In daily life, olfactory stimuli are potential generators of affective states, but also have a strong influence on social interaction. Pleasant odors have been shown to increase perceived attractiveness and pro-social behavior, whereas unpleasant body odors are often associated with negative personality traits. Since both pleasant odors and positive affective state facilitate pro-social behavior, it is conceivable that the influence of the odors on social interaction is mediated by the induced affective state elicited by the odor itself. The present thesis aims at exploring the impact of hedonic, i.e., pleasant or unpleasant, odors on the processing and evaluation of social stimuli as assessed by verbal, physiological, and behavioral indices. First, I investigate the effects of initially neutral odors which gained threatening value through an aversive conditioning procedure on social stimuli (Study 1). Second, I study the influence of naturally hedonic odors on social interaction. Third, this thesis aims at disentangling differences in the effects of an odor attributed to either a social interaction partner or the environment where the social encounter takes place (Study 2, 3, and 4). In the first study, a context conditioning procedure was applied, during which one out of two long-lasting neutral odors was paired with an unpredictable aversive unconditioned stimulus (US, i.e., white noise). This odor (CTX+) thereby gained threatening value, while another odor (CTX-) remained unpaired and therefore signaled safety. During a test session, facial stimuli were presented within both conditioned olfactory contexts. Results indicate that autonomic arousal was increased to faces when presented in the threatening odor context. Additionally, participants rated facial stimuli as more aversive when presented in the threatening odor as compared to the safety odor, indicating that faces acquire hedonic value from the odor they were presented in. Strikingly, angry facial expressions received additional processing resources when presented within a threatening olfactory context, as reflected on verbal reports and electrodermal activity (EDA). This latter finding suggests that threat-related stimuli, here angry faces, are preferentially processed within an olfactory context where a threat might happen. Considering that the hedonic value of an odor may be quite subjective, I conducted a pilot study in order to identify odors with pleasant vs. unpleasant properties for most participants. Seven odors (four pleasant and three unpleasant) were rated with respect to their valence (pleasant vs. unpleasant), arousal (arousing vs. calm), and intensity. Additionally, EDA was measured. Two pleasant (Citral and Eucalyptol) and two unpleasant (“Animalis” and Isobutyraldehyde) odors were chosen from the original seven. The unpleasant odors were rated as more negative, arousing, and intense than the positive ones, but no differences were found regarding EDA. These four odors were subsequently used in a virtual reality (VR) paradigm with two odor attribution groups. Participants of the social attribution group (n = 59) were always passively guided into the same room (an office) towards one out of two virtual agents who were either paired with the pleasant or the unpleasant odor. Participants of the contextual attribution group (n = 58) were guided into one out of two rooms which were either paired with the pleasant or the unpleasant odor and where they always met the same agent. For both groups, the agents smiled, frowned or remained with a neutral facial expression. This design allowed evaluating the influence of odor valence as a within-subjects factor and the influence of odor attribution as a between-subjects factor. Unpleasant odors facilitated the processing of social cues as reflected by increased verbal and physiological arousal as well as reduced active approach behavior. Specific influence of odor valence on emotional facial expressions was found for ratings, EDA, and facial mimicry, with the unpleasant odor causing a levelling effect on the differences between facial expressions. The social attribution group exhibited larger differences between odors than the contextual group with respect to some variables (i.e., ratings and EDA), but not to others (i.e., electrocortical potentials – ERPs – and approach behavior). In sum, unpleasant in comparison to pleasant odors diminished emotional responses during social interaction, while an additional enhancing effect of the social attribution was observed on some variables. Interestingly, the awareness that an interaction partner would smell (pleasantly or unpleasantly) boosted the emotional reactivity towards them. In Study 3, I adapted the VR paradigm to a within-subjects design, meaning that the different attribution conditions were now manipulated block-wise. Instead of an approach task, participants had to move away from the virtual agent (withdrawal task). Results on the ratings were replicated from Study 2. Specifically, the difference between pleasant and unpleasant odors on valence, arousal, and sympathy ratings was larger in the social as compared to the contextual attribution condition. No effects of odor or attribution were found on EDA, whereas heart rate (HR) showed a stronger acceleration to pleasant odors while participants were passively guided towards the agent. Instead of an approach task, I focused on withdrawal behavior in this study. Interestingly, independently of the attribution condition, participants spent more time withdrawing from virtual agents, when an unpleasant odor was presented. In sum, I demonstrated that the attribution of the odors to the social agent itself had an enhancing effect on their influence on social interaction. In the fourth and last study, I applied a similar within-subjects protocol as in Study 3 with an additional Ultimatum Game task as a measure of social interaction. Overall findings replicated the results of Study 3 with respect to HR and EDA. Strikingly, participants offered less money to virtual agents in the bad smelling room than in the good smelling room. In contrast to Study 3, no effects of odor attribution were found in Study 4. In sum, again I demonstrated that unpleasant odor may lessen social interaction not only when the interaction partner smells badly, but also in more complex interaction situations. In conclusion, I demonstrated that hedonic odors in general influence social interaction. Thus, pleasant odors seem to facilitate, while unpleasant odors seem to reduce interpersonal exchanges. Therefore, the present thesis extends the body of literature on the influence of odors on the processing of social stimuli. Although I found a direct influence of odors on social preferences as well as on the physiological and behavioral responses to social stimuli, I did not disentangle impact of odor per se from the impact of the affective state. Interestingly, odor attribution might play an additional role as mediator of social interactions such as odor effects in social interactions might be boosted when the smell is attributed to an individual. However, the results in this regard were less straightforward, and therefore further investigations are needed. Future research should also take into account gender or other inter-individual differences like social anxiety.

Chronic pain is a common problem in clinical practice, not well understood clinically, and frequently tough to satisfactorily diagnose. Because the pathophysiology is so complex, finding effective treatments for people with chronic pain has been overall less than successful and typically reduced to an unsatisfactory trial-and-error process, all of which translates into a significant burden to society. Knowledge of the mechanisms underlying the development of chronic pain, and moreover why some patients experience pain and others not, may aid in developing specific treatment regimens. Although nerve injuries are major contributors to pain chronification, they cannot explain the entire phenomenon. Considerable research has underscored the importance of the immune system for the development and maintenance of chronic pain, albeit the exact factors regulating inflammatory reactions remain unclear. Understanding the putative molecular and cellular regulator switches of inflammatory reactions will open novel opportunities for immune modulatory analgesics with putatively higher specificity and less adverse effects. It has become clear that small, non- coding RNA molecules known as microRNAs are in fact potent regulators of many thousands of genes and possibly cross-communicate between cellular pathways in multiple systems acting as so-called “master-switches”. Aberrant expression of miRNAs is now implicated in numerous disorders, including nerve injuries as well as in inflammatory processes. Moreover, compelling evidence supports the idea that miRNAs also regulate pain, and in analogy to the oncology field aid in the differential diagnosis of disease subtypes. In fact, first reports describing characteristic miRNA expression profiles in blood or cerebrospinal fluid of patients with distinct pain conditions are starting to emerge, however evidence linking specific miRNA expression profiles to specific pain disorders is still insufficient. The present thesis aimed at first, identifying specific miRNA signatures in two distinct chronic pain conditions, namely peripheral neuropathies of different etiologies and fibromyalgia syndrome. Second, it aimed at identifying miRNA profiles to better understand potential factors that differentiate painful from painless neuropathies and third, study the mechanistic role of miRNAs in the pathophysiology of pain, to pave the way for new druggable targets. Three studies were conducted in order to identify miRNA expression signatures that are characteristic for the given chronic pain disorder. The first study measured expression of miR-21, miR-146a and miR-155 in white blood cells, skin and nerve biopsies of patients with peripheral neuropathies. It shows that peripheral neuropathies of different etiologies are associated with increased peripheral miR-21 and miR-146a, but decreased miR-155 expression. More importantly, it was shown that painful neuropathies have increased sural nerve miR-21 and miR-155 expression, but reduced miR-146a and miR-155 expression in distal skin of painful neuropathies. These results point towards the potential use of miRNAs profiles to stratify painful neuropathies. The seconds study extends these findings and first analyzed the role of miR-132-3p in patients and subsequently in an animal model of neuropathic pain. Interestingly, miR-132-3p was upregulated in white blood cells and sural nerve biopsies of patients with painful neuropathies and in animals after spared nerve injury. Pharmacologically modulating the expression of miR-132-3p dose-dependently reversed pain behavior and pain aversion, indicating the pro-nociceptive effect of miR-132-3p in chronic pain. This study thus demonstrates the potential analgesic impact by modulating miRNA expression. Fibromyalgia is associated with chronic widespread pain and, at least in a subgroup, impairment in small nerve fiber morphology and function. Interestingly, the disease probably comprises subgroups with different underlying pathomechanisms. In accordance with this notion, the third study shows that fibromyalgia is associated with both aberrant white blood cell and cutaneous miRNA expression. Being the first of its kind, this study identified miR-let-7d and its downstream target IGF-1R as potential culprit for impaired small nerve fiber homeostasis in a subset of patients with decreased intra-epidermal nerve fiber density. The work presented in this thesis is a substantial contribution towards the goal of better characterizing chronic pain based on miRNA expression signatures and thus pave the way for new druggable targets.

Depressionen und Angststörungen sind die beiden häufigsten psychischen Erkrankungen. Für Angststörungen wurde in zahlreichen Untersuchungen die Bedeutung veränderter Muster in den basalen emotional-assoziativen Lernprozessen für die Ätiologie und Aufrechterhaltung der Erkrankung gezeigt. Hierzu zählen eine verstärkte Akquisitionsreaktion auf den konditionierten Stimulus, Defizite in der Inhibition der Furchtreaktion auf den Sicherheit signalisierenden Stimulus, Übergeneralisierung und Beeinträchtigungen in der Extinktion konditionierter Reaktionen. Aufgrund der hohen Prävalenzen einer Komorbidität mit Depressionen rückte in den letzten Jahren zunehmend die Untersuchung der genannten Prozesse bei Depressionen in den Fokus. Hierfür konnten bisher keine einheitlichen Ergebnisse gezeigt werden. Weiterhin wird der Subtyp der ängstlichen Depression einerseits mit hohen Prävalenzen beschrieben, andererseits zeigen Untersuchungen eine schlechtere Prognose, stärkere Einschränkungen in der Funktionalität und ein schlechteres Ansprechen auf die Therapie im Vergleich zu depressiven Patienten ohne hohes Ängstlichkeitsniveau. In dieser Arbeit wurden die Akquisition, Generalisierung und Extinktion in einem differentiellen Konditionierungsparadigma bei schwer depressiven ängstlichen und nicht ängstlich-depressiven Patienten sowie einer gesunden Kontrollgruppe untersucht. Ängstliche und nicht ängstlich-depressive Patienten zeigten ein beeinträchtigtes Sicherheitslernen in der Akquisition und Beeinträchtigungen in der Extinktion der konditionierten Furcht. Es ergaben sich keine Unterschiede hinsichtlich der Stärke der Generalisierung zwischen Patienten und den gesunden Kontrollen und es konnten keine differenzierenden Muster zwischen den ängstlich- und den nicht ängstlich-depressiven Patienten gezeigt werden. Zusammenfassend weisen die Ergebnisse auf Veränderungen im Furchtlernen bei Patienten mit Depressionen hin. Es konnten keine Belege für unterschiedliche Mechanismen im Furchtlernen von ängstlich- und nicht ängstlich-depressiven Patienten gefunden werden. Unsere Ergebnisse stützen somit die Klassifikation der ängstlichen Depression als Subtyp der Depression. Weiterhin weisen die Ergebnisse der beeinträchtigten Extinktion bei Patienten mit Depressionen darauf hin, dass Expositionselemente, welche bei der Therapie von Angststörungen als Verfahren der Wahl eingesetzt werden, auch bei der Behandlung von Depressionen integriert werden sollten, um so den Therapieerfolg zu verbessern.

Hintergrund: Eine Panikattacke beginnt typischerweise mit der Wahrnehmung einer physiologischen oder psychischen Veränderung, die von der Person als bedrohlich eingestuft wird. Während in klassischen neuroanatomischen Modellen der Panikstörung die Amygdala in der sich anschließenden aufschaukelnden Symptomatik in den Mittelpunkt gestellt wurde, erweitern aktuelle Studien dieses amygdalozentrische Bild und lenken die Aufmerksamkeit auf extratemporale neuronale Netzwerke. Dysfunktionen im neuronalen Aufmerksamkeitsnetzwerk, relevant für die Wahrnehmung und Regulierung exterozeptiver und interozeptiver Prozesse, könnten zur Entstehung einer Panikstörung beitragen. Weiterhin scheinen bestimmte Risikogenotypen für die Panikstörung wie z.B. im Adenosin Rezeptor 2A (ADORA2A) oder dem Neuropeptid S Rezeptor (NPSR1) Gen und die entsprechenden Neurotransmittersysteme in der Regulierung der Aufmerksamkeitsnetzwerke involviert zu sein. Fragestellung: Dysfunktionen im noradrenergen bottom-up Alertingnetzwerk und in der dopaminergen exekutiven top-down Aufmerksamkeitskontrolle könnten in einem neurokognitiven Entstehungsmodell der Panikstörung eine wichtige Rolle spielen. Mit Hilfe funktioneller Bildgebung soll die Funktion des neuronalen Aufmerksamkeitsnetzwerkes in einer nichtklinischen Stichprobe abhängig von genetischen Risikofaktoren und einer klinischen Stichprobe von und nach einer kognitiven Verhaltenstherapie untersucht werden. Methoden: Im nichtklinischen Teil der Untersuchung wurden in Studie 1 47 gesunde Versuchspersonen für die NPSR1 rs324981 Variante stratifiziert rekrutiert. Mittels fMRT wurde die Aktivität des Alertingnetzwerks und des Executive Control Netzwerks auf neuronaler Ebene mit dem Attentional Networt Test (ANT) untersucht. In Studie 2 wurde bei N=65 Versuchspersonen stratifiziert für die ADORA2A rs5751876 Variante als zusätzliches Verhaltensmaß die Fähigkeit zur interozeptiven Wahrnehmung in Bezug zur Konnektivität im insulären Ruhenetzwerk untersucht. Im klinischen Teil der Untersuchung (Studie 3) wurden 44 Patienten mit Panikstörung sowie eine entsprechend große und gematchte Kontrollgruppe rekrutiert. Es wurden fMRT Ruhemessungen vor und nach Abschluss einer kognitiven Verhaltenstherapie erhoben. Als zusätzliches Verhaltensmaß wurde die selbstberichtete Aufmerksamkeitskontrolle zwischen der Patienten- und der Kontrollgruppe verglichen. Ergebnisse: Träger des NPSR1 TT und des ADORA2A TT Risikogenotyps für Angst und Angsterkrankungen zeigten eine erhöhte Aktivität in Teilen des Alertingnetzwerks. Die Aktivität im Executive Control Netzwerk war arealabhängig teilweise erhöht, teilweise reduziert. Innerhalb eines interozeptiven Netzwerks zeigten Träger des ADORA2A TT Genotyps Hinweise auf eine dysfunktionale fronto-striatale-insuläre Interaktion. Im klinischen Teil der Studie zeigten Patienten mit Panikstörung eine reduzierte Konnektivität des dorsolateralen Präfrontalkortex (dlPFC) im fronto-parietalen Aufmerksamkeitsnetzwerk. Die Konnektivität innerhalb dieses Netzwerks korrelierte mit Defiziten in selbstberichteter Aufmerksamkeitskontrolle bei Patienten mit Panikstörung. Nach Abschluss der Therapie zeigte sich bei Patienten, die von der Therapie profitiert hatten, wieder eine Zunahme oder Verbesserung der Konnektivität mit dem dlPFC. Schlussfolgerung: Die Ergebnisse dieser Untersuchung betonen die Rolle dysfunktionaler interozeptiver und exterozeptiver Aufmerksamkeitsnetzwerke in der Entstehung von Angsterkrankungen. Bei Patienten mit Panikstörung sowie gesunden Versuchspersonen mit bestimmten prädisponierenden genetischen Variationen scheint eine Dysbalance des neuronalen Aufmerksamkeitsnetzwerks bzgl. der Abstimmung von bottom-up und top-down Netzwerken vorzuliegen.

Although Social Anxiety Disorder (SAD) is one of the most prevalent mental disorders, still little is known about its development and maintenance. Cognitive models assume that deviations in attentional as well as associative learning processes play a role in the etiology of SAD. Amongst others, deficits in inhibitory attentional control as well as aberrations during fear generalization, which have already been observed in other anxiety disorders, are two candidate mechanisms that might contribute to the onset and retention of SAD. However, a review of the literature shows that there is a lack of research relating to these topics. Thus, the aim of the present thesis was to examine in which way individuals with SAD differ from healthy controls regarding attentional control and generalization of acquired fear during the processing of social stimuli. Study 1 tested whether impairment in the inhibitory control of attention is a feature of SAD, and how it might be influenced by emotional expression and gaze direction of an interactional partner. For this purpose, individuals with SAD and healthy controls (HC) participated in an antisaccade task with faces displaying different emotional expressions (angry, neutral and happy) and gaze directions (direct and averted) serving as target stimuli. While the participants performed either pro- or antisaccades in response to the peripherally presented faces, their gaze behavior was recorded via eye-tracking, and ratings of valence and arousal were obtained. Results revealed that both groups showed prolonged latencies and increased error rates in trials with correct anti- compared to prosaccades. However, there were no differences between groups with regard to response latency or error rates, indicating that SAD patients did not exhibit impairment on inhibitory attentional control in comparison to HC during eye-tracking. Possible explanations for this finding could be that reduced inhibitory attentional control in SAD only occurs under certain circumstances, for example, when these individuals currently run the risk of being negatively evaluated by others and not in the mere presence of phobic stimuli, or when the cognitive load of a task is so high that it cannot be unwound by compensatory strategies, such as putting more effort into a task. As not only deviations in attentional, but also associative learning processes might be pathogenic markers of SAD, these mechanisms were further addressed in the following experiments. Study 2 is the first that attempted to investigate the generalization of conditioned fear in patients with SAD. To this end, patients with SAD and HC were conditioned to two neutral female faces serving as conditioned stimuli (CS+: reinforced; CS-: non-reinforced) and a fearful face paired with a loud scream serving as unconditioned stimulus (US). Fear generalization was tested by presenting morphs of the two faces (GS: generalization stimuli), which varied in their similarity to the original faces. During the whole experiment, self-report ratings, heart rate (HR) and skin conductance responses (SCR) were recorded. Results demonstrated that SAD patients rated all stimuli as less pleasant and more arousing, and overestimated the occurrence of the US compared to HC, indicating a general hyperarousal in individuals with SAD. In addition, ratings and SCR indicated that both groups generalized their acquired fear from the CS+ to intermediate GSs as a function of their similarity to the CS+. However, except for the HR data, which indicated that only SAD patients but not HC displayed a generalization response in this measure, most of the results did not support the hypothesis that SAD is characterized by overgeneralization. A plausible reason for this finding could be that overgeneralization is just a key characteristic of some anxiety disorders and SAD is not one of them. Still, other factors, such as comorbidities in the individuals with SAD, could also have had an influence on the results, which is why overgeneralization was further examined in study 3. The aim of study 3 was to investigate fear generalization on a neuronal level. Hence, high (HSA) and low socially anxious participants (LSA) underwent a conditioning paradigm, which was an adaption of the experimental design used study 2 for EEG. During the experiment, steady-state visually evoked potentials (ssVEPs) and ratings of valence and arousal were recorded. Analyses revealed significant generalization gradients in all ratings with highest fear responses to the CS+ and a progressive decline of these reactions with increasing similarity to the CS-. In contrast, the generalization gradient on a neuronal level showed highest amplitudes for the CS+ and a reduction in amplitude to the most proximal, but not distal GSs in the ssVEP signal, which might be interpreted as lateral inhibition in the visual cortex. The observed dissociation among explicit and implicit measures points to different functions of behavioral and sensory cortical processes during fear generalization: While the ratings might reflect an individual’s consciously increased readiness to react to threat, the lateral inhibition pattern in the occipital cortex might serve to maximize the contrast among stimuli with and without affective value and thereby improve adaptive behavior. As no group differences could be observed, the finding of study 2 that overgeneralization does not seem to be a marker of SAD is further consolidated. In sum, the conducted experiments suggest that individuals with SAD are characterized by a general hyperarousal during the exposition to disorder-relevant stimuli as indicated by enhanced arousal and reduced valence ratings of the stimuli compared to HC. However, the hypotheses that reduced inhibitory attentional control and overgeneralization of conditioned fear are markers of SAD were mostly not confirmed. Further research is required to elucidate whether they only occur under certain circumstances, such as high cognitive load (e.g. handling two tasks simultaneously) or social stress (e.g. before giving a speech), or whether they are not characteristics of SAD at all. With the help of these findings, new interventions for the treatment of SAD can be developed, such as attentional bias modification or discrimination learning.