Cognitive performance, particularly in the domains of memory and executive functioning (EF), have been shown in previous research to predict decline in everyday functioning in older adults. The goal-control model posits that episodic memory difficulties cause weak or decaying task goals that lead to the omission of every day task steps (low accomplishment). EF difficulties preclude control over the execution of task goals that lead to inefficient and error-prone performance (high errors). Intraindividual variability (IIV) in neuropsychological test performance has been proposed as a noninvasive early marker of dementia and has utility in cognitively healthy older adults. In this study we examined cross-sectional relations between within-domain IIV in memory and attention/EF with performance of everyday tasks in the lab. We expected greater memory IIV to be associated with task accomplishment (goal decay), and greater attention/EF IIV to be associated with errors (poor control over goals).

40 cognitively normal (CN) older adult (65+) participants (M age=71.44, SD=10.62; 73.8% women; 85.7% White; M education=18.55, SD=8.38) completed the Naturalistic Action Test (NAT), requiring completion of standardized, everyday tasks (i.e., breakfast, lunch) and scored for accomplishment of steps and various errors (micro-errors - misreach to distractor object, extra actions; overt -sequence, perseverative behaviors, etc.; motor errors). Within-domain IIV measures were calculated for 6 memory measures (HVLT-R, BVMT-R) and 6 attention/EF measures (Digit Span, Trail Making Test, Salthouse Letter/Pattern Comparison). First, raw scores for each test were z-transformed, then the intraindividual standard deviation of all z-scores was calculated. Bivariate Spearman’s rank-order correlations were used to examine associations between NAT performance and within-domain IIV, as well as mean performance. Linear regressions were used to examine the associations of IIV score with NAT scores, adjusting for age, sex, and mean cognitive performance.

Among CN participants, higher memory IIV was significantly associated with lower NAT accomplishment (r=-.329, p=.038), and better mean memory performance was significantly associated with fewer errors (micro-errors r=-.509, p=.003; overt r=-.438, p=.012; motor errors r=-.463, p=.008). Regression models revealed that memory IIV, after controlling for mean memory performance, age, and sex, did not significantly predict NAT performance. High attention/EF IIV was significant associated with more errors (overt r=.377, p=.016), whereas mean attention/EF performance was not significantly correlated with any NAT measures. Attention/executive function IIV significantly predicted errors (micro-errors B=4.15, p=.03; Overall model: R2=0.285, F(4, 24)=2.393, p=.079; overt B=.562, p=.032; Overall model: R2=0.371, F(4, 24)=3.543, p=.021) after adjusting for mean attention/executive function performance, age, and sex.

Consistent with the goal-control model framework, greater variability in memory was associated with lower task accomplishment, whereas greater variability in attention/EF was associated with more errors. However, only attention/EF IIV predicted NAT performance, specifically errors (micro-errors, overt errors), after adjusting for age, sex, and mean attention/EF performance. Within-domain IIV can be used to predict mild functional difficulties in cognitively healthy older adults. Future research should examine within-domain IIV in a larger sample with more diversity to maximize generalizability, and in a longitudinal design to evaluate within-domain IIV predictive validity for cognitive/functional impairment.