Human Random Number Generation (RNG) engages several components of executive functions, including working memory, response inhibition, and cognitive flexibility. Human-generated numerical sequences can be conceptualized as bio cognitive signals with chaotic properties. The complexity reduction hypothesis suggests that aging is associated with a decline in the dynamical complexity of cognitive systems. The present study aimed to quantify these dynamical changes in healthy young and older adults using nonlinear dynamical analyses.

Thirty healthy participants, including 15 young adults and 15 older adults, were recruited. Each participant performed a Random Number Generation task producing a sequence of 200 numbers (1–9). The resulting time series were analyzed using phase space reconstruction. Measures from Recurrence Quantification Analysis (RQA) and fractal analysis were extracted. Group classification was performed using a Support Vector Machine (SVM).

Compared with the young group, the older group showed significantly higher values of Determinism (DET), Mean Diagonal Line Length (L), and Laminarity (LAM), indicating increased repetitive patterns and reduced behavioral variability. The correlation dimension and Lyapunov exponent were significantly lower in the older group, suggesting reduced dynamical complexity and lower sensitivity to initial conditions. The Hurst exponent was higher in the older group, reflecting stronger long range correlations and decreased randomness. The classification accuracy between the two groups reached 91.3%.

These findings provide quantitative evidence supporting the complexity reduction hypothesis in cognitive aging. Given its simplicity and low cost, this approach may offer a promising basis for developing cognitive screening tools for older adults.