Reactivation of the Unconditioned Stimulus Inhibits the Return of Fear Independent of Cortisol

Abstract

Reconsolidation is the post-retrieval stabilization of memories, a time-limited process during which reactivated (i.e., retrieved) memories can be updated with new information, become stronger or weaker, depending on the specific treatment. We have previously shown that the stress hormone cortisol has an enhancing effect on the reconsolidation of fear memories in men. This effect was specific, i.e., limited to the conditioned stimulus (CS) that was reactivated, and did not generalize to other previously reinforced, but not reactivated CS. Based on these results, we suggested that cortisol plays a critical role in the continuous strengthening of reactivated emotional memories, contributing to their persistence and robustness. In the current study, we aimed to achieve a more generalized reconsolidation enhancement using an alternative reactivation method, i.e., by a low-intensity unconditioned stimulus (UCS) presentation instead of the more common unreinforced CS presentation. In previous studies, UCS reactivation was shown to lead to a more generalized reconsolidation effect. Therefore, we hypothesized that the combination of cortisol treatment and UCS reactivation would lead to an enhanced fear memory reconsolidation, which would generalize from previously reinforced CS to stimuli that resemble it. We tested 75 men in a 3-day fear conditioning paradigm: fear acquisition training on day 1; UCS reactivation/no reactivation and pharmacological treatment (20 mg hydrocortisone/placebo) on day 2; extinction training, reinstatement and test (of original and modified stimuli) on day 3. In contrast to our hypothesis, UCS reactivation prevented the return of fear [observed in skin conductance responses (SCR)] regardless of the pharmacological manipulation: while reinstatement to the original CS was found in the no-reactivation group, both reactivation groups (cortisol and placebo) showed no reinstatement. As the only methodological difference between our previous study and the current one was the reactivation method, we focus on UCS reactivation as the main explanation for these unexpected findings. We suggest that the robust prediction error generated by the UCS reactivation method (as opposed to CS reactivation), combined with the lower UCS intensity, has by itself weakened the emotional value of the UCS, thus preventing the return of fear to the CS that was associated with it. We call for future research to support these findings and to examine the potential of this reactivation method, or variations thereof, as a tool for therapeutic use.

Keywords: extinction; fear conditioning; glucocorticoids; reconsolidation; reinstatement; retrieval.

Figure 1

Experimental timeline. The testing was conducted on three consecutive days separated by 24-h intervals: fear acquisition training on day 1; pharmacological treatment and memory reactivation on day 2; extinction training, reinstatement and reinstatement test on day 3. The procedure was identical for the three groups on days 1 and 3 and differed between the three groups only on day 2, in which memory was either reactivated (RE+CORT, RE groups) or not reactivated (CORT group) following the administration of cortisol (RE+CORT, CORT) or placebo (RE). Skin conductance responses (SCR; illustrated by the palm) served as a measure of conditioned fear, and were recorded during all experimental phases. Seven saliva samples (illustrated by the saliva collecting devices) were used to assess salivary cortisol, and were collected during the three experimental days. CS, conditioned stimulus; CSM, conditioned stimulus modified. Bolts, representation of the unconditioned stimulus (UCS).

Figure 2

Day 1: Fear acquisition training. Mean (± SEM) SCR to the three conditioned stimuli (mean of 10 trials for CS+ and CS−). SCR to the reinforced CS+ are significantly higher (RE+CORT: 0.16 ± 0.03; RE: 0.17 ± 0.03; CORT: 0.24 ± 0.04) than the SCR to the unreinforced CS− (RE+CORT: 0.10 ± 0.02; RE: 0.14 ± 0.02; CORT: 0.18 ± 0.03; ***p < 0.001), demonstrating successful fear learning. The figure presents all three groups combined (N = 75; RE+CORT group: n = 25, RE group: n = 25, CORT group: n = 25) as no interaction and no main effect of group were found.

Figure 3

Day 3: Fear extinction training took place on the third testing day. Here, both conditioned stimuli (CS+, CS−) were presented (both not reinforced) for 10 times each. This graph illustrates mean (± SEM) SCR to both stimuli at early extinction (trials 1–5) vs. late extinction (trials 6–10). The significant difference between CS+ and CS− during early extinction (*p < 0.05) indicates fear memory retrieval (early extinction CS+: RE+CORT: 0.13 ± 0.03; RE: 0.14 ± 0.03; CORT: 0.15 ± 0.04; early extinction CS−: RE+CORT: 0.12 ± 0.03; RE: 0.11 ± 0.02; CORT: 0.12 ± 0.02); the significant reduction of response to the CS+ (**p < 0.01) and the lack of difference between the stimuli at the later phase indicate successful fear extinction (late extinction CS+: RE+CORT: 0.08 ± 0.02; RE: 0.13 ± 0.03; CORT: 0.09 ± 0.02; late extinction CS−: RE+CORT: 0.09 ± 0.03; RE: 0.14 ± 0.03; CORT: 0.12 ± 0.03). As no interaction or main effect of group were found, the graph presents the groups combined (N = 74; RE+CORT group: n = 24, RE group: n = 25, CORT group: n = 25).

Figure 4

Day 3: Analyses of the reinstatement test to the original stimuli compared the mean (± SEM) SCR to the conditioned stimuli (CS+, CS−) in the late phase of extinction (mean of the last five trials; late extinction CS+: RE+CORT: 0.08 ± 0.02; RE: 0.13 ± 0.03; CORT: 0.09 ± 0.02; late extinction CS−: RE+CORT: 0.09 ± 0.03; RE: 0.14 ± 0.03; CORT: 0.12 ± 0.03) and the first trial after reinstatement (first reinstatement trial CS+: RE+CORT: 0.10 ± 0.03; RE: 0.17 ± 0.04; CORT: 0.21 ± 0.05; first reinstatement trial CS−: RE+CORT: 0.06 ± 0.02; RE: 0.14 ± 0.04; CORT: 0.06 ± 0.03) in the three experimental groups (N = 73; RE+CORT group: n = 24, RE group: n = 25, CORT group: n = 24). In the CORT group, differential reinstatement (i.e., higher SCR after the reinstatement shocks only to CS+) was found (*p < 0.05). No reinstatement was found in the two reactivation groups, RE+CORT and RE (all p < 0.01).

Figure 5

Day 3: Analyses of the reinstatement test to the modified stimuli compared the mean (± SEM) SCR to the stimuli in the late phase of extinction (mean of last 5 trials of the original CS+ or CS−; late extinction original CS+: RE+CORT: 0.08 ± 0.02; RE: 0.13 ± 0.03; CORT: 0.09 ± 0.02; late extinction original CS−: RE+CORT: 0.09 ± 0.03; RE: 0.14 ± 0.03; CORT: 0.12 ± 0.03) and the first trial after reinstatement (modified stimulus, CS+M or CS−M; first reinstatement trial CS+M: RE+CORT: 0.11 ± 0.03; RE: 0.18 ± 0.04; CORT: 0.18 ± 0.04; first reinstatement trial CS−M: RE+CORT: 0.16 ± 0.05; RE: 0.12 ± 0.04; CORT: 0.18 ± 0.05). The analysis revealed a generalized reinstatement (i.e., to both CS+M and CS−M; **p = 0.005). Since no group effect or interaction were found, the figure presents all three experimental groups combined (N = 73; RE+CORT group: n = 24, RE group: n = 25, CORT group: n = 24).