Emotions Are Not Hardwired — So What Is Emotional Intelligence, Really?

The neuroscience has moved on. The $2 billion EI industry hasn't. What survives, what doesn't, and what it means for how we train leaders.

The Premise We Stopped Questioning

In 1995, Daniel Goleman published Emotional Intelligence: Why It Can Matter More Than IQ. The book stayed on the New York Times bestseller list for over a year, sold five million copies, and launched what Harvard Business Review called "a revolutionary, paradigm-shattering idea." Within a decade, emotional intelligence had become the dominant framework for leadership development worldwide — an industry now estimated at over $2 billion globally, built on EI assessments, coaching certifications, and corporate training programmes.

The neuroscience underneath that framework rested on a specific chain of claims. Paul Ekman's cross-cultural facial expression studies (1970s–1990s) provided the empirical base: six basic emotions — anger, disgust, fear, happiness, sadness, surprise — each with a universal, recognisable facial signature. Joseph LeDoux's amygdala research (1990s) provided the neural architecture: a fast "low road" from thalamus to amygdala that triggers emotional responses before conscious processing. Goleman wove these into a commercial framework: emotions are discrete internal events; emotionally intelligent people recognise them in themselves and others; and this recognition, combined with regulation skills, predicts leadership effectiveness more powerfully than IQ.

Thirty years later, the neuroscience has moved substantially. Three decades of fMRI data — including a landmark 2025 7-Tesla imaging study in Nature Neuroscience — have redrawn the map of how emotion works in the brain. Neuroendocrine research spanning 33 years of baboon fieldwork in Kenya has revealed how hormones and social hierarchy shape the very capacity for emotional regulation. Cross-cultural replication studies have challenged the universality of facial expression recognition. And one of the original architects of the amygdala-fear model, LeDoux himself, has publicly argued that his own work was misread — that the amygdala produces defensive survival responses, not conscious feelings of fear.

This post draws on the work of ten major researchers across three competing theoretical positions. The question is not whether emotional competence matters for leadership — the correlational evidence says it does. The question is whether the mechanism Goleman described is correct, what the updated neuroscience actually shows, and what changes when you train leaders on a map that no longer matches the territory.

Three Theoretical Camps, Ten Researchers

The science of emotion is not a settled field. It is an active, often acrimonious, multi-decade dispute among researchers who disagree on foundational questions — whether emotions are hardwired or constructed, whether they live in specific circuits or distributed networks, whether the amygdala is a fear centre or a general-purpose relevance detector.

Camp 1: Innate Circuit Theory

Paul Ekman (1934–2025) spent five decades studying facial expressions across cultures. His defining experiment used a forced-choice paradigm: participants were shown photographs of posed facial expressions and asked to match them to one of six emotion labels. The protocol was replicated across 21 cultures, including the isolated Fore tribesmen of Papua New Guinea who had minimal exposure to Western media. Agreement rates were 80–90% across literate cultures. By 1978, Ekman and Wallace Friesen had codified the Facial Action Coding System (FACS) — an anatomically-based method for decomposing facial movements into individual action units, achieving inter-rater reliability above 0.90. The protocol has been replicated nearly 200 times.

But the method has come under sustained challenge. The forced-choice format inflates agreement through elimination. When Barrett and colleagues removed the pre-set word list and used free-response methods, recognition rates among American participants dropped below 50%. When repeated with the Hadza people of Tanzania using free-response methods, participants did not match expressions to emotions as Ekman's paradigm predicted. And in a naturalistic study tracking people's actual facial expressions in daily life, participants made the stereotypical expression predicted by Ekman — smiling when happy, scowling when angry — less than 35% of the time. They were only interpreted correctly when surrounding context was taken into account.

Jaak Panksepp (1943–2017) took an entirely different empirical route. Working at Bowling Green State University and later Washington State University, Panksepp spent decades using three methodologies on vertebrate brains — electrical stimulation, pharmacological challenge, and targeted lesions — to map seven primary-process emotional command systems, each anchored in evolutionarily ancient subcortical regions conserved across mammalian species.

Crucially, Panksepp could trigger these emotional behaviours only from subcortical regions — not from the cortex. The periaqueductal gray produced the most coherent emotional responses with the lowest stimulation thresholds. This is the strongest empirical evidence that something innate and circuit-specific exists at the affective foundations of the mammalian brain. Barrett's 2017 book did not substantially engage with this data — an omission Panksepp's followers consider a critical weakness in her constructionist account.

Joseph LeDoux, now retired from NYU, built his career mapping the circuit basis of threat conditioning in rodents. His dual-pathway model showed a fast subcortical route from thalamus directly to the amygdala (enabling rapid defensive responses before conscious processing) and a slower cortical route providing contextual evaluation. This model provided the neural architecture for Goleman's "emotional hijack."

Camp 2: Body Construction Theory

Lisa Feldman Barrett, University Distinguished Professor at Northeastern and faculty at Massachusetts General Hospital/Harvard Medical School, has mounted the most comprehensive challenge to the basic emotions paradigm. With over 117,000 Google Scholar citations, her theory of constructed emotion argues that emotions are not triggered by dedicated circuits but actively constructed by the brain using three ingredients: predictions based on past experience, current interoceptive signals from the body, and available conceptual categories learned through language and culture.

Her core empirical claim, supported by a 2025 paper in Perspectives on Psychological Science (co-authored with 16 colleagues): across over 1,000 fMRI studies, no emotion-specific neural "fingerprint" has been found. Population variation within any single emotion category exceeds the consistency between categories. Even the function of a single emotion category varies situationally — anger can serve to remove an obstacle, punish, compete, identify with a group, signal care, or avoid harm.

A 2025 study in Nature Neuroscience by Barrett's group used 7-Tesla fMRI (n=90) to map the allostatic-interoceptive system with unprecedented precision, confirming over 96% of anatomical connections documented in animal tract-tracing studies. This system — spanning the default mode and salience networks — performs two core functions: allostasis (predictive regulation of the body's energy needs) and interoception (modelling signals from the body's internal organs). It is not specific to emotion — the same network underlies cognition, perception, stress regulation, and memory.

Antonio Damasio, University Professor at USC, occupies the middle ground. His somatic marker hypothesis proposes that emotional body states bias decision-making in situations of complexity and uncertainty.

The primary evidence comes from the Iowa Gambling Task. Four decks of cards: two offer high immediate payoffs with higher penalties (net loss); two offer modest payoffs with small penalties (net gain). Participants draw cards and must learn which decks are advantageous. Three key findings defined the hypothesis: healthy participants' skin conductance responses spiked before choosing risky decks approximately 25 trials before conscious awareness; patients with bilateral vmPFC damage continued choosing losing decks despite normal IQ; and amygdala-lesioned patients showed similar deficits.

Robert Sapolsky, Stanford professor of biology, neurology, and neurosurgery, and MacArthur Fellow, adds the neuroendocrine layer. His multi-timescale framework in Behave (2017) traces any behaviour across seven temporal scales: seconds (neural), minutes (sensory environment), hours-days (hormones), months (neuroplasticity), years (development), decades (culture), millennia (evolution).

Camp 3: Applied Competency Theory

Daniel Goleman built the most commercially successful framework. In competency research across approximately 200 companies, he found that over 80% of competencies distinguishing superior leadership performance were related to emotional competence, with only 20% attributable to technical skill and cognitive ability. Reuven Bar-On developed the EQ-i assessment (α=0.92, r=0.29–0.62 with performance across 100,000+ samples in 15 countries). Travis Bradberry reported EI explains 58% of performance variance across industries.

The criticism: Mayer and Salovey, who coined "emotional intelligence" in 1990, are openly critical of Goleman's mixed model. Self-report EI measures correlate strongly with neuroticism (inverse) and extroversion from the Big Five, raising the question of whether EI is a distinct construct or a repackaging of existing personality dimensions.

What Specifically Breaks

The "Recognise Emotions" Assumption

Goleman's first pillar — self-awareness through recognising emotions — assumes stable, universal categories to recognise. The empirical picture is more contested than the EI industry acknowledges. Ekman's forced-choice method is well-replicated within its paradigm, but the paradigm itself is questioned: forced-choice inflates agreement, free-response recognition is significantly weaker, and real-world facial expressions are far more variable than posed photographs.

The honest position: there is likely something innate at the subcortical level (Panksepp's affective systems) that interacts with cortical prediction and construction (Barrett's framework). But the simple "six basic emotions, each with a recognisable face" story that EI training relies on is an oversimplification the data no longer fully supports.

The Amygdala-Hijack Model

Goleman popularised LeDoux's dual-pathway finding as the "emotional hijack." But LeDoux himself now argues the amygdala produces defensive survival responses, not the conscious feeling of fear. Barrett reinterprets it as a salience and uncertainty detector. And Sapolsky adds the hormonal complication: elevated testosterone adjusts the gain on the amygdala; chronic stress enlarges it through dendritic growth while simultaneously atrophying the prefrontal cortex — the regulator degrades under the very conditions that make regulation most necessary.

The "Manage Your Emotions" Framing

Goleman's framework treats emotions as internal events — things that happen to you, which you then identify and manage, like items on a dashboard. Barrett's constructionist view says emotions are not events that happen to you but predictions your brain makes. The intervention point shifts from "after the emotion fires" to "before it is constructed."

What Survives — And the Evidence That Matters More

The uncomfortable finding: Goleman's neuroscience was substantially wrong, but his behavioural prescriptions partially work — for reasons he did not understand.

Emotional Granularity Replaces Emotional Recognition

Barrett's concept of emotional granularity — the ability to make fine-grained distinctions in affective experience — is the scientifically robust mechanism underlying what Goleman loosely called "self-awareness." The difference between someone who says "I feel bad" and someone who distinguishes between disappointed, disillusioned, and demoralised is not pedantry. It is a measurable capacity that predicts regulation, decision-making, and health outcomes.

The Hormonal Context the EI Industry Ignores

Sapolsky's research reveals that what EI programmes call "emotional regulation" occurs against a hormonal backdrop that leaders are rarely conscious of and training programmes never address.

His fieldwork with wild olive baboons in Kenya — spanning 33 consecutive summers — produced some of the most striking data in behavioural neuroendocrinology. Working in a national reserve, he anaesthetised baboons using blowgun darts under controlled conditions (same time of day, not during illness, fights, or mating), collected blood samples, and measured cortisol by radioimmunoassay.

Sapolsky's specific findings on cortisol and leadership capacity: persistent HPA axis activation impairs the prefrontal cortex's capacity for empathy, curiosity, and objective decision-making. Elevated cortisol downregulates testosterone, increasing anxiety and risk aversion. Chronic stress decreases interest in new information and makes leaders more susceptible to surrounding themselves with agreeable advisors — seeking oxytocin-releasing social support rather than challenging feedback.

Interoceptive Awareness and the Clinical Evidence

Judson Brewer's randomised controlled trial of the Unwinding Anxiety app (published in JAMA Internal Medicine, 2021) demonstrated a 67% reduction in anxiety with a number needed to treat of 1.6 and 40% generalised anxiety disorder remission. The mechanism: interoception-based mindfulness — learning to observe bodily sensations (tightness, heat, constriction) without automatically constructing a catastrophic interpretation. Not identifying and labelling "anxiety" as a basic emotion, but interrupting the predictive cycle at the body-signal level.

The Scorecard

What Changes If We Take the Updated Science Seriously

The practical implications cut five ways.

From "emotion recognition" to "concept construction." Training should shift from teaching leaders to identify six basic emotions in others' faces toward expanding their emotional vocabulary and conceptual repertoire. The mechanism of emotional granularity produces measurably better regulation, decision-making, and interpersonal effectiveness. A leader who can distinguish between disappointed, disillusioned, and demoralised will respond differently to each.

From "manage your emotions" to "manage your allostatic budget." Sleep, nutrition, exercise, social connection, and circadian rhythm directly determine the quality of the brain's predictive machinery. A sleep-deprived leader is not failing at self-regulation through lack of willpower; their brain is operating with degraded hardware. This reframes organisational wellness from "employee benefit" to "prerequisite for the emotional competence the organisation says it values."

From "empathy as psychological skill" to "empathy as physiological capacity." Cortisol-driven prefrontal atrophy impairs the ability to model others' mental states. Leadership development that teaches empathy techniques without addressing the stress conditions that degrade empathic capacity is applying a software patch to a hardware problem.

From "EI is trainable" to "emotional construction is context-dependent." The training vector is different: change the concepts (vocabulary), change the body state (physiology), change the predictions (context exposure). The intervention point shifts upstream.

The hierarchy itself is a physiological intervention. Whether a leadership position comes with agency or surveillance, predictability or chronic uncertainty, directly shapes cortisol profiles and prefrontal cortex function. The same individual can show the stress physiology of subordination or the physiology of effective regulation depending entirely on the social environment. Sapolsky's baboon troop, after the aggressive males died and the culture shifted, demonstrated this with the clarity of a controlled experiment conducted by tuberculosis.

The central theoretical question remains unresolved. Panksepp's subcortical circuits — mapped through causal electrical stimulation, conserved across mammalian species — are real. Barrett's finding that no emotion-specific neural fingerprint emerges from cortical fMRI data is also real. LeDoux's revised position — that survival circuits produce nonconscious defensive responses while conscious feelings are cortical constructions — may be the bridge. Innate subcortical systems generate core affective states, which the cortex then constructs into rich, context-dependent, language-mediated emotional experience. Damasio's somatic markers provide the body-brain interface. And Sapolsky's hormonal and hierarchical context determines whether the machinery operates at full capacity or in a degraded state.

For leaders and for organisations investing in emotional development: the destination may be roughly the same. But the map has been redrawn. And the body — its sleep, its cortisol, its social environment, its metabolic budget — is no longer optional equipment.