About HumanEval OS
Standardized, millisecond-precision cognitive benchmarks measuring the apex of human nervous system velocity, working memory, and motor dexterity.
Scientific Integrity
Every battery in HumanEval translates established neuropsychological literature—including Miller's 7±2 Law, simple reaction latency paradigms, and Kyoto University's Ayumu working memory experiments—into high-framerate web software.
Empirical Quantile Norms
Raw scores are mapped directly onto empirical population quantile distributions derived from 80M+ benchmark trials and psychometric literature, accounting for the natural ex-Gaussian positive skew of motor and reflex latencies.
HumanEval explicitly differentiates computerized task performance from general intelligence (g). Decades of double-blind RCTs demonstrate that training on computerized paradigms produces strong near-transfer, but rarely general far-transfer. We refuse to make ungrounded marketing claims about “multiplying IQ” or “rewiring your brain.” We measure and train specific, isolatable mental faculties: working memory gating, frontoparietal inhibitory control, task-set switching latency, and probabilistic calibration.