In 1990, two Yale psychologists named Peter Salovey and John D. Mayer published a paper in the journal Imagination, Cognition, and Personality that would eventually reshape popular culture in ways they could not have anticipated. The paper, titled "Emotional Intelligence," was a careful, modest scientific proposal: that there might exist a coherent mental ability to monitor one's own and others' emotions, to discriminate among them, and to use this information to guide thinking and action. Salovey and Mayer were cautious. They proposed a testable construct, drew on existing work in emotion research and cognitive psychology, and acknowledged the limits of their evidence. What they launched was a research program.
What happened next was not science — at least not primarily. In 1995, science journalist Daniel Goleman published Emotional Intelligence: Why It Can Matter More Than IQ, a book that sold more than five million copies and converted a technical hypothesis into a cultural conviction. Goleman claimed that EQ — emotional quotient — could account for up to 80 percent of the difference in people's success in life, a figure that had no serious empirical support. He argued that EI predicted leadership effectiveness, academic achievement, relationship quality, and occupational attainment well beyond what conventional IQ could explain. Corporations adopted EI competency frameworks. Schools built social-emotional learning curricula. The Harvard Business Review declared EI "the sine qua non of leadership." The claim had outrun the data by an enormous margin.
By the early 2000s, a growing body of psychologists was pushing back. Edwin Locke, Frank Landy, Kristof-Stefan Petrides, and a succession of meta-analysts had begun dismantling specific claims, questioning the construct's definitional coherence, and demanding that EI be held to the same evidentiary standards as any other proposed psychological trait. What emerged from that correction is more interesting than the popular story in either direction — EI is neither the superpower Goleman claimed nor the pseudoscientific fiction its harshest critics alleged. It is a legitimate but complicated construct whose scientific value depends almost entirely on how it is defined and measured.
"Emotional intelligence is the ability to perceive, use, understand, and manage emotions in oneself and others." — Peter Salovey & John Mayer, 1990
Three Models, Three Constructs
The confusion around emotional intelligence stems in part from the fact that the label covers at least three distinct theoretical entities that predict different outcomes, use different measurements, and command different levels of scientific respect.
| Dimension | Ability Model (Mayer, Salovey, Caruso) | Mixed Model (Goleman) | Trait Model (Petrides, Furnham) |
|---|---|---|---|
| Core definition | A cognitive ability to perceive, use, understand, and manage emotion | A composite of competencies: self-awareness, self-regulation, motivation, empathy, social skill | Emotion-related self-perceptions and dispositions assessed as personality facets |
| Primary measurement | Performance-based MSCEIT (correct answers scored against expert consensus or statistical norms) | Self-report or 360-degree rating scales (ECI, ECI-2) | Self-report questionnaires (TEIQue); part of broader personality space |
| Predictive validity | Modest but incremental above IQ and Big Five for emotional outcomes; some evidence for job performance in high-EI-demand roles | Larger apparent correlations, but heavily confounded with personality traits already captured by Big Five | Strong prediction of well-being and life satisfaction; largely overlaps with established personality constructs |
| Scientific status | Best-supported as distinct ability construct; peer-reviewed measurement with ongoing debate about scoring validity | Widely criticized as atheoretical aggregation; criticized by Locke (2005), Landy (2005) for lacking discriminant validity | Openly personality-based; well-validated within personality psychology framework; critics argue it adds little beyond existing measures |
| Main critique | Incremental validity over IQ and personality still modest; scoring consensus norms questioned | Construct validity weak; conflates unrelated traits; inflated real-world claims | Tautological overlap with neuroticism, agreeableness, and extraversion limits unique predictive contribution |
The Cognitive Science Foundation
The Salovey-Mayer model did not emerge from a vacuum. Its intellectual scaffolding rested on two prior lines of cognitive science. The first was Howard Gardner's 1983 theory of multiple intelligences, which proposed interpersonal and intrapersonal intelligences as distinct cognitive capacities, legitimizing the idea that there might be intellectual faculties beyond those measured by classical IQ tests. Salovey and Mayer cited Gardner explicitly and positioned EI as a more rigorously testable version of his interpersonal constructs.
The second foundation was the rapidly expanding literature on emotion and cognition. Antonio Damasio's somatic marker hypothesis, developed through studies of patients with ventromedial prefrontal cortex damage (most famously his patient "Elliot"), demonstrated that damage disrupting emotional processing profoundly impaired rational decision-making — a finding published across a series of papers culminating in Descartes' Error (1994). If emotion was integral to cognition rather than opposed to it, the idea that some people might be more skilled at deploying emotional information cognitively became scientifically plausible.
Robert Sternberg's triarchic theory of intelligence (1985) provided a third scaffolding: an explicit model of practical and experiential intelligence distinct from analytic ability. Salovey and Mayer inherited a field already pushing against the narrow g-factor conception of intelligence, and EI fit neatly into that broader movement.
Richard Lazarus's transactional model of stress and coping (1984) contributed the functional view of emotions as appraisals — cognitive evaluations of significance — rather than mere physiological events. If emotions encode meaning, then reading them accurately is a form of information processing, which made the notion of emotional skill conceptually coherent.
Four Case Studies in EI Research
Case Study 1: The MSCEIT and Incremental Validity In 2011, Marina Fiori and John Antonakis published a study in the Journal of Individual Differences directly testing whether ability EI, as measured by the MSCEIT, predicted outcomes above and beyond general intelligence (g). Their sample of 201 adults showed that the MSCEIT's experiential branch (perceiving and using emotions) retained incremental validity for social outcomes after controlling for g, while the strategic branch (understanding and managing emotions) showed weaker independence from verbal intelligence. The finding was important precisely because it was carefully bounded: some components of ability EI appeared to measure something distinct from IQ, but the separation was partial and dimension-specific, not global.
Case Study 2: The Meta-Analysis of EI and Job Performance David L. Van Rooy and Chockalingam Viswesvaran conducted a comprehensive meta-analysis published in the Journal of Vocational Behavior in 2004, synthesizing 69 studies and 9,522 participants. They found an uncorrected mean correlation of r = .23 between EI and performance criteria — a modest but real effect. Critically, they found that most EI measures correlated substantially with general intelligence and personality, raising the confounding problem that would dominate the field for the next decade: if EI scores track things we already measure with established tools, what unique variance are they explaining? Van Rooy and Viswesvaran concluded that EI was a valid predictor but could not yet claim unique explanatory power.
Case Study 3: The Cascading Model and the Challenge to Goleman Dana L. Joseph and Daniel A. Newman published a meta-analytic investigation in the Journal of Applied Psychology in 2010 that directly tested the mechanisms through which EI relates to job performance. Their cascading model proposed that EI predicts performance through a specific sequence: emotional perception enables emotional understanding, which enables emotional regulation, which finally affects performance through its relationship with personality and cognitive ability. Across 65 studies, they found that mixed-model EI measures largely collapsed into already-known constructs once personality and IQ were controlled, but ability EI retained some incremental validity specifically in jobs high in emotional labor demands. The paper was a careful, constructive critique — not a dismissal but a structural renegotiation of where in the causal chain EI actually sits.
Case Study 4: Trait EI and Well-Being Konstantinos Petrides and Adrian Furnham developed the Trait Emotional Intelligence Questionnaire (TEIQue) and argued explicitly that trait EI was a personality construct properly located at the lower levels of personality hierarchies, not a cognitive ability. In a 2003 paper in the European Journal of Personality, they demonstrated that trait EI predicted life satisfaction and affect balance beyond the Big Five personality factors — though the incremental variance explained was small. Their transparency about EI's personality-adjacent status was both a scientific strength and the source of the main critique: if trait EI is essentially personality by another name, its conceptual contribution depends on whether its specific facet combination offers something the Big Five architecture does not.
Intellectual Lineage
The intellectual ancestry of EI winds through several distinct disciplines. The proximate origin is Salovey and Mayer (1990), who synthesized emotion-cognition research from cognitive psychology and proposed the first formal model. Their work built on earlier ideas by David Wechsler, who had noted as early as 1943 that non-intellective factors contributed to intelligent behavior — a prophetic aside that Salovey and Mayer elevated into a research program.
Goleman's popularization drew not from cognitive science but from developmental psychology, particularly the work of Carol Gilligan on relational intelligence and Lawrence Kohlberg on moral reasoning, filtered through his own reading of Damasio and Salovey. His synthesis was journalistic rather than theoretical, which explains both its cultural resonance and its scientific fragility.
The trait model lineage runs through Hans Eysenck and the psychometric personality tradition: Petrides and Furnham explicitly positioned trait EI within the hierarchical personality framework established by Eysenck and later refined by Costa and McCrae's five-factor model. Their work connects EI to a much older tradition of individual differences research, giving it psychometric rigor while simultaneously raising questions about genuine novelty.
The critiques of EI emerged from two distinct traditions. Edwin Locke (2005), writing in Emotion, came from an industrial-organizational psychology perspective shaped by the goal-setting literature, and argued that EI was a concept assembled from dissimilar elements with no unifying theoretical principle — effectively a political and commercial construction rather than a scientific discovery. Frank Landy (2005), in Journal of Organizational Behavior, provided a historical and methodological critique, documenting the gap between EI's empirical record and the claims made in applied settings, and questioning whether the construct had yet earned its widespread use in personnel selection.
Empirical Research: What the Evidence Actually Shows
The empirical literature on EI has matured considerably since Goleman's maximalist claims. Several patterns are now reasonably well established.
First, ability EI as measured by the MSCEIT demonstrates discriminant validity from personality traits. A comprehensive review by John D. Mayer, Richard D. Roberts, and Sigal G. Barsade in the Annual Review of Psychology (2008) concluded that the MSCEIT correlates at approximately r = .35 with general intelligence, indicating meaningful but partial overlap, and shows lower correlations with Big Five traits than self-report EI measures. This supports the position that something distinct from personality and IQ is being captured — though the effect sizes remain modest.
Second, cultural and gender differences in EI measurement present interpretive complications. Several studies, including work by Brackett, Mayer, and Warner (2004) in Personality and Social Psychology Bulletin, found consistent gender differences on the MSCEIT, with women scoring higher on emotion perception and expression tasks. Whether these differences reflect genuine ability differences, cultural socialization of emotional expression norms, or artifacts of the consensus scoring method remains unresolved. Cross-cultural studies (e.g., Fernandez-Berrocal et al., 2006) have documented that EI profiles vary across national samples in ways that complicate universal claims about the construct.
Third, educational applications have produced a more encouraging literature. Research on social-emotional learning (SEL) programs, synthesized by Durlak et al. (2011) in Child Development, found that well-designed school-based SEL programs produced significant improvements in social skills, attitudes, and academic performance, with effect sizes in the range of .20 to .30 for academic achievement. However, these programs intervene on a broad range of competencies, and whether EI specifically (versus general social skills or emotion regulation capacity) drives the effects cannot be cleanly isolated.
Fourth, clinical applications have shown promise in specific populations. EI measures predict alexithymia (difficulty identifying and describing emotions), and several studies have linked lower ability EI to poorer outcomes in anxiety disorders and depression, though the causal direction and clinical utility remain active areas of investigation.
Limits, Critiques, and Nuances
The most penetrating critiques of EI do not come from hostile dismissal but from careful methodological examination. Locke (2005) argued that EI violated basic conceptual criteria for an intelligence: it lacked a common underlying process, combined traits that were neither necessarily correlated nor necessarily related to emotion per se, and was defined in ways that made it almost impossible to falsify. His critique was specifically leveled at Goleman's mixed model but raised questions that the ability model had to answer.
The scoring problem for the MSCEIT remains genuinely unresolved. The test uses consensus scoring — an answer is "correct" if it matches what most respondents choose — or expert scoring, with emotion researchers as the normative reference group. Both methods have been criticized: consensus scoring may capture cultural norms rather than emotional ability, and expert scoring assumes that researchers are more emotionally accurate than laypersons, which has not been independently validated. Roberts, Zeidner, and Matthews (2001) raised these concerns in Emotion, and subsequent researchers have not fully answered them.
The incremental validity problem is the central practical concern. Schulte, Ree, and Carretta (2004) found that when general intelligence and the Big Five were both controlled, self-report EI explained essentially no unique variance in academic performance. This pattern — EI as a redundant predictor — appears more consistently for self-report measures than for ability measures, but even ability EI's incremental contribution is small enough that its addition to personnel selection batteries is difficult to justify on efficiency grounds alone.
There is also a well-documented susceptibility of self-report EI measures to faking and social desirability responding. Grubb and McDaniel (2007) demonstrated that motivated respondents could substantially inflate their EI scores without specific coaching, a problem that undermines the construct's use in high-stakes selection contexts.
Cultural and gender issues in EI research constitute a longer-term unresolved challenge. The consensus scoring method is, by construction, culturally relative: the "correct" emotional response is defined by the majority response of the norming sample. If that sample is predominantly Western, educated, and from individualistic cultures, the scoring norms may not generalize to populations with different emotional display rules or expressive cultures. Research on this question is ongoing but remains underdeveloped relative to the construct's applied use.
Despite these limitations, EI retains genuine scientific interest in its ability model form. The construct maps onto a real phenomenon — people differ meaningfully in how well they read emotional signals, regulate their affective states, and deploy emotional information in problem-solving — and these differences have consequences for social functioning, mental health, and professional effectiveness that existing trait frameworks do not fully capture. The question is not whether emotional competence matters, which is almost certainly true, but whether EI as a formal construct provides a scientifically useful framework for understanding it beyond what we already know from personality and intelligence research.
References
Salovey, P., & Mayer, J. D. (1990). Emotional intelligence. Imagination, Cognition and Personality, 9(3), 185-211.
Goleman, D. (1995). Emotional intelligence: Why it can matter more than IQ. Bantam Books.
Mayer, J. D., Roberts, R. D., & Barsade, S. G. (2008). Human abilities: Emotional intelligence. Annual Review of Psychology, 59, 507-536.
Van Rooy, D. L., & Viswesvaran, C. (2004). Emotional intelligence: A meta-analytic investigation of predictive validity and nomological net. Journal of Vocational Behavior, 65(1), 71-95.
Joseph, D. L., & Newman, D. A. (2010). Emotional intelligence: An integrative meta-analysis and cascading model. Journal of Applied Psychology, 95(1), 54-78.
Fiori, M., & Antonakis, J. (2011). The ability model of emotional intelligence: Searching for valid measures. Personality and Individual Differences, 50(3), 329-334.
Locke, E. A. (2005). Why emotional intelligence is an invalid concept. Journal of Organizational Behavior, 26(4), 425-431.
Landy, F. J. (2005). Some historical and scientific issues related to research on emotional intelligence. Journal of Organizational Behavior, 26(4), 411-424.
Petrides, K. V., & Furnham, A. (2003). Trait emotional intelligence: Behavioural validation in two studies of emotion recognition and reactivity to mood induction. European Journal of Personality, 17(1), 39-57.
Roberts, R. D., Zeidner, M., & Matthews, G. (2001). Does emotional intelligence meet traditional standards for an intelligence? Some new data and conclusions. Emotion, 1(3), 196-231.
Brackett, M. A., Mayer, J. D., & Warner, R. M. (2004). Emotional intelligence and its relation to everyday behaviour. Personality and Individual Psychology Bulletin, 30(9), 1387-1402.
Durlak, J. A., Weissberg, R. P., Dymnicki, A. B., Taylor, R. D., & Schellinger, K. B. (2011). The impact of enhancing students' social and emotional learning: A meta-analysis of school-based universal interventions. Child Development, 82(1), 405-432.
Frequently Asked Questions
What is Emotional Intelligence?
Emotional Intelligence (EI) was originally defined by Salovey and Mayer (1990) as the ability to perceive, appraise, and express emotion accurately; to access and generate feelings that facilitate thought; to understand emotion and emotional knowledge; and to regulate emotions to promote growth. Goleman's 1995 popular reformulation expanded it to include motivation, empathy, and social skills — a move that significantly increased its appeal but weakened its scientific precision.
What are the three main models of Emotional Intelligence?
The ability model (Mayer, Salovey, Caruso) treats EI as a genuine cognitive ability measurable by performance tasks (MSCEIT). The mixed model (Goleman) combines emotional abilities with personality traits and motivational factors. The trait model (Petrides, Furnham) reconceptualizes EI as a personality trait measurable by self-report. The three models produce different measures, different predictions, and different evidence bases — they are not simply different ways of measuring the same thing.
Does Emotional Intelligence predict job performance?
Van Rooy and Viswesvaran's (2004) meta-analysis found EI predicted job performance with r = .24 — significant but modest, and smaller than general intelligence (r = .51) or conscientiousness (r = .31). Joseph and Newman's (2010) cascading model showed EI predicts performance mainly through emotional labor jobs, not across all occupations. Schulte et al. (2004) found that when personality is controlled, ability EI adds little incremental validity.
What are the main criticisms of Emotional Intelligence?
Edwin Locke (2005) argued that EI is a political and commercial concept rather than a scientific construct — that it stretches the definition of intelligence beyond coherence. Frank Landy (2005) reviewed the measurement and validity literature and found widespread overstatement of EI's predictive power. The consensus scoring problem in the MSCEIT (using group agreement rather than objective answers) has been challenged as circular. Trait EI overlaps substantially with established personality dimensions, raising discriminant validity questions.
What does Emotional Intelligence add beyond IQ and personality?
Fiori and Antonakis (2011) found that ability EI predicts academic achievement and social outcomes incrementally above general intelligence — but mainly when the measure is pure emotional ability (MSCEIT) rather than mixed-model self-report. The incremental validity is specific, not general: EI adds predictive value primarily in domains involving emotional labor, interpersonal performance, and situations where emotion recognition and regulation are central task requirements.