Empathy is often described as an unqualified good. The more deeply we feel another person’s experience, the kinder and more connected we should become.

Psychology offers a more complicated picture. Understanding someone’s perspective can improve care. Compassion can motivate help. But absorbing another person’s distress until it becomes personal distress can leave the observer overwhelmed, particularly when recovery time, control and support are scarce.

A large global analysis does not test that exhaustion mechanism directly. It does place it beside a wider and unsettling pattern. Researchers analysed 28,000 emotional-intelligence assessments from adults in 166 countries between 2019 and 2024. Average scores fell across every one of the eight competencies measured.

Empathy was one of them. So were identifying emotions, recognising habitual reactions, thinking through consequences, regulating emotion, sustaining internal motivation, exercising optimism and connecting daily choices to purpose.

The authors call the pattern an “emotional recession.” The phrase is evocative, but it is their interpretive framework, not a clinical diagnosis. The data support a narrower conclusion: different groups of assessment takers reported fewer emotional resources in 2024 than comparable groups did in 2019.

“Empathy” is an umbrella term. Cognitive empathy involves understanding another person’s perspective or mental state. Affective empathy involves sharing something of their feeling. Empathic concern and compassion add warmth, care and a wish for the other person to be well.

Personal distress is different. It is the aversive, self-focused discomfort that can arise when witnessing suffering becomes difficult to tolerate. The attention can shift from the person in need towards escaping one’s own overload.

In a widely cited review of empathy and compassion, neuroscientists Tania Singer and Olga Klimecki argued that repeatedly resonating with suffering can contribute to empathic distress, while compassion may support care without the same negative emotional burden.

This distinction stops a useful idea from turning into the claim that caring is harmful. The risk is not simply “too much empathy.” It is emotional resonance that overwhelms regulation and remains unsupported.

Klimecki and colleagues tested this distinction in an experiment on empathy and compassion training. Participants viewed videos of people suffering. After empathy training, exposure was associated with more negative feeling and activity in regions linked to pain-related processing. Compassion training later increased positive feeling and recruited a different pattern of brain activity.

The experiment was relatively small and cannot reproduce the daily pressures of nursing, teaching, parenting or caring for a relative. It did show that sharing distress and cultivating compassionate concern are psychologically separable responses, not two labels for the same state.

Newer evidence reinforces the need for precision. A 2026 meta-analysis of 29 nursing studies found that overall empathy was associated with less burnout, not more. The subdomains moved differently: personal distress had a moderate positive association with burnout, while perspective taking had a weak negative association.

Most of that literature is observational, so direction remains uncertain. Distress may contribute to exhaustion, exhaustion may make distress harder to regulate, or both may be driven by unsafe workloads and inadequate support. The best question is not whether empathy is good or bad. It is which component, under which demands, with which resources.

The new peer-reviewed “Emotional Recession” study used the Six Seconds Emotional Intelligence Assessment, a 77-item self-report questionnaire commonly completed in coaching, education and organisational-development programmes.

The researchers selected 28,000 records from a database containing 276,972 assessment takers. They balanced the analytical sample across seven global regions, adults above and below 40, and male and female respondents. Records came from 166 countries.

One methodological point is essential. The researchers did not follow the same 28,000 adults for six years. They compared annual averages from different participants. This was a repeated cross-sectional analysis, not evidence that each individual lost emotional skill over time.

Average overall emotional-intelligence scores were 5.79 per cent lower in 2024 than in 2019. Emotional literacy fell by 2.3 points, recognising patterns by 2.2, consequential thinking by 1.6, navigating emotions by 2.0, intrinsic motivation by 3.0, optimism by 3.0, empathy by 1.6 and pursuing noble goals by 2.6.

All eight differences were statistically significant. The overall standardised difference was Cohen’s d of 0.22, which is small. Empathy’s 1.6-point fall was tied for the smallest raw change, while optimism and intrinsic motivation declined most.

The period covers the pandemic and its aftermath, making prolonged uncertainty, disrupted relationships and altered work hard to ignore. Yet the study did not isolate the pandemic, remote work, social media, inflation or any other explanation.

It also did not measure empathic distress, compassion fatigue or burnout. The assessment defined empathy as recognising and appropriately responding to other people’s emotions. The older empathy literature offers a plausible account of how emotional demands can exceed available resources, but it was not tested as the mechanism behind this global trend.

The downward pattern was concentrated earlier in the 2019 to 2024 period, with later averages appearing to level rather than fall ever faster. The data show a sustained gap more clearly than an accelerating collapse.

Emotional intelligence itself is not a single universally agreed construct. A systematic review found 40 different measures across ability, trait and mixed models. This study establishes a pattern within one particular framework, not across every scientific definition of emotional skill.

Geographic reach should not be confused with population representation. Participants generally encountered the assessment through training, coaching and organisational programmes. The authors acknowledge that professionals in leadership, learning and human development, along with people in postsecondary education, were overrepresented.

The questionnaire measured self-perception rather than observed behaviour. It did not watch participants comfort someone, resolve a conflict or regulate distress under pressure. Both the emotional competencies and four reported life outcomes came from the same instrument, which can strengthen correlations through common method alone.

Two competency scales had internal-consistency estimates of 0.64 and 0.65, slightly below the conventional 0.70 benchmark. The minimum score change that would count as reliable change for one person has not been established. Raw data were not publicly available because the assessment is proprietary.

All four authors worked for Six Seconds, the organisation behind the instrument. They disclosed that relationship. Employment does not negate the findings, but it matters when an organisation analyses its own tool and gives the trend a distinctive name.

Emotional demands do not operate in a vacuum. Job Demands-Resources theory describes how sustained demands become more damaging when people lack resources such as autonomy, social support, useful feedback, adequate staffing and time to recover.

That helps explain why telling exhausted people to become more empathetic or resilient can miss the problem. Emotional literacy and perspective taking may be teachable. They cannot create an extra nurse on a short-staffed ward, make an insecure job safe or restore an evening consumed by work.

Individual capacity still matters. People can learn to notice when another person’s feeling has become their own personal distress, take perspective without losing boundaries and move from helpless resonance towards useful care. As ScienceBlog’s earlier look at emotional regulation across adulthood showed, experience can change how strongly people react to interpersonal strain even when other abilities decline.

But support is part of that capacity, not an optional reward for possessing it. Trusted colleagues, realistic demands, rest and permission to step back can determine whether sensitivity remains a route to connection or becomes a route to depletion.

The global study does not prove that humanity is losing emotional intelligence, or that empathy exhaustion caused a worldwide decline. It suggests that, within one broad but imperfect window, adults in 2024 reported less access to every emotional skill the questionnaire measured than adults did in 2019.

That is concerning without being a verdict. Emotional capacities are not bottomless, but neither are they fixed. Protecting them requires both practice within people and support around them.