Fear conditioning is a form of associative learning by which a neutral stimulus is repeatedly paired with an aversive unconditioned stimulus (US), becoming a conditioned stimulus (CS), which is capable of eliciting a conditioned fear response (CR). Although fear conditioning is an adaptive form of learning, it may become a source of pathology when anxious reactivity to a CS persists in the absence of a CS/US association.
Several fear conditioning processes have been implicated in the pathogenesis of anxiety disorders (Lissek et al., 2005 and Mineka and Zinbarg, 2006). These processes include acquisition, within-session extinction, extinction recall, conditioned inhibition, and conditioned fear-generalization. For example, an enhanced fear acquisition may be characteristic of social phobia (Lissek et al., 2008a), generalized anxiety disorder (GAD; Thayer, Friedman, Borkovec, Johnsen, & Molina, 2000), or posttraumatic stress disorder (PTSD; Orr et al., 2000 and Peri et al., 2000). Impaired within-session fear extinction has been shown for individuals with panic disorder (PD; Michael et al., 2007 and Otto et al., 2014) or GAD (Pitman & Orr, 1986), whereas impaired extinction recall could be present in PTSD (Milad et al., 2008) or obsessive-compulsive disorder (OCD; Milad et al., 2013). Moreover, conditioned inhibition deficits may characterize PTSD (Jovanovic et al., 2009 and Jovanovic et al., 2010). However, these results have not been always replicated. For example, “normal” (i.e., not enhanced) fear acquisition has been reported in PD (Michael et al., 2007), social phobia (Hermann et al., 2002 and Tinoco-González et al., 2014), or GAD (Pitman & Orr, 1986).
Conditioned fear-generalization occurs when fear CRs extend to a range of novel stimuli (generalization stimuli) that resemble the original CS. It can become maladaptive (i.e., excessive) when some of these stimuli are perceived as harmful given the similarity and (non) discriminability to previously learned stimuli (Dunsmoor, Mitroff, & LaBar, 2009).
The possible role of conditioned fear-generalization in pathological anxiety has gained increased research recognition during the last decade (Hajcak et al., 2009, Lissek, Biggs, et al., 2008, Lissek et al., 2010, Lissek et al., 2013, Vervliet et al., 2005 and Vervliet et al., 2004). Indeed, recent etiological accounts of anxiety disorders suggest that conditioned fear-generalization could be a central pathogenic marker of some anxiety disorders (Lissek, 2012), although prospective studies supporting this assumption are lacking. In several case–control studies, individuals with specific anxiety disorders have shown abnormal (i.e., enhanced) conditioned fear-generalization in comparison to healthy controls. This has been the case for PD (Lissek et al., 2010), PTSD (cited in Lissek et al., 2008b), and recently, GAD (Lissek et al., 2013). In GAD, generalization may contribute to an increase in the number of events triggering worry, the hallmark of the disorder (Greenberg, Carlson, Cha, Hajcak, & Mujica-Parodi, 2013). It may also contribute to worry about topics that only have a moderate relatedness with the original triggers (Lissek et al., 2013).
In the aforementioned study, Lissek et al. (2013) compared fear acquisition and conditioned fear-generalization using a validated experimental paradigm (Lissek et al., 2008b) among 22 patients with GAD and 26 healthy controls. GAD patients showed abnormally broad conditioned fear-generalization gradients, compared to controls, as measured by the fear-potentiated startle, despite showing similar fear acquisition. However, in a recent study no evidence for enhanced conditioned fear-generalization in subjective (risk ratings) or autonomic (pupillary response) measures was found in women with GAD in comparison to healthy controls (Greenberg et al., 2013). A limitation of these two studies on conditioned fear-generalization in GAD is that almost 50% of patients had a comorbid anxiety (Lissek et al., 2013) or depressive (Greenberg et al., 2013) disorder. This casts doubts about the specificity of conditioned fear-generalization impairments in GAD. Additionally, studies on the role of certain individual differences variables that are closely related to GAD (e.g., trait-anxiety or trait-worry) in nonclinical or subclinical individuals using fear conditioning paradigms have also provided inconsistent results. For example, high-trait anxiety has been associated with enhanced fear acquisition in some studies (Indovina, Robbins, Núñez-Elizalde, Dunn, & Bishop, 2011), but not in others (Torrents-Rodas et al., 2013). The same is true for trait-worry (Joos et al., 2012 and Otto et al., 2007). These studies used relatively different conditioning tasks and a possible explanation for the inconsistent results is that ambiguous tasks may obscure the effects of trait anxiety (Torrents-Rodas et al., 2013; see Beckers, Krypotos, Boddez, Effting, & Kindt, 2013).
Although fear conditioning processes do not feature prominently among current theoretical explanations of GAD (see Behar et al., 2009 and Newman et al., 2013), some typical GAD characteristics, such as the tendency to interpret ambiguous stimuli as threatening (see Lissek et al., 2013), suggest that these processes may have an important role in the disorder and that more studies in the field are needed. In the present study, we examined conditioned fear-acquisition and fear-generalization in GAD patients using a fear conditioning paradigm assessing fear-potentiated startle (FPS) and subjective responses. We focused our analyses on GAD participants without comorbidity but conducted additional separate analyses in GAD patients with comorbid anxiety disorders. Based on previous research, we predicted that GAD patients would show enhanced conditioned fear-generalization in comparison to healthy controls (Lissek, Biggs, et al., 2008 and Lissek, Levenson, et al., 2008), but we expected that GAD patients and healthy controls would show similar fear acquisition (Lissek, Biggs, et al., 2008 and Lissek, Levenson, et al., 2008).