Contextual control over expression of fear is affected by cortisol

Abstract

At the core of anxiety disorders is the inability to use contextual information to modulate behavioral responses to potentially threatening events. Models of the pathogenesis of anxiety disorders incorporate stress and concomitant stress hormones as important vulnerability factors, while others emphasize sex as an important factor. However, translational basic research has not yet investigated the effects of stress hormones and sex on the ability to use contextual information to modulate responses to threat. Therefore, the purpose of the present study was threefold: first, we aimed at developing an experimental paradigm specifically capable of capturing contextual modulation of the expression of fear. Second, we tested whether cortisol would alter the contextualization of fear expression. Third, we aimed at assessing whether alterations in contextualization due to cortisol were different for men and women. Healthy participants (n = 42) received placebo or hydrocortisone (20 mg) prior to undergoing a newly developed differential contextual fear-conditioning paradigm. The results indicated that people rapidly acquire differential contextual modulation of the expression of fear, as measured by fear potentiated startle (FPS) and skin conductance responses (SCR). In addition, cortisol impaired the contextualization of fear expression leading to increased fear generalization on FPS data in women. The opposite pattern was found in men. Finally, as assessed by SCR, cortisol impaired differential conditioning in men. The results are in line with models suggesting heightened vulnerability in women for developing anxiety disorders after stressful events.

Keywords: anxiety disorders; context; cortisol; fear potentiated startle; fear-conditioning; sex differences.

Figure 1

(A) Experimental conditioned stimuli and relevant comparisons. An image of a living room and an image of a garden constituted the safe (context B) and threat (context A) contexts, while two angry faces constituted the CS+ (CS1) and CS− (CS2). Several comparisons of interest concerning fear contextualization and fear generalization exist. A(CS1)+ > B(CS2)− denotes differential conditioning; A(CS1)+ > B(CS1)− denotes proper fear contextualization; B(CS1) > B(CS2) denotes impaired contextualization; A(CS1)+ > A(CS2)− denotes less fear generalization via the context; A(CS2)− > B(CS2)− denotes fear generalization via the context. (B) Timeline of an exemplary trial. Each trial started with the onset of a context (duration: 25 s), in which after 11 ± 1 s a face appeared (duration: 8 s) and again disappeared. After face offset, the context remained onscreen for another 6 s (± 1), followed by a variable inter trial interval (ITI: 9 ± 1 s). If a context startle probe was presented, it occurred after 6 ± 1 s relative to context onset. During every CS presentation a CS probe was presented after 7.5 s. ITI probes were presented in the middle of the ITI trial. (C) US-expectancies. Mean expectancy scores of the unconditioned stimulus as a function of stimulus type across all groups. Error bars represent standard errors of the mean. All participants were completely aware of the experimental contingencies.

Figure 2

Startle magnitudes. Averages of noise alone (NA) trials derived from the same block as the CS presentations were subtracted from average startle potentiation during CS presentations. Blocks consisted of two trials of each CS type. Graphs depict data as a function of Sex and Drug. Error bars represent standard errors of the mean.

Figure 3

Fear contextualization and fear generalization. Deviations from linearity reflect a significant quadratic component in generalization gradients, and are to be expected in the case of proper fear contextualization. (Panel A) Shows that men having received the cortisol pill showed proper fear contextualization (significant quadratic component), which was absent after placebo administration, resulting in increased fear generalization (absence of significant quadratic component). (Panel B) Shows that women having received the placebo pill showed proper fear contextualization (significant quadratic component), which was absent after cortisol administration, resulting in increased fear generalization (absence of significant quadratic component). Significant quadratic components in generalization gradients are depicted by ^*p < 0.05; or ^***p < 0.000.

Figure 4

Skin conductance responses. Skin conductance data were range corrected and subsequently log transformed. Blocks consist of averages of two trials of each CS type. Graphs depict data as a function of Sex and Drug. Error bars represent standard errors of the mean.