USF Libraries
USF Digital Collections

Emotion-modulated startle in major and minor depression


Material Information

Emotion-modulated startle in major and minor depression the role of mood severity in emotion reactivity
Physical Description:
Taylor-Clift, April
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:


Subjects / Keywords:
Subthreshold depression
Depression severity
Emotional responding
Eyeblink reflex
Positive affect
Dissertations, Academic -- Psychology -- Masters -- USF   ( lcsh )
non-fiction   ( marcgt )


ABSTRACT: Major depressive disorder (MDD) is a disorder defined by mood disturbance, but the deficits in emotional reactivity that accompany MDD are not yet fully characterized. Researchers have utilized the emotion-modulated startle paradigm to investigate emotional responding among depressed individuals with mixed results. Inconsistent results may be due in part to the heterogeneity of mood disorders, including variation in mood severity. The current study utilized an emotion-modulated startle procedure with 33 individuals currently experiencing a major depressive episode, 25 individuals currently experiencing a minor depressive episode (mD), and 31 healthy controls. Severity of depression, anxiety, and positive and negative mood states were ascertained on the sample. Emotion-modulated startle failed to differentiate between mood disordered individuals and healthy controls. However, results found a significant association between abnormal patterns of emotion responding and positive affect (PA), such that individuals with low PA showed exaggerated responding to unpleasant stimuli. The results suggest that PA may be an important dimension in mood disorders that underlies abnormal emotional responses.
Thesis (M.A.)--University of South Florida, 2008.
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by April Taylor-Clift.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 65 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001994093
oclc - 317322243
usfldc doi - E14-SFE0002419
usfldc handle - e14.2419
System ID:

This item is only available as the following downloads:

Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001994093
003 fts
005 20090327163529.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 090327s2008 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0002419
BF121 (Online)
1 100
Taylor-Clift, April.
0 245
Emotion-modulated startle in major and minor depression :
b the role of mood severity in emotion reactivity
h [electronic resource] /
by April Taylor-Clift.
[Tampa, Fla] :
University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 65 pages.
Thesis (M.A.)--University of South Florida, 2008.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
3 520
ABSTRACT: Major depressive disorder (MDD) is a disorder defined by mood disturbance, but the deficits in emotional reactivity that accompany MDD are not yet fully characterized. Researchers have utilized the emotion-modulated startle paradigm to investigate emotional responding among depressed individuals with mixed results. Inconsistent results may be due in part to the heterogeneity of mood disorders, including variation in mood severity. The current study utilized an emotion-modulated startle procedure with 33 individuals currently experiencing a major depressive episode, 25 individuals currently experiencing a minor depressive episode (mD), and 31 healthy controls. Severity of depression, anxiety, and positive and negative mood states were ascertained on the sample. Emotion-modulated startle failed to differentiate between mood disordered individuals and healthy controls. However, results found a significant association between abnormal patterns of emotion responding and positive affect (PA), such that individuals with low PA showed exaggerated responding to unpleasant stimuli. The results suggest that PA may be an important dimension in mood disorders that underlies abnormal emotional responses.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Advisor: Jonathan Rottenberg, Ph.D.
Subthreshold depression
Depression severity
Emotional responding
Eyeblink reflex
Positive affect
Dissertations, Academic
x Psychology
t USF Electronic Theses and Dissertations.
4 856


Emotion-Modulated Startle in Major and Minor Depression: The Role of Mood Severity in Emotion Reactivity by April Taylor-Clift A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Psychology College of Arts and Sciences University of South Florida Major Professor: Jonathan Rottenberg, Ph.D. David Drobes, Ph.D. Kristen Salomon, Ph.D. Cynthia Cimino, Ph.D. Date of Approval: March, 24, 2008 Keywords: subthreshold depression, depression severity, emotional responding, eyeblink reflex, positive affect Copyright 2008, April Taylor-Clift


Table of Contents List of Tables iii List of Figures iv Abstract v Introduction 1 Depression: Symptoms versus Diagnosis 2 Emotional Reactivity in MDD 5 Startle Paradigms and the Emo tion-Modulated Startle Response 7 Emotion Modulated Startle Response and Psychopathology 9 Emotion Modulated Startle Response and MDD 10 Specific Aims 12 Methods 15 Participants 15 Procedure Overview 17 Diagnostic Procedure 17 Ascertainment of Severity 18 Startle Sessions 18 Picture Stimuli 19 Startle recording 20 Skin Conductance and Heart Rate Recording 20 Data Reduction 21 Hypothesis Testing 22 Results 24 Overview of Results 24 Effects of Demographic Variables 24 Clinical Characteristics of the Sample 25 Subjective Ratings of Picture Valence 27 Startle Magnitude 28 Startle Magnitude and Severity Measures 30 Skin Conductance Responses 33 Relationships between Skin Conduc tance and Severity Measures 35 i


ii Cardiac Responses 36 Relationship between Cardiac Responses and Severity 38 Discussion 42 Effects of the Paradigm 43 Effects of Diagnosis and Symptom Severity 43 Affect and Hedonic Capacity 45 Conclusions 47 References 49 Appendices 61


iii List of Tables Table 1 Demographic Characteristics of the Sample 16 Table 2 Clinical Characteristics of the Sample 26 Table 3 Correlations between Severity Measures 26 Table 4 Self-Report Ratings 28 Table 5 Startle Magnitude 30 Table 6 Re-categorization of Diagnostic Groups by Median Splits 31 Table 7 Standardized Startle Responses According to Positive Affect Scores 33 Table 8 Skin Conductance Means 34 Table 9 Cardiac Measures by Diagnostic Group 37 Table 10 Cardiovascular Measures by Positive Affect 39


iv List of Figures Figure 1. Mean Standardized Startle Magnitudes Accordin g to Diagnostic Group 29 Figure 2. Startle Magnitude According to PA Group 33 Figure 3. Heart Phases (D, A1, D2) by Picture Valence 36 Figure 4. D1 Cardiac Responses by Dia gnostic Group for Each Picture Valence 39 Figure 5. A1 Cardiac Responses by Dia gnostic Group for Each Picture Valence 40 Figure 6. D1 Cardiac Responses by PA Group for Each Picture Valence 41 Figure 7. A1 Cardiac Responses by PA Group for Each Picture Valence 41


v Emotion-modulated Startle in Major and Minor Depression: The Role of Mood Severity in Emotion Reactivity April Taylor-Clift ABSTRACT Major depressive disorder (MDD) is a di sorder defined by mood disturbance, but the deficits in emotional reactivity that accompany MDD are not yet fully characterized. Researchers have utilized the emotion-m odulated startle paradigm to investigate emotional responding among depressed individu als with mixed results. Inconsistent results may be due in part to the heterogene ity of mood disorders, including variation in mood severity. The current study utilized an emotion-modulated startle procedure with 33 individuals currently experien cing a major depressive episod e, 25 individuals currently experiencing a minor depressive episode (m D), and 31 healthy controls. Severity of depression, anxiety, and positive and negativ e mood states were ascertained on the sample. Emotion-modulated startle failed to differentiate between mood disordered individuals and healthy cont rols. However, results found a significant association between abnormal patterns of emotion respondi ng and positive affect (PA), such that individuals with low PA showed exagge rated responding to unpleasant stimuli. The results suggest that PA may be an important dimension in mood disorders that underlies abnormal emotional responses.


1 Introduction Depression is one of the leading causes of disability worldwide (Murray & Lopez, 1997), and affects approximately one in five women and one in ten men in the United States (American Psychiatric Associa tion [APA], 1994). MDD symptoms include persistent sad mood and/or loss of interest or pleasure in daily ac tivities, as well as several associated somatic and cognitive sy mptoms, including loss of appetite, weight gain or loss, sleep difficulties, psychomoto r agitation or retard ation, lack of energy, feelings of worthlessness or guilt, concentr ation difficulties, and suicidal ideation (APA, 1994). MDD also has extremely high recurrence rates. Over 70% of depressed patients have more than one episode, and indeed, depr essed patients may spend only 22% of the 12 years following a major depressive ep isode without symptoms (Judd et al., 1998). Furthermore, approximately 40% of indivi duals with three or more episodes of depression may relapse within 12 to 15 weeks of recovery (Keller et al., 1992; Mueller et al., 1996). Given the recurrent nature of M DD, it seems reasonable that depressed individuals possess one or more stable vulnerability traits that predispose them to repeated episodes of this disorder. One proposed vulnerability trait i nvolves deficits in emotional responding. There are conflicting views on how MDD influences emotion. Based in part upon the assumption that moods will f acilitate reactions to like-va lenced emotions (Rosenberg, 1998), researchers have suggest ed that MDD (and the asso ciated depressed mood) may


2 facilitate negative emotions (e.g., Goli n, Hartman, Klatt, Munz, & Wolfgang, 1977; Lewinsohn, Lobitz, & Wilson, 1973) or inhibit positive emotions (Berenbaum & Oltmanns, 1992; Sloan, Strauss, Quirk, & Sa jatovic, 1997; Sloan, Strauss, & Wisner, 2001). However, as discussed in more deta il below, growing evid ence provides support for an alternative view of emotions in MDD, namely that MDD may involve inflexible responses across differing valence contexts, or emotion context insensitivity (ECI), (Rottenberg, 2005; Rottenberg, Gross, & Go tlib, 2005). The discrepancy between ECI (no mood enhanced reactivity in MDD) and th e idea of mood-facilita ted emotions raises the question of whether MDD represents a sp ecial case: Do the severe mood states in MDD have effects on emotional reactivity th at are distinct from those of milder depressed mood? To date, no studies have ex amined whether ECI exists across the range of depressed mood. This study enrolled a sample with a wide range of depression severity and collected data on several severity metr ics to examine the effects of depression severity on ongoing emotional reactivity. Emotional reactivity was probed using emotion-modulated startle paradigm, which ha s been extensively used to characterize motivation and pathophysiology of several c linical disorders (Grillon & Baas, 2003). Depression: Symptoms versus Diagnosis A longstanding debate in the literature concerns whether depression represents a distinct disease state (e.g., Judd, 1997) or whether the symptoms of depression are best thought of as existing along a continuum of severity (e.g., Cassano et al, 2004). The dimensional model of depression views the m ood and physical symptoms of depression as existing along a continuum, with th e diagnostic threshold placed somewhat arbitrarily—according to the number of symp toms and the level of impairment—at one


3 end of the continuum (e.g., Cassano et al, 2004; Prisciandaro & Roberts, 2005; Hankin, Fraley & Lahey, 2005). Many rese archers utilize dysphoric p opulations (individuals who self-report low mood, but who do necessarily m eet criteria for clinical depression) and generalize findings from this population to cl inically depressed i ndividuals (Vredenburg, Flett, & Krames, 1993). However, this methodology involves the assumption that dysphoria lies along the depression continuum and shares the same symptoms, mood states, and emotion regulation difficulties as diagnosable MDD (Coyne, 1993). There is some evidence to suggest that this generali zation is valid and that dysphoric individuals and clinically depressed indi viduals resemble one anothe r in several respects (e.g., Sweeney, Anderson, & Bailey, 1986). However, the literature usi ng the dimensional model has measured severity in several ways, and there is no consensus on which severity measure, if any, repr esents the most effective mean s to elucidate the dimensional properties of depressed mood. The disease state model, on the other hand, rejects the idea of a depressive continuum and argues instead that a clinical ly depressed person suffers from mood and physical symptoms that are qualitatively differe nt than those symptoms of an individual with ordinary sad mood (e.g., Ottowitz, Dougherty, & Savage, 2002) This model of depression generally conceptualizes clinical depression as resulti ng from neurological, neuroendocrine, or other biol ogically linked etio logies (e.g., Ottowitz, et al., 2002; Arborelius, Owens, Plotsky, & Nemeroff, 1999; Drevets, 1998). Such models presume that more severe depression connotes a mo re chronic disease pattern (Judd, 1997). For example, those individuals exhibiting more “diseased” neurological or biological processes may be more likely to suffer from ch ronic, recurring depres sion, whereas those


4 exhibiting less “diseased” neurological proce sses may be less likely to experience more chronic forms of the disease. The disease model generally pays less attention to the conceptualization of subthreshold forms of depression, although, some commentators also formulate subthreshold depression with in the disease framework (Judd, Akiskal, & Paulus, 1997). The Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) currently lists one form of subthres hold depression, termed minor depressive disorder (mD), in the Appendix as a provisi onal diagnosis meriti ng further study (APA, 1994). Minor depressive disorder is defined as a period of two or more weeks during which at least two to four of the nine sy mptoms for a major depressive episode are present, and one symptom must be either depressed mood or lack of interest or pleasure in most or all daily activities (APA, 1994). Minor depression may di ffer from MDD by its general lack of neurovege tative symptoms (Rapaport et al., 2002.), but the lack of research on mD limits strong claims about symptomatic differences. Evidence of mD prevalence rates are uncertain. Estimates vary depending on the degree of adherence to DS M-IV diagnostic criteria. Data from the nationally representative populatio n of the National Comorbidity Survey found lifetime prevalence rates for mD with no prior history of MD of 10% (Kessler, Zha o, Blazer & Swartz, 1997). Those studies that follow DSM-IV crite ria for mD found lower prevalence rates compared to those studies that diagnosed mD according to scores on depression severity scales, such as the BDI (e.g., 3.6% ve rsus 12.9%; Newman, Sheldon, & Bland, 1998; Beekman, Deeg, & van Tilburg, 1995).


5 Research indicates that mD results in impairments similar to those of MDD, a pattern which would support the continuum vi ew of depression. I ndividuals with mD often experience incomplete re solution of episodes as with MDD (Kessler et al., 1997). Minor depression can result significant f unctional disability and interfere with employment attendance to a similar degree as mild MDD (Cuijpers, de Graaf, & van Dorsselaer, 2004; Kessler et al ., 1997). There is some eviden ce that individuals with mD may use outpatient services as frequently as individuals with MDD (Gonzalez-Tejera et al., 2005). Individuals with mD are at an increased risk fo r developing MDD compared to individuals with no depressive symptoms (C uijpers et al., 2004). Indeed, the odds ratio for developing a first-time major depressive episode following a diagnosis of mD is as large as 5 (Fogel, Eaton, & Ford, 2006). Existing data is unclear as to whether mD is a transient mood state preceding or following a major depressive episode or whet her mD is a distinct disease with unique emotional correlates. Surprisingly, there have been virtually no di rect comparisons of these conditions, and no laboratory studies that compare mD and MDD on emotional characteristics. Emotional Reactivity in MDD Emotion provides an important context fo r extending the continuity-of-depression debate into an important domain of clinical functioning. There has been active research on emotional reactivity in MDD, where thr ee competing hypotheses for the emotional responding to positive and negative stimuli have emerged based on theory and research findings. These are described in turn.


6 The positive attenuation view hypothesizes that MDD individuals’ responses to positively valenced stimuli are attenuated co mpared to the responses of non-depressed controls. For instance, compared to contro ls, MDD individuals have shown diminished emotion response to pleasant film stimu li (Berenbaum & Oltmanns, 1992). Likewise, depressed individuals also report reduced emo tional responses to positive picture stimuli (Sloan, Strauss, Quirk, & Sajatovic, 1997; Sl oan, Strauss & Wisner, 2001). The negative potentiation view theorizes that the emo tional responses of MDD individuals to negatively valenced stimuli ar e potentiated or heightened co mpared to the responses of non-depressed controls. Little research has been conducted that supports this hypothesis and the negative potentiation hypothesis ha s not been supported when clinically diagnosed depression populations are used (Golin, Hartman, Klatt, Munz, & Wolfgang, 1977; Lewinsohn, Lobitz, & Wilson, 1973). Lastl y, the emotion context insensitivity (ECI) theory argues that MDD individuals’ core emotion defi cit is a general failure to exhibit context-appropriate emotional react ivity to valenced stimuli in general (Rottenberg, Kasch, Gross & Gotlib, 2002; Rottenberg, Gross, & Gotlib, 2005). There is growing evidence for ECI in MDD. For example, depressed patients have shown less electromyography (EMG) modulati on to affective stimuli (Gehricke & Shapiro, 2000) and less facial reactivity to expressive faces (Wexler, Levenson, Warrenburg, & Price, 1994). In fact, a r ecent quantitative review found broad based support for the idea that MDD individuals di splay ECI in a variety of experimental contexts (Bylsma, Morris, & Rottenberg, in press). However, Bylsma et al. also found that the effect sizes were heterogeneous. It is possible that severa l factors influence the presence or absence of ECI among MDD samples and account for the heterogenous


7 findings. The goal of the current study is to utilize the emotion-modulated startle paradigm to examine abnormalities in emotion processing and emotion reactivity in MDD, as well as to examine the role of se verity as a factor in fluencing the possible presence or absence of ECI in MDD. Startle Paradigms and the Emoti on-Modulated Startle Response Laboratory assessment of the startle response has resulted in methods for examining automatic processes to aversive stimuli in humans and other species. In its most basic form, the startle response characte rizes a defensive reflex to an aversive stimulus (e.g., a very loud sound). The st artle response includ es a cascade of evolutionarily adaptive behaviors, designed to protect the organism from harm, such as blinking of the eyes, a forward and downward movement of the head, and a drawing in of the shoulders (Landis & Hunt, 1939). In humans, the startle response is often quantified by the magnitude of the eye blink in response to the aversive stimul us, or startle probe, which is generally a brief burst of noise. Co mponents of the startle response typically occur within 20ms of the aversive stimulus. The startle paradigm has been extensiv ely studied in both animals and humans (e.g., Koch & Schnitzler, 1997), and the neurolog ical correlates of the startle response pattern are increasingly well understood (e.g., Lang, Bradley, & Cuthbert, 1998). Importantly, although the startl e response can be elicited re liably in many contexts, the magnitude of the response is influenced by th e current affective state of the organism. Specifically, numerous studies have found that emotional states elicited by affective stimuli (such as valenced pictures) reliabl y modulate the displayed amount of startle


8 amplitude in humans (Larson, Ruffalo, Nietert, & Davidson, 2000; Bradley & Lang, 2000; Bradley, Cuthbert, & Lang, 1990). In healthy subjects, the “late” startle probe that is presented anywhere from 3 to 6 seconds after the onset of the valenced stimu li elicits a greater startle response if the subject is viewing unpleasant pictures (e.g., snak es, injured humans) than if the subject is viewing pleasant (e.g., families, food) or neut ral pictures (e.g., landscape, garden tools), and pleasant pictures elicit smaller startle res ponses than neutral pictures. Thus, negative affect that is elicited by unpleas ant pictures potentiates the st artle response, while affect elicited by pleasant stimuli inhibits the st artle response. This pattern of emotionmodulated startle does not appear to result from unequal allocation of attention to the startle probe (Vrana & Lang, 1990; Bradley et al., 1990). The emotion-modulated startle response pers ists even in the pr esence of subjects’ habituation to the startle probe, and is thus thought to represent underlying motivational processes (Lang, 1995). Indeed, the typical e xplanation for this pattern of emotion reactivity is that startle res ponses are compatible with aver sive motivation (i.e., fight or flight mechanisms) and incompatible with a ppetitive motivational states (i.e., approach behaviors; Lang, Bradley & Cu thbert, 1998). This normative pa ttern of startle response can be conceptualized as normative emotion reactivity to affective stimuli. For the purposes of this proposal, a lack of context-appropriate emotion reactivity will refer to a subject’s failure to display both potentia tion in the context of unpleasant stimuli and inhibition in the context of pleasant stimuli.


9 Emotion-Modulated Startle Response and Psychopathology The emotion-modulated startle paradigm has been widely used to demonstrate atypical patterns of emotion-modulated star tle in several mental disorders (Grillon & Baas, 2003). Thus far, some evidence indica tes disorder-specific patterns of atypical startle responding. A lack of startle pote ntiation during unpleasa nt picture viewing characterizes schizophrenic patients, who also display overall deficient habituation to the acoustic startle probe (Schlenker, Cohen, & Hopmann, 1995; Taiminen, et al, 2000). Psychopaths display an abnormal pattern of startle modulation comp ared to controls, generally showing equivalent startle responses for unpleasan t stimuli and pleasant stimuli and heightened responding to neutral s timuli (Patrick, Bradley, & Lang, 1993). The modulated startle pattern of individuals with generalized anxi ety disorder (GAD) generally indicates a higher overall startle, but no consis tent pattern of reactivity differences (Grillon & Baas, 2003). However, re search with other anxiety disorders has yielded consistent patterns that seem to di fferentiate these conditions. In some severe posttraumatic stress disorder patients ther e is larger potentia tion during unpleasant pictures compared to controls and higher baseline startle responses compared to controls (Miller & Litz, 2004; Morgan, Grillon, Sout hwick & Charney, 1996). Studies have also found evidence for increased startle potentiation to feared stimuli in subjects with specific phobias and an increased baseline startle and increased startle potentiation in patients with panic disorder and social phobia (Ha mm, Cuthbert, Globisch, & Vaitl, 1997; Larsen, Norton & Walker, 2002; Cuthbert, Lang, & St rauss, 2003). Thus, emotion-modulated startle patterns may be an appropriate and sensitive means to differentiate clinical disorders (such as mD and MDD).


10 Emotion Modulated Startle Response and MDD Despite the general promise of the emoti on-modulated startle paradigm, relatively little research has been conduc ted with it in depressed populat ions. Most of the existing studies are cross sectional in nature and have generated conflicting results (reviewed below). Consistent with the approach take n in this proposal, va riation in depression severity both across and within studies may explain these discrepancies, with more severe, diagnosable depression being associated with more pronounced deficits in startle modulation. Research with a sample of distressed co llege students—as indexed by high scores on the Depression scale of th e Minnesota Multiphasic Pe rsonality Inventory—supported the negative potentiation hypothesis. Specifical ly, highly distressed individuals showed increased potentiation during negative imagery trials (Cook, Hawk, Davis, & Stevenson 1991). These results suggest that high levels of distress and possibly the presence of mD may actually potentiate respondi ng to negative stimuli. By contrast, in clinically depressed sa mples, findings from emotion-modulated startle studies are most cons istent with ECI. For instance, Dichter and colleagues found, in two independent samples, a flattened patt ern of startle (i.e., l ack of inhibition for pleasant pictures and lack of potentiati on for unpleasant pictures) among clinically depressed outpatients as compared to contro ls (Dichter et al., 2004; Dichter & Tomarken, 2008). In another early study of this type (Allen et al., 1999), c linically depressed inpatients showed a normative pattern of star tle modulation; however, additional analyses suggested that results differed when patie nts were divided into groups based on the severity of the depressive symptoms, as indexed by scores on the Beck Depression


11 Inventory (BDI). Among subjects with se vere BDI scores, startle magnitude was potentiated during pleasant pict ures but not during unpleasant pictures. In more mildly depressed subjects with mode rate and lower BDI scores, a pattern similar to the normative pattern of emotion-modulated star tle reflex startle wa s observed, indicating context appropriate emotion reactivity. Therefore, the pa ttern of modulated startle exhibited by the severely depressed group in this study seems to conform with the ECI hypothesis, whereas the pattern of startle in the more mild ly depressed sample does not support ECI. Although th ese internal analyses are lim ited by the small sample sizes involved (N’s = 7), they illustrate the poten tial importance of symptom severity in modifying startle reactivity. Kaviani and colleagues also observed e ffects of depression severity on emotionmodulated startle reflexes in a DSM-IV di agnosed, clinically de pressed group (Kaviani, Gray, Checkley, Wilson, & Kumari, 2004). No significant differences in startle modulation were found between the depresse d patients and the control group. However, when depressed participants were divide d into low and high depressed groups on the Hospital Anxiety and Depression Scale (HADS ), the low depressed group showed the normative pattern of modulated startle, which did not differ significantly from the startle pattern of controls, while the high-depressed gr oup showed a flattened pattern of startle. When depression severity was defined in terms of the number of previous depressive episodes, ECI was supported fo r a subgroup of depressed individuals who reported the most recurrent depressive epis odes (Forbes et al., 2005). That is, when depressed patients were assigned to one of three severity gr oups based on the number of previous depressive episodes, those patie nts with episodes “too numerous to count”


12 displayed a flattened pattern of modulation. The group with one or two prior episodes and the group with three or more episodes di splayed the normative pattern of startle modulation. The results of Forbes and collea gues again suggest that individuals with more severe MDD (for example, outpatients with multiple depressive episodes) lack context-appropriate emotion reactivity, whereas those with fewer episodes show intact emotion reactivity. Indeed in th is study depressed participants as a whole showed a startle modulation pattern generally similar to that of controls. In summar y, while the database of findings remains modest, MDD is sometime s associated with an absence of emotion modulated startle, and mild and moderate leve ls of MDD are sometimes associated with a more normative startle pattern. One critical limitation of prior studies of the emotionmodulated startle paradigm is that they have not fully re presented the full range of depression severity; that is, no studies have included partic ipants with both minor and major depression in the same sample. This ki nd of research design is needed to address the question of whether minor or mild m ood disturbances have effects on emotional reactivity than are distinct from those s een in more severe diagnosable depression. Specific Aims Existing research of the emotion-modulated startle pattern of individuals with MDD has produced mixed findings. The current re view suggests that this heterogeneity may be due to several factors, including the heterogeneity of depression, variations between studies in how severity is measure d, and inconsistencies in how cut off points for low and high depression severity are use d. The current study a ddressed the use of differing measures for depression severity by utilizing individuals with minor depression, as diagnosed according to DSM-IV-TR crite ria (mD; APA, 1994) and by analyzing


13 emotional responding according to several diffe rent measures of symptom severity (e.g., amount of negative affect, lack of positive a ffect, presence and seve rity of comorbid anxiety). We predicted that levels of em otion-modulated startl e would be linked to depression severity. By examining several m easures of depression severity, we could identify the measure of severity that pr ovides the clearest di fferences in contextappropriate emotional reactivity. The general aim of this study was to i nvestigate emotional reactivity across varying levels of severity. More specifi cally, we sought to cl arify whether contextappropriate emotion reactivity varies accord ing to DSM-IV diagnostic categories of depression and according to several alternative severity metrics. To achieve these aims, the following hypotheses were tested: Hypothesis 1a: Startle modulation of depressed individuals will be significantly different from the startle patterns of contro l subjects and subjects with mD, such that MDD individuals will show fl attened emotion-modulated star tle responses re lative to mD individuals and healthy controls. Hypothesis 1b: It was predicted—based on Cook et al. (1991)—that individuals with minor depression would display grea ter negative potentiation compared to the responses of all other groups Hypothesis 2 (secondary): Measures of symptom severity will differentiate those with atypical emotion-modulated startle patterns from thos e with more typical emotionmodulated startle patterns. The primary analyses focused on the ma gnitude of the startle response as a function of stimulus valence. Secondary anal yses were conducted with skin conductance


14 and heart rate. Although the l iterature provided less guidance for predictions based on skin conductance and heart rate, we had similar hypotheses for these variables and expected that analyses would parallel results from startle.


15 Methods Participants Participants were recruited from f liers and online posti ngs in and around the Tampa Bay community (Table 1). Over 460 potential participants were screened by telephone. Of those individuals, 159 were invite d into the lab to complete the SCID. Of those participants who completed a SCID, approximately 52 were excluded for failing to meet inclusion or exclusion criteria. A further 21 individuals failed to attend the scheduled interview session for a variety of reasons (e.g., scheduling changes, failing to return calls to reschedule, et c). Participants were excluded for history of a major head injury, hearing impairment, di agnosis of bipolar disorder substance abuse occurring within 6 months prior to en try into the study, or any hi story of primary psychotic symptoms. Final participants were primarily females (77%) fluent in English and between the ages of 18 and 55. The final sample approxima ted the ethnic distribution of the Tampa Bay area: 59.8% Caucasian, 16.3% African American, 10.9% Latino/ Hispanic, 6.5% Asian, & 1.1% Native American. According to diagnoses based on DSM-IV-TR criteria, participants were experiencing a cu rrent Major Depressive episode ( n = 33), a current Minor Depressive episode ( n =25), or had no past or pr esent psychopathology (i.e., no history of any Axis I disorder as asse ssed by the SCID, including past mD episodes; n = 31). Table 1 contains demographic informa tion of the sample according to diagnostic


16 group. Participants were matched on age, ethn icity, gender, education level, income, and marital status (all p s > .11 for Cramer V tests). Provisional DSM-IV-TR criteria recommend an absence of past episodes of MDD for an mD diagnosis. To improve study feasibil ity, we loosened this criterion and 36% of mD participants experienced at least one major depressive episode (MDE). In these included subjects we required a period of at least eight weeks with no residual depressive symptoms between the major depressive episod e and the minor depressive episode. In all cases of mD with a past MDE, MDEs occurred at least one year pr ior to the current mD episode. Table 1 Demographic Characteristics of the Sample Group MDD mD Controls Variable (n = 34) (n = 26) (n = 32) Age, M (SD) 29.88 (11.25)25.85 (6.88) 26.72 (7.79) % Caucasian 46.9% 73.1% 61.8% % Female 88.2% 65.4% 75% Education 5.29 (1.85) 5.23 (1.75) 5.63 (2.14) Income 5.31 (3.44) b 4.40 (3.07) b 5.93 (3.89) b % Married 17.6% 19.2% 28.1% % Antidepressants 14.7% 5.3% 3.7%c % Psychotherapy 11.8% 5.3% 0% Education was assessed on an 8-point scal e with higher numbers representing more education—a score of 5.63 re flects graduation from a 2-year or a technical college. bIncome was assessed on a 12-point scale— a score of 5.93 represents an income of between $25,000 and $34,999. cOne control participant was taking an antidep ressant to (successfully) control migraine headaches.


17 Procedure Overview Individuals responding to rese arch ads were initially screened over the phone to determine potential eligibility. Screeni ng questions were based on key diagnostic questions from the Structured Clinical Inte rview based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV-TR) Axis I Disorders, Research Version, Patient Edition with Psychotic Scr een (SCID-I/P W/ PSY SCREEN; First et al., 2002). Based on this initial screening, potential participants were invited to complete a full SCID with a clinical docto ral student. Final diagnoses for study inclusion were made based on this SCID administ ration. Participants also competed a Beck Depression Inventory-II (BDI-II; Beck, Steer & Brown, 1996) at this session. Participants deemed eligible based on th e initial SCID interview were invited to return to the lab within one week to pa rticipate in the emo tion-modulated startle procedure. The diagnostic interview was sepa rated from the startl e procedure to reduce participant burden and redu ce the likelihood that the in terview would in any way influence participants’ emotion reactivity in the startle procedure. Diagnostic Procedure The SCID is a semi-structured intervie w designed to diagnos e individuals based on the DSM-IV. Reliability and validity m easures for the SCID differ according to population and diagnosis, but reliability for diagnosing MDD is relatively high with interrater reliability kappas ranging fro m .80 to .93 (Zanarini & Frankenburg, 2001; Zanarini et al., 2000; Skre, Onstad & Torger sen, 1991). Screening wa s conducted for the following diagnoses: bipolar I and II diso rder, major depressive disorder, minor depressive disorder, dysthymic disorder, schizophrenia, schizoaffective disorder,


18 schizophreniform disorder, delusional disord er, brief psychotic disorder, substance dependence, social phobia, specific phobia, and obsessive com pulsive disorder. Ascertainment of Severity The Beck Depression Inventory (BDI-II), a well-validated, 21-item, selfadministered scale, was used to assess depres sion symptom severity. Scores range from 0 to 63 with higher scores representing more severity. Coefficient alphas for the BDI-II were high (alpha = 0.96). The test -retest reliability has also proven to be high at r = .93 (Beck, Steer, & Brown, 1996). The Beck Anxi ety Inventory (BAI) is a 21-item selfadministered questionnaire used to assess an xiety symptom severity. Symptoms are rated on a four-point scale, with higher scores in dicating more severe a nxiety symptoms (Beck, Epstein, Brown, & Steer, 1988). The internal consistency of the BAI for the current study was high (alpha = .92), and the BAI corre lates highly with the SCL-90-R Anxiety Subscale (r = .81) (Steer, Ranieri, Bec k, & Clark, 1993). The Positive and Negative Affect Schedule (PANAS) is a 20 item self-repor t scale, measuring dispositional forms of positive and negative affect (Watson, Clark, & Tellegen, 1988). The PANAS has successfully differentiated depression and a nxiety in clinical samples (Dyck, Jolly, & Kramer, 1994; Jolly, Dyck, Kramer, & Wh erry, 1994). The PANAS was also highly reliable, with a Cronbach’s alpha of 0.90 fo r positive affect and 0.88 for negative affect. Startle Session Upon arrival to the startle procedure, participants completed a second BDI-II, a BAI, and a PANAS. Following completion of questionnaires, partic ipants were seated approximately 1.5 feet from a 20-inch comput er monitor placed on a table directly in front of the participant. Prior to pictur e presentation, electrode s were placed on the


19 subjects and impedances were checked. To familiarize participants with the procedure and to habituate participants to the startle probe, two sample neutral images were shown accompanied by inter-trial and picture-view ing startle probes and the participant completed two sets of ratings. Pictures were presented in 3 blocks of 12 pictures, with each block including four pictures of each va lence. Pictures were presented in semirandom order with the constraint that no more than two pict ures in a row were of the same valence. The picture presentation se quence was as follows: (1) 2-second baseline; (2) six-second picture viewing; (3) 20 seconds to rate the valence and arousal of the picture using the self-assessment maniki n procedure (SAM; La ng, 1980; Hodges, Cook & Lang, 1985); (4) variable (15-second average) inter-trial intervals prior to presentation of the next picture. The startle response was elicited by a bina ural acoustic stimulus (50 milliseconds of white noise at 105db with an instantaneous ri se time) during nine of the twelve images in each category, and during nine of the inte r-trial intervals. Startle responses were probed at varying times between 3000 a nd 5500ms after picture onset to assess contextual affective processing and defensiv e versus appetitive motivational systems (Bradley, Cuthbert, & Lang, 1993). Probes also occurred randomly between 6000ms and 9000ms after picture presentation (inter-trial startles ) for nine of the pictures (three of each valence category). These inter-trial st artle probes were included to decrease the predictable anticipation of startle probes during picture viewing. Probe times were semirandom with the constraint that no more than two of each probe time occurred in a row. Picture Stimuli. Pictures were 36 affective pictur es designed to elicit positive, negative, or neutral affect from the Interna tional Affective Picture System (Center for the


20 Study of Emotion & Attention, 1999)1. Pictures were drawn from a recent study of startle modulation in depressed persons (Dichter & Tomarken, 2008). Pleasant and unpleasant pictures were matched on reported levels of arousal (Lang, Bradley, & Cuthbert, 2005). Following Dichter and Tomarken (2008), different sets of pictures were shown to males and females; these sets were matche d on ratings of valence and arousal. Startle Recording. Startle data were collected an d analyzed following guidelines and recommendations set forth in Blumenthal and colleagues’ committee report (2005). Stimulus control and physiologi cal data collection utilized an IBM compatible computer running VPM data acquisition and reduction software (Cook, 1997). Data were collected and stored offline for later analyses. For measurement of the ey eblink component of startle reactivity, 2 “small” (4 mm) Beckma n-type electrodes were placed 36mm apart just beneath the lower eyelid of the left eye to record the contraction of the orbicularis oculi muscle. Impedance for the electrode pa ir was less than 20 Ohms. EMG data were amplified using a high-resolu tion (A/D) converter. Next EM G signal was filtered to remove background noise and maximize the signal to noise ratio. Lastly, the signal was digitally smoothed. Skin Conductance and Heart Rate Recording Continuous physiological data, i.e., heart rate and skin conductance data, were collected and stored offline throughout the procedure. To measure cardiac activity, three “large” (8 mm) Beckman-type electrodes were placed between the pa rticipant’s wrist and elbow. Additionally, two large el ectrodes were applied to the palm of the participant’s non-dominant hand to measure skin conducta nce responses during picture display.


21 Initial heart rate decelera tion (D1) magnitude as compared to baseline and subsequent acceleration (A1) compared to ba seline were computed for each picture and averaged across valence conditions. Initial he art rate deceleration occurs within three seconds following picture onset and is genera lly conceptualized as representing sensory orienting responses, while acceleration occurs within two to five seconds following picture onset and is representative genera lly of emotional processing or defensive responding (Lacey & Lacey, 1970; Graham & Clifton, 1966; Sokolov, 1963). More specifically for the current study, the magnitude of D1 represents the degree of initial sensory orientation towards a stimulus, while the magnitude of A1 represents the degree of emotional processing of a stimulus. Initial deceleratory responses are generally greater for unpleasant than for pleasant pictures, wh ile acceleratory res ponses are generally greater in the presence of pleasant stimu li (Lang, Greenwald, Br adley, & Hamm, 1993). For skin conductance, the peak magn itude, average magnitude, and difference between baseline and peak magnitude for pi cture presentation (b etween two and six seconds following picture onset) were computed for each valence condition. Skin conductance is influenced solely by the sy mpathetic nervous system. Considerable evidence indicates that skin conductance increases in response to arou sing stimuli and is greatest in response to the most arousing pleasant and unpleasant stimuli (Bradley, Codispoti, Cuthbert, & Lang, 2001). Data Reduction EMG signals for each individual were an alyzed for onset latency and, most importantly for the purposes of the current study, for mean peak amplitude by condition. Peak magnitude was manually scored by determining peak amplitude between 20 and


22 120 ms following startle probe onset and subtracting onset EMG activity. Peak magnitude values were transformed into standardized T scores (with a mean of 50 and a standard deviation of 10) using the mean a nd standard deviation of startle responses during neutral pictures. This procedure accounts for arbitrar y individual differences in baseline startle response values while pres erving response differences between valence conditions. T scores were subsequently averaged across each valence condition. Eyeblink reflexes were excluded (treated as missing valu es) if the reflex occurred 20ms or earlier before the startle probe onset or if an unstable baseline period precluded the determination of startle onset. Trials with no perceptible eyeblink startle response were given a magnitude score of zero and were in cluded in analyses. Twenty participants— termed non-responders—were excluded because th eir data did not yield startle responses for greater than 45% of trials. A Cramer’s V analysis confirmed that there were no differences in the number of non-responders within each diagnostic group ( p = .97). Approximately 22% (7) healthy controls were non-responders, 21% (7) of MDD individuals were non-responders, and 23% (6) of mD individuals were non-responders. Hypothesis Testing A repeated measures analysis of va riance (ANOVA) was performed for each physiological variable with diagnostic gr oup (control, MDD, mD) as the betweensubjects variable and picture valence (pleasan t, neutral, unpleasant) as the within-subjects variable. Assumptions of sphericity were met except when otherwise stated. Repeated measures ANOVAs were followed by one-way ANOVAs and paired sample t tests when appropriate. Additionally, subjec ts’ ratings of picture valen ce and arousal were analyzed


23 separately using repeated-measures ANO VAs. Again, these were followed up by oneway ANOVAs and t tests when appropriate. In order to test secondary hypotheses th at measures of symptom severity would differentiate those with atypica l emotion-modulated startle patterns from those with more typical emotion-modulated startle patterns, se verity groups were created based on median splits of the severity measures for mD a nd MDD individuals comb ined. Three (severity group: high, moderate, controls ) by three (picture valence: pleasant, neutral, unpleasant) ANOVA’s were then conducted with each physiological measure of interest.


24 Results Overview of Results Initial analyses were conducted to asse ss the possible effect of demographic variables on the dependent variab les and confirm the validity of the clinical diagnostic groups. Subjective ratings of pi cture valence and arousal were then analyzed to confirm that the intended manipulati on was successful and to exam ine any possible differences between groups for these ratings. The primary results for startle are then presented, with repeated measures analysis of variance (ANOVA) of group effects on startle variables preceding repeated-measures ANOVA analyses of severity. These are followed by parallel secondary analyses of skin conductance and heart rate. Effects of Demographic Variables Repeated measures ANCOVAs were conducted for physiological variables by diagnostic group with gender, education level, ethnicity, income, treatment, and picture order entered individually as covariates. Ther e were no significant in teraction effects for these covariates on any of the physiological variables of interest including baseline physiological measures ( p s > .13). Therefore, these demographic variables were omitted from future analyses. Additionally, there were no significant effects of medication on physiological variables. However, for medica tion status, small cell sizes (see Table 1) made it difficult to assess possible medicati on effects due to low power. Because of the potential for antidepressant and anxiolytic medications to influence physiological measures, significant and trend-level eff ects were followed by analyses in which


25 medication status (as present or absent) was in cluded as a covariate. Where such analyses resulted in changes to significant fi ndings, these resu lts are reported. Clinical Characteristics of the Sample Results of the questionnaire measures are listed in Table 2. Results confirmed diagnostic categorization of groups. As exp ected, a one-way anal ysis of variance (ANOVA) confirmed that Beck Depression I nventory (BDI) scores varied significantly among all three groups [ F (2, 87) = 90.73, p < .01] such that MDD individuals had the highest BDI scores (indicati ng higher depression severity) followed by mD individuals and then control individuals. Beck Anxi ety Inventory (BAI) sc ores also differed significantly between groups [ F (2, 87) = 40.22, p < .01]. Follow-up tests indicated that MDD individuals and mD individuals endorse d significantly higher symptoms of anxiety than did healthy individuals ( p < .01), but that scores did not differ significantly between MDD and mD individuals ( p = .34). Positive Affect (PA) scores also di ffered significantly between groups [ F (2, 87) = 22.59, p < .01]. MDD and mD individuals both differed significantly from control individuals in the expected direction, showing significan tly lower levels of PA ( p < .01). MDD and mD individuals also differed significantly in terms of PA in the expected direction with mD individuals showing si gnificantly higher PA compared to MDD individuals ( p < .05). Negative Affect (NA) diffe red significantly between groups [ F (2, 87) = 17.23, p <. 01], such that MDD and mD individual had significantly higher negative affect compared to healthy controls ( p < .01). However, MDD and mD individuals not differ signifi cantly in terms of NA ( p = .15). Zero order correlations


26 between severity measures and all physio logical measures are presented in the Appendices. Table 2 Clinical Characteristics of the Sample Group MDD mD Controls ( n = 33) ( n = 25) ( n = 31) Variable Mean (SD) Mean (SD) Mean (SD) BDI 29.03 (11.25)19.52 (6.70) 2.29 (3.16) BAI 16.51 (8.73) 13.88 (8.45) 1.52 (1.84) PA 20.94 (6.27) 26.08 (6.68) 32.48 (7.60) NA 20.76 (8.47) 18772 (5.76)11.87 (2.60) Note: BDI = Beck Depression Inventory BAI = Beck Anxiety Inventory PA = positive affect NA = negative affect Correlations between clinical characteris tics are given in Table 3. All correlations among questionnaire measures reached significance and were in the expected directions. Table 3 Correlations between Severity Measures Variable BDI BAI PA NA BDI 0.69*-0.67*0.62* BAI -0.44*0.63* PA -0.40* NA p < .01 Note: BDI = Beck Depression Inventory BAI = Beck Anxiety Inventory PA = positive affect NA = negative affect


27 Subjective Ratings of Picture Valence Picture ratings of valence a nd arousal are presented in Table 3. Subjective ratings of picture valence confirmed that pictur es differed according to valence condition. Likewise, analyses of subjective arousal ra tings confirmed that pleasant and unpleasant pictures were matched on arousal. More spec ifically, a series of repeated measures ANOVAs were conducted to examine differences among subjective picture ratings between groups. Results indicated a si gnificant main effect of Valence [ F (2, 86) = 482.46, p < .01, ˆ = .85] and Arousal [ F (2, 86) = 92.21, p < .01, ˆ = .75]. Valence ratings displayed the expected linear trend, with pleasant pi ctures receiving the highest ratings and unpleasant pictures receiving the lo west ratings of valence. Arousal ratings for pleasant pictures and unpleas ant pictures were significan tly higher than ratings for neutral pictures ( p < .01), but did not differ significantly fr om each other, indicating that the valenced pictures were successfully matched on arousal. In addition, there was a small, significant group by picture valence in teraction effect for arousal ratings [ F (4, 170) = 2.76, p < .05, ˆ = .06], such that MDD individuals rated unpleasant pictures as significantly more arousing th an did mD individuals [ t (1, 56) = 2.07, p < .05] or controls [ t (1, 61) = 2.39, p < .05], but did not differ in the neut ral picture or pleasant pictures.


28 Table 4 Self-Report Ratings Group MDD mD Controls (n = 33) (n = 25) (n = 30) Valence Mean (SD) Mean (SD) Mean (SD) Pleasant 14.24 (2.38) 14.06 (2.80) 14.54 (2.59) Neutral 10.24 (1.22) 10.37 (1.22) 10.54 (1.84) Unpleasant 2.26 (2.77) 3.51 (2.78) 3.27 (2.53) Arousal Pleasant 12.60 (2.58) 12.49 (3.14) 13.47 (2.83) Neutral 6.14 (2.54) 7.35 (2.88) 6.78 (4.16) Unpleasant 14.10 (3.84) 11.74 (4.83) 11.65 (4.29) *Ratings are based on a 0-20 scale. Startle Magnitude Table 5 lists the mean startle T scores by diagnostic group. A repeated-measures ANOVA for startle responses between diagnos tic groups indicated the expected linear main effect for picture-type [ F linear (1, 69) = 17.81, p < .05, ˆ = .21]. Pairwise comparisons indicated that pleasant pictur es differed significantly from unpleasant pictures (p < .05), with larg er startle responses being elicited during unpleasant as compared to pleasant pictures. There was a trend towards lower startle responses during pleasant pictures as compared to neutral pictures ( p = .06) and towards larger startle responses during unpleasant pictures as compared to neutral pictures ( p = .05). Figure 1 shows the overall startle responses for each gr oup according to picture type. Inconsistent with our primary hypothesis, there were no significant effects for diagnostic group [ F (2, 72) = .71, p = .59], nor were there group by picturevalence linear or qua dratic trends.


29 46 47 48 49 50 51 52 53 MDDmDControls Diagnostic Group Pleasant Neutral Unpleasant * Figure 1. Mean standardized startle magnitudes according to diagnostic group. To further examine the effects of dia gnostic category on startle reactivity, we examined two other widely used startle metrics—startle amplitude and raw startle magnitude values—for possible group interaction effects. Startle amplitude is calculated in much the same way as startle magnitude with the exception that trials indicating no startle response are omitted from analyses. In this way, startle magnitude for only those pictures eliciting a response are utilized for analyses. Re sults of a repeated-measures ANOVA for startle amplitude failed to reve al significant group by picture valence interactions [ F (4, 138) = 1.13, p = .35]. Likewise a re peated measure ANOVA was conducted using raw (unstandardized) startle ma gnitude values as the dependent variable and results again indicated only significan t main effect for picture valence and no significant group by valence interactions [ F (4, 138) = 1.59, p = .18]. Table 5 presents


30 amplitude and raw magnitude scores for each group. In summary, across the various startle metrics we obtained the expected valence modulation effects. However, the hypothesis that emotion-modula tion of startle for depresse d individuals would differ significantly from that of c ontrols or that of mD i ndividuals was not supported. Table 5 Startle Magnitude MDD mD Controls ( n = 27) ( n = 20) ( n = 25) Valence Mean (SD) Mean (SD) Mean (SD) Magnitude T Scores Pleasant 48.74 (2.81) 49.52 (3.42) 48.54 (2.22) Neutral 49.73 (3.35) 50.17 (3.20) 50.10 (2.92) Unpleasant 51.94 (3.15) 50.71 (2.74) 50.93 (2.66) Amplitude T Scores Pleasant 48.54 (2.93) 49.56 (3.20) 48.67 (2.33) Neutral 49.57 (3.59) 50.19 (3.09) 49.89 (2.81) Unpleasant 52.15 (3.27) 50.49 (2.76) 50.91 (2.68) Raw Scores Pleasant 163.77 (145.88) 170.58 (161.08) 134.83 (107.49) Neutral 173.15 (151.47) 167.01 (140.02) 152.88 (122.85) Unpleasant 194.48 (180.17) 176.53 (149.32) 154.84 (134.20) Startle Magnitude and Severity Measures In order to test secondary hypotheses that measures of severity would differentiate atypical emotion-modulated startle respondi ng from typical emotion-modulated startle responding, severity groups were created according to median splits on severity measures. To establish groups according to se verity, mD and MDD individuals were first analyzed together and the median for each questionnaire score was attained. Minor and


31 major depressed individuals were then di vided into two new groups based on these median splits and compared with the healt hy control groups. Table 6 lists the resulting categorization of diagnostic groups according to severity measures. Analyses conducted with each of the three severity groups parallel those conducted with diagnostic groups. Table 6 Re-categorization of Diagnostic Groups by Median Splits Group MDD mD Controls ( n = 33) ( n = 25) ( n = 31) BDI Group Mean (SD)Mean (SD)Mean (SD) Controls 0 0 31 Moderate 10 20 0 High 23 5 0 BAI Group Controls 0 0 31 Moderate 15 14 0 High 18 11 0 PA Group Controls 0 0 31 Moderate 13 18 0 High 20 7 0 NA Group Controls 0 0 31 Moderate 16 16 0 High 17 9 0 Note: BDI = Beck Depression Inventory BAI = Beck Anxiety Inventory PA = positive affect NA = negative affect Specifically, three (severity group) by thr ee (picture valence) ANOVA’s were conducted to examine group differences in startle respons es across picture valence. It was expected that the most severe groups (e.g., high BD I, high BAI, low PA, and high NA) would


32 show atypical emotion-modulation of startle responses (i.e., a failu re to exhibit the normal pattern of valence modulation). Before testing the hypothesis, we determined that in no case were there significant differences for inter-trial interval startle responses ( p s > .25). Inconsistent with our hypothesis, severity analyses for BDI, BAI, PA, and NA revealed no significant interaction effects involving group (all ps > .38). However, severity analyses for PA revealed a trend towards group by picture valence interaction ( p = .11). In fact, after controlling for medication status, results for the groups created by median splits of PA indicated a significant linear effect for picture valence by PA group [ F linear (2, 57) = 5.36, p < .05, ˆ = .16]. Follow-up paired comparisons indicated significant differences between groups for unpleasant pictures, such th at individuals with th e lowest PA scores showed the largest startle responses during unpl easant pictures as compared to healthy controls ( p < .05) and those with medium PA scores ( p < .05). Furthermore, paired sample t tests revealed the expected linear pa ttern—i.e., a difference between pleasant and unpleasant pictures [ t (1, 48) = 3.26, p < .01]—for healthy cont rols, but not for those with mid-level PA scores. Specifically, with in the moderate PA group, startle responses for pleasant and unpleasant pictures [ t (1, 52) = -1.81, p = .08] and between unpleasant and neutral pictures [ t (1, 52) = -.45, p = .66] were not significantly different. Thus, as shown in Figure 2, mood disordered persons who were moderately low in PA were the only group not to show the expected emotionmodulation by picture valence. Table 7 lists means of standardized T scores for startle magnitude according to PA group. In sum, among severity metrics, PA alone differentia ted atypical from typical emotion-modulated startle responding, but the form of the interaction was unexpected.


33 Table 7 Standardized Startle Responses Acco rding to Positive Affect Scores Group Low Moderate Controls (n = 18) (n = 28) (n = 25) Valence Mean (SD) Mean (SD) Mean (SD) Pleasant 48.91 (3.07) 49.23 (3.18) 48.54 (2.22) Neutral 49.17 (3.39) 50.30 (3.17) 50.10 (2.92) Unpleasant 53.02 (2.51) 50.65 (2.67) 50.93 (2.66) 46 47 48 49 50 51 52 53 54 LowModerateControls PA Group Pleasant Neutral Unpleasant * Figure 2. Startle magnitude according to PA group. Skin Conductance Responses The raw means for skin conductance r eactivity, including average magnitude during picture display (Mean), peak magnitude (Peak), and the average difference between peak magnitude and the baseline conduc tance prior to pictur e display (Diff) are


34 presented in Table 8. In order to normaliz e skewness and kurtosis associated with raw skin conductance values, raw data was transfor med in a linear natural log transformation. Table 8 Skin Conductance Means MDD mD Controls Valence ( n = 34) ( n = 25) ( n = 29) Mean Mean (SD) Mean (SD)Mean (SD) Pleasant .003 (.01) -.002 (.01).001 (.01) Neutral .001 (.02) -.002 (.01)-.003 (.01) Unpleasant .016 (.03) .013 (.03) .019 (.06) Peak Pleasant .013 (.02) .006 (.01) .009 (.01) Neutral .011 (.02) .005 (.01) .005 (.01) Unpleasant .033 (.05) .029 (.05) .037 (.08) Difference Pleasant .014 (.02) .008 (01) .011 (.01) Neutral .013 (.02) .007 (.01) .008 (.01) Unpleasant .034 (.05) .031 (.05) .037 (.08) *Note: Skin conductance values expressed in microsiemens ( S) A series of repeated measures ANCOVAs was performed on transformed skin conductance scores with baseline skin conductance as a covariate to determine differences between picture types across di agnostic groups. The sphericity assumption was not met for average skin conductance, peak skin conductance, or the difference between peak skin conductance and baseline Because the Greenhouse-Geisser Epsilons were low (0.53), adjustments were made using the Greehouse-Gei sser correction. Prior studies of skin conductance patterns for em otion-modulated startle paradigms led to hypotheses of a quadratic effect for skin conduc tance, such that skin conductance would


35 be equivalently high for pleasant and unpleasan t pictures, and low fo r neutral pictures (Lang, Greenwald, Bradley, & Hamm, 1993). Anal yses yielded a significant quadratic effect for average skin conductance [ F Quadratic (1.08, 85) = 6.32, p < .05, ˆ = .07], peak skin conductance [ F Quadratic (1.08, 85) = 7.14, p < .01, ˆ = .08], and difference between peak skin conductance a nd baseline skin conductance [ F Quadratic (1.06, 85) = 8.69, p < .01, ˆ = .10]. However, pairwise comparis ons indicated that skin conductance for unpleasant pictures was higher than for pl easant pictures and neutral pictures ( p s < .05). Pairwise comparisons also indicated a similar pattern for peak skin conductance ( p < .05). Lastly, the same quadratic trend wa s observed for skin conductance difference scores with all p s < .05. Regarding emotion-modulated startle di fferences between diagnostic groups, it was hypothesized that MDD individuals w ould show a significantly different (nonquadratic) pattern of skin conductance res ponses as compared to mD and healthy controls. However, differences between groups for each skin conductance variable were not statistically significant (all p s > .90), nor were there a ny picture valence by group interactions for any of the skin conductance variables ( p s > .60) or group by quadratic trends ( p s > .52). Relationships between Skin Conductance and Severity Measures Analyses conducted with each of the three severity groups were similar to those conducted for startle responses. Specificall y, three (severity group) by three (picture valence) ANOVAs were conducted to examin e group differences in skin conductance responses across picture valence. Again, it was predicted that the most severe groups would show atypical emotion-modulation of startle responses compared to healthy


36 controls and moderately severe groups. Inconsistent with th is hypothesis, there were no significant group by valence interactions for BDI ( p s > .62), BAI ( p s > .46), PA ( p s > .67), or NA ( p s > .69). Cardiac Responses Average heart rate values for each ph ase, including D1, A1, and a secondary deceleration (D2), are displayed in Figure 3. Because D2 values are not readily interpretable in this type of emotion-modul ated paradigm, analyses focused on D1 and A1 values. Mean values by diagnostic group for D1 and A1 as a deviation from baseline values are presented in Table 9. -6 -4 -2 0 2 4 6 Pleasant Neutral Unpleasant Phase MDD mD Controls Figure 3. Heart phases (D1, A1, D2) by picture valence. A series of repeated measures ANCOVAs (with baseline heart rate entered as a covariate) were conducted to examine whet her heart phase magnitudes (D1, A1) differed


37 between diagnostic groups across picture vale nce. In line with pr evious hypotheses, it was predicted that MDD individuals would experience significantly smaller absolute heart phase magnitudes (signa ling less sensory and emotiona l processing) compared to mD individuals and healthy cont rols. However, results failed to reveal a significant main effect for heart phase for group (D1: [ F (2, 65) = .24, p = .79]; A1: [ F (2, 65) = .44, p = .65]) or for group by picture valence (D1: [ F (4, 128) = .91, p = .46]; A1: [ F (4, 128) = 1.00, p = .41]). Furthermore, there were no li near by valence or quadratic by valence effects for picture valence, signaling that he art phase values did not differ significantly between picture valences. Values for D1 by pi cture valence and A1 by picture valence for each diagnostic group are displayed in Figures 4 and 5, respectively. Table 9 Cardiac Measures by Diagnostic Group MDD mD Controls Valence ( n = 26) ( n = 17) ( n = 27) D1 Mean (SD) Mean (SD) Mean (SD) Pleasant -3.19 (2.27) -3.06 (1.73) -3.09 (1.59) Neutral -2.81 (1.75) -3.04 (1.87) -2.98 (1.69) Unpleasant -3.73 (2.16) -3.68 (2.13) -4.32 (2.27) A1 Pleasant 2.80 (2.48) 4.20 (3.14) 2.79 (2.59) Neutral 2.61 (2.20) 3.49 (2.23) 3.09 (2.81) Unpleasant 1.71 (2.28) 2.62 (1.92) 1.30 (2.25) Note: D1 = initial heart rate decele ration beats/min change from baseline A1 = heart rate acceleration beats/min change from baseline


38 Relationship between Cardiac Responses and Severity The typical pattern of cardiac respondi ng to valenced pictures involves more initial heart rate d eceleration (D1) for unpleasant pictur es—indicating a greater sensory orienting response—than for pleasant pictur es and larger subsequent acceleratory responses to pleasant pictures—representi ng greater emotional processing—than for unpleasant pictures (Lang, Greenwald, Bradle y, & Hamm, 1993). In order to test the hypothesis that more severe groups would s how equally small levels of sensory and emotional processing of valenced stimuli, a series of ANCOVA’s was conducted for each heart phase variable by picture valence for each severity group, with baseline heart rate again entered as a covariate. In addition, within group lin ear trends by valence were examined to test the hypotheses that patte rns of responding would differ across severity groups. Specifically, results of the ANCOVA failed to reveal signi ficant group by picture valence effects for BDI group (D1: [ F (4, 128) = 2.21, p = .08]; A1: [ F (4, 128) = 1.45, p = .22]), for BAI group (D1: [ F (4, 128) = 1.10, p = .36]; A1: [ F (4, 128) = 1.18, p = .32]), or for NA group (D1: [ F (4, 128) = 1.35, p = .26]; A1: [ F (4, 128) = 1.86, p = .12]). However for PA group there were the hypothesi zed group by picture valence interactions for D1 [ F (4, 128) = 4.17, p < .01, ˆ = .11] and A1 [ F (4, 128) = 3.33, p < .05, ˆ = .09]). Mean cardiac values according to PA gr oup are listed in Table 10. Although group by valence interactions for D1 were signifi cant (Figure 5), follow-up paired comparisons indicated only trend level di fference for D1 following unplea sant picture onset between those with low levels of PA and healthy controls ( p = .09), such that individuals with low PA showed higher D1 values. Additionally, th ere was a trend for i ndividuals with low


39 levels of PA to have higher A1 values duri ng unpleasant pictures as compared to healthy controls ( p = .08; Figure 6). Table 10 Cardiovascular Measures by Positive Affect Low Moderate Controls Valence ( n = 19) ( n = 22) ( n = 27) D1 Mean (SD) Mean (SD) Mean (SD) Pleasant -3.51 (2.49) -2.75 (1.64) -3.09 (1.59) Neutral -2.59 (1.49) -3.09 (2.03) -2.98 (1.69) Unpleasant -3.20 (1.63) -4.20 (2.49) -4.32 (2.27) A1 Pleasant 2.46 (2.11) 4.11 (3.25) 2.79 (2.59) Neutral 2.64 (2.31) 3.18 (2.26) 3.09 (2.81) Unpleasant 2.31 (2.22) 1.73 (2.19) 1.30 (2.25) Note: Values are change in hear t rate (in beats per minute) compared to baseline heart rate. -4.5 -4 -3.5 -3 -2.5 -2 Pleasant NeutralUnpleasant Valence MDD mD Controls Figure 4. D1 cardiac responses by diagnos tic group for each picture valence.


40 1 1.5 2 2.5 3 3.5 4 4.5 Pleasant NeutralUnpleasant Valence MDD mD Controls Figure 5. A1 cardiac responses by diagnos tic group for each picture valence. Lastly, individuals with moderate PA le vels had significantly higher A1 values compared to healthy controls ( p < .05) and compared to indivi duals with low levels of PA ( p < .05; Figure 7). Only the group differen ce in A1 for unpleasant pictures survived covariation of medication stat us and remained significant ( p < .05) In sum, these results suggest that compared to healthy controls mood disordered individuals with low PA showed smaller sensory orie nting responses and greater emotional processing during unpleasant pictures. Furthermore, results indi cate that mood disordered persons with moderate levels of PA experienced signi ficantly more emotional processing during pleasant pictures as compared to mood disord ered individuals with low PA and healthy controls.


41 -4.5 -4 -3.5 -3 -2.5 -2 PleasantNeutralUnpleasant Valence Low Moderate Controls Figure 6. D1 Cardiac responses by PA group for each picture valence. 1 1.5 2 2.5 3 3.5 4 4.5 PleasantNeutralUnpleasant Valence Low Moderate Controls Figure 7. A1 Cardiac responses by PA group for each picture valence.


42 Discussion Major depressive disorder is define d primarily by mood changes. However, research has not yet discerned how this chronic mood disturbance affects emotional responding. Furthermore, it is unclear whether subthreshold forms of depression, such as minor depression (mD), involve similar or different abnormalities in emotional responding as seen in MDD. The general aim of the current study was to examine abnormalities in emotional reactiv ity as a vulnerability and se verity marker for MDD. It was predicted that MDD individuals woul d display abnormal (blunted) patterns of emotional responding compared to healthy controls, as we ll as mD individuals. More specifically, this study was the first to examin e the relationship between emotional reactivity to pleasant and unpleas ant stimuli in a sample of carefully diagnosed MDD individuals, mD individuals, and healt hy controls. The primar y goal of the study was to discern diagnostic differences in em otion reactivity as a function of affective foreground, by utilizing the emotion-modulated st artle eyeblink reflex. To clarify the role of severity in emotional reactivity, we analyzed startle to affective pictures as a function of several severity metrics, including de pression severity, anxiety severity, and the strength of positive and negative mood states. To afford a more comprehensive assessment of psychophysiological responding, para llel analyses were conducted on heart rate and skin conductance.


43 Effects of the Paradigm Several aspects of the results increase c onfidence that that the emotion-modulated startle paradigm was successfully applied. Results yielded the expected pattern of emotion modulation for startle reactivity. Standardized star tle responses during negative pictures were significantly larger than startle resp onses occurring during pleasant pictures. Moreover, ratings of picture valence indicated th at participants found unpleasant pictures significantly less pleasant than neutral and pleasant pi ctures. Additionally, participants found pleasant pict ures to be equally arousi ng as unpleasant pictures, and both to be more arousing than neutral pictur es. Skin conductance measures also displayed the expected quadratic pattern, such that measures were highest during unpleasant pictures and lowest for neutral and pleasant pi ctures. Lastly, heart rate variables showed the expected quadratic waveform across three phases of cardiac response. Taken together, these findings indicate that the manipulati on of affective state produced the expected pattern of responses, suggesting that the in ternal validity of the paradigm was sound. Effects of Diagnosis and Symptom Severity The primary hypotheses predicted that M DD individuals would show less valence modulation when compared to mD individua ls and controls ac ross all physiological variables. However, analyses did not re veal significant group by picture valence interactions for any physiological variable of interest. In other words, in these data persons with MDD, as well as persons with mD, appeared to exhibit essentially normal valence modulation of startle, skin conductance, and heart rate responses. It is unclear why MDD partic ipants did not exhibit a lack of appropriate valence modulation across physiological indi cators, especially in light of two recent studies which


44 found blunted startle modulation in MDD (D ichter & Tomarken, 2008; Dichter et al., 2004). However, previous research on the startle in MDD is not entirely uniform. In fact, two other previous studies found essentially no rmal emotion-modulated startle patterns in MDD individuals relative to he althy controls, and only abnor mal startle modulation when depressive subgroups were analyzed (Allen et al., 1999; Kaviani et al., 2004). We anticipated this second possibility by pe rforming a wide range of subgroup analyses focusing on several metrics of symptom severity. More specifically, secondary analyses were conducted with various severity measures to test the hypothesis that th e most severe group of mood-disordered individuals would show the clear est pattern of atypical (flatt ened) patterns of emotional responding across physiological variables. Resu lts utilizing Beck Depression Inventory (BDI) scores, Beck Anxiety Inventory (BAI) sc ores, and negative affe ct scores (NA), as measured by the Positive Affect Negative Aff ect Scale (PANAS) as measures of severity were inconsistent with this hypothesis, and did not yield significant differences between severity groups. However, analyses conducted with positive affect (PA), as assessed by the PANAS, did predict differences in emoti on-modulated startle re sponding as well as heart rate responses to affec tive pictures. These findings were novel in the li terature and merit comment. Specifically, mood disordered individuals w ith the lowest PA exhibited the largest startle responses to unpleasant pictures of the three severity groups. Thus, these individuals displayed increased reactivity for unpleasant pi ctures. Surprisingly, it was those mood disordered individuals with mode rately low PA (and not the most severe group) that failed to show the typical pattern of emotion mo dulation for startle exhibited


45 by healthy persons. PA also differentiate d patterns of cardiac responding between severity groups. Mood disordered individua ls with the lowest PA showed atypical patterns of cardiac responses. The typical pattern of cardiac res ponding involves initial deceleration of heart rate in response to stimuli (D1), i ndicative of initial sensory orienting, and subsequent acceleration (A1), indicative of emotional processing (Lacey & Lacey, 1970; Graham & Clifton, 1966; Sokolov, 1963). In terms of emotion modulation, the typical pattern of heart rate responses involves heightened se nsory orientation for unpleasant compared to pleasant pictures and heightened emotional processing of pleasant as compared to unpleasant pict ures (Lang, Greenwald, Bradley, & Hamm, 1993). Mood disordered individuals with low PA showed equivalent sensory orientation for and emotional processing of both unpleas ant and pleasant pictures. This suggests deficits in attending to emotional stimuli, such that the most severe group of mood disordered individuals exhibited decreased sensory orienting for unpleasant pictures, decreased emotional processi ng of pleasant stimuli, and in creased emoti onal processing of unpleasant stimuli, resulting in a flatte ned pattern of heart rate responding across valence conditions (Figures 6 and 7). Past researchers have proposed models of emotional processing that are somewhat consis tent with this pattern of biased attending towards negative stimuli (Beck, 1967). Affect and Hedonic Capacity Given past research in this area (e.g., Allen, et al., 1999) it is perhaps surprising that PA alone, and not BDI or NA, scores exerted influence on startle modulation and cardiac responses. However, there was a significant association between positive affect and startle responding to unpleasant pictures. A lack of PA as indicated by the PANAS,


46 can be conceptualized as representing a dimensional measure of anhedonia (Clark & Watson, 1991) or, similarly, impaired hedonic capacity (Meehl, 1962). Specifically, as anhedonia is defined by a lack of interest or pleasure, a lower score on the PA scale of the PANAS indicates higher levels of impaired hedonic capacity. Previous studies have found an effect for anhedonia on startle m odulation, such that anhedonic individuals showed blunted emotional responding, as marked by a lack of startle modulation to both pleasant and unpleasant pictures (Kaviani et al., 2004; Lars on, Nitschke, & Davidson, 2007).We did not replicate this fl attened pattern of startle re sponses for persons with low PA. Instead, individuals with the lowest PA displayed higher startle responses during unpleasant pictures than did individuals with moderate PA and healthy controls. Individuals with the lowest PA also e xhibited increased emo tional processing of unpleasant pictures (as indicated by cardiac va riables). Therefore, increased emotional processing of unpleasant stimuli may be relate d to a weak appetitiv e system, which leads to greater cardiac accelerati on and higher levels of startle responding during unpleasant pictures. In general, PA and NA have often been conceptualized as two independent affective dimensions (Tellegen, Watson, & Clark, 1999). However, NA and PA have been theorized to be recipro cally related in some instan ces of disturbed mood (Meehl, 1962). For example, the theory of aversive drif t—first proposed as an explanation for the marked anhedonia seen in schizophrenia disord ers—predicts that as the ability to feel pleasure decreases along the dimension of hedonic capacity, approaching anhedonia, individuals begin paying more attention to objects in the environment with negative affective tone (Meehl, 1962; M eehl, 2001). Consistent with this interpretation, in this


47 dataset we found that reports of PA and NA we re moderately negatively intercorrelated ( r = -.4). Thus, one potentially impor tant finding of this study is that it suggests that more severe hedonic deficits in mood disorders resu lt in higher levels of emotional processing for unpleasant stimuli. Future studies should further examine this association between greater emotional processing of unpleasant s timuli and dimensional hedonic capacity, and whether greater emotional processing of negative stimuli predicts the course of depression. Conclusions The current study did not re veal the expected differen ces in emotional responding between major and minor depression. These null findings for diagnostic group differences do not allow us to decide be tween the continuum and the disease model (whether MDD represents a dis tinct mood state with chronic, disease-like deficits in emotional responding or whether it is an arbi trary point along the c ontinuum of disturbed mood). However, this study does suggest that qualitative features of depression such as symptom type (e.g., decreased hedonic capacity ) may help to explain why findings for diagnostic group are often heterogeneous in the area of emotion. Finally, there is relatively little resear ch concerning the pr edictive ability of emotional responding in MDD and no research on mD. Given that some evidence indicates that anhedonia predicts a more recurrent course of MDD (Clark et al, 1984) and nonrecovery of MDD (Kasch et al., 2002), examination of the predictive validity of anhedonia severity for the spectrum of mood di sorders is warranted. Given that the lack of positive affect seemed to be indicative of more severe de ficits in emotional processing,


48 future studies should examine whether these de ficits in emotional processing predict the development of or recove ry from mood disorders.


49 References Allen, N.B., Trinder, J. & Brennan, C. (1999) Affective startle modulation in clinical depression: Preliminary findings. Biological Psychiatry, 46 542-550. American Psychiatric Association. (1994) Diagnostic and sta tistical manual of mental disorders (4th ed.). Washington, DC: Author. Arborelius, L., Owens, M.J., Plotsky, P.M ., & Nemeroff, C.B. (1999). The role of corticotropin-releasing factor in depression a nd anxiety disorders. Journal of Endocrinology, 160 1-12. Beck, A.T., (1967). Depression: Clinical, experime ntal, and theoretical aspects. New York: Hoeber. Beck, A.T., Epstein, N., Brow n, G. & Steer, R.A. (1988). An inventory for measuring clinical anxiety: Psyc hometric properties. Journal for consulting and Clinical Psychology, 56 893-897. Beck, A.T., Steer, R.A., Ball, R., & Ranier i, W.F. (1996). Comparison of the Beck Depression Inventories-IA and – II in psychiatric outpatients. Journal of Personality Assessment, 67 588-797. Beck, A.T., Steer, R.A., & Brown, G.K. (1996). BDI-II Manual San Antonio, TX: The Psychological Corporation. Beekman, A. T. F., Deeg, D. J. H., van Tilbur g, T., Smit, J.H., Hooije r, C., van Tilberg, W. (1995). Major and minor depression in later life: A study of prevalence and risk factors. Journal of Affective Disorders, 36 65-75.


50 Berenbaum, H. & Oltmanns, T.E. (1992). Emotional experience and expression in schizophrenia and depression. Journal of Abnormal Psychology, 101 37-44. 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 1-15. Bradley, M.M., Codispoti, M., Cuthbert, B.N., & Lang, P.J. (2001). Emotion and motivation I: Defensive and appetiti ve reactions in picture processing. Emotion, 1 276-298. Bradley, M. M., Cuthbert, B. N., & Lang, P.J. (1990). Startle refl ex modification: Emotion or attention? Psychophysiology, 27 513-522. Bradley, M. M., Cuthbert, B. N. & Lang, P. J. (1993). Pictures as prepulse: Attention and emotion in Startle modification. Psychophysiology, 30 541-545. Bradley, M.M. & Lang, P.J. (2000). Affec tive reactions to acoustic stimuli. Psychophysiology, 37 204-215. Bylsma, L.M., Morris, & Rottenberg, J.R. (i n press). A Meta-Analysis of Emotional Reactivity in Major Depressive Disorder. Clinical Psychology Review. Cassano, G.B., Rucci, P., Frank, E., Fagiolini, A., Dell’Osso, L., Shear, M.K. et al. (2004). The mood spectrum in unipolar a nd bipolar disorder: Arguments for a unitary approach. American Journal of Psychiatry, 161 1264-1269. Center for the Study of Emotion and Attention [CSEA-NIMH] (1999). The international affective pictur e system: Digitized photographs. Gainesville, FL: The Center for Resear ch in Psychophysiology, University of Florida.


51 Clark, L. A., & Watson, D. (1991). Triparti te model of anxiety and depression: Psychometric evidence and taxonomic implications. Journal of Abnormal Psychology, 100(3) 316-336. Cook, E.W. III, Atkinson, L. & Lang, K.G. ( 1987). Stimulus control and data acquisition for IBM PCs and compatibles. Psychophysiology, 24 726-727. Cook, E.W., Hawk, L.W., Davis, T.L., & St evenson, V.E. (1991). Affective individual differences and startle reflex modulation. Journal of Abnormal Psychology, 100 5-13. Coyne, J. (1993). Two views of depressi on: A Just-So Theory of Depression. PsycCRITIQUES, 38 947-948. Crawford, J.R. & Henry, J.D. (2004). The Positive and Negative Affect Schedule (PANAS): Construct validity, measurement properties and normative data in a large, non-clinical sample. British Journal of Clinical Psychology, 43 245-265. Cuijpers, P., de Graaf, R., & van Dorsselaer. (2004). Minor depressi on: Risk profiles, functional disability, health care use a nd risk of developing major depression. Journal of Affective Disorders, 79 71-79. Cuthbert, B.N., Lang, P.J., & Strauss, C. (2003). The psychophysiology of anxiety disorder: Fear memory imagery. Psychophysiology, 40 407-422. Dichter, G.S. & Tomarken, A.J. (2008). The chronometry of affective startle modulation in unipolar depression. Journal of Abnormal Psychology, 117 1-15. Dichter, G.S., Tomarken, A.J., Shelton, R.C. & Sutton, S.K. (2004). Earlyand late-onset startle modulation in unipolar depression. Psychophysiology, 41 433-440.


52 Drevets, W. C. & Raichle, M. E. (1998). Reciprocal suppression of regional blood flow during emotional versus higher cognitive pr ocesses: Implications for interactions between emotion and cognition. Cognition & Emotion,12(3) 353-385. Dyck, M.J., Jolly, J.B., & Kramer, T. ( 1994). An evaluation of positive affectivity, negative affectivity, and hyperarousal as markers for assessing between syndrome relationships. Personality and Individual Differences, 17 637-646. First, M.B., Spitzer, R.L., Gibbon, M., and Williams, J.B.W.: Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition With Psychotic Screen (SCID-I/P W/ PSY SCREEN) New York: Biometrics Research, New York State Psyc hiatric Institute, November 2002. Fogel, J., Eaton, W.W., & Ford, D.E. (2006). Mi nor depression as a pred ictor of the first onset of major depressive disord er over a fifteen-year follow-up. Acta Psychiatrica Scandinavica, 113 36-43. Forbes, E.E., Miller, A. Cohn, J.F. Fox, N. A. & Kovacs, M. (2005). Affect-modulated startle in adults with childhood-onset depr ession: Relations to bipolar course and number of lifetime depressive episodes. Psychiatry Research, 134 11-25. Gehricke, J.G. & Shapiro, D. (2000). Reduced facial expression and social context in major depression: Discrepancies between f acial muscle activity and self-reported emotion. Psychiatry Research, 95 157-167. Golin, S. Hartman, S.A., Klatt, E.N., Munz, K. & Wolfgang, G.L. (1977). Effects of selfesteem manipulation on arousal and reacti ons to sad models in depressed and nondepressed college students. Journal of Abnormal Psychology, 86 435-439.


53 Gonzlez-Tejera, G., Canino, G., Ramrez, R ., Chvez, L., Shrout, P., Bird, H., Bravo, M., Martnez-Taboas, A., Ribera, J., & Ba uermeister, J. (2005). Examining minor and major depression in adolescents. Journal of Child Psychology and Psychiatry, 46 888-899. Graham, F. K. & Clifton, R. K. (1966). Hear t-rate as a component of the orienting response. Psychological Bulletin, 65(5) 305-320. Grillon, C. & Baas, J. (2003). A review of the modulation of the startle reflex by affective states and its application to psychiatry. Clinical Neurophysiology, 114 15571579. Hamm, A.O., Cuthbert, B.N., Gl obisch, J. & Vaitl, D. (1997). Fear and the startle reflex: Blink modulation and autonomic response patterns in animal and mutilation fearful subjects. Psychophysiology, 34, 97-107. Hankin, B.L., Fraley, R.C. & Lahey, B.B. (2005). Is Depression Best Viewed as a Continuum or Discrete Category? A taxometric analysis of childhood and adolescent depression in a population-based sample. Journal of Abnormal Psychology, 114 96-110. Hermens, M.L.M., van Hout, H.P.J., Terlui n, B., van der Windt, D.A., Beekman, A.T.F., van Dyck, R., & de Haan, M. (2004). Th e prognosis of minor depression in the general population: A systematic review. General Hospital Psychiatry, 26 453462. Hodges, R.L., Cook, E.W. & Lang, P.J. (1985) Individual differences in autonomic response: Conditioned association or conditioned fear? Psychophysiology, 22 545-560.


54 Jolly, J.B., Dyck, M.J., Kramer, T.A., & Whe rry, J.N. (1994). Integration of positive and negative affectivity and cogni tive content specificity: Im proved discrimination of anxious and depressive symptoms. Journal of Abnormal Psychology, 103 544552. Judd, L.L. (1997). The clinical Course of unipolar Major Depressive Disorders. Archives of General Psychiatry, 54 989-991. Judd, L.L., Akiskal, H.S., Maser, J.D., Zeller, P.J., Endicott, J., Coryell, W., Paulus, M.P., Kunovac, J.L. Leon, A.C., Mueller, T.I., Rice, J.A. & Keller, M.B. (1998). A prospective 12 year study of subs yndromal and syndromal depressive symptoms in unipolar major depressive disorders. Archives of General Psychiatry, 55 694-700. Judd, L. L., Akiskal, H. S. & Paulus M. P., (1997). The role of clinical significance of subsyndromal depression symptoms (SSD) in unipolar major depressive disorder. Journal of Affective Disorders, 45(1 Kaviani, H., Gray, J.A., Checkley, S.A ., Wilson, G.D. & Kumari, V. (2004). Affective modulation of the startle response in depression, anhedonia, and anxiety. Journal of Affec tive Disorders, 83 21-31. Keller, M.B., Lavori, P.W., Mueller, T.I ., Endicott, J., Coryell, W., Hirschfeld, R.M.A., & Shea, T. (1992). Time to recovery, chronicity, and levels of psychopathology in major depression: A 5-year prospective follow-up of 431 subjects. Archives of General Psychiatry, 49, 809-816.


55 Kessler, R.C., Zhao, S., Blazer, D.G., & Swar tz, M. (1997). Prevalence, correlates, and course of minor depression and major depression in the national comorbidity survey. Journal of Affective Disorders, 45 19-30. Koch, M. & Schnitzler, H.U. (1997). The ac oustic startle respons e in rats—circuits mediating evocation, inhi bition, and potentiation. Behavioural Brain Research, 89 35-49. Koch, M., Schmid, A. & Schnitzler, H.U. (1996). Pleasure attenuation of the startle response is disrupted by 6-hydroxydopamine lesion of the nucleus accumbens. Neuroreport, 7 1442-1446. Lacey, J.I. & Lacey, B.C. (1970). Some autonomic central nervous system interrelationships. In P. Black (Ed). Physiological correlates of emotion (pp. 205227). New York: Academic Press. Landis, C. & Hunt, W.A. (1939). The startle pattern New York: Farrar and Company, Inc. Lang, P.J. (1980). Behavioral treatment a nd bio-behavioral assessment: Computer applications. In J.B. Sidowski, J.H. Johnson & T.A. Williams (Eds). Technology in mental health care delivery systems (pp.119-137). Norwood, N.J.: Ablex. Lang, P.J. (1995). The emotion probe: St udies of motivation and attention. American Psychologist, 50 372-385. Lang, P.J., Bradley, M.M., & Cuthbert, B.N. (2005). International affective picture system (IAPS): Affective ratings of pictures and instruction manual. Technical Report A-6. University of Florida, Gainesville, FL.


56 Lang, P.J., Bradley, M.M. & Cuthbert, B.N. (1998). Emotion, motivation, and anxiety: Brain mechanisms and psychophysiology. Biological Psychiatry, 44 1248-1263. Lang P. J., Greenwald, M. K., Bradley, M. M. & Hamm, A.O. (1993). Looking at pictures: Affective, facial, vis ceral, and behavioral reactions. Psychophysiology, 30 261-273. Larsen, D.K., Norton, G.R., & Walker, J.R. ( 2002). Analysis of startle responses in patients with panic disorder and social phobia. Cognitive Behaviour Therapy, 31 156-169. Larson, C.L., Ruffalo, D., Nietert, J.Y., & Da vidson, R.J. (2000). Temporal stability of the emotion-modulated startle response. Psychophysiology, 37 92-101. Larson, C.L., Nitschke, J.B., & Davidson, R.J. (2007). Common and dis tinct patterns of affective response in dimensi ons of anxiety and depression. Emotion, 7 182-191. Lewinsohn, P.M., Lobitz, W.C. & Wilson, S. (1973). “Sensitiv ity” of depressed individuals to aversive stimuli. Journal of Abnormal Psychology, 81 259-263. Meehl, P.E. (2001). Primary and secondary hedonia. Journal of Abnormal Psychology, 110, 188-193. Meehl, P.E. (1962). Schizotaxi a, schizotypy, schizophrenia. American Psychologist, 17, 827-838. Miller, M.W. & Litz, B. T. (2004). Emotionalprocessing in Posttrau matic stress disorder II: Startle reflex modulati on during picture processing. Journal of Abnormal Psychology, 113 451-463.


57 Morgan, C.A., Grillion, C., Southwick, S. M., & Charney, D.S. (1996). Exaggerated acoustic startle reflex in Gulf war veterans with posttraumatic stress disorder. American Journal of Psychiatry, 153 64-68. Mueller, T.I., Keller, M.B., Leon, A.C., So lomon, D.A., Shea, M.T., Coryell, W., & Endicott, J.(1996). Recovery after 5 years of unremitting major depressive disorder. Archives of General Psychiatry, 53 794-799. Murray, C.J. & Lopez, A.D. (1997). Global mort ality, disability, and the contribution of risk factors: Global Bu rden of Disease Study. Lancet, 349 1436-1442. Newman, S.C., Sheldon, C. T. & Bland, R. C. (1998). Prevalence of depression in an elderly community sample: A comparison of GMS-AGECAT and DSM-IV diagnostic criteria. Psychological Medicine, 28 1339-1345. Ottowitz, W.E., Dougherty, D.D. & Savage, C. R. (2002). The neural network basis for abnormalities of attention and executive f unction in major depressive disorder: Implications for application of the medical disease model to psychiatric disorders. Harvard Review of Psychiatry, 10, (86-99). Patrick, C.J., Bradley, M.M., & Lang, P.J. (1993). Emotion in the criminal psychopath: Startle Reflex Modulation. Journal of Abnormal Psychology, 102 82-92. Prisciandaro, J.J. & Roberts, J.E. (2005) A taxometric investigation of unipolar depression in the national comorbidity survey. Journal of Abnormal Psychology, 114 718-728. Rapaport, M.H., Judd, L.L., Schettler, P.J., Yonkers, K.A., Thase, M.E., Kupfer, D.J., Frank, E., Plewes, J.M., Tollefson, G.D ., & Rush, A.J. (2002). A descriptive analysis of minor depression. American Journal of Psychiatry, 159 637-643.


58 Rosenberg, E. L. (1998). Levels of anal ysis and the organization of affect. Review of General Psychology, 2(3) 247-270. Rottenberg, J. (2005). Mood and em otion in major depression. Current Directions in Psychological Science, 14 167-170. Rottenberg, J., Gross, J.J., & Gotlib, I.H. ( 2005). Emotion context insensitivity in Major Depression Disorder. Journal of Abnormal Psychology, 114 627-639. Rottenberg, J., Kasch, K. L., Gross, J. J., & Gotlib, I. H. (2002). Sadness and amusement reactivity differentially predict concurre nt and prospective functioning in major depressive disorder. Emotion, 2 135-146. Rottenberg, J., Wilhelm, F.H., Gross, J.J ., & Gotlib, I.H. (2003). Vagal rebound during resolution of tearful crying among depr essed and nondepressed individuals. Psychophysiology, 40 1-6. Schlenker, R., Cohen, R., & Hopmann, G. ( 1995). Affective modula tion of the startle reflex in schizophrenic patients. European Archives of Psychiatry & Clinical Neuroscience, 245 309-318. Skre, I., Onstad, S., Torgersen, S. & Kringlen, E. (1991). High interrater reliability for the Structured Clinical Interview for DSM-III-R Axis I (SCID-I). Acta Psychiatrica Scandinavica, 84 167-73. Sloan, D.M., Strauss, M.E., Quirk, S.W., & Satajovic, M. (1997). Subjective and expressive emotional responses in depression. Journal of Affec tive Disorders, 46 135-141. Sloan, D.M., Strauss, M.E., & Wisner, K.L. (2001). Diminished response to pleasant stimuli by depressed women. Journal of Abnormal Psychology, 110 488-493.


59 Sokolov, E. N. (1963). Orienting re flex as a cybernetic system. Zhurnal Vysshei Deyatel’nosti, 13(5) 816-830. Somsen, R. J., Van der Molen, M. W. & Orle beke. (1983). Phasic he art rate changes in reaction time, shock avoidance and una voidable shock tasks: Are hypothetical generalizations about differe nt S1-S2 tasks justified? Psychophysiology, 20(1) 88-94. Steer, R. A., Ranieri, W. F. & Beck, A. T. (1993). Further evidence fo r the validity of the beck anxiety inventory with psychiatric outpatients. Journal of Anxiety Disorders, 7(3) 195-205. Sweeney, P. D., Anderson K. & Bailey, S. (1986 ). Attributional styl e in depression: A meta-analytic review. Journal of Personality and Social Psychology, 50(5) 974991. Taiminen, T., Jaakelainen, S., Ilonen, T., Meye r, H., Karlsson, H., Lauerma, H., et al. (2000). Habituation of the blink reflex in first episode schizophrenia, psychotic depression, and non-psychotic depression. Schizophrenia Research, 44 69-79. Tellegen, A., Watson, D., & Clark, L. A. (1999) On the dimensional and hierarchical nature of affect. Psychological Science, 10, 297-303. Vrana, S.R. & Lang, P.J. (1990). Fear im agery and the startle-probe reflex. Journal of Abnormal Psychology, 99 189-197. Vredenburg, Flett, G.L. & Krames, L. ( 1993). Analog versus clinical depression: A clinical reappraisal. Psychological Bulletin, 113 327-344.


60 Watson, D., Clark, L.A., & Tellegen, A. ( 1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 47 1063-1070. Wexler, B.E., Levenson, L., Warrenburg, S., & Pr ice, L.H. (1994). Decreased perceptual sensitivity to emotion-evoking stimuli in depression. Psychiatry Research, 51 127-138. Zanarini, M.C. & Frankenburg, F.R. (2001). Atta inment and maintenance of reliability of axis I and axis II disorders over th e course of a longitudinal study. Comprehensive Psychiatry, 42 369-374. Zanarini, M.C., Skodol, A.E., Bender, D., Do lan, R., Sanislow, C., Schaefer, E., Morey L.C., Grilo, C.M., Shea, M.T., McGlashan, T.H. & Gunderson, J.G. (2000). The collaborative longitudinal personality diso rders study: Reliability of axis I and II diagnoses. Journal of Personality Disorders, 14 291-9.


61 Appendices


62 Appendix A: Footnotes 1IAPS numbers of pictures used ar e as follows: neutral (2570, 2580, 7006, 7009, 7025, 7030, 7150, 7175, 7187, 7217, 7224, 7491 for males; and 2840, 5530, 5740, 6150, 7004, 7010, 7031, 7035, 7040, 7185, 7491, 9360 for females) pleasant (1650, 4320, 4653, 4689, 7501, 8080, 8180, 8260, 8300, 8380, 8470, 8501 for males; 4660, 5460, 5910, 7502, 8030, 8034, 8080, 8180, 8185, 8200, 8210, 8400 for females), and unpleasant (3015, 3053, 3060, 3071, 3080, 3170, 3530, 6260, 6313, 6570, 9410, 9570 for males; 2730, 3010, 3015, 3053, 3060, 3100, 3120, 6312, 9050, 9433, 9571, 9921 for females.


63 Appendix B: Additional Tables Zero Order Correlations between Startl e Responses and Severity Measures Magnitude T scores BDI BAI PA NA Pleasant 0.05 -0.05 -0.06 -0.04 Neutral 0.02 0.16 0.15 0.07 Unpleasant 0.17 -0.04 -0.25* 0.02 Raw Scores Pleasant 0.05 -0.02 0.03 0.16 Neutral 0.03 -0.00 0.11 0.17 Unpleasant 0.07 -0.01 0.01 0.18 Amplitude Pleasant 0.04 -0.07 -0.04 -0.05 Neutral 0.02 0.17 0.13 0.08 Unpleasant 0.19 0.00 -0.27* 0.07 *p < .05 Note: BDI = Beck Depression Inventory BAI = Beck Anxiety Inventory PA = Positive Affect NA = Negative Affect


64 Appendix B: Additional Tables (Continued) Zero Order Correlations between Skin Conductance and Severity Measures Valence Mean BDI BAI PA NA Pleasant 0.02 -0.04 0.09 0.13 Neutral 0.16 0.09 -0.05 0.26* Unpleasant -0.09 -0.07 -0.04 -0.05 Peak Pleasant 0.05 0.01 -0.00 0.14 Neutral 0.13 0.13 -0.12 0.26* Unpleasant -0.09 -0.05 -0.03 -0.06 Difference Pleasant 0.07 -0.03 -0.00 0.15 Neutral 0.14 0.14 -0.12 0.26* Unpleasant -0.08 -0.04 -0.04 -0.06 p < .05 Note: Skin conductance values expressed in microsiemens ( S) BDI = Beck Depression Inventory BAI = Beck Anxiety Inventory PA = Positive Affect NA = Negative Affect


65 Appendix B: Additional Tables (Continued) Zero Order Correlations between HR Phase and Severity Measures Valence D1 BDI BAI PA NA Pleasant 0.12 -0.04 0.07 -0.15 Neutral 0.03 0.04 -0.03 -0.05 Unpleasant 0.04 0.04 -0.11 -0.10 A1 Pleasant 0.02 0.03 0.06 0.06 Neutral -0.01 -0.06 0.02 0.15 Unpleasant 0.13 0.12 -0.11 0.16 Note: BDI = Beck Depression Inventory BAI = Beck Anxiety Inventory PA = Positive Affect NA = Negative Affect