Math and Verbal Fluency Across Adulthood: Insights into Aging and Individual Differences

A recent publication in Scientific Reports by Yarden Gliksman (Ruppin Academic Center), Itai Schwarz, and Sharon Naparstek (Bar-Ilan University) takes on a question that has received surprisingly little research attention given its everyday relevance: how does mathematical fluency change as we age, and which arithmetic operations are most vulnerable?

Math Fluency in Aging

Cognitive aging research has produced a large literature on memory, executive function, and verbal abilities. Mathematical functioning in later adulthood has had far less of it. That is a gap worth closing, since numeracy sits behind everyday tasks from medication management and financial decision-making to medical risk comprehension. As life expectancy rises and the older adult population grows, knowing which math skills are preserved and which decline becomes more important for research and for clinical practice.

The Study Design

The team examined 52 participants spanning younger and older adults using two measures of math fluency. The first was an oral test, capturing performance under naturalistic verbal-response conditions. The second was the Ben-Gurion University Math Fluency test (BGU-MF), a computerized paradigm previously developed by the lab that measures both accuracy and reaction times across all four basic arithmetic operations: addition, subtraction, multiplication, and division. Verbal fluency was assessed in parallel using semantic and phonological fluency tasks, which allowed a direct comparison between the mathematical and verbal domains.

Key Findings

Several findings stand out:

• Operations are not equally affected by age. Division and subtraction showed the largest age-related differences, while addition was relatively preserved across the lifespan. The dissociation suggests that arithmetic operations rely on partially distinct cognitive mechanisms, with the more demanding operations disproportionately vulnerable to age-related decline.
• Education was the strongest predictor of math performance in older adults. This aligns with the broader cognitive reserve literature, where educational attainment is consistently linked to better cognitive outcomes in later life.
• Time since retirement negatively predicted accuracy and the number of exercises solved. The longer a participant had been retired, the lower their math fluency scores, consistent with the idea that disengagement from cognitively demanding daily work may accelerate certain aspects of cognitive decline.
• Verbal fluency declined with age, but to a lesser extent than math fluency. That points to a broad age-related reduction in fluency abilities, and it suggests that arithmetic processing carries a heightened vulnerability above and beyond domain-general slowing.
• Effects were consistent across formats. The same age differences emerged in both the computerized BGU-MF and the oral math fluency tasks, supporting the validity of computerized fluency testing in older populations.

Implications for Cognitive Aging Research

The dissociation between addition (preserved) and division and subtraction (more vulnerable) reinforces theoretical models in which different arithmetic operations engage distinct retrieval, procedural, and working memory components. The more procedurally demanding operations take the brunt of age-related cognitive slowing. The role of education and retirement timing supports a use-it-or-lose-it framing for cognitive maintenance in later adulthood.

For clinicians and researchers working with older populations, the paper makes the case for including standardized, computerized arithmetic fluency assessments alongside traditional verbal fluency measures, particularly when screening for early cognitive change or characterizing individual differences in cognitive aging trajectories.

Reference

Gliksman, Y., Schwarz, I., & Naparstek, S. (2026). Math and verbal fluency across adulthood provide insights into aging and individual differences. Scientific Reports. https://doi.org/10.1038/s41598-026-41329-1