Research on attention, interruption, and cognitive performance has been accumulating for decades. But productivity content tends to treat findings selectively — citing the parts that support a framework while ignoring complications and caveats.
This article aims to present the relevant science accurately, including where evidence is strong, where it is contested, and where popular claims outrun the underlying data.
What We Know About Sustained Attention
The Ericsson Research on Deliberate Practice
Anders Ericsson, the psychologist whose research on expert performance is frequently (and often inaccurately) cited in productivity writing, documented something relevant to focus measurement: the daily practice ceiling.
In studies of world-class musicians, chess players, and other experts, Ericsson and colleagues found that deliberate practice — cognitively demanding, effortful improvement work — was typically capped at around four hours per day, even among the most advanced performers. Attempts to extend deliberate practice beyond this threshold produced declining quality rather than proportional output.
This finding applies to deep work measurement in a practical way: a target of more than four deep hours per day is likely to produce diminishing returns even for highly skilled practitioners. Self-reported deep hours above this threshold warrant scrutiny — either the definition of “deep” is too broad, or the quality is declining as the hours extend.
The four-hour ceiling is an average across studies, not a hard physiological law. Individual variation is significant. But it is a useful empirical anchor for setting realistic deep hour targets.
Ultradian Rhythms and Work Intervals
Sleep researcher Nathaniel Kleitman identified the ultradian rhythm — a roughly 90-minute cycle of arousal and recovery that governs sleep stage transitions. Subsequent researchers (including Peretz Lavie and, more recently, Andrew Huberman in synthesizing this literature) proposed that similar cycles continue during waking hours, cycling between higher and lower states of cognitive alertness.
If the ultradian framework is correct, it would suggest that approximately 90-minute work intervals are better aligned with biological rhythms than the arbitrary 25-minute Pomodoro intervals. It would also suggest that the natural troughs in the cycle — roughly every 90 minutes — are appropriate rest and recovery windows.
Important caveat: the application of ultradian rhythm research to knowledge work scheduling is more extrapolation than direct evidence. The original research established rhythms in physiological arousal; the translation to “therefore you should work in 90-minute blocks” involves assumptions that have not been directly tested in controlled workplace studies. The framework is plausible and directionally useful, but treat it as a hypothesis rather than an established principle.
The Interruption Research: What Gloria Mark’s Work Actually Found
Gloria Mark’s research at UC Irvine is probably the most-cited in productivity writing on interruptions. The numbers quoted — “it takes 23 minutes to refocus after an interruption” — have become so widely repeated that they are often presented without qualification.
What the research actually found requires more nuance.
Mark and colleagues ran observational studies in workplaces, tracking how people used their time and how long it took them to return to a primary task after an interruption. In her earlier studies, the figure cited was approximately 23 minutes — but this was the time to return to the original task, not necessarily the time to return to the pre-interruption level of cognitive engagement. Recovery time also varied substantially by interruption type, task complexity, and the individual worker.
Later Mark research (including the 2014 study “Bored Mondays and Focused Afternoons” and subsequent work on self-interruption) found that people interrupt themselves nearly as often as they are interrupted by external factors. The 2008 study co-authored with Daniela Gudith and Ulrich Klocke found that people can compensate for interruptions by working faster after returning, but at a measurable cost in increased stress.
The core finding — that interruptions carry significant recovery costs and that those costs accumulate across a workday — is robust. The specific “23 minutes” figure should be understood as a rough order-of-magnitude, not a precise measurement that applies uniformly to every interruption.
For focus measurement purposes, what matters is the directionality: distraction count per hour is a meaningful metric because each distraction carries a non-trivial cognitive cost that compounds across a session.
Attention Residue: Sophie Leroy’s Contribution
Sophie Leroy’s 2009 research on “attention residue” added an important mechanism to the interruption literature. Leroy found that when people switch from one task to another before completing the first, part of their attention remains on the incomplete task — creating a cognitive load that reduces performance on the new task.
This has direct implications for session completion rate as a metric. Sessions that end early — even if the person moves to another task that seems productive — carry an attention residue cost. Completing planned sessions matters not just for cognitive output within the session but for the quality of whatever comes next.
Leroy’s research supports the idea that session completion rate is a leading indicator of cognitive performance, not just a compliance measure.
The Limits of Self-Report
Self-report is a foundational method in psychological research and the primary data source for the Focus Dashboard metrics. It has known limitations that are worth being honest about.
Recall bias. People do not accurately remember how they spent time. Laura Vanderkam’s time diary research and the American Time Use Survey both show consistent overestimation of time spent working and underestimation of leisure time. For focus logging specifically, this means you should record sessions immediately after completing them, not at end of day.
Mood-congruent bias. How you feel when you log a session influences how you rate it. A session that went reasonably well may be rated 2 if you are in a negative mood when you log it, and 3 if you are in a positive mood. This introduces noise into quality ratings, though it is less severe for distraction counts (which you can tally as you go rather than rate retrospectively).
Consistency decay. People log more diligently in the early weeks of a tracking practice and less diligently as the novelty fades. Gaps in logging are not random — they cluster around low-motivation periods, which are exactly the periods you most want to capture. For AI analysis, this means your data set is likely biased toward better-than-average sessions.
None of these limitations invalidate self-report as a method. They mean you should interpret trends rather than absolute numbers, log immediately rather than retrospectively when possible, and view your dataset as an approximation rather than a precise record.
What Cognitive Science Says About Measuring Attention Directly
Laboratory attention measurement typically uses well-validated instruments: sustained attention tasks (like continuous performance tests), EEG-based measures of neural activity, or behavioral measures like response accuracy and reaction time.
These methods have good validity in controlled settings. They do not translate cleanly to real-world knowledge work. A 30-minute sustained attention test in a lab measures something different from 90 minutes of writing a technical document in an open-plan office. The cognitive demands, the social context, and the motivational factors are all different.
Consumer-grade EEG devices (Muse, for example) use brain electrical activity to infer cognitive states, but the inference from consumer EEG to meaningful “focus score” involves substantial signal processing and interpretation that has not been independently validated against knowledge work output quality. The devices are potentially useful for biofeedback and relaxation training; their utility for measuring focus during complex knowledge work is less established.
Wearable physiological sensors (Oura, Whoop) measure heart rate variability and related signals that correlate with autonomic arousal and recovery state. HRV has a reasonably good evidence base as a marker of recovery and readiness. Its utility as a real-time focus measurement tool is more limited — it tells you something about your physiological readiness, not your cognitive engagement during a specific session.
What This Means for Your Tracking Practice
The research supports a modest conclusion: no current method measures focus directly and reliably in real-world knowledge work settings.
What you can measure are behavioral proxies — session duration, completion, and self-reported distraction — that correlate imperfectly but meaningfully with actual cognitive performance. The key word is imperfectly. Treat your Focus Dashboard metrics as informative signals, not ground truth.
The practical upshot: self-reported session logs are the best available tool for personal focus improvement, not because they are perfect, but because they capture cognitive engagement at a level of specificity that no automated tool can match, and because their imperfections are predictable enough to work around with consistent practice.
The research base on attention and interruption consistently supports the three metrics in the Focus Dashboard. Deep hours per day reflects the evidence on sustainable cognitive work volume. Session completion rate reflects the attention residue literature. Distraction count per hour reflects the interruption recovery research.
None of them are perfect. All of them are worth tracking.
Begin with the one that feels most immediately useful — for most people, that is distraction count per hour, because it is the most granular and the most directly actionable within a single session.
Related: Complete Guide to Focus Metrics and AI · Why Focus Scores Are Misleading · Focus Metrics Framework with AI
Tags: focus research, attention science, Gloria Mark, Ericsson deliberate practice, ultradian rhythms, distraction research
Frequently Asked Questions
-
What does research say about how long people can sustain focused attention?
Research on expert deliberate practice by Anders Ericsson found that even elite performers rarely maintain peak cognitive effort beyond four hours per day. For most knowledge workers, sustained high-quality deep work above three to four hours is unusual and requires deliberate recovery management. -
How long does it take to recover from an interruption, according to research?
Gloria Mark's research at UC Irvine found it took an average of over 20 minutes to return to a pre-interruption level of engagement after a significant interruption. Shorter disruptions also carry meaningful recovery costs, though smaller ones. -
What is the ultradian rhythm and how does it affect focus?
The ultradian rhythm is a roughly 90-minute biological cycle identified by sleep researcher Nathaniel Kleitman. During waking hours, cognitive alertness appears to follow similar cycles, with peaks and troughs suggesting that 90-minute work intervals may be better aligned with biological rhythms than arbitrary shorter intervals. -
Is there scientific evidence that self-report is accurate for tracking focus?
Self-report has known biases — people rate recent events differently than distant ones, and mood affects retrospective assessment. However, for the purpose of personal trend analysis, consistent self-report is more useful than automated app tracking because it captures cognitive engagement rather than application usage.