Work-related stress, depression, and anxiety now account for 17.1 million lost working days in the UK every year, almost half of all work-related ill health cases. The average UK business loses £1,300 per employee annually to stress-related absence alone, rising to £2,500 in professional services and financial sectors. Burnout among UK workers has risen from 51% to 63% in two years. In the United States, the 2025 Aflac WorkForces Report recorded burnout at a seven-year high. McKinsey’s 2025 Women in the Workplace report found that 60% of senior women experience frequent burnout, rising to 70% among those recently promoted into leadership.
Most of the organisations generating these figures have not been sitting still. They have added wellbeing programmes, mental health days, employee assistance programmes, and access to coaching. The data is rising anyway.
Which means that what is being offered and what the nervous system actually needs are two different things, and the gap between them is widening every year. The cost of that gap is already sitting in absence figures, attrition rates, and the declining quality of thinking organisations depend on. It simply has not been attributed to the right cause.
This article is an attempt to explain what is actually happening biologically, why the standard responses are not working, and where the most effective levers for change actually sit. It will not give you an implementation manual. It will give you a different way of seeing the problem, and that tends to be where meaningful change begins.
What the nervous system does at work
The human nervous system cannot distinguish between physical danger and social threat. When a person feels at risk of humiliation, exclusion, or professional punishment, the brain’s alarm system, the amygdala, activates in exactly the same way it would if the threat were physical.
What follows is a cascade. The HPA axis, the body’s primary stress-response system connecting the brain to the adrenal glands, fires. Cortisol and adrenaline are released. Resources are redirected away from the prefrontal cortex, the part of the brain responsible for complex thinking, creativity, and sound decision-making, and towards vigilance and self-protection.
The nervous system is doing precisely what it was designed to do. The problem is that it was designed for threats that pass, not for environments where the threat is chronic and the source is the workplace itself. In a single moment of perceived threat, the quality of a person’s thinking degrades measurably. Extended across a meeting, a day, a year, the cumulative effect on performance is significant, and largely invisible to everyone, including the person experiencing it.
Why organisations don’t see it coming
Burnout does not announce itself, and the biology explains why. One of cortisol’s less-discussed effects is that it impairs self-awareness and metacognition, the capacity to observe one’s own cognitive and emotional state. People experiencing chronic stress progressively lose the ability to accurately assess how depleted they are. They normalise their condition. They adjust their baseline. They keep going, functioning at diminishing capacity, telling themselves and those around them that they are fine.
From the outside, the signals are equally easy to misread. The behaviours that accompany advanced burnout, reduced expressiveness, social withdrawal, narrowed thinking, decreased initiative, are the same behaviours that register in a performance review as disengagement, lack of drive, or attitude problems. The manager sees someone coasting. The data shows someone running on empty. By the time absence or resignation makes the problem undeniable, months or years of cumulative biological cost have already been paid.
Worth naming too is how the resilience conversation tends to play out in organisations. Resilience is treated as a fixed character trait that some people possess and others lack. The research does not support this. Resilience is a capacity, one that is built or depleted by the environment a person inhabits. Asking people to be more resilient in an environment that is actively depleting them misdiagnoses the problem and places the burden of a structural failure on the individual least equipped to solve it.
What chronic stress does to the brain
Bruce McEwen, the neuroendocrinologist who spent decades studying the long-term effects of stress on the brain, developed the concept of allostatic load: the cumulative biological cost of sustained stress on the body and brain. His research showed that when cortisol remains elevated over time, it physically alters brain structures. The hippocampus, which governs memory and learning, shrinks. The prefrontal cortex becomes less efficient. The amygdala grows more reactive, making the person progressively more sensitive to threat, and less able to regulate their response to it.
The result is a brain that is less capable of doing the work it is being asked to do. Memory weakens. Judgement deteriorates. Emotional regulation, the thing high-performing environments implicitly demand, becomes genuinely harder. Sustained exposure to stress changes tissue, and the cognitive decline that follows is a biological process with nothing to do with how committed or capable the person is.
Performance can fall by up to 35% under chronic stress — the cognitive equivalent of functioning on no sleep.
Research on stress-related cognitive impairment suggests performance can fall by up to 35% under these conditions, the cognitive equivalent of functioning on no sleep. In knowledge industries, where the entire output is the quality of people’s thinking, that is a material loss that belongs in the same conversation as revenue, retention, and risk.
The structural explanation
Wilmar Schaufeli and Arnold Bakker’s Job Demands-Resources model, developed in the early 2000s, offers a useful framework for understanding how environments produce these outcomes. The model proposes that every job can be understood through two categories.
Demands are the cognitive, emotional, and physical costs of the work: workload, time pressure, role ambiguity, emotional labour. Resources are the features of the environment that help people cope with those demands and sustain performance over time: autonomy, clarity, social support, feedback, and the sense that the relational environment is safe enough to function in.
The model’s central finding is that resources buffer. When they are present, people can absorb high demands without depleting. When they are absent, even moderate demands become unsustainable because there is nothing to draw on.
In practice, the same workload can be experienced very differently depending on the quality of the environment surrounding it. Two people with identical job descriptions, working the same hours, can have entirely different neurological experiences of that work, determined largely by whether their environment provides the resources the nervous system needs to function well. Wellbeing programmes are resources, but if the relational environment is unsafe, if demands are rising faster than resources, and if the fundamental conditions for belonging are absent, adding an Employee Assistance Programme (EAP) or a meditation app at the margins does not close the gap. It addresses the symptom while the cause continues undisturbed.
What those resources look like, and why belonging is central
To understand why the quality of the relational environment matters so much, it helps to look at research that predates the modern workplace entirely.
Marie Jahoda was an Austrian-British social psychologist whose work on unemployment in 1930s Austria identified what she called the latent functions of work: the things work provides beyond a salary. These include time structure, social connection, collective purpose, a sense of identity and status, and regular meaningful activity. Jahoda’s argument was that when people lose access to these functions, the psychological deterioration that follows is largely non-financial. It comes from losing the conditions the self needs to remain coherent.
The relevance to the modern organisation is direct. An employee does not have to be unemployed to experience latent deprivation. They simply have to work in an environment that withholds these functions while nominally providing a job. The person who feels excluded from their team, whose contributions go unrecognised, who has no sense of collective purpose in their role, that person is experiencing a form of deprivation that the nervous system registers as sustained threat.
One of the most significant resources an organisation can provide to close that gap is what Amy Edmondson, a professor at Harvard Business School, calls psychological safety: the shared belief within a team that it is safe to take interpersonal risks, to speak up, ask questions, admit uncertainty, or challenge a decision without fear of punishment or humiliation.
Edmondson’s research found that high-performing teams reported more errors than lower-performing ones, not because they made more mistakes, but because they felt safe enough to surface and discuss them. The environment had removed the cost of candour.
When psychological safety is present, people spend their cognitive energy on the work. When it is absent, that energy goes elsewhere: into monitoring tone, rehearsing phrasing, calculating whether speaking up is worth the risk. None of this appears on a timesheet. All of it is measurable in the brain.
What high standards without resources actually produces
The instinct in high-performance organisations is to drive standards up, and that instinct is right. But standards without resources produce a particular kind of performance: careful, defended, oriented towards avoiding punishment rather than contributing something of value. Mistakes get concealed rather than examined. People hold back the thinking that might invite scrutiny.
Edmondson mapped team climates across two axes, psychological safety and performance standards, and identified four zones. The apathy zone, where both are low, produces disengagement. The comfort zone, where safety is high but standards are low, produces complacency. The anxiety zone, where standards are high but safety is absent, is where most high-pressure organisations land, and where performance looks acceptable from the outside while quietly eroding from within. The learning zone, where both safety and standards are high, is where teams actually improve, where someone can say “I think we’re heading in the wrong direction” without first calculating whether the relationship can survive it. Reaching that zone requires resourcing the environment to the same degree that standards are enforced.
AI is accelerating the problem
A 2025 eight-month study by researchers at the University of California Berkeley, conducted inside a 200-person technology company, found that employees using AI tools did not work less. They worked more, at a faster pace, across a broader scope of tasks, and for longer hours, often without being asked to. The researchers identified a self-reinforcing dynamic: AI made doing more feel possible, so people did more, the new pace became the baseline, and the baseline became the expectation.
A separate 2026 study highlighted by Harvard Business Review identified what researchers termed “AI brain fry”, mental fog, slower decision-making, and exhaustion linked to the cognitive load of managing multiple AI tools, validating their outputs, and remaining accountable for quality in a system moving faster than human review can comfortably sustain.
In the language of the Job Demands-Resources model, AI is dramatically increasing demands without automatically increasing resources. It is compressing decision cycles, eroding the recovery time the nervous system needs between tasks, and creating ambient pressure: the sense that because more is now technically possible, more is now required. For people already working in environments with insufficient psychological safety, where belonging is uncertain and the relational climate is depleting rather than sustaining, AI lands in a system already running close to its limits.
There is also a subtler effect. AI creates uncertainty about role, value, and identity, precisely the latent functions Jahoda identified as essential to psychological coherence. When people are unsure whether their contribution still matters, whether their expertise remains relevant, whether their position is secure, the nervous system registers that uncertainty as threat.
What needs to change, and what it actually costs to start
The science here is well understood. The limiting factor has been organisational willingness to treat the environment as a performance variable rather than a backdrop, compounded by the assumption that changing it requires significant investment.
The most powerful interventions available are behavioural and educational, not budgetary.
Start with leadership education, not leadership programmes.
Most leadership interventions do not produce lasting change because they ask leaders to behave differently without giving them a biological understanding of why it matters. A leader who understands that their visible uncertainty gives permission for others to admit theirs, and that this directly affects the cortisol levels, cognitive function, and burnout risk of their team, is more likely to change their behaviour than one who has been told to “be more vulnerable” in a workshop. Education in the neuroscience of stress, belonging, and psychological safety costs a fraction of a leadership development programme and produces more durable results, because it changes the mental model, not just the surface behaviour. In practice, this means investing in short, science-grounded learning experiences for senior leaders and managers before asking them to lead differently. Not a two-day offsite. A conversation about biology that changes how they read what they are seeing in their teams.
Make belonging a measurable leadership responsibility.
Psychological safety travels downward through what leaders visibly do: whether they admit uncertainty, whether they invite challenge, whether dissent is met with curiosity or retribution. The signal is continuous and teams read it accurately. Closing that gap is a retention strategy, a performance strategy, and a cost-reduction strategy simultaneously. In practice, this means adding team climate to the metrics that actually get reviewed, running short pulse surveys that ask people whether they feel safe to raise concerns, and making those results part of leadership performance conversations rather than anonymising them into irrelevance. What gets measured gets attended to.
Acknowledge and redistribute invisible labour.
Research by Babcock and colleagues in the American Economic Review found that women in teams carry a disproportionate share of relational work, mentoring, conflict mediation, emotional support, onboarding, work that is cognitively expensive, rarely recognised formally, and almost never reflected in compensation or evaluation. This work is a resource the organisation is consuming without accounting for. In practice, the starting point is simply making it visible: mapping who is doing what relational work across the team, naming it in role descriptions and performance conversations, and rotating it deliberately rather than allowing it to settle on the same people by default. The cost of doing this is a conversation. The cost of not doing it accumulates on the people already carrying the most.
Design AI adoption around human cognitive capacity.
Organisations that navigate AI without compounding their burnout problem will be the ones that treat human cognitive load as a design constraint. In practice, this means asking a different question when introducing AI tools: not “how much more can we produce?” but “where does this genuinely reduce burden, and where does it quietly increase it?” It means building protected time for reflection, judgement, and deep work into team rhythms rather than assuming people will create that space themselves. And it means being honest with teams about the pace the organisation expects to sustain, because that conversation, uncomfortable as it may be, is far less costly than the attrition and absence that follows when no one has it.
The return on that investment
The cost of inaction is already being paid. Stress-related absence and poor mental health now cost UK businesses between £21.6 billion and £28 billion annually. Mental health is the single largest driver of lost working days in the country, accounting for 22.1 million days in the most recent reporting period. Attrition driven by burnout costs between 50% and 200% of an employee’s annual salary to recover from, depending on seniority.
Against those figures, the cost of educating leaders in the neuroscience of stress and belonging, of introducing structured team health measures, of redistributing invisible labour, of building AI adoption policy around human capacity, is modest. The return shows up in reduced absence, lower attrition, higher quality thinking, and teams genuinely capable of performing at the level organisations need from them. The organisations that have begun this work are doing so because they have looked at the numbers and understood that the environment they have built is expensive in ways they had not previously calculated.
The workforce is already deciding
A further pressure is building for organisations that choose not to act voluntarily.
Deloitte’s 2025 Gen Z and Millennial Survey, drawing on responses from more than 23,000 people across 44 countries, found that these two generations, projected to make up 74% of the global workforce by 2030, are seeking what the report calls a trifecta: money, meaning, and wellbeing. Only 6% of Gen Z said their primary career goal is to reach a senior leadership position. 89% of Gen Z and 92% of millennials said a sense of purpose is very or somewhat important to their job satisfaction. 44% of Gen Z and 45% of millennials have already left a role they felt lacked purpose. Around 40% of both groups have turned down an employer based on their approach to employee mental wellbeing.
Burnout rising year on year, despite a decade of wellbeing investment, is what happens when environments consistently fail to provide what the nervous system needs to function. The cost is counted in working days lost, in attrition, in the thinking that never quite gets offered, and in the talent that eventually leaves for somewhere that asks less of them for no good reason.
Organisations that understand this, and respond by treating the environment itself as the intervention, will be the ones still able to attract and retain exceptional people a decade from now. The ones that continue reaching for another programme while the structural conditions remain unchanged will keep generating the same data, and paying the same price for it.
References
- Health and Safety Executive, Work-related stress, anxiety or depression statistics in Great Britain 2024. hse.gov.uk
- Work.life, Workplace Stress in the UK: 2025 Statistics, Causes and Solutions. work.life
- Mental Health First Aid England, Key Workplace Mental Health Statistics 2024. mhfaengland.org
- Aflac, WorkForces Report 2025-2026. 15th annual survey of 1,002 employers and 2,000 employees, April-May 2025. aflac.com
- McKinsey & Company, Women in the Workplace 2025. mckinsey.com
- McEwen, B.S. (1998). Stress, adaptation, and disease: Allostasis and allostatic load. Annals of the New York Academy of Sciences, 840(1), 33-44.
- Arnsten, A.F.T. (2009). Stress signalling pathways that impair prefrontal cortex structure and function. Nature Reviews Neuroscience, 10, 410-422.
- Bakker, A.B. & Demerouti, E. (2007). The Job Demands-Resources model: State of the art. Journal of Managerial Psychology, 22(3), 309-328.
- Jahoda, M. (1982). Employment and Unemployment: A Social-Psychological Analysis. Cambridge University Press.
- Edmondson, A.C. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350-383. See also: Edmondson, A.C. (2018). The Fearless Organization. Wiley.
- UC Berkeley / Harvard Business Review, AI Doesn’t Reduce Work - It Intensifies It. Eight-month study, February 2026. hbr.org
- Bedard, J. et al. (2026). When Using AI Leads to “Brain Fry”. Harvard Business Review, March 2026. hbr.org
- Deloitte Global, 2025 Gen Z and Millennial Survey. 23,482 respondents across 44 countries, October-December 2024. deloitte.com
- Babcock, L., Recalde, M.P., Vesterlund, L. & Weingart, L. (2017). Gender differences in accepting and receiving requests for tasks with low promotability. American Economic Review, 107(3), 714-747.
- HSE / AXA UK and Cebr, Mind Health Study 2025. Workplace Insight, November 2025. workplaceinsight.net
- Health and Safety Executive, Health and Safety Statistics 2025. hse.gov.uk