Volume 31, pages 407–419 (2005)
Hot Sauce, Toy Guns, and Grafﬁti: A Critical
Account of Current Laboratory Aggression
and Mike Eslea
Department of Psychology, University of Central Lancashire, Preston PR1 2HE, United Kingdom
In 1996, Tedeschi and Quigley published a review of laboratory aggression research that included many
damning criticisms of the genre. This paper revisits Tedeschi and Quigley’s critique, and examines the
ways that subsequent researchers have addressed the weaknesses they identiﬁed. In particular, it
examines three new laboratory aggression paradigms (Hot Sauce, Bungled Procedure, and
Experimental Grafﬁti Paradigms) that have attempted to improve upon the ‘‘classic’’ paradigms
(Teacher/Learner & Essay Evaluation Paradigms, Competitive Reaction Time Game, Point
Subtraction Aggression Paradigm, and Bobo Modelling Paradigm). In each case, this review will
show, that although some aspects of the new designs are indeed improvements, many of Tedeschi and
Quigley’s arguments still apply. In conclusion, this investigation will identify a number of factors that
future laboratory aggression researchers should consider. These include: The perceptions and
motivations of the aggressor; the apparent distance between the aggressor and the target; the
availability of non-aggressive response options; the problems of demand characteristics, permissive cues
and agentic shift; the differences between proactive and reactive aggression, and the distinction between
overt and covert forms of aggression. Aggr. Behav. 31:407–419, 2005.r2005 Wiley-Liss, Inc.
Keywords: laboratory aggression; hot sauce paradigm; bungled procedure paradigm; experimental grafﬁti
This paper focuses on the laboratory-controlled measurement of reactive aggression (e.g.,
anger expression, vengeful hostility), which is often viewed as a reaction to frustration or
provocation. It differs from proactive aggression (e.g., bullying, domination, coercive acts) in
that the former is a frustration response [Berkowitz, 1989] associated with a lack of self
control and the latter being less emotional and more likely driven by the expectation of
reward [Bandura, 1983]. A number of aggression models have been proposed to describe and
explain what factors might mediate the relationship between provocation/frustration and
reactive aggression. According to Berkowitz’s  Cognitive-Neoassociationist Model of
Affective Aggression, frustration leads to aggression by initiating negative affect, which
Correspondence to: Dominik Ritter, Department of Psychology, University of Central Lancashire, Preston PR1
2HE, United Kingdom. E-mail: email@example.com or firstname.lastname@example.org
Received 19 February 2003; amended version accepted 27 February 2004
Published online 16 February 2005 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ab.20066
r2005 Wiley-Liss, Inc.
in turn is linked to aggression through an associative network of emotions (e.g., anger, fear),
cognitions (e.g., hostile thoughts), and expressive motor responses (e.g., aggression, escape).
Similarly, Anderson et al.’s  General Affective Aggression Model states that negative
affect is a consequence of any aversive antecedent, and that it might mediate the activation or
simply be a concomitant of hostile thoughts and physiological arousal. However, unlike
Berkowitz’s model, Anderson et al.  argued that affect, cognition, and physiological
arousal are all parallel processes engendered by aversive antecedent conditions, although the
activation of one element tends to increase the accessibility of other elements in that network.
In 1996, Tedeschi and Quigley published a review of laboratory aggression research that
included many damning criticisms of the genre, including the distinct stimulus-response
nature of most past and present laboratory aggression research, and the use of inadequate
paradigms that were developed at the height of behaviourism (1960s), when researchers were
primarily concerned with observable behaviour at the expense of mediating processes such as
cognition and emotion.
Early Aggression Models include the Frustration-Aggression Hypothesis by Dollard et al.
, according to which frustration produces a condition of readiness to aggress, and that
aggression is always preceded by frustration. Early laboratory aggression research and classic
laboratory aggression paradigms might have been useful in identifying different kinds of
frustrating experiences (e.g., disappointment, irritation, annoyance, punishment) that might
contribute to the instigation of aggressive behaviour, but they might tell us little about why
some people behave aggressively, and others do not, under similar circumstances, whether
people perceive their own behaviour to be as aggressive as the experimenter does, and what
other cognitive and affective factors might contribute to the display of aggressive behaviour.
It is these cognitive and affective mediators, as well as other factors (e.g., the distance
between the aggressor and the target; the availability of non-aggressive response options; the
problems of demand characteristics, permissive cues, and agentic shift; the differences
between proactive and reactive aggression; and the distinction between overt and covert
forms of aggression), that have been largely ignored by most researchers studying aggression
in the laboratory.
CLASSIC LABORATORY AGGRESSION PARADIGMS
Tedeschi and Quigley  noted that although thousands of laboratory experiments
have been conducted to examine aggressive behaviour, the bulk of these have followed one of
just four main paradigms. Tedeschi and Quigley dub these ‘‘The Big Four’’ [p. 165]. These are
the Teacher/Learner, Essay Evaluation, Competitive Reaction Time Game, and Bobo
Modelling Paradigms. Another popular paradigm, not covered by Tedeschi and Quigley
, is the Point Subtraction Aggression Paradigm. For readers unfamiliar with these, a
brief critical summary of each now follows.
In the Teacher/Learner Paradigm [Buss, 1961], participants play the role of a teacher to a
learner (actually a confederate) in a memory task. Typically, the participant is located in a
separate room where he receives information about the learner’s responses on the memory
task (correct or incorrect) and then ‘‘punishes’’ incorrect responses by delivering electric
shocks. The Essay Evaluation Paradigm [Berkowitz et al., 1962] is similar: instead of
punishing responses on a memory task, participants are asked to ‘‘evaluate’’ essays
supposedly written by a confederate. Tedeschi and Quigley  pointed out that these two
408 Ritter and Eslea
paradigms share several weaknesses. Both involve cover stories that could lead participants
to give shocks for prosocial (even altruistic) reasons (to help teach, for example) and not in
order to cause harm. Both paradigms involve an authority ﬁgure (the experimenter) being
permissive of the aggressive responses. Both require considerable distance between
participant and confederate (usually separate rooms). Both give participants only aggressive
response options and no other way of responding. Supposedly aggressive behaviour in the
Teacher/Learner and Essay Evaluation Paradigms is therefore likely to be very different from
real world aggression.
The next ‘‘Big Four’’ paradigm is the Competitive Reaction Time Game [Taylor, 1967].
Participants are told that they are to play a game against another player (actually a
confederate), in which they must press a button as quickly as possible after a particular
signal. Before each trial, the players decide what severity of electric shock will be applied to
the other participant in the event that he is slower. In fact, winning, losing, and confederate
choice of shock intensity are manipulated by the experimenter. The main advantage of this
paradigm compared to the Teacher/Learner and Essay Evaluation Paradigms is the absence
of a cover story, so the delivery of shocks cannot be said to have a prosocial motivation.
However, as Tedeschi and Quigley  note, the iterative nature of the game may
encourage aggressive responding for other reasons, such as reciprocity (fairness) and
deterrence (social control), rather than desire to harm. There is also the possibility that very
competitive participants might use more severe shocks, not because they want to harm their
opponent, but to negatively affect their reaction rates and so ‘‘win’’ the game. Also, many of
the disadvantages mentioned in the previous section are also found here: distance between
participant and confederate, lack of alternative response options, and experimenter
permissiveness of aggressive responding.
The last of Tedeschi and Quigley’s  ‘‘Big Four,’’ and probably the most famous, is
the Bobo Modelling Paradigm [Bandura, 1973], in which participants (usually children)
watch an adult playing aggressively with a large inﬂatable toy clown (‘‘Bobo’’). Many
variables can be manipulated, such as whether the behavior of the adult model is rewarded or
punished. The outcome variables are usually either the children’s own use of aggressive
behaviours when they are subsequently allowed to play with Bobo, or their recall of the
behaviour of the model. These studies have generated a great deal of interesting data about
the circumstances in which children imitate behaviours they have seen, but it is not clear
whether these behaviours really deserve to be considered aggressive. Instead, Tedeschi and
Quigley  argue convincingly that children in Bobo experiments are merely enjoying
There is another laboratory aggression paradigm, not included in Tedeschi and Quigley’s
 critique, but which is worthy of inclusion here as a ‘‘classic,’’ since it has been used in
a great many studies since its ﬁrst publication in 1981. We refer to the Point Subtraction
Aggression Paradigm [Cherek, 1981]. Participants are seated at a computer and told that
pressing a particular button (usually 100 times) will earn them a sum of money (usually 10
cents or equivalent). Alternatively, they can press a second button (usually 10 times) to
deduct the same sum from another participant supposedly playing the same game in a
different room. An on-screen counter displays their running total, and allows them to see
when their (ﬁctitious) opponent has deducted 10 cents from their pot. In fact, subtractions
are made on a predetermined basis by the computer. The outcome (aggression) variable is
therefore simply the number of presses of the participant’s own subtraction button. The
paradigm also lends itself well to further design manipulations: for example, in some versions
Laboratory Aggression Paradigms 409
of the game, there is a third button which participants can use to protect their total from
subtraction for a certain length of time [Cherek et al., 2000]. Unlike in the Teacher/Learner,
Essay Evaluation, and Competitive Reaction Time Game Paradigms, the participant is not
forced to respond aggressively (they could simply continue hitting their earn button), but as
in the other paradigms there is no non-aggressive way to interact with the opponent. The
problems of (apparent) distance between aggressor and target, and of experimenter
permissiveness of aggressiveness, are found here, too.
To summarise, Tedeschi and Quigley  identiﬁed reasons to be sceptical of all major
aggression paradigms claiming to be laboratory analogues of real world behaviour. In
particular, they argued that insufﬁcient attention had been paid to the motivations behind
apparently aggressive behaviour.
NEW LABORATORY AGGRESSION PARADIGMS
In recent years, several researchers have developed new methods to measure aggressive
behaviour under controlled laboratory conditions, in an attempt to overcome the limitations
of classic laboratory aggression paradigms discussed above. The next section introduces three
such attempts. These are the Hot Sauce, Bungled Procedure, and Experimental Grafﬁti
Hot Sauce Paradigm
Lieberman et al.  developed this new method in which hot chili sauce administration
is used as a measure of physical aggression. In this procedure participants are required to
determine the amount of hot sauce to be (purportedly) consumed by another person who
allegedly does not like spicy foods and who has provoked the participant beforehand, usually
either by giving them a noxious juice sample, or writing a ‘‘world-view threatening’’ essay
(strongly opposing and insulting the political convictions of the participant).The advantages
of this method are: that it does not require expensive and elaborate equipment (as in the case
of electric shock paradigms); the observed behaviour is easily quantiﬁable (amount of sauce)
and less likely to be interpreted as competitive; that it is ecologically valid as spicy food has
been used in real world assaults [BBC News, 2001] and child abuse cases [Galletta, 2002;
JSOnline Milwaukee Journal Sentinel, 2000]. McGregor et al.  established the validity
of this new paradigm utilising theoretical perspectives, such as Terror Management Theory.
In their study there were two ways of responding (derogation or hot sauce administration) as
a means of reacting to provocation. Both ways of responding seemed to serve the same
function, to defend one’s own worldview. If participants were able to use derogation as a
defence mechanism, they administered less hot sauce than if they were allowed to use
derogation only after administering hot sauce. Furthermore, convergent validity of this
paradigm was also provided by the correlation between amount of hot sauce and both the
overall and physical aggression scores on the Buss and Perry  Aggression Questionnaire.
Bungled Procedure Paradigm
In a similar attempt to add face validity to a laboratory aggression paradigm, Russell et al.
[1996, 2002] developed a paradigm in which participants are given the opportunity to shoot at
a human target with a pellet or paintball gun. The target is a woman, and the task is
410 Ritter and Eslea
presented as a ‘‘novel form of male entertainment’’ [Russell et al., 1996]. Aggression in this
paradigm is operationalised as the power of the gun chosen (from an array of guns of varying
power) multiplied by the number of pellets elected to be used to shoot at the target. In reality,
participants never actually shoot at the target, as they are told that there has been a mistake
(the ‘‘Bungle’’ of the title), that they are in fact in the control condition, and that they are not
therefore required to shoot at the target. The great advantage of this paradigm is its
considerable face validity, as many real world violent acts, such as homicide, involve ﬁrearms.
It differs from other paradigms in that the participants are not provoked beforehand, so
responding under these conditions could be regarded as proactive (instrumental) aggression
(future investigations employing this paradigm might make use of provocation to explore
reactive aggression). This paradigm allows various manipulations to explore participants’
potentially aggressive intentions in relation to more serious acts of violence. However, the
procedure only allows assessment of potentially aggressive intentions, as the actual aggressive
behaviour never takes place.
The third new paradigm is Norlander et al.’s  Experimental Grafﬁti and Tearing
Procedure. This is another interesting attempt at a face-valid aggression paradigm that uses a
real-world behaviour as its outcome variable: physical damage to another’s property. Unlike
the paradigms discussed above, which focus on direct aggression (albeit often at a distance),
this method elicits an indirect form of aggression. For example, damage to property has been
found to be used aggressively by prisoners [Ireland, 1999] and spouses seeking revenge for
inﬁdelity [Archer, 2001].
In the Experimental Grafﬁti and Tearing Paradigm [Norlander et al., 1998], participants
perform two tasks. First, they are provided with an illustration of ‘‘Adam and Eve in the
Garden of Paradise’’ and instructed to draw upon it. Judges then rate the amount of
‘‘grafﬁti’’ thus added, the degree of destruction caused, and any aggressive or sexual content
in the additions. Next, participants are provided with an illustration of ‘‘Samson and the
Lion’’ (chosen for its ‘‘strongly aggressive character, which is intended to provoke
participants to exhibit aggressiveness’’ [p. 202]) and instructed to tear apart the illustration
into a number of pieces. Participants are then instructed to place all the pieces in an envelope
that is half the size of the illustration. There is no time limit and the number of pieces
produced is monitored as a dependent variable. Factors under investigation usually include
sex differences, the inﬂuence of alcohol, and the effect of frustration (one group is given an
impossible task beforehand). Indirect support for the validity of this paradigm has been
provided by Kortynk and Perkins  who demonstrated that young men who had
consumed alcohol wrote and sketched more grafﬁti, compared with men who had not drunk
alcohol, which they interpreted as the tendency for alcohol to increase vandalism behaviour,
which is indeed a common observation in the real world [Goldstein, 1996].
CRITICIMS OF LABORATORY AGGRESSION PARADIGMS
So far, this study has summarised ﬁve ‘‘classic’’ paradigms, Tedeschi and Quigley’s 
critique thereof, and three new attempts to study aggression under controlled conditions. In
the remaining sections we consider each of the common problems in turn, beginning with the
issue of mediating variables.
Laboratory Aggression Paradigms 411
A shortcoming of laboratory aggression research, even in studies employing new
paradigms, is that it has maintained its distinct stimulus-response character. According to the
General Affective Model of Aggression [Anderson et al., 1995] and the Cognitive
Neoassociationist Model of Affective Aggression [Berkowitz, 1993] one’s affect and
cognitions, as well as one’s physiological arousal, play important roles in mediating the
relationship between any external stimulation (e.g., interpersonal provocation, frustration,
pain) and one’s behaviour (e.g., aggression, escape). However, many researchers attempting
to measure aggressive behaviour in the laboratory still seem to ignore mediating variables,
such as affect (e.g., anger, fear) and cognitions (e.g., hostile thoughts). The lack of interest in
participants’ motivations, as criticized by Tedeschi and Quigley , is particularly
striking, since aggressive behaviour in response to provocation (reactive aggression) has been
deﬁned as the intentional harm of another person. How can researchers be certain that their
participants actually intend to harm the other person?
Social Cognition and Motivation
The point about divergent perceptions between participants and experimenters was raised
earlier in relation to the Teacher/Learner Paradigm and Essay Evaluation Paradigm and their
use of cover stories [Tedeschi and Quigley, 1996]. Participants are either informed that they
are teaching another person by punishing incorrect responses in a learning task and by giving
verbal reinforcements for correct responses, or that shocks are the means by which another
person’s intellectual or creative product is evaluated. Thus, if participants believe and act
according to the cover stories, then their aim is not to harm, but to teach and to evaluate the
performance of the other person.
It is highly questionable whether participants in the Bungled Procedure Paradigm perceive
their behaviour as aggressive or are motivated by the desire to harm another person. It was
found that participants perceive this task primarily as an entertaining, exciting, and enjoyable
experience [Russell et al., 1996, 2002], with these hedonic ratings being more important
predictors of the supposedly aggressive responses than personality and other aggression
measures. It should be acknowledged, though, that Russell et al.  found that the
number of shots was related to antisocial tendencies and psychopathy. However, despite its
attractive face validity and experimental realism (participants reported disappointment when
prevented from shooting at the human target), there was no relation between the physical
aggression scale of the Buss and Perry  Aggression Questionnaire (AQ) and the
aggression measure. The key factors appeared to be psychopathic and antisocial tendencies,
rather than physical aggression. In the Russell et al.  study, psychotic tendencies were
negatively related to estimates of hurting, thus indicating that these individuals underestimate
the pain that their actions cause the victims of their aggression. However, if participants think
that being shot at would not be painful, can their behaviour really be regarded as aggression?
Furthermore, antisocial tendencies were found to be unrelated to aggressive responding,
although Arms et al. [unpublished data] found a relationship between the aggression measure
and the physical aggression and anger scale of the AQ.
Unlike most paradigms, the confederate in the Bungled Procedure Paradigm is a real
visible person, so many of the problems of distance between aggressor and victim are
avoided. However, the confederate does wear protective gear and this could convey to
participants that being shot at does not really cause any harm. Just as a Bobo doll is designed
412 Ritter and Eslea
to be bashed, protective gear seems designed to be shot at. Participants are thus much more
likely to regard their behaviour as play aggression or rough-and-tumble-play.
Regarding the Experimental Grafﬁti and Tearing Paradigm, Norlander et al. 
discovered that women scored higher on elaboration than men, which made these authors
wonder whether, at least for the female sample, scrawling grafﬁti was a creative expression.
The lack of correlation between elaboration and scrawling-grafﬁti in both males and females
contradicts this possibility. But would participants necessarily regard their behaviour as
intentionally harmful to someone’s property? In this paradigm the value attached to the
artwork might not be regarded as very high. Why would one regard the tearing of a piece of
paper as aggressive? It might be a good idea to convey to participants that it is a piece of
work of considerable value, e.g., painted by another participant. This way one could
manipulate the worth of the artwork and make its destruction more serious and personal. It
may be that this procedure falls into the Bobo Modelling Paradigm trap noted above: that
the responses are really rough-and-tumble play, not aggression.
In relation to the Hot Sauce Paradigm, Lieberman et al.  argued that many of their
participants indicated during debrieﬁng that they thought that the sauce was quite hot and
reported allocating large amounts, because they did not like the target. This, however, is a
rather vague statement about the number of participants behaving in such a manner. A more
detailed analysis of such motives would provide stronger evidence for the validity of this
measure to assess intentional harm of another person.
There have been a number of studies, dating back to the early 1970s, highlighting the
importance of exploring participants’ motivations in laboratory aggression paradigms. For
example, Baron and Eggleston  found a positive correlation between self-ratings of
altruism and the intensity of shocks administered to the learner. Rule and Nesdale 
found that the most intense shocks were given to the confederate when the participant was
not insulted and when they were told that shocks facilitated the learner’s performance.
Despite this knowledge about the importance of mediating variables such as motivations,
and the frequent criticisms in relation to this issue [e.g., Tedeschi and Quigley 1996, 2000],
many researchers (including those who developed the new paradigms) still fail to manipulate,
or control, any of the multiple goals that may motivate aggression when studying aggressive
behaviour in the laboratory. It is true that administering electric shocks, sound bursts, or hot
sauce to someone is more or less harmful to the other person. This, however, does not mean
that these behaviours, which have similar outcomes (although varying in degree), are caused
by similar motivations. The problem of laboratory aggression paradigms is that they are
deﬁned by the outcome of a particular behaviour rather than its underlying motivation. We
argue that it is vital to explore participants’ motivations underlying their behaviour in the
laboratory if we wish to generalise to behaviour in the real world.
Affect (e.g., anger) is known to play a major role in reactive aggression [Berkowitz, 1993],
when one retaliates for being provoked. However, as with motivation, affective states have
also largely been ignored in laboratory aggression research. From studies employing the Hot
Sauce Paradigm, it is not clear whether the negative affect as produced by the provocations
(bad juice or worldview-threatening essay), was maintained throughout the experiments, as
affect was only being measured straight after the provocation, but not prior to, during, or just
after retaliation. In fact, the mood induction in McGregor et al.’s  study was not
Laboratory Aggression Paradigms 413
successful, as there was no signiﬁcant difference between the ‘‘Mortality Salience’’ (MS)
group (participants who are made to think about their own death) and the ‘‘Exam Control’’
group (participants who are made to think about taking an exam), thus casting serious doubt
over whether participants’ behaviour in that study could be regarded as hostile physical
aggression. In fact, MS participants in their third study were in a better mood than controls,
which these authors explain as the denial of these participants’ own distress. Again, it is
questionable whether the administration of hot sauce could be regarded as hostile physical
aggression, since participants in this particular study were, in fact, in a much better mood
than controls. This ﬁnding is contrary to what one might expect from Berkowitz’s 
Neassociationist Model of Affective Aggression, according to which negative affect has a
direct inﬂuence on emotions (e.g., anger), cognitions (e.g., hostile thoughts), and expressive
motor patterns (e.g., physical aggression).
In relation to the Bungled Procedure Paradigm, it seems unlikely that ‘‘Samson and the
Lion’’ could really be considered a provocation in the same sense as electric shocks, point
subtraction, or worldview threatening essays. There was no manipulation check, so one
cannot be certain whether participants really felt provoked/angered by the picture. At most,
some arousal and/or perhaps some priming of aggressive responses may have occurred. There
is evidence that people might be cognitively primed with aggressive themes when exposed to
stimuli of a violent nature. For example, if exposed to a ‘‘weapon,’’ participants tend to
behave more aggressively than if no weapon is present [Berkowitz and LePage, 1967; Maass
and Koehnken, 1989].
The new paradigms still fail to give participants a choice among means of responding (to
provocations) and only provide them with physical forms of aggressive behaviour. In the Hot
Sauce Paradigm, for example, participants are not given alternatives (e.g., different tasting
foods); their only choice in the McGregor et al.  experiments, for example, is to
administer hot sauce. They can neither choose between a variety of different tastes, nor the
degree of spiciness. Furthermore, participants in the Bungled Procedure Paradigm might be
under the impression that if they chose not to shoot at the target their data would be deemed
invalid. Thus, not choosing to shoot at the human target is unlikely to be regarded as a non-
aggressive response option on the part of the participant (in contrast to the experimenter).
Instead, it could be seen as dropping out of the experiment. Furthermore, as participants in
the Bungled Procedure Paradigm are informed that it is about men’s impressions of a novel
form of entertainment, they might see themselves forced to shoot at the human target so that
they can report on their impression. It might be better to offer various options (e.g., targets,
inanimate objects, etc.) to the participant and assess the number of times the participant
would choose to shoot at the human target, or to assess which body parts the participant
would shoot at. A headshot could be regarded as more aggressive than a shot at the legs,
because of the former one’s greater potential for serious injury. On the other hand, large
body parts such as the torso might be preferred by participants, not because of the greater
likelihood of damage, but because it is an easier target to hit, compared to a relatively small
one (e.g., the head).
414 Ritter and Eslea
By failing to give alternative ways of responding, one does not know how participants
would react if they had other choices. Interestingly, Hokanson , for example, found
that by giving female participants a choice between rewarding and punishing a confederate
who had insulted them, they frequently chose to use rewards.
Distance Between Participant & Confederate
Another problem relates to the distance between the participant and the (often
‘‘hypothetical’’) opponent/confederate. Any noxious stimulation (e.g., hot sauce, electric
shocks) is delivered from a distance, where the aggressor and the victim cannot see or hear
each other. However, it needs to be acknowledged that in the Bungled Procedure Paradigm
the confederate is a real visible person, so many of the problems of distance between
aggressor and victim are avoided. However, this paradigm only allows one to infer the
participant’s intentions, as they do not come to actually shoot at the confederate.
Interpersonal conﬂict outside the laboratory often involves close proximity, and bodily
harm involves physical contact between the harm-doer and the victim, making it difﬁcult to
determine if and to what extent the behaviour in these paradigms relates to actual violence in
the real world. We know from the famous Milgram  studies on obedience that by
merely increasing the distance between aggressor and victim, the threshold for behaving
aggressively decreases, which might be due to emotional detachment (one does not see/hear
the victim’s suffering), less danger for the self, and less strenuousness. In other words, it
might be easier to drop a bomb from a plane onto a city, than to beat someone to death.
Obedience, Permissive Cues, and Demand Characteristics
It can be argued that the chief instigation for using noxious stimuli in these new aggression
experiments is the command of a legitimate authority, and thus many aggression experiments
could be regarded as variations of the Milgram  experiments on authority. Gottfredson
and Hirschi  argued that participants in laboratory experiments might be agreeable and
compliant individuals sensitive to social norms, thus measuring conformity to an authority
ﬁgure and/or compliance to demand cues of experimenters [Orne, 1962]. What occurs under
these conditions is known as ‘‘agentic shift’’ (acting as an agent for an authority ﬁgure, and
thus removing personal responsibility), which increases the likelihood of behaving
aggressively. In the Norlander et al.  study, for example, women scored higher on
scrawling-grafﬁti and destruction in comparison to men, and there was no sex-difference on
aggression or tearing. It is therefore questionable whether this paradigm really assesses
destructive behaviour, which is usually attributed to men [Goldstein, 1996]. As women tend
to be more conforming in general [Collin et al., 1994; Eagly and Chrvala, 1986], it suggests
that this procedure might in fact induce conformist behaviour.
Another problem is the excess of permissive cues, such as the experimenter’s approval or
the apparent harmlessness of the behaviour. In the Experimental Grafﬁti Paradigm, for
example, participants are speciﬁcally instructed to draw on pictures or tear apart illustrations
into a number of pieces. The Hot Sauce Paradigm suffers from similar demand
characteristics, as participants are instructed to administer ‘‘as much or as little’’ hot sauce
to someone who has just provoked them, thus potentially guiding them as to how to behave
in this unfamiliar situation. It is noteworthy, though, that Russell et al.  acknowledged
the excess of permissive cues in the Bungled Procedure Paradigm (e.g., the experimenter’s
approval), which according to Leonard and Taylor  can increase men’s aggression
Laboratory Aggression Paradigms 415
towards women. An exception to this is a study by Kortynk and Perkins  assessing
scrawling grafﬁti as spontaneous behaviour. Here participants were not speciﬁcally told to
scrawl grafﬁti, thus reducing demand characteristics. Participants were simply given the
opportunity to scrawl grafﬁti by providing appropriate surroundings in the form of posters
and signs, some of which were covered in grafﬁti. During the experimental session the
experimenter was (supposedly) required to leave the laboratory for a while, leaving the
participant to his/her own devices.
Overt vs. Covert Aggression
Another important issue is the distinction between covert (more indirect and disguised)
and overt acts (more direct and obvious) of aggression. In the Competitive Reaction Time
Paradigm the severity of electric shock could be regarded as overt aggressive behaviour, while
the length of shocking time could be regarded as a more covert form of aggressive behaviour.
Similarly, in the Bungled Procedure Paradigm the number of shots and the power of the gun
chosen are likely to represent two different types of aggression, with the former likely to be a
more overt and the latter likely to be a more covert act of aggression. Finally, in the Hot
Sauce Paradigm the amount of hot sauce appears to be a relatively overt form of aggression
(although the cups used were covered with lids so the experimenters could not see the amount
of sauce being dispensed, they could presumably feel the mass). It would be interesting to
replicate the procedure giving participants a covert way to aggress (e.g., by giving them a
range of spice intensities to choose from).
Laboratory aggression paradigms are necessarily constrained (at least in recent years) by
the acceptability of angering or frustrating participants, and thus it is difﬁcult to determine
whether the relatively mild provocations under laboratory conditions bear any resemblance
to the provocations occurring in real life (i.e., severe verbal and physical abuse). These severe
forms of provocations might elicit behaviour not being produced by relatively mild forms in
Another problem is that although it is desirable to add face validity to laboratory
aggression paradigms by creating some surface resemblance to real world situations, there is
a danger that, in doing so, one might prime participants to behave aggressively in a similar
situation encountered outside the laboratory. For example, participants in the Teacher/
Learner Paradigm are unlikely to later ﬁnd themselves in a situation where they could give
shocks to another person. On the other hand, it is quite possible that Hot Sauce participants
may later ﬁnd themselves preparing a meal for an irritating spouse or ungrateful child. This
point emphasises the importance of thorough debrieﬁng.
WHAT MAKES A GOOD LABORATORY AGGRESION PARADIGM?
In summary, all the aggression paradigms discussed here have strengths and weaknesses.
The most common weakness are the cover stories used to disguise the real purpose of the
studies, which often leads to serious confounds, such as changing the participants’
perceptions of the ‘‘aggressive’’ behaviour into either prosocial behaviour (such as teaching)
or competitive behaviour (which may be superﬁcially aggressive, but which involves no desire
416 Ritter and Eslea
to just harm). Furthermore, the ‘‘aggressive’’ behaviour is almost always explicitly permitted
by the experimenter, leading to excessive demand characteristics, and the confounding of
aggression with conformity and agentic shift. A good laboratory aggression paradigm would
give participants the opportunity to aggress, without drawing attention to that opportunity in
a permissive way. Ideally, such a paradigm should allow aggression to emerge spontaneously
(proactive aggression) as well as following provocation (reactive aggression).
Almost all aggression paradigms to date have problems with regard to the supposed targets
or victims of the aggression, either because of the distance involved (e.g., the victim is
supposed to be in another room) or because the behaviour does not seem likely to cause real
harm (e.g., the victim is only a doll, or wears protective clothing). A good paradigm would
involve direct contact with a real confederate, using a behaviour that is unambiguously
harmful in appearance, and that cannot be confused with rough-and-tumble play. Ideally, the
aggressive behaviour will be one of a range of available response options, including prosocial
and communicative responses as well as merely non-aggressive behaviours. Paradigms in
which aggression is the only way to interact with the opponent can tell us little about how
people elect to aggress in the real world.
The majority of laboratory aggression studies have explored interpersonal physical
aggression. It thus appears to be reasonable to also assess other forms of aggressive
behaviour, such as verbal aggression [Berkowitz, 1970; Wheeler and Caggiula, 1966]; written
aggression, (providing negative evaluations of a provocateur; Fisher and Harris, 1976;
Rohsenow and Bachorowski, 1984); impersonal aggression, (property destruction; Norlander
et al., 1998; Nation and Cooney, 1982), and indirect physical displaced aggression (delivering
noxious sound blasts to an unknown future participant; Sherrod et al., 1979). Furthermore, it
seems sensible to operationalise aggression as aggressive responding that bears some
resemblance to aggressive behaviour in the real world (i.e., hot sauce, shooting at someone).
As a result, one would not have to make inferences from behaviour observed under
controlled laboratory conditions (i.e., button pressing) to behaviour outside the laboratory.
Most importantly, we concur with Tedeschi and Quigley’s  conclusion that such
paradigms must pay attention to the perceptions, beliefs, and motivations of the participant.
It needs to be acknowledged, though, that the new paradigms have advanced towards more
ecologically valid measures, which provide participants with more familiar tasks that they
can relate to (e.g., destroying property, shooting at a target) in comparison to rather abstract
and unfamiliar tasks, such as pressing different kinds of buttons to earn and deduct points,
and pressing buttons to deliver electric shocks to someone. Furthermore, the behaviours
under investigation in these new paradigms bear closer resemblance to real world aggressive
acts such as vandalism and assault, removing the problem inherent in classic laboratory
aggression paradigms of having to make inferences from behaviour inside the laboratory to
outside the laboratory. Finally, most researchers studying aggression in the laboratory now
provide participants with a non-aggressive behavioural option, which appears to be a direct
response to Tedeschi and Quigley’s  criticisms. However, these options are
unfortunately not real alternatives, as they simply involve not responding at all (e.g., not
administering hot sauce, stating that one wishes not to shoot at the human target). It would
thus be desirable to provide participants with real alternatives such as being able to behave in
a benevolent way towards the aggressor, to talk to the aggressor, to consult the experimenter,
or to just leave the laboratory.
Does the ideal paradigm exist? Not yet. Research inevitably involves compromise, and it
may be that such a paradigm is not even possible, but there is no reason why most of the
Laboratory Aggression Paradigms 417
suggestions here could not be incorporated into a single design. Such a paradigm could
certainly have a striking impact upon aggression research.
Anderson CA, Deuser WE, DeNeve KM. 1995. Hot
temperatures, hostile affect, hostile cognition, and
arousal: tests of a general model of affective
aggression. Pers Soc Psychol Bull 21:434–448.
Archer J. 2001. A strategic approach to aggression. Soc
Bandura A. 1983. Self-efﬁcacy determinants of antici-
pated fears and calamities. J Pers Soc Psychol
Baron RA, Eggleston RJ. 1972. Performance on the
‘‘aggression machine’’: Motivation to help or harm?
Psychon Sci 26:321–322.
BBC News. 2001. Seven hurt in hot sauce attack.
August 2003 from http://news.
Berkowitz L. 1970. Aggressive humor as a stimulus to
aggressive responses. J Pers Soc Psychol 16:710–717.
Berkowitz L. 1989. Frustration-aggression hypothesis:
examination and reformulation. Psychol Bull
Berkowitz L, Corwin R, Heironimus M. 1962. Film
violence and subsequent aggressive tendencies. Pub-
lic Opin Q 27:217–229.
Berkowitz L, LePage A. 1967. Weapons as aggression-
eliciting stimuli. J Pers Soc Psychol 7:202–207.
Buss AH. 1961. The psychology of aggression,
New York: Wiley.
Buss AH, Perry M. 1992. The Aggression Questionnaire.
J Pers Soc Psychol 63:452–459.
Cherek DR. 1981. Effects of smoking different doses of
nicotine on human aggressive behavior. Psychophar-
Cherek DR, Lane SD, Dougherty DM, Moeller FG,
White S. 2000. Laboratory and Questionnaire
Measures of Aggression Among Female Parolees
With Violent or Nonviolent Histories, Aggress
Collin CA, Di Sano F, Malik R. 1994. Effects of
confederate and subject gender on conformity
in a color classiﬁcation task. Soc Behav Pers
Dollard J, Doob LW, Miller NE, Mowrer OH, Sears
RR. 1939. Frustration and aggression. New Haven,
CT: Yale University Press.
Eagly AH, Chrvala C. 1986. Sex differences in con-
formity: Status and gender role interpretations.
Psychol Women Q 10:203–220.
Fisher JL, Harris MB. 1976. Modelling, arousal, and
aggression. J Soc Psychol 100:219–226.
Galletta J. 2002. Remembering Melisha Gibson, Chat-
tanooga Times Free Press. Retrieved 15
2003 from http://www.msnbc.com/local/ctfp/
CTFP2981913.asp?cp1 = 1
Goldstein AP. 1996. The psychology of vandalism.
New York: Plenum Press.
Gottfredson MR, Hirschi T. 1993. A control theory
interpretation of psychological research on aggres-
sion. In: Felson RB, Tedeschi JT, editors. Aggression
and violence: Social interactionist perspectives.
Washington, DC, US: American Psychological
Association. Xiii: 265. p 47–68.
Hokanson JE. 1970. Psychophysiological evaluation of
the catharsis hypothesis: In: Megargee EI, Hokanson
JE, editors. The dynamics of aggression. New York:
Harper & Row. p 74–86.
Ireland JL. 1999. Bullying behaviours among male and
female prisoners: a study of adult and young
offenders. Aggress Behav 25:161–178.
JSOnline Milwaukee Journal Sentinel. 2000. Man faces
child abuse charge in hot-sauce case. Retrieved 15
August 2003 from http://www.jsonline.com/WI/
Kortynk NX, Perkins DV. 1983. Effects of alcohol
versus expectancy for alcohol on the incidence of
grafﬁti following an experimental task. J Abnorm
Leonard KE, Taylor SP. 1983. Exposure to pornogra-
phy, permissive and nonpermissive cues, and male
aggression toward females. Motivation and Emotion
Lieberman JD, Solomon S, Greenberg J, McGregor HA.
1999. A hot new way to measure aggression: Hot
sauce allocation. Aggress Behav 25:331–348.
Maass A, Koehnken G. 1989. Eyewitness identiﬁcation:
Simulating the ‘‘weapon effect’’. Law Hum Behav
Martin NS. 1994. Child killing a losing battle in S.
Florida. Sun Sentinel.
McGregor HA, Lieberman JD, Greenberg J, Solomon S,
Arndt J, Simon L, Pyszczynski T. 1998. Terror
management and aggression: Evidence that mor-
tality salience motivates aggression against world-
view-threatening others. J Pers Soc Psychol 74:
Milgram S. 1963. Behavioral study of obedience.
J Abnorm Soc Psychol 67:376.
Nation JR, Cooney JB. 1982. The time course of
extinction-induced aggressive behavior in humans:
418 Ritter and Eslea
Evidence for a stage model of extinction. Learn
Norlander T, Gustafson R. 1997. Effects of alcohol on
picture drawing during the veriﬁcation phase of
the creative process, Creativity Research Journal
Norlander T, Nordmarker A, Archer T. 1998. Effects of
alcohol and frustration on experimental grafﬁti.
Scand J Psychol 39:201–207.
Orne MT. 1962. On the social psychology of the
psychological experiment: With particular reference
to demand characteristics and their implications. Am
Rohsenow DJ, Bachorowski JA. 1984. Effects of alcohol
and expectancies on verbal aggression in men and
women. J Abnorm Psychol 93:418–432.
Rule BG, Nesdale AR. 1974. Differing functions of
aggression. J Pers 42:467–481.
Russell GW, Arms RL, Loof SD, Dwyer RS. 1996.
Men’s aggression toward women in a bungled
procedure paradigm. J Soc Behav Pers 11:729–738.
Russell GW, Arms RL, Dwyer RS, Josuttes D.
2002. Females as targets of male aggression in
the bungled procedure paradigm. University of
Sherrod DR, Moore BS, Underwood B. 1979. Environ-
mental noise, perceived control, and aggression.
J Soc Psychol 109:245–252.
Taylor SP. 1967. Aggressive behavior and physio-
logical arousal as a function of provocation
and the tendency to inhibit aggression. J Pers 35:
Tedeschi J, Quigley B. 1996. Limitations of laboratory
paradigms for studying aggression. Aggress Viol
Laboratory Aggression Paradigms 419