You opened ChatGPT. Then Claude. Then Perplexity. Then your notes app. Then back to ChatGPT. You did this in four minutes. Your brain did not.
Your brain is still on Tab 1 while your fingers are on Tab 4. That leftover cognitive residue is not a metaphor — it is a measurable neurological phenomenon with a name, a paper trail, and a cost that compounds every single time you jump.
This issue is about that phenomenon: Attention Residue. What it is, why AI tools amplify it in ways we haven't seen before, and what the science says you can actually do about it.
Think of it like a whiteboard. Sometimes we erase something on it but some of the ink remains and stays as a trace in the background.
— Gloria Mark · Professor of Informatics, UC Irvine · Attention Span (2023)The 47-second number is not a typo. In 2004, when Gloria Mark first began tracking attention duration in real office environments, the average was 2.5 minutes. By 2012 it had dropped to 75 seconds. Her most recent measurements put it at 47 seconds — with one replication finding the median at just 40 seconds. That is not a trend. That is a collapse.
And this was before the average knowledge worker had five AI assistants open in parallel tabs.
Leroy, 2009 The term "attention residue" was coined by Sophie Leroy, University of Washington professor, in a 2009 paper published in Organizational Behavior and Human Decision Processes.
Leroy's definition is precise: the persistence of cognitive activity about a Task A even though one has stopped working on Task A and is now performing Task B. It is not the same as distraction — distraction is external. Attention residue is internal. Your own mind creates the interference.
Leroy's experiments revealed a cruel irony: the more unfinished or time-pressured the interrupted task felt, the stronger the residue on the next task. You don't just lose time. You lose quality of cognition on everything that follows.
Normal tab-switching interrupts your attention. AI tab-switching does something worse — it rewards the interruption. Every time you jump from Claude to ChatGPT to Perplexity, you receive a compelling output. Variable reward schedules. Immediate feedback. The neurochemistry of your brain treats each AI response as a small win, reinforcing the very switching behaviour that is depleting your cognition. It is the slot machine problem, dressed in a productivity costume.
The Zeigarnik Effect adds a second layer. Named after Soviet psychologist Bluma Zeigarnik, who in 1927 observed that waiters remembered incomplete orders far better than completed ones — your brain keeps unfinished tasks "active" in working memory. Every half-finished prompt in a different AI tab is an open loop consuming mental RAM you never budgeted for.
Read this list for 10 seconds, then scroll past. Don't re-read it.
Come back in 5 minutes — you'll remember the incomplete items more vividly than the completed ones. That's Zeigarnik. Your AI tabs exploit this exact mechanism.
Leroy's follow-up research proposed the simplest high-ROI intervention in the literature. When you must switch tasks, write one sentence: where you were, and what you planned to do next. This offloads the open loop from working memory to external storage. The brain stops background-processing the abandoned task. Recovery time drops from 23 minutes to near-zero. It takes 12 seconds.
Five tactics backed by research, not intuition. Each one targets a specific mechanism in the attention residue chain.
Choose one AI tool per cognitive session. The efficiency gain from using "the right AI for each subtask" is almost always smaller than the residue cost of switching. Research: multitasking students retained 28% less information despite feeling more confident. Pick your tool before the session, not during it.
Leroy's "Ready-to-Resume Plan" in practice: before every tab switch, type one line — "I was doing [X] and about to [Y]." Save it anywhere visible. This offloads the open loop to external storage. Recovery time drops from 23 minutes to near-zero.
Block your day into 25-minute "attention epochs" assigned to exactly ONE cognitive context — including one AI tool. Research found that 25 minutes is the minimum threshold for the incubation stage of analytical work. Under that, you're in permanent warm-up mode.
At the end of every AI session, rate yourself: was I fully present, or was I carrying residue? Gloria Mark found a bidirectional loop — fragmented attention causes stress, and stress further fragments attention. The audit breaks the loop by making the cost visible to you.
Never abandon an AI conversation mid-thread without a deliberate closure act. Paste the key output somewhere permanent and type "Done for now." The Zeigarnik Effect activates on incomplete interactions — explicit closure signals task completion to your brain's goal-tracking system.
Lower = less cognitive residue risk per session. Based on context-reload depth requirements:
This month's influence move uses the attention residue principle offensively — in communication and persuasion. If you understand when someone's attention residue is high, you can time your message for maximum cognitive impact.
Research shows that people are most receptive to reframing precisely when they have just switched contexts — their existing mental model is momentarily destabilised. Here's how to engineer this in pitches, emails, and negotiations...
Start your pitch or email with a specific, sharp interruption to their current mental context. Not with your proposition — with theirs. Name what they were probably just thinking about before reading this. This creates a controlled residue event...
Two users. Same AI tools. Completely different cognitive outcomes this week.
Multitasking feels productive. It creates a subjective sense of momentum. A 2024 replication of Mark's research found that users in high-switch states rated their own productivity significantly higher than focused users — despite measurably lower output quality. You will feel sharper while getting worse. That gap is the enemy.
- 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
- Mark, G., Gudith, D., & Klocke, U. (2008). The cost of interrupted work: More speed and stress. Proceedings of CHI '08. ACM. DOI: 10.1145/1357054.1357072
- Mark, G. (2023). Attention Span: A Groundbreaking Way to Restore Balance, Happiness and Productivity. Hanover Square Press.
- 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. Organization Science, 29(3), 380–397.
- Zeigarnik, B. (1927). Über das Behalten von erledigten und unerledigten Handlungen. Psychologische Forschung, 9, 1–85.
- University of Michigan (2024). Multitasking during lectures reduces information retention by 28%. Replication study, Department of Cognitive Science.