This study was conducted at the Anxiety Disorders Clinic, an outpatient research clinic, at the New York State Psychiatric Institute (NYSPI) and Columbia University Medical Center, and was approved by the NYSPI Institutional Review Board. Following informed consent, 67 unmedicated participants diagnosed with an anxiety disorder or OCD completed questionnaires and a differential fear conditioning paradigm followed by extinction. Participants were randomized to either standard extinction or NFE. Participants returned to the lab 24 h later to test post-extinction recall and reinstatement [skin conductance in response to conditioned stimulus (CS)].

Participants were recruited via advertisements and referrals from physicians. Participants included adults aged 18–50 with a DSM-5 diagnosis of OCD, social anxiety disorder (SAD), generalized anxiety disorder (GAD), and/or specific phobia (SP). Diagnoses were made by trained clinicians (e.g., doctoral student with master’s degree, clinical psychologist, or psychiatrist) using a structured clinical interview (SCID; First and Spitzer, 1996). Participants were excluded if they used psychotropic medication in the last 4 weeks (8 weeks for selective serotonin reuptake inhibitors), had current diagnosis of major depressive disorder, or lifetime diagnosis of any psychotic disorder, bipolar disorder, or alcohol, or substance use disorder. Other exclusions included: acute suicidal risk; major medical or neurological problems that might interfere with study procedures or data interpretation or increased risk of participation (assessed via brief meeting with MD or Nurse Practitioner, e.g., cardiovascular disease, seizure disorder, head trauma); inability to refrain from caffeine (for 4 h) or nicotine (for 24 h) without withdrawal symptoms. Medication and drug use was established via self-report in the brief meeting with MD or Nurse Practitioner.

Following informed consent, participants completed a series of questionnaires, including the 21-item Beck Depression Inventory II (BDI-II; Beck et al., 1996) to assess depressive symptomatology, the 40-item State-Trait Anxiety Inventory (STAI; Spielberger et al., 1983) to assess both state (current, in the moment) and trait (how one typically feels) anxiety, and the Intolerance of Uncertainty Scale (IUS; scores range from 27 to 135; Buhr and Dugas, 2002) to assess response to uncertainty.

Participants completed the acquisition, extinction, and extinction recall, and reinstatement task over the course of 2 days. PsyLab (Contact Precision Instruments, Cambridge, MA, USA) was used to collect skin conductance data and run the acquisition, extinction, and recall tasks. On Day 1, two Ag/AgCl electrodes were affixed to the hypothenar eminence of the participants’ left hand, to assess skin conductance, and two Ag/AgCl electrodes were affixed to the participants’ right wrist, to deliver shocks. SignaGel Electrode Gel (Parker Laboratories, Inc., Fairfield, NJ, USA), a highly conductive saline gel, was used¹. The shock was generated by the SHK1 Pain Stimulation Shocker (Contact Precision Instruments, Cambridge, MA, USA) and lasted 200 ms. A shock work-up was conducted to determine the level of shock that each participant found to be highly annoying, but not painful (Boucsein et al., 2012).

Following the work-up, participants were instructed to sit comfortably and pay attention to the images displayed on the screen. They were told they may or may not receive shocks, and that there would be an association between pictures and shocks, but they would need to learn it themselves. Participants wore headphones (Sennheiser PRO, Sennheiser electronic GmbH & Co. KG, Wennebostel, Wedemark, Germany) to block out noise and to deliver the novel tone to participants randomized to the NFE group.

Conditioned stimuli (CS) included two angry male faces (following Dunsmoor et al., 2015)². Each trial included a CS displayed for 6 s, followed by a 12 s intertrial interval. The trial order was pseudorandomized so that no more than three trials of the same type occurred in a row. The conditioning session began with 10 habituation trials (five each of CS+ and CS-) to diminish initial orienting responses. This was followed by the first run of fear conditioning that included four CS+ trials that co-terminated with a shock to the wrist, seven CS+ trials unpaired with shock, and seven CS- trials. The second run of fear-conditioning included four CS+ trials paired with shock, eight CS+ trials unpaired with shock, and eight CS- trials. Conditioning was identical between groups. One rationale for using partial CS-US pairing is that continuous (100%) CS-US pairing rates can lead to a rapid decrease in conditioned responses (Grady et al., 2016) that would potentially obscure the effect of the extinction manipulation.

Following conditioning, the standard extinction group (EXT) underwent two runs of extinction that each included 10 CS+ trials unpaired with shock and 10 CS- trials. Following Dunsmoor et al. (2015), there was a very short pause (less than 1 min) between the first and second run of fear conditioning and extinction. For participants randomized to the NFE group, all CS+ extinction trials co-terminated with a low-volume 440-Hz tone for 1.5 s, delivered binaurally through headphones. Both the 200 ms shock (during acquisition for both groups) and the 1.5 s tone (during extinction for the NFE group) co-terminated with CS+. Thus, the onset of the tone during extinction preceded the onset of the shock during acquisition. The dB level of the tone was not recorded. The tone was meant to be perceptible but not loud or aversive. All of the subjects in the NFE condition were asked at the end of the experiment if they heard the tone, and all reported that they did.

Participants returned to the lab the following day. No new instructions were given for Day 2 that would indicate any departure from the procedures from the previous day. Electrodes were reattached and shock intensity was set at the level determined on Day 1. The recall included 10 CS+ trials unpaired with shocks (or tones) and 10 CS- trials. After these 20 CS trials, participants received three unsignaled shocks to the wrist to reinstate conditioned responses. The reinstatement test included 10 CS+ trials unpaired with shock and 10 CS- trials. See Figure 1 for an illustration of the phases of the experiment.

Skin conductance responses (SCR) were collected using PsyLab (Contact Precision Instruments, Cambridge, MA, USA) at 500 Hz. Responses were calculated according to previous criteria (Dunsmoor et al., 2015) using a validated automated MATLAB (Mathworks, Inc., Natick, MA, USA) script (Green et al., 2014). CS+ trials paired with the shock during fear conditioning were not included in the analysis to avoid potential confounds introduced by the electrical shock. An SCR was considered related to the CS if the trough-to-peak deflection occurred within a 0.5–6.0 s time window starting at CS onset, if the responses lasted between 0.5 and 5.0 s, and if the response was greater than 0.2 microsiemens. If a response on a trial did not meet these criteria, it was scored as zero. Raw SCR values were square-root transformed to normalize the distribution. To account for different patterns of SCR over the course of learning, we divided the data into early and late phases. Early and late fear acquisition were defined as the first and second run of conditioning, respectively. Early extinction was defined as the first run of extinction. Late extinction was defined as the last 3 CS+ and 3 CS- trials, in order to capture responses toward the end of training (Dunsmoor et al., 2015). Early recall and reinstatement were likewise defined as the first three CS+ and three CS- trials, as prior research shows that extinction manipulations tend to affect early presentations of CS+ at extinction-retention tests (Milad et al., 2009; Schiller et al., 2010; Dunsmoor et al., 2015).

Following prior fear conditioning studies (Duits et al., 2016), repeated measures Analyses of Variance (ANOVAs) were run separately for each phase of the experiment (early acquisition, late acquisition, late extinction, early recall, early reinstatement). In each ANOVA, Group (EXT vs. NFE) was included as a between-subjects factor and stimulus (CS+, CS-) was included as a within-subjects factor. We followed-up significant two-way interactions with t-tests. Statistical significance was defined as p < 0.05.

Sixty-seven participants provided consent. Three were excluded from the analysis [did not return for the second day of study (n = 1), no longer met inclusion criteria (n = 1), technical errors (n = 1)]. An additional 13 participants were not included in the analysis due to a failure to show evidence of conditioned learning, as defined by no positive difference between mean SCRs to the CS+ vs. CS- during late conditioning (i.e., a difference of 0 or a greater mean SCRs to the CS- than CS+).

This resulted in 51 participants (60.80% female, M age = 25.23, SD age = 4.82, age-range = 18–41) included in the analysis: 27 in the EXT group and 24 in the NFE group. Participants in the two groups did not differ significantly in demographic or clinical characteristics (see Table 1 for M, SD, and tests for differences between groups).

As expected, STAI scores suggest elevated state (M = 44.14, SD = 9.39) and trait anxiety (M = 52.20, SD = 9.53) in both groups. These scores are approximately one and two standard deviations, respectively, above scores of healthy individuals (Spielberger et al., 1983). IUS scores reveal elevated levels of intolerance of uncertainty (M = 77.98, SD = 17.75); this mean score is approximately three standard deviations above a college sample without anxiety disorders (Freeston et al., 1994). BDI-II scores suggest minimal depression (M = 14.24, SD = 8.81).

The sample was free from psychotropic medications at the time of the study. Most participants (37 participants) had never used psychotropic medications. Of the 14 participants that had, weeks since the last dose ranged from 5 to 150.86 (M = 84.43, SD = 52.71), except for one participant who had a 2 mg dose of clonazepam 3 weeks prior to consent.

See Figure 2 for trial by trial SCR to CS+ and CS-, separated by Group. See Figure 3 for means and standard errors of the means for each phase of the experiment, separated by Group. ANOVAs for early and late fear acquisition revealed significant effects of Stimulus (CS+, CS-), demonstrating that participants acquired conditioned SCRs to the CS+ compared to the CS- (early fear conditioning: F_(1,49) = 23.31, p < 0.001, ηp2 = 0.32; late fear conditioning: F_(1,49) = 19.56, p < 0.001, ηp2 = 0.29). There was no effect of Group for early fear acquisition (F_(1,49) = 0.49, p = 0.487, ηp2 = 0.01), but results revealed a Stimulus by Group (NFE, EXT) interaction for early fear acquisition (F_(1,49) = 4.72, p = 0.035, ηp2 = 0.09). Follow-up paired sample t-tests and visual inspection of means demonstrated that both groups successfully acquired conditioned fear during early fear conditioning (EXT: t₍₂₆₎ = 2.40, p = 0.024, d = 0.25; NFE: t₍₂₃₎ = 4.07, p < 0.001, d = 0.52), although the difference in response to CS+ to CS- was larger for the NFE group than the EXT group. For late fear acquisition, there was no effect of Group (F_(1,49) = 1.55, p = 0.219, ηp2 = 0.03) or Stimulus by Group interaction (F_(1,49) = 2.10, p = 0.154, ηp2 = 0.04), suggesting that during the second run of conditioning, participants successfully acquired similar levels of conditioned fear, regardless of group.

Contrary to expectations, the ANOVA for late fear extinction revealed a significant effect of Stimulus (F_(1,49) = 4.17, p = 0.047, ηp2 = 0.08), indicating that SCRs remained elevated to the CS+ compared to the CS-. That is, participants still evinced heightened SCRs to the CS+ vs. the CS- in the last few trials of extinction, despite having been presented with 17 preceding CS+ trials during extinction without shock. Notably, there was no main effect of Group (F_(1,49) = 0.12, p = 0.724, ηp2 = 0.002) or Stimulus by Group interaction (F_(1,49) = 0.46, p = 0.502, ηp2 = 0.01). Overall, the extinction results suggest that conditioned fear was not extinguished in this anxiety sample, regardless of group.

Interpretation of the 24-h recall and reinstatement tests is complicated by the fact that neither group fully extinguished on Day 1. The ANOVA for 24-h recall revealed a significant effect of Stimulus (F_(1,49) = 38.63, p < 0.001, ηp2 = 0.44), with no effect of Group (F_(1,49) = 0.03, p = 0.863, ηp2 = 0.001) or stimulus by Group interaction (F_(1,49) = 3.88, p = 0.055, ηp2 = 0.07). For the reinstatement test, there was no effect of stimulus (F_(1,49) = 4.02, p = 0.050, ηp2 = 0.08), Group (F_(1,49) = 0.23, p = 0.637, ηp2 = 0.01), or stimulus by Group interaction (F_(1,49) = 0.01, p = 0.910, ηp2 < 0.001).

We also conducted an additional test to determine if there was an increase in differential responding from the end of extinction to the beginning of the 24-h recall test, despite incomplete extinction across participants. Specifically, we conducted a repeated measures ANOVA with Group (EXT vs. NFE) as a between subjects factor, and stimulus (CS+, CS-) and Phase (Late Extinction, Early Recall) as within subjects factors. Results revealed no effect of Group (F_(1,49) = 0.10, p = 0.752, ηp2 = 0.002), Stimulus by Group interaction (F_(1,49) = 0.31, p = 0.580, ηp2 = 0.01), Phase by Group interaction (F_(1,49) = 0.03, p = 0.870, ηp2 = 0.001), or Phase by Stimulus by Group interaction (F_(1,49) = 2.67, p = 0.109, ηp2 = 0.05). Additionally, there was no correlation between IUS and spontaneous recovery (as measured by SCR to CS+ during early trials of 24 h recall) in the entire sample (r₍₄₉₎ = 0.10, p = 0.508), or in either group (EXT group: r₍₂₅₎ = 0.23, p = 0.242; NFE group: r₍₂₂₎ = −0.13, p = 0.545).

At the end of the experiment, we asked participants how often the shock followed the CS+ and CS- face. 86.3% of subjects correctly responded that the CS+ face was sometimes paired with shock, and 72.5% correctly responded that the CS- face was never paired with shock. We did not exclude subjects for incorrect contingency awareness, given that the question was asked retrospectively at the end of the experiment on Day 2.