Table 1
Demographic Characteristics.
| EXCLUSION STATUS | STUDY | DEMOGRAPHIC AND CLINICAL MEASURES | ||||
|---|---|---|---|---|---|---|
| N | SEX (% FEMALE) | AGE (SD) | GAD-7 (SD) | PHQ-8 (SD) | ||
| Pre-Exclusion | Study 1 | 47 | 28 (59.6%) | 22.8 (1.5) | 4.7 (4) | 5.1 (4.3) |
| Study 2 | 68 | 43 (64.2%) | 23.1 (1.7) | 4.7 (4.8) | 5.4 (5.5) | |
| Study 3 | 120 | 83 (69.2%) | 22.8 (1.7) | 6.5 (5.3) | 6.3 (5.1) | |
| Combined | 235 | 154 (65.8%) | 22.9 (1.7) | 5.6 (4.9) | 5.8 (5.1) | |
| Between Group Comparison | H = 1.49, p = 0.476 | H = 1.5, p = 0.472 | H = 7.17, p = 0.028 | H = 3.68, p = 0.159 | ||
| Medium-Exclusion | Study 1 | 35 | 21 (60%) | 22.8 (1.5) | 4.9 (3.6) | 5 (3.9) |
| Study 2 | 37 | 24 (64.9%) | 23.1 (1.7) | 6.3 (5.3) | 7.2 (5.8) | |
| Study 3 | 73 | 51 (69.9%) | 22.8 (1.6) | 6.3 (5.3) | 5.8 (4.5) | |
| Combined | 145 | 96 (66.2%) | 22.8 (1.6) | 6 (5) | 6 (4.8) | |
| Between Group Comparison | H = 1.06, p = 0.588 | H = 0.7, p = 0.705 | H = 0.76, p = 0.685 | H = 2.31, p = 0.316 | ||
| Pre-Medium Exclusion Comparison | W = 16898, p = 0.938 | W = 16061, p = 0.846 | W = 15402, p = 0.363 | W = 15589.5, p = 0.47 | ||
| Strict-Exclusion | Study 1 | 0 | – | – | – | – |
| Study 2 | 29 | 18 (62.1%) | 22.9 (1.6) | 6.2 (5.3) | 7.2 (6.3) | |
| Study 3 | 59 | 43 (72.9%) | 22.7 (1.6) | 6.7 (5.6) | 6.1 (4.6) | |
| Combined | 88 | 61 (69.3%) | 22.8 (1.6) | 6.6 (5.4) | 6.4 (5.2) | |
| Between Group Comparison | H = 1.06, p = 0.304 | H = 0.16, p = 0.686 | H = 0.08, p = 0.776 | H = 0.13, p = 0.715 | ||
| Medium-Strict Exclusion Comparison | W = 6181.5, p = 0.625 | W = 6304, p = 0.745 | W = 6060, p = 0.52 | W = 6153.5, p = 0.649 | ||
[i] Bold ~ p < 0.05.
Kruskal-Wallis test used for between group comparisons; Wilcoxon rank-sum test used for exclusion comparisons.
GAD-7 – Generalized Anxiety Disorder seven item scale; PHQ-8 – Patient Health Questionnaire eight item scale.
Figure 1
Task design. The upper panel illustrates the format of each fear conditioning trial. A fixation cross is used for the inter-trial interval (ITI). CS is presented on screen, before a US expectancy scale is presented for participants to enter their response. If the trial is reinforced with a US, this is played through participant headphones following the US expectancy rating screen. The next ITI and trial then commences. The lower panel illustrates the specific design and trial order of the paradigm in this experiment. The acquisition phase includes twenty four trials, twelve each of CS+ and CS– trials. These are presented in a pseudorandom order. CS+ is reinforced with US on 75% of occasions, i.e. nine of the twelve CS+ trials, also in a pseuodorandom order. Participants must essentially discern which CS is threatening, and which is safe. An extinction phase of thirty six trials (eighteen CS+ and eighteen CS–) follows the acquisition phase after a ten minute break. Here participants must learn that the threat CS is now safe.
Figure 2
Model specification and fit. Expectancy Rating (ER); Conditioned Stimulus (CS); Unconditioned Stimulus (US); Associative Value (V); Prediction Error (PE); Learning Rate (LR). An illustration of how the trial-by-trial structure of the fear conditioning paradigm conforms to the generative Rescorla Wagner model proposed to model underlying cognitive behaviour. A Bayesian plate diagram for the base Rescorla Wagner model (1) is depicted. As a hierarchical model, group level hyperpriors (mean, μ, standard deviation, σ) influence subject level parameters, in this model the learning rate (LR) and lapse, in the larger box. These subject level parameters determine the conditional variables (V, p) within the smaller box, representing an individual trial.
Table 2
Correlations of fear conditioning descriptive measures with anxiety severity, and depression (n = 145).
| PHASE | DESCRIPTIVE MEASURE | GAD-7 | PHQ-8 | ||
|---|---|---|---|---|---|
| ρ [95% CI] | P-VALUE | ρ [95% CI] | P-VALUE | ||
| Acquisition | CS+ Whole Phase Mean | –0.05 [–0.22, 0.12] | 0.56 | 0.00 [–0.18, 0.17] | 0.97 |
| CS– Whole Phase Mean | 0.20 [0.02, 0.35] | 0.02 | 0.25 [0.09, 0.40] | 0.00 | |
| CS Discrimination | –0.16 [–0.32, 0.01] | 0.06 | –0.17 [–0.33, 0.00] | 0.04 | |
| Extinction | CS+ Whole Phase Mean | 0.18 [0.00, 0.34] | 0.03 | 0.21 [0.03, 0.36] | 0.01 |
| CS– Whole Phase Mean | 0.16 [–0.01, 0.33] | 0.06 | 0.20 [0.06, 0.36] | 0.01 | |
| CS Discrimination | 0.16 [–0.00, 0.32] | 0.06 | 0.08 [–0.09, 0.24] | 0.34 | |
[i] GAD-7 – Generalized Anxiety Disorder seven item scale; PHQ-8 – Patient Health Questionnaire eight item scale.
Figure 3
Posterior predictive check. A) Group level posterior predictive check (PPC) comparing model generated data from the winning five-learning rate model, to real participant data. The shaded area represents the 50% highest density interval (HDI), with the solid red line indicating the median generated expectancy rating. B) A correlation matrix of generated whole phase means (y axis) to real whole phase means (x axis). The left to right diagonal compares like with like, indicating a better fit in the extinction phase compared to more volatile acquisition phase.
Table 3
Learning rate correlations with anxiety severity, in both medium and strict excluded samples.
| SPEARMAN CORRELATIONS ρ [95% CI] BETWEEN MODEL PARAMETERS AND MEASURES | ||||
|---|---|---|---|---|
| PARAMETER | GAD-7 | PHQ-8 | ||
| MEDIUM (N = 145) | STRICT (N = 88) | MEDIUM (N = 145) | STRICT (N = 88) | |
| Acquisition CS+ Learning Rate (US+) | –0.07 [–0.25, 0.09] | –0.15 [–0.37, 0.06] | –0.03 [–0.19, 0.13] | –0.01 [–0.21, 0.19] |
| Acquisition CS+ Learning Rate (US–) | –0.10 [–0.27, 0.07] | –0.30 [–0.51, –0.08] | –0.06 [–0.21, 0.10] | –0.20 [–0.40, 0.03] |
| Acquisition CS– Learning Rate | –0.22 [–0.35, –0.07] | –0.32 [–0.49, –0.14] | –0.14 [–0.30, 0.02] | –0.14 [–0.33, 0.05] |
| Extinction CS+ Learning Rate | –0.21 [–0.37, –0.04] | –0.33 [–0.52, –0.13] | –0.23 [–0.38, –0.06] | –0.26 [–0.43, –0.04] |
| Extinction CS– Learning Rate | –0.07 [–0.23, 0.09] | –0.15 [–0.34, 0.07] | –0.10 [–0.27, 0.07] | –0.13 [–0.34, 0.08] |
| Lapse Rate | –0.06 [–0.23, 0.13] | –0.13 [–0.36, 0.07] | –0.16 [–0.31, 0.02] | –0.22 [–0.41, –0.02] |
| CS+ Initial Value | 0.05 [–0.12, 0.21] | 0.08 [–0.12, 0.29] | 0.04 [–0.11, 0.21] | 0.02 [–0.19, 0.24] |
| CS– Initial Value | 0.04 [–0.13, 0.20] | –0.07 [–0.29, 0.13] | 0.13 [–0.04, 0.29] | 0.00 [–0.21, 0.21] |
| Extinction CS– Jump Value | –0.06 [–0.22, 0.12] | –0.08 [–0.30, 0.15] | –0.07 [–0.22, 0.09] | –0.07 [–0.27, 0.13] |
[i] Bold ~ p < 0.05.
GAD-7 – Generalized Anxiety Disorder seven item scale; PHQ-8 – Patient Health Questionnaire eight item scale.
Table 4
Learning rate parameters correlated with the residualised scores of GAD-7 and PHQ-8 (removing shared variance), and the shared variance between the GAD-7 and PHQ-8 scales.
| PARAMETER | MEDIUM CRITERIA (N = 145) | STRICT CRITERIA (N = 88) | ||||
|---|---|---|---|---|---|---|
| GAD-7 RESIDUAL | PHQ-8 RESIDUAL | SHARED VARIANCE | GAD-7 RESIDUAL | PHQ-8 RESIDUAL | SHARED VARIANCE | |
| Acquisition CS+ Learning Rate (US+) | –0.03 [–0.21, 0.15] | 0.02 [–0.13, 0.18] | –0.05 [–0.22, 0.11] | –0.16 [–0.39, 0.08] | 0.13 [–0.10, 0.34] | –0.07 [–0.27, 0.16] |
| Acquisition CS+ Learning Rate (US–) | –0.06 [–0.24, 0.12] | –0.01 [–0.17, 0.16] | –0.08 [–0.23, 0.09] | –0.21 [–0.41, 0.02] | 0.03 [–0.19, 0.25] | –0.26 [–0.45, –0.05] |
| Acquisition CS– Learning Rate | –0.14 [–0.29, 0.04] | 0.01 [–0.16, 0.19] | –0.18 [–0.34, –0.02] | –0.28 [–0.46, –0.07] | 0.16 [–0.04, 0.36] | –0.22 [–0.41, –0.02] |
| Extinction CS+ Learning Rate | –0.06 [–0.22, 0.10] | –0.09 [–0.26, 0.08] | –0.24 [–0.39, –0.09] | –0.17 [–0.38, 0.05] | –0.03 [–0.25, 0.19] | –0.31 [–0.48, –0.12] |
| Extinction CS– Learning Rate | 0.06 [–0.10, 0.21] | –0.09 [–0.24, 0.07] | –0.10 [–0.25, 0.07] | 0.05 [–0.16, 0.25] | –0.08 [–0.30, 0.13] | –0.16 [–0.35, 0.04] |
[i] Bold ~ p < 0.05.
Values show Spearman correlation ρ [95% CI]. GAD-7 and PHQ-8 correlations ~ Medium ρ = 0.71, Strict ρ = 0.74.
GAD-7 – Generalized Anxiety Disorder seven item scale; PHQ-8 – Patient Health Questionnaire eight item scale.
Table 5
Steiger’s Z Test Results.
| COMPUTATIONAL PARAMETER | DESCRIPTIVE WHOLE PHASE MEAN | T [95% CI] | P-VALUE |
|---|---|---|---|
| Acquisition CS+ Learning Rate (US+) | Acquisition CS+ | 0.28 [0.00, 0.56] | 0.78 |
| Acquisition CS+ Learning Rate (US–) | Acquisition CS+ | 0.44 [0.00, 0.88] | 0.66 |
| Acquisition CS– Learning Rate | Acquisition CS– | 0.23 [0.00, 0.46] | 0.82 |
| Extinction CS+ Learning Rate | Extinction CS+ | 0.81 [0.00, 1.62] | 0.42 |
| Extinction CS– Learning Rate | Extinction CS– | –1.51 [–3.02, –0.00] | 0.13 |
[i] Steiger’s Z direct comparison of correlations between anxiety severity and computational parameters, against anxiety severity and descriptive whole phase mean measures.