Figure 1
Example of a DDM Decision Process. The DDM describes the accuracy and reaction time of decisions using a basic mechanism of evidence accumulation with a drift-diffusion process. In a given trial, an individual will continuously extract noisy sensory evidence from the presented stimulus. This noisy evidence is accumulated over time, pushing a decision variable towards one of two decision boundaries. Once enough evidence has been sampled to push the decision variable across a boundary, a decision is made and the respective response initiated. This process is described by the following parameters: drift rate, which represents the rate of evidence accumulation towards either decision boundary; boundary separation, which represents the distance between the two decision boundaries; bias, which represents a priori bias towards one or another decision boundary at the start of the trial; and non-decision time, which represents time spent on decision-independent processing.
Table 1
Demographics and Characteristics of the Sample.
| TOTAL | MIN–MAX | |
|---|---|---|
| Age, Years | 39 ± 12 | 19–71 |
| Sex | ||
| Female | 73 (49) | – |
| Male | 77 (51) | – |
| OSPAN | 0.88 | 0.29–1 |
| PI-WSUR | ||
| Total | 22 ± 22 | 0–121 |
| Checking | 8 ± 8 | 0–34 |
| Contamination | 9 ± 9 | 0–38 |
| Grooming | 1 ± 3 | 0–12 |
| Obsessional Impulses | 1 ± 4 | 0–30 |
| Obsessional Thoughts | 3 ± 4 | 0–21 |
| Y-BOCS-SR | ||
| Total | 7 ± 6 | 0–22 |
| Obsessions | 4 ± 4 | 0–13 |
| Compulsions | 3 ± 3 | 0–13 |
| Y-BOCS-SR—State | ||
| Total | 3 ± 5 | 0–18 |
| Obsessions | 2 ± 3 | 0–10 |
| Compulsions | 2 ± 2 | 0–16 |
[i] Values are mean ± SD or n (%).
OSPAN, partial unit score for letter recall in the Operation Span task; PI-WSUR, Padua Inventory-Washington State University Revision; Y-BOCS-SR, Yale-Brown Obsessive Compulsive Scale Self Report standard version; Y-BOCS-SR—State, Yale-Brown Obsessive Compulsive Scale Self Report modified to assess state-level symptoms experienced during completion of experimental tasks.
Figure 2
Distributions of Questionnaire Total Scores. Distributions of total scores for the PI-WSUR, standard Y-BOCS-SR (“YBOCS”), and our modified Y-BOCS-SR—State (“YBOCS-State) questionnaires. PI-WSUR scores ranged from 0 to 121 (M = 22, SD = 22). YBOCS scores ranged from 0 to 22 (M = 7, SD = 6). YBOCS-State scores ranged from 0 to 18 (M = 4, SD = 5).
Table 2
Correlations of WM and Psychometric Questionnaires.
| OSPAN | PI-WSUR | YBOCS | YBOCS-STATE | |
|---|---|---|---|---|
| OSPAN | 1 | 0.029 | –0.002 | 0.13 |
| PI-WSUR | 0.029 | 1 | 0.477* | 0.312* |
| YBOCS | –0.002 | 0.477* | 1 | 0.499* |
| YBOCS-State | 0.13 | 0.312* | 0.499* | 1 |
[i] Notes. * = p < .001 (fdr-corrected).
OSPAN = partial unit score for letter recall in the Operation Span task; PI-WSUR = Padua Inventory-Washington State University Revision; YBOCS = Yale-Brown Obsessive Compulsive Scale Self Report standard version; YBOCS-State = Yale-Brown Obsessive Compulsive Scale Self Report modified to assess state-level symptoms experienced during completion of experimental tasks.
Figure 3
Difficulty in the RDMT is Modulated by Uncertainty Level of the Motion Stimulus. Uncertainty level represents motion stimulus coherence (Low, 45% coherence; Medium, 20% coherence; High, 7.5% coherence). Medium uncertainty trials were more accurate than high uncertainty trials (t(149) = 26.7, p = 1.2e–58), and low uncertainty trials were more accurate than medium uncertainty trials (t(149) = 10.6, p = 7.8e–20). Similarly, medium uncertainty trials had shorter reaction times than high uncertainty trials (t(149) = –14.4, p = 6.9e–30), and low uncertainty trials had shorter reaction times than medium uncertainty trials (t(149) = –14.7, p = 6.8e–31).
Table 3
Posterior Median and Central 95% Credible Interval for Main Group-level Parameters in Model 1.
| MEDIAN | LOWER | UPPER | |
|---|---|---|---|
| Drift rate | |||
| Subject, SD | 0.41 | 0.37 | 0.47 |
| Group, Stim uncertainty level | |||
| Low | 1.80 | 1.73 | 1.87 |
| Medium | 1.20 | 1.13 | 1.27 |
| High | 0.41 | 0.34 | 0.47 |
| Group, Diff between uncertainty levels | |||
| Low – Medium | 0.60 | 0.57 | 0.63 |
| Medium – High | 0.80 | 0.77 | 0.82 |
| Boundary separation | |||
| Subject, SD | 0.56 | 0.50 | 0.63 |
| Group, Stim uncertainty level | |||
| Low | 1.93 | 1.84 | 2.02 |
| Medium | 2.15 | 2.06 | 2.24 |
| High | 2.45 | 2.36 | 2.54 |
| Group, Diff between uncertainty levels | |||
| Low – Medium | –0.22 | –0.24 | –0.20 |
| Medium – High | –0.30 | –0.32 | –0.28 |
| Non-decision time | |||
| Subject, mean | 0.38 | 0.36 | 0.40 |
| Subject, SD | 0.13 | 0.11 | 0.14 |
[i] Effects are considered significant when the median 95% CI does not include 0. As such, all of the effects and contrasts listed in Table 2 above are significant.
Figure 4
Posterior Median and 95% CI for Model 3a and Model 3b Regression Coefficients. Panel (A) shows parameters for Model 3a; Panel (B) shows parameters for Model 3b. “Low” = low uncertainty trials at 45% dot motion coherence; “Med” = medium uncertainty trials at 20% coherence; “High” = high uncertainty trials at 7.5% coherence. For each model, the following parameters are shown for each level of stimulus uncertainty: group level drift rate (“Drift”); effects of subject-level scores on the Y-BOCS-SR—State (“YBOCS-State”), Padua Inventory (“PI-WSUR”), and/or the standard Y-BOCS-SR (“YBOCS”) on drift rate (Drift); group level decision boundary separation (“Boundary”); effects of subject-level scores on the YBOCS-State, PI-WSUR, and/or YBOCS on boundary separation (Boundary). Also shown is the group level mean for non-decision time. The median estimate for each contrast’s posterior distribution is represented by a black dot, and the 95% CI is represented by a red line. An effect is considered significant if the 95% CI does not overlap with 0.
Figure 5
Contrasts of YBOCS-State versus YBOCS or PI-WSUR Drift Rate Coefficients for Models 3a and 3b (Posterior 95% CI). “Low” = low uncertainty trials at 45% dot motion coherence; “Med” = medium uncertainty trials at 20% coherence; “High” = high uncertainty trials at 7.5% coherence. Differences (“Contrasts”) between the posterior distribution of the drift rate regression coefficients for YBOCS-State versus PI-WSUR (Model 3a; coral color), and for YBOCS-State versus the standard YBOCS (Model 3b; cyan color) across uncertainty levels. Contrasts were computed by subtracting the corresponding distributions from one another. The median estimate for each contrast’s posterior distribution is represented by a black dot, and the 95% CI is represented by a red line. A contrast is considered significant if the 95% CI does not overlap with 0.
Figure 6
Posterior Median and 95% CI for Model 5 Regression Coefficients. Effects of subject-level scores on drift rate. “Low” = low uncertainty trials at 45% dot motion coherence; “Med” = medium uncertainty trials at 20% coherence; “High” = high uncertainty trials at 7.5% coherence. The median estimate for each parameter/regression coefficient’s posterior distribution is represented by a black dot, and the 95% CI is represented by a red line. An effect is considered significant if the 95% CI does not overlap with 0.