Effects of Anticipatory Anxiety on Emotion Processing

Effects of Anticipatory Anxiety on Emotion Processing

Dissertation

zur Erlangung des akademischen Grades des Doktors der Naturwissenschaften (Dr. rer. nat.)

an der Universität Konstanz

Mathematisch-Naturwissenschaftliche Sektion Fachbereich Psychologie

Vorgelegt von

Florian Bublatzky Konstanz, September 2009

Tag der mündlichen Prüfung: 30.11.2009 1. Referent: Prof. Dr. Harald T. Schupp 2. Referentin: Prof. Dr. Johanna Kissler

Konstanzer Online-Publikations-System (KOPS) URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-98479

URL: http://kops.ub.uni-konstanz.de/volltexte/2009/9847/

We are highly adaptive creatures. The predictable becomes, by definition, background, leaving the attention uncluttered, the better to deal with the random or unexpected.

(Ian McEwan, Enduring Love)

Table of Contents

Acknowledgements IV

Summary V

Zusammenfassung VIII

Chapter 1 11

General Introduction

Chapter 2 49

Additive Effects of Threat-of-shock and Picture Valence on Startle Reflex Modulation

Chapter 3 61

The Interaction of Anticipatory Anxiety and Emotional Picture Processing: An Event- Related Brain Potential Study

Chapter 4 89

Pictures Cueing Threat: Brain Dynamics in Viewing Explicitly Instructed Danger Cues

Chapter 5 112

General Discussion

References

List of Contributions

Acknowledgements

I am deeply grateful to many friends and colleagues for their multifaceted support and long-term collaboration. It goes beyond the scope of the acknowledgements to mention all of them and all they did. However, I would like to highlight …

… my supervisor and mentor Harald Schupp. I owe him a special dept for his guidance in the depths of scientific work, and the plenty of support and challenges.

… my actual and former colleagues Jessica Stockburger, Ralf Schmälzle, Tobias Flaisch, Peter Peyk, Margarita Stolarova, Frank Häcker, Alexander Barth, Ines Krug, and Sabine Widmann-Schmid for their support in countless specific and general affairs.

… Beate Hofer for acting “Dr. Mabuse” … and thanks to all those who suffered from the anticipation of threat for scientific purposes.

… my Spanish friends and colleagues Mamen Pastor, Pedro Guerra and Jaime Vila (et al., 2008) for sharing the joy of hard scientific work in combination with simply having a lot of fun … muchísimas gracias a vosotros.

… the DFG, LGFG and DAAD for providing the necessaries for such a work.

Besides work, I would like to express my very personal thanks to Nils, Jessy, Ralf, little Sunshine, Pedro, Freda and Tabea … for simply being like you are.

And finally my family, to whom this is dedicated.

Thank you.

Summary

The present thesis examined the impact of anticipatory anxiety on emotion processing. For this purpose, two well-established research lines were integrated: The

“threat-of-shock” and “picture viewing” paradigm. The exposure to pictorial stimuli has been shown to be associated with perceptual, physiological and behavioral changes varying as a function of hedonic picture valence and arousal. Based on an evolutionary perspective, it is assumed that evolved motivational systems – mediating approach and avoidance – selectively facilitate the perceptual processing of high-significant environmental information in order to prepare and execute fast and appropriate actions in a given situation. Furthermore, in favor of dynamic adjustment to phylogenetic and ontogenetic core objectives, these systems easily acquire contingencies of environmental conditions and events. In this context, the threat-of-shock paradigm has been shown to activate the defensive motivational system by means of verbal instructions about aversive environmental contingencies. That is, merely the verbal announcement about imminent danger is sufficient to elicit the defensive state of anticipatory anxiety.

However, little is known about the impact of aversive anticipation on emotional perception and responding. To investigate this research question two aspects of main importance were pursued: (1) The predictive value of picture materials for threat of aversive events (i.e., pictures were either unrelated or predictive for threat-of-shock), and (2) the focus of processing stage (i.e., measurement of perceptual-evaluative processing vs.

physiological response preparation and motor output).

Study I examined the impact of sustained threat signals on psychophysiological responses (defensive startle reflex and skin conductance) to picture materials which were unrelated to threat-of-shock conditions. To this end, pleasant, neutral, and unpleasant pictures were presented as a continuous stream during contextually signaled threat-of-

shock and safety. Regarding hedonic picture valence and threat-of-shock manipulation previous findings were replicated. Of main interest, anticipating aversive events and viewing unrelated affective pictures additively modulated defensive activation.

Specifically, despite overall potentiated startle blink magnitude during threat-of-shock conditions, the startle reflex remained sensitive to hedonic picture valence. In addition, changes in skin conductance level and self-report data support the notion of an enhanced aversive state under conditions of instructed threat-of-shock. Overall, defensive activation by physical threat appeared to operate independently from reflex modulation by unrelated picture media.

Study II focused on perceptual and attentional processes mutually involved in anticipatory anxiety and affective picture processing. Utilizing an experimental protocol similar to Study I, the main finding was that pleasant pictures mismatching the current aversive context elicited a sustained negative difference potential over visual processing regions. In contrast, unpleasant and neutral picture processing did not vary as a function of threat-of-shock. Furthermore, in both safety and threat-of-shock conditions, emotional pictures elicited an enlarged early posterior negativity (EPN) and late positive potential (LPP) indicative for preferential emotion processing. These data demonstrated that the activation of the fear/anxiety network exerted valence-specific effects on unrelated affective picture processing.

Complementary to Study I and II, Study III examined the perceptual processing of picture materials which were explicitly instructed to cue electric shocks. In different experimental runs, the same picture categories served as a threat-cue (predictive for threat- of-shock), safe-cue (predictive for no shock), and control cue (no threat-of-shock). Results fully replicated selective emotion processing as indicated by the EPN and LPP.

Furthermore, instructed threat-cues elicited pronounced modulations for the P1, P2 and

LPP amplitudes irrespective of the implicit hedonic picture valence. Of main interest, the differentiation between threat-, safe- and control cues varied across the visual processing stream. Depending on the predictive picture value and elapsed processing time, stimulus processing progressively gained high-accuracy information about environmental conditions.

Integrating both threat-of-shock and picture viewing paradigm the present thesis demonstrated the mutual impact of aversive anticipation and emotion processing. At the level of perceptual-evaluative processes, motor/behavioral output and self-report data, key findings regarding both manipulations were fully replicated. Of main interest, the threat-of- shock manipulation revealed valence-specific and valence-unspecific effects on picture processing. These findings depend critically on (1) whether the picture content is predictive or unrelated to threat-of-shock, and (2) which processing stage is focused (e.g., perceptual-evaluative, or behavioral/motor output). Rather predictable threat-of-shock was related to enhanced vigilance, more elaborate stimulus processing, and physiological response patterns irrespective of hedonic picture content. In contrast, viewing pictures unrelated to contextually signaled threat, exhibited valence-specific effects on electrocortical picture processing, and revealed valence-sensitive effects on the defensive reflex activity. Overall, the anticipation of real-world aversive events dynamically modulated information processing stream and psychophysiological response systems in order to flexibly adjust the organism to environmental contingencies.

Zusammenfassung

Die vorliegende Arbeit untersucht den Einfluss von Erwartungsangst auf emotionale Verarbeitungs- und Reaktionsprozesse. Zu diesem Zweck wurden zwei etablierte Forschungslinien integriert: Das „Bildbetrachtungsparadigma“ und das „verbal instruierte Bedrohungsparadigma“ (Threat-of-shock). Die Darbietung von Bildern mit unterschiedlichem Valenz- und Erregungsgehalt geht mit Veränderungen der Wahrnehmung sowie spezifischen physiologischen und Verhaltensprozessen einher. Von einer evolutionären Perspektive ausgehend wird angenommen, dass entwicklungsgeschichtlich alte motivationale Annäherungs- und Vermeidungssysteme selektiv die Verarbeitung bedeutsamer Umweltinformationen fördern und entsprechende physiologische Systeme zur Ausübung schneller situationsadäquater Reaktionen vorbereiten. Um eine dynamische Anpassungen an phylogenetisch und ontogenetisch grundlegende Ziele zu ermöglichen, können Zusammenhänge zwischen Bedrohungsreizen, unangenehmen Ereignissen und entsprechenden Reaktionen schnell und einfach erlernt werden. In diesem Kontext konnte mit Hilfe des Threat-of-shock Paradigmas gezeigt werden, dass motivationale Systeme durch verbale Instruktionen über aversive Umweltzusammenhänge moduliert werden. In anderen Worten, die rein verbale Ankündigung drohender Gefahr ist ausreichend um Erwartungsangst auszulösen.

Bezüglich des Einflusses von Erwartungsangst auf emotionale Wahrnehmungs- und Reaktionsprozesse ist wenig bekannt. Um dieser Fragestellung nachgehen zu können, wurden zwei Aspekte von wesentlicher Bedeutsamkeit variiert: (1) Der Vorhersagewert der Bildmaterialien bezüglich der Bedrohung durch elektrische Schocks (die Bilder waren entweder irrelevant oder prädiktiv für Schockgefahr), und (2) die fokussierte Verarbeitungsebene (d.h. die Messung perzeptuell-evaluativer vs. physiologisch- motorischen Verarbeitungsprozesse).

Studie I untersuchte den Einfluss von kontextuellen Bedrohungssignalen auf psychophysiologische Reaktionen (defensiver Lidschlag-Reflex und Hautleitfähigkeit) beim Betrachten von Bildern, die nicht prädiktiv für Schockgefahr waren. Zu diesem Zweck wurden schnell wechselnde angenehme, neutrale und unangenehme Bilder während lang anhaltender Bedrohungs- und Sicherheitsbedingungen dargeboten. Frühere Befunde zur Modulation des Schreckreflexes durch den emotionalen Bildgehalt und der instruierten Bedrohungssituation wurden repliziert. Von besonderem Interesse war, dass trotz einer generellen Potenzierung defensiver Schutzreflexe unter Bedrohungsbedingungen der defensive Lidschluss-Reflex sensitiv für die hedonische Valenz der Bildinhalte war. Die Ergebnisse weisen darauf hin, dass motivationale Systeme durch die Betrachtung von Bildern und anhaltenden aversiven Erwartungsprozessen unabhängig moduliert werden können.

In einem ähnlichen Experimentaldesign untersuchte Studie II den gemeinsamen Einfluss von Erwartungsangst und emotionaler Bildverarbeitung auf perzeptuelle und Aufmerksamkeitsprozesse mit Hilfe von ereignis-korrelierten Hirnpotentialen (EKP). Die Resultate zeigten speziell für angenehme Bilder eine frühe und lang anhaltende Negativierung über visuellen Verarbeitungsarealen unter Schockgefahr im Vergleich zur Sicherheitsbedingung. Dieser valenzspezifische Befund wird als verstärkte Aufmerksamkeitslenkung und Verarbeitung von situations-inkongruenten Reizen interpretiert. Darüber hinaus wurden emotionsspezifische EKP Komponenten („early posterior negativity, EPN“ und „late positive potential, LPP“) unter beiden Bedingungen repliziert.

Aufbauend auf diesen Befunden, wurde in Studie III die perzeptuelle Verarbeitung von Bildmaterialien untersucht, die explizit als Prädiktor für elektrische Schocks instruiert

wurden. Zu diesem Zweck wurde in verschiedenen Durchgängen der implizit emotionale Bedeutungsgehalt der Bildkategorien jeweils als Schockhinweis-, Sicherheits- und Kontrollreize verbal moduliert. Das Betrachten von Bildern unter möglicher Schockgefahr war mit frühen Effekten einer erhöhten P1-Komponente assoziiert, die als unspezifisch erhöhte Vigilanz in Gefahrensituationen interpretiert wurde. Darauffolgende Verarbeitungsstufen (P2-Komponente und vor allem die späten positiven Potentiale, LPP) differenzierten zunehmend akkurat zwischen Schock- und Sicherheitssignalen sowie Bildverarbeitung unter Kontrollbedingungen. In Abhängigkeit des prädiktiven Gehaltes und der abgelaufenen Verarbeitungszeit, erlangen Prozesse der sensorischen Aufnahme und Reizevaluation zunehmend genauere Informationen über Umweltbedingungen.

Zusammenfassend zeigt die vorliegende Arbeit den wechselseitigen Einfluss von Erwartungsangst und emotionalen Verarbeitungsprozessen. Hauptbefunde des Bildbetrachtungs- und des verbal instruierten Bedrohungsparadimas wurden repliziert.

Bezüglich der Interaktion von emotionalen und aversiven Erwartungsprozessen, zeigten sich valenz-spezifische und valenz-unabhängige Ergebnismuster in Abhängigkeit der Vorhersagekraft (distinkter Schockhinweisreiz vs. kontextuelle Bedrohungssignale) und der Messebene (perzeptuell-evaluativ vs. motorisch-behavioral). Unabhängig vom impliziten emotionalen Gehalt der Bilder, geht die Verarbeitung prädiktiver Schockhinweisreize mit generell erhöhter Vigilanz, verstärkter Reizevaluation und psychophysiologischen Reaktionsprozessen einher. Demgegenüber steht die Betrachtung bedrohungsirrelevanter Bilder im Zusammenhang mit valenz-spezifischen perzeptuellen Verarbeitungsmuster und valenz-sensitiven psychophysiologischen Reaktionsprozessen.

Insgesamt werden Prozesse der Informationsverarbeitung und psychophysiologischer Reaktionsvorbereitung dynamisch an wechselnde Umweltbegebenheiten angepasst und durch aversive Erwartungsprozesse moduliert.

Chapter 1

General Introduction

General Introduction

1.1 Anticipatory Anxiety

Whether to achieve appetitive or to avoid aversive outcomes, human behavior is goal directed and future oriented. Depending on the motivational impact of approaching events, the anticipation and preparation of appropriate behavior is crucial. According to this perspective, phylogenetic core objectives, such as sustaining life and organism integrity in the face of threat are universal to living organisms. According to their evolutionary significance, such functions are suggested to be based on primitive evolved systems connecting significant environmental information (e.g., fast movements, proximity, key threat indicators) directly with appropriate actions (Cacioppo, Gardner, &

Berntson, 1999; Lang, Bradley, & Cuthbert, 1997; Öhman, Flykt, & Lundqvist, 2000;

Tooby & Cosmides, 1992). Moreover, in favor of adaptation and dynamic adjustment to ontogenetic relevance and individual goals, these systems are “prepared” for learning new associations between potential dangers and appropriate actions (Bandura, 1977; Hugdahl &

Johnsen, 1989; Öhman & Mineka, 2001; Seligman, 1971).

Noxious stimulation leads to physiological, emotional and behavioral changes of critical adaptive significance. For instance, the experience of pain minimizes injury by motivating withdrawal or escape. Furthermore, in order to anticipate and avoid future harm, learning about threatening antecedents and cues is of vital interest (Darwin, 1872;

Ploghaus et al., 1999). As a model of fear learning, a wealth of animal and human research utilized classical conditioning paradigms, in which neutral stimuli acquire emotional properties by being paired with aversive events (e.g., electrical shocks; Davis, 1992;

Fanselow, 1994; Hamm & Vaitl, 1996; LeDoux, 2002; Rescorla, 1988). However, recent research demonstrated that the direct encounter with a negative reinforcer is not a necessary condition for learning. For instance, both humans and other mammal animals

can learn by observing the actions of others (Bandura, 1977; Olsson, Nearing, & Phelps, 2007). Even more efficient and associated with fewer risks than learning by direct or observed experiences is learning by means of various forms of communication (Rachman, 1977; Olsson & Phelps, 2004). For example, you may be told to beware a particular neighborhood because it is known to be dangerous. Accordingly, you might avoid or actually fear this area without ever having been there before, and moreover, even without ever having been victim of a crime yourself. Knowledge in this case is mediated by verbal communication. By contrast with classical conditioning, this example points to a rather abstract neural representation of the associations between stimulus (neighborhood as contextual threat cue), event (assault or injury), and response (avoidance behavior; Lang, 1979; Olsson & Phelps, 2007).

In the cognitive domain, such associative representations have been suggested to be organized as propositional networks, i.e., logical relationships between concepts (Anderson

& Bower, 1973; Kieras, 1978). Following this approach, Peter J. Lang (1979) proposed a bio-informational theory integrating propositional network and physiological response patterns (e.g., heart rate, electrodermal activity). Specifically, the mental representation of environmental settings can be evoked by verbal instructions (e.g., imagery scripts) activating propositional structures which comprise various levels coding semantic meaning, stimulus representation and response programs (Lang, 1979; Kissler, Assadollahi, & Herbert, 2006). For instance, in a spider phobic individual the encounter with a spider might include following propositions regarding the stimulus and situational context (e.g., long legs; moving fast towards me; no one else is here), overt and covert responses (e.g., self-referent statements “I’m scared”; physiological responses such as tachycardia and sweating; avoidance behavior) and the interpretive semantic meaning (e.g.,

“spiders are disgusting/dangerous”). In this context, aversive learning by means of verbal

communication is assumed to be crucially based on the anticipation and imagery of the potential aversive outcomes and consequences. Accordingly, fear/anxiety networks are suggested to be activated even when lacking concrete external reinforcement such as electric shocks during the acquisition phase and later actual threat cue encounter.

Focusing on aversive anticipatory processes, proceedings in basic research are of particular relevance for clinical research on the etiology and treatment of anxiety disorders.

Summarized in diagnostic classification systems (e.g., Diagnostic and Statistical Manual of Mental Disorders, DSM-IV by the APA, 1994; International Statistical Classification of Diseases and Related Health Problems, ICD-10 by the WHO, 2006), the anxious apprehension of aversive events is accompanied by a broad spectrum of symptoms including subjective (e.g., diffuse unpleasant feelings and worries, loss of control), behavioral (e.g., avoidance) and physiological responses (e.g., heart palpitation, muscle tension, sweating). Whereas the experience of fear is a normal and necessary reaction to threatening situations, however, a state of disorder emerges when fear becomes greater than that warranted by the situation. For instance, individuals with generalized anxiety syndrome (GAS) suffer from multifaceted and sustained worries about possible outcomes of daily events. In panic disorder, key elements refer to suffering panic attacks (i.e., intense episodes of fear, physiological arousal, feelings of impending catastrophes), and the constant state of anxious apprehension directed at the reoccurrence of panic attacks and their possible consequences.

With regard to the eliciting conditions, symptomatology, and response foci, a distinction between fear and anxiety has been proposed (Barlow, Chorpita, & Turovsky, 1996; Davis, 1998; Grillon, 2002). Fear is an immediate and cue-specific aversive sensation tied to clear response patterns preparing either fight or flight behavior in an emergency situation (e.g., in spider phobics viewing spiders; Cannon, 1929; Hamm,

Cuthbert, Globisch, Vaitl, 1997; Lang et al., 1997). On the other hand, more generalized anxiety is elicited by the mere anticipation of aversive events (e.g., in GAS patients worrying about various topics). The imagination about what might happen in an uncertain situation is perceived as a sustained, diffuse, and free-floating aversive state in response to an imprecise or unknown threat (Barlow et al., 1996).¹

Neural Representation of Fear and Anxiety

Although the focus of the response is different (immediate vs. anticipated threat), fear and anxiety are closely interrelated depending on the activation of evolutionarily primitive subcortical circuits (Davis, 1998; Lang, Davis & Öhman, 2000). Based on evolved modules tailored to solve recurrently encountered problems in biogenesis, these systems are presumed to be organized motivationally to promote survival (Lang et al., 1997; Öhman & Mineka, 2001; Tooby & Cosmides, 1992). Thus, activated by unconditioned aversive stimuli, corresponding systems organize autonomic and hormonal changes, motor behavior and attentional processes (Lang & Davis, 2006; LeDoux, 2000).

A key structure for the acquisition and expression of fear/anxiety is the amygdala, a limbic structure in the medial anterior temporal lobe which consists of several distinct cell ensembles, nuclei (e.g., central, basolateral, medial) and extended areas (e.g., bed nucleus of the stria terminalis, BNST). The basolateral amygdala mediates sensory information from the thalamus, hippocampus and widespread cortical areas. On the other side, rather efferent structures such as the central amygdala nucleus and the BNST project to several hypothalamic and brainstem areas subserving more specialized functions and controlling organ systems (see Figure 1.1; Davis, 1998). Thus, the encounter with stimuli predictive for threat is suggested to change neural transmission in the amygdala producing somatic,

1 In the following, the terms fear and anxiety will be used in reference to this distinction.

autonomic and endocrine signs of fear and anxiety (for detailed reviews see: Aggleton, 2000; Davis, 2002; Davis & Whalen, 2001; Lang & Davis, 2006).

Anatomical target Effect of stimulation Physiological and behavioral signs Lateral

hypothalamus

sympathetic activation tachycardia, SCR, paleness, pupil dilation, blood pressure elevation N. paraventricularis ACTH release corticosteroid release, "stress response"

Nervus vagus parasympathetic activation bradycardia, urination, defecation, ulcers N. parabrachialis increased respiration panting

Locus coeruleus, tegmental area

activation of dopamine, norepinephrine,

acetylcholine systems

behavioral and EEG arousal, increased vigilance

N. reticularis pontis

caudalis increased reflexes increased startle responses Central grey cessation of behavior freezing

Fig. 1.1. Fear vs. anxiety: Hypothetical illustration of the differential involvement of the amygdala and BNST including anatomical target areas and related functions (from Davis, 1992, 1998).

Furthermore, human and animal research suggested the amygdaloid complex differentially involved in cue-specific fear (e.g., central amygdala nucleus) and more context-related anxiety (e.g., BNST; cf. Davis, 1998; Grillon, 2002; Hasler et al., 2007).

For instance, amygdala lesions were observed to suppress both explicit and contextual fear conditioning. In contrast, lesions of the hippocampus or BNST specifically suppress contextual fear conditioning (Kim & Fanselow, 1992; McNish, Gewirtz & Davis, 1997;

Phillips & LeDoux, 1992).

Considerably less is known about the neural base of fear learning by means of verbal communication or observation (Cook & Mineka, 1990; Mineka & Cook, 1993;

Olsson et al., 2007; Olsson & Phelps, 2004, 2007). Recent research revealed similar neural circuitry involved in various ways of learning (e.g., insular cortex, amygdala, anterior cingulate cortex). However, accounting for cognitive social inference processes and language representation in observational and verbal learning, the representation of CS-US associations is suggested to recruit additional structures (e.g., medial prefrontal cortex, visual cortex; Mineka & Cook, 1993; Olsson, Nearing, & Phelps, 2007; Olsson & Ochsner, 2008; Olsson & Phelps, 2007; Pulvermüller, 2005).

Threat-of-shock Paradigm: The Anticipation of Aversive Events

The “threat-of-shock” paradigm has been established as a distinguished method to induce fear and anxiety in humans (Deane, 1969; Grillon, Ameli, Woods, Merikangas, &

Davis, 1991; Reiman, Fusselman, Fox, & Raichle, 1989). This experimental protocol comprises the explicit instruction to expect an aversive event (e.g., electrical shock) during

‘threat’ but not during ‘safe’ conditions. A special feature of this protocol is that participants do not necessarily need to experience the aversive event. Accordingly, the mere verbal announcement of threat imminence has been shown to reliably elicit

physiologic response patterns resembling results from classical fear conditioning studies (Olsson & Phelps, 2004).

Recent research in patients and healthy participants worked out basic parameters of the threat-of-shock protocol prompting aversive anticipation. In order to elicit a pronounced fear/anxiety response, the anticipated event needs to be sufficiently aversive and perceived as non-controllable (Grillon, Baas, Lissek, Smith, & Milstein, 2004).

Depending on the proximity of a potential threat, physiological systems have been shown to respond differently in fostering either sensory intake (remote) or preparing avoidance actions (imminent; Cornwell, Echiverri, Covington, & Grillon, 2008; Grillon et al., 2004;

Hasler et al., 2007; Lang, Davis, & Öhman, 2000). Closely related, the amount of predictability is an important factor evoking either fear or anticipatory anxiety (Armfield, 2006; Grillon et al., 2004, Grillon, Baas, Cornwell, & Johnson, 2006; Seligman, 1968;

Zvolensky, Eifert, Lejuez, Hopko, & Forsyth, 2000). For instance, when signaled by an explicit cue (e.g., red square) an aversive event appears rather predictable because the possible incidence of threat is limited in time (Mol, Baas, Grillon, van Ooijen, &

Kenemans, 2007). In contrast, extending a threat period in time, putative aversive events are less locked to the cue onset and therefore less predictable. In transfer, proximity and predictability can be modulated as rather spatial parameters (e.g., picture size, spatial distance; Löw, Lang, Smith, & Bradley, 2008; DeCesarei & Codispoti, 2006). In this case, an explicit cue is a spatially focused signal distinct from its background. In contrast, contextual cues (e.g., darkness) are linked to experimental settings, establishing a rather sustained background information (Grillon & Ameli, 1998). Taken together, depending on threat-cue parameters (e.g., proximity, predictability), the threat-of-shock manipulation provides a paradigm to provoke both cue-specific fear and anticipatory anxiety.

The activation of the fear/anxiety circuitry by means of the threat-of-shock manipulation is revealed by a broad array of response measures including self-report, behavior and physiology. In the following sections, research focusing on the impact of verbally instructed threat will be reviewed with respect to different psychophysiological measurements (e.g., motor and perceptual parameters). In addition, corresponding clinical research will be outlined briefly.

Peripheral Psychophysiology of Anxiety

Physiological response systems are sensitive to the signal value of environmental stimuli (Öhman, Hamm, & Hugdahl, 2000). As an indicator of sympathetic activation, the electrodermal activity (e.g., skin conductance response, SCR) has been shown to be enlarged under threat-of-shock as compared to safety conditions (Bradley, Moulder, &

Lang, 2005; Chattopadhyay, Cooke, Toone, & Lader, 1980). Moreover, this pattern resembles findings from studies using differential conditioning and observational learning paradigms (Olsson & Phelps, 2004). As an indicator for cardiovascular activity, variations in heart rate strongly depend on threat imminence (Lang, Davis & Öhman, 2000). Thus, a biphasic pattern from orienting (deceleration; ‘fear bradycardia’) to defense (acceleration;

action mobilization) has been described in animal and human research (Bradley, 2009;

Graham, 1979; Lang et al., 1997). Accordingly, passively viewing threat signals has been shown to be associated with cardiac deceleration (Bradley et al., 2005), whereas heart rate acceleration has been observed during intense fear to immediate threat (Deane, 1969;

Hamm et al., 1997; Melzig, Weike, Zimmermann, & Hamm, 2007).

In the presence of threat-of-shock cues, defensive reflexes are clearly potentiated as measured by the orbicularis eyeblink response to auditory startle probes (Baas, Kenemans, Böcker, & Verbaten, 2002; Böcker, Baas, Kenemans, & Verbaten, 2004; Bradley et al.,

2005; Funayama, Grillon, Davis, & Phelps, 2001; Grillon et al., 1991; Grillon, Ameli, Merikangas, Woods, & Davis, 1993; Grillon & Davis, 1995; Lissek et al., 2007).

Furthermore, this defensive motor behavior is especially pronounced under conditions of uncertainty. For instance, anticipating unpredictable (not reliably cued) in contrast to predictable (reliably cued) aversive events caused potentiated startle responses (Grillon et al., 2004). Pursuing the temporal specifity of this threat effect, Grillon and colleagues (1993) instructed only the last 10 s of threat-of-shock periods (50 s) as putative shock time windows. Replicating previous research, overall eyeblink startle responses were greater during threat than safety periods. However, this effect became progressively larger the longer each threat period lasted, and then abruptly decreased with the offset of the threat signal (Grillon et al., 1993). Thus, narrowing the critical time period of potential threat, defensive reflex system is increasingly prepared to respond.

Aversive activation in the threat-of-shock paradigm can be seen in virtually all individuals (Grillon et al., 1991), and remains observable across repeated experimental sessions (Baas, Grillon et al., 2002; Bitsios, Philpott, Langley, Bradshaw, & Szabadi, 1999). Furthermore, with respect to the power and stability of threat-of-shock effects, habituation processes across time are of particular interest. Several studies observed threat- of-shock effects as relatively sustained (Chattopadhyay et al., 1980; Cornwell et al., 2007;

Grillon et al., 1991; Grillon et al., 1993; Mol et al., 2007). Specifically, potentiated startle reflexes were sustained across blocks of few trials in some studies resulting in non- significant threat-of-shock by block interactions (Grillon et al., 1991; Grillon et al., 1993).

However, a fMRI study reported significant within block habituation, i.e., attenuated amygdala activation (Phelps et al., 2001). Taken together, the available evidence supports the notion that verbally mediated threat-of-shock is a powerful tool to examine fear/anxiety in humans.

The neural structures mediating the defensive eyeblink reflex are well-documented in animal and human research (see for detailed reviews Davis, 1992; LeDoux, 2000).

Accordingly, startle studies in clinical population (e.g., lesion patients) provide inferential information about the functional and anatomical connectivity in human emotions.

Regarding neural substrates involved in emotional picture viewing and threat-of-shock, Funayama and colleagues (2001) reported impaired modulation of fear responding due to instructed threat in patients with left unilateral temporal lobectomy (LTL) but not in patients with right temporal lobe damage. The opposite pattern of startle modulation was observed when participants viewed emotional pictures. Correspondingly, a recent fMRI study by Phelps and colleagues (2001), observed left-sided amygdala activation during threat-cue trials. These hemispheric differences might refer to the aspect of awareness of a CS-US association, not learned by experience but verbal instructions (Funayama et al., 2001; Morris, Öhman & Dolan, 1998; Olsson & Phelps, 2004).

As referring to clinical implications, several studies examined physiological response measures in individuals suffering from fear/anxiety. For instance, abnormal startle responses have been shown in individuals with PTSD (Grillon & Baas, 2003).

Although, exaggerated startle response is not considered a symptom in anxiety disorders other than PTSD, startle methodology is informative in identifying stimuli and conditions that are anxiogenic to anxiety patients (e.g., in specific phobias). Thus, phobic individuals have been shown to exhibit potentiated startle responses specifically to their feared objects (e.g., spiders in spider phobics) but not to other fear-relevant or non-related stimuli (e.g., snakes; Globisch, Hamm, Esteves, & Öhman, 1999; Hamm et al., 1997). Considerably less is known regarding other anxiety disorders (Grillon & Baas, 2003). For instance, panic disorder patients show similar startle magnitudes and SCL changes to threat-of-shock and darkness manipulations as compared to healthy controls (Grillon, Ameli, Goddard, Woods,

& Davis, 1994; Melzig et al., 2007). However, overall anxious patients have been observed to be highly sensitive to aversive contexts (e.g., shock electrode placement; participation in a stressful experiment; Grillon & Ameli, 1998).

Neuroimaging of Anxiety

Methods like functional magnetic resonance imaging (fMRI) allow to trace non- invasively increased brain activity based on blood oxygenation variation (blood oxygenation level dependent, BOLD-effect; Logothetis, Pauls, Augath, Trinath, Oeltermann, 2001; Ogawa, Lee, Kay, & Tank, 1990). Thus, measuring regionally specific metabolic changes associated with experimental manipulations provides high resolution insights in the anatomic locus of anxiety-related processes.

Neuroimaging studies revealed neural substrates of verbally induced expectations of aversive events. Similar to the anticipation of aversive pictures (Simmons, Matthews, Stein, & Paulus, 2004; Nitschke, Sarinopoulos, Mackiewicz, Schaefer, & Davidson, 2006), threat-of-shock compared to safety signals were found to be associated with modulated blood flow in the amygdala, anterior cingulate cortex (ACC), insular, and prefrontal cortices (Chua, Krams, Toni, Passingham, & Dolan, 1999; Dalton, Kalin, Grist &

Davidson, 2005; Hasler et al., 2007; Phelps et al., 2001; Ploghaus et al., 1999; Simpson, Drevets, Snyder, Gusnard, & Raichle, 2001). Furthermore, recent studies observed the anticipatory activation of relevant primary (SI) and secondary sensory processing areas (SII), indicating modulated somatosensory and exteroceptive sensory-discrimative processing of threat cues, during the anticipation of electrical stimulation and pain (Berns et al., 2006; Bornhövd et al., 2002; Carlsson et al., 2006).

Given the choice of waiting for an electric shock or getting it over with quickly, many people prefer the latter, even when the short delay would imply higher voltage

stimulation (Berns et al., 2006). In this context, shock intensity was associated with increased BOLD changes within insular, ACC and somatosensory cortices, however, the delay of aversive events did not modulate responses in these areas (Berns et al., 2006;

Bornhövd et al., 2002). Furthermore, recent research reported unpredictability as an important factor mediating habituation of neural activity (e.g., in the amygdala) and emotional behavior (Bornhövd et al., 2002; Grillon et al., 2006; Herry et al., 2007). Thus, the predictability of aversive events may play a major role in the processing of threat signals gating anticipatory processes. While predictable threat provides correct temporal estimations of events, these conditions are reported as less aversive (e.g., lower in anxiety, negative valence and pain intensity). Associated with enhanced activity in sensory cortices and posterior insular, predictable conditions have been suggested to modulate coping processes attenuating focus on danger, and selective attention to the sensory input (Berns et al., 2006; Bornhövd et al., 2002; Carlsson et al., 2006). In contrast, the more aversive context of unpredictable threat was associated with brain activity in affective pain processing areas (e.g., anterior insular, orbitofrontal cortex) and parietal cortex suggested to indicate enhanced alertness and sustained attention during unpredictable threat (Bornhövd et al., 2002; Carlsson et al., 2006). Accordingly, neural activation patterns associated with acute pain and its anticipation are closely connected (Ploghaus et al., 1999).

Furthermore, changes in perceived anxiety and psychophysiological responding have been shown to be correlated with neural activity in anticipatory anxiety. For instance, neural-peripheral coupling under threat-of-shock has been observed between amygdala, insular and prefrontal BOLD changes in relation to self-reported anxiety, cardiac contractility, and skin conductance responses (Carlsson et al., 2006; Dalton et al., 2005;

Phelps et al., 2001). However, to establish models of neural-peripheral connectivity, future

translational research is needed to integrate findings from animal and human research (cf.

Herry et al., 2007).

A special case of anticipatory anxiety refers to the anticipation of social situations comprising threat of social evaluation (e.g., giving a speech; Lorberbaum et al., 2004;

Tillfors, Furmark, Marteinsdottir, & Fredrikson, 2002). For instance, Cornwell, Johnson, Berardi, and Grillon (2006) observed correlations between trait social anxiety and startle reactivity during speech anticipation. Furthermore, in patients suffering from social phobia the anticipation of social threat has been observed to be associated with greater subcortical and limbic (e.g., amygdala, insular) activation, as well as modulated blood flow in ACC and prefrontal cortices (Lorberbaum et al., 2004; Tilfors et al., 2002). Regarding the involvement of prefrontal structures, recent research highlighted the regulatory impact on emotion, anticipation and decision making (Davidson, Jackson & Kalin, 2000; Jackson, Malmstadt, Larson, & Davidson, 2000; Ochsner et al., 2004). For instance, Bechara, Damasio, Damasio and Anderson (2000) observed that patients with lesions of the ventromedial prefrontal cortices exhibited difficulties in the anticipation of future pleasant and unpleasant consequences and therefore showed impaired choice behavior. Taken together, increasing evidence suggested that the anticipation of aversive events modulates brain activity in widespread cortical networks involved in stimulus representation (e.g., pain), motivational/emotional, and attentional processes.

Electrophysiology of Anxiety

The presentation of distinct stimuli evokes electrical brain potentials reflecting a multitude of ongoing perceptual and cognitive processes (Rugg & Coles, 1995). These event-related potentials (ERP) can be extracted out of the broader spontaneous EEG activity by means of averaging procedures (e.g., Luck, 2005) strengthening the ratio of

event-related to non-related signals. Although less specific in spatial localization of generator sources, ERP methodology provides high temporal resolution insights in brain dynamics.

Threat of aversive events increased several ERP components, including early sensory components related to physical stimulus features (e.g., spatial frequency) and later components like the P3 wave indicative of increased selective attention (Baas, Kenemans et al., 2002; Baas, Milstein, Donlevy & Grillon, 2006; Böcker et al., 2001, 2004; Löw et al., 2008; Scaife, Groves, Langley, Bradshaw, & Szabadi, 2006). Recent studies showed that threat-of-shock modulated the processing of neutral stimuli at very early sensory processing stages. For instance, the brain stem wave V to simple click sounds, which is partly generated by the inferior colliculus, was enlarged when presented during threat-of- shock compared to safety periods (Baas et al., 2006). In accordance with clinical studies, these results suggest that the fear system already modulates perceptual processes within the first 10 ms after stimulus onset (Baas et al., 2006). In addition, as revealed by a MEG study, the processing of auditory stimuli deviant from a previously established sound context was associated with greater right auditory and inferior parietal activity (Cornwell et al., 2007). Furthermore, indicating anxiety-related sensitization to environmental changes, the magnetic mismatch negativity (MMNm) in response to stimulus deviance was positively correlated with increased anxiety levels under threat-of-shock (Cornwell et al., 2007).

In the visual modality, Baas, Kenemans and colleagues (2002) reported early occipital threat-of-shock effects in interaction with physical stimulus features (e.g., spatial frequency). Relative to safety cues, high frequency threat cues were associated with increased occipital negativity whereas low spatial frequency cues were accompanied with more positivity around 60-100 ms poststimulus. Subsequent stimulus processing revealed

threat modulation in the P3 complex and slow waves. Specifically, visual threat cues were associated with pronounced late positive potentials (LPP; 300 - 500 ms) over parietal sensorsites resembling modulations due to the intrinsic emotional value of more complex stimuli, e.g., IAPS pictures (Baas, Kenemans et al., 2002; Schupp, Junghöfer, Weike, &

Hamm, 2003b, Schupp, Flaisch, Stockburger, & Junghöfer, 2006). The LPP component is indicative for attention processes related to conscious recognition and elaborate processing of motivational significant stimuli (Del Cul, Baillet, & Dehaene, 2007; Nieuwenhuis, Aston-Jones, & Cohen, 2005, Schupp et al., 2006). Thus, enhanced LPP amplitudes have been consistently observed in studies utilizing complex pictures depicting threatening contents (e.g., mutilation, human and animal threat; Cuthbert, Schupp, Bradley, Birbaumer,

& Lang, 2000; Keil et al., 2002; Palomba, Angrilli, & Mini, 1997; Schupp et al., 2000, 2003b, 2006). Furthermore, the LPP component has been observed to increase for looming stimuli signaling reward and loss of money (Löw et al., 2008). Integrating these findings, the LPP component varies as a function of motivational stimulus significance mediated by instructed threat-contingencies and affective picture media.

As indicated by slow potentials, further studies observed differential processing for threat and safety cues in later processing stages related to sustained perceptual operations and memory processes (Böcker et al., 2001; Baas, Kenemans et al., 2002; Ritter &

Ruchkin, 1992). For instance, Böcker and colleagues (2001) observed a negative potential shift over medial frontal sensor sites (FCz) developing around 600 ms poststimulus evoked by threat cues indicating possible shocks. This frontal negativity resembles an anticipatory component observed in S1-S2 attention paradigms (e.g., Böcker et al., 2001; Rockstroh, Elbert, Canavan, Lutzenberger, & Birbaumer, 1989). This contingent negative variation (CNV) develops between two contingent events S1 and S2, particularly if S2 requires a

distinct response (cf. shock cue and subsequent aversive shock; Böcker et al., 2001; for detailed reviews see Brunia, 1999; Rockstroh et al., 1989).

Complementary evidence regarding the electrocortical processing of anxiety-related stimuli comes from research in clinical anxiety disorders. For instance, suggesting increased vigilance in individuals with spider and social phobia, generally enhanced P100 amplitudes were reported irrespective of the picture content (Michalowski et al., 2009;

Kolassa, Musial, Kolassa, & Miltner, 2006, 2007). Furthermore, increased EPN and LPP amplitudes were reported in animal phobics (e.g., spider phobics) during viewing their feared animal (e.g., spider pictures) in contrast to fear-relevant (e.g., snakes) or non-related picture contents (Kolassa, Musial, Mohr, Trippe, & Miltner, 2005; Michalowski et al., 2009; Miltner et al., 2005). In combination with behavioral data, these studies suggest enhanced perceptual processing and biased behavioral responses to idiosyncratically feared stimulus materials (Kolassa et al., 2006, 2007; Miltner, Krieschel, Hecht, Trippe, & Weiss, 2004; Öhman, Flykt, & Esteves, 2001).

Summary

It is proposed that fear and anxiety depend on the activation of evolutionary grown primitive subcortical circuits, including the amygdala and its neural projection areas. In favor of adaptation and dynamic adjustment to phylogenetic and ontogenetic core objectives, these systems easily acquire contingencies of environmental threat cues, aversive events and appropriate responses. Accordingly, converging evidence reveals the broad impact of merely verbally instructed threat. Threat-of-shock as compared to safety conditions are perceived as more threatening, aversive and higher in arousal, and if possible, such conditions are avoided. Furthermore, the perceptual stimulus processing, as well as autonomic and somatic response parameters are modulated by aversive anticipation

processes. Taken together, similar to explicit fear conditioning and observational learning, aversive conditions mediated by means of verbal communication activate fear/anxiety networks regulating perception, cognition, and avoidance behavior.

Whereas the current section focused on the impact of instructed fear and anxiety on emotional experience, physiological responding and behavior, in the following the relation between emotion and attention will be outlined. Main findings regarding the impact of viewing complex natural pictures varying in emotional valence and arousal will be reviewed.

1.2 Emotion and Attention

Emotions are considered here as evolved modules organizing and modulating multifaceted functions such as sensory intake, information processing, and various responding levels (e.g., subjective experiences, physiology and behavior). The functionality and hardware of such processes has been shaped in order to appropriately cope the recurrent encounter with adaptive problems in phylogenesis and ontogenesis (Cacioppo et al., 1999; Öhman & Mineka, 2001; Tooby & Cosmides, 1992). Based on the universal principles of approach and avoidance, basic motivational systems (appetitive and aversive) are assumed to provide the fundament for the evolvement of multifaceted human emotion response repertoire (Buck, 1994; Lang et al., 1997; Öhman et al., 2000). For instance, procreation, sustaining organisms’ integrity and the adaptation to environmental conditions are vital objectives motivating specific feelings and behavior. Moreover, various functions developed to organize human being and behavior in relation to their social and cultural contexts (e.g., communication, emotion expression and perception;

Ekman, 1999; Izard, 1977; Scherer, 1984).

Human emotion and behavior is embedded in environmental settings, thus, any action strongly depends on perceptual information up-take, and therefore, is gated through attentional processes. Not only the detection of action cues but also the fine-tuning and re- adjustment of ongoing responses require iterative situational updates. Accordingly, a seemingly endless stream of sensory and interoceptive input is needed to be encoded and integrated by information processing systems. Apparently, sensoric information processing needs to be selective. According to this perspective, both emotion and attention deal with processing priorities organizing adequate responding in a highly complex environment (Oatley & Johnson-Laird, 1987).

Recent attention theories commonly refer to limited capacity of processing resources leading to the necessity of input selection (Broadbent, 1958; Desimone &

Duncan, 1995; Deutsch & Deutsch, 1963; Hopfinger, Luck & Hillyard, 2004; Kastner &

Ungerleider, 2000; Treisman, 1969). In detail, selective attention refers to the aspect that not all stimuli are equally processed, i.e., some stimuli, events or attributes are given longer, more effective and elaborate processing than others (Compton, 2003; Kastner &

Ungerleider, 2000, 2001; Pashler, 1998). Currently, two broader selection mechanisms have been proposed in visual attention research: Firstly, explicitly directed attention, which implies voluntarily and goal-directed attention towards task-relevant locations, visual features, objects or higher-order semantic categories (Beck & Kastner, 2005; Hillyard &

Anllo-Vento, 1998; Posner, 1980). That is, if one actively directs attention (top-down) to a particular location in a complex scene, processing of information at the attended location will be facilitated and processing of irrelevant objects will be suppressed (Beck & Kastner, 2005). Secondly, selective attention may be sensory-driven (bottom-up) by stimulus characteristics such as physical key features (e.g., velocity, location; Corbetta, Miezin, Dobmeyer, Shulman, & Petersen, 1990; Luck, Woodman, & Vogel, 2000; Luck & Ford, 1998; Posner, 1980). Similarly, stimuli depicting implicit emotional/motivational significant information (e.g., related to procreation or life-sustaining topics) have been suggested to guide selective attention and receive preferential processing (Compton, 2003;

Stockburger, Schmälzle, Flaisch, Bublatzky, & Schupp, 2009; Lang et al., 1997; Schupp et al., 2003b, 2006; Vuilleumier, 2005). Unlike voluntary attention, these attentional processes are assumed to operate rather spontaneously, that is, in passive viewing conditions without explicit instruction to attend.

In this context, natural pictures depicting emotional and motivational significant scenes (e.g., mutilation, threatening or erotic scences) are useful stimulus materials to

examine the interaction of both implicit and explicit attention processes under conditions of competition (Kastner & Ungerleider, 2000; Pessoa & Ungerleider, 2004; Schupp, Stockburger, Bublatzky et al., 2007, 2008; Vuilleumier, Armony, Driver, & Dolan, 2001) and cooperation (Compton, 2003; Compton et al., 2003; Schupp, Stockburger, Codispoti et al., 2007). For instance, research on visual search revealed an ‘attention capture’ by features such as movement or abrupt onsets (Yantis & Jonides, 1984, 1990) and emotional salient cues (Hansen & Hansen, 1988; Öhman, Flykt, & Esteves, 2001). Similarly, more complex emotional stimuli have been shown to attract attention and benefit from a processing advantage over neutral stimuli (Calvo & Lang, 2004; Öhman, Lundqvist, &

Esteves, 2001; Schupp et al., 2006). That is, environmental information depicting emotionally significant meaning guide selective attention and receive enhanced processing (e.g., Lang, et al., 1997; Cacioppo et al., 1999; Bradley, Codispoti, Cuthbert, & Lang, 2001; Öhman, Flykt, & Lundqvist, 2000; Hamm, Schupp, & Weike, 2003; Schupp et al., 2003b; Vuilleumier, Armony, & Dolan, 2003). Building upon this, explicitly directing attention to (or away from) emotional significant stimuli/features has been shown to modulate perceptual processing, physiological responding and behavior (Schupp, Stockburger, Codispoti et al., 2007).

Picture Viewing Paradigm

Humans perceive their environment predominantly visually. Thus, the presentation of pictures depicting real life contents provides a distinguished methodological approach to evoke content-related emotional states and to establish a contextual environment in an experimental laboratory setting. Since stimulus features (e.g., physical characteristics, presentation time) can be held controlled, responses measured during passive picture viewing strongly relate to processes elicited by the stimulus content. Furthermore,

concurrently performed active tasks (e.g., count emotional cues) allow to examine the interface of emotional and attentional processes.

To promote these benefits in emotion research, Lang and colleagues collected a set of naturalistic photographs (International Affective Picture System, IAPS; Lang, Bradley,

& Cuthbert, 2008) standardized with respect to subjective evaluations (e.g., Self- Assessment-Manikin, SAM; Bradley & Lang, 1994). As illustrated in Figure 1.2, a multitude of pictures depicting neutral and emotion-related scenes can be organized regarding their emotional value in an affective space spanned by basic dimensions of emotion: Valence and arousal (Bradley, 2000; Bradley & Lang, 2000). Thus, the IAPS set is designed to allow (1) for experimental control in stimulus selection, (2) comparisons of results from different studies, and (3) exact replications within and across research labs.

Fig. 1.2. The affective space as defined by means of pleasure and arousal for IAPS pictures (Lang et al., 1997). Vectors indicate hypothetic motivational systems organizing affective evaluations (from Schupp, Cuthbert, et al., 2004).

Until today, a large body of research utilized the IAPS with different methodological approaches, psychophysiological measurements and parameters. In the

following subsections, major results are reviewed focusing on affective modulations as revealed by peripheral measurements, neuroimaging and electrophysiological parameters.

Peripheral Psychophysiology of Emotion

To promote the survival of individuals and species, fast and accurate responses to environmental changes are essential (Öhman & Mineka, 2001). In this context, emotions are discussed as action dispositions (Frijda, 1986, Lang et al., 1997), providing the readiness to act and respond. Although not all emotional experiences yield an overt behavior, physiological activity in behavior relevant organ systems are reflected in psychophysiological parameters.

When viewing pictures with emotional and neutral contents, several response systems covary with affective valence and/or arousal, as defined by evaluative judgements (see Figure 1.3; Bradley et al., 2001). Accomplishing expressive and signal functions, valence sensitive responses are shown in facial muscle movements (Ekman & Friesen, 1975). For instance, zygomatic “smile” EMG activity increases linearly with rated pleasantness, and conversely, judged unpleasantness with the corrugator “frown” EMG activity (Bradley et al., 2001; Dimberg, 1990). As a cardio-vascular parameter, heart rate generally prompts marked deceleration during passive picture viewing suggested to indicate orienting, whereas pleasant in comparison to unpleasant pictures show greater accelerative responses related to motor preparation (Bradley et al., 2001; Lacey, 1967;

Palomba et al., 1997; Vila et al., 2003). Other response parameters covary with rated arousal rather than affective valence. For instance, regardless of picture valence, electrodermal activity and viewing time increase linearly with rated arousal of picture contents (Calvo & Lang, 2004; Greenwald, Cook, & Lang, 1989; Lang, Bradley, &

Cuthbert, 1990).

Fig. 1.3. Mean skin conductance changes and startle blink responses when viewing specific pleasant, neutral and unpleasant picture contents (from Bradley et al., 2001).

Of special interest, the defensive startle reflex (measured by the orbicularis eyeblink EMG activity) evoked by acoustic startle probes covaries with both rated picture valence and arousal in an interactive way (Berg & Balaban, 1999; Hamm, Greenwald, Bradley & Lang, 1993; Lang et al., 1997; Vrana, Spence & Lang, 1988). In reference to viewing neutral pictures or baseline startle reflex, the eyeblink activity decreases with rated

pleasantness and increases with unpleasantness. This dichotomy is even more pronounced for highly arousing than for stimuli or contextual settings rated lower in arousal (see Figure 1.3; Bradley et al., 2001; Schupp, Cuthbert, et al., 2004). Moreover, valence modulated blink reflex has been shown to be sensitive to individual differences and vary with clinical symptoms (Grillon & Baas, 2003; Hamm et al., 1997). Therefore, the affective eyeblink modulation supports the motivational priming hypothesis assuming that fundamental motivational approach and avoidance systems are linked to action programs, which are more responsive in the presence of congruently valenced environmental stimuli and contexts (Lang et al., 1997). Since this pattern is observable in humans and other animals, the startle reflex serves as an excellent variable to outline and test fundamental models of emotion (Lang & Davis, 2006).

Neuroimaging of Emotion

Research in mapping neural structures underlying emotional processes and experiences forged ahead dramatically with the advent of neuroimaging techniques.

Distinct neuronal activation patterns are observed consistently in viewing pictures of emotional significant in comparison to neutral pictures. Specifically, enhanced activity for affective materials (pleasant and unpleasant) is demonstrated in brain areas involved in visual information processing, for instance the occipital and fusiform gyri (see Figure 1.4;

e.g., Bradley et al., 2003; Junghöfer, Schupp, Stark, & Vaitl, 2005, Junghöfer, Peyk, Flaisch, & Schupp, 2006; Lang, Bradley, Fitzsimmons, et al., 1998; Sabatinelli, Flaisch, Bradley, Fitzsimmons, & Lang, 2004, Sabatinelli, Bradley, Fitzsimmons, & Lang, 2005).

Furthermore, signal changes in visual cortex vary with rated stimulus arousal (high arousing > low arousing > neutral stimuli; Bradley et al., 2003; Junghöfer et al., 2005).

Apart from modality specific processing areas, emotional modulation has been reported in

several cortical and sub-cortical structures suggested to mediate emotional processing and expression, such as the amygdala (e.g., Hadjikhani & de Gelder, 2003; Hamann, Herman, Nolan, & Wallen, 2004; Junghöfer et al., 2005; Norris, Chen, Zhu, Small, & Cacioppo, 2004; Sabatinelli et al., 2005; Vuilleumier et al., 2001; Whalen et al., 1998), insular (e.g., Junghöfer et al., 2005; Simmons et al., 2004) and orbito-frontal cortex (e.g., Karama et al., 2002; Murphy, Nimmo-Smith, & Lawrence, 2003; Nitschke et al., 2004; Vuilleumier et al., 2001).

Fig. 1.4. Illustration of in- creased BOLD-activation in extrastriate cortex while viewing emotional arousing pictures (from Sabatinelli et al., 2004).

Electrophysiology of Emotion

Further studies used event-related brain potentials to determine the temporal brain dynamics in cortical processing of emotional stimuli. With regard to emotion theories, recent ERP research strongly relates to attentional processes mediating fast and accurate extraction of emotional/motivational information. Consistent with this view, a large number of studies demonstrated reliable modulations in electrocortical parameters evoked by the emotional stimulus significance. Specifically, the processing of emotional picture cues has been shown to be associated with several ERP waveforms differentiating from neutral cues. In the following sections, findings regarding most consistently observed differential processing patterns, an increased early posterior negativity (EPN), and enhanced late positive potential (LPP) are reviewed.

The Early Posterior Negativity (EPN)

Differentiating the processing of emotional as opposed to neutral visual stimulus materials, the EPN is a negative going waveform developing between ~100-150 ms and maximally pronounced ~200-300 ms after stimulus onset. The topography of the EPN is temporo-occipital with a less pronounced corresponding polarity reversal over fronto- central sensorsites (see Figure 1.5). Furthermore, estimates of the neural generators by current source density (CSD; Junghöfer, Elbert, Leiderer, Berg, & Rockstroh, 1997) and L2-minimum-norm procedures (L2-MMN; Hämäläinen & Ilmoniemi, 1994; Hauk, Keil, Elbert, & Müller, 2002) suggest EPN sources over occipito-temporal-parietal sites (Junghöfer, Bradley, Elbert, & Lang, 2001; Schupp, Stockburger, Codispoti et al., 2007;

Schupp et al., 2006). Of special interest, this early indicator of selective emotion processing co-varies with the level of emotional arousal. Specifically, pictures depicting highly arousing pleasant and unpleasant contents (e.g., erotica, mutilation) prompt more pronounced EPN amplitudes than same valenced stimuli rated lower in emotional arousal (e.g., romance, pollution; Junghöfer et al., 2001; Schupp et al., 2003a, 2003b; Schupp, Junghöfer, Weike, & Hamm, 2004). Further evidence, which suggests that the EPN component is driven by the emotional value of visual stimuli, refers to the antecedent and eliciting conditions. Besides naturalistic photographs of pleasant and unpleasant scenes, also facial expressions and even more abstract cues like emotion-related hand gestures and written words have been shown to be associated with selective processing patterns as indicated by the EPN (Flaisch, Schupp, Renner, & Junghöfer, 2009; Herbert, Junghöfer, &

Kissler, 2008; Kissler, Herbert, Peyk, & Junghöfer, 2007; Schupp, Öhman et al., 2004).

Fig. 1.5. (a) ERP waveforms for pleasant, neutral and unpleasant pictures and (b) topography plots of the EPN (from Schupp et al., 2003a).

Furthermore, the EPN occurs spontaneously in passive viewing conditions (Schupp et al., 2003b; Schupp, Junghöfer et al., 2004), and while participants hold unrelated task goals in mind (Schupp et al., 2003a; Schupp, Stockburger, Bublatzky et al., 2007, 2008).

Thus, indicative for stimulus-driven processing, the EPN effect appears relatively unaffected by stimulus repetition and presentation features such as varying exposure times (Junghöfer et al., 2001; Peyk, Schupp, Keil, Elbert, & Junghöfer, 2009; Schupp et al., 2005, 2006). Accordingly, rather than referring only to low-level perceptual stimulus characteristics (e.g., figure/ground compositions, picture size; cf. Bradley, Hamby, Löw, &

Lang, 2007; DeCesarei & Codispoti, 2006), the emotional arousal modulation appears

secondary to feature-based stimulus identification as indicated by P1-waveform (Schupp, Stockburger, Schmälzle et al., 2008).

However, the EPN appears not as a fully automated processing default. For instance, directing explicit attention to concurrently presented visual task stimuli (i.e., superimposed grid patterns) has been observed to reduce – albeit not abolish – EPN amplitude (Schupp, Stockburger, Bublatzky et al., 2007). Interestingly, indicating no crossmodal resource competition, explicit attention directed to the auditory modality did not markedly reduce EPN amplitude for concurrently presented emotional stimuli (Schupp, Stockburger, Bublatzky et al., 2008). In contrast to competition designs, paying explicit attention to emotional significant rather than neutral cues did not further potentiate the EPN (Schupp, Stockburger, Codispoti et al., 2007).

In summary, recent research suggested the EPN to reflect the facilitated processing of visual scenes depicting information of emotional and motivational significance (Schupp et al., 2006). Accordingly, motivational systems regulating motor output (i.e., favoring approach or avoidance dispositions) were suggested to modulate already early sensory encoding stages facilitating detection and categorization of significant stimuli (Schupp et al., 2003b).

The Late Positive Potential (LPP)

Subsequent to early perceptual encoding stages, pronounced late positive potentials (LPP) over parietal sensorsites have been consistently observed for emotional in comparison to neutral cues (maximum ~400-600ms; Cacioppo, Crites, Gardner, &

Berntson, 1994; Cuthbert et al., 2000; Del Cul et al., 2007; Keil et al., 2002; Palomba et al., 1997; Schupp et al., 2000, 2003b). Similar to the EPN modulation, the LPP varies as a function of affective arousal. Specifically, the higher a picture is evaluated in emotional

arousal, the stronger pronounced is the differentiation between emotional and neutral cues (see Figure 1.6; Schupp et al., 2000; Schupp, Cuthbert et al., 2004; Schupp, Junghöfer et al., 2004). Furthermore, the affective modulation of the LPP has been shown to be a robust phenomenon relatively unaffected by stimulus familiarity, physical characteristics (e.g., stimulus size, complexity) and terms of presentation (e.g., mixed-blocked presentation, exposure time; Bradley et al., 2007; Codispoti, Ferrari, & Bradley, 2006, 2007; DeCesarei

& Codispoti, 2006; Pastor et al., 2008). Extending the LPP time window up to several seconds, a prolonged positive slow wave over centro-parietal sensorsites differentiates between emotional and neutral cues (Cuthbert et al., 2000; Pastor et al., 2008). In contrast to the EPN modulation, the LPP amplitude has been observed to be enhanced by actively paying attention to affective stimuli (Schupp, Stockburger, Codispoti et al., 2007).

Fig. 1.6. ERP waveforms for a representative parietal sensor when viewing specific pleasant, neutral and unpleasant picture contents illustrating LPP and slow waves. Note positivity is plotted downwards (from Schupp, Cuthbert et al., 2004).

Seen from a broader perspective, emotionally arousing stimuli are assumed to activate motivational brain circuits, prompting a sustained attentional set (Bradley, 2000;

Lang et al., 1997; Hamm et al., 2003; Schupp, Cuthbert et al., 2004). Thought to display processes of stimulus evaluation subsequent to perceptual encoding, the LPP has been established as a sensitive measure of attention indicative for capacity-limited processing, representation in working memory and conscious recognition (Donchin & Coles, 1988;

Kranczioch, Debener, & Engel, 2003; Luck et al., 2000; Nieuwenhuis et al., 2005; Schupp et al., 2006).

An Integrative Perspective of Selective Attention

In terms of polarity, latency and topography, both the EPN and LPP component closely resemble findings from cognitive psychology on explicit selective attention processes. That is, paying attention to distinct stimulus features (e.g., color, shape) or higher-order categorizations based on semantic meaning (e.g., animal vs. non-animal) has been shown to be associated with a selection negativity (SN; occipital, ~120 ms poststimulus) and the P3 component (parietal, ~300-600 ms; Codispoti, Ferrari, Junghöfer,

& Schupp, 2006; Delorme, Rousselet, Mace, & Fabre-Thorpe, 2004; Donchin & Coles, 1988; Hillyard & Münte, 1984; Johnson, 1988; Kok, 2001; Nieuwenhuis et al., 2005;

Picton, 1992; Potts & Tucker, 2001; Schupp et al., 2003a, 2006; Smid, Jakob, & Heinze, 1999).

Integrating both the emotion and cognitive approach, implicit emotional and explicitly instructed stimulus significance are suggested to guide selective attention and draw processing resources towards meaningful environmental stimuli. Based on sensory and low-level perception (e.g., feature detection and integration; Treisman, 2006; Treisman

& Gelade, 1980), early perceptual-evaluative processes are suggested to extract potentially

significant stimuli from sensory input overflow. At this early transitory processing stage, ERP components such as EPN and SN may serve as a measure of stimulus evaluation in perceptual representation regions, i.e., striate and extrastriate visual cortex (Junghöfer et al., 2001; Schupp et al., 2003b, 2006). Significant information, already tagged for preferential processing, is suggested to access higher-order capacity limited processing stages (as reflected by LPP and P3 components) implicated in elaborate high-accuracy processing, stimulus consolidation and conscious recognition (Kranczioch, Debener, &

Engel, 2003; Öhman, 1986; Sergent, Baillet, Dehaene, 2005; Schupp, Stockburger, Codispoti et al., 2007; Schupp et al., 2006; Wickens, Kramer, Vanasse, & Donchin, 1983;

Vogel, Luck, & Shapiro, 1998). In transfer to a natural complex environment, emotionally significant stimuli presumably receive preferential processing throughout both early and late processing stages. Thus, accurate situational information based on motivationally guided attention, provides the fundament to efficiently organize appropriate behavior and action (Lang et al., 1997; Öhman et al., 2000).

Summary

Based on fundamental motivational systems (defensive and appetitive), emotions are considered as action dispositions providing the readiness to adequately act and respond in a given situation. As a method to elicit emotions and simulate environmental conditions in a laboratory setting, the picture viewing paradigm is introduced and related experimental research reviewed.

According to the motivated attention theory, emotional significant stimuli guide selective attention and receive enhanced processing. Suggested to reflect such attentional processes, the early posterior negativity and late positive potentials for emotional as opposed to neutral picture cues are discussed. Building upon facilitated processing of

emotional significant stimuli, physiological systems are primed to respond fast and accurately within an environmental context. This motivational priming account is supported by a broad array of data showing context-dependent modulation in defensive reflexes (e.g., startle response), autonomic (e.g., heart rate, skin conductance) and behavioral parameters (e.g., reaction time, avoidance behavior).