Coping styles in animals: current status in behavior and stress-physiology (J.M. Koolhaas)

Coping styles in animals: current status in behavior and stress-physiology

J.M. Koolhaas

a,*, S.M. Korteb, S.F. De Boera, B.J. Van Der Vegta, C.G. Van Reenenb,

H. Hopster

b, I.C. De Jonga,b, M.A.W. Ruisb, H.J. Blokhuisb

a

Department of Animal Physiology, University of Groningen, P.O. Box 14, 9750 AA Haren, The Netherlands

b

DLO-Institute for Animal Science and Health (ID-DLO), Department of Behavior, Stress Physiology and Management, P.O. Box 65, 8200 AB Lelystad,

The Netherlands

Received 1 May 1999

Abstract

This paper summarizes the current views on coping styles as a useful concept in understanding individual adaptive capacity and

vulnerability to stress-related disease. Studies in feral populations indicate the existence of a proactive and a reactive coping style. These

coping styles seem to play a role in the population ecology of the species. Despite domestication, genetic selection and inbreeding, the same

coping styles can, to some extent, also be observed in laboratory and farm animals. Coping styles are characterized by consistent behavioral

and neuroendocrine characteristics, some of which seem to be causally linked to each other. Evidence is accumulating that the two coping

styles might explain a differential vulnerability to stress mediated disease due to the differential adaptive value of the two coping styles and

the accompanying neuroendocrine differentiation.

q1999 Elsevier Science Ltd. All rights reserved.

Keywords:

Coping; Aggression; Stress; Disease; Corticosterone

1. Introduction

Psychosocial factors have long been recognized as important

in health and disease both in man and in animals. It is

not the physical characteristics of a certain aversive stimulus

but rather the cognitive appraisal of that stimulus, which

determines its aversive character and whether a state

commonly described as stress is induced. The impact of

aversive stimuli or stressors is determined by the ability of

the organism to cope with the situation [1,2]. Several definitions

of coping can be given [3]. In the present paper, we

prefer to use the term coping as the behavioral and physiological

efforts to master the situation [3,4]. Successful

coping depends highly on the controllability and predictability

of the stressor [5,6]. A consistent finding across

species is that whenever environmental stressors are too

demanding and the individual cannot cope, its health is in

danger. For this reason, it is important to understand the

mechanisms and factors underlying the individual’s capacity

to cope with environmental challenges. A wide variety

of medical, psychological and animal studies demonstrate

that individuals may differ in their coping capacities.

Factors that have been shown to affect the individual’s

coping capacity include genotype, development, early

experience, social support, etc. Since many studies in

humans indicate that coping mechanisms are important in

health and disease [7], researchers have tried for a long time

to determine the individual vulnerability to stress-related

diseases using estimates of the individual coping capacity.

One approach concerns attempts to classify coping

responses into distinct coping styles. A coping style can

be defined as a coherent set of behavioral and physiological

stress responses which is consistent over time and which is

characteristic to a certain group of individuals. It seems that

coping styles have been shaped by evolution and form

general adaptive response patterns in reaction to everyday

challenges in the natural habitat. The concept of coping

styles has been used in a wide variety of animal species

(see Table 1). Despite the widespread use of the concept,

it is not without debate [8]. This is due to several flaws in the

studies using the concept. First, not all studies fulfill the

criterion of coping style as a coherent set of behavioral

and physiological characteristics because only one parameter

has been studied. Second, the extent to which clearly

distinct coping styles can be distinguished has been questioned

[8,9]. Special attention will be given here to the

frequency distribution of coping styles in a population, the

consistency over time and the one-dimensional character of

the concept of coping styles. Finally, one may wonder to

PERGAMON

Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

NEUROSCIENCE AND

BIOBEHAVIORAL

REVIEWS

NBR 376

0149-7634/99/$ - see front matter

q 1999 Elsevier Science Ltd. All rights reserved.

PII: S0149-7634(99)00026-3

www.elsevier.com/locate/neubiorev

* Corresponding author. Tel.:

131-50-3632338; fax: 131-50-3632331.

E-mail address:

koolhaas@biol.rug.nl (J.M. Koolhaas)

what extent the concept of coping styles is really related to

the individual vulnerability to stress-mediated disease.

This review will discuss these major issues and it will be

argued that the clustering of various behavioral characteristics

may to some extent be causally related to differences

in (re)activity of the neuroendocrine system.

2. Behavioral characteristics of coping styles

Much of our current thinking on coping styles is derived

from the work of Jim Henry [10]. He suggested, on the basis

of social stress research in animals and man, that two stress

response patterns may be distinguished. The first type, the

active response, was originally described by Cannon [11] as

the fight-flight response. Behaviorally, the active response is

characterized by territorial control and aggression. Engel

and Schmale [12] originally described the second type of

stress response as the conservation-withdrawal response.

This response pattern is characterized behaviorally by

immobility and low levels of aggression.

These ideas led to the hypothesis that the individual level

of aggressive behavior, i.e. the tendency to defend the home

territory, is related to the way individual males react to

environmental challenges in general. The hypothesis was

tested by Benus [13] using male house mice that were

genetically selected for either short attack latency (SAL)

or long attack latency (LAL). Also when other indices of

aggressive behavior are taken into account, the SAL males

are considered extremely aggressive whereas the LAL

males have very low levels of intermale aggressive behavior

[14]. The results of a series of experiments not only in mice,

but also in rats, suggest the existence of at least two coping

styles, which are summarized in Table 2. We prefer to use

the terms proactive coping rather than active coping and

reactive rather than passive coping (see further). Several

conclusions can be drawn from these results. First, the individual

level of aggressive behavior is indeed related to the

way in which the animals react to a wide variety of environmental

challenges. Second, it seems that aggressive males

have a more proactive type of behavioral response, whereas

non-aggressive or reactive males seem to be more adaptive

and flexible, responding only when absolutely necessary.

An important fundamental question is whether the two

types of behavior patterns can be considered to represent

styles of coping in the sense that they are both aimed at

successful environmental control [15]. Several experiments

indicate that the different behavior patterns can indeed be

considered as coping styles aimed at environmental control.

This is, for example, shown in a recent experiment using

wild-type rats. This strain of rats shows a large individual

variation in aggressive behavior similar to the variation in

wild house mice. After being tested for their tendency to

defend the home cage against an unfamiliar male conspecific,

the males were tested in a shock prod defensive burying

test. In this test, the animal is confronted with a small,

926

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

Table 1

Overview of the species in which a strong individual differentiation has been observed that may reflect coping styles. The plus signs give a rough indication of

the number of parameters on which the individual differentiation is based, i.e.

1 indicates a single parameter study, 1 1 indicates a multi-parameter study

Species Behavioral parameters Physiological parameters Reference

Mouse (

Mus musculus

domesticus

)

1 1 1 1

[13]

Rat (

Rattus norvegicus) 1 1 1 1 [77]

Pig (

Sus scrofa) 1 1 1 1 [20,37]

Tree shrew (

Tupaja belangeri) 1 1 1 1 [78]

Cattle (

Bos taurus) 1 11 [23]

Great tit (

Parus major) 1 1 [19]

Chicken (

Gallus domesticus) 1 11 [79]

Beech marten (

Martes foina) 1 [80]

Stickleback (

Gasterosteus

aculeatus

)

1 1

[24]

Rainbow trout (

Oncorhynchus

mykiss

)

1 1

[81]

Rhesus monkey (

Macaca

mulatta

)

1 1 1 1

[82]

Human (

Homo sapiens) 1 1 1 1 [26]

Octopus (

Octopus rubescens) 1 1 [25]

Table 2

Summary of the behavioral differences between proactive and reactive

male rats and mice

Behavioral characteristics

Proactive Reactive References

Attack latency Low High [14]

Active avoidance High Low [70,83]

Defensive burying High Low [84], this paper

Nest-building High Low [85]

Routine formation High Low [16]

Cue dependency Low High [17,84]

Conditioned immobility Low High [17]

Flexibility Low High [77]

electrified prod in its home cage. Because this prod is a

novel object, the experimental animal will explore it by

sniffing at the object. Consequently, the animal receives a

mild but aversive shock. As soon as it has experienced the

shock, the animal has two options to avoid further shocks. It

may either hide in a corner of the cage to avoid further

contact with the shock prod, or it may actively bury the

shock prod with the bedding material of the cage. Under

these free choice conditions, aggressive males spend most

of the test-time (10 min) burying (Fig. 1) while non-aggressive

males show immobility behavior. Notice, however, that

the two types of responding are equally successful in avoiding

further shocks. In this particular test, successful coping

can be defined operationally as avoidance of further shocks.

The terms active and passive coping are frequently used to

indicate the differences between the two styles. However,

these terms may lead to some confusion, because the terms

do not properly describe the fundamental differences. A

very fundamental difference seems to be the degree in

which behavior is guided by environmental stimuli

[16,17]. Aggressive males easily develop routines, i.e. a

rather intrinsically driven rigid type of behavior. Nonaggressive

males in contrast are more flexible and react to

environmental stimuli all the time. For that reason, we

prefer to use the terms proactive coping and reactive coping.

This differential degree of flexibility may explain why

aggressive males are more successful under stable colony

conditions, whereas non-aggressive males do better in a

variable or unpredictable environment, for example during

migration [18].

It is important to emphasize that the differentiation in

coping styles may not be expressed equally clearly in all

challenging situations. In particular, tests that measure

aspects of initiative or proactivity seem to be most discriminative.

This holds, for example, for latency measures

such as the attack latency test in males or the defensive

burying test, which allow the animal a choice between

proactive and reactive coping. Although female mice

usually do not show territorial aggression, females of the

short attack latency selection line show much more defensive

burying than female mice of the long attack latency

selection line. This supports our view that aggression is

only one of a larger set of behavioral characteristics that

make up the proactive coping style.

3. Distinct coping styles: distribution and consistency

over time

The concept of coping style and the way it is generally

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

927

Fig. 1. Correlation

.R . 0:72. between attack latency score in rats as

measured in the resident intruder paradigm and the percentage of time

spent burying in the defensive burying test of 10 min duration.

Fig. 2. Frequency distribution of attack latency scores (seconds) obtained in the 5th to the 12th generation of laboratory bred wild-type male rats

.N . 2500. in

different age classes (postnatal days).

presented in the literature suggests that there are distinct

phenotypes, which are more or less stable over time in

their response to stressors. The early studies by Oortmerssen

and colleagues, on a feral population of house mice suggest

a bimodal distribution of male phenotypes as measured by

the individual latency to attack a standard intruder into the

home cage [18]. The idea of bimodal distributions has been

strengthened by the fact that the phenotypical differences

appeared to have a rather strong genetic component. Genetic

selection for either of the extremes of the variation in a

certain behavioral or physiological characteristic generally

results in distinct genotypes within a few generations. Many

studies on coping styles and individual vulnerability to

stress mediated disease are based on the use of such genetic

selection lines. Selection lines have rather stable characteristics,

which are relatively insensitive to environmental

influences. However, there is confusion in the literature on

this issue, because several investigators using unselected

strains of animals were unable to find clearly distinct and

stable coping styles. The main problem seems to be the large

diversity in origin, age and gender of the experimental

animals involved. In the few studies that consider feral

populations, distinct phenotypes are found. Both in wild

house mice and in a small bird, the great tit (

Parus

major

), latency measures seem to have a bimodal distribution

[18,19]. However, one has to realize that the distribution

is not truly bimodal because latency measures are

generally finite, i.e. above a certain time the latency is set

to the maximum value. This leads to an accumulation of

individual scores in the distribution curve at this maximum

value. Nevertheless, the individual behavioral scores in wild

populations are certainly not normally distributed. An

analysis of a large database of aggressive behavior of a

population of 2500 laboratory-bred adult male wild rats

reveals an age-dependent change in the distribution of

attack latencies. Above a certain age, three peaks emerge

in the frequency distribution, with a clear intermediate

group (see Fig. 2). This difference with the wild situation

may be explained by the fact that there is little or no selection

pressure in the laboratory. It is tempting to consider the

possibility that intermediate animals are less successful in

nature. Few studies address the survival value of distinct

proactive and reactive coping styles. However, recent, yet

unpublished studies in feral populations of birds indicate

that the fitness of different coping styles depends on the

stability of the environment in terms of social structure

and food availability.

Many studies use laboratory strains of animals or heavily

domesticated farm animals like pigs. Usually, individual

behavioral scores are normally distributed in these studies.

For example, several studies in pigs show that the distribution

of individual scores in the back-test is normally distributed

[20,21]. Moreover, it is hard to tell how a certain inbred

or domesticated strain relates to the original and presumably

functional distribution of its wild ancestors. However, it is

intriguing that the extremes of this normal distribution still

fulfill the criteria for proactive and reactive coping styles,

both behaviorally and physiologically [20]. Although the

discussion on the shape of the distribution curve is important

from an evolutionary point of view, it does not seem to

matter much when individual vulnerability to stress-related

diseases is concerned. Afterall, it has been repeatedly shown

that the extremes in a population, irrespective of the detailed

distribution curve, may differ not only quantitatively, but

also qualitatively in their behavioral and physiological

response pattern to stress (see Table 1). Evidence has been

found in different species that the behavioral and physiological

response of individual animals to a specific stressor is

consistent over time. In pigs, for example, individual gilts

that displayed relatively long latency times to contact a

novel object and spent relatively little time near the object

during their first exposure showed a similar response when

re-tested one week later [22]. Also in dairy cows, consistency

was measured in behavior, in heart rate and in plasma

cortisol concentrations when individual animals were

repeatedly tested in a novel environment test over one

week. Moreover, consistent stress responses to the same

test were also found for cardiac and adrenocortical

responses over one year [23].

Another important issue concerns the one-dimensional

character of the concept of coping styles. Several studies

have used a factor analytical approach to reduce the sources

of individual variation in a population to a limited number

of components [24–26]. This statistical approach usually

reveals two or three factors that explain a considerable

part of the individual variation. Although some of these

factors relate to trait characteristics or aspects of personality

similar to coping styles, others may relate to state variables

such as stress and fear. These studies, both in humans and in

animals, emphasize the multidimensional character of individual

(personality) traits. However, aspects of aggression

such as hostility, impulsivity, anger or proactivity are often

found as an important dimension. In an experimental study

in Roman High and Roman Low avoidance lines of rats,

Steimer et al. [27] includes the dimension of emotional

reactivity as a second trait characteristic. By correlating

behavior of individual animals in a number of coping

style and emotion/anxiety related tests, he found evidence

for two independent dimensions, i.e. coping style and

emotional reactivity. Individual behavioral profiles calculated

either on indices of emotional reactivity or on indices

of exploratory activity as an indirect measure of coping style

resulted in different clustering of individuals. These two

dimensions together might explain individual vulnerability

to anxiety.

4. Neuroendocrine characteristics of coping styles

Differences in coping style have been observed in male

rodents during both social and non-social stressful conditions

(see Table 2). Coping styles are not only characterized

928

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

by differences in behavior but also by differences in physiology

and neuroendocrinology. As mentioned earlier, tests

that measure aspects of initiative or proactivity seem to be

most discriminative. The defensive burying tests in rodents

is such a test, which allows the animal a choice between

proactive and reactive coping. In general, defensive burying

is accompanied by high plasma noradrenaline and relatively

low plasma adrenaline and corticosterone, while freezing

behavior is associated with relatively low plasma noradrenaline

and high plasma corticosterone levels [28,29]. In a

strain of wild-type rats, the more aggressive males showed

the highest levels of burying behavior and showed a larger

catecholaminergic (both plasma noradrenaline and adrenaline)

reactivity after electrified prod exposure and after social

defeat than did the non-aggressive rats [30]. Previously, it

was shown that during social defeat the more competitive

proactive male rats reacted with higher responses of blood

pressure and catecholamines than the more reactive rats. In

addition, these competitive males had higher baseline levels

of noradrenaline [31]. The same holds for strain differences.

The aggressive Wild type-rats responded to social defeat

with larger sympathetic (plasma noradrenaline levels) reactivity

and concomitantly lower parasympathetic reactivity

(as measured by increased heart rate response and decreased

heart rate variability) than the less aggressive Wistar rats

[32]. Thus, proactive coping rodents show in response to

stressful stimulation a low HPA-axis reactivity (low plasma

corticosterone response), but high sympathetic reactivity

(high levels of catecholamines). In contrast, reactive coping

rodents show higher HPA axis reactivity and higher parasympathetic

reactivity (Table 3).

Differences in endocrine activity have also been observed

for HPA axis and gonadal axis activity under baseline conditions.

In aggressive mice, reduced circadian peak plasma

corticosterone levels have been observed as compared to

non-aggressive mice [33]. In mice of the short attack latency

selection line and in wild-type male rats, high baseline levels

of testosterone have been observed [34,35].

There is a growing body of evidence that similar coping

styles can be found in farm animals as well. Hessing and

colleagues showed that male castrated pigs could be characterized

as high resistant or low resistant at an early age

(1–2 weeks) by means of a back-test (manual restraint) [36].

In this back-test, a piglet is put on its back and the number of

bouts of resistance is used to characterize the animal. The

high-resistant pigs made more escape attempts and mean

heart rate frequency was higher than in low-resistant pigs

[36]. At three and at eight weeks of age, the high-resistant

ones were less inhibited in approaches to novel objects in an

open field. But the high-resistant pigs spent less time in

exploring the novel object than low-resistant pigs [36].

Heart rate frequency of high-resistant pigs was also substantially

increased in reaction to a falling novel object, while

heart rate frequency of low-resistant animals was only

slightly increased or even decreased (bradycardia), suggesting

that parasympathetic reactivity was higher in low-resistant

pigs [37]. Hessing and colleagues did not find clear

differences in HPA axis reactivity between the high- and

low-resistant animals, although basal plasma cortisol levels

were higher in low-resistant than in high-resistant pigs and

this was accompanied by adrenal hypertrophy. Recently,

however, Ruis et al. [20,21] showed clear differences in

HPA axis reactivity in high- and low-resistant female

pigs. The low-resistant animals had higher HPA axis reactivity

than the high-resistant ones. This was shown by higher

salivary cortisol responses to a novel environment test, to

routine weighing at 25 weeks of age, and to administration

of a high dose of ACTH [20]. Interestingly, the low-resistant

animals with high HPA axis reactivity at 24 weeks of age,

showed less aggression in group-feeding competition tests,

hesitated longer to leave their home pens and to contact a

human than did high-resistant animals. Altogether, pigs that

showed high resistance in the back-test and low HPA-axis

reactivity and high sympathetic reactivity in response to

stressful stimulation are thought to be representatives of

the proactive coping style. In contrast, pigs that showed

low resistance in the back-test and high HPA axis reactivity

and high parasympathetic reactivity are thought to be representatives

of the reactive coping style.

Recently, itwas shown that laying hens, fromtwo lines with

high or low propensity to feather peck, also show individual

differences in physiological and behavioral responses to stress

that are similar to the described coping styles. During manual

restraint (keeping the bird on its side by hand for 8 min), the

high feather pecking line showed more resistance and higher

mean heart rate frequency and lower parasympathetic reactivity

than the low feather pecking line [38,39]. HPA axis reactivity

(plasma corticosterone levels) was highest in the low

feather pecking line, while the sympathetic reactivity (plasma

noradrenaline levels) was the highest in the high feather pecking

line. These data suggest that chickens of the high feather

pecking line are representatives of the proactive coping style,

whereas birds of the low feather pecking line are representatives

of the reactive coping style.

5. Causal relationship between neuroendocrine and

behavioral characteristics of coping

One may wonder to what extent the behavioral and

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

929

Table 3

Summary of the physiological and neuroendocrine differences between

proactive and reactive animals

Physiological and neuroendocrine characteristics

Proactive Reactive References

HPA axis activity Low Normal [33,38,56,70,86]

HPA axis reactivity Low High [20,28,56,87]

Sympathetic reactivity High Low [38,77,88]

Parasympathetic reactivity Low High [37,39]

Testosterone activity High Low [34,35]

physiological characteristics are causally related. Of course,

it is highly unlikely that all differences in coping style can

be reduced to one single causal factor. However, evidence is

accumulating that a differential HPA axis reactivity may

explain some of the behavioral differences. In different

species, freezing behavior as part of the reactive coping

response can be observed in response to an inescapable

stressor or predator. In rats, a large number of studies

have shown that corticosteroids play a permissive role in

this fear-induced freezing behavior. Adrenalectomy (ADX)

impaired the duration of fear-induced freezing compared to

sham-ADX controls, suggesting the involvement of adrenal

hormones. This behavioral deficit in ADX animals could be

restored by the application of corticosterone [40]. In line

with these experiments, treatment with metyrapone, a corticosteroid

synthesis inhibitor, reduced fear-induced freezing

behavior, suggesting that corticosterone is a key hormone in

the expression of fear-induced immobility [41]. Since corticosterone

can bind to both the mineralocorticoid and glucocorticoid

receptor, further experiments were performed to

find out which specific receptor type was involved. Intracerebroventricularly

administered mineralocorticoid receptor

antagonist RU28318 reduced the fear-induced freezing

response, whereas the glucocorticoid receptor antagonist

RU38486 was without effect [42]. The modulation of freezing

via a mineralocorticoid receptor-dependent mechanism

did not come as a surprise. Limbic mineralocorticoid receptors

bind corticosterone with about 10 times higher affinity

than glucocorticoid receptors, and low circulating levels of

the corticosteroid hormone almost completely occupy

mineralocorticoid receptors [43,44]. The biological background

could be that a glucocorticoid receptor-dependent

mechanism would have been too slow because glucocorticoid

receptors are only occupied at much higher hormone

levels that are reached several minutes after the stressor. In

nature, the permissive steroid mineralocorticoid receptor

action makes an immediate freezing response possible

during a sudden appearance of a predator. There is a growing

body of evidence that corticosteroids also play a role in

fear-induced behavioral inhibition in farm animals. In

laying hens, it has been shown that the birds with the shortest

tonic immobility response have the lowest corticosterone

levels [45]. Further, chronic administration of corticosterone

moderately increased plasma levels of corticosterone

and prolonged the tonic immobility reaction in hens

suggesting a causal role for corticosteroids [46]. Also, in

dairy calves, a positive correlation was observed between

plasma cortisol levels and the latency to approach a novel

object (van Reenen, unpublished observation).

6. Coping styles and differences in disease vulnerability

The concept of coping styles implies that animals have a

differential way to adapt to various environmental conditions.

Negative health consequences might arise if an animal

cannot cope with the stressor or needs very demanding

coping efforts. Sustained over-activation of various

neuroendocrine systems may lead to specific types of

pathology. Hence, in view of the differential neuroendocrine

reactivity and neurobiological make-up, one may expect

different types of stress-pathology to develop under conditions

in which a particular coping style fails. Although there

are only a limited number of studies performed concerning

pathology in relation to the type of coping style adopted,

there are some indications that the two coping styles differ

in susceptibility to develop cardiovascular pathology, ulcer

formation, stereotypies and infectious disease.

6.1. Cardiovascular disease

Various studies emphasize the differences between the

two coping styles in autonomic balance. Because of the

role of the two branches of the autonomic nervous system

in cardiovascular control, one may expect in conditions of

over-activation of these systems, a differential vulnerability

for various types of cardiovascular pathology as well.

Indeed, a number of experiments found evidence that the

proactive coping animal is more vulnerable to develop

hypertension, atherosclerosis and tachyarhythmia due to

the high sympathetic reactivity [32,37,47–49]. However,

hypertension has never been observed after conditions of

uncontrollable stress. In social groups, hypertension generally

develops in dominant or subdominant males that have

difficulties to maintain their social position. Therefore, it

seems that these types of cardiovascular pathology only

develop under conditions of threat to control rather than

loss of control [15]. The reactive coping style seems to be

characterized by a shift in the autonomic balance towards a

higher parasympathetic tone and reactivity as can be

observed by a strong bradycardia response in reaction to a

sudden unpredicted stressor. Although there have been no

systematic studies of the cardiovascular consequences of

this characteristic, one may suggest that these types of

animals are more vulnerable to sudden cardiac death due

to bradyarhythmia.

6.2. Gastric ulceration and stereotopies

There are numerous studies to indicate that the controllability

of stressors is an extremely important factor in ulcer

formation. The development of ulcers is low when animals

are able to actively control or predict the stressor or divert

their attention away from the stressor. For example, if rats

can terminate the inescapable shock, or can chew wood

during inescapable shock [50], or can bite on a wooden

stick during cold restraint stress less, ulcers are observed

[51].

The classical studies of Weiss [50] showed that the development

of ulcers was high when the number of active

coping attempts was high in the absence of informational

feedback or with negative informational feedback present.

In the experimental animal that could actively control the

930

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

aversive shock by either pressing a lever during the warning

signal or during the shock itself, the total length of stomach

wall erosions was much smaller than in the yoked partner,

which received exactly the same number of shocks, but

could not control them. Also, when a feedback tone was

given after each correct avoidance–escape response, the

amount of gastric ulceration was further reduced. However,

when brief punishment shock was given to the avoidance–

escape and yoked animals whenever an avoidance–escape

was made, then the avoidance–escape group showed more

severe ulcer formation than the yoked partners. Further, in

the absence of informational feedback, a positive correlation

was observed between the number of active coping

attempts and the amount of gastric ulceration.

In line with these results is an observation in Roman high

avoidance (RHA) and Roman low avoidance (RLA) rats,

which can be considered to represent the proactive and

reactive coping style, respectively. It was shown that RHA

rats, after stress of food-deprivation for five days, had more

stomach lesions than RLA rats [52]. A negative correlation

between attack latency in the intruder test and gastric

ulceration induced by restraint-in-water stress [53], also

suggests that animals that prefer a proactive coping style

are more vulnerable to the formation of ulcers during

uncontrollable stress. In rat colonies, dominant animals

that are usually representatives of the proactive coping

style are reported to develop stomach wall erosions when

they have lost their leading position (social outcast) after

frequent attacks by other colony members [15].

Another example of a possible relationship between

behavioral coping characteristics and pathology has been

found in veal calves. It was shown that veal calves fed

only with milk developed tongue-playing as a stereotypy

[54]. However, not all calves did this with the same intensity.

Those calves that developed a lot of oral stereotypies

showed less stomach wall ulcers when slaughtered at 20

weeks of age. However, calves that did not develop

tongue-playing, all had such ulcers at the same slaughter

age [54]. Recently these results were confirmed in a larger

study involving 300 veal calves (van Reenen et al., in

preparation). Also in tethered breeding sows that were

housed individually, two separate groups could be distinguished:

some cows spent up to 80% of their active time in

this behavior while others hardly developed stereotypies.

Surprisingly, the sows that showed less initial resistance

in the back-test were the ones to develop high levels of

stereotypy later on [55]. Recently, it was shown that high

levels of stereotypies are associated with a reduced sympathetic

activation caused by the chronic stress of tethering as

was shown by a decrease in heart rate during bouts of stereotyped

behavior. In this view, stereotypies help the animal to

cope with the adverse situation of tethering [56].

There is increasing evidence that individual animals that

adopt the proactive or reactive coping style differ in

sensitivity of the dopaminergic system and consequently

they may differ in vulnerability to the development of

stereotypies. For instance, in mice, the dopamine receptor

agonist apomorphine produced a greater enhancement of

stereotyped behavior in proactive coping animals than in

reactive coping animals, suggesting that proactive coping

animals may be associated with a more sensitive dopaminergic

system [57]. Similar correlations were found in rat

lines previously selected for high and low expression of

stereotyped behavior (gnawing) in response to apomorphine.

The apomorphine-susceptible rats showed more

proactive coping behavior (fleeing), whereas the apomorphine-

unsusceptible rats showed more reactive behavior

(freezing) in reaction to an open-field [58]. A similar relation

between coping style and stereotypy has been demonstrated

in pigs. Individual proactive (high resistant) and

reactive (low resistant) coping pigs can be distinguished

in the back-test in which the reaction to manual restraint

is measured [59]. Recently it was shown that the high-resistant

pigs have a higher oral stereotypic response (snout

contact-fixation with floor) to apomorphine as compared

to low-resistant pigs [60]. Thus, also in pigs there is a relationship

between coping style, sensitivity of the dopaminergic

system and development of stereotypies. Moreover, it

has been shown that the dopaminergic-sensitivity factor, i.e.

the latency to initiate stereotypic gnawing induced by

apomorphine, also predicted ulcerogenic vulnerability [61].

The underlying mechanism of increased vulnerability of

proactive coping animals to develop stereotypies is not well

understood. Here it is hypothesized that altered HPA-axis

regulation plays a crucial role in the development of stereotypies.

In farm animals it has been suggested that stereotypies

are performed to lower the state of arousal and anxiety

and to lower corticosteroid levels; however, not all studies

show this correlation [62]. A possible explanation for the

conflicting data may be the differential effects of corticosteroid

hormones at the stage when the stereotypy starts to

develop and at the stage when a full-blown stereotypy

continues to exist. It is hypothesized that stress levels of

corticosteroids may enhance the acquisition and expression

of stereotypies, whereas an already developed stereotypy

may reduce corticosteroid levels. This is supported by the

following two examples in rodents. First, amphetamine activates

dopamine pathways and induces stereotyped behavior

(e.g. gnawing) that can be potentiated by high levels of

corticosterone [63]. This suggests that brain glucocorticoid

receptors are involved. Moreover, corticosteroids sensitize

the dopaminergic system, probably through binding to the

glucocorticoid receptors [64]. Second, dopamine-depleting

lesions of the caudate-putamen are associated with a reduction

in stereotyped behavior but an enhanced corticosterone

response [65]. Thus, glucocorticoids via glucocorticoid

receptors may play an important role in the sensitization

of the dopaminergic system. Interestingly, apomorphinesusceptible

rats do differ in glucocorticoid receptor and

mineralocorticoid receptor expression in different brain

nuclei and have higher (and more prolonged) plasma

ACTH and total plasma corticosterone responses than

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

931

apomorphine-unsusceptible rats [58]. Further studies are

needed to investigate whether these differences in corticosteroid

receptor expression are responsible for the differences

in sensitivity in the dopaminergic system and

whether this is the underlying mechanism which, under

conditions of severe stress, increases the vulnerability of

the proactive coping animal to develop stereotypies.

6.3. Immunological defense during coping or non-coping

Contemporary psychoneuroimmunology emphasizes

the role of the HPA axis and the sympathetic branch of

the autonomic nervous system in communication between

the brain and the immune system [66]. In view of the differential

reactivity of these two systems in the two coping

styles, one may expect to see differences in the immune

system as well. Indeed, several studies in rats and mice

demonstrate that individual differentiation in coping is an

important factor in stress and immunity. In the social stress

models in particular, the individual level of social activity

seems to be an important explanatory variable in some

studies [67,68]. Although these studies do not specifically

address the issue of coping styles, it is tempting to consider

the possibility that these socially active animals represent

the proactive coping style. Sandi et al. [69] specifically

addressed the question of the significance of individual

differentiation in emotional responsiveness to the differentiation

in immunology. They used the Roman-high (RHA)

and low-avoidance (RLA) rats that have been genetically

selected on the basis of their active avoidance behavior [70].

These selection lines have been shown to differ in a number

of behavioral and neuroendocrine stress responses in a

similar way as the proactive and reactive coping styles

mentioned above. It was shown that the NK cell activity

and the proliferation response of splenocytes to mitogenic

stimulation was lowest in the RLA males, a difference that

was even more pronounced after the stress of active shockavoidance

learning. Other evidence that emotionality may

interact with the immunological response has been found in

dairy cows. During endotoxin mastitis in cows that were

socially isolated, animals that were selected one year earlier

for a strong adrenocortical response to isolation, showed a

significantly larger reduction in peripheral blood lymphocyte

numbers than cows that were previously classified as

weak responders [71].

In a study of pigs, Hessing [72] demonstrated that aggressive,

resistant pigs had a higher in vivo and in vitro cell

mediated immune response to specific and non-specific antigens

than non-aggressive, non-resistant pigs. After stress,

the aggressive, resistant pigs showed the strongest immunosuppression.

This difference in immunological reactivity in

relation to coping style may explain the differential disease

susceptibility associated with social rank in group-housed

pigs after challenge with Aujeszky virus. These observations

in pigs are consistent with similar data obtained in

colony housed male rats [67]. Finally, a recent observation

shows that proactive coping male rats are more vulnerable

to the experimental induction of the autoimmune disease

EAE (experimental allergic encephalomyelitis), which is

considered to be an animal model for multiple sclerosis in

humans. This high vulnerability seems to be due to the high

sympathetic reactivity in the proactive coping males [73].

7. Concluding comments

The concept of coping style has been frequently used in

many studies and in an increasing number of species.

However, only a few studies have a sufficiently broad

approach to the individual behavioral and physiological

characteristics and their consistency over time to be conclusive

on the generality of the typology across species. Nevertheless,

the available literature makes it tempting to consider

the possibility that the distinctions between proactive and

reactive coping styles represent rather fundamental biological

trait characteristics that can be observed in many

species. Species or strains may differ in their degree of

differentiation depending on the strength of the selection

pressure in nature or in the laboratory (genetic selection,

domestication), but the extremes within a certain population

differ generally in the same behavioral and physiological

parameters and in the same direction. This may be partially

due to the possibility that some of the characteristics share a

common causal physiological basis. The few studies in feral

populations suggest that the individual differentiation in

coping style may be highly functional in population

dynamics. Phenotypes within one species seem to have a

differential fitness depending on the environmental conditions

such as population density, social stability, food availability,

etc. This idea is strengthened by the ecological

studies of Wilson and co-authors [74,75]. Although these

authors use the term shyness and boldness to indicate individual

differences within a population, they argue that this

differentiation represents adaptive individual differences in

resource use and response to risk.

Different coping styles are based on a differential use of

various physiological and neuroendocrine mechanisms.

The general impression is that these mechanisms vary in

the same direction consistently over species. However, the

degree of variation and the organizational level at which the

variation is expressed may differ. It is likely that genetic

selection will artificially exaggerate trait characteristics up

to a level, which may not normally be found in a natural

population. There are certainly more dimensions that may

account for the individual differentiation in behavior and

physiology. It will be a major challenge for behavioral

physiologists to refine the scales for individual differences

in order to improve their predictive power for health,

welfare and disease.

Little is currently known about the origin of coping

styles. The few genetic selection lines that have been sufficiently

characterized both behaviorally and physiologically

932

J.M. Koolhaas et al. / Neuroscience and Biobehavioral Reviews 23 (1999) 925–935

indicate a strong genetic basis. Some recent studies suggest

that perinatal factors might play a role as well. However, the

use of genetic selection lines may overestimate the role of

the genotype. Indeed, the fact that cross-fostering and

embryo transfer did not affect aggressive behavior in our

selection lines of mice indicates that these lines are devoid

of any perinatal plasticity [76]. Unfortunately, the large

number of recent studies on the influence of perinatal factors

in adult stress-reactivity rarely considers a sufficiently wide

spectrum of behavioral and physiological characteristics to

be conclusive on the effects on coping styles as a coherent

set of characteristics. The same holds for the influence of

adult (social) experiences. Clearly, stress at an adult age

may produce enduring changes in behavior and physiology.

Whether it changes coping styles as a trait characteristic is

virtually unknown. So far, we prefer to consider coping

styles as rather stable trait characteristics originating from

genetic factors in combination with epigenetic factors early

in life. Experiences in adult life may alter the state of the

animal for a long period of time as expressed in some behavioral

and physiological parameters, but they do not seem to

change the coping style as a whole. In contrast, coping style

is not a rigid characteristic that allows the individual only to

respond according to one specific coping style in all situations.

The absence of sawdust in the defensive burying test,

for example, also elicits freezing behavior in the proactive

animal. In this case, the environment was restrictive and

consequently the animal did not follow its preferred coping

style.

The available evidence so far confirms the idea that

coping style helps to determine individual vulnerability

for stress-related disease. First, the concept implies that

animals may be differentially adapted to different environmental

conditions. Second, the differences in physiological

reactivity make the two coping styles vulnerable to different

types of disease. Hence, in our view, psychopathology can

only be understood as a function of the individual coping

style and the environmental demands. Understanding this

complex relationship is of crucial importance in understanding

human and animal health and welfare.