Paper Review: Attention Residue and the Ready-to-Resume Plan

Paper 1: Leroy, S. (2009). “Why Is It So Hard to Do My Work? The Challenge of Attention Residue When Switching Between Work Tasks.” Organizational Behavior and Human Decision Processes, 109(2), 168-181. DOI: 10.1016/j.obhdp.2009.04.002

Paper 2: Leroy, S. & Glomb, T. M. (2018). “Tasks Interrupted: How Anticipating Time Pressure on Resumption of an Interrupted Task Causes Attention Residue and Low Performance on Interrupting Tasks and How a ‘Ready-to-Resume’ Plan Mitigates the Effects.” Organization Science, 29(3), 380-397. DOI: 10.1287/orsc.2017.1184

Also relevant: Leroy, S. & Schmidt, A. M. (2016). “The Effect of Regulatory Focus on Attention Residue and Performance During Interruptions.” Organizational Behavior and Human Decision Processes, 137, 218-235.

Author: Sophie Leroy, now Associate Dean and Professor at UW Bothell School of Business. Previously at University of Minnesota Carlson School of Management. Has spent 17+ years on this research line.

Status: Real research with real data. The 2009 paper won the Academy of Management Best Paper Award (MOC division). The 2018 paper has 78 Scopus citations. Cal Newport popularized the attention residue concept in Deep Work, making Leroy’s work one of the most widely cited pieces of organizational psychology in the productivity literature.

TL;DR

When you switch tasks, your brain does not cleanly switch with you – cognitive residue from the previous task lingers and degrades performance on the new task. This is worse when the old task was unfinished and you expect time pressure when returning to it. The intervention is dead simple: before switching, spend less than a minute writing down where you were and what you will do when you come back. This “Ready-to-Resume Plan” provides enough cognitive closure that attention residue drops significantly and performance on the new task improves – participants were 79% more likely to make the optimal decision on the interrupting task when they used the plan. This is the theoretical blueprint for session-resumption tooling.

The Core Concept: Attention Residue

Leroy defines attention residue as “the persistence of cognitive activity about a Task A even though one stopped working on Task A and currently performs a Task B.” Your brain keeps processing the old task in the background, like browser tabs consuming memory.

This is not just metaphor. In the 2009 experiments, residue was measured using a lexical decision task (Experiment 1) – flash strings of letters on screen, decide as fast as possible if they form real words. The words were related to Task A. Participants experiencing attention residue were faster to recognize Task A-related words, proving that the old task was literally still active in their cognitive workspace.

The key insight is not that switching is hard. It is that the residue from the old task actively impairs the new task. You are not just losing time – you are performing worse because cognitive resources are split.

What Makes Residue Worse

Three factors increase attention residue (established across the full research arc):

1. Task incompletion. Unfinished tasks create much more residue than finished ones. This connects directly to the Zeigarnik effect – incomplete tasks create persistent cognitive tension.

2. Anticipated time pressure on return. The 2018 paper’s key contribution: it is not just whether the task is done. If you know you will have to rush when you come back (5 minutes vs. 15 minutes to complete), you carry more residue into the new task. Your brain keeps the old task “warm” because it knows resumption will be stressful.

3. Prevention framing. The 2016 paper (Leroy & Schmidt) found that tasks framed around avoiding losses (“don’t mess this up”) create more residue than tasks framed around achieving gains (“here’s an opportunity”). Prevention-focused mindsets are stickier.

What Reduces Residue

Two factors reduce attention residue:

1. Task completion under time pressure. Counterintuitive: finishing a task is not enough. You need to finish it under some time pressure to fully disengage. Low-intensity, open-ended completion leaves residue even when the task is “done.” The deadline creates cognitive closure.

2. The Ready-to-Resume Plan. More on this below.

Study Design: The 2009 Paper (2 Experiments)

Experiment 1

Participants worked on word puzzles (Task A), then switched to reviewing resumes and making hiring decisions (Task B). Two manipulated variables:

• Task completion: Some finished the puzzles, some were interrupted
• Time pressure: Some had time pressure on Task A, some did not

Between tasks, a lexical decision task measured whether Task A words were still cognitively activated. Result: interrupted participants and low-time-pressure completers showed significant attention residue. Only those who finished under time pressure had clean cognitive transitions.

Experiment 2

Same task switching paradigm but removed the lexical decision task (which itself could contaminate results by introducing a third task). Directly measured performance on Task B (the resume/hiring decision).

Result: attention residue from Task A caused measurably worse performance on Task B. Participants who were still mentally chewing on the word puzzles made worse hiring decisions.

The finding that even completing a task is not enough is the non-obvious result. Most productivity advice says “finish what you start.” Leroy shows that finishing without urgency leaves residue too. The brain needs a forcing function to let go.

Study Design: The 2018 Paper (4 Studies)

Study 1: Field Survey

202 working professionals in the Midwest. Self-reported experiences when interrupted at work. Established that the time-pressure-on-resumption phenomenon exists in real workplaces, not just labs.

Study 2: Laboratory

Participants interrupted from Task A to work on Task B. Word association measures confirmed that subjects continued thinking about Task A, impairing Task B performance. The critical manipulation: participants told they would have either 5 minutes or 15 minutes to complete Task A when they returned. The 5-minute group (high anticipated time pressure) showed significantly greater attention residue and worse performance on the interrupting task.

Study 3: Laboratory (n=66)

Introduced the Ready-to-Resume Plan intervention. Resume/hiring decision task. Participants who wrote a brief plan before switching performed significantly better on Task B.

Study 4: Laboratory (n=44)

Replicated Study 3 results. Participants using the Ready-to-Resume Plan:

• Made better hiring decisions
• Recalled more information from resumes
• Were 79% more likely to correctly choose the top candidate
• Showed significantly reduced attention residue

Total participants across all 4 studies: ~312+ (202 + unspecified + 66 + 44)

The Ready-to-Resume Plan: Exactly What It Is

The intervention takes less than one minute. Before switching to the interrupting task, you write down:

1. Where you are in the current task
2. What you planned to do next when you return
3. What challenges remain
4. What actions you must postpone but will resume later

That is it. Not a full project plan. Not a detailed brain dump. A few sentences: where I was, what is next, what is hard.

Why It Works (Mechanism)

The plan works through implementation intentions – a concept from Gollwitzer’s research. When you form an if-then plan (“when I return, I will do X”), your brain treats the plan as a commitment that reduces the need to keep the goal actively monitored. The cognitive tension from the Zeigarnik effect (unfinished tasks demanding attention) is partially resolved not by finishing the task but by making a concrete plan for its completion.

Masicampo & Baumeister (2011, JPSP) confirmed this mechanism independently: making specific plans for unfulfilled goals eliminates the cognitive interference those goals normally cause. The brain accepts a good plan as a proxy for completion.

Leroy’s metaphor: it is like closing browser tabs. You are not deleting the work – you are bookmarking it so your brain stops keeping it in active memory.

What the Plan Does NOT Do

The research specifically did not examine whether the plan helps you perform better when you return to the interrupted task. It only measured performance on the interrupting task. This is an important gap. The plan was studied as a “switching away” tool, not a “switching back” tool.

Limitations and Criticisms

Real Limitations

1. Lab settings with student participants. Studies 2-4 used undergraduates reviewing fake resumes. The ecological validity gap is real. A student interrupted from word puzzles to review resumes is not the same as a programmer interrupted from debugging to attend a meeting. The field study (Study 1) helps, but it is survey data, not behavioral measurement.

2. Small sample sizes. n=66 and n=44 for the intervention studies. These are small by modern standards. The 79% improvement figure comes from 44 participants. Worth noting, not disqualifying.

3. Task complexity is low. Word puzzles and resume reviews are relatively simple tasks. Real knowledge work involves building complex mental models over hours. Whether a 60-second plan can externalize enough context from a 3-hour debugging session is an open question.

4. No long-term follow-up. The studies measured immediate effects. Does the plan still work after days or weeks? When you return to a project after a weekend, is a Tuesday afternoon plan still useful on Monday morning? Unknown.

5. No ADHD population. All participants were presumably neurotypical (or at least unscreened). Attention residue is a universal human phenomenon, but ADHD brains may experience it differently – possibly more intensely (weaker cognitive control = harder to suppress residue) or possibly less intensely (if ADHD involves faster cognitive “eviction” of previous tasks, which some research suggests).

6. Resumption performance not measured. As noted above, the plan was only tested for its effect on the interrupting task. Whether it helps you get back to the original task faster or better is assumed but not demonstrated in this research.

Not Really Limitations

• The effect sizes are solid. 79% improvement in decision accuracy is not noise.
• The field study grounds the lab findings in real workplace experience.
• The theoretical framework (Zeigarnik + implementation intentions + attention residue) is well-established from independent research lines.
• The concept has been replicated/extended in Leroy’s 2016 paper with Schmidt (regulatory focus moderator, 3 studies) and is consistent with Masicampo & Baumeister’s independent findings on plan-making.

Design Implications for Session-Resumption Tools

This research is the theoretical backbone for building session-resumption tooling for knowledge workers. But the research needs to be adapted, not copied directly.

What Leroy Studied vs. What Session-Resumption Tools Need

Leroy studied	Tool builders need
Plan made at moment of interruption	Plan made at session end (or at context-window limits)
Plan for same-day return	Plan for return after hours or days
Simple task context	Complex multi-file programming context
Written by the person doing the work	Could be auto-generated by AI
Used to improve performance on the NEW task	Used to improve performance on RETURN

The gap between “write a 60-second note before switching tasks” and “auto-generate a context summary when a coding session ends” is significant. But the mechanism is the same: externalize the state so the brain can let go.

Design Principles for Session-Resumption Tooling

1. The plan must contain both state and intention.

“Where I was” alone is not enough. Leroy’s plan requires both location (where you stopped) AND direction (what you will do next). A git diff tells you what changed but not what you were about to do. A session-end hook needs to capture:

• What was being worked on (state)
• What the next step would have been (intention)
• What was tricky or unresolved (challenge)

This maps directly to Leroy’s four elements: where you are, what to do next, what challenges remain, what to postpone.

2. Brevity is load-bearing.

The plan works because it takes less than a minute. If a session-end summary is longer than 5 bullet points, it becomes a cognitive burden itself. Three bullet points is the right target. Leroy confirms this is the right instinct.

3. Auto-generation changes the equation.

Leroy’s participants wrote their own plans, which forced them to actively reflect. Auto-generating the plan (from git log, persistent context, conversation history) removes the reflection step but also removes the burden. For ADHD, removing burden probably matters more than preserving reflection. The re-entry moment (reading the summary) can provide the reflection Leroy intended.

4. The plan is for the RETURN, not the departure.

Leroy studied the plan’s effect on switching away. Session-resumption tools care about switching back. The Masicampo & Baumeister research suggests the plan also frees cognitive resources during the gap (your brain stops nagging about the unfinished work). But the primary value is the re-entry: when you start a new session, the first thing you see is “here is where you left off and what comes next.”

5. Time pressure findings map to ADHD perfectionism.

Leroy’s finding that anticipated time pressure increases residue is relevant to ADHD anxiety patterns. If a session ends with “I have so much left to do and not enough time,” the residue will be intense and possibly paralyzing. The session-end summary should include a note of progress (“you finished X and Y”) alongside the “next steps” – not just a todo list but a “you are in a good place to continue” signal.

Concrete Implementation Spec

Based on this research, an ideal session-end summary should capture:

## Session Summary (auto-generated)

**Working on:** [project/branch/files touched]
**Progress:** [what was accomplished this session]
**Next step:** [the most likely next action]
**Open question:** [what was unresolved or tricky]

This maps to Leroy’s four elements:

• “Where you are” = Working on + Progress
• “What to do next” = Next step
• “What challenges remain” = Open question
• “What to postpone” = implicit (anything not in Next step)

A session-start hook then surfaces this as the opener:

Last session (Tuesday, 3:42 PM):
- Working on podcast episode script
- Finished: TTS formatting pass, source citations
- Next: Record featured segment, needs research file
- Open: Unsure if API endpoint pricing changed

Three to four lines. Takes 5 seconds to read. Provides both state and direction. This is the Ready-to-Resume Plan, automated.

What Questions Remain

1. Does auto-generation work as well as self-writing? Leroy’s mechanism partially depends on the act of writing forcing reflection. Auto-generation skips that step. Does reading an auto-summary provide equivalent cognitive closure? Testable.

2. Does the plan help across longer gaps? Leroy tested same-session effects. Does a plan written Tuesday help with re-entry on Thursday? On Monday? After two weeks? The Altmann & Trafton (2002) goal activation model predicts decay, so longer gaps may need richer summaries.

3. Does ADHD change the equation? Attention residue is universal, but ADHD working memory deficits (Kasper et al., 2012: very large magnitude central executive impairments) may mean the plan needs to externalize more context than neurotypical users need. A 3-line summary might not be enough for re-entry when working memory cannot fill in the gaps.

4. Implementation intentions + ADHD: Gawrilow & Gollwitzer (2008) showed that implementation intentions (if-then plans) actually normalize response inhibition in ADHD children to control levels. This suggests the mechanism behind the Ready-to-Resume Plan might be especially effective for ADHD – the if-then structure automates action control that ADHD brains struggle to maintain internally. Promising but untested for task resumption specifically.

5. Does the departure summary reduce inter-session anxiety? The Masicampo & Baumeister finding (plans eliminate cognitive interference from unfulfilled goals) predicts that a good session-end summary should reduce the “I should be working on that” mental noise between sessions. Relevant for ADHD rumination patterns. Testable by self-report.

Related Papers Worth Noting

Foundational:

• Masicampo, E. J. & Baumeister, R. F. (2011). “Consider It Done! Plan Making Can Eliminate the Cognitive Effects of Unfulfilled Goals.” JPSP, 101(4), 667-683. – Independently confirms the plan-as-closure mechanism.
• Gollwitzer, P. M. (1999). “Implementation Intentions: Strong Effects of Simple Plans.” American Psychologist. – The theoretical base for why if-then plans work.

ADHD-specific:

• Gawrilow, C. & Gollwitzer, P. M. (2008). “Implementation Intentions Facilitate Response Inhibition in Children with ADHD.” Cognitive Therapy and Research. – If-then plans normalize ADHD inhibitory control. Exciting implication for session-resumption design.

Programmer-specific:

• Parnin, C. & Rugaber, S. (2011). “Resumption Strategies for Interrupted Programming Tasks.” Software Quality Journal. – Only 10% of programming sessions resume in <1 minute. 56% navigate to other locations first. Direct evidence that programmers need “where was I?” tooling.
• Mark, G., Gudith, D. & Klocke, U. (2008). “The Cost of Interrupted Work: More Speed and Stress.” CHI ‘08. – 23 minutes 15 seconds average to fully resume after interruption.

Leroy’s full arc:

• Leroy & Schmidt (2016). Regulatory focus paper. Prevention-framed tasks create stickier residue. Adds nuance.
• Leroy (2024 ongoing study, UW Bothell). Still actively researching attention and focus – worth monitoring for new publications.

Bottom Line

This is the best-evidenced intervention available for session-resumption tooling. The Ready-to-Resume Plan is not a productivity hack – it is a cognitive mechanism grounded in implementation intention theory, the Zeigarnik effect, and 17 years of Leroy’s empirical work. The fact that it works in under 60 seconds and requires only “where I was + what is next” makes it implementable as a session hook today.

The ADHD angle is unexplored in Leroy’s work but theoretically promising: if implementation intentions can normalize inhibitory control in ADHD children (Gawrilow & Gollwitzer), they may also reduce attention residue. Auto-generation instead of self-writing trades the reflection benefit for the burden-reduction benefit. For ADHD, that is probably the right trade.

The question worth testing: does automated re-entry context measurably decrease resumption time?