Daily Negative Affect and Reaction Time Inconsistency in Emerging Adults: Ecological Momentary Assessment Study

Abstract Background Anxiety and mood disorders, characterized by elevated negative affect (NA) and cognitive impairments, are highly prevalent among college students. Within-person (WP) NA variability, which captures moment-to-moment fluctuations in NA, provides unique insights into emotional processes that are not reflected in mean NA levels. Cognitive variability, particularly reaction time (RT) inconsistency, is increasingly recognized as a sensitive marker of cognitive health and functional integrity. Although prior research links NA to cognitive variability, the short-term dynamics of these associations in naturalistic settings remain understudied. College students provide an ideal population for examining these dynamics using ecological momentary assessment (EMA). Objective This study investigated the association between WP NA and RT inconsistency, hypothesizing that higher WP fluctuations in NA would predict increased RT inconsistency. We also examined the moderating roles of practice effects and covariates, including neuroticism, insomnia, and sex. Methods Using EMA, 99 university students completed morning and evening assessments over 14 days, including a cognitive task measuring RT inconsistency (standard deviation in trial-level RT) and self-reported NA. Multilevel modeling was used to assess WP fluctuations in NA and their impact on RT inconsistency, accounting for time (session number), between-person differences in NA, and covariates such as sleep problems, neuroticism, age, sex, and use of a touch device. Results WP fluctuations in NA significantly predicted increased RT inconsistency (exp(β)=1.022, 95% CI 1.008‐1.037; P =.007), supporting the hypothesis that NA variability disrupts cognitive performance. Male students exhibited lower RT inconsistency than female students, with a small effect size (exp(β)=0.824, 95% CI 0.694‐0.977; P =.049). Finally, EMA sessions were inversely associated with RT inconsistency, with a stronger effect up to session 3 (exp(β)=0.930, 95% CI 0.879‐0.985; P =.03) than after session 3 (exp(β)=0.986, 95% CI=0.979‐0.992; P