Daily life fluctuations in affect predict within-person changes in a real-world measure of cognitive processing speed

Background: Considerable research points to a deleterious effect of negative affect oncognition. However, most evidence comes from experimental induction paradigms thatcannot separate between- and within-person processes and have unclear implications forcognitive performance in the real world. Ecological momentary assessment (EMA)studies can address this gap, allowing us to generate time-series data that can distinguishtrait and state-like mechanisms, provide temporal evidence for causation, and bridge thedivide between lab and life. Here, we developed a microlongitudinal design to examinethe between- and within-person (contemporaneous and time-lagged) relationshipsbetween affect and a real-world measure of cognitive processing speed. Methods: We analyzed three separate EMA datasets. Across these studies, a total of 914 participants(70.89% female) between 18 and 82 years tracked negative and positive affect 2 or 3times daily, for 6 or 8 weeks, completing between 63 and 126 assessments each. Weused a recently validated method to derive cognitive processing speed from digitalquestionnaire response time (DQRT). Multilevel vector autoregressive models were usedfor analysis. Results: Affect and DQRT were related between-person; people with higheraverage negative affect were slower in responding to survey items overall; with theopposite for positive affect (partial correlation range: -0.299 – 0.541, P-FDR-corrected <0.05). This was observed for 36/37 affective items assessed. At the within-person level,DQRT was slower when negative affect increased and positive affect decreased (partialcorrelation range: -0.135 – 0.129, P-FDR-corrected < 0.05; significant in 34/37 items). Inlagged analyses, higher negative affect (and lower positive affect) predicted slower DQRTat the next time point (5 to 12 hours later) (β estimate range: -0.027 to 0.067, P-FDR-corrected < 0.05) for 27/37 items. The strongest predictors of future DQRT were feelingsof worry and anxiety and there was no evidence for reverse temporal causation.Conclusion: We identified a potential causal relationship where negative affect predictsslower survey completion times. This finding may inform mechanistic accounts ofcognitive deficits in mental health disorders. Future work should examine how theseresults compare to more standard tasks for assessing information processing.