Our study grew from an interest in moving beyond the simple question of whether people perform more slowly when they feel fatigued. Instead, we wanted to examine the dynamics of fatigue and cognitive performance as they unfold in daily life. In particular, we focused on what we call the capacity to “override” fatigue: moments when cognitive performance is better than would be expected given a person’s current level of fatigue.

Answering this question required more data than any single study could readily provide. We therefore conducted an individual participant data coordinated analysis using six intensive longitudinal studies. Participants repeatedly reported their fatigue and completed a brief Symbol Search task on smartphones as they went about daily life. We applied the same analytical framework separately to each study and then meta-analyzed the estimates, allowing us to identify replicated patterns while preserving meaningful differences among studies.

An important part of the work happened before any analysis began. To assemble the collection of studies, I contacted several principal investigators and asked whether they would be willing to share their datasets. Each dataset had its own access requirements, and the relevant data-access protocols had to be completed before the data could be used. The investigators kindly made their data available and also shared valuable insights about the populations, measures, and study contexts represented in their datasets.

This made the project more than an effort to increase statistical power. Bringing together the datasets also brought together researchers with expertise in different clinical and non-clinical populations, all connected by a common interest in fatigue. Their knowledge helped us interpret differences across studies with greater care. The coordinated analysis thus combined data with the contextual expertise needed to understand them.

Across the six studies, greater-than-usual fatigue and faster-than-usual processing speed each predicted a reduced capacity to override fatigue before the next assessment. The fatigue finding was intuitive: when people already felt more fatigued than usual, they appeared to have less capacity to sustain performance relative to their fatigue later. The processing-speed finding was more surprising. Faster performance may sometimes reflect greater investment of resources in the present. Much like a runner maintaining an unusually fast pace early in a race, a person performing faster than usual may have fewer resources available later.

These findings do not mean that fatigue and processing speed are interchangeable. Their concurrent association was negative but small: people tended to perform slightly more slowly when they felt more fatigued, but high-fatigue moments were not necessarily low-performance moments. This distinction may help explain why both greater fatigue and faster performance could independently signal less capacity to override fatigue later.

Several findings generalized across clinical and non-clinical populations. The concurrent relationship between fatigue and processing speed and the effect of current fatigue on subsequent fatigue override were notably consistent despite differences in samples and designs. Clinical or mixed samples, however, showed a more negative ratio of change in processing speed to change in fatigue, suggesting greater sensitivity to fatigue changes. This aligns with research on fatigability in multiple sclerosis, fibromyalgia, and diabetes, although further research is needed before drawing causal conclusions.

Why these findings matter

The results may contribute to research on how to schedule rest and sustain performance among workers, students, athletes, and people living with chronic conditions. Most importantly, they point toward possible early-warning signals. Feeling more fatigued than usual may indicate that the capacity to maintain performance is likely to decline over the next several hours. Scheduling rest or reducing energy expenditure at that point-before performance becomes harder to sustain-could be more useful than waiting for clear impairment to appear.

The phrase “more fatigued than usual” is important. Some fatigue is normal, and the same absolute level may have different meanings for different people. Interventions may therefore benefit from responding to changes relative to an individual’s own typical level rather than relying solely on a universal cutoff. This could inform more personalized approaches to activity pacing and energy conservation, areas in which intervention findings have often been mixed. Future studies should directly test whether momentary, individualized recommendations improve fatigue or functional outcomes.

Faster-than-usual processing speed may provide another useful signal. If periods of especially strong performance require increased resource investment, they may not be indefinitely sustainable. Recognizing this could help people reserve their highest-performance periods for tasks that most require them, such as complex decisions, important meetings, or demanding academic work, and plan recovery around those periods.

Our conclusions remain bounded by the available data. The fatigue questions, Symbol Search procedures, and assessment schedules varied somewhat across studies; only datasets available for secondary analysis could be included; and we could not consistently examine sleep, despite its importance for both fatigue and cognition. Future work should distinguish general, mental, and physical fatigue and incorporate repeated sleep measures.

Still, the consistency of several effects across diverse studies suggests that everyday changes in fatigue and performance contain information about what may happen next. Just as importantly, the project demonstrates the value of coordinated analysis: sharing data did not simply create a larger evidence base- it created a collaboration in which different forms of population-specific expertise could converge on a common scientific question.