for unpleasant compared to threat pictures. This slow wave effect was qualified
by planned contrasts that revealed a significant Group x Valence interaction, with
controls exhibiting significantly more negative slow waves than high-fear participants for
threat relative to unpleasant pictures. Put another way, this indicates that the slow waves
of high-fear participants were more positive than controls with respect to unpleasant slow
waves. Figures 3-15 and 3-16 show the overall effects of all picture valence categories
on ERP amplitudes for controls and high-fear participants, while Figure 3-17 shows
grand-averaged ERP waveforms that reveal the Group x Valence interaction where high-
fear participants show more positive slow wave for threat pictures (relative to unpleasant)
Effects of Picture Valence on ERPs:
.1 M Unpleas.
E U Threat
Early LPP Late LPP Slow wave
Figure 3-15. Mean ERP amplitudes for the early and LPP and slow wave for controls.
Error bars represent standard errors.
Effects of Picture Valence on ERPs:
-2 Early LPP Late LPP Slow wave
Figure 3-16. Mean ERP amplitudes for the early and LPP and slow wave for high-fear
participants. Error bars represent standard errors.
- High-Fear Threat
- High-Fear Unpleas.
0 250 500 750 1000
Figure 3-17. Grand-averaged ERPs for controls and high-fear participants during threat
and unpleasant interference picture processing in the at high WM load.
The enhanced positivity exhibited in high-fear participants' slow waves for the
threat pictures is intriguing because it may account for group differences in sustained
attention to the threat picture interference. The slow wave time window used for the
present analysis was taken from 660-760 ms following picture onset, but as Figure 3-17
shows, the group difference in threat picture slow waves increases dramatically beyond
this point. If the time window for slow wave analysis was extended to 1000 ms
following picture onset, the group difference found in the slow waves would likely be
much more powerful than the present analyses indicate. Unfortunately, activity recorded
after 750 ms is confounded by the fact that a fixation point replaced the pictures at this
time, making it difficult to interpret this activity as purely reflecting sustained attention
and not some degree of response to the picture offset.
One final analysis was conducted to determine if the increased positivity seen in
the fear participants' slow waves for the threat pictures was associated with behavioral
symptomatology reported by the participants in the self-report measures. To accomplish
this, mean slow wave amplitudes for threat pictures were correlated with mean scores on
the SNAQ and SPQ, STAI-trait, STAI-state, and BDI. Correlations revealed that slow
wave responses to the threat pictures were moderately correlated to phobic symptoms
reported in the SNAQ/SPQ, r(27) = .363, p = .053, and the state anxiety reported in the
STAI, r(27) = .408, p < .05. Correlations further revealed associations between threat
slow waves and trait anxiety reported in the STAI, [r(27) = .417, p < .05 and depressive
symptoms reported in the BDI, r(27) = .402, p < .05. These findings suggest that
sustained attentional processing to interfering threat pictures is associated with behavioral
symptoms of negative affect (including specific fear), which may account for group
differences in WM impairment caused by threat picture interference.
To summarize the ERP results, WM load manipulation had a strong effect on the
neural activity devoted to processing the interfering pictures. This effect mirrors that
seen in the WM task data. Low WM load trials evoked more positive early and late LPPs
compared to the high load, suggesting that less attentional resources were available to
process the pictures during high load trials. Differential load-related activity was centro-
parietal in its scalp distribution; however, this effect diminished over time and was not
significant in the slow wave components. With regard to the valence effects of the
interfering pictures, LPPs had larger positive amplitudes for emotionally salient
compared to neutral interfering pictures. This also reflects effects that were seen in the
accuracy and RT data, suggesting that the greater the positivity of the interfering picture
response, the greater the disruption in the maintenance of the WM representation. All
ERP components showed greater centro-parietal positivity for negative pictures
(unpleasant and threat) compared to neutral. Slow waves were most positive for
unpleasant pictures in the centro-parietal region, yet controls and high-fear participants
differed in their slow waves such that high-fear participants showed more positive
activity for the interfering threat pictures than controls. The enhanced slow wave
positivity for threat pictures was associated with self-reported behavioral symptoms of
negative affect, including specific fear of the threatening content.
This study investigated the effects of affective interference on visual working
memory (WM). The main objectives were: 1) to determine if emotionally-arousing
interference degrades the active maintenance of task-relevant representations in WM and,
if so, 2) whether this degradation is disproportionately greater to interfering
evolutionarily-relevant threat stimuli. Additionally, this study sought to determine if
high-fear participants showed evidence of greater WM-related disruption to clinically-
relevant interfering stimuli. Evidence from WM task performance and
electrophysiological data converge to suggest that the representations in WM were
differentially modulated by task-irrelevant affective interference. Changes in task
performance attributable to both WM load and affective interference appear to be
associated with rapidly changing neural processes. Taken together, these findings raise
new questions about interactions between affect and higher-level cognitive processes like
WM, and provide insight into the disruptive nature of fear and anxiety.
Interference Effects on WM Task Performance
WM load had very strong effects on both accuracy and RT, with participants
making more errors and taking longer to respond to the probe during the high- than low-
load conditions. This indicates that participants had more difficulty with the more
challenging 3-diamond cues and engaged more WM resources to actively maintain
representations of the higher load stimuli. Valence of the interfering picture had different
group-related effects on accuracy and probe RT, as controls made more errors and took
less time to respond to the probe during trials that involved pleasant compared to neutral
picture interference; high-fear participants took more time to respond to the probe during
threat picture trials and showed a trend-level increase compared to controls in errors
during the threat and unpleasant relative to pleasant picture interference.
The finding that controls made more errors and took less time to respond during
pleasant picture trials is somewhat surprising at first; however, controls rated pleasant
pictures as the most arousing category of pictures, significantly higher than neutral.
Correlations between error rates and probe RTs for the pleasant picture trials were not
significant, suggesting that they did not exhibit a speed-accuracy trade-off for the
pleasant picture trials. Therefore, it appears instead that the controls' WM task
performance was influenced by arousal in such a way as to cause greater interference in
WM representations while at the same time facilitating faster processing of the probe.
Ashby (1999) proposed that moderate levels of positive affect enhance WM, but the
pleasant pictures used in this study may have been too arousing for this benefit to be
realized. This finding fits a pattern similar to that of the classic Yerkes-Dodson law
(Yerkes & Dodson, 1908) that performance is related to arousal in an inverted U-shaped
pattern such that too low or high arousal leads to impairment in cognitive performance.
In this light, the controls' arousal ratings may suggest that the pictures were too arousing
and may have overwhelmed the PFC with excessive levels of dopamine (Arnsten, 1998;
Goldman-Rakic, 1996), thereby disrupting WM function and leading to more impulsive
Task-related findings for high-fear participants fit some of the predictions that
their WM performance would be most impaired during trials that involved threat pictures.
Increased reaction time in probe responses may be indicative of a reduced cognitive
capacity to actively maintain the representation of the cue stimuli and/or execute the
recognition decision and response. Since high-fear participants rated threat pictures as
the most arousing, this pattern also appears to fit the pattern of the Yerkes-Dodson law.
However, it is interesting that this effect is opposite to that seen in controls where
pleasant pictures actually facilitated faster response to the probe. This may suggest that
the underlying neural processing involved in threat-based affective interference in high-
fear participants may be functionally different from those involved in pleasant affective
interference in controls. It seems plausible that this threat-based effect may be linked to a
greater level of disengaging attention bias invoked by the threatening stimuli (Fox et al.,
2001), which may not occur for pleasant pictures. This would account for the ERP
findings of more positive slow waves to threat pictures in high-fear participants.
Nevertheless, the dissociation between different valence-specific arousal effects cannot
be fully explained with these data alone.
Picture Response Data
Participants detected interfering threat pictures during the WM task faster than the
other categories, and high-fear participants took longer than controls to detect interfering
pictures as a whole. This is an interesting finding, as it shows that threat pictures
successfully captured the participants' attention better than any other type of picture -
regardless of the their level of fear and that higher levels of anxiety has a cost on overall
efficiency of attentional processing. This supports previous work that has found that
threat cues evoke automatic attention biases (Ohman, Flykt, & Esteves, 2001), and may
be explained by neural pathways that are present for both healthy and anxious individuals
that route threatening information directly to the amygdala for automatic processing
(Ledoux, 1996). It appears that the picture detection response (which involved a
keypress) may have made an impact on the late LPP component, as the peak latencies for
each valence group seem to coincide somewhat with the picture detection RTs.
While making their ratings at the end of the experiment, participants chose to
view unpleasant pictures the longest prior to making their ratings. One likely explanation
for this is that the unpleasant pictures were more complex and required more sustained
attention in order to rate them. Instead of simple pictures of snakes, smiling babies, or
non-affective items such as an ironing board, unpleasant pictures included some complex
scenes such as accidents, natural disasters, and saddening events. It is also possible that
they depicted more novelty and provoked more curiosity in the participants. Participants'
ratings of valence and arousal differentiated neutral, pleasant, and unpleasant pictures in
line with the expected patterns, revealing that the key manipulation of picture valence
was successfully executed. High-fear participants rated threat pictures as significantly
more arousing and less pleasant than controls, while controls rated pleasant pictures as
significantly more arousing than other picture categories.
ERP Reflections of Interference Stimulus Processing
WM Load Effects on Interference-Related Activity
As predicted, there was a clear load-related effect on ERPs to the task-irrelevant
interfering stimuli. Both the early and late LPP was smaller during the high- than low-
load condition. This finding suggests that fewer resources were available during the
high-load maintenance interval to process interfering stimuli and extends previous work
that observed a similar compromise of attentional resources with regard to picture
processing and startle modulation (Schupp et al., 1997). However, the WM load
modulation of picture-related LPPs decayed over time and did not affect the slow wave
ERP component. Load-related WM modulation of interference picture processing was
most pronounced for the earlier stages of attentional processing and,
electrophysiologically, was less reflected during later stages of more sustained attention.
Valence-Related Effects Interference Activity
As predicted, the valence characteristics of the interfering stimuli influenced the
amplitudes of the early and late LPP and slow wave. LPPs were larger for emotionally-
arousing than neutral interference pictures, which supports previous research that
affective pictures selectively engage attention based on motivational importance
(Cuthbert et al., 2000; Schupp et al., 2003, Baas et al., 2001). Additionally, all three ERP
components were larger in amplitude to negative (unpleasant and threat) than pleasant
pictures. This finding is consistent with the hypothesis that the unpleasant and threat
pictures recruited more attentional processing than did pleasant pictures. Slow waves
were most positive for unpleasant pictures, suggesting that unpleasant pictures engaged
more sustained attention than the other picture categories. Slow waves were very
responsive to differences in picture valence category, but they appeared to be cut short by
the offset of the picture; that is, there is a clear offset potential evoked by picture
termination, and this offset potential disrupted the slow wave making it difficult to
evaluate the duration of differential processing associated with emotional valence. A
significant Group x Valence interaction revealed that controls and high-fear participants
differed in their slow waves. High-fear participants showed more positive activity than
controls for threat pictures relative to unpleasant. Unpleasant pictures still evoked the
most positive slow waves in both groups, but high-fear participants exhibited
dramatically more sustained processing of the threat pictures compared to controls. This
finding may help to explain the increased WM impairment for high-fear participants
caused by the threat picture interference. Surprisingly, no ERP effects seemed to
correlate with the controls' WM impairments caused by the pleasant picture interference.
Interestingly, the scalp-voltage distribution maps shown in Figure 3-13 suggest
that emotionally arousing versus neutral scalp-voltage difference maps became
progressively more frontal over time. While this effect has not yet been quantitatively
examined, there are several potentially important implications. First, the choice of
electrode site for measuring valence-related effects may not have been optimal for
sensitively detecting differences in the different components. Second, the progressively
more frontal distribution over time may be consistent with the hypothesis that
emotionally-arousing interference stimuli both disrupt active maintenance of task-
relevant representations in WM and, furthermore, that representations of these task-
irrelevant interfering stimuli become preferentially processed over representations of the
task-relevant stimuli. That is, it is possible that representations of the task-irrelevant
interference stimuli may actually become task relevant and displace those of the task-
relevant stimuli in the dlPFC.
The data suggest that affective interference disrupts the active maintenance of
representations in WM and that this impairment is associated with the arousal
characteristics of the interfering stimuli. High-fear participants exhibited greater
allocation of sustained attention for threat pictures than controls, which corresponded
with a longer RT to make recognition decisions during the WM task. This may suggest
that high-fear participants have difficulty disengaging their attention from clinically-
relevant threat stimuli, which may lead to maintenance of their fear. Controls also
showed task-related patterns of WM impairment from interfering pleasant pictures
(which they rated as highly arousing), but these impairments did not show the neural
signature that was detected for an overall attention bias for threatening pictures or the
threat-induced WM impairments in high-fear participants.
Taken with the task performance data, the ERP results from this study suggest
several important findings. First, high- and low-fear individuals both show selective
processing of effectively arousing stimuli, even in the context of a challenging cognitive
task. Second, clinically-relevant threat interference has the capacity to disrupt neural
processes of sustained attention that are needed for higher-level cognitive processes such
as WM, and these changes in sustained attention are significantly associated with
behavioral symptoms of negative affect, including specific fear. Third, when attentional
resources are devoted to a challenging visual WM task, they are effectively subtracted
from the attentional capacity that is available for the processing of incoming affective
pictures, indicating that WM can diminish the degree of affective processing resources
that are available. Taken together, these ERP effects provide novel insights into neural
processes involved in attention, WM, and emotion that have not been previously
examined. Future studies are needed to clarify the nature of these cognitive-affective
interactions and expand these findings to other domains of cognitive and affective
Alternative Explanations and Possible Limitations
The results of this study offer new hope into better understanding nature of
affective interference of WM and the potential role that anxiety plays on higher-level
cognition. However, alternative explanations and possible limitations need to be
considered. The operational definitions associated with any form of emotion research are
complex. Lang (1998) has postulated that emotions can be measured on two dimensions:
valence and arousal. As a result, this study sought to define affective stimuli along these
two dimensions. However, it is possible that the pictures varied on other criteria that
were not controlled (e.g., brightness, complexity, human content) and may have affected
the outcome. It is also possible that the sample selected for participation may have had
life experiences that were not assessed that may have influenced their reactions to the
pictures. Group differences may have also been influenced by demand characteristics
that were placed on the high-fear participants (i.e., special attention given to them during
recruitment and interview because of their higher levels of fear). Alternatively, it is also
possible that the "high-fear" participants were not fearful enough for robust group effects
to emerge, as other studies have found individuals with spider or snake phobias score
much higher on the SNAQ/SPQ than the high-fear participants in this study. The mean
SNAQ/SPQ scores for high-fear participants in this study was 10.6, while snake and
spider phobics have obtained mean scores over 23 in other studies (e.g., Fredrikson,
Although this study repeated pictures during the WM task, it did not have a
sufficient number of trials to examine potential habituation effects; that is, repeated
exposure to each of the pictures may reasonably be assumed to be associated with
decreased arousal to subsequent exposures. However, given the limited number of trials
per condition, there was not adequate signal-to-noise ratio in the ERP data to examine
effects of repeated exposure. An additional concern is that the baseline accuracy
condition for comparing groups was measured as performance no-interference trials, yet
this condition was randomly mixed with the four categories of picture interference,
making the no-interference condition a bit of an "odd-ball" that may not truly represent
baseline WM performance. Lastly, this study had a small sample size that unfortunately
could not be completely balanced on gender in each group due to the low numbers of
available high-fear individuals.
Several additional potential limitations of the ERP data must be considered. First,
terms like "late positive potential" have somewhat limited interpretational value aside
from discussions of general attentional resources. Part of the reason for this is that
volume conduction of electrical signals in the brain makes it very difficult to map sources
of brain activity with precise spatial resolution, although new technologies are emerging
that make source localization more feasible and effective. A considerable amount of ERP
variability exists from participant to participant, and some participants exhibit the same
ERP components at slightly different latencies. Grand-averaging subjects together in
certain cases might actually average out a key component of interest. Also, due to the
number of trials that were available, no ERP analyses were performed on incorrect trials,
meaning that accuracy could not be considered as a variable in the ERP analysis.
Another concern emerged after analyzing the data taken from the centro-parietal
electrode, as the sensitivity to affective picture characteristics shifted more frontally and
may have been more pronounced if the slow wave had been analyzed from a more frontal
electrode site. It appears that 750 ms may not have not been a long enough picture
presentation to assess full slow wave effects. Finally, since analysis of the cue and probe
waveforms went beyond the scope of this project, it is difficult to completely interpret the
picture processing ERPs without a more complete understanding of the neural responses
to the task-relevant stimuli associated with the WM task.
Despite its limitations, this study provides important new information about the
interfering effects of emotional arousal on active maintenance of task-relevant
representations within WM. For the sake of complexity, and to address the primary aims
of the current research, task-relevant stimulus-evoked activity (i.e., cues and probes) was
not examined. These data will be examined in the future to more fully elucidate the role
of affective interference on WM-related processes. Additionally, as the arousal level
ii li/hi/ a valence category has been shown to modulate startle probe and LPP amplitudes
during emotional picture viewing (i.e., larger LPP and startle with higher arousal ratings
within category; Schupp et al., 1997), it would be useful to examine arousal-related
effects more parametrically. By necessity, in the current study, arousal and emotional
versus neutral stimuli are confounded. That is, pleasant and unpleasant stimuli were both
rated as high in arousal, while neutral stimuli were rated as low in arousal.
While localization of the neural sources giving rise to the observed interference
effects was not a primary aim of the study, such information will certainly be useful.
Findings by Perlstein et al. (2002) and Gray et al. (2002) have indeed shown that WM-
related dlPFC activity is modulated by emotionally arousing contexts. Future efforts will
be aimed at applying source localization routines to the current data using Brain Electric
Source Analysis (BESA; Scherg, 1990) to determine if the effects observed in the current
study are similarly localizable to the dlPFC. Also, the paradigm used for the present
study is currently being adapted for the functional MRI context, in order to facilitate
examination of brain activity that offers a higher degree of spatial resolution. Future
extensions of the project could also include measuring different types of specific fear or
anxiety (e.g., needle phobia, PTSD, etc.), to learn whether the findings in the current
study generalize to other clinical populations. Furthermore, this paradigm could
potentially be used to assess behavioral and electrophysiological effects of therapy that
focuses on reduction of fear symptoms in phobic individuals.
This study sought to investigate the effects of affective interference on visual
working memory (WM), testing whether affective interference impairs representations in
working memory (WM) and the degree to which this interference is due to effects of
arousal brought about by the clinical or evolutionary relevance of threat in participants
with high levels of fear. Performance data from the WM task revealed that WM
performance is modulated by affective interference and that arousal plays a key role in
that interference, causing high-fear participants to respond more slowly as a result of
threat interference and controls to make more errors and respond more quickly as a result
of pleasant picture interference. Subtle differences in sustained attention appear to be the
only source of group differences in the neural effects associated with threat-based
interference of WM, but no neural effects appear to explain the effects of pleasant picture
WM interference in controls. As a whole, participants demonstrated selective detection
of threatening stimuli, which corresponded to enhanced positivity in late positive
potentials measured at the centro-parietal region of the scalp. Importantly, ERP results
showed that increased WM load and the processing of affective interference
compromised the neural resources of attention. In the end, these findings help to better
illustrate the complexity of cognitive-affective interactions, and indicate that anxiety can
play a key role in disrupting higher-level cognitive processes involved in WM.
LIST OF REFERENCES
Adolphs, R., Gosselin, F., Buchanan, T.W., Tranel, D., Schyns, P., & Damasio, A.R.
(2005). A mechanism for impaired fear recognition after amygdala damage.
Nature, 433 (7021), 68-72.
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental
disorders (4th ed., text revision). Washington, DC: Author.
Armony, J.L. & Dolan, R.J. (2002). Modulation of spatial attention by fear-conditioned
stimuli: An event-related fMRI study. Neuropsychologia, 40, 817-826.
Armony J.L. & Ledoux, J.E. (1997). How the brain processes emotional information.
Annals of the New York Academy of Sciences, 821, 259-270.
Arnsten, A.F.T. (1998). Catecholamine modulation of prefrontal cortical cognitive
function. Trends in Cognitive Sciences, 2 (11), 436-447.
Ashby, F.G., Isen, A.M., & Turken, A.U. (1999). A neuropsychological theory of
positive affect and its influence on cognition. Psychological Review, 106 (3),
Baas, J.M.P., Kenemans, J.L., Bocker, K.B.E., & Verbaten, M.N. (2002). Threat-
induced cortical processing and startle potentiation. Neuroreport, 13, 133-137.
Baddeley, A.D. (1986). Working memory. Oxford: Oxford University Press.
Beck, A.T., Steer, R.A., & Brown, G.K. (1996). Manual for the BDI-II. San Antonio,
TX: The Psychological Coorperation.
Bertrand, 0., Perrin, F., & Pernier, J. (1985). A theoretical justification of the average-
reference in topographic evoked potential studies. Electroencephalography and
Clinical Neurophysiology, 62, 462-464.
Brown, T. A., DiNardo, P. A., & Barlow, D. H. (1994). Anxiety disorders interview
schedule for DSM-IV (ADIS-IV). Albany, NY: Graywind Publications.
Carlsson, K., Petersson, K.M., Lundqvist, D., Karlsson, A., Ingvar, M. & Ohman, A.
(2004). Fear and the amygdala: Manipulation of awareness generates differential
cerebral responses to phobic and fear-relevant (but not feared) stimuli. Emotion,
4 (4), 340-353.
Cohen, J.D., Perlstein, W.M., Braver, T.S., Nystrom, L.E., Noll, D.C., Jonides, J. &
Smith, E.E. (1997). Temporal dynamics of brain activation during a working
memory task. Nature, 386, 604-608.
Coles, M.G.H. & Rugg, M.D. (1995). Event-related brain potentials: an introduction. In
M.G.H. Coles & M.D. Rugg (Eds.), Electrophysiology of mind. Event-related
potentials and cognition (pp. 1-35). Oxford, England: Oxford University Press.
Compton, R.J. & Banich, M.T. (2003). Paying attention to emotion: An fMRI
investigation of cognitive and emotional Stroop tasks. Cognitive, Affective, &
Behavioral Neuroscience, 3 (2), 81-96.
Curtis, C.E. & D'Esposito, M. (2003). Persistent activity in the prefrontal cortex during
working memory. Trends in Cognitive Sciences, 7 (9), 415-423.
Cuthbert, B.N., Schupp, H.T., Bradley, M.M., Birbaumer, N., & Lang, P.J. (2000).
Brain potentials in affective picture processing: Covariation with autonomic
arousal and affective report. Biological Psychology, 52, 95-111.
Dolan, R.J. (2002). Emotion, cognition, and behavior. Science, 298, 1191-1194.
Donchin, E. & Coles, M.G. (1988). Is the P300 component a manifestation of context
updating? Behavioral and Brain Sciences, 11, 357-427.
Fabiani, M., Gratton, G., & Coles, M.G.H. (2000). Event-related potentials. In J.T.
Cacioppo, L.G. Tassinary & G.G. Bemston (Eds.), Handbook ofpsychphysiology
(2nd ed.). Cambridge, England: Cambridge University Press.
Fox, E. (1993). Allocation of visual attention and anxiety. Cognition and Emotion, 8,
Fox, E., Russo, R., Bowles, R., & Dutton, K. (2001). Do threatening stimuli draw or
hold visual attention in subclinical anxiety? Journal of Experimental Psychology:
General, 130 (4), 681-700.
Fuster, J.M. (1997). The prefrontal cortex: Anatomy, physiology, and neuropsychology
of the frontal lobe (3rd ed.). New York: Raven.
Gray, J.R., Braver, T.S., & Raichle, M.E. (2002). Integration of emotion and cognition
in the lateral prefrontal cortex. Proceedings of the National Academy of Sciences,
Goldman-Rakic, P.S. (1987). Circuitry of primate prefrontal cortex and regulation of
behavior by representational memory. In F. Plum (Ed.), Handbook of physiology;
The nervous system (pp. 373-414). Bathesda, MD: American Physiological
Goldman-Rakic, P.S. (1996). Regional and cellular fractionation of working memory.
Proceedings of the National Academy of Sciences, 93, 13473-13480.
Hermans, D., Vansteenwegen, D., & Eelen, P. (1999). Eye movement registration as a
continuous index of attention deployment: Data from a group of spider anxious
students. Cognition and Emotion, 13, 419-434.
Huynh, H. & Feldt, L.S. (1976). Estimation of the box correction for degrees of
freedom from sample data in the randomized block and split plot designs.
Journal of Educational Statistics, 1, 69 -82.
Ikeda, M., Iwanaga, M., & Seiwa, H. (1996). Test anxiety and working memory system.
Perceptual Motor \/Aill, 82, 1223-1231.
Johnson, V.S., Miller, D.R., & Burleson, M.H. (1986). Multiple P3s to emotional
stimuli and their theoretical significance. Psychophysiology, 23, 684-693.
Keil, A., Bradley, M. M., Hauk, 0., Rockstroh, B., Elbert, T., & Lang, P. J. (2002).
Large-scale neural correlates of affective picture processing. Psychophysiology,
Keppel, G. (1982). Design and analysis: A researcher's handbook, (2nd ed.). Englewood
Cliffs, NJ: Prentice-Hall.
Kindt, M. & Brosschot, J.F. (1997). Phobia-related cognitive bias for pictorial and
linguistic stimuli. Journal ofAbnormal Psychology, 106 (4), 644-648.
Klorman, R., Hastings, J.E., Weerts, T.C., Melamed, B.G., & Lang, P.J. (1974).
Psychometric description of some specific-fear questionnaires. Behavior
Therapy, 5, 401-409.
Kutas, M. & Dale, A. (1995). Electrical and magnetic readings of mental functions. In
M.D. Rugg (Ed.), Cognitive Neuroscience (pp 187-242). Cambridge, MA: MIT
Lane, R.D., Reiman, E.M., Bradley, M.M., Lang, P.J., Ahem, G.L., Davidson, R.J., &
Schwartz, G.E. (1997). Neuroanatomical correlates of pleasant and unpleasant
emotion. Neuropsychologia, 35, 1437-1444.
Lang, P. J. (1980). Behavioral treatment and bio-behavioral assessment: Computer
applications. In J. B. Sidowski, J. H. Johnson, & T. A. Williams (Eds.),
Technology in mental health care delivery systems (pp. 119-137). Norwood, NJ:
Lang, P.J., Bradley, M.M., & Cuthbert, B.N. (1997). Motivated attention: Affect,
activation, and action. In P. Lang, R.F. Simmons, & M. Balaban (Eds.), Attention
and orienting: Sensory and motivational processes: Hillsdale, NJ: Erlbaum
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1998). International affective picture
system (IAPS): Technical manual and affective ratings. Gainesville, FL: The
Center for Research in Psychophysiology, University of Florida.
Lavric, A., Rippon, G., & Gray, J.R. (2003). Threat-evoked anxiety disrupts spatial
working memory performance: An attentional account. Cognitive Therapy and
Research, 27 (5), 489-504.
Ledoux, J. E. (1996). The emotional brain. New York, NY: Simon & Schuster.
Low, A., Rockstroh, B., Harsch, S., Berg, P., & Cohen, R. (2000). Event-related
potentials in a working-memory task in schizophrenics and controls.
Schizophrenia Research, 46, 175-186.
Luck, S.J., Woodman, G.F., & Vogel, E.K. (2000). Event-related potential studies of
attention. Trends in Cognitive Sciences, 4 (11), 432-440.
MacLeod, C.M. (1991). Half a century of research on the Stroop effect: An integrative
review. Psychological Bulletin, 190, 163-203.
MacLeod, C.M, Matthews, A., & Tata, P. (1986). Attentional bias in emotional
disorders. Journal ofAbnormal Psychology, 95, 15-20.
Miller, E.K. & Cohen, J.D. (2001). An integrative theory of prefrontal cortex function.
Annual Review ofNeuroscience, 24, 167-202.
Morris, J.S., Frith, C.D., Perrett, D.O., Rowland, D., Young, A.W., Calder, A.J., &
Dolan, R.J. (1996). A differential neural response in the human amygdala to
fearful and happy facial expressions. Nature, 383 (6603), 812-815.
Ohman, A., Flykt, A., & Esteves, F. (2001). Emotion drives attention: Detecting the
snake in the grass. Journal of Experimental Psychology: General, 130 (3), 466-
Ohman, A. & Mineka, S. (2001). Fear, phobias, and preparedness: Toward an evolved
module of fear and fear learning. Psychological Review, 108, 483-522.
Ohman, A. & Mineka, S. (2003). The malicious serpent: Snakes as a prototypical
stimulus for an evolved module of fear. Current Directions in Psychological
Science, 12 (1), 5-9.
Ohman, A. & Soares, J.J.F. (1998). Emotional conditioning to masked stimuli:
Expectancies for aversive outcomes following nonrecognized fear-relevant
stimuli. Journal of Experimental Psychology: General, 127, 69-82.
Ohman, A., & Soares, J.J.F. (1993). On the automatic nature of phobic fear:
Conditioned electrodermal responses to masked fear-relevant stimuli. Journal of
Abnormal Psychology, 102. 121-132.
Ohman, A., & Soares, J.J.F. (1994). Unconscious anxiety: Phobic responses to masked
stimuli. Journal ofAbnormal Psychology, 103, 231-240.
Perlstein, W.M., Elbert, T., & Stenger, V.A. (2002). Dissociation in human prefrontal
cortex of affective influences on working memory-related activity. Proceedings
of the National Academy of Sciences, 99, 1736-1741.
Posner, M.I., Inhoff, A.W., Friedrich, F.J., & Cohen, A. (1987). Isolating attentional
systems: A cognitive-anatomical analysis. Psychobiology, 15, 107-121.
Ratcliff, R. (1993). Methods for dealing with reaction time outliers. Psychological
Bulletin, 114, 510-532.
Runchkin, D.S., Johnson, R., Mahaffey, D., & Sutton, S. (1988). Toward a functional
categorization of slow waves. Psychophysiology, 25, 339-353.
Sadock, B.J. & Sadock, V.A. (2003). Kaplan and Sadock's synopsis of psychiatry:
Behavioral sciences/clinical psychiatry (9th ed.). Philadelphia, PA: Lippincott
Williams & Wilkins.
Scherg, M. (1990). Fundamentals of dipole source potential analysis. In F. Grandori &
M. Hoke (Eds.), Auditory evoked magnetic fields and electric potentials.
Advances in audiology (Vol. 6, pp. 65-78). Basel: Karger.
Schimmack, U. (2005). Attentional interference effects of emotional pictures: Threat,
negativity, or arousal? Emotion, 5 (1), 55-66.
Schupp, H.T., Cuthbert, B.N., Bradley, M.M., Birbaumer, N., & Lang, P.J. (1997).
Probe P3 and blinks: Two measures of affective startle modulation.
Psychophysiology, 34, 1-6.
Schupp, H.T., Junhhofer, M., Weike, A.I., & Hamm, A.O. (2003). Emotional
facilitation of sensory processing in the visual cortex. Psychological Science, 14
Schupp, H.T., Junhhofer, M., Weike, A.I., & Hamm, A.O. (2004). The selective
processing of briefly presented affective pictures: An ERP analysis.
Psychophysiology, 41, 441-449.
Smith, E.E. & Jonides, J. (1999). Storage and executive processes in the frontal lobes.
Science, 283, 1657-1661.
Spielberger, C. S., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the state trait
anxiety inventory. Palo Alto, CA: Consulting Psychologists Press.
Tucker, D.M., Liotti, M., Potts, G.F., Russell, G.S., & Posner, M.I. (1994).
Spatiotemporal analysis of brain electrical fields. Human Brain Mapping, 1, 134-
Vrana, S., Roodman, A., & Beckham, J. (1995). Selective processing of trauma-relevant
words in post-traumatic stress disorder. Journal ofAnxiety Disorders, 9, 515-530.
Williams, J.M.G., Watts, F.N., Macleod, C., & Matthews, A. (1997). Cognitive
psychology and emotional disorders, (2nd ed.). Chichester, England: Wiley.
Williamson, S.J. & Kaufman, L. (1990). Theory of neuroelectric and neuromagnetic
fields. In F. Grandor, M. Hokle, & G.L. Romani (Eds.), Auditory Evoked
Magnetic Fields and Electric Potentials. Advances in Audiology (Vol. 6, pp. 1-
39). Basel, Switzerland: Karger Publications.
Yamasaki, H., LaBar, K.S., & McCarthy, G. (2002). Dissociable prefrontal brain
systems for attention and emotion. Proceedings of the National Academy of
Sciences, 99, 11447-11451.
Yerkes, R.M. & Dodson, J.D. (1908). The relation of strength of stimulus to rapidity of
habit-formation. Journal of Comparative Neurology and Psychology, 18, 459-
David Stigge-Kaufman received his B.A. degree from Bethel College in North
Newton, Kansas, in 1998, majoring in biology and psychology. He plans to receive his
M.S. from the University of Florida in 2005, and then plans to continue his doctoral study
in clinical psychology, concentrating his clinical and research training in