Motor speed does not impact the drift rate: a computational HDDM approach to differentiate cognitive and motor speed

Abstract

The drift diffusion model (DDM) is a widely applied computational model of decision making that allows differentiation between latent cognitive and residual processes. One main assumption of the DDM that has undergone little empirical testing is the level of independence between cognitive and motor responses. If true, widespread incorporation of DDM estimation into applied and clinical settings could ease assessment of whether response disruption occurs due to cognitive or motor slowing. Across two experiments, we manipulated response force (motor speed) and set size to evaluate whether drift rates are independent of motor slowing or if motor slowing impacts the drift rate parameter. The hierarchical Bayesian drift diffusion model was used to quantify parameter estimates of drift rate, boundary separation, and non-decision time. Model comparison revealed changes in set size impacted the drift rate while changes in response force did not impact the drift rate, validating independence between drift rates and motor speed. Convergent validity between parameter estimates and traditional assessments of processing speed and motor function were weak or absent. Widespread application, including neurocognitive assessment where confounded changes in cognitive and motor slowing are pervasive, may provide a more process-pure measurement of information processing speed, leading to advanced disease-symptom management.

Characterising and differentiating cognitive and motor speed in older adults: a longitudinal birth cohort study

“…This puts a high focus on cognitive abilities, linguistic or visuospatial abilities and lower-order abstraction skills (29). Motor IPS as assessed by RTTs, on the other hand, focuses on the efficiency of speeded motor responses (30). Contrary to the LCT, the behavioural responses in RTTs only involve simple motor processes (i.e., pressing specific buttons with specific fingers), from which reaction times statistics can be measured.…”

Sleep deprivation affects interference control: A diffusion model analysis.

“…Note that there is another way to determine which diffusion model parameters change between experimental conditions (e.g., Koob et al, 2023; Sandry & Ricker, 2022), which relies on model comparisons and does not require statistical analyses on model parameters. To test whether our conclusion is stable and unaffected by the method of how we determine which parameters vary across conditions, we performed model comparisons (see Appendix B) and compared the results of model comparisons with those of fitting the full model allowing all parameters to vary between the two sleep conditions that were reported above.…”

Non-decision time: The Higgs Boson of decision.

“…Only a few of these studies are designed to specifically manipulate sensory and/or motor delays and assess the effect on model parameters. Manipulation of sensory conduction delay (Servant et al, 2016; Weindel et al, 2021), motor difficulty (Sandry & Ricker, 2022; Voss et al, 2004; Weindel et al, 2021), or response modality (Gomez et al, 2015) lead to the expected variations in NDT model estimates but, crucially, changes are also observed in the decision parameters. Reciprocally, when those same studies also manipulate factors expected to selectively modulate decision parameters (speed–accuracy trade-off or task difficulty), changes in NDT are also often reported.…”