Task-switching at work

Switching cost is universal

Subjects asked to alternate between two simple tasks — classify a digit as odd/even, then classify a letter as vowel/consonant — show measurable cost on switch trials: slower response times, more errors, in the hundreds of milliseconds per switch (Monsell 2003¹). The cost persists even with preparation time (the “residual switch cost”). Rubinstein, Meyer & Evans 2001² established the four-process model and the additivity of switch costs across component operations.

The knowledge-work translation comes from Mark, Gonzalez & Harris 2005³. Observational study of information workers: average task duration before interruption was 11 minutes; re-orientation after interruption took 23+ minutes. Mark, Gudith & Klocke 2008⁴ extended the findings. The 23-minute figure is the cost to regain the cognitive state the work requires, not the time to remember where you were. This is universal, not an ADHD finding.

The ADHD gradient — set-shifting deficits

Boonstra, Oosterlaan, Sergeant & Buitelaar 2005⁷ meta-analysed executive function in adult ADHD. Set-shifting — the operation underlying task switching — showed reliable medium deficits across paradigms. Hervey et al. 2004⁸ reached convergent conclusions earlier.

The gradient operates two ways. Each switch costs more in adult ADHD per the meta-analytic finding. And more switches happen — adults with ADHD initiate more switches in environments that permit it, downstream of the impulsivity / distractibility profile (Snyder, Miyake & Hankin 2015⁹). The visible workday cost is cost-per-switch × switches-per-day; both factors are elevated. The product is the load the reader experiences as “I can never get into anything.”

Cognitive Load Theory — the frame

Sweller 1988¹⁰ established Cognitive Load Theory in instructional design. Working-memory capacity is limited; load that exceeds it impairs performance. The framework separates load that's inherent to the task, load imposed by the environment, and load devoted to productive learning.

The application. The modern knowledge-work environment imposes substantial extraneous load — notifications, ambient conversation, meeting-imposed context switches, async messaging that demands sync-like response times. This load is independent of the work itself. In adult ADHD, the smaller central-executive capacity documented at working memory failures means a given extraneous-load level consumes a larger share of available capacity.

Open-plan offices — empirically failed

Bernstein & Turban 2018⁵ ran two field experiments at large firms transitioning from cubicle/private offices to open-plan layouts. Face-to-face interaction dropped roughly 70% after the move; electronic communication rose correspondingly; productivity got worse. The hypothesis that open layouts increase collaboration was falsified by direct measurement. Kim & de Dear 2013⁶ found open-plan offices scored worse on noise privacy and overall satisfaction across worker types.

The reader who finds the open-plan office substantially harder than their peers is not being precious. The cost is real, universally documented, and amplified by the ADHD gradient — ambient conversation and unpredictable interruption consume the already-smaller central-executive capacity faster.

Meeting density — the back-to-back problem

Each meeting is a switch into a new context with its own intrinsic load. Back-to-back meetings impose switches without recovery windows. The Mark 2005 re-orientation cost is built into every transition; for adult ADHD readers the per-switch cost is amplified per Boonstra 2005, and cumulative load across a meeting-dense day exceeds capacity earlier than for non-ADHD peers.

The tactical implication: calendar batching of similar work, focus blocks that protect uninterrupted time, and meeting-free days are mechanism-aligned interventions that reduce switches and per-switch cost. The Safren et al. 2010¹⁴ CBT-for-adult-ADHD protocol’s calendar / task-list externalisation is the closest RCT-evidenced delivery vehicle for this in clinical context.

The multitasking myth, named

What observers call multitasking is rapid task switching, and the switching imposes substantial cost (Rubinstein 2001; Monsell 2003). The reader who has been told “you just can’t multitask” was given a half-truth. The universal finding is that no onecan. The accurate self-description is “switching costs me more per switch and I’m doing more switches.” The first framing produces shame; the second produces tactics.

The supertasker exception. Watson & Strayer 2010¹¹ identified roughly 2.5% of the general population who showed no measurable cost from dual-tasking under specific paradigms. Supertaskers are rare; the reader asking the question is overwhelmingly not one.

What helps — graded honestly

None of these has a task-switching-specific RCT in adult ADHD; all are mechanism-aligned with the cognitive-load and working-memory literatures.

Calendar batching and focus blocks.Concentrate same-type work into single blocks; protect uninterrupted time. Reduces switches per day, attacking both halves of the gradient. Mark 2005’s 23-minute re-orientation cost compounds across switches; eliminating switches reclaims the cost.

Written-not-verbal communication where possible. Verbal communication imposes real-time central-executive load — phonological loop holding speech, meaning processing, response preparation, turn-taking monitoring all at once. Written communication permits asynchronous processing — read at your own pace, re-read, respond after consideration. Operational forms: agendas distributed in advance, meeting recaps in writing, complex decisions made in written documents, one-on-one updates via written status rather than standups.

Async-over-sync where possible.Email, written documents, ticket systems externalise the timing of engagement. Sync demands engagement at the sender’s preferred time. The boundary between async and sync in modern workplaces has eroded; Slack with notifications and read-receipt expectations functions as sync. The tactic is not “more async” but “treat async as actually async, with response windows of hours not seconds.”

Single-tab discipline. Each open tab is a held intention with associated central-executive cost (Smith 2003¹² cost-of-prospective-memory paradigm operating on tabs-as-cues). Closing tabs reduces the held-intention load. The open tab is a failed offload — the reader is using the browser as memory but paying the held-load cost rather than the offload-benefit cost (Risko & Gilbert 2016¹³).

Notifications off as the default. Every notification has the potential to trigger an interruption with 23-minute re-orientation cost. Turning notifications off converts the default from interrupt-driven to checked-on-schedule.

Workplace accommodations under the Americans with Disabilities Act (ADA). Written meeting recaps, advance agendas, written-rather-than-verbal instructions, private workspaces, noise-cancelling headphones — all documented in Job Accommodation Network (JAN) guidance. Full treatment at ADHD at work.

Stimulant medication. Partial effect on the underlying capacity. Working memory and sustained attention improve by small-to-moderate effect sizes; set-shifting is among the cognitive domains where stimulant effects are documented. Effect is partial; tactics still required.

What doesn’t

• “Just focus.” Asks the over-loaded central executive to do its job. The intervention has to operate on load, not on focus.
• Pomodoro without protected blocks. A Pomodoro inside an interrupt-rich environment with notifications on does not solve the underlying load problem. The intervention is the protected block; Pomodoro is one heuristic for structuring it.
• “Multitask better.” Reproduces the myth. Switching costs are universal; better multitasking is faster switching, which is still switching with cost.
• Wholesale productivity systems imported from elsewhere. Getting Things Done (GTD) style systems help to the extent they externalise; they become harmful when maintenance overhead exceeds offloading benefit. The reader updating their task system instead of doing the task is in failure mode.
• Open-plan office “focus zones” or quiet pods. Partial mitigations of an empirically falsified design. The cost is structural to the layout; adding mitigation rooms does not solve it.