Motivation and Emotion, VoL 17, No.

3, 1993

Pain and Aggression: Some Findings and

Implications
Leonard Berkowitz 1
University of Wisconsin-Madison

Consistent with the cognitive-neoassociationistic conception of anger and

emotional aggression, a wide variety of studies with animal as well as human
subjects demonstrate that pain often gives rise to an inclination to hurt an
available target, and also, at the human level, that people in pain are apt to
be angry. However, and also in accord with the present formulation, these
"primitive" angry~aggressive reactions can be suppressed, intensified, or
modified by cognitive processes.

One of the widespread illusions in our society is that suffering is ennobling.

Perhaps because of the hope that there is compensation for the blows re-
ceived from life, many persons want to think that good will somehow come
out of the bad. And so, looking for the silver lining, we might tell ourselves
that people who have met with some tragic circumstance or who have un-
dergone a painful occurrence will at least become better human beings even
if they are not completely returned to their earlier happier condition. The
victims' troubles will lessen their arrogance and selfishness and teach them
to be more sympathetic to the others around them.
The thought is c o m f o r t i n g - but, unfortunately, actually has little em-
pirical justification. Few of us become more virtuous and kinder as a result
of our suffering. Quite to the contrary, a mounting body of evidence indi-
cates that people afflicted by a decidedly unpleasant occurrence are all too
apt to become hostile and aggressive.
Summarizing these findings, I have proposed a cognitive-neoassocia-
tionistic conception of anger and emotional aggression (Berkowitz, 1983,

1Address all correspondence to Leonard Berkowitz, University of Wisconsin-Madison,

Wisconsin.

277

0146-7239/93/0900-0277507.00/0© 1993 Plenum PublishingCorporation

278 Berkowitz

Unpleasant Event

Negative Affect

Aggression-Related Tendenciea Escape.Related Tendencies

(Aggression-a ssocia ted expres.~ive- (Escape- or avoidance-related
motor responses, physiological expressive-motor responses,
reactions, thoughts, and memories) physiological reactions, thoughts,
and memories)

Rudimentary Anger
1
Rudimentary Fear

More Elaborated, "Higher-Order" Thinking

fI'houghts dealing with attnburlons, anticipated
outcome of event, soda! rules about appropriate
emotion in situation, conceptions of the nature of
a given emotion, etc.)

Differentiated Feelings

Irritation, Armc ance, or Anger Fear

Fig. 1. The Cognitive-neoassociationistic conception of the development of

differentiated feelings following an average event.

1989, 1993a, 1993b) holding that strong unpleasant sensations give rise to
both flight and fight reactions. People experiencing intense negative affect
may seek to escape or avoid the aversive situation, but they presumably
also have anger-related feelings, thoughts, and memories. The model is
summarized in Fig. 1. Simply put, it contends that immediately after an
aversive event is encountered and before any control processes go into op-
eration, the resulting negative affect activates angry feelings, hostile
thoughts and memories, and aggressive motor reactions automatically and
with little thought and attention because of the associative ties linking these
various components of the "anger/aggression syndrome" (Berkowitz,
1993b). As a consequence, the afflicted organism theoretically is inclined
to strike at some target.
Although many different kinds of unpleasant feelings seem to have
this effect, according to a growing number of studies, the results also sug-
Pain and Aggression 279

gest that aggressive reactions are especially likely to occur when the nega-
tive affect is fairly intense, when suitable targets for aggression are avail-
able, and when response t e n d e n c i e s stronger than the aggressive
inclinations have not been evoked (Berkowitz, 1993a, p. 65). These aggres-
sion-facilitating conditions often exist when the person is experiencing
physical pain, and the present paper will therefore focus on how pain might
give rise to an instigation to aggression.
This emphasis on the aggressive effects of pain can be very informa-
tive. The research on this topic certainly can give us a better understanding
of emotional aggression. But more than this, it also has important impli-
cations for the analysis of emotions generally, and I will spell out some of
these implications at the end of this article.

ANIMAL RESEARCH

Animal experiments showing pain-evoked aggression date back at

least to 1939, although the best known investigations of this phenomenon
were carried out by Azrin, Hutchinson, Ulrich, and their associates (cf.
Azrin, Hutchinson, & Hake, 1966; Hutchinson, 1983; Ulrich, 1966; also cL
Berkowitz, 1982). Some of the principal findings obtained by these studies
are especially worth considering. Besides pointing to the continuity of ani-
mal life and the role of biological influences in some aspects of human
reactions to pain, these results highlight conceptual issues with which I will
be concerned throughout this paper.
Generally speaking, in these studies when two animals were cooped
up together in a small chamber and exposed to noxious stimuli (such as
physical blows, electric shocks, loud noises, or intense heat), they frequently
began to fight. There is suggestive evidence that aggression was most likely
to occur under intermediate levels of this noxious stimulation. Furthermore,
it is also important to note that, when the aversively generated aggression
arose, this was a specific, recognizably aggressive reaction and not merely
an intensification of whatever action tendency was ongoing at the time (cf.
Hutchinson, Renfrew, & Young, 1971). Since this aggressive reaction took
place with a fair degree of regularity, without any prior experience with
fighting, and tended to persist even in the absence of obvious rewards, Ul-
rich and Azrin (1962) at first termed it "reflexive aggression."
More recent research has demonstrated, however, that aggression is
not an inevitable response to painful stimulation (Azrin, Hutchinson, &
Hake, 1966; Berkowitz, 1982; Hutchinson, 1983; also of. Bandura, 1973).
Many afflicted animals prefer to avoid or escape from the noxious event
rather than attack someone. Nevertheless, the studies also indicate the ani-
280 Berkowitz

mals are more likely to assault an available target when they do not know
how to get away from the pain source. Can it be, as I have proposed else-
where (e.g., Berkowitz, 1990, 1993a, 1993b), that the painful occurrence
activates both flight and fight tendencies, although one of these inclinations
may mask the other urge on many occasions? Prior learning clearly can
influence the relative strengths of these tendencies. Those animals who
have previously learned that aggression is apt to be punished and/or who
have developed nonaggressive reactions to the noxious stimulation are rela-
tively unlikely to attack an appropriate target when they are hurt (cf. Berk-
owitz, 1982, pp. 261-262).
Animal research also provides important information about the goals
of the aversively stimulated aggression. Disputing the Azrin/Hutchinson/U1-
rich analysis, several investigators (e.g., Blanchard, Blanchard, & Takahishi,
1978) have argued that the pain-evoked aggression is defensive behavior
rather than an appetitive-like attack where there is a specific aim to injure
or even destroy some target. For these critics the pained animals are ba-
sically only trying to protect themselves (cf. Archer, 1990). Taking a related
position, other psychologists (e.g., Bandura, 1973; Potegal, 1979) have held
that the pain-evoked assault is an attempt to eliminate or lessen the noxious
stimulation and that this behavior is maintained by negative reinforcement.
These interpretations of aversivety stimulated aggression have been
countered by still other investigators. Some (Archer, 1990; Brain, 1981;
Hutchinson, 1983) have noted that (in both laboratory and naturalistic stud-
ies) clearly offensive as well as defensive aggression can be seen at times.
And similarly, according to other findings, pained animals can strive to in-
jure their target as well as to escape or avoid the noxious stimulation. Thus,
where Knutson, Fordyce, and Anderson (1980) demonstrated that pain-elic-
ited aggression can be reinforced by the cessation of the unpleasant event,
the same experiment also showed that pained animals were inclined to fight
even when they had not learned that their attacks could end the aversive
occurrence. Also pointing to appetitive aggression, Azrin, Hutchinson, and
McLaughtin (1965) reported that the afflicted animals may even exert effort
(e.g., pull a chain) in order to obtain a suitable target for their aggression.
I will suggest below that suffering humans may also exhibit this kind of
appetitive aggression; they too may be inclined to strike at an appropriate
target, at least partly for the sake of doing harm.
At any rate, despite the seeming exceptions that have occasionally
been reported, Moyer (1976) concluded in his review of the pertinent lit-
erature that "there can be no doubt that under certain circumstances pain
can lead to an intense attack. Such behavior has been demonstrated in the
monkey . . . the cat . . . the rat . . .and the gerbil . . ." (p. 200). Moyer
also observed that "The most effective stimulus for eliciting this kind of
Pain and Aggression 281

aggression is a sharp, sudden pain . . . . Other types of aversive stimulation,

such as cold, loud noises, and heat, are much less effective in producing
aggression" (p. 201). This last comment does not mean, it should be noted,
that cold, noise, heat, and other decidedly unpleasant conditions do not
activate hostility and aggressive inclinations at all. These emotional reac-
tions may be relatively weak at times, but they can still occur (cf. Anderson,
1989; Berkowitz, 1982, 1983).

SOME RESEARCH WITH HUMANS

Observations of People

Turning from animals to humans, it is by no means unusual to find

reports of an association between pain and anger in medical and psycho-
logical studies of people suffering from intense pain (cf. Sternbach, 1978).
Many different kinds of pain patients, such as, for example, people tor-
mented by severe episodic headaches (Hatch et al., 1992; Mathew, 1990),
or by a lasting injury to their spinal cords (Summers, Rapoff, Varghese,
Porter, & Palmer, 1991), are often described as relatively angry and/or hos-
tile. But of course, these observations in themselves are difficult to inter-
pret. The causal direction is unclear in many of these instances; we cannot
say whether the pain produced or was exacerbated by the anger, and in-
deed, many authorities maintain that anger contributes to the pain expe-
rience (cf. Sternbach, 1978; Wade, Price, Hamer, Schwartz, & Hart, 1990).
Other factors that might possibly account for the pain-anger association
obviously are also uncontrolled.
It is therefore necessary to look to controlled laboratory experiments
in order to determine whether painful events can give rise to anger, hostility,
and emotional aggression. The suffering established in these investigations
obviously is not especially severe or prolonged, and typically is also far more
controllable and more clearly due to some external cause than most cases
of naturally occurring pain. Nevertheless, if the relatively weak pain levels
employed in this laboratory research can evoke anger and aggression, we
can have greater confidence that whatever anger and aggression are dis-
played by people in agony are at least partly the result of their suffering,

Experiments on the Aggressive Consequences of Pain

From my perspective the pain-activated aggressive reaction is emo-

tional behavior in that it is generated by strong affect and is also often
282 Berkowitz

i,nol
aggressiveness
disposition
.......

Fig. 2. Influences affecting the strength of impulsive aggression.

impulsive in nature since the response frequently is stronger than was con-
sciously intended. Figure 2 lists a number of factors that have already been
identified as affecting the intensity of impulsive aggression (cf. Berkowitz,
1974, 1993b). The present paper will consider only a few of these possible
influences.
Stimuli Associated with Pain Source. As Fig. 2 indicates, classical con-
ditioning can operate in pain-related aggression by humans just as it can
influence animal fighting. Much as the mere presentation of stimuli that
had been repeatedly paired with the occurrence of painful events can pro-
voke animals to attack each other (e.g., Hutchinson et al., 1971; Vernon
& Ulrich, 1966), according to Fraczek and his associates in Poland (cited
in Berkowitz, 1982, pp. 273-274), the sight of situational details that had
been associated with an earlier painful incident can intensify the strength
of an assault upon an available target. Some of the university students in
one of Fraczek's experiments learned to associate the color green with the
receipt of electric shocks. Soon afterwards, when all of the participants were
asked to punish someone else, half of these people found that the
"weapon" apparatus they were to employ in delivering the shocks was
painted green, whereas the remaining subjects found that the apparatus
Pain and Aggression 283

was a different color. The most punitive people were those whose "weapon"
had the color-mediated connection with their earlier pain.
What Is the Aim of the Pain-Elicited Aggression? Just as painful events
apparently evoke appetitive aggression in animals, studies with humans in-
dicate that the afflicted persons often want to do more than merely lessen
their suffering; they frequently are also motivated to hurt someone, and
may assault an available target even when that individual is not responsible
for their pain. In Feshbach's (1964) terms, they are prompted to engage
in hostile aggression, where their primary aim is to inflict injury, and do
not only want to carry out instrumental aggression.
An experiment conducted by Berkowitz and Embree (1987) provided
indirect evidence for this contention by suggesting that people can be ag-
gressively inclined as long as they are suffering even though they realize it
is possible to end their pain soon. All of the mate subjects in this study
were required to keep one of their hands in a tank of water that was either
painfully cold or at room temperature as they supervised a fellow student's
work on an assigned task. Half of the participants were given permission
to take their hand from the water whenever they wanted so that they could
easily escape from the painful situation, whereas pressure was placed on
the other subjects not to take their hand from the water. Even though the
permission to withdraw from the painful stimulation had apparently less-
ened the irritability and annoyance felt by the men in the cold-water con-
dition, according to questionnaire ratings made at the end of the session,
these people were the most punitive to their partners as they supervised
their partners' work. Still feeling pain, they evidently were still aggressively
inclined.
Much better, and more direct, support for the notion of a pain-acti-
vated desire to hurt can be found in two experiments by Berkowitz, Co-
chran, and Embree (1981) making use of the same cold-stressor procedure
that had been employed in the previously mentioned study. As in that in-
vestigation, each female subject was required to keep one hand in water
that was either very cold or at a more comfortable room temperature as
she evaluated another woman's solutions to a series of business problems.
Cross-cutting the water temperature variation, half of the people were in-
formed that any punishment they gave the other woman would help that
person (by motivating her to do better), whereas the others were led to
think the punishment would be hurtful (in that it was likely to disrupt the
woman's problem solving).
Once these conditions were established, the participants heard their
partners advance their problem solutions (a fixed set of answers used for
all subjects), and the subjects delivered their "evaluations" ranging any-
where from five rewards (each a five-cent coin) to five punishments (each
284 Berkowitz

20

10

[ ] No. of
r, punishments

[ ] No. of
rewards
Physicallyvery unpleasant Physicallylessunpleasant
Punishment Punishment Punishment Punishment
hurts helps hurts helps
Experimental conditions

Fig. 3. Total number of rewards and punishments administered in cold-pressor situation as a

function of water temperature and supposed effect of punishment. (Data from Berkowitz et
al., 1981, Expt. 1.)

an unpleasant blast of noise). Figure 3 shows the mean number of rewards

and punishments that were administered in each of the four experimental
conditions over all "trials" of the first experiment in the series.
It is clear, first of all, that the participants were generally unwilling
to punish their partners, particularly when they were reminded that the
punishment would be harmful. As can be seen in the figure, those exposed
to the less unpleasant water temperature and told that punishment would
hurt their partners typically delivered the most rewards and fewest punish-
ments. This reluctance to do harm apparently was much weaker for the
suffering subjects exposed to the painfully cold water, but particularly when
they believed they could hurt their targets. And so, presumably because
they wanted to inflict some injury, the pained subjects thinking their pun-
ishment would be harmful administered the fewest rewards and most pun-
ishments.
Angry Persons Are Also Responsive to Signs of Pain. If we can assume
that deliberately provoked people are also in pain (at least to some degree),
still other studies have demonstrated that the afflicted (i.e., angered) per-
sons are motivated to hurt (cf. Berkowitz, 1982, pp. 280-281). In one of
the clearest of these experiments, Baron (1977) reported that angered sub-
jects' assaults on their tormentors rose in intensity when they found that
their initial attacks were hurting those individuals. By contrast, the indica-
Pain and Aggression 285

tions of the victims' suffering served to reduce the intensity of the punish-
ment delivered by nonprovoked subjects. It is very much as if the informa-
tion about the victims' pain had signaled to the angered participants that
they were fulfilling their aim of hurting their tormentors, and thus inten-
sified their "appetitive" attacks.
In this last-mentioned case the pain cue had acquired its aggression-
heightening value by being associated with the provocateur's agony. Ac-
cording to still other research, even the sight of someone other than the
immediate tormentor suffering from a beating can stimulate an angry per-
son to be more aggressive than he or she otherwise would be. When the
deliberately insulted subjects in an experiment by Hartmann (1969) were
shown a brief movie depicting two strangers fighting, they subsequently
were more aggressive to the individuals who had provoked them if the
movie they had seen had focused on the fight victim's pained reactions
rather than on the fight winner's attacks.
Even seemingly neutral stimuli having no obvious painful meaning
can have this kind of effect as a result of having previously been associated
with a tormentor's injury. An experiment by Swart and Berkowitz (1976)
is illustrative. Subjects who had been provoked by one of the experimenter's
accomplices then watched this person receive electric shocks as a certain
light flashed. When they later were given an opportunity to punish someone
else in response to various light signals, they assaulted this other individual
more intensely when the signal light was the same as the one that previously
had been paired with the earlier injury to the provocateur.

Some Effects Of Cognitions

The findings reported so far indicate that painful occurrences can give
rise to aggressive reactions independently of the appraisals and attributions
emphasized by traditional cognitive theorizing. This does not mean, of
course, that cognitions never have a major role in pain-elicited aggression.
Students of human pain (e.g., Melzack, 1973) have frequently noted how
the pain-producing sensory input can be modified by thoughts, interpreta-
tions, and prior experience before it activates motivational systems.
Attributions and Perceived Control. Attribution research has repeatedly
demonstrated that people's beliefs as to what caused a noxious event can
greatly influence the experienced unpleasantness of this stimulation. Sum-
marizing many of the findings obtained in this research, writers such as
Ferguson and Rule (1983) and Weiner (1985), among others, have pro-
posed that the victims are especially apt to become aggressive when they
think some external agent had controlled the noxious occurrence, inten-
286 Berkowitz

tionally and wrongfully seeking to harm them. I have little doubt that
thoughts of this kind can drastically affect emotional reactions to painful
events. For me, however, and contrary to the more extreme forms of tra-
ditional cognitive theorizing, these attributions (1) are not necessary for
anger and emotional aggression to arise, and (2) heighten the likelihood
of anger and aggression because they intensify the negative affect that is
experienced (Berkowitz, 1989, 1993a, 1993b). Simply put, a noxious event
is much more unpleasant when we believe someone has deliberately and
malevolently sought to make us suffer.
Research also suggests that the aggression-enhancing effect of noxious
stimulation can be lessened by the perception of control over this stimu-
lation. In Geen's experiments on this topic (cf. Geen, 1990, pp. 67-69), as
an example, subjects were exposed to unpleasant noise at the time that
they were supposed to punish another person who had provoked them ear-
lier. The people who had been told earlier that they could terminate the
aversive noise whenever they desired were less aroused physiologically and
also less punitive than their counterparts who did not have this perceived
control over the disturbing noise or who were only able to predict its offset.
The subjects' belief that they could end the noxious occurrence apparently
had lessened its experienced unpleasantness, and this led to a relatively
low level of aversion-instigated aggression.
Coping with the Pain Source. More than just thinking they can lessen
the pain they experience, sometimes people can do things to reduce the
hurt they feel, even when the pain source is within them, so that they then
are less angry and emotionally aggressive. Consider women in childbirth.
According to a field experiment conducted by Leventhal, Leventhal,
Shacham, and Easterling (1989), the women in this study who were about
to give birth generally reported not only more pain but also more anger
and fear as their labor progressed and their contractions became more in-
tense. More pertinent to my present point, those participants in the ex-
periment who had been taught to monitor their contractions and had
attended LaMaze classes indicated lower levels of anger as well as pain
than their counterparts in the control condition. The former apparently
had coped more effectively with the relatively painful situation, thereby
lessening their felt distress and their concomitant anger.
Pain Expectations. Thoughts other than those dealing with the cause
of the aversive stimulation and/or its control can also affect the person's
emotional reactions to noxious events. Highlighting one of these additional
possibilities, Leventhal, Brown, Shacham, and Engquist (1979) showed that
the experienced unpleasantness of aversive stimulation is increased when
people expect to suffer from this stimulation. Berkowitz and Thome (1987)
followed this lead and demonstrated that the heightened displeasure pro-
Pain and Aggression 287

duced by pain expectations can also give rise to fairly strong attacks upon
an available target.
In this experiment, following the standard procedure described ear-
lier, the subjects (undergraduate women) were asked to keep one hand in
a tank of water, supposedly as part of an investigation of harsh environ-
mental conditions on supervision. Two-thirds of the participants were to
find that the water was quite cold (7°C) whereas the other third were to
find the water was at a less unpleasant room temperature (23°C). But just
before they immersed their hand in the water they were given preparatory
information. One of the cold-water groups and all of the people about to
be exposed to the warmer water were warned that they might feel pain.
Subjects about to undergo the other cold-water condition were told only
that they might experience tingling sensations in their hand and the word
pain was not used.
Immediately after getting this information, each participant was told
to place her hand in the water, and as in our earlier studies, she was then
required to evaluate another student's solutions to a series of business-re-
lated problems by administering rewards and punishments to her. At the
end of the 6-min-long session the subjects rated how they had experienced
the water temperature and what their feelings had been while their hand
was in the water.

25

~22~aPunlshmnts

!E
~F~gReuards

~RIR Index

l~Istress

Cold Pain Warm-Pain

Cold-Sensatlon

Fig. 4. Reactions to cold-pressor situation as a function of expectation. AIA

= annoyance, irritation, anger.
288 Berkowitz

The latter ratings confirmed the Leventhal group's earlier findings:

As can be seen in Fig. 4, the women exposed to the cold water and who
had expected to feel pain reported experiencing the strongest distress. Go-
ing beyond this, however, and in accord with the present cognitive-neoas-
sociationistic analysis, these particular subjects also indicated the greatest
level of felt anger on our AIA index composed of their ratings of "annoy-
ance," "irritation," and "anger." But most important of all, as Fig. 4 also
shows, a relatively high level of aggression accompanied this felt anger.
The women in the cold-water-pain expectation condition delivered signifi-
cantly fewer rewards and more punishments to their partners than did the
people in the other groups. This occurred, it should be noted, even though
the subjects had not been antagonized in any way by their partners and
the punishment they administered was no more effective than giving re-
wards in allowing them to escape quickly from the water.
Some Seeming Exceptions: The Self-Regulation of Emotional Reactions.
Despite all this evidence that painful stimulation frequently produces ag-
gressive reactions, it is clear that suffering persons do not always act ag-
gressively and, indeed, are frequently depressed rather than assaultive. The
present analysis can accommodate these seeming exceptions fairly readily.
Besides raising the obvious possibility that the pain-elicited instigation to
aggression can be inhibited, it also recognizes that the noxious event often
evokes stronger flight than fight tendencies; humans as well as members
of many animal species often prefer to escape from unpleasant situations
rather than attack available targets. A host of factors--genetic, learned,
and situational--can determine the relative strengths of these opposing
tendencies (cf. Berkowitz, 1982, 1983).
Furthermore, as Fig. 1 implies, the cognitive-neoassociationistic model
also holds that thought processes brought into operation after the initial
"primitive" emotional reactions can lead to nonangry feelings and nonag-
gressive behavior (Berkowitz, 1989, 1993a, 1993b). Repeating my earlier
summary of this formulation, the negative affect produced by an aversive
event presumably activates angry feelings, hostile thoughts and memories,
and aggressive urges automatically and with little thought and attention be-
cause of the associative ties linking these various components of the "an-
ger/aggression syndrome" (Berkowitz, 1993b). But, the model also says,
"higher-level" control mechanisms can then intervene to regulate w modify,
intensify, or suppress m the automatic responses, depending upon the avail-
able information and the perceived requirements of the situation.
The leading analyses of how controlled processing can regulate auto-
matic, associative response tendencies (e.g., Schneider, Dumais, & Shiffrin,
1984) generally contend that control mechanisms are activated when at-
tention is directed toward the ongoing automatic reactions. Consistent with
Pain and Aggression 289

these formulations, Berkowitz and Troccoli (1990) proposed that people

are apt to exert some control over the aggressive inclinations that have
been automatically elicited in them by negative affect when they become
aware of their unpleasant feelings; as long as the negative affect and the
resulting aggressive impulses are not too intense, the attention given to
these feelings presumably sets controlled processing into operation so that
the afflicted persons tend to act in accord with all of the information avail-
able to them at the time.
Berkowitz and Troccoli (1990) have now carried out four experiments
supporting this argument, two of them reported in their initial article on
this topic (Berkowitz & Troccoli, 1990), and another summarized elsewhere
(Berkowitz, 1993a, pp. 357-358). Their fourth study is especially pertinent
to the present paper since it deals with the self-regulation of hostile ten-
dencies evoked by the cold-pressor situation.
As in the other cold-pressor experiments reviewed here, the male sub-
jects immersed their hands in a tank of water that was either very cold or
at a more comfortable room temperature, but this time they listened to a
brief audio-tape-recorded autobiographical statement that had supposedly
been made by a job applicant. With their hands still in the water, half of
the men in each of the two water temperature conditions were then asked
to rate and think about their feelings at the time (the self-focus group),
whereas the other participants' attention was diverted from themselves by
asking them to describe the average student on the campus (other focus
group). A minute later and with their hands still immersed in the tank, all
of the subjects were then required to indicate their impressions of the job
applicant's personality on an adjective checklist. We established a measure
of the participants' hostility from this checklist by counting the number of
socially disapproved personal qualities that were attributed to the job ap-
plicant.
Figure 5 summarizes the significant interaction that was obtained on
this hostility measure. As can be seen, the cold water led to stronger ex-
pressed hostility than the warmer water only when the men's attention had
been directed away from themselves. By contrast, under self-directed at-
tention the people exposed to the painfully cold water were actually less
hostile to the job applicant than were those in the more comfortable water
temperature condition, as if the former were "leaning over backwards" try-
ing to keep their feelings from affecting their judgments. (Much the same
pattern has also been obtained in several of our other studies on this topic.)
All in all and considering all of our experiments employing a variety of
pain-inducing procedures and hostility measures, there is good reason to
believe that painful events are most likely to produce open hostility and
290 Berkowitz

H o s t ~ l l t ~ level
20

I'.i

10

~ 2 Le_~is cold
o#ter

1 V e r w co]o
u~ter
Self focus Other focus

Condl t runs

Fig. $. Number of bad traits attributed to partner in cold-pressor situation as a

function of attention direction.

aggression when these presumably automatic reactions are not controlled

by self-directed attention-activated processes.

SOME IMPLICATIONS FOR EMOTION THEORIZING

In my view, the studies reported here, as well as the other relevant

investigations that could have been cited (cf. Anderson, 1989; Berkowitz,
1983, 1989, 1993a, 1993b), have important implications for theorizing about
emotions generally. As I have argued elsewhere (Berkowitz, 1993b), any
truly comprehensive analysis in this area must recognize that emotional re-
actions can occur without the relatively complicated appraisals and attri-
butions postulated by the traditional cognitive formulations. "Bottom-up"
processes initiated by psychologically relatively simple sensations can give
rise to emotional feelings and actions independently of considerations of
what caused these sensations and/or what their personal significance might
be. Humans share many of the qualities possessed by members of other
animal species and physiological mechanisms and associative processes can
affect them at times much as they influence other animals.
This is not to say, of course, that the "higher-level" cognitive functions
have only a trivial role in emotional phenomena or that "top-down" proc-
esses can safely be ignored. Elaborate cognitions can do a great deal to
determine what is felt and what actions are taken. Emotion theories must
Pain and Aggression 291

recognize the contributions made by all of these levels and processes, the
"higher-order" ones and those of a biologically more primitive nature.
One final comment is warranted here. Many emotion theorists, espe-
cially those with a traditional cognitive orientation, are all too likely to
reify linguistic analyses so that they regard whatever linguistic differentia-
tions they can establish as being real and significant. Thus, since they can
distinguish verbally between a "mood" and an "emotion," they frequently
insist these concepts are actually different phenomena. These theorists
might thus maintain that the research summarized in this paper involves
moods (or perhaps feelings) but not emotions. And furthermore, since they
(and many laypersons) can think of differences between annoyance, irrita-
tion, and anger, they also hold that these actually are different emotional
states. Following their linguistic analysis, then, these writers might then con-
tend that the present results have to do with annoyance or irritation but
not anger.
The empirical research I and other social psychologists have con-
ducted on the anger/aggressive consequences of aversive events argue
against these arbitrary distinctions. They have repeatedly shown that ratings
of felt annoyance, irritation, and anger are highly intercorrelated, much as
if the feeling states reflected by these ratings have a good deal in common
(although the feelings might perhaps differ in intensity), and furthermore,
that these measures generally have considerable construct validity. The
findings also demonstrate how reasonable it is to say the subjects had emo-
tional reactions to the situations confronting them. In general, perhaps be-
cause of their metatheoretical predilections, traditional cognitive theorists
in the emotion realm may have relied much too greatly on linguistic as
well as on cognitive analyses.

REFERENCES

Anderson, C. A. (1989). Temperature and aggression: Ubiquitous effects of heat on

occurrence of human violence. Psychological Bulletin, 106, 74-96.
Archer, J. (1990). Pain-induced aggression: An ethological perspective. Current Psychology:
Research and Rev&ws, 8, 298-306.
Azrin, N. H., Hutchinson, R. R., & Hake, D. F. (1966). Extinction-induced aggression. Journal
of the Experimental Analysis of Behavior, 9, 191-204.
Azrin, N. H., Hutchinson, R. R., & McLaughlin, R. (1965). The opportunity for aggression
as an operant reinforcer during aversive stimulation. Journal of the Experimental Analysis
of Behavior, 8, 171-180.
Bandura, A. (1973). Aggression: A social learnh~ganalysis. Englewood Cliffs, NJ: Prentice-Hall.
Baron, R. A. (1977). Human aggression. New York: Plenum Press.
Berkowitz, L. (1974). Some determinants of impulsive aggression: Role of mediated
associations with reinforcements for aggression. Psychological Review, 81, 165-176.
292 Berkowitz

Berkowitz, L. (1982). Aversive conditions as stimuli to aggression. In L. Berkowitz (Ed.),

Advances in experimental social psychology (Vol. 15, pp. 249-288). New York: Academic
Press.
Berkowitz, L. (1983). Aversively stimulated aggression: Some parallels and differences in
research with animals and humans. American Psychologist, 38, 1t35-1144.
Berkowitz, L. (1989). The frustration-aggression hypothesis: An examination and
reformulation. Psychological Bulletin, 106, 59-73.
Berkowitz, L. (1990). On the formation and regulation of anger and aggression: A
cognitive-neoassociationistic analysis. American Psychologist, 45, 494-503.
Berkowitz, L. (1993a). Aggression: lts causes, consequences, and controL New York:
McGraw-Hill.
Berkowitz, L. (1993b). Towards a general theory of anger and emotional aggression:
Implications of the cognitive-neoassociationistic perspective for the analysis of anger and
other emotions. In R. S. Wyer, Jr., & T. K. Srull (Eds.), Advances hl social cognition:
11ol. 6. Perspectives on anger and emotion. Hillsdale, NJ: Erlbaum.
Berkowitz, L., Cochran, S., & Embree, M. (1981). Physical pain and the goal of aversively
stimulated aggression. Journal of Personality and Social Psychology, 40, 687-700.
Berkowitz, L., & Embree, M. C. (1987). The effect of escape possibility on aversively
stimulated aggression. Journal of Research in Personality, 21, 405-416.
Berkowitz, L., & Theme, P. R. (1987). Pair expectation, negative affect, and angry aggression.
Motivation and Emotion, 11, 183-193.
Berkowitz, L., & Troccoli, B. T. (1990). Feelings, direction of attention, and expressed
evaluations of others. Cognition and Emotion, 4, 305-325.
Blanehard, R. J., Blanchard, D. C., & Takahishi, L. K. (1978). Pain and aggression in the
rat. Behavioral Biology, 23, 291-305.
Brain, P. F. (1981). Differentiating types of attack and defense in rodents. In P. F. Brain and
D. Benton (Eds.), Multidisciplinary approaches to aggression research (pp. 53-78).
Amsterdam/New York/Oxford; Elsevier North-Holland.
Ferguson, T. J., and Rule, B. G. (1983). An attributional perspective on anger and aggression.
In R. G. Geen and E. I. Donnerstein (Eds.), Aggression: Theoretical and empirical reviews
(Vol. 1, pp. 41-74). New York: Academic Press.
Feshbach, S. (1964). The function of aggression regulation of aggressive drive. Psychological
Review, 71, 257-272.
Geen, R. G. (1990). Human aggression. Milton Keynes, UK: Open University Press.
Hartmann, D. P. (1969). Influence of symbolically modeled instrumental aggression and pain
cues on aggressive behavior. Journal of Personality and Social Psychology, 11, 280-288.
Hatch, J. P., Moore, P. J., Borcherding, S., Cyr-Provost, M., Boutros, N. N., & Seleshi, E.
(1992). Electromyographic and affective responses of episodic tension-type headache
patients and headache-free controls during stressful task performance. Journal of
Behavioral Medicine, 15, 89-112.
Hutchinson, R. R. (1983). The pain-aggression relationship and its expression in naturalistic
settings. Aggressive Behavior, 9, 229-242.
Hutchinson, R. R., Renfrew, J. W., & Young, G. A. (1971). Effects of long-term shock and
associated stimuli on aggressive and manual responses. Journal of the Experimental
Analysis of Behavior, 15, 141-166.
Knutson, J. F., Fordyce, D. J., & Anderson, D. J. (1980). Escalation of irritable aggression:
Control by consequences and antecedents. Aggressive Behaviot, 6, 347-359.
Leventhal, E. A., Leventhal, H., Shacham, S., & Easterling, D. V, (1989). Active coping
reduces reports of pain from childbirth. Journal of Consulth~g and Clinical Psychology, 57,
365-371.
Leventhal, H., Brown, D., Shacham, S., & Engquist, S. (1979). The effects of preparatory
information about sensations, threat of pain, and attention on cold presser distress.
Journal of Personality and Social Psychology, 37, 970-985.
Mathew, N. T. (1990). Advances in cluster headache. Neurologic Clinics, 8, 867-890.
Melzack, R. (1973). The Puzzle of Pain. New York: Basic Books.
Moyer, K. E. (1976). The psychobiology of aggression. New York: Harper & Row.
Pain and Aggression 293

Potegal, M. (1979). The reinforcing value of several types of aggressive behavior: A review.
Aggressive Behavior, 5, 353-373.
Schneider, W., Dumais, S. T., & Shiffrin, R. M. (1984). Automatic and control processing
and attention. In R. Parasuraman & D. R. Davies (Eds.), VaHelies of attention (pp 1-27).
New York: Academic Press.
Sternbach, R. A. (1978). The psychology of Pah~. New York: Raven Press.
Summers, J. D., Rapoff, M. A., Varghese, G., Porter, K., & Palmer, R, E. (1991). Psychosocial
factors in chronic spinal cord injury pain. Pain, 47, 183-189.
Swart, C., & Berkowitz, L. (1976). The effect of a stimulus associated with a victim's pain on
later aggression. Journal of Personality and Social Psychology, 33, 623-631.
Ulrich, R. E. (1966). Pain as a cause of aggression. American ZoologL~t, 6, 643-662.
Ulrich, R. E., & Azrin, N. H. (1962). Reflexive fighting in response to aversive stimulation.
Journal of the Experimental Analysis of Behavior, 5, 511-520.
Vernon, W., & Ulrich, R. (1966). Classical conditioning of pain-elicited aggression. Science,
152, 668.
Wade, J. B., Price, D. D., Hamer, R. M., Schwartz, S. M., & Hart, R. P, (1990). An emotional
component analysis of chronic pain. Pain, 40, 303-310.
Weiner, B. (1985). An attributional theory of achievement motivation and emotion.
Psychological Review, 92, 548-573.