Respond to two (2) colleagues

Emotion Review
Vol. 8, No. 4 (October 2016) 290 –300

© The Author(s) 2016
ISSN 1754-0739
DOI: 10.1177/1754073916639667
er.sagepub.com

In 1990, two of us proposed the existence of a new intelligence,
called “emotional intelligence.” Drawing on research findings in
the areas of emotion, intelligence, psychotherapy, and cognition,
we suggested that some people might be more intelligent about
emotions than others (Salovey & Mayer, 1990, p. 189). We called
attention to people’s problem solving in areas related to emotion:
recognizing emotions in faces, understanding the meanings of
emotion words, and managing feelings, among others. We argued
that, collectively, such skills implied the existence of a broader,
overlooked capacity to reason about emotions: an emotional
intelligence (Cacioppo, Semin, & Berntson, 2004; Haig, 2005).
We later characterized the problem-solving people carried out as
falling into four areas or “branches” (Mayer & Salovey, 1997).

In the present article, we revisit the theoretical aspects of our
ability model of emotional intelligence, update the model so as
to enhance its usefulness, and examine its implications. We
begin by considering a set of principles that guide our thinking
about emotional intelligence. After discussing these principles,
we revise the four-branch model slightly. We then locate emo-

tional intelligence amidst related “broad” intelligences, taking
care to distinguish emotional intelligence from personal and
social intelligences, and elucidate examples of reasoning for
each one of these intelligences. We wrap up by considering the
influence of the model and its implications for the future.

Seven Principles of Emotional Intelligence
We will describe a set of principles that have guided our theoriz-
ing about emotional intelligence. Together, these principles—
guidelines really—succinctly represent how we think about
emotional intelligence.

Principle 1: Emotional Intelligence Is a Mental
Ability

Like most psychologists, we regard intelligence as the capacity
to carry out abstract reasoning: to understand meanings, to grasp
the similarities and differences between two concepts, to formulate

The Ability Model of Emotional Intelligence:
Principles and Updates

John D. Mayer
Department of Psychology, University of New Hampshire, USA

David R. Caruso
Yale College Dean’s Office, Yale University, USA

Peter Salovey
Office of the President and Department of Psychology, Yale University, USA

Abstract

This article presents seven principles that have guided our thinking about emotional intelligence, some of them new. We have
reformulated our original ability model here guided by these principles, clarified earlier statements of the model that were unclear,
and revised portions of it in response to current research. In this revision, we also positioned emotional intelligence amidst other
hot intelligences including personal and social intelligences, and examined the implications of the changes to the model. We
discuss the present and future of the concept of emotional intelligence as a mental ability.

Keywords
ability measures, broad intelligences, emotional intelligence, personal intelligence, social intelligence

Author note: The authors gratefully acknowledge the assistance of Jessica Hoffmann, Zorana Ivcevic, Kateryna Sylaska, and Ethan Spector, whose comments on an earlier draft
strengthened this work in key areas.
Corresponding author: John D. Mayer, Department of Psychology, University of New Hampshire, 10 Library Way, Durham, NH 03824, USA. Email: [email protected]

639667 EMR0010.1177/1754073916639667Emotion ReviewMayer et al. The Ability Model of Emotional Intelligence
research-article2016

SPEcIAl SEctIon: EmotIonAl IntEllIgEncE

mailto:[email protected]

http://crossmark.crossref.org/dialog/?doi=10.1177%2F1754073916639667&domain=pdf&date_stamp=2016-08-18

Mayer et al. The Ability Model of Emotional Intelligence 291

powerful generalizations, and to understand when generaliza-
tions may not be appropriate because of context (Carroll, 1993;
Gottfredson, 1997). We agree also that intelligence can be
regarded as a system of mental abilities (Detterman, 1982).

Regarding how people reason about emotions, we proposed
that emotionally intelligent people (a) perceive emotions accu-
rately, (b) use emotions to accurately facilitate thought, (c)
understand emotions and emotional meanings, and (d) manage
emotions in themselves and others (Mayer & Salovey, 1997).

Principle 2: Emotional Intelligence Is Best
Measured as an Ability

A foundation of our thinking is that intelligences are best meas-
ured as abilities—by posing problems for people to solve, and
examining the resulting patterns of correct answers (Carroll,
1993; Mayer, Panter, & Caruso, 2012). (Correct answers are
those that authorities identify within the problem-solving area.)
The best answers to a question can be recognized by consulting
reference works, convening a panel of experts, or (more contro-
versially for certain classes of problems), by identifying a gen-
eral consensus among the test-takers (Legree, Psotka, Tremble,
& Bourne, 2005; MacCann & Roberts, 2008; Mayer, Salovey,
Caruso, & Sitarenios, 2003).

People are poor at estimating their own levels of intelli-
gence—whether it is their general intelligence or their emo-
tional intelligence (Brackett, Rivers, Shiffman, Lerner, &
Salovey, 2006; Paulhus, Lysy, & Yik, 1998). Because people
lack knowledge of what good problem-solving actually
entails, they estimate their abilities on other bases. These
include a mix of general self-confidence, self-esteem, misun-
derstandings of what is involved in successful reasoning, and
wishful thinking. These nonintellectual features add con-
struct-irrelevant variance to people’s self-estimated abilities,
rendering their judgments invalid as indices of their actual
abilities (Joint Committee, 2014).

Principle 3: Intelligent Problem Solving Does
Not Correspond Neatly to Intelligent Behavior

We believe there is a meaningful distinction between intelli-
gence and behavior. A person’s behavior is an expression of that
individual’s personality in a given social context (Mischel,
2009). An individual’s personality includes motives and emo-
tions, social styles, self-awareness, and self-control, all of which
contribute to consistencies in behavior, apart from intelligence.
Among the Big Five personality traits, for example, extraver-
sion, agreeableness, and conscientiousness correlate near zero
with general intelligence. Neuroticism correlates at r = −.15,
and openness about r = .30 (DeYoung, 2011). The Big Five
exhibit correlations of similar magnitude with emotional intel-
ligence: Neuroticism correlates r = −.17 with emotional intelli-
gence and openness r = .18; extraversion and conscientiousness
correlate with emotional intelligence between r = .12 and .15,
and agreeableness, r = .25 (Joseph & Newman, 2010). These
correlations indicate the relative independence of intelligences

from socioemotional styles. They confirm what everyday obser-
vation suggests: that emotionally stable, outgoing, and consci-
entious people may be emotionally intelligent or not.

Similarly, a person may possess high analytical intelligence
but not deploy it— illustrating a gap between ability and achieve-
ment (Duckworth, Quinn, & Tsukayama, 2012; Greven, Harlaar,
Kovas, Chamorro-Premuzic, & Plomin, 2009). Intelligence tests
tend to measure potential better than the typical performance of
everyday behavior. Many people with high levels of intelligence
may not deploy their ability when it would be useful (Ackerman
& Kanfer, 2004). For these reasons, the prediction from intelli-
gence to individual instances of “smart” behavior is fraught with
complications and weak in any single instance (Ayduk &
Mischel, 2002; Sternberg, 2004). At the same time, more emo-
tionally intelligent people have outcomes that differ in important
ways from those who are less emotionally intelligent. They have
better interpersonal relationships both in their everyday lives and
on the job—as articles in this issue and elsewhere address
(Fernández-Berrocal & Extremera, 2016; Izard et al., 2001;
Karim & Weisz, 2010; Lopes, 2016; Mayer, Roberts, & Barsade,
2008; Mayer, Salovey, & Caruso, 2008; Nathanson, Rivers,
Flynn & Brackett, 2016; Roberts et al., 2006; Rossen & Kranzler,
2009; Trentacosta, Izard, Mostow, & Fine, 2006)

Although intelligences predict some long-term behavioral
outcomes, predicting any individual behavior is fraught with
uncertainty because of the other personality—and social— vari-
ables involved (Funder, 2001; Mischel, 2009).

Principle 4: A Test’s Content—the Problem
Solving Area Involved—Must Be Clearly
Specified as a Precondition for the
Measurement of Human Mental Abilities

Establishing the content of the area. To measure emo-
tional intelligence well, tests must sample from the necessary
subject matter; the content of the test must cover the area of
problem-solving (Joint Committee, 2014). A test of verbal
intelligence ought to sample from a wide range of verbal
problems in to assess a test-taker’s problem-solving
ability. Test developers therefore must cover the key areas of
verbal problem-solving required, such as understanding
vocabulary, comprehending sentences, and other similar
skills. The specification of a problem-solving area—vocabu-
lary, sentence comprehension, and the like for verbal reason-
ing—defines the intelligence and its range of application. The
content specification is designed to ensure that the test sam-
ples a representative group of problems.

Subject matter differs from ability. Once the test’s content
is established, the test can be used to identify a person’s men-
tal abilities. People’s problem-solving abilities are reflected
by the correlational (or covariance) structure of the responses
they make to the test items. People’s abilities are revealed
when a group of scores on test items rise and fall together
across a sample of individuals. Note that the mental abilities
measured by a test are independent to some degree from the

292 Emotion Review Vol. 8 No. 4

nature of the problems to be solved. That is, a person’s abili-
ties will not necessarily correspond directly to the different
types of content in a subject area—a matter we consider fur-
ther in the next principle.

Principle 5: Valid Tests Have Well-Defined
Subject Matter That Draws Out Relevant
Human Mental Abilities

People exhibit their reasoning abilities as they solve problems
within a given subject area. As such, a test’s validity depends
both on the content it samples and the human mental abilities it
elicits. From this perspective, test scores represent an interaction
between a person’s mental abilities and the to-be-solved prob-
lems. If the test content is poorly specified, the items will misrep-
resent the domain, and any hoped-for research understanding of
mental abilities may be inconclusive. If problem-solving
domains overlap too much with other areas, ability factors redun-
dant with other areas may emerge; if the test content is too broad,
eclectic sets of ability factors may arise, and if the content is too
narrow the test may fail to draw out key mental abilities. A gar-
bage in, garbage out process will replace good measurement.

As implied in the previous lines, human abilities do not
necessarily map directly onto test content: The abilities people
use to solve problems have their own existence independent of
the organization of the subject matter involved. In the intelli-
gence field, a test of verbal knowledge may ask a person ques-
tions about nonfiction passages, fiction, poetry, and instruction
manuals. Despite the diversity of material, people use just one
verbal intelligence to comprehend them all. On the other hand,
the skill to identify what is missing in a picture and the skill to
rotate an object in space (in our minds) may appear to draw on
the same visual understanding. However, identifying the miss-
ing part of a picture draws primarily on perceptual-organiza-
tional intelligence whereas the object-rotation task draws
primarily on spatial ability, and these mental abilities are dis-
tinct (Wai, Lubinski, & Benbow, 2009). As applied to emo-
tional intelligence, we need both to describe accurately the
emotional problem solving that people undertake and the abil-
ities people employ to solve those problems—which are two
different matters (Joint Committee, 2014).

Principle 6: Emotional Intelligence is a Broad
Intelligence

We view emotional intelligence as a “broad” intelligence. The
concept of broad intelligences emerges from a hierarchical view
of intelligence often referred to as the Cattell–Horn–Carroll or
“three-stratum model” (McGrew, 2009). In this model, general
intelligence, or g, resides at the top of the hierarchy, and it is
divided at the second stratum into a series of eight to 15 broad
intelligences (Flanagan, McGrew, & Ortiz, 2000; McGrew,
2009). The model is based on factor-analytic explorations of
how mental abilities correlate with one another. Such analyses
suggest that human thinking can be fruitfully divided into areas
such as fluid reasoning, comprehension-knowledge (similar to

verbal intelligence), visual-spatial processing, working mem-
ory, long-term storage and retrieval, and speed of retrieval. The
three-stratum model also includes at its lowest level more spe-
cific mental abilities. For example, the broad intelligence,
“comprehension-knowledge” includes the specific ability to
understand vocabulary and general knowledge about the world.

Broad intelligences fall into subclasses (McGrew, 2009;
Schneider & Newman, 2015). One class of broad intelli-
gences reflects basic functional capacities of the brain such as
mental processing speed and the scope of working memory. A
second class of broad intelligences includes members identi-
fied by the sensory system they relate to, including auditory
intelligence and tactile/physical intelligence. Still others may
reflect subject matter knowledge such as verbal intelligence.
Mental abilities in late adolescence and adulthood may be
shaped and strengthened into “aptitude complexes” by educa-
tional pursuits and interests to form domain-specific knowl-
edge such as in mathematics, sciences, or government and
history (Ackerman & Heggestad, 1997; Rolfhus & Ackerman,
1999).

Emotional intelligence fits such descriptions of a broad
intelligence. MacCann, Joseph, Newman, and Roberts (2014)
collected data on 702 students who took a wide range of intel-
ligence tests, including one of emotional intelligence, over an
8-hour testing period. Using confirmatory factor analysis,
MacCann et al. (2014) found that emotional intelligence, indi-
cated by three of the four branches of the Mayer, Salovey,
Caruso Emotional Intelligence Test (MSCEIT; Mayer, Salovey,
& Caruso, 2002), fits well among other known broad intelli-
gences within the second-stratum of the Cattell–Horn–Carroll
model. In a reanalysis of the same data, Legree et al. (2014)
were also able to fit emotional intelligence into the Cattell–
Horn–Carroll framework; they included all four branches of
the MSCEIT as indicators of emotional intelligence by cor-
recting for the different response scales used across the test’s
subtasks (Legree et al., 2014).

Principle 7: Emotional Intelligence is a
Member of the Class of Broad Intelligences
Focused on Hot Information Processing

We believe that the broad intelligences—especially those
defined by their subject matter—can be divided into hot and
cool sets. Cool intelligences are those that deal with relatively
impersonal knowledge such as verbal-propositional intelli-
gence, math abilities, and visual-spatial intelligence. We view
hot intelligences as involving reasoning with information of
significance to an individual—matters that may chill our hearts
or make our blood boil. People use these hot intelligences to
manage what matters most to them: their senses of social
acceptance, identity coherence, and emotional well-being.
Repeated failures to reason well in these areas lead to psychic
pain which—at intense levels—is coprocessed in the same
brain centers that process physical pain (Eisenberger, 2015).
By thinking clearly about feelings, personality, and social
groups, however, people can better evaluate, cope with, and

Mayer et al. The Ability Model of Emotional Intelligence 293

predict the consequences of their own actions, and the behavior
of the individuals around them.

Emotional intelligence falls within this category because
emotions are organized responses involving physical changes,
felt experiences, cognitions, and action plans—all with strong
evaluative components (Izard, 2010). Social intelligence is
another member of the category (Conzelmann, Weis, & Süß,
2013; Hoepfner & O’Sullivan, 1968; Weis & Süß, 2007; Wong,
Day, Maxwell, & Meara, 1995). Social intelligence is “hot”
because social acceptance is fundamentally important to us;
among social animals, group exclusion is a source of primal
pain (Baumeister & Leary, 1995). Finally, personal intelli-
gence—an intelligence about personality—is a newly proposed
member of this group (Mayer, 2008, 2014; Mayer et al., 2012).
Personal intelligence is a hot intelligence because our sense of
self is a primary source of inner pleasure and pain—ranging
from self-satisfaction and pride on the positive side to self-
loathing and suicidal thoughts and action on the negative side
(Freud, 1962; Greenwald, 1980).

Summary and Applications

In this section, we described seven principles that guide our
thinking about emotional intelligence. We employed some of
these principles—notably that emotional intelligence is an abil-
ity and a hot intelligence—from the outset of our work. We also
introduced some new principles, such as those concerning broad
intelligences. In the next section, we review the four-branch
model of emotional intelligence and present an updated view of
our model and of our present thinking, recognizing that these
principles could lead to other models as well.

the Four-Branch model: original and
Revised
In this section of the article, we briefly revisit our 1997 four-
branch model of emotional intelligence and then proceed to
renew it—as well as to clarify its range of usefulness in the con-
text of the field’s current understanding of intelligences. More
specifically, we (a) add more abilities to the model, (b) distin-
guish the four-branch model of problem-solving content from
the structure of human abilities relevant to emotional intelli-
gence, (c) relate emotional intelligence to closely allied broad
intelligences, (d) examine the key characteristics of the prob-
lem-solving involved, and (e) more clearly distinguish between
areas of problem-solving and areas of human mental abilities.

The Four-Branch Model of Emotional
Intelligence

Our four-branch ability model distinguished among four areas
of problem-solving necessary to carry out emotional reasoning:
The first was (a) perceiving emotions, which we regarded as
computationally most basic. We then proceeded through the
increasingly integrated and more cognitively complex areas of
(b) facilitating thought by using emotions, (c) understanding

emotions, and (d) managing emotions in oneself and others
(Mayer & Salovey, 1997). (We referred to these problem-solv-
ing areas as branches after the line drawing in our original dia-
gram.)

Each branch represents a group of skills that proceeds devel-
opmentally from basic tasks to more challenging ones. The
Perceiving Emotions branch leads off with the “ability to iden-
tify emotions in one’s physical states, feelings, and thoughts,”
and proceeds to such developmentally advanced tasks (as we
saw them then) as the ability to discriminate between truthful
and dishonest expressions of feeling. The parallel developmen-
tal progression in the Understanding branch begins with the
ability to label emotions and progressed to more challenging
tasks such as understanding “likely transitions among emo-
tions,” such as from anger to satisfaction.

Update 1. The Four-Branch Model Includes
More Instances of Problem-Solving Than Before

Table 1 recapitulates the four branches of the original model in
its four rows, from perceiving emotions to managing emotions
(see left column). To the right, we have included many of the
original types of reasoning that illustrated each branch, some-
times rewriting them for clarity. Within a row, each set of abili-
ties is arranged (approximately) from the simplest to the most
complex skills, from bottom to top.

Based on research since 1997, we have added several areas
of problem solving to this revised model that initially we over-
looked. For example, the “Understanding Emotion” area origi-
nally included the abilities to label emotions, to know their
causes and consequences, and to understand complex emotions.
To those original areas of understanding, we have added emo-
tional appraisal and emotional forecasting—topics that have
experienced increased research attention and that have been
directly related to emotionally intelligent reasoning (see also
Barrett, Mesquita, & Gendron, 2011; Dunn, Brackett, Ashton-
James, Schneiderman, & Salovey, 2007; MacCann & Roberts,
2008)—as well as a sensitivity to cultural contexts (Matsumoto
& Hwang, 2012). As others have pointed out, reasoning in an
individual area is not necessarily discrete; rather, problem-solv-
ing activities can spill or cascade into one another. For example,
emotion perception is often helpful to accurate emotion under-
standing (see Joseph & Newman, 2010).

Update 2: The Mental Abilities Involved
in Emotional Intelligence Remain To Be
Determined

When we first proposed the four-branch model, we believed it
could reasonably correspond to four mental ability factors in the
area (Mayer & Salovey, 1997). That said, the content domains
are independent of the mental abilities within the domain (by
Principles 4 and 5). In fact, the four-branch model is not well
reflected in the factor structure of our ability-based measures
(Legree et al., 2014; Maul, 2011; Palmer, Gignac, Manocha, &
Stough, 2005; Rossen, Kranzler, & Algina, 2008).

294 Emotion Review Vol. 8 No. 4

From an empirical standpoint, tasks on the Mayer–
Salovey–Caruso Emotional Intelligence Test (MSCEIT) have
been represented by between one and three factors (Legree
et al., 2014; MacCann et al., 2014). Those theorists who favor
a three-factor model have argued for dropping Branch 2,
Facilitating Thought Using Emotions—which describes how
drawing on emotions can enhance cognition. Critics contend
that confirmatory factor models of the MSCEIT fit branches
1, 3 and 4 of the model reasonably well, but not Branch 2
(Joseph & Newman, 2010).

We agree that a mental ability factor of Facilitating Thought
has not reliably emerged from studies of the MSCEIT. This may
be a failure of the test construction, or because people solve
such problems using their ability at emotional understanding (or
another ability) rather than any reasoning distinctly related to
facilitating thought.

Although Facilitating Thought may fail to emerge as a dis-
crete mathematical factor on the MSCEIT, we believe it makes
sense to retain the branch in our four-branch model of problem
solving areas. Being able to draw on emotions to facilitate
thought is part of an overall emotional intelligence: Knowing

that sadness can promote the performance of detail-oriented
work—and that creativity burgeons with happiness—seems to
us integral to the construct (Isen, Daubman, & Nowicki, 1987),
and additional findings point to the idea that people use inner
emotional states to solve problems (Cohen & Andrade, 2004;
Leung et al., 2014). Moreover, tasks that involve Facilitating
Thought do correlate with scores of overall emotional intelli-
gence on the MSCEIT.

The four-branch model of emotional intelligence demar-
cates emotional problem-solving overall. We no longer expect,
however, that the specific mental abilities involved in emo-
tional intelligence will necessarily coincide with the specific
problem solving areas described by the four-branch model.

Update 3. Emotional Intelligence Is a Broad,
Hot Intelligence and Invites Comparisons With
Personal and Social Intelligences

In our early works we sometimes wrote that emotional intelli-
gence was similar to social intelligence (Mayer & Salovey,
1993; Salovey & Mayer, 1990) and at other times we described

table 1. The four-branch model of emotional intelligence, with added areas of reasoninga.

The Four Branches Types of Reasoning

4. managing emotions • Effectively manage others’ emotions to achieve a desired outcomeb

• Effectively manage one’s own emotions to achieve a desired outcomeb

• Evaluate strategies to maintain, reduce, or intensify an emotional responseb

• Monitor emotional reactions to determine their reasonableness
• Engage with emotions if they are helpful; disengage if not
• Stay open to pleasant and unpleasant feelings, as needed, and to the information they convey

3. Understanding emotions • Recognize cultural differences in the evaluation of emotionsc

• Understand how a person might feel in the future or under certain conditions (affective forecasting)c

• Recognize likely transitions among emotions such as from anger to satisfaction
• Understand complex and mixed emotions
• Differentiate between moods and emotionsc

• Appraise the situations that are likely to elicit emotionsc

• Determine the antecedents, meanings, and consequences of emotions
• Label emotions and recognize relations among them

2. Facilitating thought using
emotiond

• Select problems based on how one’s ongoing emotional state might facilitate cognition
• Leverage mood swings to generate different cognitive perspectives
• Prioritize thinking by directing attention according to present feeling
• Generate emotions as a means to relate to experiences of another personc

• Generate emotions as an aid to judgment and memory

1. Perceiving emotion • Identify deceptive or dishonest emotional expressionsb

• Discriminate accurate vs. inaccurate emotional expressionsb

• Understand how emotions are displayed depending on context and culturec

• Express emotions accurately when desired
• Perceive emotional content in the environment, visual arts, and musicb

• Perceive emotions in other people through their vocal cues, facial expression, language, and behaviorb

• Identify emotions in one’s own physical states, feelings, and thoughts

Note. aThe bullet-points are based on Mayer and Salovey (1997) except as indicated in superscripts b and c. Within a row, the bulleted items are ed approximately from
simplest to most complex, bottom to top. The four-branch model depicts the problem-solving areas of emotional intelligence and is not intended to correspond to the factor
structure of the area.
bAn ability from the original model was divided into two or more separate abilities.
cA new ability was added.
d Note that the Branch 2 abilities can be further divided into the areas of generating emotions to facilitate thought (the bottom two bulleted items) and tailoring thinking to
emotion (the top three bulleted items).

Mayer et al. The Ability Model of Emotional Intelligence 295

emotional intelligence as sui generis—it did not appear to be
like any other intelligence—surely nothing in the Cattell–Horn–
Carroll model as originally formulated. Neither of these posi-
tions appear helpful today.

Today, we believe there exists a group of hot intelligences of
which emotional intelligence is a member. Two other candi-
dates for this group are social intelligence and personal intelli-
gence (see Principle 7). Some of these intelligences are better
understood than others.

Social intelligence has been the most challenging to measure
(Conzelmann et al., 2013; Romney & Pyryt, 1999; Wong et al.,
1995). Work conducted early in the 20th century indicated that
social intelligence correlated so highly with general intelligence
as to be indistinguishable from it (Wyer & Srull, 1989). Recent
research bears this out: Conzelmann et al. (2013) found that
both social memory and social perception appeared to blend
into general intelligence, consistent with earlier studies. They
also found, however, more promising evidence for an independ-
ent social understanding task.

Another currently researched member of this group is per-
sonal intelligence: the capacity to reason about personalities—
both one’s own and the personalities of others. There …