Exploring the relationship between mean performance and within-person variability on smartphone-based cognitive testing in adults across the lifespan

Cognitive performance is classically measured through measures of central tendency. However, intraindividual cognitive variability (IIV) also holds important information about cognitive functioning. Smartphone-based ecological momentary cognitive testing (EMCT) can capture IIV across days. This study examined predictors of IIV, including demographics, affect, and mean performance, in EMCTs completed on the NeuroUX platform among two US-based samples: adults and older adults with high rates of mild cognitive impairment. The adult sample (<i>n</i> = 375) completed eight EMCTs assessing memory, processing speed, executive functioning, and working memory; each test was self-administered five times over ten days. The older adult sample (<i>n</i> = 94) completed three EMCTs assessing memory, processing speed/executive function, and working memory at three difficulty levels; each test at each difficulty level was self-administered five times over thirty days. Mean performance demonstrated the strongest association with IIV across groups. In the adult group, better mean performance was associated with less variability on tests of memory, executive functioning, and two out of three tests of processing speed. On tests of working memory, better mean performance was associated with greater variability, possibly due to the difficulty of these measures at higher performance levels. In the older adult sample, better mean performance was associated with two of three versions of the memory and working memory EMCTs and all difficulty levels of the processing speed/executive function task. Better average performance was associated with greater consistency across most measures. Broader and more diverse data collection methods like EMCT can provide valuable insights into cognitive functioning beyond traditional mean-based measures.