Download NEURAL MECHANISMS SUPPORTING THE LEARNING

NEURAL MECHANISMS SUPPORTING THE
LEARNING-RELATED EMOTIONAL RESPONSE TO A THREAT
by
KIMBERLY H. WOOD
DAVID C. KNIGHT, COMMITTEE CHAIR
EDWIN W. COOK, III
RAJESH K. KANA
ADRIENNE C. LAHTI
KRISTINA M. VISSCHER
A DISSERTATION
Submitted to the graduate faculty of The University of Alabama at Birmingham,
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy
BIRMINGHAM, ALABAMA
2013
Copyright by
Kimberly H. Wood
2013
NEURAL MECHANISMS SUPPORTING THE
LEARNING-RELATED EMOTIONAL RESPONSE TO A THREAT
KIMBERLY H. WOOD
BEHAVIORAL NEUROSCIENCE
ABSTRACT
Successful regulation of the emotional response to a threat allows one to react
more effectively under threatening conditions. The prefrontal cortex (PFC) and amygdala
are key brain regions that mediate the regulation and expression of emotion. We
employed Pavlovian fear conditioning to investigate the neural mechanisms that
influence the emotional response to a threat. These procedures were designed to
investigate conditioned diminution of the unconditioned response (UCR). The specific
aims were to better understand the role of associative learning, expectation,
controllability, and predictability in modulating UCR expression.
This project employed functional magnetic resonance imaging (fMRI) to assess
the magnitude of the threat-related response within the PFC, cingulate cortex, inferior
parietal lobule, insula, amygdala, and hippocampus. We also investigated the peripheral
expression of emotion indexed via skin conductance response (SCR) and startle eye-blink
electromyography (EMG) during differential fear conditioning. To assess the effect of
expectation of an impending threat, volunteers provided a continuous self-report measure
of UCS expectancy throughout the conditioning sessions. We also examined whether
individual differences in anxiety level influenced the emotional response to a threat.
In general, we observed a relationship between anxiety level and the threat-related
neurophysiological response. Conditioned UCR diminution within the neurophysiological
response was also observed. More specifically, the threat-related fMRI signal response
iii
and SCR expression was diminished on predictable vs. unpredictable trials. However, the
opposite pattern was observed in the EMG data. An enhanced startle-eyeblink response
was observed for predictable compared to unpredictable trials. Further, controllability
affected the threat-related fMRI signal response within the ventromedial PFC and
hippocampus. The unconditioned SCR elicited by the threat paralleled the fMRI signal
response within several brain regions that showed UCR diminution. A negative
relationship was observed between UCS expectancy and the threat-related response
within several brain regions that showed conditioned UCR diminution. In summary, we
observed learning-related changes in the emotional response to a threat within regions of
the PFC, amygdala, and hippocampus. The current findings suggest that these brain areas
support learning-related processes that modulate the emotional response to a threat.
Keywords: fMRI, emotion, fear conditioning, threat response, prefrontal cortex,
amygdala
iv
DEDICATION
This dissertation is dedicated to my husband and daughters, Robert, Krista, and Kayli
Wood. Thank you for the encouragement and support during this academic journey over
the last fourteen years. I could not have done this without you!
v
ACKNOWLEDGEMENTS
Thank you to my mentor Dr. David C. Knight for his guidance and generosity throughout
my graduate career.
Thanks to my dissertation committee for their time and feedback in the development of
this project.
I appreciate all past and present members of our lab. Thank you all for the much needed
laughs.
Special thanks to Josh and Muriah for their contribution to data
acquisition and analysis.
I would like to thank the Behavioral Neuroscience Directors and past and present
students. To Drs. Randich and Amthor, it has been such a privilege to be under
your academic advisement and your student. Thank you to the Behavioral
Neuroscience students for your advice.
Thank you to the Department of Psychology for their administrative and financial support
during my graduate training.
A special thanks to my husband and daughters, Robert, Krista, and Kayli for their love
and encouragement throughout my academic career. Words cannot express how
truly thankful I am to have such an amazing family.
Thank you to all my family, church family, and friends for your support and
encouragement throughout my graduate career.
vi
TABLE OF CONTENTS
Page
ABSTRACT ....................................................................................................................... iii
DEDICATION .....................................................................................................................v
ACKNOWLEDGEMENTS ............................................................................................... vi
LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... ix
LIST OF ABBREVIATIONS ............................................................................................ xi
INTRODUCTION ...............................................................................................................1
Modulation of the UCR ............................................................................................2
PFC-amygdala network ...........................................................................................5
Controllability and predictability ............................................................................6
Anticipation and anxiety ..........................................................................................7
Specific Aims ............................................................................................................9
NEURAL MECHANISMS UNDERLYING THE CONDITIONED
DIMINUTION OF THE UNCONDITIONED FEAR RESPONSE..................................12
NEURAL SUBSTRATES UNDERLYING LEARNING-RELATED
CHANGES OF THE UNCONDITIONED FEAR RESPONSE .......................................58
CONTROLLABILITY AND PREDICTABILITY DIMINISH
THE NEURAL RESPONSE TO A THREAT...................................................................93
SUMMARY .....................................................................................................................124
GENERAL LIST OF REFERENCES .............................................................................128
APPENDIX: IRB APPROVAL FORM..........................................................................134
vii
LIST OF TABLES
Table
Page
NEURAL MECHANISMS UNDERLYING THE CONDITIONED
DIMINUTION OF THE UNCONDITIONED FEAR RESPONSE
1
Regions showing conditioned diminution of the UCR ................................................46
2
Regions showing change over time .............................................................................47
3
Regional activity varying with trait anxiety.................................................................48
4
Regions showing a relationship between
anticipatory and threat-related activity ........................................................................49
NEURAL SUBSTRATES UNDERLYING LEARNING-RELATED
CHANGES OF THE UNCONDITIONED FEAR RESPONSE
1
Regions showing conditioned diminution of the UCR ............................................... 85
2
Regions showing potentiation of the UCR ..................................................................86
3
Regions showing change over time .............................................................................87
4
Regions showing a relationship between
anticipatory and threat-related activity ........................................................................88
CONTROLLABILITY AND PREDICTABILITY DIMINISH
THE NEURAL RESPONSE TO A THREAT
1
Demographics and group characteristics ...................................................................117
2
Regions showing conditioned diminution of the UCR ..............................................118
3
Regional activity varying with state anxiety ..............................................................119
viii
LIST OF FIGURES
Table
Page
NEURAL MECHANISMS UNDERLYING THE CONDITIONED DIMINUTION
OF THE UNCONDITIONED FEAR RESPONSE
1
Conditioning procedure ...............................................................................................50
2
UCS expectancy and unconditioned SCR....................................................................51
3
UCR diminution within the fMRI signal response ......................................................52
4
Stimulus x trial interaction within the ventromedial PFC ...........................................53
5
Trait anxiety and the unconditioned fMRI signal response .........................................54
6
Relationship between amygdala and unconditioned SCR ...........................................55
7
Relationship between anticipatory and threat-related activity .....................................56
NEURAL SUBSTRATES UNDERLYING LEARNING-RELATED
CHANGES OF THE UNCONDITIONED FEAR RESPONSE
1
Conditioned and unconditioned stimuli ...................................................................... 89
2
UCS expectancy and unconditioned SCR....................................................................90
3
UCR diminution within the fMRI signal response ......................................................91
4
Relationship between anticipatory and threat-related activity .............................................92
CONTROLLABILITY AND PREDICTABILITY DIMINISH
THE NEURAL RESPONSE TO A THREAT
1
Acquisition phase .......................................................................................................120
2
UCS Expectancy, unconditioned SCR, and EMG response ......................................121
ix
3
Conditioned UCR diminution within the fMRI signal response ...............................122
4
Regions showing predictability x controllability interaction .....................................123
x
LIST OF ABBREVIATIONS
CS
conditioned stimulus
CS+
CS paired with the unconditioned stimulus
CS−
CS presented alone
CR
conditioned response
SCR
skin conductance response
EMG
electromyography
UCR
unconditioned response
UCS
unconditioned stimulus
xi
INTRODUCTION
Fear is considered an important defense mechanism due to its evolutionary role in
survival (Kim & Jung, 2006; LeDoux, 2003). Consequently, certain environmental
stimuli have become innately hardwired over our evolutionary history to induce fear (e.g.
loud noises, darkness). However, fear can also be rapidly associated with neutral stimuli,
thereby permitting animals to adapt to an ever changing environment (Domjan, 2005;
Kim & Jung, 2006; LeDoux, 2003). This adaptation of the fear response has been
observed in a wide range of species and response systems using Pavlovian fear
conditioning (Davis, 1992; Domjan, 2005; Helmstetter & Bellgowan, 1994; Kim & Jung,
2006). Further, the ability to form associations between a dangerous event and the cues
that predict it allows an organism to more effectively minimize the impact of an
impending threat. For example, conditioned hypoalgesia, (ie. decreased pain sensitivity)
is observed to painful stimuli during Pavlovian conditioning (Bellgowan & Helmstetter,
1996; Helmstetter, 1992).
This reduction in the response to a threat is also observed during Pavlovian fear
conditioning when the unconditioned response (UCR) is diminished to predictable
compared to unpredictable presentations of an unconditioned stimulus (UCS) (Domjan,
2005). From a functional perspective, it is the response to the threat itself (i.e. the UCS)
that is the most important component of Pavlovian conditioning (Domjan, 2005; Pavlov,
1927). Further, fear responses in Pavlovian conditioning closely resemble characteristic
traits of human anxiety disorders, thus, understanding the biological mechanisms of fear
1
conditioning may elucidate the behavioral and physiological abnormalities of emotion
regulation in fear-related disorders (Davis, 1992; Kim & Jung, 2006; LeDoux, 2007;
Maren, 2001; Milad et al., 2007).
Pavlovian conditioning, often referred to as classical conditioning, is one of the
oldest and most straightforward paradigms to study fear-related processes (Domjan,
2005; LeDoux, 1998; Pavlov, 1927). It is considered a model system to investigate the
neurobiological mechanisms of learning (Fanselow & Ledoux, 1999; Helmstetter &
Bellgowan, 1993; Kim & Jung, 2006; Maren, 2001). During Pavlovian fear conditioning,
a neutral conditioned stimulus (CS) is paired with an aversive UCS. The conditioned
response (CR) produced by the CS is often used to index fear expression. Traditionally,
CR expression is taken as evidence that an association between the CS and UCS has been
formed. In contrast, the UCR is often considered an automatic, reaction to the aversive
UCS that does not require associative learning. Although CR expression is evidence of
adapting to environmental change, there are also associative learning-related changes in
the UCR to the threat itself. Further, learning-related changes in the UCR, produced by
the UCS, directly impacts survival and therefore may be the most biologically relevant
feature of Pavlovian conditioning (Domjan, 2005; Pavlov, 1927). Thus, a better
understanding of the neural mechanisms that mediate learning-related changes in the
UCR is warranted.
Modulation of the UCR
Learning-related changes in the UCR have been observed in a variety of response
systems. These changes can be exhibited as an increase or decrease in the UCR to the
2
UCS. For example, prior work has shown potentiation of the startle response during fear
conditioning (Grillon et al., 1991), while unconditioned skin conductance response (SCR)
diminishes as associative learning develops during Pavlovian fear conditioning (Baxter,
1966; Kimmel, 1967; Marcos & Redondo, 1999) . More specifically, the magnitude of
the unconditioned SCR decreases as the CS and UCS are repeatedly paired (Baxter,
1966). Although some of this early work could be influenced by habituation of the UCR,
other research indicates UCR amplitude is decreased to paired compared to unpaired
presentations of the CS and UCS (Kimmel, 1967), and UCR magnitude is smaller to
predictable compared to unpredictable UCS presentations (Lykken et al., 1972; Peeke &
Grings, 1968). These findings indicate the reduction in UCR amplitude during Pavlovian
conditioning cannot be solely explained by a simple non-associative learning process (i.e.
habituation). Instead, the findings suggest that presentation of the CS+ (i.e. stimulus that
predicts the UCS) modulates UCR expression via associative learning processes (Baxter,
1966; Kimmel, 1967; Knight et al., 2010, 2011; Marcos & Redondo, 1999). This
phenomenon is generally referred to as conditioned UCR diminution.
UCR diminution has been investigated using differential conditioning procedures
that consist of a training session in which one CS is paired with the UCS (CS+) and a
second CS is presented alone (CS−), followed by a testing session where both the CS+
and CS− are paired with the UCS (Knight et al., 2011; Marcos & Redondo, 1999; Wood
et al., 2012). This work has demonstrated greater UCR diminution when the UCS follows
the CS+ compared to when the UCS follows the CS− on test trials (Knight et al., 2011;
Marcos & Redondo, 1999; Wood et al., 2012). This prior work indicates that UCR
diminution is in part mediated by an associative learning process in which the CS+ gains
3
discriminative control over the UCR (Baxter, 1966; Kimmel, 1967; Knight et al., 2010;
Marcos & Redondo, 1999). Others have suggested that conscious expectations modify
UCR expression (Dunsmoor et al., 2008; Knight et al., 2010; Rust, 1976). For example,
greater UCR diminution has been observed when participants expect a UCS compared to
when the UCS is unexpected (Dunsmoor et al., 2008; Knight et al., 2010). Additionally,
graded increases in UCS expectancy are paralleled by graded decreases in unconditioned
SCR magnitude (Dunsmoor et al., 2008; Knight et al., 2010). These findings suggest that
associative learning processes and conscious UCS expectancies modulate the expression
of UCRs.
Few brain imaging studies have used functional magnetic resonance imaging
(fMRI) to investigate the neural substrates that support conditioned UCR diminution. In
this previous research, UCR diminution has been observed within the fMRI signal
response of the dorsolateral prefrontal cortex (dlPFC), anterior cingulate cortex (ACC),
and amygdala (Dunsmoor et al., 2008; Wood et al., 2012). Further, these studies also
found that UCS expectancy varied with the amplitude of the fMRI signal response within
several brain regions that showed learning-related changes (Dunsmoor et al., 2008; Wood
et al., 2012). For example, UCS expectancy increased during the CS and the amplitude of
the fMRI signal response to the UCS within these brain regions decreased (Dunsmoor et
al., 2008). Similar findings have been observed within ventromedial (vmPFC) and
dorsomedial (dmPFC) prefrontal cortices as well (Knight et al., 2010; Wood et al., 2012).
These studies also found that when the magnitude of the unconditioned fMRI signal
response within the dmPFC, dlPFC, and insula increased, a larger autonomic response
(e.g. SCR) was produced. Similar to the learning-related changes observed within the
4
fMRI signal response, as UCS expectancy increased the magnitude of unconditioned
SCRs decreased (Knight et al., 2010). These findings suggest that diminution of the
behavioral UCR is, at least, partly mediated by via prefrontal brain regions that support
expectancy-related processes (Dunsmoor et al., 2008; Knight et al., 2010; Wood et al.,
2012).
PFC-amygdala network
Prior work has implicated interconnected dorsal and ventral neural systems that
independently process the properties of stimuli (Kahnt et al., 2011; Morgane et al., 2005),
including emotionally valenced stimuli (Viinikainen et al., 2010). The ventral system
consists of ventral regions of the PFC such as the orbitofrontal cortex (OFC) and vmPFC,
as well as insular cortex, and subgenual and pregenual regions of the ACC (Phillips et al.,
2003). The ventral system also includes subcortical brain structures such as the amygdala,
ventral striatum, thalamus, and hypothalamus (Phillips et al., 2003). The ventral system is
essential in the assessment of the emotional significance of stimuli within the
environment and the production of the emotional response (Phillips et al., 2003). The
dorsal system consists of dmPFC, dlPFC, dorsal and caudal regions of the ACC, and the
hippocampus (Phillips et al., 2003). The dorsal system is responsible for integrating
cognitive processes and executive planning functions that can be influenced by emotional
input (Büchel et al., 1998; Phillips et al., 2003). The dorsal and ventral systems also differ
in the type of responses produced. The dorsal system is important for effortful cognitive
regulation of emotional responses, whereas the ventral system contributes to the
generation of automatic responses (Phillips et al., 2003). However, the role of these
5
dorsal and ventral systems in the modulation of the UCR is currently unknown.
These dorsal and ventral neural systems also play an important role in human fear
conditioning (Phillips et al., 2003). During fear acquisition, ACC activity increases as the
CS and UCS pairing rate increases (Dunsmoor et al., 2007). The amygdala also plays a
key role in forming the CS and UCS association (Helmstetter, 1992; LeDoux, 2007), as
well as detecting changes in the relationship between the CS and UCS (Knight et al.,
2004). Further, the PFC modulates subcortical brain regions, like the amygdala, which
controls expression of the emotional response during fear conditioning (Delgado et al.,
2008; Knight et al., 2005). For example, prior work has demonstrated similarities in the
pattern of activity within the fMRI signal response of the dmPFC, dlPFC, and vmPFC
that correspond with SCR expression during fear conditioning (Knight et al., 2010).
These findings demonstrate learning-related changes within the dorsal and ventral neural
systems during human Pavlovian fear conditioning. However, it remains unclear as to
how these brain areas specifically contribute to conditioned diminution of the UCR.
Controllability and predictability
Prior work suggests that top-down mechanisms support contingency learning and
emotional regulation (Dunsmoor et al., 2008; Kim & Jung, 2006; Knight et al., 2010),
and that both cortical and subcortical brain regions process aversive stimuli (Delgado et
al., 2008; Milad et al., 2004). Cortical and subcortical brain regions (e.g. dlPFC, ACC,
insula, and amygdala) also respond differentially based on the ability to control a UCS in
both humans (Salomons et al., 2004, 2007; Wiech et al., 2006) and animals (Amat et al.,
2005; Baratta et al., 2007, 2008; Maier et al., 2006). Prior animal research suggests that
6
the medial PFC (mPFC) inhibits the response of subcortical brain areas (e.g. the
amygdala) when a UCS is controllable (Baratta et al., 2007; Foa et al., 1992; Maier et al.,
2006). More importantly, exposure to a controllable UCS alters the impact of that UCS
when encountered in the future (Amat et al., 2005; Baratta et al., 2007; Maier et al.,
2006). For example, exposure to a controllable UCS interferes with subsequent fear
conditioning, whereas exposure to an uncontrollable UCS potentiates the conditioned
fear-response (Baratta et al., 2007; Maier et al., 2006). Similar investigations of
controllability have been conducted in neuroimaging studies using fMRI. However, in
this prior work participants received trials of both controllable and uncontrollable
presentations of a painful stimulus (Salomons et al., 2004, 2007; Wiech et al., 2006).
These studies found greater activation within dlPFC, ACC, insula, and the amygdala in
response to an uncontrollable UCS compared to a controllable UCS. Additionally,
increased activation within dlPFC, ACC, and vmPFC has been observed in anticipation
of an uncontrollable UCS compared to a controllable UCS (Salomons et al., 2007).
However, these studies did not include unpredictable presentations of controllable and
uncontrollable aversive stimuli (Salomons et al., 2004, 2007; Wiech et al., 2006).
Therefore, it remains unclear as to how the ability to control a UCS during Pavlovian fear
conditioning may influence the PFC-amygdala circuit and effect UCR diminution.
Anticipation and anxiety
A number of neuroimaging studies have investigated anticipation of aversive
events using a variety of paradigms (Critchley et al., 2001; Herwig et al., 2007; Nitschke
et al., 2006; Volz et al., 2003). These studies have suggested the amygdala, dlPFC, insula,
and ACC support anticipatory processes (Schienle et al., 2010). For example, the ACC
7
and amygdala respond to cues that predict aversive stimuli during fear conditioning
(Büchel et al., 1998; Davis & Whalen, 2001; Knight et al., 1999; Nitschke et al., 2006;
Phillips et al., 2003) and show increased activity when participants are not certain of the
outcome (Critchley et al., 2001; Davis & Whalen, 2001; Dunsmoor et al., 2007; Volz et
al., 2003). Anticipation of a possible threat occurring in the future is a key feature of
anxiety-related disorders (Davis et al., 2009; Grillon, 2002; Nitschke et al., 2009, 2006).
Therefore, human neuroimaging studies have also investigated the PFC-amygdala circuit
as a function of anxiety. For example, during cued and contextual fear conditioning,
recruitment and sustained activation of vmPFC was observed among participants with
low trait anxiety, but not participants with high trait anxiety (Indovina et al., 2011).
Furthermore, vmPFC activity was inversely related to fear conditioned SCRs for
participants with low trait anxiety, but not those with high trait anxiety (Indovina et al.,
2011). Prior work has also shown that anxiety level influences ACC activity that is
required to attend and respond to threatening stimuli (Klumpp et al., 2011). Specifically,
high trait anxiety was associated with decreased activation within the ACC (Klumpp et
al., 2011; Sehlmeyer et al., 2011) in conjunction with an exaggerated amygdala response
(Sehlmeyer et al., 2011). Taken together, these findings support previous research that
suggests that anxiety level influences top-down regulatory control of the fear response
(Kim et al., 2011; Milad et al., 2009; Rauch et al., 2006). Given that anxiety level affects
anticipatory reactivity and the response to aversive stimuli, differences in the level of
anxiety individuals experience may also influence UCR diminution.
8
Specific Aims
In summary, prior Pavlovian conditioning research suggests that regions of the
PFC modulate the emotional response controlled by the amygdala (Delgado et al., 2008;
Kim & Jung, 2006; Milad et al., 2006, 2009; Rauch et al., 2006). Therefore, PFC
modulation of the amygdala may also mediate learning-related changes in UCR
expression during fear conditioning (Dunsmoor et al., 2008; Knight et al., 2010).
However, there remains a critical gap in our understanding of the mechanisms that
contribute to conditioned diminution of the UCR. Specifically, partially independent
associative learning and expectancy-related processes appear to influence UCR
diminution (Kimmel, 1967; Marcos & Redondo, 1999). However, the neural circuitry
that supports these processes remains unclear. This project used a multi-modal approach
to investigate learning-related changes in the emotional response evoked by a threat. We
combined neurophysiological, behavioral, and self-assessment measures to examine UCR
diminution during Pavlovian fear conditioning. These measures included a continuous
self-report of conscious UCS expectancy, unconditioned SCR, and startle eye-blink
electromyography (EMG) response produced by the UCS. By incorporating these indices,
the effect of UCS expectation on the magnitude of brain (i.e. fMRI) and autonomic
responses (i.e. SCR and EMG) to the UCS can be observed. Additionally, self-assessment
measures (i.e. anxiety level) were used to evaluate processes that affect UCR diminution.
The primary objective of this project was to attain a better understanding of the
neurophysiological mechanisms that effect conditioned UCR diminution using
Pavlovian fear conditioning. These studies investigated the role that associative
learning, UCS expectancy, controllability, predictability, and anxiety level play in
9
the conditioned diminution of the UCR. The central hypothesis was that the
magnitude of unconditioned dlPFC, dmPFC, vmPFC, and amygdala response is
mediated by associative learning, expectation, controllability, and predictability of
the UCS. We also expected that anxiety level would influence the magnitude of the
threat-related neurophysiological response.
Manuscript 1 – In this study, we used fMRI to compare threat-related brain
activity and SCR expression to predictable and unpredictable presentations of the UCS
using CSs that were easy to discriminate. This differential conditioning procedure was
followed by a test phase that consisted of UCS presentations that followed the CS+ and
CS−, as well as presentations of the UCS alone to assess associative learning and
expectancy-related processes that support conditioned UCR diminution. We expected to
observe learning-related changes in the unconditioned fMRI signal response within
regions of the PFC and amygdala that influence the emotional response (indexed via
SCR). We also assessed the relationship between threat-related activity within brain
regions that showed UCR diminution and individual differences in trait anxiety.
Manuscript 2 – The purpose of this study, was to better understand the neural
substrates that support associative learning processes that influence UCR expression in
the absence of differential UCS expectancies. The differential conditioning procedure
was also followed by a test phase that consisted of UCS presentations that followed the
CS+ and CS−, as well as presentations of the UCS alone. However, this paradigm used
CSs that were difficult to discriminate to assess associative learning independent of
10
expectancy-related processes that influence the threat-related fMRI signal and SCR
expression. We hypothesized UCR diminution would be observed independent of
conscious UCS expectancies. Further, we expected to observe learning-related changes
within the PFC-amygdala circuit that mediate the emotional response (indexed via SCR).
Manuscript 3 – In this study, we used predictable and unpredictable
presentations of the UCS to investigate conditioned UCR diminution. We also assessed
the effect of UCS controllability on UCR diminution. For this study, there were two
groups that consisted of yoked pairs where one group (Controllable Condition) could
terminate the UCS, and the other group (Uncontrollable Condition) could not terminate
the UCS. We also assessed the influence of UCS expectancy and state anxiety in the
modulation of the threat-related fMRI signal, SCR production, and EMG response. We
hypothesized that controllability and predictability would influence the magnitude of the
unconditioned neurophysiological response. Further, we expected to observe a
relationship between anxiety level and the magnitude of the threat-related emotional
response produced within the PFC-amygdala circuit, as well as emotional expression
(indexed via SCR and EMG).
11
12
NEURAL MECHANISMS UNDERLYING THE CONDITIONED DIMINUTION
OF THE UNCONDITIONED FEAR RESPONSE
by
KIMBERLY H. WOOD, LAWRENCE W. VER HOEF, AND DAVID C. KNIGHT
NeuroImage 60, 787–799
Copyright
2012
by
Elsevier
Used by permission
Format adapted and errata corrected for dissertation
Abstract
Recognizing cues that predict an aversive event allows one to react more effectively
under threatening conditions, and minimizes the reaction to the threat itself. This is
demonstrated during Pavlovian fear conditioning when the unconditioned response
(UCR) to a predictable unconditioned stimulus (UCS) is diminished compared to the
UCR to an unpredictable UCS. The present study investigated the functional magnetic
resonance imaging (fMRI) signal response associated with Pavlovian conditioned UCR
diminution to better understand the relationship between individual differences in
behavior and the neural mechanisms of the threat-related emotional response. Healthy
volunteers participated in a fear conditioning study in which trait anxiety, skin
conductance response (SCR), UCS expectancy, and the fMRI signal were assessed.
During acquisition trials, a tone (CS+) was paired with a white noise UCS and a second
tone (CS−) was presented without the UCS. Test trials consisted of the CS+ paired with
the UCS, CS− paired with the UCS, and presentations of the UCS alone to assess
conditioned UCR diminution. UCR diminution was observed within the dorsolateral
PFC, dorsomedial PFC, cingulate cortex, inferior parietal lobule (IPL), anterior insula,
and amygdala. The threat-related activity within the dorsolateral PFC, dorsomedial PFC,
posterior cingulate cortex, and inferior parietal lobule varied with individual differences
in trait anxiety. In addition, anticipatory (i.e. CS elicited) activity within the PFC showed
an inverse relationship with threat-related (i.e. UCS elicited) activity within the PFC, IPL,
and amygdala. Further, the emotional response (indexed via SCR) elicited by the threat
was closely linked to amygdala activity. These findings are consistent with the view that
13
the amygdala and PFC support learning-related processes that influence the emotional
response evoked by a threat.
Key words: fMRI, learning, conditioning, unconditioned response, amygdala, prefrontal
cortex, emotion, fear, anxiety, skin conductance
14
Introduction
The ability to identify and successfully respond to dangers within one’s
environment is critical to survival. Learning the relationship between a threat and the
cues that predict it allows one to more effectively react under threatening conditions and
avoid or minimize harm (Domjan, 2005; Franchina, 1969; Helmstetter & Bellgowan,
1993; Kamin, 1954; Kim & Jung, 2006). Learning the cues that predict an aversive event
results in the production of a fear conditioned response (CR) in anticipation of the threat.
It is this CR that is often the primary focus of conditioning studies. Although the CR
produced by the warning cue is often used as evidence of learning, there are also
learning-related modifications to the unconditioned response (UCR) produced by the
threat itself. From a functional perspective, it is important to understand these innate
UCRs to naturally occurring threats due to their biological relevance for survival
(Domjan, 2005).
During Pavlovian conditioning a conditioned stimulus (CS) is paired with an
aversive unconditioned stimulus (UCS). Associative learning is apparent when the CS
generates a CR. The UCR elicited by the UCS is typically considered a reflexive,
unlearned response. However, prior work has demonstrated learning-related changes in
the UCR. For example, behavioral studies have demonstrated a reduction in UCR
amplitude to predictable compared to unpredictable UCS presentations (Baxter, 1966;
Knight et al., 2011; Lykken et al., 1972; Lykken & Tellegan, 1972; Peeke & Grings,
1968). More specifically, UCR amplitude is decreased when the UCS follows a CS
presentation compared to presentations of the UCS alone (Baxter, 1966; Kimmel, 1967).
This phenomenon is generally referred to as UCR diminution. Prior studies indicate that
15
UCR diminution is partly mediated by an associative learning process (Baxter, 1966;
Kimmel, 1967; Knight et al., 2011; Marcos & Redondo, 1999; Redondo & Marcos,
2002). In addition, UCR diminution appears to be modulated by conscious UCS
expectancies, such that as expectation of the UCS increases, UCR amplitude decreases
(Dunsmoor et al., 2008; Knight et al., 2010, 2011; Rust, 1976; Sarinopoulos et al., 2010).
Diminution of the UCR during Pavlovian conditioning is consistent with formal
learning theory that suggests; 1) learning occurs when there is a discrepancy between an
expectation and the outcome, 2) the CS gains discriminative control over the UCR to the
UCS during conditioning, and 3) that the UCR is diminished by a predictable compared
to unpredictable UCS (Rescorla, 1988; Rescorla & Wagner, 1972; Wagner & Brandon,
1989). For example, prior work has demonstrated a decrease in the magnitude of brain
activation once the relationship between a CS and UCS has been learned and is
predictable (Casey et al., 2000; Dunsmoor et al., 2008; Fletcher et al., 2001; Knight et al.,
2010; Linnman et al., 2011). In contrast, brain activity increases when an outcome
violates expectancies (Casey et al., 2000; Fletcher et al., 2001; Knight et al., 2010).
Only a few brain imaging studies have investigated the neural substrates that
mediate UCR diminution. Dunsmoor et al. (2008) explored UCR diminution during
Pavlovian fear conditioning with functional magnetic resonance imaging (fMRI), and
demonstrated conditioned diminution of the UCR within the dorsolateral prefrontal
cortex (PFC), dorsomedial PFC (including the anterior cingulate cortex), and amygdala.
Similar findings have been observed within the ventromedial PFC, dorsomedial PFC,
insula, and inferior parietal lobule (IPL) as well (Knight et al., 2010). Further, these
studies found that as UCS expectancy increased (during the CS), the amplitude of the
16
fMRI signal response to the UCS decreased within many of these brain regions
(Dunsmoor et al., 2008; Knight et al., 2010). Related work suggests that anticipatory
activity within the dorsomedial PFC modulates insula and amygdala responses to
aversive stimuli (Sarinopoulos et al., 2010). Further, this work suggests that UCS
probability influences the magnitude of brain activation produced by an aversive UCS
(Dunsmoor et al., 2008; Sarinopoulos et al., 2010). Taken together, these studies suggest
that diminution of the behavioral UCR is at least partly mediated by UCS expectancies
that are supported by prefrontal brain regions (Dunsmoor et al., 2008; Knight et al., 2010;
Sarinopoulos et al., 2010). In addition, the unconditioned fMRI signal response within the
dorsomedial PFC, dorsolateral PFC, and insula appears to vary with the magnitude of
unconditioned SCRs (Knight et al., 2010). These findings suggest these brain regions
modulate the expression of threat-related emotional responses.
A number of neuroimaging studies have investigated brain activity during the
anticipation of aversive events (Büchel et al., 1998; Critchley et al., 2001; Knight et al.,
1999; LaBar et al., 1998; Nitschke et al., 2006). This work indicates that the PFC, insula,
and amygdala respond in anticipation of aversive stimuli and show increased activity
when participants have uncertain expectations (Critchley et al., 2001; Davis & Whalen,
2001; Dunsmoor et al., 2007; Volz et al., 2003). For instance, anticipatory responses
within the insula and dorsolateral PFC are larger when participants are uncertain of
whether a UCS will be presented (Dunsmoor et al., 2007). The medial PFC also appears
to be important during unpredictable situations. This region may be essential for the
development of coping strategies and response selection when conditions are
unpredictable (Schienle et al., 2010). Consistent with this view, previous research has
17
demonstrated the importance of the PFC in emotion-related behavior (Delgado et al.,
2008; Ochsner et al., 2002).
Prior work also suggests that individual differences in anxiety levels influence
activity within the neural circuitry that mediates fear-related processes (Klumpp et al.,
2011). More specifically, conditioning research indicates that anxiety levels affect
anticipatory responses within the amygdala and PFC (Indovina et al., 2011). Others have
demonstrated that activity within brain regions that support regulatory control processes
is influenced by the level of trait anxiety (Basten et al., 2011; Sehlmeyer et al., 2011). For
example, an inverse relationship has been observed between trait anxiety and
dorsomedial PFC activity during fear extinction (Sehlmeyer et al., 2011). Other research
has demonstrated differences in the functional connectivity of the dorsomedial PFC and
amygdala in high compared to low anxious individuals (Kim et al., 2011). Taken
together, these findings suggest that individual differences in trait anxiety may modulate
activity within the brain regions that support UCR diminution, and in turn, may effect the
peripheral expression of the emotional response.
Prior work suggests that interconnected dorsal and ventral brain systems process
distinct aspects of emotional stimuli (Kahnt et al., 2011; Morgane et al., 2005;
Viinikainen et al., 2010). A dorsal system that consists of dorsomedial and dorsolateral
PFC appears to be responsible for integrating executive planning functions and higher
level cognitive processes that can be influenced by emotional input (Phillips et al., 2003).
Further, this dorsal system appears to be important for the effortful cognitive regulation
of the emotional response (Delgado et al., 2008; Herwig et al., 2007; Ochsner et al.,
2002). A ventral system that includes the amygdala and insula appears to play an
18
important role in assessing the significance of stimuli and producing the emotional
response (Cheng et al., 2003, 2006, 2007; Knight et al., 2005; Phillips et al., 2003).
However, the role of these dorsal and ventral systems in the threat-related emotional
response is in need of further study.
The present study investigated trait anxiety, UCS expectancy, unconditioned
SCR, and fMRI signal responses during Pavlovian conditioned UCR diminution to better
understand the processes that influence learning-related changes in the emotional
response to a threat. Prior research indicates that UCR diminution is mediated by
associative learning (Baxter, 1966; Kimmel, 1967; Knight et al., 2010; Marcos &
Redondo, 1999; Redondo & Marcos, 2002) and conscious expectations of the UCS
(Dunsmoor et al., 2008; Knight et al., 2010, 2011; Rust, 1976). Based on previous
research, we expected a decrease in the unconditioned fMRI signal to develop with
associative learning and expectations of the UCS (Dunsmoor et al., 2008; Knight et al.,
2010; Sarinopoulos et al., 2010). Given the importance of the amygdala and dorsolateral,
dorsomedial, and ventromedial PFC in emotion (Delgado et al., 2008; Dunsmoor et al.,
2008; Kim & Jung, 2006; Ochsner et al., 2002; Sarinopoulos et al., 2010), we
hypothesized that these brain regions would show UCR diminution during Pavlovian
conditioning. Further, we expected activity within these brain regions to vary with
individual differences in UCS expectancy and trait anxiety. In turn, we expected
amygdala activity to vary with the learning-related modulation of unconditioned SCRs
(Cheng et al., 2003, 2006, 2007; Knight et al., 2005).
19
Materials and Methods
Participants: Twenty-four healthy right-handed volunteers participated in this study and
were included in the behavioral data analyses [12 male, 12 female; age = 20.83 ± 0.64
years (mean ± SEM); range = 19-33 years]. Three participants were excluded from the
fMRI analysis due to poor data quality leaving a total of twenty-one participants (12
male, 9 female; age = 20.81 ± 0.71 years; range = 19-33 years). All subjects provided
written informed consent in compliance with the University of Alabama at Birmingham
Institutional Review Board.
State-Trait Anxiety Inventory: Participants completed the State-Trait Anxiety Inventory
(STAI; Form Y) for Adults (Spielberger, 1983) prior to the conditioning session. The
STAI consists of self-assessment scales that measure state and trait anxiety in terms of
negative affect in general (Grös et al., 2007). Scores on the state scale reflect current
anxiety levels, while trait anxiety scores reflect a relatively long-term predisposition for
anxiety (Spielberger, 1983).
Conditioned and unconditioned stimuli: Participants were exposed to a differential fear
conditioning procedure in which the conditioned and unconditioned stimuli were
presented through MR-compatible pneumatic headphones. Two tones (700 & 1300 Hz;
10s duration; 20s ITI) served as the CSs and a loud (100db) white-noise served as the
UCS (0.5s duration). The UCS coterminated with one tone (CS+), whereas the second
tone was presented alone (CS−) during acquisition trials. A total of thirty-two trials of
each CS were presented over four 590s acquisition blocks (8 trials of each CS were
20
presented in each block). Additionally, each acquisition block contained one set of 3 test
trials that consisted of UCS presentations that coterminated with the CS+ (CS+UCS) and
CS− (CS−UCS), as well as presentations of the UCS alone (Figure 1). Thus, acquisition
blocks consisted of 19 trials (8 CS+ & 8 CS− acquisition trials, as well as 1 CS+UCS, 1
CS−UCS, & 1 UCS alone test trial). Test trials during the acquisition blocks were
presented on trials 17-19 of blocks 1 and 2; trials 13-15 of block 3; and trials 9-11 of
block 4. The final acquisition block was followed by a 920s block of 30 test trials (10
CS+UCS trials, 10 CS−UCS trials, 10 UCS alone trials). In total, there were 14 test trials
for each stimulus (4 from the acquisition blocks, 10 from the test block). We anticipated
that the UCS expectancy ratings to these stimuli would rapidly increase once the test
block began. Therefore, test trials were included during acquisition blocks in an effort to
maintain differential UCS expectancies (i.e. CS+ vs. CS−) for a greater number of trials.
The 14 test trials were grouped into the first 7 test trials (Early test trials) and the last 7
test trials (Late test trials) for further analysis. The test trials from the acquisition blocks
were included in this analysis and binned in this manner to increase the number of test
trials for this contrast, and to reflect the learning-related changes observed in UCS
expectancy that developed during the study. The stimuli were counterbalanced and
presented in a pseudorandom order such that no more than two trials of the same stimulus
were consecutively presented.
UCS expectancy: UCS expectancy was used to measure expectation of the UCS and
assess whether the relationship between the CS and UCS had been learned. Using
Presentation software (Neurobehavioral Systems, Inc.; Albany, CA), a UCS expectancy
21
rating scale was presented on an IFIS-SA LCD (Invivo Corp.; Gainesville, FL) video
screen located above the subject's head and viewed through a mirror attached to the RF
coil. An MRI compatible joystick (Current Designs; Philidelphia, PA) was used to
monitor subjects’ expectancy of receiving the UCS. The joystick controlled a rating bar
which was presented throughout the conditioning session on the video screen. Subjects
were instructed to rate their UCS expectancy on a moment-by-moment basis using a
continuous scale from 0 to 100 (0 = certain the UCS would not be presented, 50 =
uncertain whether the UCS would be presented, 100 = certain the UCS would be
presented) to reflect their current UCS expectancy. UCS expectancy was calculated as the
average response (1s sample) at UCS onset. Additional details on this methodology have
been published previously (Knight & Wood, 2011).
Skin conductance response: An MRI compatible physiological monitoring system
(Biopac Systems; Goleta, CA) was used to collect SCR data. SCR was sampled
(2,000Hz) with a pair of disposable radio-translucent electrodes (1cm diameter, Biopac
Systems; Goleta, CA) from the distal phalanx of the middle and ring fingers of the
nondominant hand. SCR data were processed using Biopac AcqKnowledge 3.9 software.
A 1Hz low pass digital filter was applied and SCR data were resampled at 125Hz.
Unconditioned SCRs were calculated as the maximum SCR during the 10s following the
UCS presentation as compared to baseline (average SCR during 5s prior to CS onset).
Functional MRI: Structural and functional imaging was completed on a 3 Tesla Siemens
Allegra scanner. High-resolution anatomical images (MPRAGE) were obtained in the
22
sagittal plane using a T1 weighted series (TR=2300ms, TE=3.9ms, flip angle=12⁰,
FOV=25.6cm, matrix=256 x 256, slice thickness=1mm, 0.5mm gap) to serve as an
anatomical reference. Blood oxygen level dependent fMRI of the entire brain was
conducted using a gradient-echo echoplanar pulse sequence in an oblique-axial
orientation (TR=2000ms, TE=30ms, flip angle=70º, FOV=24cm, matrix=64 x 64, slice
thickness=4mm, no gap) during each block of stimulus presentations. Functional image
processing was performed with the Analysis of Functional NeuroImages (AFNI) software
package (Cox, 1996). Echo-planar time series data were corrected for slice timing offset,
motion corrected, concatenated, reregistered to the fifth volume of the first imaging
block, and spatially blurred using a 4mm full-width-at-half-maximum Gaussian filter.
Functional MRI data were analyzed at the individual subject level using the input
from all stimuli in a multiple linear regression using a gamma variate hemodynamic
response function. Regressors to account for brain activity not related to the UCR on test
trials included reference waveforms for the CS+ and CS− during acquisition, UCS during
acquisition, CS+ and CS− on test trials, joystick movement, and head motion parameters.
On average, less than one millimeter of movement occurred during the scanning session
(0.76 ± 0.07). The regressors of interest for this study modeled the unconditioned fMRI
signal response to UCS presentations during each type of test trial (i.e. CS+UCS,
CS−UCS, and the UCS alone). Separate reference waveforms were used for Early and
Late test trials in this analysis. Percent signal change on test trials was used as an index of
the magnitude of the unconditioned fMRI signal response produced by the UCS.
Functional maps reflecting percent signal change were converted to the Talairach and
Tournoux stereotaxic coordinate system for group analyses (Talairach & Tournoux,
23
1988).
Based on prior work (Dunsmoor et al., 2008; Knight et al., 2010), group level
analyses were restricted using an anatomical mask, to the PFC, cingulate cortex, IPL,
insula, amygdala, and hippocampus to reduce the number of voxel-wise comparisons. We
conducted a repeated-measures ANOVA to test for a main effect of stimulus (CS+UCS,
CS−UCS, and UCS alone) and trial (Early vs. Late test trials), as well as a stimulus x trial
interaction. A voxel-wise threshold of p< 0.05 (corrected) was employed by using an
uncorrected threshold of p< 0.007 and a cluster volume larger than 620mm3 (11 voxels of
3.75 x 3.75 x 4.00mm dimension). These threshold criteria were used to correct for
multiple comparisons. These threshold criteria were based on Monte Carlo simulations
that were used to reject smaller clusters of activation produced by chance alone (false
positives) (Forman et al., 1995; Saad et al., 2006). The Monte Carlo simulation program
(AlphaSim) (Cox et al., 1996; Saad et al., 2006) considers multiple factors to determine
the combination of cluster volume and voxel-wise uncorrected p threshold. The
simulations are based on: (1) the volume of tissue being studied (restricted to brain
regions within our anatomical mask in accordance with our a priori hypotheses); (2)
voxel size (using the 3.75 x 3.75 x 4.00mm dimensions in which the voxels were
originally acquired); (3) spatial smoothness of the data; and (4) the alpha level (Saad et
al., 2006). Given our a priori hypotheses and the relatively small volume of the amygdala,
we used a voxel-wise threshold of p< 0.007 and a cluster volume larger than 170mm3 (3
voxels of 3.75 x 3.75 x 4.00mm dimension) for this area. Monte Carlo simulations
indicated that the cluster size and p< 0.007 threshold criteria result in a FWE corrected
significance threshold of p< 0.05. Follow-up t-test comparisons were conducted in SPSS
24
on the mean percent signal change activation passing the significance threshold (p< 0.05
corrected) for the ANOVA.
Two different analysis procedures (i.e. correlation and multiple linear regression)
were completed to investigate the relationship between our behavioral measures (i.e. trait
anxiety, UCS expectancy, and SCR) and the fMRI signal from brain regions that
demonstrated UCR diminution in the ANOVA (i.e. functional regions of interest; ROI).
The correlation analysis compared the mean percent signal change from all voxels within
an ROI to behavioral measures (p< 0.05 Bonferroni corrected). Separate correlation
analyses were completed for trait anxiety scale scores, mean UCS expectancy ratings, and
mean unconditioned SCR amplitude to determine whether these behavioral measures
varied with ROI activity. In addition, a voxel-wise multiple linear regression analysis was
conducted to compare these behavioral measures to the unconditioned fMRI signal
response within the ROI. Thus, the regression analysis was also limited to the functional
ROI identified in the ANOVA. The linear model included participant’s trait anxiety scale
score, mean unconditioned SCR amplitude, and mean UCS expectancy rating during each
type of test trial (i.e. CS+UCS, CS−UCS, and UCS alone) to assess the relationship
between these behavioral measures and the unconditioned fMRI signal response.
Regressors that coded for stimulus type were also included in this analysis to determine
whether the behavioral measures explained unique variance in the data. One participant
was excluded from the regression analysis because the trait anxiety questionnaire was not
completed. AlphaSim (Cox et al., 1996; Saad et al., 2006) was used to conduct Monte
Carlo simulations limited to the amygdala and functional ROI from our repeatedmeasures ANOVA that demonstrated a main effect of stimulus or stimulus x trial
25
interaction. A voxel-wise threshold of p< 0.007 and a cluster volume larger than 394mm3
(7 voxels of 3.75 x 3.75 x 4.00mm dimension) was employed, resulting in a FWE
corrected significance threshold of p< 0.05.
Although the correlation and multiple linear regression analyses (described
above) used to compare behavioral measures (i.e. trait anxiety, UCS expectancy, and
SCR) to the fMRI signal within the function ROI were similar, there were important
differences between the two analyses. First, separate correlation analyses were completed
for trait anxiety, UCS expectancy, and unconditioned SCR to determine whether these
behavioral measures varied with ROI activity. However, the use of separate analyses
cannot determine whether the observed brain-behavior relationships are independent.
Therefore, the multiple linear regression analysis was completed using a linear model that
included regressors for trait anxiety, UCS expectancy, unconditioned SCR, and stimulus
type to determine whether these behavioral measures explained unique variance in ROI
activity (e.g. independent of stimulus type). Second, the correlation analysis compared
each behavioral measure to the average signal from all voxels within a ROI, whereas the
regression analysis was completed on a voxel-wise basis (restricted to each ROI). Thus
the correlation analyses assessed activity within an ROI as a whole, while the regression
analysis evaluated activity within an ROI on a voxel-wise basis.
Prior work suggests that PFC activity regulates the emotional response to aversive
stimuli (Delgado et al., 2008; Sarinopoulos et al., 2010). Therefore, we completed an
additional voxel-wise multiple regression analysis to identify PFC areas in which
anticipatory activity varied with the unconditioned fMRI signal response obtained from
the amygdala and functional ROI from our ANOVA. This analysis was restricted to
26
CS+UCS and CS−UCS trials because a CS was not presented during UCS alone trials to
elicit an anticipatory response. This analysis included a regressor representing trial type
(CS+UCS & CS−UCS), a regressor for the amplitude of the unconditioned fMRI signal
response from our functional ROI, and a regressor for the interaction of trial type and
unconditioned fMRI signal response amplitude. This analysis was restricted to the PFC
and cingulate using an anatomical mask. As indicated by Monte Carlo simulations, a
voxel-wise threshold of p< 0.007 and a cluster volume larger than 620mm3 (11 voxels of
3.75 x 3.75 x 4.00mm dimension) was employed resulting in a FWE corrected
significance threshold of p< 0.05.
Results
UCS expectancy: Repeated measures ANOVA revealed significant differences in UCS
expectancy during test trials. Results showed a main effect for stimulus type (F[1,23] =
58.41, p< 0.05), a main effect for trial (F[1,23] = 63.43, p< 0.05), and a stimulus by trial
interaction (F[1,23] = 38.90, p< 0.05). UCS expectancy was greater during Early test
trials on CS+UCS presentations than on CS−UCS (t[23] = 6.74, p< 0.05) and UCS alone
trials (t[23] = 7.55, p< 0.05). UCS expectancy did not differ for CS−UCS and UCS alone
presentations during the Early test trials (t[23] <1.00). On Late test trials however, UCS
expectancy to the CS−UCS was greater than expectancy for the UCS alone (t[23] = 4.40,
p< 0.05). UCS expectancy to the CS+UCS remained greater than to the UCS alone (t[23]
= 6.45, p< 0.05) and CS−UCS (t[23] = 2.01, p< 0.05) on Late test trials (Figure 2a).
Skin conductance response: Repeated measures ANOVA also revealed significant
differences in unconditioned SCR expression during test trials. There was a main effect
27
for stimulus type (F[1,23] = 4.77, p< 0.05). However, there was no trial effect (F< 1.00)
or stimulus by trial interaction (F< 1.00). T-test comparisons revealed a significantly
diminished unconditioned SCR for CS+UCS trials (mean ± SEM [adjusted for between
subject variance (Loftus & Masson, 1994)]: 0.07 ± 0.02) as compared to CS−UCS trials
(0.14 ± 0.02; t[23] = -1.77, p< 0.05) and UCS alone trials (0.11 ± 0.01; t[23] = -2.18, p<
0.05). There was not a significant difference in unconditioned SCR between the CS−UCS
and UCS alone (t < 1.00) (Figure 2b).
Functional MRI: The fMRI data analysis indicated that several brain regions showed
diminution of the unconditioned fMRI signal response (Table 1; Figures 3-4). UCR
diminution was observed within the dorsolateral PFC, dorsomedial PFC, anterior insula,
IPL, and posterior cingulate cortex (PCC). In each of these regions, the unconditioned
fMRI signal response demonstrated a main effect for stimulus type (F[20] > 5.63; p<
0.05 corrected). A main effect for trial (i.e. Early vs. Late test trials) was observed within
the dorsolateral PFC, dorsomedial PFC, PCC, and anterior insula (Table 2). A stimulus x
trial interaction was observed within ventromedial PFC (Talairach coordinates: 12, 49,
14; volume: 656mm3, Table 1; Figure 4). T-test comparisons were completed on the
mean fMRI signal from each volume of activation that passed the significance threshold
(p< 0.05 corrected) for the main effect of stimulus type. All regions showed a diminished
UCR on CS+UCS trials compared to the UCS alone. Most of these areas also showed a
reduction in UCR amplitude on CS+UCS trials compared to CS−UCS trials. The left IPL
and right anterior insula were the only regions that did not show a diminished UCR on
CS+UCS compared to CS−UCS trials. In addition, the left dorsolateral PFC and left IPL
28
showed reduced UCR amplitude on CS−UCS trials compared to presentations of the
UCS alone (Table 1).
Percent signal change data from each of the clusters of activation that met our
significance criteria were correlated with trait anxiety, UCS expectancy, and
unconditioned SCR amplitude measures (Table 1; p< 0.05 Bonferroni corrected). The
mean percent signal change within these brain regions did not vary with trait anxiety or
unconditioned SCR. However, there was a significant correlation observed between UCS
expectancy ratings and activity within the bilateral dorsolateral PFC, dorsomedial PFC,
bilateral IPL, bilateral insula, and PCC (Table 1).
A voxel-wise multiple linear regression analysis that accounted for stimulus type
was also conducted to determine areas in which brain activity varied with trait anxiety,
UCS expectancy, and SCR production. This analysis was restricted to regions that
showed UCR diminution from the ANOVA with one exception. The amygdala was
included, based on prior work demonstrating its role in learning-related SCR production
(Cheng et al., 2003, 2006, 2007; Knight et al., 2005) even though it did not meet the
significance criteria for the ANOVA. This analysis demonstrated that inter-subject trait
anxiety levels explained unique variability in the activation observed within the
dorsolateral PFC, dorsomedial PFC, PCC, and IPL (Table 3; Figure 5). There were no
brain regions that showed a relationship with UCS expectancy that met our significance
criteria. Unconditioned SCR amplitude varied with UCS−related activity within the left
(r = 0.45) and right (r = 0.42) amygdala (Talairach coordinates and volume: left; -25, -4,
-15 and 511mm3, right; 25, -4, -16 and 460mm3; Figure 6 a & b). Because our initial
ANOVA did not demonstrate UCR diminution within the amygdala that met our
29
significance criteria, we created functional ROI from the bilateral amygdala volumes that
were associated with unconditioned SCR production. These functional ROI were then
used to assess UCR diminution within the amygdala. UCR diminution was observed
within the bilateral amygdala (Figure 6 c & d). The amplitude of the unconditioned fMRI
signal response within the amygdala was diminished on CS+UCS trials compared to UCS
alone trials (left; t[20] = -2.175; right; t[20] = -1.860, p< 0.05). The UCR on CS−UCS
trials fell at an intermediate level and did not differ from the UCR to CS+UCS or UCS
alone trials (Figure 6 c & d).
We also completed a group level regression analysis to determine whether
anticipatory PFC activity varied with threat-related activity obtained from the amygdala
and functional ROI from our ANOVA (regions depicted in Figures 3, 4, & 6 and Table
1). Anticipatory activity (i.e. the CR) within the dorsolateral PFC and dorsomedial PFC
showed a negative relationship with threat-related (i.e. the UCR) activity on CS+UCS,
but not CS−UCS trials within many of our functional ROI. This effect was observed
between anticipatory right dorsolateral PFC activity and threat-related activity within the
left dorsolateral PFC (Figure 7b, d) and right IPL (Figure 7a). A similar pattern was
observed between anticipatory activity within 2 separate areas of the left dorsolateral PFC
and the threat response within an adjacent area of the left dorsolateral PFC (Figure 7c and
d). Finally, the same basic pattern was observed between anticipatory dorsomedial PFC
activity and threat-related responses within the left dorsolateral PFC (Figure 7g and h),
ventromedial PFC (Table 4), and bilateral amygdala (Figure 7e and f). Threat-related
activity within the remaining functional ROI, including the right dorsolateral PFC,
30
dorsomedial PFC, left IPL, bilateral insula, and PCC did not vary with anticipatory
activity in the PFC.
Discussion
The ability to identify and quickly respond to a threat is critical to survival.
Associative learning processes allow one to predict impending threats to more effectively
avoid or escape danger (Franchina, 1969; Kamin, 1954). Further, these processes can
minimize the reaction to the threat itself (Dunsmoor et al., 2008; Knight et al., 2010;
Marcos & Redondo, 1999). For example, conditioned hypoalgesia (decreased sensitivity
to painful stimuli) develops during fear conditioning to reduce the pain produced by
noxious stimuli (Helmstetter, 1992; Helmstetter & Bellgowan, 1993). A similar process
appears to diminish the emotional response elicited by aversive stimuli during fear
conditioning (Dunsmoor et al., 2008; Knight et al., 2010; Marcos & Redondo, 1999;
Sarinopolous et al., 2010). The present study investigated trait anxiety, UCS expectancy,
unconditioned SCR, and fMRI signal responses during Pavlovian conditioned UCR
diminution to better understand the processes that influence learning-related changes in
the emotional response to a threat.
In the current study, we observed conditioned diminution of the unconditioned
SCR. The magnitude of the unconditioned SCR was diminished on CS+UCS trials
compared to the UCR produced during CS−UCS and UCS alone trials (see Figure 2b).
These data demonstrate that the emotional response to an aversive stimulus is reduced
when it is predictable. These findings are consistent with prior behavioral research that
has demonstrated a reduction in UCR magnitude when the UCS follows a CS+ compared
31
to when the UCS follows a CS− or is presented alone (Baxter, 1966; Kimmel, 1967;
Knight et al., 2011; Lykken et al., 1972; Marcos & Redondo, 1999; Redondo & Marcos,
2002). Taken together, these findings support the view that UCR diminution during
Pavlovian conditioning is in part mediated by an associative learning process.
Learning-related changes in UCS expectancy were also observed in the present
study. UCS expectancy ratings on CS+UCS trials were consistently high across Early and
Late test trials, indicating that participants expected the UCS to follow the CS+. UCS
expectancy ratings to the UCS alone were relatively low on both Early and Late test
trials. In contrast, UCS expectancy ratings on CS−UCS test trials increased over the
course of the study such that expectancy ratings on Late test trials were larger to the
CS−UCS than UCS alone (Figure 2a). These findings demonstrate that participants
learned that there was a change in the relationship between the CS− and UCS over the
course of the study. The increase in UCS expectancy ratings on CS−UCS trials indicates
participants learned that the CS− no longer predicted the absence of the UCS, but instead
that the UCS would follow the CS−. Previous research indicates that UCS expectancy
modulates unconditioned autonomic responses (Dunsmoor et al., 2008; Knight et al.,
2010, 2011). These studies have demonstrated that as UCS expectancy increases,
unconditioned SCRs decrease (Dunsmoor et al., 2008; Knight et al., 2010, 2011). This
prior work is generally consistent with investigations of the neural mechanisms of error
detection. For example, brain activity decreases as the relationship between a CS and
UCS becomes predictable (Fletcher et al., 2001). However, activity increases when an
outcome violates expectations (Casey et al., 2000; Fletcher et al., 2001). Therefore,
conscious expectations that a threat is imminent may play an important role in UCR
32
diminution. Interestingly however, the unconditioned SCR produced in response to
CS−UCS trials did not diminish over the course of the study (Figure 2b). Further,
unconditioned SCR amplitude did not differ between the CS−UCS and UCS alone on
Late test trials, as would be predicted if UCR diminution was solely mediated by changes
in UCS expectancy. Instead, the magnitude of the unconditioned SCR on CS−UCS trials
remained elevated. This finding suggests that UCR diminution is not solely mediated by
UCS expectancy. Instead, associative learning processes independent of UCS expectancy
also appear to influence UCR diminution. This conclusion is consistent with prior work
demonstrating that both UCS expectancy and an expectancy independent conditioning
process influence UCR diminution (Knight et al., 2011). For example, unconditioned
SCR diminution is greater to CS+UCS than CS−UCS presentations even after
participants have been explicitly informed that the UCS would follow both CS+ and CS−
presentations on test trials (Marcos & Redondo, 1999). Further, greater UCR diminution
has been observed to CS+UCS than CS−UCS trials with equivalent UCS expectancy
ratings (Knight et al., 2011). Thus, while there is strong evidence from prior work that
demonstrates expectations of the UCS modulate UCR expression, learning-related
processes that are independent of expectancy also provide discriminative control over the
UCR. In general, these findings are consistent with prior work that has demonstrated a
dissociation between UCS expectancy and SCR expression (Balderston & Helmstetter,
2010; Knight et al., 2003, 2006, 2009; Schultz & Helmstetter, 2010).
In the present study, UCR diminution was observed in both SCR and fMRI data.
Unconditioned SCR magnitude was diminished on CS+UCS compared to UCS alone
trials, and this response pattern was paralleled by the fMRI signal response within the
33
dorsolateral PFC, dorsomedial PFC, anterior insula, IPL, PCC, and amygdala. The
unconditioned SCR and fMRI signal within most of these brain regions also showed
diminution on CS+UCS compared to CS−UCS trials (Table 1; Figures 2 & 3).
Diminution of the unconditioned fMRI signal response within each of these brain regions
has been observed in previous neuroimaging studies (Dunsmoor et al., 2008; Knight et
al., 2010). For example, the UCR diminution observed within the dorsomedial PFC in the
present study overlaps with the anterior cingulate (Dunsmoor et al., 2008) and
dorsomedial PFC (Knight et al., 2010) activations observed in prior work. Further, the
UCR diminution observed within dorsolateral PFC (Dunsmoor et al., 2008; Knight et al.,
2010), IPL (Dunsmoor et al., 2008; Knight et al., 2010), PCC (Knight et al., 2010), insula
(Knight et al., 2010), and amygdala (Dunsmoor et al., 2008) in prior studies largely
overlaps with the activity observed in the present study. The conditioning procedures
employed by these prior studies differ somewhat from those used in the present study.
For example, Knight et al. (2010) demonstrated diminution of the unconditioned SCR
and fMRI signal response by varying the volume of auditory CS presentations above and
below the perceptual detection threshold. The UCR was diminished when the UCS
followed a perceived versus unperceived CS+ (Knight et al., 2010). Dunsmoor et al.
(2008) investigated differences in unconditioned SCR and fMRI signal amplitude using
continuous reinforcement (i.e. 100% pairing of the CS and UCS) and partial
reinforcement (i.e. 50% pairing of the CS and UCS) conditioning procedures, and
demonstrated greater UCR diminution during continuous than partial reinforcement
(Dunsmoor et al., 2008).
34
The methodology of the present research is similar to that used in the Dunsmoor
et al. (2008) study in that the CS− was partially paired with the UCS during acquisition.
More specifically, the CS− was paired with the UCS on one (i.e. the CS−UCS test trial)
out of every 9 trials (approximately 11% CS and UCS pairing rate) during acquisition
blocks. The CS− was then paired with the UCS on every trial during the test block (100%
pairing rate). Given the nature of this type of study, the CS− must be paired with the UCS
to assess UCR diminution. However, by pairing the CS− and UCS the contingencies
between the two stimuli change, and participants learn that the CS− is paired with the
UCS on a percentage of the trials. Such learning is apparent in the UCS expectancy
ratings depicted in Figure 2a. The intermittent pairing of the CS− and UCS during
acquisition was intended to increase the number of CS−UCS trials in which low UCS
expectancy was reported because prior work indicates that UCS expectancies modulate
UCR amplitude (Knight et al., 2010, 2011). This effect is also illustrated in the current
study. A significant negative correlation was observed between UCS expectancy and
activity within several brain regions that showed UCR diminution (Table 1). However,
this relationship was not apparent in our group level regression analysis. Together, these
findings indicate UCS expectancy varies with the fMRI signal within these ROI, but does
not explain unique variance in the fMRI data when other regressors (i.e. stimulus type)
are included in the model. In combination with prior research (Dunsmoor et al., 2008;
Knight et al., 2010), the current findings support the view that differences in the
emotional response (indexed with SCR) to predictable versus unpredictable threats are
mediated by associative learning processes that are supported by a network of brain
regions that includes the amygdala and PFC (Figures 3, 4, & 6).
35
Prior work suggests that prefrontal processes support contingency learning
(Knight et al., 2004; Phillips et al., 2003), modulation of the emotional response (Delgado
et al., 2008; Milad et al., 2004; Ochsner et al., 2002), and the integration of emotional
information (Phillips et al., 2003). In line with work that suggests an interconnected
dorsal and ventral system processes and responds to emotional stimuli (Phillips et al.,
2003), we found evidence that suggests these dorsal and ventral systems mediate the
UCR diminution observed during Pavlovian conditioning. The dorsal system appears to
integrate cognitive processes that can be affected by emotional input (Phillips et al.,
2003), and the present study demonstrated UCR diminution in multiple dorsal system
brain regions (i.e. dorsolateral PFC, dorsomedial PFC, and IPL; Figure 3). Thus, the
emotion-related processes supported by this dorsal system may mediate UCR diminution.
We also observed UCR diminution within the amygdala as well as a strong relationship
between amygdala activity and the unconditioned SCRs that showed conditioned UCR
diminution (Figure 6). This finding is consistent with the view that a ventral neural
system assesses the emotional significance of environmental stimuli and controls
expression of the emotional response (Phillips et al., 2003).
The amplitude of the fMRI signal response within many of the brain regions
showing UCR diminution fluctuated with inter-subject variations in trait anxiety. The
voxel-wise multiple linear regression analysis demonstrated that as trait anxiety
increased, UCR activity increased within the dorsolateral PFC, dorsomedial PFC, PCC,
and IPL (Figure 5 & Table 3). Although the multiple regression analysis demonstrated a
relationship between trait anxiety and the magnitude of ROI activity, a similar effect was
not observed in the correlation analysis (Table 1). The regression analysis indicates that
36
activity within a subset of the voxels within these ROI vary with trait anxiety. This effect
was not observed in the correlation analysis, because the voxels that showed a
relationship between trait anxiety and ROI activity were averaged together with voxels
that did not show the same relationship, diluting the effect. In general, the multiple
regression findings are generally consistent with prior work suggesting that trait anxiety
modulates activity within these brain regions as threatening stimuli are assessed (Klumpp
et al., 2011). Related work suggests that trait anxiety levels influence the activity within
brain regions that support regulatory control processes (Basten et al., 2011; Sehlmeyer et
al., 2011). For example, dorsomedial PFC activity shows a negative relationship with trait
anxiety during the extinction of conditioned fear (Sehlmeyer et al., 2011). Other research
has shown that functional connectivity between the dorsomedial PFC and the amygdala is
weaker in high compared to low anxious individuals (Kim et al., 2011). The present
findings are consistent with prior work that suggests trait anxiety modulates activity
within dorsal brain regions that support emotion-related processes.
Previous research has also shown that changes in the magnitude of the
unconditioned fMRI signal response within the dorsolateral PFC, dorsomedial PFC, and
insula are paralleled by changes in autonomic activity (Knight et al., 2010). In the
present study, we found the amplitude of UCR activity within the amygdala varied with
the amplitude of the unconditioned SCR. Our data suggest that the amygdala also plays
an important role in the learning-related modulation of the unconditioned SCR. These
findings are consistent with prior research that demonstrated the amygdala and PFC
influence the expression of the emotional response (Cheng et al., 2003, 2006; Delgado et
al., 2008; Knight et al., 2005; Milad et al., 2004, 2007; Sarinopoulos et al., 2010).
37
Prior work suggests that the PFC regulates activity within brain regions (e.g.
amygdala & insula) that support emotional expression (Delgado et al., 2008; Milad et al.,
2004, 2007; Sarinopoulos et al., 2010). For example, anticipatory PFC activity is
negatively correlated with amygdala and insula responses to aversive stimuli
(Sarinoplous et al., 2010). The findings from the present study are relatively consistent
with this prior work. In the present study, an inverse relationship was observed between
anticipatory PFC and threat-related activity within several brain regions showing UCR
diminution. As anticipatory dorsomedial and dorsolateral PFC activity increased, threatrelated activity within the dorsolateral PFC, ventromedial PFC, IPL, and amygdala
decreased on CS+UCS trials (Table 4; Figure 7). These findings are generally consistent
with the view that the PFC supports emotion-related processes, and that anticipatory
activity affects threat-related responses within the PFC-amygdala circuit.
In summary, learning-related changes in the unconditioned SCR and fMRI signal
response were observed during Pavlovian fear conditioning. The current findings
replicate prior work demonstrating UCR diminution using functional neuroimaging in
conjunction with behavioral measures (e.g. SCR and UCS expectancy ratings) during
conditioning (Dunsmoor et al., 2008; Knight et al., 2010). Investigating the
neurobiological processes that mediate conditioned UCR diminution may provide a
starting point to better understand the emotional dysregulation that characterizes many
anxiety disorders (Grillon, 2002; Davis et al., 2009; Kim & Jung, 2006; Milad et al.,
2006). For example, prior investigations of the neurobiological markers of anxiety
suggest that insufficient top-down inhibitory control (Klumpp et al., 2011; Nitschke et al.,
2006; Rauch et al., 2006; Schienle et al., 2010) may result in hypersensitivity of the
38
subcortical brain areas (e.g. the amygdala) that control the peripheral expression of
emotion (Etkin & Wager, 2007; Milad et al., 2006, 2009; Rauch et al., 2006). Several
brain regions that integrate cognitive processes and modulate emotional responses
exhibited UCR diminution (i.e. dorsolateral PFC and dorsomedial PFC). UCR diminution
was also observed within the PCC, anterior insula, and IPL. Furthermore, many of the
brain regions that demonstrated UCR diminution in this study have also been implicated
in regulating autonomic responses associated with affective states and emotional
behavior. The findings from the present study are consistent with the view that
anticipatory activity within the PFC may inhibit threat-related activity within this neural
circuit. In turn, the learning-related changes in UCR amplitude within these brain regions
may interact with the amygdala to modulate unconditioned SCR production. The present
findings provide a better understanding of the psychological processes and neural
circuitry that mediate the threat-related emotional response.
Acknowledgements: This research was supported by the University of Alabama at
Birmingham Faculty Development Grant Program.
39
References
Balderston, N. L., & Helmstetter, F. J. (2010). Conditioning with masked stimuli affects
the timecourse of skin conductance responses. Behavioral Neuroscience, 124(4),
478–489.
Basten, U., Stelzel, C., Fiebach, C.J. (2011). Trait anxiety modulates the neural efficiency
of inhibitory control. Journal of Cognitive Neuroscience, 10, 3132-3145.
Baxter, R. (1966). Diminution and recovery of the UCR in delayed and trace classical
GSR conditioning. Journal of Experimental Psychology, 71(3), 447-451.
Büchel, C., Morris, J., Dolan, R.J., & Friston, K.J. (1998). Brain systems mediating
aversive conditioning: an event-related fMRI study. Neuron, 20, 947-957.
Casey, B. J., Thomas, K.M., Welsch, T.F., Badgaiyan, R.D., Eccard, C.H., Jennings, J.R.,
& Crone, E.A. (2000). Dissociation of response conflict, attentional selection and
expectancy with functional magnetic resonance imaging. Proceedings of the
National Academy of Sciences, 97(15), 8728–8733.
Cheng, D. T., Knight, D. C., Smith, C. N., Stein, E. A., & Helmstetter, F. J. (2003).
Functional MRI of human amygdala activity during Pavlovian fear conditioning:
Stimulus processing versus response expression. Behavioral Neuroscience, 117
(1), 3-10.
Cheng, D. T., Knight, D. C., Smith, C. N., Stein, E. A., & Helmstetter, F. J. (2006).
Human amygdala activity during the expression of fear responses. Behavioral
Neuroscience, 120 (6), 1187-1195.
Cheng, D. T., Richards, J., & Helmstetter, F. J. (2007). Activity in the human amygdala
corresponds to early, rather than late period autonomic responses to a signal for
shock. Learning & Memory, 14 (7), 485-490.
Cox, R.W. (1996). AFNI: Software for analysis and visualization of functional magnetic
resonance neuroimages. Computers and Biomedical Research, 29, 162-173.
Critchley, H.D., Mathia, C.J., & Dolan, R.J. (2001). Neural activity in the human brain
relating to uncertainty and arousal during anticipation. Neuron, 29, 537-545.
Davis, M., Walker, D.L., Miles, L., & Grillon, C. (2009). Phasic vs. sustained fear in rats
and humans: role of the extended amygdala in fear vs. anxiety.
Neuropsychopharmacology Reviews, 1-31.
Davis, M., & Whalen, P.J. (2001). The amygdala: vigilance and emotion. Molecular
Psychiatry, 6, 13-34.
40
Delgado, M.R., Nearing, K.I., LeDoux, J.E., & Phelps, E.A. (2008). Neural circuitry
underlying the regulation of conditioned fear and its relation to extinction.
Neuron, 59, 829-838.
Domjan, M. (2005). Pavlovian conditioning: A functional perspective. Annual Review of
Psychology, 56, 179-206.
Dunsmoor, J.E., Bandettini, P.A., & Knight, D.C. (2007). Impact of continuous versus
intermittent CS−UCS pairing on human brain activation during Pavlovian fear
conditioning. Behavioral Neuroscience, 121, 635-642.
Dunsmoor, J.E., Bandettini, P.A., & Knight, D.C. (2008). Neural correlates of
unconditioned response diminution during Pavlovian conditioning. NeuroImage,
40, 811-817.
Etkin, A. & Wager, T.D. (2007). Functional neuroimaging of anxiety: A meta-analysis of
emotional processing in PTSD, social anxiety disorder, and specific phobia.
American Journal of Psychiatry, 164, 1476-1488.
Fletcher, P. C., Anderson, J. M., Shanks, D. R., Honey, R., Carpenter, T. A., Donavan,
T., Papadakis, N., & Bullmore, E.T. (2001). Responses of human frontal cortex to
surprising events are predicted by formal associative learning theory. Nature
Neuroscience, 4(10), 1043-1048.
Forman, S.D., Cohen, J.D., Fitzgerald, M., Eddy, W.F., Mintun, M.A., & Noll, D.C.
(1995). Improved assessment of significant activation in functional magnetic
resonance imaging (fMRI): use of a cluster size threshold. Magnetic Resonance in
Medicine, 33, 636–647.
Franchina, J. J. (1969). Escape behavior and shock intensity: Within subject versus
between-groups comparisons. Journal of Comparative and Physiological
Psychology, 69(2), 241–245.
Grillon, C. (2002). Startle reactivity and anxiety disorders: Aversive conditioning,
context, and neurobiology. Biological Psychiatry, 52, 958-975.
Grös, D.F., Antony, M.M., Simms, L.J., & McCabe., R.E. (2007). Psychometric
Properties of the State–Trait Inventory for Cognitive and Somatic Anxiety
(STICSA): Comparison to the State–Trait Anxiety Inventory (STAI).
Psychological Assessment 19(4), 369-381.
Hartley, C.A., & Phelps, E.A. (2010). Changing fear: The neurocircuitry of emotion
regulation. Neuropsychopharmacology, 35, 136-146.
Helmstetter, F.J. (1992). The amygdala is essential for the expression of conditional
hypoalgesia. Behavioral Neuroscience, 106(3), 518-528.
41
Helmstetter, F. J., & Bellgowan, P. S. (1993). Lesions of the amygdala block conditional
hypoalgesia on the tail flick test. Brain Research, 612(1–2), 253–257.
Herwig, U., Baumgartner, T., Kaffenberger, T., Bruhl, A., Kottlow, M., Schreiter-Gasser,
Abler, B., Jäncke L., & Rufer M. (2007). Modulation of anticipatory emotion and
perception processing by cognitive control. NeuroImage, 37, 652-662.
Indovina, I., Robbins, T.W., Núñez-Elizalde, A.O., Dunn, B.D., & Bishop, S.J. (2011).
Fear-conditioning mechanisms associated with trait vulnerability to anxiety in
humans. Neuron, 69, 563-571.
Kahnt, T., Heinzle, J., Park, S.Q., & Haynes, J.D. (2011). Decoding different roles for
vmPFC and dlPFC in multi-attribute decision making. NeuroImage, 56, 709-715.
Kamin, L. J. (1954). Traumatic avoidance learning: The effects of CS−US interval with a
trace-conditioning procedure. Journal of Comparative and Physiological
Psychology, 47(1), 65–72.
Kim, M.J., Gee, D.G., Loucks, R.A., Davis, F.C., & Whalen, P.J. (2011). Anxiety
dissociates dorsal and ventral medial prefrontal cortex functional connectivity
with the amygdala at rest. Cerebral Cortex, 7, 1667-1673.
Kim., J.J., & Jung, M.W. (2006). Neural circuits and mechanisms involved in Pavlovian
fear conditioning: A critical review. Neuroscience and Biobehavioral Reviews,
30, 188-202.
Kimmel, E. (1967). Judgments of UCS intensity and diminution of the UCR in classical
GSR conditioning. Journal of Experimental Psychology, 73, 532-543.
Klumpp, H., Ho, S.S., Taylor, S.F., Phan, K.L., Abelson, J.L., & Liberzon, I. (2011).
Trait anxiety modulates anterior cingulate activation to threat interference.
Depression and Anxiety, 28, 194-201.
Knight, D. C., Smith, C. N., Stein, E. A., & Helmstetter, F. J. (1999). Functional MRI of
human Pavlovian fear conditioning: Patterns of activation as a function of
learning. NeuroReport, 10 (17), 3665-3670.
Knight, D.C., Nguyen, H.T., & Bandettini, P.A. (2003). Expression of conditional fear
with and without awareness. Proceedings of the National Academy of Sciences,
100(25), 15280-15283.
Knight, D.C., Cheng, D.T., Smith, C.N., Stein, E.A., & Helmstetter, F.J. (2004). Neural
substrates mediating human delay and trace fear conditioning. Journal of
Neuroscience, 24, 218-228.
Knight, D. C., Nguyen, H. T., & Bandettini, P. A. (2005). The role of the human
amygdala in the production of conditioned fear responses. NeuroImage, 26, 11931200.
42
Knight, D.C., Nguyen, H.T., & Bandettini, P.A. (2006). The role of awareness in delay
and trace fear conditioning in humans. Cognitive, Affective, and Behavioral
Neuroscience, 6(2), 157-162.
Knight, D.C., Waters, N.S., & Bandettini, P.A. (2009). Neural substrates of explicit and
implicit fear memory. NeuroImage, 45, 208-214.
Knight, D.C. Waters, N.S., King, M.K., & Bandettini, P.A. (2010). Learning-related
diminution of unconditioned SCR and fMRI signal responses. NeuroImage, 49,
843-848.
Knight, D. C., Lewis, E. P., & Wood, K. H. (2011). Conditioned diminution of the
unconditioned SCR. Behavioral Neuroscience, 124 (4), 626-631.
Knight, D. C. & Wood, K. H. (2011). Investigating the neural mechanisms of aware and
unaware fear memory with fMRI. Journal of Visualized Experiments (56), e3083,
DOI:10.39791/2083.
LaBar, K.S., Gatenby, J.C., Gore, J.C., LeDoux, J.E., & Phelps, E.A. (1998). Human
amygdala activation during conditioned fear acquisition and extinction: a mixedtrial fMRI study. Neuron, 20, 937-945.
Linnman, C., Rougemont-Bücking, A., Beucke, J. C., Zeffiro, T. A., & Milad, M. R.
(2011). Unconditioned responses and functional fear networks in human classical
conditioning. Behavioral Brain Research, 221 (1), 237-245.
Loftus, G.R., & Masson, M.E.J. (1994). Using confidence intervals in within-subject
designs. Psychonomic Bulletin & Review, 4, 476-490.
Lykken, D.T., Macindoe, Y., & Tellegen, A. (1972). Perception: Autonomic responses to
shock as a function of predictability in time and locus. Psychophysiology, 9, 318333.
Lykken, D.T., & Tellegan, A. (1972). On the validity of the perception hypothesis.
Psychophysiology, 11, 125.132.
Marcos, J.L., & Redondo, J. (1999). Effects of conditioned stimulus presentation on
diminution of the unconditioned response in aversive classical conditioning.
Biological Psychology, 50, 89-102.
Milad, M.R., Vidal-Gonzalez, I., & Quirk, G.J. (2004). Electrical stimulation of medial
prefrontal cortex reduces conditioned fear in a temporally specific manner.
Behavioral Neuroscience, 118, 389-394.
Milad, M.R., Rauch, S.L., Pitman, R.K., & Quirk, G.J. (2006). Fear extinction in rats:
Implications for human brain imaging and anxiety disorders. Biological
Psychology, 73, 61-71.
43
Milad, M.R., Quirk, G.J., Pitman, R.K., Orr, S.P., Fischl, B., & Rauch, S.L. (2007). A
role for the human dorsal anterior cingulate cortex in fear expression. Biological
Psychology, 62, 1191-1194.
Milad, M.R., Pitman, R.K., Ellis, C.B., Gold, A.L., Shin, L.M., Lasko, N.B., Zeidan,
M.A., Handwerger, K., Orr, S.P., & Rauch, S.L. (2009). Neurobiological basis of
failure to recall extinction memory in posttraumatic stress disorder. Biological
Psychiatry, 66, 1075-1082.
Morgane, P.J., Galler, J.R., & Mokler, D.J. (2005). A review of systems and networks of
the limbic forebrain/limbic midbrain. Progress in Neurobiololgy, 75, 143–160.
Nitschke, J.B., Dixon, G.E., Sarinopoulos, I., Short, S.J., Cohen, J.D., Smith, E.E.,
Kosslyn, S.M., Rose, R.M., & Davidson, R.J. (2006). Altering expectancy
dampens neural response to aversive taste in primary taste cortex. Nature
Neuroscience, 9, 435-442.
Ochsner, K.N., Bunge, S.A., Gross, J.J., & Gabrieli, J.D. (2002). Rethinking feelings: an
fMRI study of the cognitive regulation of emotion. Journal of Cognitive
Neuroscience, 14, 1215-1229.
Peeke, S.C., & Grings, W.W. (1968). Magnitude of UCR as a function of variability in
the CS−UCS relationship. Journal of Experimental Psychology, 77, 1-6.
Phillips, M.L., Drevets, W.C., Rauch, S.L., & Lane, R. (2003). Neurobiology of emotion
perception I: The neural basis of normal emotion perception. Biological
Psychiatry, 54, 504-514.
Rauch, S.L., Shin, L.M., & Phelps, E.A. (2006). Neurocircuitry models of posttraumatic
stress disorder and extinction: Human neuroimaging research—past, present, and
future. Biological Psychiatry, 60, 376–382.
Redondo, J., & Marcos, J.L. (2002). Effects of CS−US interval on unconditioned
response diminution in human heart rate classical conditioning. Journal of
Psychophysiology, 17, 30-38.
Rescorla, R. A. (1988). Behavioral studies of Pavlovian conditioning. Annual Review of
Neuroscience, 11, 329-352.
Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: variations
in the effectiveness of reinforcement and nonreinforcement. In A. H. Black & W.
F. Prokasy (Eds.), Classical conditioning II: current research and theory (pp. 6499). New York: Appleton-Century-Crofts.
Rust, J. (1976). Unconditioned response diminution in the skin resistance response. The
Journal of General Psychology, 95, 77-84.
44
Saad, Z.S., Chen, G., Reynolds, R.C., Christidis, P.P., Hammett, K.R., Bellgowan, P.S.F.,
& Cox, R.W. (2006). Functional imaging analysis contest (FIAC) analysis
according to AFNI and SUMA. Human Brain Mapping, 27, 417-424.
Sarinopoulos, I., Grupe, D.W., Mackiewicz, K.L., Herrington, J.D., Lor, M., Steege, E.E.,
& Nitschke, J.B. (2010). Uncertainty during anticipation modulates neural
responses to aversion in human insula and amygdala. Cerebral Cortex, 20, 929940.
Schienle, A., K chel, A., Ebner, F., Reishofer, G., & Schäfer, A. (2010). Neural
correlates of intolerance of uncertainty. Neuroscience Letters, 479 (3), 272-276.
Schultz, D.H. & Helmstetter, F.J. (2010). Classical conditioning of autonomic fear
responses is independent of contingency awareness. Journal of Experimental
Psychology, 36(4), 495-500.
Sehlmeyer, C., Dannlowski, U., Schöning, S., Kugel, H., Pyka, M., Pfleiderer, B.,
Zwitserlood, P., Schiffbauer, H., Heindel, W., Arolt, V., & Konrad, C. (2011).
Neural correlates of trait anxiety in fear extinction. Psychological Medicine, 41,
789-798.
Spielberger, C.D. (1983). State-Trait Anxiety Inventory for Adults. Redwood City, CA:
Mind Garder.
Talairach, J., & Tournoux, P. (1988). Co-planar Stereotaxic Atlas of the Human Brain.
Thieme, New York.
Viinikainen, M., Jääskeläinen, I.P., Alexandrov, Y., Balk, M.H., Autti, T., & Sams, M.
(2010). Nonlinear relationship between emotional valence and brain activity:
Evidence of separate negative and positive valence dimensions. Human Brain
Mapping, 31, 1030-1040.
Volz, K.G., Schubotz, R.I., & von Cramon, D.Y. (2003). Predicting events of varying
probability: uncertainty investigated by fMRI. NeuroImage, 19, 271-280.
Wagner, A. R. & Brandon, S. E. (1989). Evolution of a structured connectionist model of
Pavlovian conditioning (AESOP). In S. B. Klein & R. R. Mowrer (Eds.),
Contemporary Learning Theories: Pavlovian Conditioning and the Status of
Traditional Learning Theory. (pp. 149-189). Hillsdale, NJ: Erlbaum.
45
Table 1. Regions showing conditioned diminution of the UCR.
Talairach
coordinates
Region
Vol (mm3)
Main Effect of Stimulus type
Dorsolateral PFC
x
y
CS+UCS vs.
CS−UCS
CS−UCS vs.
UCS alone
CS+UCS vs.
UCS alone
Trait
SCR
UCS
Expectancy
z
t
t
t
r
r
r
Right
3,668
30.3
44.0
27.4
− 6.03
n.s.
− 6.37
0.16
− 0.01
− 0.28
Right
14,229
40.3
12.8
36.6
− 4.23
n.s.
− 5.18
0.19
− 0.01
− 0.42*
Left
11,710
− 37.7
10.2
38.1
− 3.26
− 2.96
− 4.67
0.12
0.06
− 0.48*
Left
1,110
− 32.6
50.1
19.2
− 3.20
n.s.
− 4.75
0.19
0.01
− 0.38*
Left
908
− 21.1
41.0
40.2
− 4.53
n.s.
− 2.89
0.27
0.00
− 0.28
22,771
1.0
20.3
38.6
− 3.84
n.s.
− 5.07
0.23
0.12
− 0.47*
3,210
42.2
− 52.5
42.6
− 3.65
n.s.
− 4.77
0.33
0.06
− 0.38*
Dorsomedial PFC
Inf. Parietal Lobule
Right
Right
892
48.5
− 42.4
25.0
− 3.50
n.s.
− 3.68
0.15
0.04
− 0.27
Left
4,672
− 42.2
− 51.1
42.9
n.s.
− 4.69
− 5.29
0.22
0.08
− 0.54*
Right
1,398
38.4
15.3
0.3
n.s.
n.s.
− 3.70
0.24
0.08
− 0.41*
Left
4,731
− 37.5
13.7
− 0.8
− 3.54
n.s.
− 4.67
0.28
0.26
− 0.49*
11,381
− 1.7
− 34.4
32.5
− 4.08
n.s.
− 5.05
0.21
0.06
− 0.39*
12.0
48.9
13.6
− 4.58
n.s.
n.s.
0.05
0.13
− 0.26
Anterior Insula
Posterior Cingulate
Stimulus x Trial Interaction
Ventromedial PFC
656
Early test Trials
Late test Trials
n.s.
n.s.
n.s.
0.14
0.10
0.09
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center of mass for areas of activation. Significance criteria:
ANOVA F[21] > 5.63, p < 0.05 (corrected); t[20] p < 0.05 (corrected). Significance criteria for two-tailed correlations: * indicates p < 0.05
(corrected).
46
Table 2. Regions showing change over time.
Region
3
Talairach coordinates
x
z
y
Hemisphere
Vol (mm )
Right
2,379
22.0
42.2
29.3
Right
1,308
22.7
18.4
47.0
Left
9,230
-40.4
11.6
33.2
Dorsomedial PFC
Left
22,211
-2.2
31.2
27.2
Posterior Cingulate
Left
5,628
-1.3
-47.6
26.2
Anterior Insula
Right
1,991
38.7
18.6
0.1
Dorsolateral PFC
Left
2,068
-35.9
13.4
-1.7
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center
of mass for areas of activation. Significance criteria: F[21] > 9.01; p < 0.05 (corrected).
47
Table 3. Regional activity varying with trait anxiety.
Talairach coordinates
Region
3
Hemisphere
Vol (mm )
Right
702
36.2
9.4
46.8
0.33
Right
429
36.7
28.1
36.8
0.38
Dorsomedial PFC
Left
1,155
− 2.4
14.4
50.3
0.34
Posterior Cingulate
Left
711
− 1.1
− 43.5
24.6
0.35
Inf. Parietal Lobule
Right
992
42.2
− 53.6
43.3
0.39
Dorsolateral PFC
x
y
z
Trait
r
Left
744
− 40.1 − 59.4
42.6
0.33
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center of
mass for areas of activation. Significance criteria: p < 0.05 (corrected).
48
Table 4. Regions showing a relationship between anticipatory and threat-related activity.
UCR DIMINUTION FUNCTIONAL ROIs
ANTICIPATORY BRAIN ACTIVATION
Talairach coordinates
Region
Vol
(mm3)
x
y
Talairach coordinates
z
Region
Main Effect of Stimulus type
Dorsolateral PFC
Vol
(mm3)
x
y
z
749
29.1
21.8
44.7
Dorsomedial PFC2g
8,864
7.9
− 9.2
41.3
Dorsomedial PFC2dh
4,177
7.9
12.3
41.4
Dorsolateral PFC
Left
11,710
− 37.7
10.2
38.1
Left
1,110
− 32.6
50.1
19.2
Left
908
− 21.1
41.0
40.2
↔
↔
↔
Right1b
Dorsolateral PFC
Right1d
Inf. Parietal Lobule
Right
2,114
30.1
14.4
42.1
Left
d
1,219
− 35.1
20.9
34.9
Left
c
1,035
− 33.0
46.3
20.3
638
27.0
20.3
41.0
1,002
7.4
20.9
56.6
Dorsomedial PFC2f
645
16.2
− 5.7
45.9
Dorsomedial PFC2e
12,121
6.7
− 13.1
45.4
Dorsolateral PFC
3,210
42.2
− 52.5
42.6
↔
Right1a
Stimulus x Trial Interaction
↔
Ventromedial PFC
656
12.0
48.9
13.6
Dorsomedial PFC
Regression of Unconditioned SCR
Amygdala
Right
460
24.9
− 4.3
− 16.1
Left
511
− 24.9
− 4.4
− 14.8
↔
↔
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center of mass for areas of
activation. Significance criteria: t[20] > 2.85, p < 0.05 (corrected). A stimulus (CS+UCS vs CS−UCS) x UCR
amplitude (from ROI on left side of table) interaction was observed in the anticipatory response (i.e. the CR) within
the dorsolateral and dorsomedial PFC (right side of the table). Numbers (1 & 2) denote areas of overlap within the
regions showing an interaction effect. Letters (a-h) correspond to images presented in Figure 7.
49
Acquisition Trials
Test Trials
CS+
CS+
UCS
UCS
CS-
CS-
UCS
0
10 20 30 40
Time (Seconds)
UCS
50
No CS
UCS
0
10 20 30 40
Time (Seconds)
50
Figure 1. Conditioning procedure. Acquisition blocks consisted of CS+ (8 trials), CS− (8
trials), and test trials (1 CS+UCS, 1 CS−UCS, and 1 UCS alone trial). Four acquisition
blocks were presented, followed by a block of test trials. The test trial block consisted of
10 CS+UCS trials, 10 CS−UCS trials, and 10 UCS alone trials. Stimuli were
counterbalanced and presented in a pseudorandom order such that no more than two trials
of the same stimulus were consecutively presented.
50
UCS Expectancy
100
a
80
60
40
CS+UCS
CS-UCS
UCS alone
20
SCR
0.20
b
0.15
0.10
0.05
1-7
8-14
Trials
Figure 2. UCS Expectancy and Unconditioned SCR. a) Learning-related differences in
UCS expectancy. UCS expectancy on Early test trials (1-7) was higher to the CS+UCS
than to CS−UCS and UCS alone trials. UCS expectancy on Late test trials (8-14)
remained high for the CS+UCS, while UCS expectancy during the CS−UCS increased
such that ratings were greater than those for the UCS alone. b) Learning-related changes
in unconditioned SCR expression were also observed. Unconditioned SCRs were
diminished on CS+UCS trials compared to CS−UCS and UCS alone trials. No
differences were observed between CS−UCS and UCS alone trials. Error bars reflect
SEM after adjusting for between-subject variance (Loftus & Masson, 1994).
51
Posterior Cingulate
0.5
0.4
fMRI signal (% )
fMRI signal (% )
Dorsomedial PFC
*
*
0.3
0.2
0.1
CS+UCS
CS-UCS
0.5
0.4
0.1
CS+UCS
UCS alone
fMRI signal (%)
fMRI signal (% )
x = -5
*
CS-UCS
*
0.2
0.1
0.4
0.3
*
*
0.2
0.1
0.0
CS+UCS
CS-UCS
UCS alone
CS+UCS
CS-UCS
0.5
*
0.4
z = 47
*
0.3
UCS alone
IPL
fMRI signal (% )
fMRI signal (% )
IPL
0.2
0.1
*
0.4
*
0.3
0.2
0.1
0.0
CS+UCS
CS-UCS
UCS alone
CS+UCS
CS-UCS
*
0.5
0.4
0.3
0.2
Right
0.1
CS+UCS
CS-UCS
UCS alone
Insula
UCS alone
Left
z = -5
fMRI signal (% )
Insula
fMRI signal (% )
UCS alone
Dorsolateral PFC
0.5
0.3
*
0.2
Dorsolateral PFC
0.4
*
0.3
0.5
*
0.4
*
0.3
0.2
0.1
CS+UCS
CS-UCS
UCS alone
Figure 3. UCR diminution within the fMRI signal response. Significant diminution of the
unconditioned fMRI signal response was observed within several brain regions (see
Table 1) including the prefrontal cortex (PFC), inferior parietal lobule (IPL), and insula
during test trials. UCR amplitude within each of these regions was reduced when the
UCS followed the CS+ (i.e. CS+UCS trials) compared to when the UCS was presented
alone. Many of these regions also showed diminished UCRs on CS+UCS versus
CS−UCS trials. Graphs reflect the mean amplitude (% signal change) of all voxels within
volumes of activation. Error bars reflect SEM after adjusting for between-subject
variance (Loftus & Masson, 1994). Asterisk indicates significant difference.
52
CS+UCS
CS-UCS
UCS alone
fMRI signal (% )
0.4
x = 11
0.2
0.0
-0.2
1-7
8-14
Trials
Figure 4. Stimulus x trial interaction within the ventromedial PFC. The unconditioned
fMRI signal response within the ventromedial PFC was larger to the CS−UCS than
CS+UCS and UCS alone on Early, but not Late test trials. Graph reflects the mean
amplitude (% signal change) of all voxels within the volume of activation. Error bars
reflect SEM after adjusting for between-subject variance (Loftus & Masson, 1994).
53
Dorsolateral PFC
Dorsomedial PFC
1.0
fMRI signal (% )
fMRI signal (% )
1.0
0.5
0.0
-0.5
0.0
-0.5
20
25
30
35
40
45
20
25
30
35
Trait Anxiety
Trait Anxiety
IPL
IPL
40
45
40
45
1.0
0.5
Right
Left
z = 45
0.0
-0.5
fMRI signal (% )
1.0
fMRI signal (% )
0.5
0.5
0.0
-0.5
20
25
30
35
40
45
20
Trait Anxiety
25
30
35
Trait Anxiety
Figure 5. Trait anxiety and the unconditioned fMRI signal response. UCR magnitude
within several regions (see Table 3) including the dorsolateral PFC, dorsomedial PFC,
and IPL varied with inter-subject differences in trait anxiety, such that as anxiety level
increased UCR magnitude increased.
54
Right
Left
y = -6
fMRI signal (% )
fMRI signal (% )
a
0.4
0.2
0.0
-0.2
0.0
0.1
0.2
0.4
b
0.2
0.0
-0.2
0.3
0.0
0.15
c
*
0.10
0.1
0.2
0.3
SCR
fMRI signal (% )
fMRI signal (% )
SCR
0.05
0.00
-0.05
0.15
d
*
0.10
0.05
0.00
-0.05
CS+UCS
CS-UCS
UCS alone
CS+UCS
CS-UCS
UCS alone
Figure 6. Relationship between amygdala and unconditioned SCR. a & b) A linear
relationship was observed between the unconditioned SCR and the unconditioned fMRI
signal response within the amygdala. These findings suggest the amygdala plays an
important role in the control of unconditioned SCRs during Pavlovian conditioning. c &
d) The volumes of left and right amygdala activation that correlated with unconditioned
SCR were used as functional ROI to determine whether the amygdala UCR was
diminished on CS+UCS trials. Learning-related changes in the unconditioned fMRI
signal response were observed within the amygdala. The amplitude of the unconditioned
fMRI signal response was diminished on CS+UCS trials compared to UCS alone trials.
Error bars reflect SEM after adjusting for between-subject variance (Loftus & Masson,
1994).
55
Relationship between Anticipatory and Threat-related Activity
dmPFC CR↔L. Amyg UCR
dlPFC CR↔IPL UCR
r = -.72, p<0.05;
r = .27, p = n.s.
a
r = -.39, p = n.s.;
e
z = 38
x = 13
dlPFC CR↔dlPFC UCR
r = -.71, p<0.05;
dmPFC CR↔R. Amyg UCR
r = .47, p<0.05
b
r = -.58, p<0.05;
x = 13
dlPFC CR↔dlPFC UCR
r = -.58, p<0.05;
dmPFC CR↔dlPFC UCR
r = .62, p<0.05
c
r = -.50, p<0.05;
r = .57, p<0.05
g
z = 21
x = 13
dlPFC CR↔dlPFC UCR
r = -.71, p<0.05;
dmPFC CR↔dlPFC UCR
r = .34, p = n.s.
d
r = -.70, p<0.05;
r = .41, p = n.s.
h
z = 38
x = 13
Right
Left
Dorsolateral PFC
CS+UCS
CS-UCS
Dorsomedial PFC
r = 0.37, p = n.s.
dlPFC UCR
dlPFC UCR
r = .56, p<0.05
f
z = 38
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
r = .67, p<0.05
r = -0.71, p < 0.05
-0.2
0.0
0.2
0.4
0.6
0.8
r = 0.41, p = n.s.
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
r = -0.70, p < 0.05
-0.2
dlPFC CR
0.0
0.2
0.4
dmPFC CR
56
0.6
0.8
Figure 7. Relationship between anticipatory and threat-related activity. Threat-related
activity, extracted from the ROI depicted in Table 1 & Figures 3, 4, & 6, was included in
a regression analysis to investigate differences in the relationship between anticipatory
activity and threat-related activity (% signal change) on CS+UCS and CS−UCS trials.
Differences (CS+UCS vs CS−UCS) in the relationship between anticipatory (i.e. CR) and
threat-related (i.e. UCR) activity were observed in several areas of the PFC (a-h). These
findings suggest that prefrontal anticipatory activity inhibits the threat-related response
on CS+UCS trials within many of the brain regions that showed UCR diminution.
Correlation values comparing anticipatory and threat-related responses on CS+UCS (blue
circles) and CS−UCS (red triangles) trials are presented above the brain images.
Correlation values above image (d) represent the activation observed within the left
dlPFC, while correlation values for the right dorsolateral (dlPFC) and dorsomedial
(dmPFC) prefrontal cortex are presented in the bottom graphs. Talairach coordinates for
the depicted areas of activation are presented in Table 4 and labeled with the letters (a-h)
corresponding to each image above.
57
58
NEURAL SUBSTRATES UNDERLYING LEARNING-RELATED
CHANGES OF THE UNCONDITIONED FEAR RESPONSE
KIMBERLY H. WOOD, DYSTANY KUYKENDALL, LAWRENCE W. VER HOEF,
AND DAVID C. KNIGHT
Submitted to The Open Neuroimaging Journal
Format adapted for dissertation
Abstract
The ability to predict an impending threat during Pavlovian conditioning diminishes the
emotional response that is produced once the threat is encountered. Diminution of the
threat response appears to be mediated by somewhat independent associative learning
and expectancy-related processes. Therefore, the present study was designed to better
understand the neural mechanisms that support the associative learning processes that
influence the threat-related emotional response. Healthy volunteers participated in a
Pavlovian fear conditioning procedure in which trait anxiety, expectation of the
unconditioned stimulus (UCS expectancy), skin conductance response (SCR), and
functional magnetic resonance imaging (fMRI) signal were assessed. UCS expectancy
ratings were higher on predictable trials of the UCS compared to unpredictable trials.
Threat-related SCR expression was diminished on predictable trials vs. unpredictable
trials of the UCS. UCR diminution was also observed within left dorsolateral PFC,
dorsomedial PFC, ventromedial PFC, and left anterior insula, whereas potentiation of the
threat-related fMRI signal response was observed within left dorsolateral PFC, inferior
parietal lobule (IPL), and posterior insula. A negative relationship was observed between
UCS expectancy and the threat-related response within dorsomedial PFC, ventromedial
PFC, and anterior insula. Threat-related activity within dorsomedial PFC and left IPL
varied with unconditioned SCR. Finally, anticipatory activity within the PFC, posterior
cingulate, and amygdala showed an inverse relationship with threat-related activity
within the brain regions that showed UCR diminution. The current findings suggest that
the PFC and amygdala support learning-related processes that modulate the emotional
response to a threat.
59
Key words: fMRI, learning, conditioning, unconditioned response, prefrontal cortex,
emotion, fear, anxiety, skin conductance
60
Introduction
Fear is considered an important defense mechanism due to its evolutionary role in
survival [1–3]. The ability to form associations between a dangerous event and the cues
that predict it allows an organism to better adapt to a changing environment [1,4]. An
important aspect of this type of associative learning (i.e. Pavlovian fear conditioning) is
that it allows an organism to more effectively avoid, escape, or minimize the impact of an
impending threat [3–7]. Thus, from a functional perspective it is the response to the threat
itself (i.e. unconditioned stimulus: UCS) that directly impacts survival and therefore may
be the most biologically relevant feature of Pavlovian fear conditioning [4].
During Pavlovian fear conditioning, a neutral conditioned stimulus (CS) is paired
with an aversive UCS. The conditioned response (CR) produced by the CS is typically
used to index fear expression. For example, an increase in transient SCR expression
during presentation of the CS. Traditionally, CR expression is taken as evidence that an
association between the CS and UCS has been formed. In contrast, the unconditioned
response (UCR) is generally considered an automatic, unlearned reaction to the aversive
UCS. However, prior work has shown that unconditioned skin conductance responses
(SCRs) diminish as associative learning develops during Pavlovian fear conditioning [8–
12]. For example, UCR amplitude is decreased to paired compared to unpaired
presentations of the CS and UCS [9-14]. UCR diminution also appears to be mediated by
conscious UCS expectancies [10,15–17]. For example, greater UCR diminution has been
observed when participants expect a UCS compared to when the UCS is unexpected
[10,15]. This phenomenon is generally referred to as conditioned UCR diminution.
61
A few brain imaging studies have investigated the neural correlates of conditioned
UCR diminution [10,15,17,18]. These studies have demonstrated conditioned diminution
of UCR activity within the dorsolateral PFC (dlPFC), dorsomedial PFC (dmPFC),
ventromedial PFC (vmPFC), anterior cingulate cortex (ACC), posterior cingulate cortex
(PCC), inferior parietal lobule (IPL), anterior insula, and amygdala [10,15,17,18]. These
findings are consistent with prior research that suggests the PFC regulates the emotional
response [19,20]. This prior work indicates that the PFC projects to the amygdala and
provides regulatory control over emotion-related processes during fear conditioning
[1,19,21]. Further, the threat-related fMRI signal response within the PFC and amygdala
influences the autonomic response (e.g. SCR) that is produced [10,15,17,22]. This
process appears to be critical for normal, healthy emotional function.
Converging lines of research indicate that healthy emotion regulation relies upon
the PFC [19,20,23–25], and that anxiety disorders may be linked to insufficient
regulatory control from the PFC. Further, PFC dysregulation is associated with increased
amygdala reactivity [26–31] and an exaggerated emotional response to threats
[11,32,33]. For example, prior work has shown that participants with low trait anxiety
exhibit greater vmPFC activation compared to participants with high trait anxiety during
cued fear conditioning [34]. In contrast, individuals with high trait anxiety showed a
diminished vmPFC response that was associated with greater fear conditioned SCRs
compared to participants with low trait anxiety [34]. Furthermore, our prior work has
demonstrated that unconditioned fMRI signal responses from several brain regions
fluctuate with individual differences in trait anxiety level [17]. Specifically, trait anxiety
varied with dlPFC, dmPFC, PCC, and IPL activity such that, as anxiety level increased
62
the threat-related fMRI signal response within these brain regions increased [17]. Taken
together, these studies suggest that anxiety level affects the magnitude of anticipatory and
threat-related brain activation, which in turn influences the peripheral expression of
emotion.
Associative learning and expectancy processes are additional factors that
influence the response produced by a threat. Prior work has demonstrated a decrease in
the magnitude of brain activation once a cue-outcome relationship is established and
predictable [35,36]. However, when an outcome violates expectations (i.e. surprise event)
an increase in the magnitude of brain activity is observed [35,36]. Further, the magnitude
of brain activation in response to an aversive stimulus is dependent upon the expectation
of whether the aversive outcome is a certainty or only a possibility [15,38]. Prior work
has demonstrated that the amplitude of the threat-elicited fMRI signal response within
regions of the PFC, insula, cingulate, IPL, and amygdala varies with UCS expectancy
[10,15,17,38]. More specifically, as UCS expectancy increases during CS presentations,
the amplitude of the threat response decreases. These findings suggest that conditioned
diminution of the UCR is in part mediated by UCS expectancies that are supported by
regions of the PFC. In turn, these learning-related changes in the brain’s response to a
threat appear to modify the peripheral emotional response that is expressed. For example,
several studies have demonstrated that as UCS expectancy increases, the magnitude of
unconditioned SCRs decrease [10,11,15]. Further, these findings parallel the learningrelated changes observed within the unconditioned fMRI signal response [10,15,17,38].
Taken together, these findings suggest that conscious expectation of an imminent threat
may play an important role in modulation of the emotional response produced. However,
63
prior work has also demonstrated threat-elicited SCRs that did not diminish as
expectation increased [17]. For example, unconditioned SCRs produced by a UCS that
followed a CS− did not differ from SCRs to a UCS presented alone even though UCS
expectancies differed between these conditions [17]. Prior work has also demonstrated
greater UCR diminution to a UCS that followed a CS+ compared to a UCS presented
after a CS− even when UCS expectancy ratings were equivalent [11]. In addition,
diminished unconditioned SCRs have been observed to a UCS that followed a CS+
compared to a UCS that followed a CS− even after participants had been explicitly
informed that the UCS would follow both CS presentations [12]. Taken together, these
studies suggest that modulation of the threat-related emotional response is not solely
mediated by conscious expectations. Instead, the findings suggest that associative
learning processes independent of UCS expectancy also influence UCR diminution.
Given that prior work suggests UCR diminution is in part mediated by
expectancy independent processes, the present study was designed to better understand
associative learning processes that influence UCR expression in the absence of
differential UCS expectancies. The aim of this study was to determine the neural
substrates that support expectancy-independent conditioned diminution of the threatrelated emotional response. Based on previous research, we expected a decrease in the
unconditioned fMRI signal to develop with associative learning independent of UCS
expectancy [11]. Given the importance of the amygdala, dlPFC, dmPFC, and vmPFC in
the regulation and expression of emotion [1,19,20,38], we hypothesized that these brain
regions would show UCR diminution during Pavlovian conditioning. Further, we
expected activity within these brain regions to vary with individual differences in trait
64
anxiety. In turn, we expected amygdala activity to vary with the learning-related
modulation of unconditioned SCRs [17,22,39–41].
Materials and Methods
Participants: Twenty-one healthy right-handed volunteers participated in this study [8
male, 13 female; age = 23.05 ± 0.82 years (mean ± SEM); range = 19-34 years]. All
participants were included in the UCS expectancy and fMRI data analyses. However,
four non-responsive (SCR < 0.05 uSiemens) participants were excluded from SCR data
analyses. Thus, a total of seventeen participants were included in the SCR analyses (7
male, 10 female; age = 23.59 ± 0.95 years; range = 19-34 years). Fifteen participants
were included in the secondary fMRI data analyses that examined the relationship
between behavior and brain activation (6 male, 9 female; age = 23.87 ± 1.06 years; range
= 19-34 years). The six participants excluded from these secondary fMRI analyses
consisted of the four participants with non-responsive SCR and two participants that did
not complete the trait anxiety assessment. All subjects provided written informed consent
in compliance with the University of Alabama at Birmingham Institutional Review
Board.
State-Trait Anxiety Inventory: Prior to the conditioning session, participants completed
the State-Trait Anxiety Inventory (STAI; Form Y) for Adults [42]. The STAI is a selfassessment questionnaire that measures state and trait anxiety in terms of general
negative affect [43]. The state scale reflects anxiety level at the current moment, whereas
the trait scale reflects anxiety level experienced in general [42].
65
Conditioned and unconditioned stimuli: Participants were exposed to a differential fear
conditioning procedure in which the conditioned and unconditioned stimuli were
presented through MR-compatible pneumatic headphones. Two tones (1025 and 1050
Hz; 10 s duration; 20 s ITI) that were difficult to discriminate served as the CSs. Our pilot
work indicated that stimuli presented at these frequencies can be differentiated when
presented back-to-back, but are difficult to discriminate when separated by a 20 s ITI. A
loud (100db) white-noise served as the UCS (0.5 s duration). The UCS coterminated with
one tone (CS+) and the second tone was presented alone (CS−) during the acquisition
phase. The acquisition phase consisted of four 590 s blocks. A total of thirty-two trials of
each CS were presented during the acquisition phase (8 trials of each CS were presented
in each block). Additionally, each acquisition block contained one set of 3 test trials that
consisted of UCS presentations that coterminated with the CS+ (CS+UCS) and CS−
(CS−UCS), as well as presentations of the UCS alone (Figure 1). Thus, acquisition
blocks consisted of 19 trials (8 CS+ & 8 CS− acquisition trials, as well as 1 CS+UCS, 1
CS−UCS, & 1 UCS alone test trial). Test trials during the acquisition blocks were
presented on trials 17-19 of blocks 1 and 2; trials 13-15 of block 3; and trials 9-11 of
block 4. The acquisition phase was followed by a 920 s test phase that consisted of 30 test
trials (10 CS+UCS trials, 10 CS−UCS trials, 10 UCS alone trials). In total, there were 14
test trials for each stimulus (4 from the acquisition phase, 10 from the test phase). The 14
test trials were grouped into the first seven test trials (Early test trials) and the last seven
test trials (Late test trials) for further analysis. The test trials were binned in this manner
to evaluate learning-related changes in UCS expectancy in a manner consistent with our
prior work using CS presentations that were easy to discriminate [17]. The stimuli were
66
counterbalanced and presented in a pseudorandom order such that no more than two trials
of the same stimulus were consecutively presented.
UCS expectancy: UCS expectancy was used to measure expectation of the UCS and
assess whether the relationship between the CS and UCS had been learned. Using
Presentation software (Neurobehavioral Systems, Inc.; Albany, CA), a UCS expectancy
rating scale was presented on an IFIS-SA LCD (Invivo Corp.; Gainesville, FL) video
screen located behind the subject's head and viewed through a mirror in front of the
participant attached to the RF coil. An MRI compatible joystick (Current Designs;
Philidelphia, PA) was used to monitor subjects’ expectancy of receiving the UCS. The
joystick controlled a rating bar which was presented throughout the conditioning session
on the video screen. Subjects were instructed to rate their UCS expectancy on a momentby-moment basis using a continuous scale from 0 to 100 (0 = certain the UCS would not
be presented, 50 = uncertain whether the UCS would be presented, 100 = certain the UCS
would be presented) to reflect their current UCS expectancy. UCS expectancy was
calculated as the average response (1 s sample) at UCS onset. Additional details on this
methodology have been published previously [44].
Skin conductance response: An MRI compatible physiological monitoring system
(Biopac Systems; Goleta, CA) was used to collect SCR data as described in prior work
[44]. SCR was sampled (2,000 Hz) with a pair of disposable radio-translucent electrodes
(1cm diameter, Biopac Systems; Goleta, CA) from the distal phalanx of the middle and
ring fingers of the nondominant hand. SCR data were processed using Biopac
67
AcqKnowledge 4.1 software. A 1 Hz low pass digital filter was applied and SCR data
were resampled at 250 Hz. Unconditioned SCRs were limited to those that occurred
within 10 s following the UCS presentation. Unconditioned SCRs smaller than 0.05
uSiemens were scored as 0.
Functional MRI: Structural and functional imaging was completed on a 3 Tesla Siemens
Allegra scanner. High-resolution anatomical images (MPRAGE) were obtained in the
sagittal plane using a T1 weighted series (TR=2300 ms, TE=3.9 ms, flip angle=12⁰,
FOV=25.6 cm, matrix=256 x 256, 160 slices, slice thickness1 mm, 0.5 mm gap) to serve
as an anatomical reference. Blood oxygen level dependent fMRI of the entire brain was
conducted using a gradient-echo echoplanar pulse sequence in an oblique-axial
orientation (TR=2000 ms, TE=30 ms, flip angle=70º, FOV=24 cm, matrix=64 x 64,
interleaved acquisition, 34 slices, slice thickness=4 mm, no gap) during each block of
stimulus presentations. Functional image processing was performed with the Analysis of
Functional NeuroImages (AFNI) software package [45]. Echo-planar time series data
were corrected for slice timing offset, motion corrected, concatenated, reregistered to the
fifth volume of the first imaging block, and spatially blurred using a 4 mm full-width-athalf-maximum Gaussian filter.
Functional MRI data were analyzed at the individual subject level using the input
from all stimuli in a multiple linear regression using a gamma variate hemodynamic
response function. Regressors to account for brain activity not related to the UCR on test
trials included reference waveforms for the CS+ and CS− during acquisition, UCS during
acquisition, CS+ and CS− on test trials, joystick movement, and head motion parameters.
68
The regressors of interest for this study modeled the unconditioned fMRI signal response
to UCS presentations during each type of test trial (i.e. CS+UCS, CS−UCS, and the UCS
alone). Separate reference waveforms were used for Early and Late test trials in this
analysis. Percent signal change on test trials was used as an index of the magnitude of the
unconditioned fMRI signal response produced by the UCS. Functional maps reflecting
percent signal change were converted to the Talairach and Tournoux stereotaxic
coordinate system for group analyses [46].
Based on prior work [10,15,17], group level analyses were restricted using an
anatomical mask, to the PFC, cingulate cortex, IPL, insula, and amygdala to reduce the
number of voxel-wise comparisons. We conducted a repeated-measures ANOVA to test
for a main effect of stimulus (CS+UCS, CS−UCS, and UCS alone) and trial (Early vs.
Late test trials), as well as a stimulus x trial interaction. A voxel-wise threshold of p <
0.05 (corrected) was employed by using an uncorrected threshold of p < 0.005 and a
cluster volume larger than 510 mm3 (9 voxels of 3.75 x 3.75 x 4.00 mm dimension).
These threshold criteria were based on Monte Carlo simulations that were used to reject
smaller clusters of activation produced by chance alone (false positives) [47,48]. Followup t-test comparisons were conducted in SPSS on the mean percent signal change
activation passing the significance threshold (p < 0.05 Bonferonni corrected) for the
ANOVA.
Two different analysis procedures (i.e. correlation and multiple linear regression)
were completed to investigate the relationship between our behavioral measures (i.e. trait
anxiety, UCS expectancy, and SCR) and the unconditioned fMRI signal response from
brain regions that demonstrated learning-related changes in the ANOVA (i.e. functional
69
regions of interest; ROI). Although these analyses are similar there are important
differences (see [17] for additional discussion). In short, separate correlation analyses
assessed the relationship between each of our behavioral measures and the mean percent
signal change within an ROI as a whole, while the regression analysis evaluated these
relationships on a voxel-wise basis. The four participants without measurable SCR data
and the two participants without a trait anxiety score were excluded from this regression
analysis because there were no data points to include in the model. AlphaSim [45,47] was
used to conduct Monte Carlo simulations limited to the functional ROI from our
repeated-measures ANOVA that demonstrated a main effect of stimulus or stimulus x
trial interaction. A voxel-wise threshold of p < 0.005 and a cluster volume larger than
225mm3 (4 voxels of 3.75 x 3.75 x 4.00mm dimension) was employed, resulting in a
FWE corrected significance threshold of p < 0.05. Prior work has demonstrated a
relationship between the magnitude of the fMRI signal response within the amygdala and
SCR production during Pavlovian fear conditioning [10,17,22,39–41]. Therefore an
anatomical mask was employed to include the amygdala in the group level regression
analysis.
Prior work suggests that the PFC and amygdala produce and regulate the
emotional response to aversive stimuli [17, 19,38]. Therefore, we completed an
additional voxel-wise multiple regression analysis to identify anticipatory PFC, cingulate,
and amygdala activity that varied with the unconditioned fMRI signal response obtained
from the functional ROI from our ANOVA. This analysis was restricted to CS+UCS and
CS−UCS trials because a CS was not presented during UCS alone trials to elicit an
anticipatory response. This analysis included a regressor representing trial type (CS+UCS
70
& CS−UCS), a regressor for the amplitude of the unconditioned fMRI signal response
from our functional ROI, and a regressor for the interaction of trial type and
unconditioned fMRI signal response amplitude. As indicated by Monte Carlo
simulations, a voxel-wise threshold of p < 0.005 and a cluster volume larger than 510
mm3 (9 voxels of 3.75 x 3.75 x 4.00 mm dimension) was employed resulting in a FWE
corrected significance threshold of p < 0.05. Given our a priori hypotheses and the
relatively small volume of the amygdala, we used a voxel-wise threshold of p < 0.005
and a cluster volume larger than 112 mm3 (2 voxels of 3.75 x 3.75 x 4.00 mm dimension)
for this area (p < 0.05 corrected).
Results
UCS expectancy: Repeated measures ANOVA revealed significant differences in UCS
expectancy during the test trials. Results showed a main effect for stimulus type (F[1,20]
= 38.30, p < 0.05) and a main effect for trial (F[1,20] = 33.68, p < 0.05). There was no
stimulus by trial interaction (F < 1.00). UCS expectancy was greater during Early test
trials on CS+UCS [mean ± SEM (adjusted for between subject variance [49]): 65.85 ±
2.86; t[20] = 5.55, p < 0.05] and CS−UCS (70.60 ± 3.97; t[20] = 5.48, p < 0.05) than on
UCS alone trials (34.31 ± 3.89). UCS expectancy did not differ for CS+UCS and
CS−UCS (t[20] = -1.00) presentations during Early test trials. UCS expectancy was
greater during Late test trials on CS+UCS (86.09 ± 2.84; t[20] = 6.20, p < 0.05) and
CS−UCS (91.83 ± 2.97; t[20] = 6.29, p < 0.05) compared to UCS alone (48.58 ± 4.56)
trials. During the Late test trials, UCS expectancy was also greater during CS−UCS
presentations than on CS+UCS presentations (t[20] = -2.31, p < 0.05) (Figure 2a).
71
Skin conductance response: Repeated measures ANOVA also revealed significant
differences in unconditioned SCR amplitude during the test trials. There was a main
effect for stimulus type (F[1,16] = 4.85, p < 0.05). There was also a trend for a main
effect of trial (F[1,16] = 4.44, p = 0.051), and a trend for a stimulus by trial interaction
(F[1,16] = 4.39, p = 0.052). T-test comparisons revealed a significantly diminished
unconditioned SCR for CS+UCS (1.07 ± 0.08; t[16] = -3.05, p < 0.05) and CS−UCS
trials (1.38 ± 0.17; t[16] = -2.04, p < 0.05) compared to the UCS alone (1.88 ± 0.22)
during Early test trials. There was not a significant difference in unconditioned SCR
between CS+UCS and CS−UCS trials (t[16] = -1.49) during Early test trials. Also, there
was not a significant difference in unconditioned SCR between CS+UCS trials (0.98 ±
0.20) and CS−UCS trials (0.90 ± 0.18; t < 1.00) during Late test trials. Unconditioned
SCR during CS+UCS (t[16] = -1.08) and CS−UCS (t[16] = -1.38) trials also did not
differ from UCS alone trials (1.30 ± 0.15) during the Late test trials (Figure 2b).
Functional MRI: Repeated measures ANOVA revealed significant differences in the
magnitude of the unconditioned fMRI signal response within several brain regions
(Tables 1 and 2; Figure 3). UCR diminution was observed within the dlPFC, dmPFC,
vmPFC, and anterior insula replicating prior work [10,15,17] (Table 1; Figure 3). We
also observed potentiation of the UCR within several brain regions including the dlPFC,
IPL, and posterior insula (Table 2). Within each of these brain regions the unconditioned
fMRI signal response demonstrated a main effect for stimulus type (F[20] > 6.06; p <
0.05 corrected). A main effect for trial (i.e. Early vs. Late test trials) was observed within
72
the dlPFC, IPL, and anterior insula (Table 3). A stimulus x trial interaction was observed
within left IPL (Table 2). T-test comparisons were completed on the mean fMRI signal
from each volume of activation that passed the significance threshold (p < 0.05 corrected)
for the main effect of stimulus type. There was no difference in the unconditioned fMRI
signal response for CS+UCS compared to CS−UCS trials within the functional ROIs
identified by the ANOVA. The unconditioned fMRI signal response was diminished for
the CS+UCS and CS−UCS compared to the UCS alone within left dlPFC, dmPFC,
vmPFC, and left anterior insula (Table 1). However, greater activation was observed for
CS+UCS and CS−UCS trials compared to UCS alone trials within left dlPFC and
bilateral IPL. Additionally, greater activation for CS−UCS vs. UCS alone trials was
observed within bilateral posterior insula (Table 2).
A correlation analysis was performed on the mean percent signal change from
each of the functional ROIs from the ANOVA and trait anxiety, UCS expectancy, and
unconditioned SCR amplitude measures (Tables 1 and 2; p < 0.05 Bonferroni corrected).
The percent signal change within these brain regions did not vary with trait anxiety or
unconditioned SCR. These findings replicate prior work that used CS presentations that
were easy to discriminate during Pavlovian fear conditioning [17]. However, there was a
significant correlation observed between UCS expectancy ratings and activity within
dmPFC, vmPFC, left IPL, and left anterior insula (Tables 1 and 2), generally consistent
with prior research [17].
A voxel-wise multiple linear regression analysis was conducted to evaluate brain
activity within the functional ROI that varied with individual differences in behavior. The
amygdala was also included in this analysis because prior work suggests this brain region
73
mediates learning-related changes in SCR production [22,39–41]. The regression model
accounted for stimulus type, trait anxiety, UCS expectancy, and unconditioned SCR. This
analysis demonstrated that unconditioned SCR amplitude explained unique variability in
the activation observed within the left dmPFC (r = .34; Talairach coordinates: -7.5, -16.3,
50; volume: 280 mm3) and left IPL (r = .28; Talairach coordinates: -41.5, -36.2, 42.3;
volume: 237 mm3), but not within the amygdala. There were no brain regions within the
functional ROI that varied with UCS expectancy or trait anxiety.
We also conducted a group level regression analysis to investigate whether
anticipatory activation (i.e. the CR) within the PFC, cingulate, and amygdala varied with
threat-related activity (i.e. the UCR) from brain regions that demonstrated UCR
diminution (Table 1). Anticipatory activity within dlPFC, dmPFC, vmPFC, ventrolateral
PFC (vlPFC), posterior cingulate, and amygdala showed a negative relationship with the
threat-related fMRI signal response (i.e. UCR) within many of the brain regions in which
conditioned UCR diminution was observed (Table 4). Anticipatory activation within left
vlPFC was negatively correlated with the threat response within each of the brain areas
that showed UCR diminution. This effect was also observed between anticipatory
activation within the right vlPFC and threat-related activity within each of the functional
ROI that showed UCR diminution, except for the left dlPFC (Table 4). A negative
relationship was also observed between anticipatory activity within vmPFC and the
threat-related response within left dlPFC, dmPFC, and left anterior insula (Figure 4a and
c). A similar pattern was observed between anticipatory activation within dlPFC, dmPFC,
vlPFC, and PCC and the threat-related response within vmPFC (Figure 4e). Finally,
74
anticipatory amygdala activity showed a similar negative relationship with the threatrelated response within dmPFC and left anterior insula (Figure 4b and d).
Discussion
Learning the relationship between a threat and the cues that predict it is critical to
survival. These cue-threat relationships are established when there is a discrepancy
between expectation and outcome [48], by associative learning and expectancy-related
processes that may somewhat independently influence the magnitude of the threatelicited response [10–12,15,17]. Therefore, the present study used CS presentations that
were difficult to discriminate to investigate the neural processes that support learningrelated changes in the UCR in the absence of differential UCS expectancies to better
understand associative learning processes that mediate UCR diminution.
Learning-related differences in UCS expectancy and unconditioned SCR
expression were observed in the present study. By design, UCS expectancy ratings were
high on CS+UCS and CS−UCS trials. In contrast, UCS expectancy on UCS alone trials
were rated around 50 (Figure 2a). These findings demonstrate that participants expected
the UCS following the CS+ and CS−, but remained uncertain about the timing of the
UCS alone during the conditioning session. UCS expectancy was also associated with the
amplitude of unconditioned SCR expression. Specifically, unconditioned SCR amplitude
was diminished on CS+UCS and CS−UCS trials (when UCS expectancy was high)
compared to the UCS alone (when UCS expectancy was lower) on Early test trials
(Figure 2b). These findings are consistent with prior work that has shown a decreased
UCR when the UCS is predictable vs. unpredictable [8,9,11,49]. Based on our pilot data,
75
we also expected to observe greater unconditioned SCR diminution during CS+UCS than
CS−UCS trials in this study. Findings of differential SCR, with equivalent UCS
expectancy, would allow us to address questions related to learning independent of
conscious expectations. However, no difference in unconditioned SCR expression was
observed between CS+UCS and CS−UCS trials. Therefore, there is no evidence of
conditioned UCR diminution that is independent of UCS expectancy in the present study.
The lack of differential SCRs on CS+UCS vs. CS−UCS trials is likely due to the lack of
discriminative control gained by the CS. Given that the CS+ and CS− were difficult to
discriminate, evidenced by high expectancy ratings, participants appear to have
interpreted the acquisition phase as a 50% reinforcement schedule of a single CS rather
than separate presentations of a CS+ and CS−. Therefore, the lack of differential
unconditioned SCRs during CS+UCS compared to CS−UCS trials appears to be due to a
deficit in learning the CS discrimination.
Although there is no evidence of learning, independent of expectancy, we did
observe conditioned UCR diminution in other contrasts in the present study.
Unconditioned SCR amplitude was diminished on CS+UCS and CS−UCS trials
compared to UCS alone trials. This response pattern was also observed in the fMRI
signal within the left dlPFC (z = 35.0), dmPFC, vmPFC, and left anterior insula. These
findings are generally consistent with prior studies that have demonstrated UCR
diminution within these brain regions [10,15,17]. However, contrary to our prior work, an
enhanced threat-related fMRI response was observed within other brain regions including
a more superior region of left dlPFC (z = 51.7), bilateral IPL, and bilateral posterior
insula. Similar to the SCR data, there were no differences between CS+UCS and
76
CS−UCS trials within any of the brain regions that demonstrated learning-related changes
in the unconditioned fMRI signal response (Tables 1 and 2). Taken together, the present
findings replicate prior work that has shown learning-related changes in brain activation
that resemble the pattern of the emotional response produced [15,17,22,39,40].
Prior work suggests that UCS expectancy modulates the amplitude of the threatrelated fMRI signal response [10,15,17]. Prior work has also shown that conscious
expectation influences the emotional response evoked by a threat [10,11,15–17]. Similar
results were also observed in the current study. Consistent with our prior work, a negative
relationship was observed between UCS expectancy and brain activity in regions that
showed UCR diminution [17]. Specifically, a negative relationship was observed between
UCS expectancy and brain activation within dmPFC, vmPFC, and left anterior insula
(Table 1). However, a positive relationship was observed between UCS expectancy and
the magnitude of the threat-related fMRI signal response within left IPL (Table 2). This
relationship between the unconditioned fMRI signal response and UCS expectancy was
not observed in our group level multiple linear regression analysis. The regression
analysis was conducted on a voxel-wise basis and accounted for additional measures of
interest (e.g. stimulus type). These findings suggest that UCS expectancy varied with the
mean percent signal change of the unconditioned fMRI signal response within the ROIs
as a whole. However, UCS expectancy did not explain unique variance within the
unconditioned fMRI signal response within these brain regions. These findings replicate
prior conditioning work that employed CS presentations that were easy to discriminate
[17]. In addition, our group level regression analysis revealed that the amplitude of the
unconditioned SCR explained unique variance in the fMRI data. Specifically, a positive
77
relationship was observed between threat-related SCR production and brain activation
within the dmPFC and left IPL. Taken together these findings support previous research
that suggests that regions of the PFC support associative learning processes [51,52], and
affect the peripheral emotional response [19–21].
The PFC appears to support top-down processes that are important for emotion
regulation. Prior work has demonstrated a negative relationship between anticipatory
PFC activity and the threat-related response within the amygdala [17,38]. Specifically, as
anticipatory PFC activity increases, the neural response to the threat decreases [17]. In
turn, the threat-elicited fMRI signal appears to mediate the learning-related changes
observed in the peripheral emotional response to a threat [11,15,17]. The data from the
current study are generally consistent with this prior work. In the present study, a
negative relationship was observed between anticipatory activation within the PFC and
threat-related activity within brain regions that showed UCR diminution (Table 1).
Specifically, as anticipatory activity within dlPFC, dmPFC, vlPFC, vmPFC, and the
amygdala increased the threat-related response within left dlPFC, dmPFC, vmPFC, and
anterior insula decreased (Table 4, Figure 4). Further, activity within dmPFC and left IPL
was positively correlated with threat-related SCR expression. These findings demonstrate
that anticipatory brain activity affects the response elicited by a threat, which in turn
affects the emotional response.
In summary, conditioned diminution of the unconditioned SCR and fMRI signal
response was observed during Pavlovian fear conditioning. UCR diminution was
observed within left dlPFC, dmPFC, vmPFC, and left anterior insula. Consistent with
prior work, many of the brain regions that showed learning-related changes in the
78
unconditioned fMRI signal response varied with UCS expectancy and unconditioned
SCR production [10,15,17]. Further, a negative relationship was observed between the
brain regions that showed UCR diminution and anticipatory PFC activiation (i.e. dlPFC,
dmPFC, vlPFC, and vmPFC). This finding supports prior work that suggests top-down
mechanisms provided by the PFC inhibit the emotional response evoked by a threat. The
findings from the current study provide a better understanding of the neural mechanisms
that support associative learning processes that modulate the threat-related response.
Acknowledgements: This research was supported by the University of Alabama at
Birmingham Faculty Development Grant Program and NIH R01 MH098348 (DCK).
79
References
1.
Kim JJ, Jung MW. Neural circuits and mechanisms involved in Pavlovian fear
conditioning: A critical review. Neuroscience & Biobehavioral Reviews.
2006;30(2):188–202.
2.
LeDoux J. The amygdala. Current biology: CB. 2007;17(20):868–74.
3.
Mineka S, Öhman A. Born to fear: non-associative vs associative factors in the
etiology of phobias. Behaviour research and therapy. 2002 Feb;40(2):173–84.
4.
Domjan M. Pavlovian conditioning: a functional perspective. Annual review of
psychology. 2005 Jan;56:179–206.
5.
Franchina JJ. Escape behavior and shock intensity: within subject versus betweengroups comparisons. Journal of comparative and physiological psychology.
1969;69(2):241–5.
6.
Helmstetter FJ, Bellgowan PS. Lesions of the amygdala block conditional
hypoalgesia on the tail flick test. Brain research. 1993;612(1-2):253–7.
7.
Kamin LJ. Traumatic avoidance learning: the effects of CS−US interval with a
trace-conditioning procedure. Journal of comparative and physiological
psychology. 1954;47(1):65–72.
8.
Baxter R. Diminution and recovery of the UCR in delayed and trace classical GSR
conditioning. Journal of experimental psychology. 1966;71(3):447–51.
9.
Kimmel E. Judgments of UCS intensity and diminution of the UCR in classical
GSR conditioning. Journal of experimental psychology. 1967;73:532–43.
10.
Knight DC, Waters NS, King MK, Bandettini PA. Learning-related diminution of
unconditioned SCR and fMRI signal responses. NeuroImage. Elsevier B.V.; 2010
Jan 1;49(1):843–8.
11.
Knight DC, Lewis EP, Wood KH. Conditioned diminution of the unconditioned
skin conductance response. Behavioral neuroscience. 2011 Aug;125(4):626–31.
12.
Marcos JL, Redondo J. Effects of conditioned stimulus presentation on diminution
of the unconditioned response in aversive classical conditioning. Biological
psychology. 1999 Jun;50(2):89–102.
13.
Lykken DT, Macindoe I, Tellegen A. Preception: Autonomic Response to Shock
as a Function of Predictability in Time and Locus. Psychophysiology.
1972;9(3):318–33.
80
14.
Peeke SC, Grings WW. Magnitude of UCR as a function of variability in the CS–
UCS relationship. Journal of experimental psychology. 1968;77:1–6.
15.
Dunsmoor JE, Bandettini PA, Knight DC. Neural correlates of unconditioned
response diminution during Pavlovian conditioning. NeuroImage. 2008 Apr
1;40(2):811–7.
16.
Rust J. Unconditioned response diminution in the skin resistance response. The
Journal of General Psychology. 1976;95:77–84.
17.
Wood KH, Ver Hoef LW, Knight DC. Neural mechanisms underlying the
conditioned diminution of the unconditioned fear response. NeuroImage. Elsevier
Inc.; 2012 Mar;60(1):787–99.
18.
Linnman C, Rougemont-Bücking A, Beucke JC, Zeffiro TA, Milad MR.
Unconditioned responses and functional fear networks in human classical
conditioning. Behavioural brain research. Elsevier B.V.; 2011 Aug 1;221(1):237–
45.
19.
Delgado MR, Nearing KI, Ledoux JE, Phelps EA. Neural circuitry underlying the
regulation of conditioned fear and its relation to extinction. Neuron. 2008 Sep
11;59(5):829–38.
20.
Ochsner KN, Bunge SA, Gross JJ, Gabrieli JD. Rethinking feelings: an fMRI
study of the cognitive regulation of emotion. Journal of cognitive neuroscience.
2002;14:1215–29.
21.
Milad MR, Vidal-Gonzalez I, Quirk GJ. Electrical stimulation of medial prefrontal
cortex reduces conditioned fear in a temporally specific manner. Behavioral
neuroscience. 2004 Apr;118(2):389–94.
22.
Knight DC, Nguyen HT, Bandettini PA. The role of the human amygdala in the
production of conditioned fear responses. NeuroImage. 2005 Jul 15;26(4):1193–
200.
23.
Basten U, Stelzel C, Fiebach CJ. Trait Anxiety Modulates the Neural Efficiency of
Inhibitory Control. Journal of cognitive neuroscience. 2011;(10):3132–45.
24.
Klumpp H, Ho SS, Taylor SF, Phan KL, Abelson JL, Liberzon I. Trait anxiety
modulates anterior cingulate activation to threat interference. Depression and
anxiety. 2011 Mar;28(3):194–201.
25.
Nitschke JB, Sarinopoulos I, Mackiewicz KL, Schaefer HS, Davidson RJ.
Functional neuroanatomy of aversion and its anticipation. NeuroImage. 2006 Jan
1;29(1):106–16.
81
26.
Etkin A, Wager TD. Functional Neuroimaging of Anxiety: A Meta-Analysis of
Emotional Processing in PTSD, Social Anxiety Disorder, and Specific Phobia.
American Journal of Psychiatry. 2007;164(October):1476–88.
27.
Milad MR, Rauch SL, Pitman RK, Quirk GJ. Fear extinction in rats: implications
for human brain imaging and anxiety disorders. Biological psychology. 2006
Jul;73(1):61–71.
28.
Milad MR, Pitman RK, Ellis CB, Gold AL, Shin LM, Lasko NB, et al.
Neurobiologial basis of failure to recall extinction memory in posttraumatic stress
disorder. Biological Psychiatry. 2009;66(12):1075–82.
29.
Rauch SL, Shin LM, Phelps EA. Neurocircuitry models of posttraumatic stress
disorder and extinction: human neuroimaging research--past, present, and future.
Biological psychiatry. 2006 Aug 15;60(4):376–82.
30.
Sehlmeyer C, Dannlowski U, Schöning S, Kugel H, Pyka M, Pfleiderer B, et al.
Neural correlates of trait anxiety in fear extinction. Psychological medicine. 2011
Apr;41(4):789–98.
31.
Somerville LH, Kim H, Johnstone T, Alexander AL, Whalen PJ. Human amygdala
responses during presentation of happy and neutral faces: Correlations with state
anxiety. Biological psychiatry. 2004 May 1;55(9):897–903.
32.
Cook EW, Davis TL, Hawk LW, Spence EL, Gautier CH. Fearfulness and startle
potentiation during aversive visual stimuli. Psychophysiology. 1992
Nov;29(6):633–45.
33.
Grillon C, Ameli R, Goddard A, Woods SW, Davis M. Baseline and fearpotentiated startle in panic disorder patients. Biological psychiatry. 1994;35:431–
9.
34.
Indovina I, Robbins TW, Núñez-Elizalde AO, Dunn BD, Bishop SJ. Fearconditioning mechanisms associated with trait vulnerability to anxiety in humans.
Neuron. 2011 Feb 10;69(3):563–71.
35.
Casey BJ, Thomas KM, Welsh TF, Badgaiyan RD, Eccard CH, Jennings JR, et al.
Dissociation of response conflict, attentional selection, and expectancy with
functional magnetic resonance imaging. Proceedings of the National Academy of
Sciences of the United States of America. 2000 Jul 18;97(15):8728–33.
36.
Fletcher PC, Anderson JM, Shanks DR, Honey R, Carpenter TA, Donovan T, et al.
Responses of human frontal cortex to surprising events are predicted by formal
associative learning theory. Nature neuroscience. 2001 Oct;4(10):1043–8.
82
37.
Dunsmoor JE, Bandettini PA, Knight DC. Impact of continuous versus intermittent
CS-UCS pairing on human brain activation during Pavlovian fear conditioning.
Behavioral neuroscience. 2007 Aug;121(4):635–42.
38.
Sarinopoulos I, Grupe DW, Mackiewicz KL, Herrington JD, Lor M, Steege EE, et
al. Uncertainty during anticipation modulates neural responses to aversion in
human insula and amygdala. Cerebral cortex. 2010 Apr;20(4):929–40.
39.
Cheng DT, Knight DC, Smith CN, Stein EA, Helmstetter FJ. Functional MRI of
human amygdala activity during Pavlovian fear conditioning: Stimulus processing
versus response expression. Behavioral Neuroscience. 2003;117(1):3–10.
40.
Cheng DT, Knight DC, Smith CN, Helmstetter FJ. Human amygdala activity
during the expression of fear responses. Behavioral neuroscience. 2006
Dec;120(6):1187–95.
41.
Cheng DT, Richards J, Helmstetter FJ. Activity in the human amygdala
corresponds to early, rather than late period autonomic responses to a signal for
shock. Learning & Memory. 2007;14(7):485–90.
42.
Spielberger C. State–Trait Anxiety Inventory for Adults. Redwood City, CA: Mind
Garden; 1983.
43.
Grös DF, Antony MM, Simms LJ, McCabe RE. Psychometric properties of the
State-Trait Inventory for Cognitive and Somatic Anxiety (STICSA): Comparison
to the State-Trait Anxiety Inventory (STAI). Psychological assessment. 2007
Dec;19(4):369–81.
44.
Knight DC, Wood KH. Investigating the neural mechanisms of aware and unaware
fear memory with FMRI. Journal of visualized experiments: JoVE. 2011
Jan;(56):1–6.
45.
Cox RW. AFNI: software for analysis and visualization of functional magnetic
resonance neuroimages. Computers and biomedical research, an international
journal. 1996 Jun;29(3):162–73.
46.
Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of the Human Brain. New
York: Thieme; 1988.
47.
Forman S, Cohen J, Fitzgerald M, Eddy W, Mintun M, Noll D. Improved
assessment of significant activation in functional magnetic resonance imaging
(fMRI): use of a cluster-size threshold. Magnetic resonance in medicine.
1995;33(5):636–47.
83
48.
Saad ZS, Chen G, Reynolds RC, Christidis PP, Hammett KR, Bellgowan PSF, et
al. Functional imaging analysis contest (FIAC) analysis according to AFNI and
SUMA. Human brain mapping. 2006 May;27(5):417–24.
49.
Loftus GR, Masson MEJ. Using confidence intervals in within-subject designs.
Psychonomic Bulletin & Review. 1994 Dec;1(4):476–90.
50.
Rescorla RA. Behavioral studies of Pavlovian conditioning. Annual review of
neuroscience. 1988;11:329–52.
51.
Orlebeke JF, Van Doornen LJ. Preception (UCR diminution) in normal and
neurotic subjects. Biological psychology. 1977 Mar;5(1):15–22.
52.
Knight DC, Smith CN, Cheng DT, Stein EA, Helmstetter FJ. Amygdala and
hippocampal activity during acquisition and extinction of human fear conditioning.
Cognitive, affective & behavioral neuroscience. 2004 Sep;4(3):317–25.
53.
Phillips ML, Drevets WC, Rauch SL, Lane R. Neurobiology of emotion perception
I: the neural basis of normal emotion perception. Biological Psychiatry. 2003
Sep;54(5):504–14.
84
Table 1. Regions showing conditioned diminution of the UCR.
Region
Vol (mm3)
Talairach coordinates
x
y
z
CS+UCS vs.
CS−UCS
CS−UCS vs.
UCS alone
CS+UCS vs.
UCS alone
t
t
t
Trait
r
SCR
r
UCS
Expectancy
r
Dorsolateral PFC
3739
−46.3
10.8
35.0
n.s.
−4.23
−3.81
0.04
−0.04
−0.17
Dorsomedial PFC
5670
−1.6
24.5
45.9
n.s.
−4.52
−4.18
0.04
−0.04
−0.36*
Ventromedial PFC
515
−0.9
52.7
−0.5
n.s.
−4.33
−3.96
−0.02
0.05
−0.50*
Left
Anterior Insula
Left
1046
−34.6 16.3
1.7
n.s.
−3.73
−3.99
0.06
0.18
−0.48*
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center of mass for areas of activation.
Significance criteria: ANOVA F[20] > 6.06, p < 0.05 (corrected); t[20] p < 0.05 (corrected). Significance criteria for two-tailed
correlations: * indicates p < 0.05 (corrected).
85
Table 2. Regions showing potentiation of the UCR.
Region
Vol (mm3)
Talairach coordinates
x
y
z
CS+UCS vs.
CS−UCS
CS−UCS vs.
UCS alone
CS+UCS vs.
UCS alone
t
t
t
Trait
r
SCR
r
UCS
Expectancy
r
Main effect of stimulus type
Dorsolateral PFC
1286
−7.3
−14.8
51.7
n.s.
3.93
3.57
0.15
0.14
0.30
Right
954
56.3
−29.3
28.4
n.s.
3.82
3.45
0.32
−0.06
0.28
Left
2618
−39.2
−40.9
51.0
n.s.
3.85
3.44
0.15
−0.07
0.43*
Left
1054
−60.9
−27.0
27.7
n.s.
4.20
3.39
0.28
−0.01
0.34
Right
3155
41.1
−7.7
10.8
n.s.
5.62
n.s.
0.16
−0.02
0.28
Left
2106
−41.9
−14.2
12.6
n.s.
4.29
n.s.
0.22
−0.12
0.31
−42.8
−34.4
38.7
n.s.
n.s.
4.11
0.14
−0.06
Left
Inf. Parietal Lobule
Posterior Insula
Stimulus x trial interaction
Inf. Parietal Lobule
Early test trials
577
0.33
Late test trials
n.s.
n.s.
n.s.
0.05
0.09
0.20
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center of mass for areas of activation.
Significance criteria: ANOVA F[20] > 6.06, p < 0.05 (corrected); t[20] p < 0.05 (corrected). Significance criteria for two-tailed
correlations: * indicates p < 0.05 (corrected).
86
Table 3. Regions showing change over time.
Talairach coordinates
Region
Dorsolateral PFC
Inf. Parietal Lobule
Hemisphere
3
Vol (mm )
x
y
z
Right
1072
23.7
44.0
32.0
Left
591
-45.8
9.0
29.5
Right
553
51.0
-41.9
24.7
Anterior Insula
Left
1258
-44.4
21.0
0.9
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center
of mass for areas of activation. Significance criteria: F[20] > 10.00; p < 0.05 (corrected).
87
Table 4. Regions showing a relationship between anticipatory and threat-related activity.
UCR DIMINUTION FUNCTIONAL ROIs
Talairach
Vol
coordinates
Region
Dorsolateral PFC
Left
3
(mm )
x
y
ANTICIPATORY BRAIN ACTIVATION
Vol
z
↔
3739
−46.3
10.8
35.0
3
Region
Ventrolateral PFC
(mm )
x
817
−37.4
29.9
−10.4
665
5.1
50.7
−9.0
536
32.8
42.6
15.4
Right
1997
35.2
21.3
−8.2
Right
1016
20.9
60.6
1.6
1483
−37.0
25.8
−9.1
601
1.4
51.2
−4.7
149
22.4
−1.7
−17.8
10761
−1.1
24.1
33.9
Right
3911
35.2
33.8
29.5
Right
2593
46.4
7.9
30.6
Right
663
34.8
−3.6
48.2
Left
2993
−33.4
26.8
34.8
Left
1513
−49.8
9.0
30.0
Left
555
−24.5
13.8
50.9
Left
Ventromedial PFC
Dorsomedial PFC
5670
−1.6
24.5
45.9
↔
Talairach coordinates
y
z
Dorsolateral PFC
Right
Ventrolateral PFC
Left
a
Ventromedial PFC
Amygdala
Rightb
Ventromedial PFC
515
−0.9
52.7
−0.5
↔
Dorsomedial PFCe,f
Dorsolateral PFC
Ventrolateral PFC
Right
1906
41.1
32.2
7.2
Left
1523
−43.3
30.6
3.1
5898
−3.0
−32.8
32.9
Left
983
−37.2
21.8
−9.7
Right
1994
36.4
22.3
−8.1
975
2.3
50.5
−3.6
Posterior cingulate
↔
Anterior Insula
Left
1046
−34.6
16.3
1.7
e
Ventrolateral PFC
c
Ventromedial PFC
Amygdala
Rightd
182
23.9
−1.2 −15.8
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for the center of mass for areas of
activation. Significance criteria: t[20] > 2.98, p < 0.05 (corrected). The UCR amplitude (from ROI on left side of
table) varied with the anticipatory response (i.e. the CR) within the dorsolateral, dorsomedial PFC, ventromedial
PFC, and posterior cingulate (right side of the table). Exploratory analysis of amygdala and hippocampal activation
small volume correction applied t[20] > 2.98, 112 mm3, p < 0.05 (corrected) (right side of the table). Letters (a-f)
correspond to images and graphical representation presented in Figure 4.
88
Acquisition Phase
Test Phase
CS+
CS+
UCS
UCS
CS-
CS-
UCS
UCS
0
10
20 30
Time (s)
40
50
No CS
UCS
0
10
20 30
Time (s)
40
50
Figure 1. Conditioned and unconditioned stimuli. The acquisition blocks consisted of
CS+ (8 trials), CS− (8 trials), and test trials (1 CS+UCS, 1 CS−UCS, and 1 UCS alone
trial). The acquisition phase consisted of four acquisition blocks, followed by the test
phase. The test phase consisted of 10 CS+UCS trials, 10 CS−UCS trials, and 10 UCS
alone trials. Stimuli were counterbalanced and presented in a pseudorandom order such
that no more than two trials of the same stimulus were consecutively presented.
89
UCS Expectancy
100
a
80
60
40
CS+UCS
CS-UCS
UCS alone
20
Unconditioned SCR
2.5
b
2.0
1.5
1.0
0.5
1-7
8-14
Trials
Figure 2. UCS expectancy and unconditioned SCR. a) Learning-related differences in
UCS expectancy. UCS expectancy on Early and Late test trials (1–7) was higher during
CS+UCS and CS−UCS trials vs. UCS alone trials. b) Learning-related changes in
unconditioned SCR expression were also observed. During Early test trials unconditioned
SCRs were diminished on CS+UCS and CS−UCS trials compared to UCS alone trials.
No differences were observed between CS+UCS and CS−UCS trials. During Late test
trials no differences were observed between CS+UCS, CS−UCS, or UCS alone trials.
90
0.5
Dorsolateral PFC
*
fMRI signal (% )
fMRI signal (% )
Dorsomedial PFC
*
0.4
0.3
0.2
0.1
CS+UCS
CS-UCS
*
0.4
*
0.3
0.2
0.1
0.0
UCS alone
CS+UCS
CS-UCS
UCS alone
z = 38
*
0.2
Anterior Insula
*
fMRI signal (% )
fMRI signal (% )
Ventromedial PFC
0.1
0.0
-0.1
-0.2
CS+UCS
CS-UCS
UCS alone
Right
Left
*
0.5
*
0.4
0.3
0.2
0.1
CS+UCS
CS-UCS
UCS alone
z = -2
Figure 3. UCR diminution within the fMRI signal response. Significant diminution of the
unconditioned fMRI signal response was observed within the prefrontal cortex (PFC) and
anterior insula during test trials. UCR amplitude within these brain regions was reduced
when the UCS followed the CS+ (i.e. CS+UCS trials) and CS− (i.e. CS−UCS trials)
compared to when the UCS was presented alone. There was no difference on CS+UCS
versus CS−UCS trials. Graphs reflect the mean amplitude (% signal change) of all voxels
within volumes of activation. Error bars reflect SEM after adjusting for between-subject
variance [49]. Asterisk indicates significant difference.
91
Relationship between Anticipatory
and Threat-related Activity
vmPFC and R. Amyg CR ↔ dmPFC UCR
r = −.60; p < 0.05
r = −.53; p < 0.05
a
b
x=2
y = -3
vmPFC and R. Amyg CR ↔ Ant. Insula UCR
r = −.66; p < 0.05
r = −.58; p < 0.05
c
d
x=2
y = -2
dmPFC and PCC CR ↔ vmPFC UCR
r = −.55; p < 0.05
r = −.57; p < 0.05
f
0.2
vmPFC UCR
e
x=2
0.0
-0.2
-0.4
CS+UCS
CS-UCS
-0.4
-0.2
0.0
0.2
0.4
dmPFC CR
Figure 4. Relationship between anticipatory and threat-related activity. Threat-related
activity, extracted from the ROI depicted in Table 1 and Figure 3, was included in a
regression analysis to investigate differences in the relationship between anticipatory
activity (i.e. CR) and threat-related activity (i.e. UCR) on CS+UCS and CS−UCS trials.
A negative relationship between anticipatory and threat-related activity (% signal change)
was observed in several areas of the PFC, cingulate, and amygdala (a-e). Correlation
values comparing the anticipatory and threat-related response within these brain areas are
presented above the brain images. The correlation value above image (e) represents
activation observed between PCC CR and vmPFC UCS. The correlation value for
dmPFC CR and vmPFC UCR is presented in graph (f). Talairach coordinates for the
depicted areas of activation are presented in Table 4 and labeled with letters (a-e)
corresponding to each image above.
92
93
CONTROLLABILITY AND PREDICTABILITY DIMINISH
THE NEURAL RESPONSE TO A THREAT
KIMBERLY H. WOOD, KENTON H. BOWEN, JOSHUA R. SHUMEN,
MURIAH D. WHEELOCK, LAWRENCE W. VER HOEF, AND DAVID C. KNIGHT
In preparation for Journal of Neuroscience
Format adapted for dissertation
Abstract
The ability to predict and control stressful events influences our emotional response to
future stressors. Prior animal research has demonstrated a diminished emotional response
to predictable and controllable stressors, whereas unpredictable and uncontrollable
stressors result in an enhanced emotional response. The present study was designed to
better understand the effect of predictability and controllability on the threat-related
emotional response. Two groups of healthy volunteers participated in a Pavlovian fear
conditioning study during functional magnetic resonance imaging (fMRI). Similar to
prior animal research, the groups consisted of yoked pairs where one group (Controllable
Condition; CC) was able to terminate the unconditioned stimulus (UCS), and the other
group (Uncontrollable Condition; UC) was not able to terminate the UCS. We also
assessed the influence of state anxiety and UCS expectancy on the modulation of threatrelated skin conductance response (SCR), startle eye-blink electromyography (EMG),
and fMRI signal response. The threat-related fMRI signal response was diminished on
predictable compared to unpredictable trials within the dorsolateral prefrontal cortex
(PFC), dorsomedial PFC, ventromedial PFC, ventrolateral PFC, and posterior cingulate
for both CC and UC groups. A predictability x controllability interaction was observed
within ventromedial PFC and left hippocampus. Specifically, the threat-related response
within these brain regions was diminished on predictable vs. unpredictable trials for the
CC group. The current findings suggest the ventromedial PFC plays a key role in
modulating the emotional response to a controllable stressor. Further, these data provide a
better understanding of the influence of predictability and controllability in the
modulation of the threat-related response.
94
Introduction
The ability to predict and control stressful events reduces the emotional response
to future stressors (Amat et al., 2005; Baratta et al., 2007; Maier, 1986; Maier & Watkins,
2010; Weinberg et al., 2010). Moreover, exposure to unpredictable and uncontrollable
stressful events is an important trigger in the development of anxiety-related disorders
(Chorpita & Barlow, 1998). A stressor’s predictability is influenced by the environmental
cues that signal it, whereas controllability is related to the ability to avoid, terminate, or
moderate a stressor (Foa et al., 1992). In general, the emotional response to stress is
diminished when stressors are predictable and controllable (Etkin, 2009; Maier et al.,
2006), whereas exposure to unpredictable and uncontrollable stress is linked to anxietylike behaviors (Weinberg et al., 2010).
The effect of predictability and controllability on brain-behavior relationships has
primarily been investigated using animal models. This prior work has often employed a
yoked Pavlovian fear conditioning paradigm. Typically in these studies, one group has
the ability to terminate the unconditioned stimulus (UCS), whereas the yoked group
cannot terminate the UCS (Maier et al., 2006). During Pavlovian conditioning a
conditioned stimulus (CS) is paired with an aversive UCS. Associative learning is
apparent when the CS generates a conditioned response (CR). The unconditioned
response (UCR) elicited by the UCS is typically considered a reflexive, response that
does not require associative learning. Prior animal research has shown that exposure to a
controllable UCS interferes with subsequent fear conditioning, whereas exposure to an
uncontrollable UCS enhances the conditioned fear-response (Baratta et al., 2007; Maier
et al., 2006). The ventromedial prefrontal cortex (vmPFC) is a key brain region that
95
modulates the conditioned fear response to controllable stressors (Baratta et al., 2007;
Maier et al., 2006).
Prior human neuroimaging work also suggests that the prefrontal cortex (PFC)
provides regulatory control over the amygdala during emotion-related tasks. Further,
insufficient top-down regulatory control by the vmPFC results in an exaggerated
amygdala response (Milad et al., 2006; Rauch et al., 2006). Although many prior
conditioning studies have investigated brain activity in anticipation of a threat, relatively
few studies have investigated the emotional response to the threat itself (Dunsmoor et al.,
2008; Knight et al., 2010; Wood et al., 2012). However, the limited work that has been
completed has demonstrated learning-related changes in UCR expression within the
fMRI signal response of the dorsolateral PFC (dlPFC), dorsomedial PFC (dmPFC),
vmPFC, and amygdala (Dunsmoor et al., 2008; Knight et al., 2010; Wood et al., 2012)
that parallel the emotional response (indexed via SCR) (Knight et al., 2010; Wood et al.,
2012). More specifically, UCR amplitude is diminished to predictable compared to
unpredictable UCS presentations (Baxter, 1966; Dunsmoor et al., 2008; Knight et al.,
2010, 2011; Lykken et al., 1972; Lykken & Tellegen, 1974; Peeke & Grings, 1968;
Wood et al., 2012). This phenomenon is generally referred to as conditioned UCR
diminution. Taken together this prior work suggests that predictability influences UCR
expression. However, there has been limited research on the effect of controllability on
UCR expression.
The current study investigated the role of controllability and predictability in
conditioned UCR diminution. To our knowledge, this is the first human neuroimaging
study to employ a yoked procedure, similar to prior animal studies (Baratta et al., 2008;
96
Maier & Watkins, 2010), to investigate the effect of controllability and predictability on
UCR expression. We hypothesized that controllability and predictability would diminish
the magnitude of the unconditioned neurophysiological response.
Materials and Methods
Experimental Design: Participants were exposed to a differential fear conditioning
procedure during fMRI in which the CS and UCS were presented through MRcompatible pneumatic headphones. The study consisted of yoked pairs of subjects, in
which one group received a controllable UCS (Controllable Condition; CC) and the
second group received an uncontrollable UCS (Uncontrollable Condition; UC). CC
participants had the ability to terminate the UCS, whereas UC participants could not
terminate the UCS.
Participants: A total of fifty-four (27 CC and 27 UC) healthy right-handed volunteers
participated in this study [28 female, 26 male; age = 23.39 ± 0.77 years (mean ± SEM);
range = 18-38 years]. Participants in the two groups were matched on gender, ethnicity,
age, and level of education (Table 1). There were no significant differences between the
two groups based on these factors. All subjects provided written informed consent in
compliance with the University of Alabama at Birmingham Institutional Review Board.
State-Trait Anxiety Inventory: Participants completed the State-Trait Anxiety Inventory
(STAI; Form Y) for Adults (Spielberger, 1983) after the conditioning session. The STAI
consists of a self-assessment measure of state and trait anxiety in terms of general
97
negative affect (Grös et al., 2007). Scores on the state scale reflect anxiety level at the
current moment, whereas trait anxiety scores reflect a relatively long-term predisposition
for anxiety (Spielberger, 1983).
Conditioned and unconditioned stimuli: Two tones (700 and 1300 Hz; 10 s duration; 20 s
ITI) served as the CSs and a loud (100 dB) white-noise served as the UCS (duration: 0.56.0 s in 0.5 s increments). The UCS coterminated with one tone (CS+UCS) and the
second tone was presented alone (CS−) during acquisition (two 960 s blocks). To assess
conditioned diminution of the UCR, the acquisition phase also included presentations of
the UCS alone. A total of 24 CS+UCS, 24 CS−, and 24 UCS alone trials were presented
during acquisition (Figure 1). The stimuli were counterbalanced and presented in a
pseudorandom order such that no more than two trials of the same stimulus were
consecutively presented.
UCS duration: CC participants were informed that the UCS would last between 0.5-6.0 s,
and that they had the ability to control the duration of the UCS. They were informed that
they could terminate the UCS by pressing a button on the joystick. In doing so, CC
participants determined the duration of the UCS for themselves as well as their matched
UC counterpart. UC participants were also informed that the UCS would last between
0.5-6.0 s. Given that UC participants did not have the ability to control the UCS, they
were instructed to make a button press when the UCS ended, to control for motor activity
associated with the button presses made by their match in the CC group.
UCS expectancy: UCS expectancy was used as a measure of conscious expectation of the
98
UCS. Presentation software (Neurobehavioral Systems, Inc.; Albany, CA) was used to
present a UCS expectancy rating scale on an IFIS-SA LCD (Invivo Corp.; Gainesville,
FL) video screen located above the subject's head and viewed through a mirror attached
to the RF coil. An MRI compatible joystick (Current Designs; Philadelphia, PA) was
used to monitor subjects’ expectancy of receiving the UCS. The joystick controlled a
rating bar which was presented throughout the conditioning session on the video screen.
Subjects were instructed to rate their UCS expectancy from moment to moment using a
continuous scale from 0 to 100 (0 = certain the UCS would not be presented, 50 =
uncertain whether the UCS would be presented, 100 = certain the UCS would be
presented) to reflect their current UCS expectancy. UCS expectancy was calculated as the
average response (1 s sample) at UCS onset. Additional details on this methodology have
been published previously (Knight & Wood, 2011).
Skin conductance:
An MRI compatible physiological monitoring system (Biopac
Systems; Goleta, CA) was used to collect SCR data. SCR was sampled (10 kHz) with a
pair of disposable radio-translucent dry electrodes (EL509, Biopac Systems; Goleta, CA).
Isotonic recording electrode gel (Gel101, Biopac Systems; Goleta, CA) was applied to
the electrodes which were then affixed to the thenar and hypothenar eminences of the left
palm. SCR data were processed using Biopac AcqKnowledge 4.1 software. A 1 Hz low
pass digital filter was applied and SCR data were resampled at 250 Hz. Unconditioned
SCRs were limited to those that occurred within 10 s following the UCS presentation.
Unconditioned SCRs smaller than 0.05 uSiemens were scored as 0. Data were then
square root transformed prior to statistical analyses.
99
Electromyography: The MRI compatible physiological monitoring system (Biopac
Systems; Goleta, CA) was also used to collect EMG data. EMG was sampled (10 kHz)
with a pair of disposable radio-translucent electrodes (1 cm diameter, Biopac Systems;
Goleta, CA) from the orbicularis oculi muscle below the left eye. The first electrode was
placed directly below the left pupil while the second was placed laterally to the first
electrode as per previous committee report guidelines (Blumenthal et al., 2005). EMG
data were processed using Biopac AcqKnowledge 4.1 software. Following guidelines for
digital filtering (Cook & Miller, 1992) and EMG denoising (Blumenthal et al., 2005) a
Fast Fourier Transform was used to assess and remove frequency domains where noise
occurred (Comb Band Stop filter at fMRI fundamental frequency ≈ 17.0 Hz, 60 Hz Notch
filter, 28-400 Hz Kaiser-Bessel Band Pass filter). The EMG signal was resampled at 1000
Hz then rectified and integrated (20 ms time constant) for scoring. Responses were scored
as the peak-valley difference with the valley occurring in the first 20 ms after the UCS
and the peak occurring within the 21-150 ms window following the UCS (Blumenthal et
al., 2005). Negative responses were scored as a zero.
Functional MRI: Structural and functional imaging was completed on a 3 Tesla Siemens
Allegra scanner. High-resolution anatomical images (MPRAGE) were obtained in the
sagittal plane using a T1 weighted series (TR=2300 ms, TE=3.9 ms, flip angle=12⁰,
FOV=25.6 cm, matrix=256 x 256, slice thickness=1 mm, 0.5 mm gap) to serve as an
anatomical reference. Blood oxygen level dependent fMRI of the entire brain was
conducted using a gradient-echo echoplanar pulse sequence in an oblique-axial
orientation (TR=2000 ms, TE=30 ms, flip angle=70º, FOV=24 cm, matrix=64 x 64, slice
100
thickness=4 mm, no gap) during each block of stimulus presentations. Functional image
processing was performed with the Analysis of Functional NeuroImages (AFNI) software
package (Cox, 1996). Echo-planar time series data were corrected for slice timing offset,
motion corrected, concatenated, reregistered to the fifth volume of the first imaging
block, and spatially blurred using a 4 mm full-width-at-half-maximum Gaussian filter.
Functional MRI data were analyzed at the individual subject level using the input
from all stimuli in a multiple linear regression using a gamma variate hemodynamic
response function. Regressors to account for brain activity not related to the UCR
included reference waveforms for the CS+ and CS−, joystick movement, button presses,
and head motion parameters. The regressors of interest for this study modeled the
unconditioned fMRI signal response to UCS presentations during the CS+UCS and UCS
alone. Percent signal change was used as an index of the magnitude of the unconditioned
fMRI signal response produced by the UCS. Functional maps reflecting percent signal
change were converted to the Talairach and Tournoux stereotaxic coordinate system for
group analyses (Talairach & Tournoux, 1988).
Based on prior work (Dunsmoor et al., 2008; Knight et al., 2010; Wood et al.,
2012), an anatomical mask was used to restrict group level analyses to the PFC, cingulate
cortex, IPL, insula, amygdala, and hippocampus to reduce the number of voxel-wise
comparisons. We conducted a repeated-measures ANOVA to test for a main effect of
predictability (CS+UCS vs. UCS alone) and controllability (CC vs. UC), as well as a
predictability x controllability interaction. A voxel-wise threshold of p < 0.05 (corrected)
was employed by using an uncorrected threshold of p < 0.005 and a cluster volume larger
than 563 mm3 (10 voxels of 3.75 x 3.75 x 4.00 mm dimension). Given our a priori
101
hypotheses and relatively small volume of the amygdala and hippocampus we used a
voxel-wise threshold of p < 0.005 and a cluster volume larger than 112 mm3 (2 voxels
3.75 x 3.75 x 4.00 mm dimension) to assess activity within these brain regions. These
threshold criteria were used to correct for multiple comparisons based on Monte Carlo
simulations that were used to reject smaller clusters of activation produced by chance
alone (Forman et al., 1995; Saad et al., 2006) and result in family-wise error corrected
significance of threshold p < 0.05. Follow-up t-test comparisons were conducted in SPSS
on the mean percent signal change activation passing the significance threshold (p < 0.05
corrected) for the ANOVA.
A secondary multiple linear regression analysis was completed to investigate the
relationship between our behavioral measures (i.e. state anxiety, UCS expectancy, SCR,
EMG) and the unconditioned fMRI signal response from brain regions that demonstrated
learning-related changes in the ANOVA (i.e. functional regions of interest; ROI). The
regression analysis evaluated these relationships on a voxel-wise basis. AlphaSim (Cox,
1996; Saad et al., 2006) was used to conduct Monte Carlo simulations limited to the
functional ROI from our repeated-measures ANOVA that demonstrated a main effect of
predictability, a main effect of controllability, or predictability x controllability
interaction (p < 0.05; corrected). Prior work has demonstrated a relationship between the
magnitude of the fMRI signal response within the amygdala, SCR, and EMG response
during Pavlovian fear conditioning (Cheng et al., 2003, 2006, 2007; Knight et al., 2005,
2010; van Well et al., 2012; Wood et al., 2012). Additionally, prior human fear
conditioning studies have demonstrated learning-related changes within the hippocampus
(Knight et al., 2004, van Well et al., 2012). Therefore an anatomical mask was employed
102
to include the amygdala and hippocampus in the group level regression analysis. A voxelwise threshold of p < 0.005 and cluster volume larger than 112 mm3 (2 voxels 3.75 x 3.75
x 4.00 mm dimension) was employed, resulting in a family-wise error corrected
significance threshold of p < 0.05.
Results
UCS duration: Given that CC and UC participants were yoked, there were no differences
in UCS duration between the groups. In addition, there were no differences in UCS
duration between the CS+UCS (mean = 3.51 ± 0.41; range = 0.5 – 6.0 s) and UCS alone
(mean = 3.62 ± 0.40; range = 0.5 – 6.0 s; t[26] = -1.38, n.s.) trials.
UCS expectancy: Repeated measures ANOVA revealed significant differences in UCS
expectancy. Results showed a main effect for predictability (F[1,52] = 32.17, p < 0.05),
but no main effect for controllability (F < 1.00) or predictability x controllability
interaction (F < 1.00). UCS expectancy was greater on CS+UCS (mean ± SEM: 75.55 ±
2.93) than on UCS alone (60.47 ± 3.56) trials for the CC group (t[26] = 3.22, p < 0.05).
The UC group also showed greater UCS expectancy on CS+UCS (76.95 ± 2.95) than
UCS alone (57.92 ± 2.70; t[26] = 5.04, p < 0.05) trials (Figure 2a). The aim of this study
was to investigate the effect of controllability and predictability on UCR expression.
Therefore, contrasts including the CS− were not conducted because a UCS was not
presented on CS− trials.
103
Skin conductance: Repeated measures ANOVA also revealed significant differences in
unconditioned SCR expression. There was a main effect for predictability (F[1,52] =
15.71, p < 0.05), but no main effect for controllability (F = 1.75) or a predictability x
controllability interaction (F < 1.00). T-test comparisons revealed a significantly
diminished unconditioned SCR for CS+UCS trials (0.55 ± 0.09) compared to UCS alone
trials (0.65 ± 0.12; t[26] = -2.23, p < 0.05) for CC participants. The same pattern was
observed for UC participants. Unconditioned SCRs were diminished on CS+UCS trials
(0.71 ± 0.09) compared to UCS alone trials (0.85 ± 0.09; t[26] = -3.44; p < 0.05) (Figure
2b).
Electromyography: Repeated measures ANOVA also revealed significant differences in
the startle-eyeblink response. There was a main effect for predictability (F[1,52] = 11.28,
p < 0.05), but no main effect for controllability (F = 1.11) or predictability x
controllability interaction (F = 2.80). T-test comparisons revealed a significantly
enhanced startle-eyeblink response on CS+UCS trials (289.65 ± 39.39) compared to UCS
alone trials (191.32 ± 27.69; t[26] = 2.97, p < 0.05) for CC participants. There was no
difference in EMG response for UC participants on CS+UCS trials (218.87 ± 23.55)
compared to UCS alone trials (185.92 ± 22.48; t[26] = 1.57, n.s.) (Figure 2c).
Functional MRI: Repeated measures ANOVA revealed significant differences in the
magnitude of the unconditioned fMRI signal response within several brain regions (Table
2, Figures 3-4). In each of these regions the unconditioned fMRI signal response
demonstrated a main effect for predictability (F[53] > 8.61; p < 0.05 corrected). There
104
was not a main effect for controllability. A predictability x controllability interaction was
observed within vmPFC and left hippocampus (Table 2; Figure 3). T-test comparisons
were completed on the mean fMRI signal from each volume of activation that passed the
significance threshold (p < 0.05 corrected) for the main effect of stimulus type. All
regions showed a diminished UCR on CS+UCS vs. UCS alone trials. Post hoc contrasts
on the predictability x controllability interaction were conducted on the mean fMRI
signal from the vmPFC and left hippocampus. These t-test comparisons revealed a
diminished UCR on CS+UCS vs. UCS alone trials for the CC group within the vmPFC
(t[26] = -3.93; p < 0.05 corrected) and left hippocampus (t[26] = -4.38; p < 0.05
corrected). The fMRI signal response on CS+UCS vs. UCS alone trials for the UC group
was not significantly different once corrected for multiple comparisons (vmPFC t[26] =
2.03; p = n.s.; left hippocampus t[26] = 2.70, p = n.s.). There were no group differences
revealed in the post hoc contrasts. Specifically, no differences were observed between the
CC group compared to the UC group on CS+UCS or UCS alone trials. The fMRI signal
response on CS+UCS trials for the CC group vs. UC group was not significantly different
once corrected for multiple comparisons (vmPFC t[26] = -2.35; p = n.s.; left
hippocampus t[26] = -2.32; p = n.s.). Similarly, there was no difference in the fMRI
signal response on UCS alone trials for the CC group vs. UC group (vmPFC t[26] = 1.66,
p = n.s.; left hippocampus t[26] = 1.24, p = n.s.).
A voxel-wise multiple linear regression analysis was conducted separately on the
averaged [predictable (i.e. CS+UCS) and unpredictable (i.e. UCS alone)] threat-related
fMRI signal response, as well as the difference (predictable – unpredictable) in activation
between these trial types. The linear regression model accounted for state anxiety, UCS
105
expectancy, SCR production, and EMG response. These analyses were restricted to the
functional ROI identified from the ANOVA with one exception. The subcortical
structures of interest (i.e. amygdala and hippocampus) were also included based on prior
human fear conditioning studies that have demonstrated learning-related changes within
these brain areas (Bach et al., 2011; Knight et al., 2005, 2004; van Well et al., 2012). The
regression analysis on the averaged threat-related fMRI signal response demonstrated that
state anxiety level explained unique variability in the activation observed within dmPFC,
vmPFC, and posterior cingulate (PCC) (Table 3). There were no brain regions that
showed a relationship with UCS expectancy, SCR, or EMG that met our significance
criteria. No other significant differences were observed.
Discussion
The predictability and controllability of stressors impacts our emotional response
to stressful events. Typically, the emotional response to a stressor is diminished when
events are predictable and controllable (Etkin, 2009; Maier et al., 2006). However,
exposure to unpredictable and uncontrollable stress results in an enhanced stress response
and is linked to anxiety-like behaviors (Weinberg et al., 2010). The present study
employed a unique Pavlovian fear conditioning procedure to investigate the effect of
predictability and controllability on the emotional response to a threat. During Pavlovian
conditioning, participants are typically exposed to an uncontrollable UCS. However, this
study consisted of two groups, a CC group that could terminate the UCS and a traditional
UC group that could not terminate the UCS. Another novel quality of the current study
was that participants were yoked, similar to paradigms used in animal research (Baratta et
106
al., 2008; Maier & Watkins, 2010). Additionally, the current procedure included
predictable and unpredictable presentations of the UCS to investigate conditioned UCR
diminution. We also assessed state anxiety, UCS expectancy, unconditioned SCR, startle
eye-blink EMG, and the fMRI signal response during Pavlovian conditioned UCR
diminution to better understand the processes that modulate the threat-related response.
In the current study, learning-related changes were observed in our behavioral
data. Both CC and UC groups demonstrated high UCS expectancy ratings for predictable
compared to unpredictable trials (Figure 2a). These findings demonstrate that participants
learned that the CS+ predicted the UCS and remained uncertain of the timing of UCS
alone presentations. As anticipated, there were no group differences in UCS expectancy.
Similar to prior work, conditioned diminution of the unconditioned SCR was also
observed in the present study. The amplitude of unconditioned SCR was diminished on
predictable compared to unpredictable trials for both the CC and UC groups. These
findings are consistent with prior behavioral studies that have demonstrated a reduction in
UCR magnitude to predictable compared to unpredictable UCS presentations (Baxter,
1966; Kimmel, 1967; Knight et al., 2011; Lykken et al., 1972; Marcos & Redondo,
1999). Learning-related changes within the EMG response were also observed in the
current study. Specifically, the EMG response was enhanced for the CC group on
predictable compared to unpredictable trials. This finding is similar to prior work that
demonstrated an enhanced startle response during a CS compared to baseline startle
(Grillon & Davis, 1997). There were no differences in EMG response for predictable vs.
unpredictable trials for the UC group. Additionally, no group differences were observed
in threat-related SCR expression or EMG response. Group differences in threat-elicited
107
SCR and EMG may have been masked by the degree of between-subject variability often
observed within these measures. Taken together, these findings are similar to previous
human fear conditioning studies and suggest that predictability influences UCR
expression (Baxter, 1966; Kimmel, 1967; Knight et al., 2011; Lykken et al., 1972;
Marcos & Redondo, 1999; Wood et al., 2012).
Similar to prior work, the threat-related fMRI signal response paralleled
unconditioned SCR expression. Specifically, a diminished fMRI signal response within
dlPFC, dmPFC, vmPFC, vlPFC, and PCC was observed to predictable compared to
unpredictable presentations of the UCS. These findings replicate prior research designed
to investigate learning-related changes within the threat-related fMRI signal response
(Wood et al., 2012). This pattern was not observed between the EMG data and the threatrelated fMRI signal response. Although there were no group differences in the startleeyeblink response, the CC group did demonstrate an enhanced startle-eyeblink response
to predictable (i.e. CS+UCS) compared to unpredictable (i.e. UCS alone) trials. The UC
group did not show learning-related changes in the EMG response. Specifically, no
difference in startle-eyeblink response was observed for predictable vs. unpredictable
UCS presentations for the UC group. Somewhat consistent with this weak behavioral
effect, controllability influenced the threat-related fMRI signal response within the
vmPFC and left hippocampus, such that the response within these brain regions was
diminished when the threat was both predictable and controllable. However, post hoc
analyses of the predictability x controllability interaction within the vmPFC and left
hippocampus did not reveal significant group differences. Specifically, there was no
difference in the amplitude of the fMRI signal response for the CC group compared to the
108
UC group on predictable or unpredictable trials. Since there was not a predictability x
controllability interaction observed in our behavioral data (i.e. UCS expectancy, SCR, or
EMG), the predictability x controllability interaction observed within the vmPFC and left
hippocampus is difficult to interpret. Given our focus on brain-behavior relationships, the
interaction effect observed in the imaging data would be more compelling if a similar
pattern was observed in the behavioral data.
Prior animal studies have shown that the vmPFC plays an important role in
modulating conditioned fear responses to controllable stressors (Baratta et al., 2007,
2008; Maier et al., 2006). Additionally, recent human neuroimaging research has
demonstrated controllability enhances anticipatory fear activity within the vmPFC (Kerr
et al., 2012). However, to our knowledge this is the first study to demonstrate that
controllability affects the vmPFC response to the threat itself. Further, prior work has
shown that the vmPFC is an important region that provides regulatory control over the
amygdala during emotion-related tasks (Milad et al., 2006; Rauch et al., 2006). Perhaps,
the exaggerated EMG response among the CC group is partly due to the diminished
threat-related fMRI signal response observed within the vmPFC. For example, prior
work suggests that regions of the PFC modulate subcortical brain areas (e.g. the
amygdala) that control the peripheral expression of emotion (Etkin & Wager, 2007; Kim
& Jung, 2006; Milad et al., 2004; Phillips et al., 2003). Therefore, diminished threatrelated activation within the vmPFC may contribute to the enhanced EMG response
observed on predictable trials for the CC group.
Prior work also suggests that anxiety level influences top-down mechanisms that
support regulatory control processes (Basten et al., 2011; Delgado et al., 2008; Klumpp et
109
al., 2011; Nitschke et al., 2006; Ochsner et al., 2002; Sehlmeyer et al., 2011). For
example, prior work suggests that anxiety level influences the magnitude of anticipatory
brain activation within the vmPFC (Indovina et al., 2011). More specifically, high anxiety
is associated with diminished vmPFC activity and an enhanced emotional response
(indexed via SCR) (Indovina et al., 2011). We have previously demonstrated anxiety
level varied with the threat-related fMRI signal response within the dlPFC, dmPFC, PCC,
and IPL (Wood et al., 2012). In this prior work, as anxiety level increased the threatrelated fMRI signal response within these brain regions also increased (Wood et al.,
2012). A positive relationship was observed between state anxiety and the threat-related
fMRI signal response within dmPFC, vmPFC, and PCC in the current study. Taken
together, these findings suggest that healthy emotion regulation relies upon the PFC
(Basten et al., 2011; Delgado et al., 2008; Nitschke et al., 2006; Ochsner et al., 2002).
In summary, learning-related changes were observed within the threat-related
neurophysiological response during Pavlovian fear conditioning. Conditioned UCR
diminution was observed within dlPFC, dmPFC, vmPFC, vlPFC, and PCC. Threatelicited SCR expression paralleled activation within the brain regions that showed UCR
diminution. However, the opposite pattern was observed for the EMG response in
relation to brain activity. Similar to our prior work, the predictability of a threat affected
the threat-related neurophysiological response that was elicited. Further, the
controllability of a threat also influenced the fMRI signal response within the vmPFC and
left hippocampus. Specifically, the magnitude of the threat-related response within these
brain regions was diminished when the threat was both controllable and predictable. This
finding suggests that the controllability and predictability of a threat impacts the
110
magnitude of both cortical and subcortical brain activation in response to a threat. To our
knowledge, this is the first human study to demonstrate the impact of predictability and
controllability on the threat-elicited neural response. However, a behavioral component
showing a similar pattern is lacking in the current results. Our inability to detect a
significant effect may be due to the degree of between-subject variability in the
psychophysiological data. Future studies, should examine the influence of predictability
and controllability on the threat-elicited emotional response using a within-subjects
design. Finally, the data suggest that the vmPFC and hippocampus play a key role in
modulating the fear response to a controllable stressor (Baratta et al., 2008; Kerr et al.,
2012; Maier et al., 2006). Taken together, the findings from the current study provide
evidence that both predictability and controllability of a stressor influences the
neurophysiological response to a threat.
Acknowledgements: This research was supported by the University of Alabama at
Birmingham Faculty Development Grant Program and NIH R01 MH098348 (DCK).
111
References
Amat, J., Baratta, M., Paul, E., Bland, S., Watkins, L., & Maier, S. (2005). Medial
prefrontal cortex determines how stressor controllability affects behavior and dorsal
raphe nucleus. Nature neuroscience, 8(3), 365–71. doi:10.1038/nn1399
Bach, D. R., Weiskopf, N., & Dolan, R. J. (2011). A stable sparse fear memory trace in
human amygdala. The Journal of neuroscience: the official journal of the Society for
Neuroscience, 31(25), 9383–9. doi:10.1523/JNEUROSCI.1524-11.2011
Baratta, M. V, Christianson, J. P., Gomez, D. M., Zarza, C. M., Amat, J., Masini, C. V,
Watkins, L. R., et al. (2007). Controllable versus uncontrollable stressors bidirectionally modulate conditioned but not innate fear. Neuroscience, 146(4), 1495–
503. doi:10.1016/j.neuroscience.2007.03.042
Baratta, M. V, Lucero, T. R., Amat, J., Watkins, L. R., & Maier, S. F. (2008). Role of the
ventral medial prefrontal cortex in mediating behavioral control-induced reduction
of later conditioned fear. Learning & memory (Cold Spring Harbor, N.Y.), 15(2),
84–7. doi:10.1101/lm.800308
Basten, U., Stelzel, C., & Fiebach, C. J. (2011). Trait Anxiety Modulates the Neural
Efficiency of Inhibitory Control. Journal of cognitive neuroscience, (10), 3132–
3145.
Baxter, R. (1966). Diminution and recovery of the UCR in delayed and trace classical
GSR conditioning. Journal of experimental psychology, 71(3), 447–451.
Blumenthal, T. D., Cuthbert, B. N., Filion, D. L., Hackley, S., Lipp, O. V, & Van Boxtel,
A. (2005). Committee report: Guidelines for human startle eyeblink
electromyographic studies. Psychophysiology, 42(1), 1–15. doi:10.1111/j.14698986.2005.00271.x
Cheng, D. T., Knight, D. C., Smith, C. N., & Helmstetter, F. J. (2006). Human amygdala
activity during the expression of fear responses. Behavioral neuroscience, 120(6),
1187–95. doi:10.1037/0735-7044.120.5.1187
Cheng, D. T., Knight, D. C., Smith, C. N., Stein, E. A., & Helmstetter, F. J. (2003).
Functional MRI of human amygdala activity during Pavlovian fear conditioning:
Stimulus processing versus response expression. Behavioral Neuroscience, 117(1),
3–10. doi:10.1037/0735-7044.117.1.3
Cheng, D. T., Richards, J., & Helmstetter, F. J. (2007). Activity in the human amygdala
corresponds to early, rather than late period autonomic responses to a signal for
shock. Learning & Memory, 14(7), 485–490. doi:10.1101/lm.632007.sion
112
Chorpita, B. F., & Barlow, D. H. (1998). The development of anxiety: the role of control
in the early environment. Psychological bulletin, 124(1), 3–21.
Cook, E. W. 3rd, & Miller, A. M. (1992). Methodology Digital Filtering: Background
and Tutorial for Psychophysiologists. Psychophysiology, 29(3), 350–367.
Cox, R. W. (1996). AFNI: software for analysis and visualization of functional magnetic
resonance neuroimages. Computers and biomedical research, an international
journal, 29(3), 162–73.
Delgado, M. R., Nearing, K. I., Ledoux, J. E., & Phelps, E. A. (2008). Neural circuitry
underlying the regulation of conditioned fear and its relation to extinction. Neuron,
59(5), 829–38. doi:10.1016/j.neuron.2008.06.029
Dunsmoor, J. E., Bandettini, P. A., & Knight, D. C. (2008). Neural correlates of
unconditioned response diminution during Pavlovian conditioning. NeuroImage,
40(2), 811–7. doi:10.1016/j.neuroimage.2007.11.042
Etkin, A. (2009). Functional Neuroanatomy of Anxiety: A Neural Circuit Perspective.
Behavioral neurobiology of Anxiety and its treatment, Current topics in Behavioral
Neurosciences 2 (pp. 251–277). doi:10.1007/7854
Etkin, A., & Wager, T. D. (2007). Functional Neuroimaging of Anxiety: A MetaAnalysis of Emotional Processing in PTSD, Social Anxiety Disorder, and Specific
Phobia. American Journal of Psychiatry, 164(October), 1476–1488.
Foa, E. B., Zinbarg, R., & Rothbaum, B. O. (1992). Uncontrollability and
unpredictability in post-traumatic stress disorder: an animal model. Psychological
bulletin, 112(2), 218–38.
Forman, S., Cohen, J., Fitzgerald, M., Eddy, W., Mintun, M., & Noll, D. (1995).
Improved assessment of significant activation in functional magnetic resonance
imaging (fMRI): use of a cluster-size threshold. Magnetic resonance in medicine,
33(5), 636–647.
Grillon, C., & Davis, M. (1997). Fear-potentiated startle conditioning in humans: Explicit
and contextual cue conditioning following paired versus unpaired training.
Psychophysiology, (34), 451–458.
Grös, D. F., Antony, M. M., Simms, L. J., & McCabe, R. E. (2007). Psychometric
properties of the State-Trait Inventory for Cognitive and Somatic Anxiety
(STICSA): Comparison to the State-Trait Anxiety Inventory (STAI). Psychological
assessment, 19(4), 369–81. doi:10.1037/1040-3590.19.4.369
113
Indovina, I., Robbins, T. W., Núñez-Elizalde, A. O., Dunn, B. D., & Bishop, S. J. (2011).
Fear-conditioning mechanisms associated with trait vulnerability to anxiety in
humans. Neuron, 69(3), 563–71. doi:10.1016/j.neuron.2010.12.034
Kerr, D. L., McLaren, D. G., Mathy, R. M., & Nitschke, J. B. (2012). Controllability
Modulates the Anticipatory Response in the Human Ventromedial Prefrontal Cortex.
Frontiers in Psychology, 3(December), 1–11. doi:10.3389/fpsyg.2012.00557
Kim, J. J., & Jung, M. W. (2006). Neural circuits and mechanisms involved in Pavlovian
fear conditioning: A critical review. Neuroscience & Biobehavioral Reviews, 30(2),
188–202.
Kimmel, E. (1967). Judgments of UCS intensity and diminution of the UCR in classical
GSR conditioning. Journal of experimental psychology, 73, 532–543.
Klumpp, H., Ho, S. S., Taylor, S. F., Phan, K. L., Abelson, J. L., & Liberzon, I. (2011).
Trait anxiety modulates anterior cingulate activation to threat interference.
Depression and anxiety, 28(3), 194–201. doi:10.1002/da.20802
Knight, D. C., Lewis, E. P., & Wood, K. H. (2011). Conditioned diminution of the
unconditioned skin conductance response. Behavioral neuroscience, 125(4), 626–
31. doi:10.1037/a0024324
Knight, D. C., Nguyen, H. T., & Bandettini, P. A. (2005). The role of the human
amygdala in the production of conditioned fear responses. NeuroImage, 26(4),
1193–200. doi:10.1016/j.neuroimage.2005.03.020
Knight, D. C., Smith, C. N., Cheng, D. T., Stein, E. A., & Helmstetter, F. J. (2004).
Amygdala and hippocampal activity during acquisition and extinction of human fear
conditioning. Cognitive, affective & behavioral neuroscience, 4(3), 317–25.
Knight, D. C., Waters, N. S., King, M. K., & Bandettini, P. A. (2010). Learning-related
diminution of unconditioned SCR and fMRI signal responses. NeuroImage, 49(1),
843–8. doi:10.1016/j.neuroimage.2009.07.012
Knight, D. C., & Wood, K. H. (2011). Investigating the neural mechanisms of aware and
unaware fear memory with FMRI. Journal of visualized experiments: JoVE, (56), 1–
6. doi:10.3791/3083
Lykken, D. T., Macindoe, I., & Tellegen, A. (1972). Preception: Autonomic Response to
Shock as a Function of Predictability in Time and Locus. Psychophysiology, 9(3),
318–333.
Lykken, D. T., & Tellegen, A. (1974). On the validity of the preception hypothesis.
Psychophysiology, 11(2), 125–32.
114
Maier, S. F. (1986). Stressor controllability and stress-induced analgesia. Annals of the
New York Academy of Sciences, 467, 55–72.
Maier, S. F., Amat, J., Baratta, M. V, Paul, E., & Watkins, L. R. (2006). Behavioral
control, the medial prefrontal cortex, and resilience. Dialogues in Clinical
Neuroscience, 8(4), 397–407.
Maier, S. F., & Watkins, L. R. (2010). Role of the medial prefrontal cortex in coping and
resilience. Brain research, 1355, 52–60. doi:10.1016/j.brainres.2010.08.039.
Marcos, J. L., & Redondo, J. (1999). Effects of conditioned stimulus presentation on
diminution of the unconditioned response in aversive classical conditioning.
Biological psychology, 50(2), 89–102.
Milad, M. R., Rauch, S. L., Pitman, R. K., & Quirk, G. J. (2006). Fear extinction in rats:
implications for human brain imaging and anxiety disorders. Biological psychology,
73(1), 61–71. doi:10.1016/j.biopsycho.2006.01.008
Milad, M. R., Vidal-Gonzalez, I., & Quirk, G. J. (2004). Electrical stimulation of medial
prefrontal cortex reduces conditioned fear in a temporally specific manner.
Behavioral neuroscience, 118(2), 389–94. doi:10.1037/0735-7044.118.2.389
Nitschke, J. B., Sarinopoulos, I., Mackiewicz, K. L., Schaefer, H. S., & Davidson, R. J.
(2006). Functional neuroanatomy of aversion and its anticipation. NeuroImage,
29(1), 106–16. doi:10.1016/j.neuroimage.2005.06.068
Ochsner, K. N., Bunge, S. A., Gross, J. J., & Gabrieli, J. D. (2002). Rethinking feelings:
an fMRI study of the cognitive regulation of emotion. Journal of cognitive
neuroscience, 14, 1215–1229.
Peeke, S. C., & Grings, W. W. (1968). Magnitude of UCR as a function of variability in
the CS–UCS relationship. Journal of experimental psychology, 77, 1–6.
Phillips, M. L., Drevets, W. C., Rauch, S. L., & Lane, R. (2003). Neurobiology of
emotion perception I: the neural basis of normal emotion perception. Biological
Psychiatry, 54(5), 504–514. doi:10.1016/S0006-3223(03)00168-9
Rauch, S. L., Shin, L. M., & Phelps, E. A. (2006). Neurocircuitry models of
posttraumatic stress disorder and extinction: human neuroimaging research--past,
present, and future. Biological psychiatry, 60(4), 376–82.
doi:10.1016/j.biopsych.2006.06.004
Saad, Z. S., Chen, G., Reynolds, R. C., Christidis, P. P., Hammett, K. R., Bellgowan, P.
S. F., & Cox, R. W. (2006). Functional imaging analysis contest (FIAC) analysis
according to AFNI and SUMA. Human brain mapping, 27(5), 417–24.
doi:10.1002/hbm.20247
115
Sehlmeyer, C., Dannlowski, U., Schöning, S., Kugel, H., Pyka, M., Pfleiderer, B.,
Zwitserlood, P., et al. (2011). Neural correlates of trait anxiety in fear extinction.
Psychological medicine, 41(4), 789–98. doi:10.1017/S0033291710001248
Spielberger, C. (1983). State–Trait Anxiety Inventory for Adults. Redwood City, CA:
Mind Garden.
Talairach, J., & Tournoux, P. (1988). Co-planar Stereotaxic Atlas of the Human Brain.
New York: Thieme.
Van Well, S., Visser, R. M., Scholte, H. S., & Kindt, M. (2012). Neural substrates of
individual differences in human fear learning: Evidence from concurrent fMRI, fearpotentiated startle, and US-expectancy data. Cognitive, affective & behavioral
neuroscience. doi:10.3758/s13415-012-0089-7
Weinberg, M. S., Grissom, N., Paul, E., Bhatnagar, S., Maier, S. F., & Spencer, R. L.
(2010). Inescapable but not escapable stress leads to increased struggling behavior
and basolateral amygdala c-fos gene expression in response to subsequent novel
stress challenge. Neuroscience, 170(1), 138–148.
doi:10.1016/j.neuroscience.2010.06.052.
Wood, K. H., Ver Hoef, L. W., & Knight, D. C. (2012). Neural mechanisms underlying
the conditioned diminution of the unconditioned fear response. NeuroImage, 60(1),
787–99. doi:10.1016/j.neuroimage.2011.12.048
116
Table 1. Demographics and group characteristics.
Controllable
Uncontrollable
Measures
Condition
Condition
Male/Female
13/14
13/14
Age
23.37 ± 1.10
23.41 ± 1.11
range
18-38
18-38
Education (years)
14.89 ± 0.49
15.26 ± 0.59
range
12-22
12-21
State Anxiety
33.30 ± 1.47
35.30 ± 1.49
Trait Anxiety
36.93 ± 1.71
38.63 ± 1.23
117
t
p
−0.03
n.s.
−0.68
n.s.
−1.05
−0.81
n.s.
n.s.
Table 2. Regions showing conditioned diminution of the UCR.
Region
Vol (mm3)
Main Effect of Predictability
Dorsolateral PFC
Right
747
Left
4034
Left
834
Dorsomedial PFC
13196
Talairach coordinates
x
y
z
CS+UCS vs.
UCS alone
t
31.9
−32.3
−26.6
3.1
0.2
0.0
45.7
13.2
57.3
53.1
31.8
40.1
−4.03
−4.58
−4.29
−4.75
Ventromedial PFC
Right
Left
2245
975
7.6
−3.2
38.8
36.5
7.0
6.8
−4.45
−4.12
Ventrolateral PFC
Left
705
−28.2
16.2
−15.8
−4.33
Posterior Cingulate
5902
2.0
−33.3
29.8
−4.67
Predictability x Controllability Interaction
Ventromedial PFC
837
2.0
39.6 −12.7
CC Group
−3.93
UC Group
n.s.
Hippocampus
586
−33.1
−23.4 −11.5
CC Group
−4.39
UC Group
n.s.
Note. Location, volumes, and coordinates from Talairach and Tournoux (1988) for
the center of mass for areas of activation. Significance criteria: ANOVA F[53] >
8.61, p < 0.05 (corrected); t[53] p < 0.05 (corrected).
118
Table 3. Regional activity varying with state anxiety.
Talairach coordinates
Region
Hemisphere Vol (mm3)
x
y
z
Dorsomedial PFC
Left
306
−5.3
29.7 36.9
Ventromedial PFC
Bilateral
1211
5.0
38.8
7.1
Posterior Cingulate
Bilateral
2256
−0.6 −33.9 27.9
Note. Location, volumes, and coordinates from Talairach and Tournoux
(1988) for the center of mass for areas of activation.
119
Acquisition Phase
CS+
UCS
CSUCS
No CS
UCS
0
10 20 30 40
Time (Seconds)
50
Figure 1. Acquisition Phase. During the acquisition phase, participants received a total
of 24 presentations of each stimulus (24 CS+UCS, 24 CS−, and 24 UCS alone). The
CS+ coterminated with the UCS, the CS− was presented alone, and UCS alone trials
were not preceded by a CS.
120
UCS Expectancy
100
a
CC
UC
80
60
40
20
µSiemens
1.0
b
0.8
0.6
0.4
0.2
350
c
µVolts
300
250
200
150
Predictable
Unpredictable
Figure 2. UCS expectancy, unconditioned SCR, and EMG response. a) Both the CC and
UC groups demonstrated learning-related differences in UCS expectancy. UCS
expectancy was higher to the predictable (i.e. CS+UCS) compared to unpredictable (i.e.
UCS alone) trials. b) Learning related changes in unconditioned SCR expression were
also observed for both the CC and UC groups. Unconditioned SCRs were diminished on
predictable vs. unpredictable trials. c) The CC group demonstrated learning-related
changes within the EMG response. The EMG response was enhanced on predictable
trials compared to unpredictable trials for the CC group. There were no differences in
EMG response for the UC group. No group differences were observed in UCS
expectancy, unconditioned SCR, or EMG.
121
x=6
0.3
0.2
Ventromedial PFC
0.1
Predictable
Unpredictable
0.20
*
fMRI signal (% )
0.25
fMRI signal (% )
fMRI signal (% )
0.4
Dorsomedial PFC
*
Posterior Cingulate
*
0.20
0.15
0.10
Predictable
0.15
Unpredictable
0.10
0.05
Predictable
Unpredictable
Figure 3. Conditioned UCR diminution within the fMRI signal response. Significant
diminution of the unconditioned fMRI signal response was observed within several brain
regions (see Table 2) including prefrontal cortex (PFC) and Posterior Cingulate (PCC).
No differences were observed between the CC and UC groups. UCR amplitude within
each of these brain areas was reduced on predictable (i.e. CS+UCS) compared to
unpredictable (i.e. UCS alone) trials. Graphs reflect the mean amplitude (% signal
change) of all voxels within volumes of activation. Asterisk indicates significant
difference.
122
Ventromedial PFC
fMRI signal (% )
0.10
0.05
0.00
-0.05
-0.10
Predictable
Unpredictable
x=3
Hippocampus
fMRI signal (% )
0.10
Right
0.05
0.00
-0.05
CC
UC
-0.10
Left
Predictable
y = -21
Unpredictable
Figure 4. Regions showing predictability x controllability interaction. The unconditioned
fMRI signal response within ventromedial PFC and left hippocampus was diminished on
predictable trials compared to unpredictable trials for the CC group, but not the UC
group. Graphs reflect the mean amplitude (% signal change) of all voxels within the
volume of activation.
123
SUMMARY
Typically, the response to a warning cue in anticipation of a threat is the primary
focus of Pavlovian conditioning studies. Although this anticipatory response is often used
as an index of associative learning, there are also associative learning-related changes in
the unconditioned response (UCR) produced by the threat itself. From a functional
perspective, it is important to understand learning-related changes in these innate UCRs
to naturally occurring threats due to their biological relevance for survival (Domjan,
2005). This project employed Pavlovian fear conditioning to investigate the influence of
associative learning, expectation, predictability, and controllability in the modulation of
the neurophysiological response to a threat.
In this project, we observed conditioned UCR diminution, using fMRI in
conjunction with behavioral measures (e.g. SCR expression and UCS expectancy ratings)
during the conditioning procedure, such that responses were diminished on predictable
compared to unpredictable presentations of the UCS. More specifically, we observed
conditioned diminution of the UCR within many of the same brain regions (e.g. dlPFC,
dmPFC, and vmPFC) in each of the studies. Further, UCS expectancy varied with activity
within several brain areas that showed UCR diminution. These findings demonstrate that
our emotional response is minimized to aversive events that are predictable. Further, our
studies have identified the neural circuitry that likely mediates this effect. Specifically,
the PFC regulates amygdala activity and the amygdala produces the peripheral emotional
response via projections to brainstem structures. These findings demonstrate that the
PFC-amygdala circuit mediates the emotional response to a threat.
124
We also investigated UCS controllability, in addition to predictability, to better
understand the neurophysiological mechanisms that affect conditioned UCR diminution.
In general, we observed conditioned UCR diminution in the final study within many of
the same brain areas (e.g. dlPFC, dmPFC, and vmPFC) as our initial investigations of
predictability. A novel finding in the final study was the affect of controllability on the
threat-related fMRI signal response within the vmPFC and hippocampus. Specifically, a
diminished response was only observed when the UCS was both predictable and
controllable. Given that predictability and controllability moderate the emotional
response to stress, the current findings identify specific brain regions that may mediate
the resilience to stress associated with predictable and controllable aversive events.
Studying the biological mechanisms that contribute to learning-related changes in
the fear response provides a starting point to better understand emotion dysregulation,
such as that observed in anxiety disorders (Davis et al., 2009; Grillon, 2002; Kim & Jung,
2006; Milad et al., 2006). Prior investigations of the neurobiological markers of anxiety
suggest that insufficient top-down regulatory control (Kim et al., 2011; Klumpp et al.,
2011; Nitschke et al., 2006; Rauch et al., 2006; Schienle et al., 2010) results in
hypersensitivity of subcortical brain areas (e.g. the amygdala) (Etkin & Wager, 2007;
Milad et al., 2007, 2006). The current project extends this prior work by demonstrating
that the predictability and controllability of a threat affects brain activity within the neural
circuitry that regulates and expresses emotion. Further, this project demonstrates that
individual differences in anxiety level influence the threat-elicited activity within these
brain regions.
125
There are some limitations to this project, specifically in Aims #2 and #3. For
example, prior work has demonstrated learning independent of expectation, in
anticipation of a threat (Knight et al., 2003, 2006; Schultz & Helmstetter, 2010). The goal
in Aim #2 was to assess UCR diminution independent of conscious expectations of the
threat. However, differences in the threat response were not observed independent of
expectation. Therefore, we were not able to address Aim #2 in this project. In Aim #3 we
expected to observe an increased threat-elicited response for participants in the UC group
compared to the CC group. However, no group differences were observed in the data.
One possibility for this result is the large between-subject variability that is typically
observed in these (i.e. SCR and startle EMG) psychophysiological measures.
Additionally, the effect of controllability has primarily been observed in anticipation of a
threat. However, Aim #3 investigated the influence of controllability on the threat-related
response. The impact of controllability on the threat response may be more difficult to
demonstrate compared to the influence of controllability in anticipation of a threat. Future
studies may better elucidate the affect of controllability on threat-elicited activation by
using a within-subject paradigm. Finally, learning-related changes in the
neurophysiological response to a threat were observed in a healthy population. A future
study of interest is to examine learning-related changes in the threat response using a
patient population (e.g. post-traumatic stress disorder or major depressive disorder).
Given the aberrant emotional regulation that is often found in anxiety disorders, we
would expect to observe diminished activation within the PFC and increased activation of
the amygdala in these groups compared to healthy individuals.
126
The work completed for this project provides a better understanding of the
neurophysiological mechanisms that influence the response to a threat. Specifically, we
have demonstrated learning-related changes in the neurophysiological response elicited
by a threat. The primary focus of fear conditioning studies has typically been centered
around the anticipatory response to a threat. This prior work has contributed to the
development of cognitive, behavioral, and drug interventions (Jovanovic & Ressler,
2010; Milad et al., 2009; Ressler et al., 2004; Vervliet et al., 2004). However, the current
studies have focused on the response to the threat itself. By demonstrating the influence
of predictability and controllability in modulating the emotional response to a threat we
have expanded the potential for novel treatments. For example, the findings from this
project have revealed learning-related changes in the threat response of specific brain
areas. Further, the psychophysiological response to a threat also showed learning-related
changes. Future investigation of the threat-related response to predictable and/or
controllable threats may provide additional insight for new cognitive, behavioral, or drug
interventions.
In summary, this project has contributed to a better understanding of the neural
circuitry that supports fear-related processes. The findings from these studies demonstrate
learning-related changes within the PFC-amygdala network that control the emotional
response to a threat. Further, this work demonstrates that regions of the PFC regulate the
emotional response controlled by the amygdala. Overall, the current findings provide a
better understanding of the neural mechanisms that mediate the emotional response to a
threat.
127
GENERAL LIST OF REFERENCES
Amat, J., Baratta, M., Paul, E., Bland, S., Watkins, L., & Maier, S. (2005). Medial
prefrontal cortex determines how stressor controllability affects behavior and dorsal
raphe nucleus. Nature neuroscience, 8(3), 365–71. doi:10.1038/nn1399
Baratta, M. V, Christianson, J. P., Gomez, D. M., Zarza, C. M., Amat, J., Masini, C. V,
Watkins, L. R., et al. (2007). Controllable versus uncontrollable stressors bidirectionally modulate conditioned but not innate fear. Neuroscience, 146(4), 1495–
503. doi:10.1016/j.neuroscience.2007.03.042
Baratta, M. V, Lucero, T. R., Amat, J., Watkins, L. R., & Maier, S. F. (2008). Role of the
ventral medial prefrontal cortex in mediating behavioral control-induced reduction
of later conditioned fear. Learning & memory (Cold Spring Harbor, N.Y.), 15(2),
84–7. doi:10.1101/lm.800308
Baxter, R. (1966). Diminution and recovery of the UCR in delayed and trace classical
GSR conditioning. Journal of experimental psychology, 71(3), 447–451.
Bellgowan, P. S., & Helmstetter, F. J. (1996). Neural systems for the expression of
hypoalgesia during nonassociative fear. Behavioral neuroscience, 110(4), 727–36.
Büchel, C., Morris, J., Dolan, R. J., & Friston, K. J. (1998). Brain systems mediating
aversive conditioning: an event-related fMRI study. Neuron, 20(5), 947–57.
Critchley, H. D., Mathias, C. J., & Dolan, R. J. (2001). Neural activity in the human brain
relating to uncertainty and arousal during anticipation. Neuron, 29(2), 537–45.
Davis, M. (1992). The role of the amygdala in fear and anxiety. Annual review of
neuroscience, 15, 353–75. doi:10.1146/annurev.ne.15.030192.002033
Davis, M., & Whalen, P. J. (2001). The amygdala: vigilance and emotion. Molecular
psychiatry, 6(1), 13–34.
Davis, M., Walker, D. L., Miles, L., & Grillon, C. (2009). Phasic vs sustained fear in rats
and humans: role of the extended amygdala in fear vs anxiety.
Neuropsychopharmacology: official publication of the American College of
Neuropsychopharmacology, 35(1), 1–31. doi:10.1038/npp.2009.109
Delgado, M. R., Nearing, K. I., Ledoux, J. E., & Phelps, E. A. (2008). Neural circuitry
underlying the regulation of conditioned fear and its relation to extinction. Neuron,
59(5), 829–38. doi:10.1016/j.neuron.2008.06.029
128
Domjan, M. (2005). Pavlovian conditioning: a functional perspective. Annual review of
psychology, 56, 179–206. doi:10.1146/annurev.psych.55.090902.141409
Dunsmoor, J. E., Bandettini, P. A., & Knight, D. C. (2007). Impact of continuous versus
intermittent CS-UCS pairing on human brain activation during Pavlovian fear
conditioning. Behavioral neuroscience, 121(4), 635–42. doi:10.1037/07357044.121.4.635
Etkin, A., & Wager, T. D. (2007). Functional Neuroimaging of Anxiety: A MetaAnalysis of Emotional Processing in PTSD, Social Anxiety Disorder, and Specific
Phobia. American Journal of Psychiatry, 164(October), 1476–1488.
Fanselow, M. S., & Ledoux, J. E. (1999). Pavlovian fear conditioning occurs in the
basolateral amygdala, 23, 229–232.
Foa, E. B., Zinbarg, R., & Rothbaum, B. O. (1992). Uncontrollability and
unpredictability in post-traumatic stress disorder: an animal model. Psychological
bulletin, 112(2), 218–38.
Grillon, C. (2002). Startle reactivity and anxiety disorders: Aversive conditioning,
context, and neurobiology. Biological psychiatry, 52, 958–975.
Grillon, C., Ameli, R., Woods, S., Merikangas, K., & Davis, M. (1991). Fear-potentiated
startle in humans: Effects of anticipatory anxiety on the acoustic blink reflex.
Psychophysiology, 28(5), 588–595.
Helmstetter, F. J. (1992). The amygdala is essential for the expression of conditional
hypoalgesia. Behavioral neuroscience, 106(3), 518–28.
Helmstetter, F. J., & Bellgowan, P. S. (1993). Lesions of the amygdala block conditional
hypoalgesia on the tail flick test. Brain research, 612(1-2), 253–257.
Helmstetter, F. J., & Bellgowan, P. S. (1994). Effects of muscimol applied to the
basolateral amygdala on acquisition and expression of contextual fear conditioning
in rats. Behavioral neuroscience, 108(5), 1005–9.
Herwig, U., Baumgartner, T., Kaffenberger, T., Brühl, A., Kottlow, M., Schreiter-Gasser,
U., Abler, B., et al. (2007). Modulation of anticipatory emotion and perception
processing by cognitive control. NeuroImage, 37(2), 652–62.
doi:10.1016/j.neuroimage.2007.05.023
Indovina, I., Robbins, T. W., Núñez-Elizalde, A. O., Dunn, B. D., & Bishop, S. J. (2011).
Fear-conditioning mechanisms associated with trait vulnerability to anxiety in
humans. Neuron, 69(3), 563–71. doi:10.1016/j.neuron.2010.12.034
129
Jovanovic, T., & Ressler, K. J. (2010). How the neurocircuitry and genetics of fear
inhibition may inform our understanding of PTSD. The American journal of
psychiatry, 167(6), 648–62. doi:10.1176/appi.ajp.2009.09071074
Kahnt, T., Heinzle, J., Park, S. Q., & Haynes, J.-D. (2011). Decoding different roles for
vmPFC and dlPFC in multi-attribute decision making. NeuroImage, 56(2), 709–15.
doi:10.1016/j.neuroimage.2010.05.058
Kim, J. J., & Jung, M. W. (2006). Neural circuits and mechanisms involved in Pavlovian
fear conditioning: A critical review. Neuroscience & Biobehavioral Reviews, 30(2),
188–202.
Kim, M. J., Gee, D. G., Loucks, R. A., Davis, F. C., & Whalen, P. J. (2011). Anxiety
dissociates dorsal and ventral medial prefrontal cortex functional connectivity with
the amygdala at rest. Cerebral cortex (New York, N.Y.  : 1991), 21(7), 1667–73.
doi:10.1093/cercor/bhq237
Kimmel, E. (1967). Judgments of UCS intensity and diminution of the UCR in classical
GSR conditioning. Journal of experimental psychology, 73, 532–543.
Klumpp, H., Ho, S. S., Taylor, S. F., Phan, K. L., Abelson, J. L., & Liberzon, I. (2011).
Trait anxiety modulates anterior cingulate activation to threat interference.
Depression and anxiety, 28(3), 194–201. doi:10.1002/da.20802
Knight, D. C., Lewis, E. P., & Wood, K. H. (2011). Conditioned diminution of the
unconditioned skin conductance response. Behavioral neuroscience, 125(4), 626–
31. doi:10.1037/a0024324
Knight, D. C., Nguyen, H. T., & Bandettini, P. A. (2003). Expression of conditional fear
with and without awareness. Proceedings of the National Academy of Sciences of
the United States of America, 100(25), 15280–3. doi:10.1073/pnas.2535780100
Knight, D. C., Nguyen, H. T., & Bandettini, P. A. (2005). The role of the human
amygdala in the production of conditioned fear responses. NeuroImage, 26(4),
1193–200. doi:10.1016/j.neuroimage.2005.03.020
Knight, D. C., Nguyen, H. T., & Bandettini, P. A. (2006). The role of awareness in delay
and trace fear conditioning in humans. Cognitive, affective & behavioral
neuroscience, 6(2), 157–62.
Knight, D. C., Smith, C. N., Cheng, D. T., Stein, E. A., & Helmstetter, F. J. (2004).
Amygdala and hippocampal activity during acquisition and extinction of human fear
conditioning. Cognitive, affective & behavioral neuroscience, 4(3), 317–25.
130
Knight, D. C., Smith, C. N., Stein, E. A., & Helmstetter, F. J. (1999). Functional MRI of
human Pavlovian fear conditioning: patterns of activation as a function of learning.
Neuroreport, 10(17), 3665–70.
Knight, D. C., Waters, N. S., King, M. K., & Bandettini, P. A. (2010). Learning-related
diminution of unconditioned SCR and fMRI signal responses. NeuroImage, 49(1),
843–8. doi:10.1016/j.neuroimage.2009.07.012
LeDoux, J. (1998). Fear and the brain: Where have we been, and where are we going?
Biological psychiatry, 44(12), 1229–38.
LeDoux, J. (2003). The emotional brain, fear, and the amygdala. Cellular and molecular
neurobiology, 23(4-5), 727–38.
LeDoux, J. (2007). The amygdala. Current biology: CB, 17(20), 868–874.
Lykken, D. T., Macindoe, I., & Tellegen, A. (1972). Preception: Autonomic Response to
Shock as a Function of Predictability in Time and Locus. Psychophysiology, 9(3),
318–333.
Maier, S. F., Amat, J., Baratta, M. V, Paul, E., & Watkins, L. R. (2006). Behavioral
control, the medial prefrontal cortex, and resilience. Dialogues in Clinical
Neuroscience, 8(4), 397–407.
Marcos, J. L., & Redondo, J. (1999). Effects of conditioned stimulus presentation on
diminution of the unconditioned response in aversive classical conditioning.
Biological psychology, 50(2), 89–102.
Maren, S. (2001). Neurobiology of Pavlovian fear conditioning. Annual review of
neuroscience, 24, 897–931. doi:10.1146/annurev.neuro.24.1.897
Milad, M. R., Pitman, R. K., Ellis, C. B., Gold, A. L., Shin, L. M., Lasko, N. B., Zeidan,
M. A., et al. (2009). Neurobiologial basis of failure to recall extinction memory in
posttraumatic stress disorder. Biological Psychiatry, 66(12), 1075–1082.
doi:10.1016/j.biopsych.2009.06.026.
Milad, M. R., Quirk, G. J., Pitman, R. K., Orr, S. P., Fischl, B., & Rauch, S. L. (2007). A
role for the human dorsal anterior cingulate cortex in fear expression. Biological
psychiatry, 62(10), 1191–4. doi:10.1016/j.biopsych.2007.04.032
Milad, M. R., Rauch, S. L., Pitman, R. K., & Quirk, G. J. (2006). Fear extinction in rats:
implications for human brain imaging and anxiety disorders. Biological psychology,
73(1), 61–71. doi:10.1016/j.biopsycho.2006.01.008
131
Milad, M. R., Vidal-Gonzalez, I., & Quirk, G. J. (2004). Electrical stimulation of medial
prefrontal cortex reduces conditioned fear in a temporally specific manner.
Behavioral neuroscience, 118(2), 389–94. doi:10.1037/0735-7044.118.2.389
Morgane, P. J., Galler, J. R., & Mokler, D. J. (2005). A review of systems and networks
of the limbic forebrain/limbic midbrain. Progress in neurobiology, 75(2), 143–60.
doi:10.1016/j.pneurobio.2005.01.001
Nitschke, J. B., Sarinopoulos, I., Mackiewicz, K. L., Schaefer, H. S., & Davidson, R. J.
(2006). Functional neuroanatomy of aversion and its anticipation. NeuroImage,
29(1), 106–16. doi:10.1016/j.neuroimage.2005.06.068
Nitschke, J. B., Sarinopoulos, I., Oathes, D. J., Johnstone, T., Whalen, P. J., Davidson, R.
J., & Kalin, N. H. (2009). Anticipatory activation in the amygdala and anterior
cingulate in generalized anxiety disorder and prediction of treatment response. The
American journal of psychiatry, 166(3), 302–10.
doi:10.1176/appi.ajp.2008.07101682
Pavlov, I. (1927). Conditioned reflexes. Oxford University Press, London.
Peeke, S. C., & Grings, W. W. (1968). Magnitude of UCR as a function of variability in
the CS–UCS relationship. Journal of experimental psychology, 77, 1–6.
Phillips, M. L., Drevets, W. C., Rauch, S. L., & Lane, R. (2003). Neurobiology of
emotion perception I: the neural basis of normal emotion perception. Biological
Psychiatry, 54(5), 504–514. doi:10.1016/S0006-3223(03)00168-9
Rauch, S. L., Shin, L. M., & Phelps, E. A. (2006). Neurocircuitry models of
posttraumatic stress disorder and extinction: human neuroimaging research--past,
present, and future. Biological psychiatry, 60(4), 376–82.
doi:10.1016/j.biopsych.2006.06.004
Ressler, K. J., Rothbaum, B. O., Tannenbaum, L., & Anderson, P. (2004). Cognitive
Enhancers as Adjuncts to Psychotherapy: use of D-Cycloserine in phobic individuals
to facilitate extinction of fear. Archives of general psychiatry, 61, 1136-1144.
Salomons, T. V, Johnstone, T., Backonja, M.-M., & Davidson, R. J. (2004). Perceived
controllability modulates the neural response to pain. The Journal of neuroscience:
the official journal of the Society for Neuroscience, 24(32), 7199–203.
doi:10.1523/JNEUROSCI.1315-04.2004
Salomons, T. V, Johnstone, T., Backonja, M.-M., Shackman, A. J., & Davidson, R. J.
(2007). Individual differences in the effects of perceived controllability on pain
perception: critical role of the prefrontal cortex. Journal of cognitive neuroscience,
19(6), 993–1003. doi:10.1162/jocn.2007.19.6.993
132
Schienle, A., Köchel, A., Ebner, F., Reishofer, G., & Schäfer, A. (2010). Neural
correlates of intolerance of uncertainty. Neuroscience letters, 479(3), 272–6.
doi:10.1016/j.neulet.2010.05.078
Schultz, D. H., & Helmstetter, F. J. (2010). Classical conditioning of autonomic fear
responses is independent of contingency awareness. Journal of experimental
psychology. Animal behavior processes, 36(4), 495–500. doi:10.1037/a0020263
Sehlmeyer, C., Dannlowski, U., Schöning, S., Kugel, H., Pyka, M., Pfleiderer, B.,
Zwitserlood, P., et al. (2011). Neural correlates of trait anxiety in fear extinction.
Psychological medicine, 41(4), 789–98. doi:10.1017/S0033291710001248
Vervliet, B., Vansteenwegen, D., & Eelen, P. (2004). Generalization of extinguished skin
conductance responding in human fear conditioning. Learning & memory, 11(5),
555–8. doi:10.1101/lm.77404
Viinikainen, M., Jääskeläinen, I. P., Alexandrov, Y., Balk, M. H., Autti, T., & Sams, M.
(2010). Nonlinear relationship between emotional valence and brain activity:
evidence of separate negative and positive valence dimensions. Human brain
mapping, 31(7), 1030–40. doi:10.1002/hbm.20915
Volz, K. G., Schubotz, R. I., & Von Cramon, D. Y. (2003). Predicting events of varying
probability: uncertainty investigated by fMRI. NeuroImage, 19(2), 271–280.
doi:10.1016/S1053-8119(03)00122-8
Wiech, K., Kalisch, R., Weiskopf, N., Pleger, B., Stephan, K. E., & Dolan, R. J. (2006).
Anterolateral prefrontal cortex mediates the analgesic effect of expected and
perceived control over pain. The Journal of neuroscience: the official journal of the
Society for Neuroscience, 26(44), 11501–9. doi:10.1523/JNEUROSCI.2568-06.2006
133
APPENDIX
IRB APPROVAL FORM
134
135