Interruptions in brain work are expensive: three ways to calm the cognitive load of the workday

What an interruption does to working memory

Working memory keeps information active without external input, and this maintenance relies on sustained prefrontal neuronal activity (Wang et al. 2011). The maintenance is fragile: when attention moves to another task, the representation of the previous task weakens. It does not wait its turn, it must be reassembled.

The speed of reassembly depends on processing speed, a central marker of cognitive performance (Salthouse 2010) which together with executive functioning predicts cognitive change (PMC 2025). When interruptions are frequent, processing speed becomes the limiting factor for the whole workday: there is enough time, but not enough capacity.

Intervention data on brain work from the Finnish Institute of Occupational Health, in particular the SujuKE project, show that interruptions and fragmented information are central load factors in brain work and that they can be influenced at workplace level. That finding matters, because it moves the discussion from individual traits to the structure of work.

Load does not stop when the workday ends

A repeated load peak is not a neutral event. The relationship between cortisol and brain structure is dose dependent: serum cortisol is inversely related to hippocampal volume (Frontiers in Aging Neuroscience 2023), and the association of social stress, the cortisol awakening response and grey matter volume has also been examined in more recent work (PMC 2025).

Over longer horizons high cortisol has been linked to the risk of dementia and Alzheimer’s disease (Ouanes & Popp 2019) and to hippocampal atrophy and disease progression (White et al. 2023). Chronic stress has also been examined as a trigger of Alzheimer’s pathology through neuroinflammation and tau pathology (Ávila-Villanueva et al. 2020).

In working life this shows directly. High job strain is associated with cognitive decline in prospective data (Singh-Manoux et al. 2016), and burnout is associated with measurable decline in cognitive performance, particularly in executive function (Gavelin et al. 2021). The concept of allostatic load describes the mechanism: continuous adaptation consumes regulatory systems (McEwen & Akil 2020).

Method 1: treat protected time as a production resource

Uninterrupted time is not a privilege but a production resource. A practical rule is that every expert gets at least two uninterrupted 90 minute blocks per day, and those blocks are as binding in the calendar as a client meeting.

Crucially, the rule must be team level rather than individual. If only one person protects their time, they transfer interruptions to others and total load does not fall. When a team agrees on a shared deep work window, the total number of interruptions drops, because askers naturally wait for the next window.

Channel discipline in communication belongs here too. When urgency is defined by the channel rather than by words, working memory does not have to evaluate the priority of every message separately. Priority evaluation is itself cognitive work, and automating it structurally frees capacity for the actual task.

Method 2: break up sitting and maintain cerebral blood flow

While fighting interruptions it is easy to forget that the brain is an organ that needs flow. Carter and colleagues (2018) showed that breaking prolonged sitting with short walks prevents the decline in cerebral blood flow that develops during sitting. The effect is therefore visible in physiology, not only in perceived alertness.

Sitting is an independent risk factor over longer horizons. High sedentary behaviour has been associated with the risk of cognitive decline (Wheeler et al. 2017), and increasing sedentary behaviour has been examined in relation to neurodegeneration over seven years of follow-up (Gogniat et al. 2025).

The practical design is simple: a three minute walk every half hour is a cheaper intervention than anything else in this article. It requires no system, no budget and no permission. At organisational level, designing meetings so that movement is possible is the structural version of the same solution.

Method 3: move the working memory load into external structure

The third method rests on the principle that working memory should not be used for storage. Everything held in mind so that it is not forgotten consumes exactly the capacity needed for solving.

In practice this means unfinished items are written out of mind into a reliable place, and that the entry contains the next concrete step rather than only the topic. The difference matters: a topic requires thinking again, a step does not.

The same principle applies to meetings. When a decision is written visibly during the meeting and the next step is assigned to a person, no collective memory effort is needed afterwards. From the cognitive ergonomics perspective this is work redesign: the structure is changed to match the real limits of human information processing.

How this appears in organisational metrics

The effect of interruption load does not appear as its own line in the financial statements, but it appears in errors, rework, delayed decisions and ultimately absence. In Finnish working life data the development of mental health based sickness absence has been a central monitored phenomenon, and load management relates to it directly.

A second mechanism is cognitive reserve. Reserve is a modifiable property that protects performance under load (Stern et al. 2020) and whose significance across the life course has been examined more broadly (Livingston et al. 2024). The structure of work therefore shapes both this quarter’s result and long-term brain health.

The practical conclusion: managing interruptions is not a comfort question. It is a shared variable of productivity and health, and it is within the organisation’s control.

The four most common mistakes in managing interruptions

Managing interruptions rarely fails because the methods are wrong. It fails because the methods are aimed at the individual while the exposure is created by the structure.

• The solution is left to the individual. When concentration is defined as a character trait, the structure is never touched and the exposure never changes.
• Protected time is agreed but not entered in the calendar. An invisible agreement always loses to a visible meeting invitation.
• All channels are treated as equally urgent. When urgency is not defined structurally, every message is evaluated separately, and evaluation is itself cognitive work.
• Output is measured but load is not. Output holds for a while even as capacity falls, so an output metric alone notices the problem too late.

A six week rollout that survives contact with reality

Structural change succeeds when it is small enough to stay in use and visible enough to be measured. Six weeks is a short enough period to commit to and long enough to see an effect.

Weeks one and two: measure the baseline and log the sources of interruption for two days without changing anything. Weeks three and four: introduce one structural change, for example a shared uninterrupted window for the team. Weeks five and six: hold the change steady and run the follow-up measurement.

What is not done matters equally. Three simultaneous changes are not attempted, because change itself creates load and overlapping changes make the effect impossible to interpret.

Key findings

Interruptions are a property of structure, and structure is a decision. A summary of where to start.

• The price of an interruption is its duration plus rebuilding the train of thought in working memory (Wang et al. 2011).
• A repeated load peak is not neutral: the relationship between cortisol and brain structure is dose dependent (Ouanes & Popp 2019).
• Uninterrupted time is a production resource and must be protected as a team, not alone.
• Breaking up sitting maintains cerebral blood flow and supports performance the same day (Carter et al. 2018).
• Working memory should not be used for storage. Write out the next step, not just the topic.

Measurement makes load manageable

The problem with load is its invisibility. The experience of load does not scale linearly with what is actually happening to cognitive performance, and self-assessment is therefore an imprecise basis for decisions, even though it works for screening (Crook & Feher 2018).

The NeuroAudit™ Method produces a Neuro Performance Index (NPI™) value and decomposes load into a twelve pillar structure. At organisational level this means the effect of interruption load can be measured before and after a structural change. The conversation then moves from opinions to numbers, and structural change can be justified in the same language as any other investment.