No robust differences in fear conditioning between patients with fear-related disorders and healthy controls

Highlights

• Patients with fear disorders, PTSD, and controls were exposed to a fear acquisition, immediate extinction and recall task.

• No robust differences in associative fear acquisition, extinction and recall between groups were detected.

• There was anecdotal evidence for altered startle responses during extinction in patients with PTSD.

• Results question the diagnostic validity of abnormalities in fear learning in patients with fear-related disorders.

Abstract

Fear conditioning and extinction serve as a dominant model for the development and maintenance of pathological anxiety, particularly for phasic fear to specific stimuli or situations. The validity of this model would be supported by differences in the physiological or subjective fear response between patients with fear-related disorders and healthy controls, whereas the model's validity would be questioned by a lack of such differences.

We derived pupillometry, skin conductance response and startle electromyography as well as unconditioned stimulus expectancy in a two-day fear acquisition, immediate extinction and recall task and compared an unmedicated group of patients (n = 73) with phobias or panic disorder and a group of patients with posttraumatic stress disorder (PTSD, n = 21) to a group of carefully screened healthy controls (n = 35).

Bayesian statistics showed no convincing evidence for a difference in physiological and subjective responses between the groups during fear acquisition, extinction learning or recall. Only the PTSD subgroup had altered startle reactions during extinction learning.

Our data do not provide evidence for general differences in associative fear or extinction learning in fear-related pathologies and thereby question the diagnostic validity of the associative fear learning model of these disorders.

Introduction

About a 100 years ago, Watson and his colleagues were among the first to experimentally research emotional learning. They were investigating putative processes underlying the development of the rich emotional and behavioural repertoire seen in adults. In their observations, children showed only fundamental modes of emotional response (love, rage, and fear) that were elicited by very few and specific situations. One of their hypotheses was that “by the method of conditioned reflexes, emotional reactions can be called out by situations (stimuli) which do not at first call them out” (Watson & Morgan, 1917). Their most famous experiment (Watson & Rayner, 1920), known as the “Little Albert” study, provided experimental evidence that fear could be conditioned to a previously neutral white rat (conditioned stimulus, CS) by pairing it with a loud noise (unconditioned stimulus, US). Although Watson and Rayner (1920) did not include an explicit extinction phase, they noted that little Albert's fear response diminished after the passage of time and had to be freshened up by presenting the white rat again together with the loud noise. Due to limited time, they could not finish their research on experimental procedures to again remove little Albert's conditioned emotional responses. However, they already suggested repeatedly confronting Albert with the CS, possibly resulting in a fatigue of the original reflex and thereby provided an experimental design to induce extinction learning.

Since then, research on human fear acquisition and extinction learning has extended its focus to investigating the development of pathological fear (Duits et al., 2015; Lissek et al., 2005). Fear acquisition and extinction learning serve as the main experimental models for the acquisition and maintenance of anxiety and fear-related disorders, offering explanatory pathways through which anxiety and fear develop and persist (Milad & Quirk, 2012; Norrholm & Jovanovic, 2018; Pittig, Treanor, LeBeau, & Craske, 2018). Notably, exposure therapy is grounded in the principles of extinction training and is the treatment of choice in most anxiety disorders (Barlow, 2018; Bouton, 2004; Marin et al., 2017; Milad & Quirk, 2012; Tolin, 2010).

During fear acquisition, a formerly neutral stimulus becomes a danger cue (CS+) by pairing it with an aversive event (the US). This leads to the acquisition of a conditioned fear response towards the CS+. In differential fear conditioning, another safe stimulus is presented (CS-) that is never followed by a US. During extinction learning, the CS+ is repeatedly presented without the US, leading to a decrease in the conditioned fear response. Upon re-exposure to the CS + after extinction during a recall session, a return of the conditioned fear response is frequently observed (Vervliet, Baeyens, Van den Bergh, & Hermans, 2013). The strength of human fear and extinction learning is quantified by comparing the response to the CS + to the response to the CS- across fear acquisition, extinction, and recall sessions (Lonsdorf et al., 2017). Physiological responses, such as an increase in pupil dilations, skin conductance response (SCR), or startle electromyography (EMG) are recorded as readouts for a differential fear response. Ratings of US expectancy are typically employed as subjective measure of fear learning (Norrholm et al., 2006).

Theories on pathological fear include abnormalities in conditioned fear as a key etiological feature (Barlow, 2018; Mineka & Oehlberg, 2008; Norrholm et al., 2015). Importantly, fear acquisition and fear extinction are thought to be independent learning processes relying on partly distinct neural structures (Bouton, 2004; Quirk & Mueller, 2008). In differential fear conditioning procedures, pathological fear might therefore be reflected in an increased acquisition of the fear response to the CS+, a generalized fear response to the CS-, reduced inhibitory safety learning during extinction, or a combination thereof (Lissek et al., 2005).

To investigate whether patients with anxiety disorders indeed show differences in fear acquisition or extinction and to summarize the large number of studies on this subject, Duits et al. (2015) conducted a meta-analysis that expanded upon a previous meta-analysis by Lissek et al. (Lissek et al., 2005). In short, patients with anxiety disorders (mostly PTSD) showed a weak-to-moderate increase in fear responding to the conditioned safety cue (d ∼ 0.4). During fear extinction, patients responded more strongly to the unreinforced CS + than control participants, indicating impaired extinction learning (d ∼ 0.35). Although the authors reported no significant moderators of these effects, the effect size for extinction (CS + minus CS-) for studies employing electric shock as the US was around zero. The effect size for other US types was about 0.2, which was driven by three studies using disorder-specific US (e.g., images of phobic objects or situations), that revealed a large effect of d ∼ 0.7 (Lissek et al., 2008; Schweckendiek et al., 2011; Wessa & Flor, 2007). In line with such subtle group differences in differential fear learning, another recent meta-analysis comparing anxious and non-anxious youth also did not detect differences in differential fear acquisition or extinction learning (Dvir, Horovitz, Aderka, & Shechner, 2019). This places doubt on the notion that pathological anxiety is a consequence of a general impairment in extinction learning.

Here we report a comparison of an unmedicated and clinically thoroughly characterized group of patients with acute fear-related disorders without PTSD, a group of PTSD patients and a group of healthy controls, who have never had a fear-related disorder, in a two-day classical fear acquisition, extinction and recall paradigm. We hypothesized that the diagnosis of a fear-related disorder or PTSD is reflected in altered fear- and extinction learning as well as in a different responding during recall. We were specifically interested in various complementary psychophysiological readouts, including pupillometry, the SCR, and startle EMG, allowing us to draw conclusions on the specific mechanisms underlying possible group differences.