Evolution, emotions, and emotional disorders.

doi:10.1037/A0013503

Evolution, Emotions, and Emotional Disorders

Randolph M. Nesse and Phoebe C. Ellsworth

University of Michigan

Emotions research is now routinely grounded in evolution,

but explicit evolutionary analyses of emotions remain rare.

This article considers the implications of natural selection

for several classic questions about emotions and emotional

disorders. Emotions are special modes of operation shaped

by natural selection. They adjust multiple response param-

eters in ways that have increased ﬁtness in adaptively

challenging situations that recurred over the course of

evolution. They are valenced because selection shapes spe-

cial processes for situations that have inﬂuenced ﬁtness in

the past. In situations that decrease ﬁtness, negative emo-

tions are useful and positive emotions are harmful. Selec-

tion has partially differentiated subtypes of emotions from

generic precursor states to deal with specialized situations.

This has resulted in untidy emotions that blur into each

other on dozens of dimensions, rendering the quest for

simple categorically distinct emotions futile. Selection has

shaped ﬂexible mechanisms that control the expression of

emotions on the basis of an individual’s appraisal of the

meaning of events for his or her ability to reach personal

goals. The prevalence of emotional disorders can be at-

tributed to several evolutionary factors.

Keywords: emotions, natural selection, evolution, adapta-

tion, appraisal

H

appenstance events can shift the subsequent his-

tory of life. If ancestors of the hippopotamus had

not browsed vegetation in ever-deeper water 50

million years ago, there would be no whales today (Gin-

gerich, Raza, Arif, Anwar, & Zhou, 1994). While such

events are rarely predictable, they can be reliably explained

by Charles Darwin’s and Alfred Russel Wallace’s great

idea, natural selection. Four-legged whale ancestors that

could swim better underwater had more offspring; over

thousands of generations, their descendants gradually be-

came superb aquatic athletes.

Happenstance occurs in intellectual evolution as well.

After ﬁnishing The Descent of Man, and Selection in Re-

lation to Sex (Darwin, 1871), Darwin realized that materi-

als he had long collected on emotions could be organized to

refute Charles Bell’s earlier claim that the elaborate mus-

culature of the human face was evidence of Divine design.

He quickly wrote The Expression of the Emotions in Man

and Animals (Darwin, 1872/1965), emphasizing the phy-

logenetic consistency of emotional expressions from ani-

mals to humans. The book is, as advertised, about expres-

sion, and it says little about the selective forces that

produced emotions, leaving a persisting anti-Darwinian

legacy for emotions research (Fridlund, 1992).

However, Darwin clearly recognized that evolution

shaped not only the physical characteristics of an organism

but also its mental processes and behavioral repertoires.

The knowledge that natural selection shaped the brain

mechanisms that mediate motivation and emotions offers a

solid foundation on which a modern theory of emotions is

being built.

Although current psychological theories of emotion

differ widely in many particulars, almost all now agree that

emotions are adaptive responses that arise from mecha-

nisms shaped by selection (Plutchik, 2003). It is now hard

to imagine that just four decades ago emotions were gen-

erally seen as products of learning unrelated to natural

selection. It took Ekman’s, Izard’s and Eibl-Eibesfeldt’s

studies of cross-cultural consistency in emotional expres-

sion to overthrow that view (Eibl-Eibesfeldt, 1983; Ekman

& Davidson, 1994; Izard, 1991). In retrospect, it is obvious

that learning cannot be the whole story and that emotions

would not exist unless they were useful. Evolution is not an

alternative to other theories of emotions; it is the common

foundation for all. Many of its contributions are so simple

that they are not always recognized. To highlight the con-

tinuing importance of Darwin’s theory of natural selection

for emotions, we consider its implications for several clas-

sic questions.

What Emotions Are

Deﬁnitions of emotions typically describe proximate as-

pects such as physiology, subjective experience, or facial

expression, often emphasizing one or another component

(Ekman & Davidson, 1994; Izard, 2007). An evolutionary

approach deﬁnes what emotions are in terms of how they

came to exist. Emotions are modes of functioning, shaped

by natural selection, that coordinate physiological, cogni-

tive, motivational, behavioral, and subjective responses in

patterns that increase the ability to meet the adaptive chal-

lenges of situations that have recurred over evolutionary

time (Nesse, 1990). They are adaptations that are useful

only in certain situations (Underwood, 1954). Like pain

and sweating, they remain latent until an evolved mecha-

nism detects cues associated with the situation in which

they are advantageous.

Randolph M. Nesse and Phoebe C. Ellsworth, Department of Psychology,

University of Michigan.

Correspondence concerning this article should be addressed to Ran-

dolph M. Nesse, Department of Psychology, University of Michigan, East

Hall, Room 3018, 530 Church Street, Ann Arbor, MI 48109-1043. E-mail:

nesse@umich.edu

129February–March 2009

●

American Psychologist

Vol. 64, No. 2, 129–139 DOI: 10.1037/a0013503

Unlike simpler adaptations, however, emotions are

not unimodal responses to speciﬁc situations, like sweating

in response to overheating. Instead, emotions adjust mul-

tiple component processes to create an organized response

to the adaptive challenges of a given situation. For instance,

appraisals that indicate a nearby predator arouse an emer-

gency response that adjusts and coordinates many aspects

of physiology and behavior. Physiology may inﬂuence

cognition, and cognition may inﬂuence feeling, which may

inﬂuence behavior and physiology in a complex, recursive

sequence.

Neuroscience and psychological investigations of

emotions focus almost exclusively on proximate questions

about (a) what a trait is like and how it works and (b) how

it develops over the course of an individual’s life. How-

ever, a proximate explanation is only half the story (Mayr,

1988; Tinbergen, 1963). The other half of a complete

explanation requires answers to evolutionary questions

about (c) how the trait developed over time in the history of

the species and (d) what evolutionary factors shaped the

trait. Taken together, these are Tinbergen’s four questions,

the undisputed foundation for all research in animal behav-

ior and behavioral ecology (Dewsbury, 1999; Tinbergen,

1963). Pursuing all four together will speed progress in

research on emotions (Alessi, 1992; Ketelaar & Clore,

1997; Nesse, 1999).

What Different Emotions Exist?

How many emotions exist, and what are they? This ques-

tion has been a source of enduring controversy (Ekman,

1992a; Ekman & Davidson, 1994; Oatley, Keltner, & Jen-

kins, 2006; Plutchik, 2003). Some theories postulate just

two basic states—positive and negative; others postulate a

small set of “basic” emotions; and still others argue for a

potentially inﬁnite number. All theorists agree, however,

that valence is a necessary quality of emotions: Emotions

are about pleasure and pain, approach and avoidance (Bar-

rett, 2006b; Ekman, 1992b; Rolls, 2005).

Many one-celled organisms can do only two things—

keep swimming in the same direction or tumble randomly

before setting off again. In combination with a 0.5-s mem-

ory, this allows movement toward food (Adler, 1975; Kosh-

land, 1980). The algorithm is simple: If the food concen-

tration is higher than it was a half second ago, move

forward; otherwise, tumble. The ability to detect danger,

such as excessive heat or acid, shaped the other primal

behavior— escape. Many bacteria can swim only at one

speed, but in most organisms, valence can also vary in

intensity.

Valence and intensity are essential features of almost

all theories of emotions (Smith & Ellsworth, 1985). Va-

lence is at the very root of emotion and motivated behavior

(Barrett, 2006b; Schlosberg, 1952; Wundt, 1897), deﬁning

an opposition that has been described as approach/avoid-

ance, positive/negative affect (Huppert & Whittington,

2003; Schlosberg, 1952; Tellegen, Watson, & Clark, 1999),

promotion/prevention (Higgins, 1997), and the behavioral

approach system (BAS)/behavioral inhibition system (BIS)

(Gray, 1987). This opposition has led many to propose

circumplex models that array various emotions on these

dimensions (see Figure 1 for an example).

Many theorists, however, believe that these two di-

mensions are insufﬁcient to describe the universe of emo-

tional experience (Fontaine, Scherer, Roesch, & Ellsworth,

2007). Darwin (1872/1965), and many before him (Sorabji,

Randolph M.

Nesse

Figure 1

A Circumplex Model of Affect

ACTIVATION

alert

tense

excited

nervous

elated

stressed

upset

happy

UNPLEASANT

PLEASANT

contented

sad

serene

de

p

ressed

relaxed

bored

calm

DEACTIVATION

Note. Adapted from “The Circumplex Model of Affect: An Integrative Ap-

proach to Affective Neuroscience, Cognitive Development, and Psychopathol-

ogy” by J. Posner, J. A. Russell, and B. S. Peterson, 2005, Development and

130 February–March 2009

●

American Psychologist

2000), described a small number of qualitatively distinct

emotions as innate, universal natural kinds. Categorical

theories of emotion remain prominent (Ekman, 1992a,

1992b). Different theorists have different lists of basic

emotions, but all include fear and anger, and most include

joy and sorrow. Some include additional emotions, such as

surprise (Plutchik, 2003), contempt (Ekman, 1992a; Izard,

1991; Tomkins, 2008), interest (Izard, 1991; Panksepp,

1998), shame and guilt (Izard, 1991; Tomkins, 2008), and

acceptance (Plutchik, 2003; Tomkins, 2008).

Modern evolutionary approaches explain speciﬁc

emotions as coordinated states that give ﬁtness advantages

in speciﬁc situations that recurred over evolutionary time

(Nesse, 1990; Tooby & Cosmides, 1990). These views

have changed from a strict modular conception to one that

increasingly emphasizes emotions as prototypes without

sharp boundaries; they share overlapping elicitors, func-

tions, and physiological and cognitive characteristics

(Nesse, 1998; Russell & Fehr, 1994). In contrast, Cosmides

and Tooby have argued that selection has shaped thousands

of discrete domain-speciﬁc mental modules to deal with

different situations (Cosmides & Tooby, 1994) and that

emotions are superordinate programs that coordinate the

modules (Cosmides & Tooby, 2000).

Some theorists reject the idea of categorically distinct

emotions, arguing for a multidimensional space with a

potentially inﬁnite number of emotions (Barrett, 2006a;

Frijda, 1994, 2006; Scherer, Schorr, & Johnstone, 2001;

Smith & Ellsworth, 1985). The space contains clusters of

common and closely related feelings (such as anger, indig-

nation, and annoyance), which sometimes overlap with

other feelings that might be classiﬁed as separate emotions

by a discrete emotions theorist (such as anger, aversion,

and contempt), and sparsely populated regions where there

are feelings familiar to only a few cultures or individuals.

This point of view is compatible with our own evolutionary

perspective.

Just as this evolutionary perspective rejects the idea of

sharply distinct basic emotions, it also rejects a sharp

distinction between emotions and moods (Beedie, Terry, &

Lane, 2005). It is useful to distinguish short-lived emotions

aroused by speciﬁc cues from moods that may last for days

or weeks without speciﬁc causes. Also, compared with

moods, emotions have more prominent facial and physio-

logical changes, and they may increase ﬁtness by some-

what different routes. They are similar, however, in that

both are special states aroused in the situations where they

have tended to increase ﬁtness.

The Origins of Different Emotions

Speciﬁc emotions partially differentiated from more primal

generic states because they improved ability to cope with

speciﬁc kinds of threats and opportunities (Ellsworth,

2007; Nesse, 2004). Figure 2 is a hypothetical phylogeny of

emotions. Note the lack of sharp differentiation among

Phoebe C.

Ellsworth

Figure 2

A Possible Phylogeny of Emotions

Note. Reprinted from “Natural Selection and the Elusiveness of Happiness” by

R. M. Nesse, 2004, Philosophical Transactions of the Royal Society of London

Publishing.

131February–March 2009

●

American Psychologist

emotions. Individuals better able to recognize and adapt to

the challenges of survival-relevant situations tended to live

longer and reproduce more than other individuals. The

mechanisms that determine when an emotion occurs

evolved in conjunction with the components of the emo-

tion. New emotional variations are useless unless they are

expressed in situations where they are advantageous. De-

tection of signiﬁcant situations is useless without the ability

to express the appropriate emotion.

Different emotions are not deﬁned by different func-

tions or mechanisms, or speciﬁc stimuli, or brain modules,

or even by particular points in dimensional space. To the

extent that there are different emotions, they correspond to

different situations that have recurred over the course of

evolutionary time. They consist of whatever changes

tended to increase ﬁtness in the relevant situation. If each

situation were sharply distinct, and if the adaptive re-

sponses were different, then selection would shape distinct

basic emotions to match each situation. However, there is

overlap in both the characteristics of situations and the

patterns of response that are adaptive responses, so emo-

tions do not have clear boundaries.

For example, confronting a snake and confronting a

bear are similar situations, as are the adaptive responses.

Finding a fruit tree or a ﬁeld of grain evokes positive

overlapping responses. An evolutionary view of their ori-

gins strongly suggests that emotions are not susceptible to

clear deﬁnitions or crisp taxonomies. The absence of a

designer and millions of years of tiny sequential changes

have shaped a mind that is not just complex but indescrib-

able by words and concepts simple enough to be satisfying.

The emotions are neither discrete entities nor points on a

few dimensions; they are overlapping point-clouds in an

N-dimensional space. It should be no surprise that observ-

ers in different cultures discern some similar patterns or

that they recognize and label constellations of points some-

what differently (Wierzbicka, 1999).

All tangible analogies are inadequate, but something

is required. Within the usual analogy of mind as computer,

the emotions are like software programs that adjust input,

output, memory, processing, and display to cope effec-

tively with a particular kind of task (Ekman, 1992b; Nesse,

1994). However, unlike software programs, emotions were

not designed for speciﬁc functions. They are closer to the

programmable states on an electronic keyboard that adjust

the pitch, volume, tone, instruments, background rhythm,

distortion, and much more to constellations appropriate for

playing rock, blues, classical, soul, tango, and overlapping

genres.

The conclusion is disquieting—the clear taxonomy of

emotions sought for so long by so many may not exist. No

precise description of emotions and their subtypes can be

accurate. Although frustrating, this conclusion can liberate

us from a fruitless quest so we can turn our attention to the

somewhat indistinct structure of emotions, their functions,

and the mechanisms that regulate when and how intensely

they are expressed.

The Functions of Emotions

Although emotions have sometimes been regarded as mal-

adaptive, most contemporary researchers assume that they

confer selective advantages (Oatley & Jenkins, 1996;

Plutchik, 2003). We have described emotions as special

processes that enhance ﬁtness in certain situations. Most

emotion researchers have been content to leave it at that

and devote their research to other more proximate ques-

tions. The few who have attempted to develop evolutionary

theories of emotion have generally taken a taxonomical,

functional approach, proposing that the differentiation of

emotions corresponds to the functions they serve. Positive

emotions motivate the organism to take advantage of en-

vironmental opportunities and to recognize when it has

succeeded in doing so. Negative emotions motivate the

organism to avoid misfortune by escaping, attacking, or

preventing harm or by repairing damage when it has al-

ready occurred.

Different emotions have sometimes been deﬁned by

their more speciﬁc functions: Fear motivates escape from

danger; anger motivates attack; joy motivates continuing

on the present course or, if the object has been attained,

ceasing to strive for it; disgust motivates avoidance, vom-

iting, and more metaphorical expulsion; interest motivates

exploration; lust motivates seduction and sexual inter-

course; sorrow motivates calling for help or giving up on

fruitless endeavors, and so on (Gross & Keltner, 1999;

Plutchik, 2003). It is worthwhile distinguishing beneﬁts

that come from communication, arousal, motivation, mem-

ory, and action intentions.

It is tempting to offer a speciﬁc function as an evolu-

tionary explanation for each emotion. However, just as the

components of emotions are best thought of as parts of one

complex pattern, the various functions of an emotion are

best understood in terms of how they together increase

ﬁtness. One emotion has many functions, and any given

function is served by many emotions. Different emotions

do not correspond to different speciﬁc functions; instead,

they correspond to the adaptive challenges encountered in

different situations.

Regulation of Emotion Elicitation

An evolutionary approach is sometimes thought to empha-

size “innate” responses to universal cues such as snakes,

smiles, and darkness and to imply that emotions are ﬁxed

action patterns rigidly elicited like reﬂexes in response to

ﬁxed cues. In fact, an evolutionary perspective explains

why the mechanisms that regulate emotion elicitation are

so ﬂexible and varied.

A looming image has been followed by harm often

enough to arouse an innate response of fear and ﬂight

(Schiff, Caviness, & Gibson, 1962). Rabbits without an

innate fear of foxes have an often-fatal anxiety disorder.

However, even fear of snakes is not innate in primates but

is only a cue especially conducive to fear conditioning

(Mineka, Keir, & Price, 1980). Classical conditioning of

emotions allows organisms to experience affect that

slightly anticipates an event. Fear two seconds before a

132 February–March 2009

●

American Psychologist

danger is far better than two seconds after. Accordingly,

fear can be conditioned more easily, and extinguished less

easily, to cues such as snakes and spiders (Ohman &

Mineka, 2001). The capacity for operant conditioning of-

fers advantages that are even more obvious. A tendency to

repeat whatever works is the most general behavioral ad-

aptation imaginable. Emotions aroused by reward-associ-

ated cues have obvious utility (Rolls, 2005). If no reward is

forthcoming, motivation declines and disengages goal pur-

suit (Klinger, 1975), a pattern that is important for under-

standing the utility of low mood.

Although conditioning adjusts emotions to situations

better than ﬁxed responses can, simple learning cannot

come close to the effectiveness of human cognition (Good-

son, 2002). Our cognitive capacities allow inference about

the future, providing a huge advantage. Internal represen-

tations of external objects combine with causal schemas to

create expectations about the future and about the likely

consequences of alternative courses of action.

These expectations have predictably powerful inﬂu-

ences on emotions. The capacity to anticipate the future

also makes it possible to conceive of a goal and pursue it

with ﬂexible strategies over many days or weeks. Most

human behavior involves goal pursuit, and speciﬁc kinds of

goal-relevant situations arise repeatedly (Diener & Fujita,

1995; Nesse, 1990, 2004; Oatley & Jenkins, 1996). Oppor-

tunities arouse desire and excitement. With steady progress

toward the goal, optimism and effort are worthwhile. Frus-

tration is useful to test the scope of an obstacle and ways to

overcome it (Oatley & Duncan, 1994). In situations where

progress is impossible, low mood disengages effort (Carver

& Scheier, 1990; Klinger, 1975). Failure causes disappoint-

ment, success causes pleasure. Individuals whose behavior

is adjusted by appropriate emotions in these situations have

a selective advantage. Figure 3 summarizes some emotions

that arise in the pursuit of goals. These are not discrete

categories of emotion; they are central tendencies, and

because situations overlap, so do emotions.

Of course, seeking a rabbit for dinner is different from

seeking admiration from one’s group or affection from a

possible mate. So, the kinds of emotions associated with

goal pursuit became specialized to deal with different goals

in different domains. For instance, signs that a sexual

partner is interested in someone else arouse jealousy, a wild

and inconsistent mixture of fear, anger, and desire to please

(Buss, 2000; Daly, Wilson, & Weghorst, 1982). We rec-

ognize and deﬁne jealousy because it is aroused by a

particular situation. Difﬁculty in agreeing whether it is a

distinct emotion or a combination of other emotions should

be no surprise. It is neither. It is a special process that tends

to work. Working, in this evolutionary sense, means that

jealousy tends to increase reproductive success, even

though it may harm an individual’s interests.

Emotions are often elicited in situations where they

are useless. This is an inevitable and adaptive outcome.

Consider a signal detection analysis of the costs and ben-

eﬁts of panic in a particular situation. If the cost of a false

alarm is low, for instance, 200 kcal and 10 minutes, and the

cost of not experiencing panic in the presence of a real

danger is high, say, 200,000 kcal of damage on average,

then a normal system will express many false alarms. In

this hypothetical case, the optimal system will express a

panic attack whenever a cue indicates a greater than 1 in

1,000 chance that a predator is present. So, 999 out of 1,000

responses will be false alarms that are perfectly normal and

useful in the long run. This “smoke detector principle” is

crucial for understanding apparently unnecessary anxiety

and depression (Nesse, 2005).

Figure 3

Emotions for Situations That Arise in Goal Pursuit

Domain

Before

Usual

progress

Fast

progress

Slow

progress

Success Failure

Physical D

D

esire Engagement Flow Frustration Pleasure Pain

Opportunity

Social E

E

xcitement Friendship Pride

Anger

Low mood

Happiness Sadness

Physical F

F

ear Coping Confidence Despair Relief Pain

Threat

Social A

A

nxiety

Defensive

arousal

Confidence Anger Pride

Shame

Embarrassment

133February–March 2009

●

American Psychologist

Evolution, emotions, and emotional disorders.

Figures

Citations

Compassion: An Evolutionary Analysis and Empirical Review

An integrative and functional framework for the study of animal emotion and mood

Pain and emotion: a biopsychosocial review of recent research.

Disgust: Evolved Function and Structure

Annual Research Review: Prenatal stress and the origins of psychopathology: an evolutionary perspective

References

Self-efficacy: toward a unifying theory of behavioral change.

The Evolution of Cooperation

Self-efficacy: Toward a unifying theory of behavioral change☆☆☆

The Descent of Man, and Selection in Relation to Sex

The Evolution of Cooperation

Related Papers (5)

Emotion and Adaptation

The role of positive emotions in positive psychology. The broaden-and-build theory of positive emotions.

What Good Are Positive Emotions

Development and validation of brief measures of positive and negative affect: The PANAS scales.

An argument for basic emotions