Motor Inhibitory Latency Predicts Sequential Risk Architecture: Dimensional Cognitive Chronometry Across Healthy and Clinical Cohorts

Categorical psychiatric taxonomies assume that clinical diagnoses reflect discrete neurocognitive dysfunctions. Substantial comorbidity and within-category heterogeneity challenge this assumption, supporting the Research Domain Criteria (RDoC) focus on continuous dimensions. We investigated whether continuous motor inhibitory chronometry predicts sequential risk choice architecture across psychiatric cohorts. Analyzing the UCLA Consortium for Neuropsychiatric Phenomics dataset (N = 257: 120 healthy controls, 49 schizophrenia spectrum, 48 bipolar disorder, and 40 ADHD), we evaluated Stop-Signal Task reaction time (SSRT) and rapid-pacing Balloon Analogue Risk Task (BART) performance using hierarchical regression, Huber robust M-estimation, bootstrap resampling (B = 10,000), and Bayesian model comparison. Categorical DSM diagnoses explained minimal variance in sequential risk tolerance (R² = 2.25%, F(3, 253) = 1.94, p = 0.123), with Bayesian comparison yielding extreme evidence for categorical null equivalence (BF₀₁ = 208.51). In contrast, continuous motor braking latency robustly predicted reduced sequential risk tolerance (Model 3: R² = 7.78%, SSRT b = -0.9925, p = 0.0034; Median Go RT b = -0.512, p = 0.0301). Downweighting heavy-tailed clinical residuals via Huber robust M-estimation confirmed this negative association (b_robust = -0.7535, z = -2.58, p = 0.0100; 95% BCa bootstrap CI [-2.031, -0.012]). Diagnostic moderation remained an underpowered exploratory trend (ΔR² = +2.53%, p = 0.076). Slower motor braking predicts heightened risk aversion rather than reckless disinhibition, reflecting frontostriatal pacing efficiency and decisional hesitation under compounding uncertainty. Quantitative behavioral chronometry provides an objective transdiagnostic endophenotype capturing executive variations obscured by categorical diagnostic boundaries. Results: Surface-level risk propensity showed no significant differences across categorical diagnoses (Model 0: F(3, 253) = 1.94, p = 0.123; mean pumps = 4.73 to 4.91). Bayesian model comparison provided extreme evidence for categorical null invariance (BF01 = 208.51; Lee & Wagenmakers, 2013). In the full additive specification, continuous motor inhibitory latency predicted reduced sequential risk tolerance (Model 3: R² = 7.78%, Adj R² = 5.57%, ΔR² = +3.22%, SSRT b = -0.9925, p = 0.0034; Median Go RT b = -0.512, p = 0.0301). Exploratory diagnostic moderation modeling (Model 4: ΔR² = +2.53%, ΔF(3, 247) = 2.320, p = 0.076) revealed a trend-level interaction pattern: while stratified subgroup regressions demonstrated robust negative coupling in healthy controls (b = -2.421, SE = 0.678, p = 0.0005) and an attenuated slope across psychiatric cohorts (ADHD: b = -0.170, p = 0.838; Bipolar: b = -0.130, p = 0.849; Schizophrenia: b = -0.730, p = 0.213), the omnibus interaction remained marginal, requiring validation in higher-powered samples. Omitting an extreme high-leverage case (Participant #113, D113 = 0.305; Cook, 1977) reduced standard OLS to trend level (b = -0.560, p = 0.0778), but Huber robust M-estimation confirmed a significant negative slope (brobust = -0.7535, SE = 0.2924, z = -2.58, p = 0.0100, 95% CI [-1.33, -0.18]), corroborated by 10,000 bootstrap resamples (95% BCa CI [-2.03, -0.01], pboot = 0.047).