The Science of Single-Tasking: What Research Actually Shows

The case for single-tasking is sometimes presented as obvious wisdom. It is worth treating it as an empirical question instead: what does the research actually show, what are the limitations of that research, and what can we confidently apply from it?

Attention Residue: Sophie Leroy’s Core Finding

The most directly applicable research for knowledge workers is Sophie Leroy’s 2009 work on attention residue, published in Organizational Behavior and Human Decision Processes.

Leroy’s experimental design was straightforward. Participants were given a first task (Task A) to work on, then asked to switch to a second task (Task B). Some participants were interrupted mid-Task A, while others were given enough time to reach a natural stopping point. Both groups then performed Task B.

The key finding: participants who were interrupted before completing Task A showed measurably impaired performance on Task B. Leroy attributed this to attention residue—the continued cognitive processing of Task A that competed with Task B for attentional resources.

A secondary finding was particularly relevant: cognitive completeness mattered more than time. Participants who had reached a natural stopping point on Task A—even if the task itself was not fully complete—showed less residue than those who were simply cut off mid-stream. This suggests that closure, not just time, is what allows attention to fully release.

Limitation to note: Leroy’s research was conducted in controlled laboratory settings with defined tasks. Workplace tasks are typically more open-ended and have less clear stopping points. The attention residue effect likely applies in real work contexts, but the magnitude may differ from lab estimates.

The Nass, Ophir, and Wagner Study: What Heavy Multitaskers Actually Do Worse

Eyal Ophir, Clifford Nass, and Anthony Wagner published their Stanford multitasking study in 2009 in the Proceedings of the National Academy of Sciences. It is one of the most cited studies in attention research.

The participants were divided into heavy and light media multitaskers based on a questionnaire about their habitual multi-stream media use. Both groups then completed three cognitive tasks:

Filtering task: Distinguish relevant from irrelevant items in a visual display. Heavy multitaskers were more distracted by irrelevant stimuli.

Task-switching task: Alternate between two categorization rules. Heavy multitaskers were slower to switch—the opposite of what multitasking practice was expected to produce.

Working memory task: Hold a sequence of letters in memory while performing a secondary task. Heavy multitaskers performed worse.

The study did not find a single cognitive advantage for heavy multitaskers. The researchers noted that this was unexpected, and that they could not definitively establish causation. Heavy multitasking might impair cognitive filtering, or people with lower cognitive filtering ability might be more prone to heavy multitasking.

Limitation to note: This was a correlational study, not an experiment. The causation question—does multitasking damage attention, or do people with attention management challenges tend to multitask more—remains partially open. Subsequent research supports a bidirectional relationship, but the Nass study alone cannot establish causal direction.

David Meyer’s Task-Switching Cost Research

David Meyer and his colleagues at the University of Michigan have produced some of the most methodologically rigorous work on task-switching, including the concept of “executive control processes” that must reconfigure when you shift between tasks.

Their work, including a comprehensive review published in Psychological Science (2001), distinguishes between two components of switching cost:

Goal shifting: The cognitive process of disengaging from one task’s goals and engaging with another’s.

Rule activation: Activating the mental rules and response mappings for the new task while inhibiting those of the previous task.

Both processes take time. For simple, well-practiced tasks, the total switching cost may be under a second and largely automatic. For complex tasks—analytical reasoning, writing, strategic planning—the reconfiguration can take significantly longer and involves genuine cognitive work, not just a brief pause.

Meyer’s research group’s estimate that switching costs can consume 20 to 40 percent of productive time for complex tasks is based on experimental studies measuring performance on cognitively demanding tasks under switching versus non-switching conditions. The estimate is a range precisely because the cost varies with task complexity, switching frequency, and individual differences.

Limitation to note: Laboratory tasks, while cognitively demanding, differ from real knowledge work in complexity and duration. The 20–40 percent estimate should be treated as an order-of-magnitude indication rather than a precise figure. It is sufficient to take seriously; it is not a precise measurement of real-world work loss.

Working Memory Capacity and the Queue Problem

The theoretical foundation for why open loops impair focus comes partly from working memory research.

George Miller’s foundational 1956 paper estimated working memory capacity at seven plus or minus two chunks. More recent work, particularly by Nelson Cowan, suggests that for complex, multi-component information the effective capacity is closer to four items.

When you have five incomplete tasks, an unread message you are aware of, and a decision you need to make by end of day, that is at minimum seven items competing for the four-item capacity. Working memory overflows. Attention becomes divided not by choice but by the limitations of the cognitive architecture.

Zeigarnik effect research—first described by Bluma Zeigarnik in the 1920s—adds a further mechanism: incomplete tasks create a persistent cognitive signal that draws attention back toward them. This effect is well-replicated and explains why partially completed work intrudes on unrelated thinking. It also has a known resolution: if you offload the item to a trusted external system, the signal diminishes. You do not need to complete the task; you need to trust that it will not be forgotten.

Limitation to note: The Zeigarnik effect has been replicated but shows variability across conditions. The “offloading to trusted system” resolution is supported by more recent research (including E. J. Masicampo and Roy Baumeister’s 2011 work) but the effect size and conditions for reliable offloading are still being characterized.

What Flow State Research Adds

Mihaly Csikszentmihalyi’s research on flow—optimal experience characterized by complete absorption in a challenging, skill-matched activity—is frequently cited in the context of deep work and single-tasking.

Flow states, in Csikszentmihalyi’s research, are associated with reduced activity in the prefrontal cortex (the brain region associated with self-monitoring and working memory), as well as subjective reports of effortless attention, time distortion, and high intrinsic reward.

For present purposes, two findings are directly relevant:

1. Flow states require extended uninterrupted engagement. They do not develop in five-minute windows. The initial investment before reaching flow may require fifteen to thirty minutes of focused work.

2. Interruptions destroy flow states. Re-entry after interruption does not resume from where flow left off—it requires restarting the accumulation process.

This provides a mechanistic account of why task-switching is especially costly for knowledge workers: beyond the switching cost itself, each switch potentially resets a flow accumulation that had been underway.

Limitation to note: Flow state measurement relies heavily on self-report. The neurological correlates are not fully characterized. Csikszentmihalyi’s research framework is well-replicated in terms of the subjective experience, but claims about the neural mechanisms should be treated with appropriate caution.

What the Research Supports and What It Does Not

Robustly supported:

• Attention residue is a real phenomenon that impairs performance on subsequent tasks.
• Task-switching carries measurable cognitive costs that scale with task complexity.
• Heavy multitaskers show worse performance on cognitive filtering and working memory tasks than light multitaskers.
• Working memory capacity is limited; open loops consume that capacity.
• Offloading incomplete tasks to trusted external systems reduces their attentional pull.

Supported but with caveats:

• The 20–40 percent switching cost estimate is directionally reliable but not precisely generalizable to all work contexts.
• Flow states require extended uninterrupted engagement, but the specific time requirements vary by individual and task.

Not established:

• That multitasking ability can improve with practice (the evidence leans against this).
• That the brain can genuinely perform two cognitively demanding tasks simultaneously without degradation to both.

What the Research Does Not Tell You to Do

Research on attention and task-switching describes mechanisms, not prescriptions. No study tells you exactly how to structure your workday.

What the research does tell you is which variables matter: open-loop count, switching frequency, working memory load, and time available for uninterrupted engagement. A workday designed with those variables in mind—fewer open loops, less frequent switching, lower working memory load, more uninterrupted time—will produce better cognitive output. The specific method for achieving that is an implementation question, not an empirical one.

The One Thing Lock is one implementation. There are others. The research gives you the principles; you design the practice.

Read Sophie Leroy’s 2009 paper on attention residue—it is publicly available and accessible without a paywall on her faculty page—and notice which parts of your current workday most closely resemble the interrupted condition she studied.

Related:

• The Complete Guide to Single-Tasking with AI Support
• Why Multitasking Feels Productive
• 5 Single-Tasking Approaches Compared
• Managing Attention in the AI Age

Tags: attention research, cognitive science, task-switching, single-tasking, attention residue