Experimental validation of graded arithmetic and linguistic working-memory tasks using behavioral and cerebral hemodynamic measures
Graded cognitive tasks are relevant only when increasing nominal difficulty produces measurable changes in performance and a convergent physiological response. This study evaluated a five-level arithmetic paradigm and a five-level linguistic paradigm designed to engage working memory through separate exposure, retention, and response intervals. Thirty-eight healthy young adults completed both paradigms. Behavioral manipulation validity was assessed from accuracy and completion time; linguistic item-level responses were additionally modeled to detect nonlinear floor effects. Bilateral middle cerebral artery blood-flow velocity was recorded with transcranial Doppler ultrasonography. Baseline-normalized mean velocity change was the primary physiological endpoint, with positive and net area under the curve, peak response, time to peak, and early slope as secondary endpoints. Accuracy declined and completion time increased strongly across levels in both domains, with a substantially steeper accuracy decline in the linguistic paradigm. Although the levels were nominally ordered, the higher linguistic levels did not produce equidistant increases due to the ground effect. Transcranial Doppler models showed significant level-dependent changes in mean velocity, positive and net area under the curve, time to peak, and early slope. However, domain-by-level and domain-by-level-by-hemisphere interactions were not significant. Behavioral and physiological measures therefore converged mainly at the level of the graded experimental manipulation, rather than as distinct domain-specific or lateralized hemodynamic signatures. These findings lend support to paradigms as a multimodal framework for manipulating cognitive load experimentally, while also indicating upper-level language saturation and physiological quality-control decisions that should guide further improvement.