Sei sulla pagina 1di 33

The Journal of Forensic Psychiatry & Psychology

ISSN: 1478-9949 (Print) 1478-9957 (Online) Journal homepage: http://www.tandfonline.com/loi/rjfp20

False inflated self-esteem and violence: a


systematic review and cognitive model

Julian S. Walker & Jenifer A. Bright

To cite this article: Julian S. Walker & Jenifer A. Bright (2009) False inflated self-esteem and
violence: a systematic review and cognitive model, The Journal of Forensic Psychiatry &
Psychology, 20:1, 1-32, DOI: 10.1080/14789940701656808

To link to this article: http://dx.doi.org/10.1080/14789940701656808

Published online: 02 Mar 2009.

Submit your article to this journal

Article views: 561

View related articles

Citing articles: 21 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rjfp20

Download by: [University of Nottingham] Date: 07 November 2015, At: 08:00


The Journal of Forensic Psychiatry & Psychology
Vol. 20, No. 1, February 2009, 132

RESEARCH ARTICLE
False inflated self-esteem and violence: a systematic review and
cognitive model
Julian S. Walkera* and Jenifer A. Brightb
a
Fromeside (Avon & Wiltshire Mental Health Partnership), Bristol, UK;
b
Maudsley Hospital, London, UK
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Traditionally, much research into violence has focussed on risk factors


rather than on perpetrators perspectives on their violent acts and the
powerful psychological influences on those individuals violent beha-
viour. In forensic settings, the most popular model for working with
violence has been anger management, which uses a cognitive behaviour-
al approach to explain how triggers may cause anger and violence via a
series of information processing biases. Interestingly, an area that
receives less attention in the cognitive behavioural literature on violence
and anger is the role of embarrassment and humiliation (disrespect),
and their opposites respect and pride (or healthy self-esteem). However,
psychodynamic perspectives put humiliation at the centre of causes of
violence, coupled with coping and social problem-solving deficits
resulting from disrupted attachments. Despite the absence of a focus
on self-esteem in models of and treatments for aggression and violence,
there is some recent research evidence that suggests a complicated
relationship between the two, but generally favours a link between low
self-esteem and violence. This paper systematically reviews studies from
the last 20 years evaluating the relationship between self-esteem and
violence. A theoretical model is subsequently presented in an attempt to
integrate ideas about self-esteem, machismo, and violence. It is
proposed that important cognitions relating to violence also relate to
self-esteem and the (arrogant or aggressive) protection of low self-esteem
in the face of humiliation. Violence can be seen as a macho response
which allows the perpetrator to express and discharge unpleasant
feelings associated with threat, and simultaneously serves a social
function by injuring the victim who provoked the assault, demonstrating
strength to others, and restoring some level of pride (saving face).
Keywords: self-esteem; violence; aggression; cognitive model;
formulation

*Corresponding author. Email: julian.walker@awp.nhs.uk

ISSN 1478-9949 print/ISSN 1478-9957 online


! 2009 Taylor & Francis
DOI: 10.1080/14789940701656808
http://www.informaworld.com
2 J.S. Walker and J.A. Bright

Introduction and definitions


Violence is a continuing problem in society at large, and the size and cost of
the problem (in terms of victim injury and distress, police time, and financial
costs) is considerable and estimated to run to many millions of pounds per
year in UK. For example, domestic violence (which accounts for a quarter
of all violence) was estimated by Stanko, Crisp, Hale, and Lucraft (1998) to
cost 5 m over one year in a single borough of London, including all costs
incurred by police, civil justice, housing, refuge, and health care services.
The development of psychological models and treatments specifically
targeted at individuals with violence problems is important for psychiatric,
Downloaded by [University of Nottingham] at 08:00 07 November 2015

custodial, and forensic services, and potentially has considerable implica-


tions for costs both financial and victim costs. Recent political debates
over the treatability of patients with personality disorder (Department of
Health, 2003) could be usefully informed by research focussed on violence
and associated personality dysfunction which also begins to address
questions over the viability of engaging such individuals in voluntary
psychological therapy.
The following definitions have been put forward for aggression and
violence: [A]ggression refers to the intention to hurt or gain advantage over
other people, without necessarily involving physical injury; violence involves
the use of strong physical force against another person, sometimes impelled
by aggressive motivation (Howells & Hollin, 1989, p. 4). Parrott and
Giancola (2007) have developed a taxonomy of aggression dierentiating
several dimensions including direct or indirect aggression, to person or
object. Parrott and Giancola note that the target should be motivated to
avoid violence (excluding, for example, consensual sadomasochism).
Clearly, aggression does not necessarily lead to violence; it can be verbal
or physical. Indeed, aggression can be expressed without direct physical
contact between aggressor and victim (e.g., by telephone), and it could be
argued that there are even non-contact violent crimes such as threats to kill
or conspiracy to murder.
Bush (1995) asserts that criminal violence is not associated with any
single disease or disorder. It does not have a single cause and it cannot be
cured with any single treatment (p. 139). However, Bush puts forward three
generalizations which oer a framework within which to consider violence.
First, violence is not an isolated and distinct form of criminal behaviour;
diversity rather than specialization is the hallmark of criminals. Second,
violence is a learned behaviour and can be understood as a coping
mechanism that is often the only response people feel they have. Third,
patterns of violence and criminal behaviour are embedded in habits of
thinking, which may be erroneous belief systems that maintain the
behaviour when they go unchallenged.
The Journal of Forensic Psychiatry & Psychology 3

A cognitive focus on violence


Historically, psychological approaches to violence have tended to focus on
the attributes of violent individuals who engage in angry or hostile
aggression rather than instrumental aggression (Howells & Hollin, 1989).
The particular emphasis in recent decades has been on anger management as
the best available treatment approach, with growing empirical evidence in
support (Beck & Fernandez, 1998) which is nonetheless limited by the
quality of studies (Walker & Bright, in press).
Anger as a risk factor for violence has been researched and described by
Novaco, most notably in Monahan and Steadman (1994). Novaco and
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Welsh (1989) maintain: [W]hile anger is neither necessary nor sucient for
aggression to occur, it is nonetheless a central activator of both individual
and collective violence and is significantly involved in a range of mental
health disturbances (p. 39). They go on to propose that problems with anger
control and regulation are important risk factors in relation to harming
others. One important aspect of anger in relation to violent behaviour is that
it may be identified and described by the aggressor and is seen as a central
factor in the loss of control and subsequent violence. Furthermore, there are
now recognized ways of treating the problem through anger management
training (Williams & Barlow, 1998). However, a more detailed evaluation of
the literature suggests that anger management may be less potent as an
intervention for violence than it initially appears (Walker & Bright, 2008).
Violent beliefs and thinking (cognitions) have received less research
attention than cognitions in sex oenders (Abel, Gore, Holland, & Camp,
1989; Blumenthal, Gudjonsson, & Burns, 1999; Hayashino, Wurtele, &
Klebe, 1995; McGrath, Cann, & Konopasky, 1998; Vanhouche & Vertom-
men, 1999; Ward et al., 1997). The majority of the studies of violence and
aggression have focussed on risk factors (Bjrkly, 1997; Monahan, 1984,
1997; Monahan & Steadman, 1994) and anger (Greene, Coles, & Johnson,
1994; Novaco, 1997; Novaco & Welsh, 1989). The factors relevant to
violence are exceptionally complex (Monahan & Steadman, 1994), and
include multiple causal and exacerbating factors, ameliorating or protective
factors, individual dierences, and the influences of other unique factors
such as the environment and the victim at the time the violent act took place.
This problem has lead McGuire and Priestly (1995) to suggest that the whole
focus of risk assessment needs to shift towards understanding what makes a
particular person commit a particular violent act at a particular time.
In the cognitive behavioural literature this approach formulating the
antecedent causes and maintaining factors in a persons thinking and
behaviour has long been considered important. Cognitive models (e.g.,
Beck, Freeman et al., 1990; Beck, Freeman, Davis et al., 2004) suggest that
adaptive and maladaptive deeply held beliefs (schemas) are laid down in
4 J.S. Walker and J.A. Bright
childhood through various processes including modelling, the result of
trauma and abuse, and other positive and negative significant events or
patterns in upbringing. The second layer to the model comprises rules
(dysfunctional assumptions) that govern when and how strongly these
deeply held beliefs are activated. The triggering of these rules results from
everyday experiences which may link to automatic thoughts that are fleeting,
often aect-laden, and relate strongly to the corresponding rule and schema.
Dierent constellations of beliefs, rules, and thoughts together with their
related symptoms correspond to dierent disorders such as depression
and anxiety disorders. These negative constellations can be enduring and
damaging, producing chronic self-defeating behaviour patterns such as
Downloaded by [University of Nottingham] at 08:00 07 November 2015

antisocial and violent behaviour. The behavioural problems of these patients


are the result of them trying to cope with a pathological belief system that
they may be only partially aware of, and unable to describe.

The development of psychological theories of violence


Historically, theories of violence can be traced back to religious teachings
and texts on human nature and the nature of evil and destructiveness, which
locate the propensity for violence within human nature rather than view it as
a product of experience or circumstance.
Twentieth-century psychological theorists regarded aggression as rooted
in instinct for example, Freuds theory of the death instinct as the root of
destructiveness (Freud, 1920) or Lorenzs (1966) understanding of survival-
based fighting instinct. The death instinct is managed by psychological
defences which keep aggression in check, but when defences are over-used or
under-developed, forms of aggression and violence may result. The idea of
instinctive aggression was similar: taking this view, aggression may have
been adaptive at an earlier evolutionary stage but ultimately becomes pent-
up aggression requiring a target to become the object of discharged
aggression. This innate approach to aggression dismisses the more
developmental pathways to aggression which have received more attention
and presented more compelling evidence in recent years (see de Zulueta,
2006, for a review).
The development of drive theory by theorists such as Dollard, Doob,
Miller, Mowrer, and Sears (1939) and Berkowitz (1974) lead to an
understanding of aggression as a response to frustration (reactive violence),
and began the emphasis on anger as a risk factor for violence emphasized by
Novaco (1994) and others during the latter part of the twentieth century.
Since anger management came to dominate clinical and forensic settings,
there does not seem to have been another landmark clinical theory of adult
violence, and perhaps because of this prison treatment groups for violent
criminals, particularly in the UK, continue to be based on anger manage-
ment principles (Hollin & Palmer, 2006). Other theories have been
The Journal of Forensic Psychiatry & Psychology 5
developed, but none have been as widely accepted or used in forensic
contexts as anger management.
For example, Becks work with feuding couples (Beck, 1999) suggests an
algorithm for anger: the perpetrator feels diminished or oended; the
perpetrator believes that the oence was unjustified; the oence was seen as
intentional; the oensive act/comment was seen as characteristic of that
person (fundamental attribution error); and the oending person is
deserving of punishment (thus a blind person or a child is less deserving
of punishment for bumping into you than an ordinary adult). The
importance of feelings and beliefs is typical of a Beckian cognitive model,
and is not only consistent with but adds to Novacos model of anger, which
Downloaded by [University of Nottingham] at 08:00 07 November 2015

emphasizes the importance of perceptions and information processing. Less


fully explored are the aetiology and mechanism for developing the extreme
sensitivity to being oended that characterizes violent individuals and which
may therefore be relevant to assessment and treatment.
An alternative to the broad approach of anger management investigates
the innate predisposition to violence in certain repeatedly violent oenders.
Problems with emotion attributions (particularly guilt) and associated
problems with inhibiting violent behaviour have been suggested by Blair
et al. (1995) to explain increased levels of violent aggression by psychopaths
in particular. Blairs (1995) violence inhibition mechanism model is based on
animal studies of submissive behaviour and associated decreases in
aggression by the attacker. The model has been developed further with
small numbers of psychopaths, and may explain the tendency of
psychopaths not to respond to the pain of others with guilt and empathy
in the way normal controls would (Blair et al., 1995). Interestingly, Blair
et al. (1995) found no dierence between psychopaths responses to
embarrassment and those of normal controls. However, it is not possible
to draw firm conclusions about the absence of a relationship from such a
small study (n 25). In contrast, in a similar small-scale study of
psychopaths, Morrison and Gilbert (2001) found extremely high levels of
internalized shame and sensitivity to humiliation in psychopaths, coupled
with a tendency to assume dominance in response. A further study by
Gilbert and Miles (2000) suggested that externalizing or blaming others
mitigated the eect of shame. Therefore, it is possible that what underpins
aggression, even in extreme violence-prone individuals, is the absence or
disabling of a violence inhibition mechanism (associated with guilt and
empathy); in turn, the disabling of the violence inhibition mechanism may
be triggered by shame and humiliation. Such a proposal is described from a
psychodynamic perspective by Gilligan (1996), and a complementary
trauma/attachment-based perspective is described by de Zulueta (2006).
Authors such as de Zulueta (2006) and Gilligan (1996) have proposed
that childhood development, in particular early experiences of trauma,
neglect, and shaming, can result in specific constellations of thoughts,
6 J.S. Walker and J.A. Bright
emotions, memories, and reactions that increase propensity to violence.
Humiliation and shame are seen as core triggers and vulnerabilities for
violence, and as such are focussed on by these authors in more depth than
generally oered in the cognitive behavioural literature.

Humiliation, shame, and embarrassment


Despite the attention to cognitions within anger management approaches,
the main focus appears to be on triggers, automatic thoughts, and
interpretation biases rather than on deeper (in particular self-referent)
cognitions like core beliefs. The issue of the core beliefs underpinning
Downloaded by [University of Nottingham] at 08:00 07 November 2015

aggression and violence does not seem to have been adequately addressed by
contemporary authors in the same way that the core beliefs of worthlessness
in depression and vulnerability in anxiety are focussed upon in general adult
psychological formulations.
Psychodynamic approaches place deeply held (unconscious) ideas at the
core of formulation and treatment, and one of the most influential and
respected writers on this topic in recent years (Gilligan, 1996) has presented
a compelling psychodynamic formulation for violence which puts shame,
poor self-esteem, and humiliation at the centre of formulating violence-
proneness. Gilligan (1996) starts with the proposition that psychoanalyti-
cally, all behaviour, including violent behaviour, whether it is labelled as
bad or mad, is psychologically meaningful (p. 9) and therefore can be
understood. He goes on to suggest that violence is a means to an end, used
to attain justice by punishing those whom they feel have punished them,
unjustly (p. 18); thus, violence is seen by the perpetrator as revenge for
being wronged unjustly by the victim. What constitutes the wrong in
Gilligans view is usually a personally meaningful insult of some kind that
causes an overwhelming sense of shame. The violent person cannot cope
with this shame due to a lack of self-love, self-esteem, or a healthy sense of
pride which would enable them to survive the inevitable rejections and
humiliations which even the most fortunate people cannot avoid (p. 47).
When this happens to someone lacking the non-violent skills to restore their
damaged sense of self-esteem, violence becomes a way of restoring that
pride. Gilligan adds that these preconditions dramatically increase the
likelihood of violent impulses being acted upon, particularly when the
feelings that would normally inhibit the acting out of those feelings, such as
love and/or guilt, are absent (p. 114).
The characteristics of a potentially violent confrontation can gradually
be pieced together. Information processing is important in that it strongly
influences how the environment is perceived; thus when an insult, rejection,
humiliation, or verbal attack is experienced as deeply shaming, and
inhibition of guilt and empathy are absent (or disabled at the crucial
moment in some way), violence becomes a way of restoring the balance for
The Journal of Forensic Psychiatry & Psychology 7
the violent person pride for themselves and punishment for the victim.
Cognitive behavioural therapy (CBT) and anger management approaches
have traditionally dealt with triggers, information processing, and elevated
arousal, without necessarily working at the level of personal meaning,
humiliation, and damaged pride (or low self-esteem).
The definition and dierentiation of shame, humiliation, and guilt are
taken up by Gilbert (1998). Internal shame is defined as shame which relates
to negative self-evaluations; external shame relates to being scorned,
ridiculed, judged, or disgraced in front of others. Shame can have numerous
foci including ones body, feelings, failures, and belonging to a discredited
social group. From an evolutionary perspective, Gilbert relates shame to
Downloaded by [University of Nottingham] at 08:00 07 November 2015

psychobiological and behavioural mechanisms which underpin submissive


behaviour. Shame can be seen as a more global self-referent evaluation than
guilt, which is seen to be related to a particular behaviour or transgression.
Thus a deep sense of shame may be avoided or defended against in such a
way that minor humiliating experiences cannot be easily integrated or
buered by a stable and positive inner sense of the self. These experiences
can only be dealt with by ejecting the unpleasant feelings into the person
causing the humiliation perhaps a subconscious goal achieved by the very
conscious and physical means of (violent) attack. Embarrassment is
described by Gilbert (1998) as milder than shame, but still linked to self-
focussed evaluations and aects which are negative and self-conscious.
Humiliation is also seen as similar to shame, but with important dierences:
one is humiliated by another and put in a relatively powerless position,
accompanied by feelings of injustice, blaming of both others and the self,
and a sense of injury or harm to the self. Furthermore, Gilbert (1998) notes:
[A]nother aspect that marks the humiliated person can be their feelings of
hatred, desires to retaliate, and ease in becoming aggressive (p. 260).
Morrison and Gilbert (2001) report one of the few studies of shame and
anger in a forensic setting. They dierentiated between primary and
secondary psychopaths (Blackburn, 1998; Blackburn & Lee-Evans, 1985),
and found that primary psychopaths showed unexpectedly high internalized
shame but assumed a dominant and threatening approach, whereas
secondary psychopaths showed even higher internalized shame and more
negative self-evaluations, and (despite seeking dominance) were more
sensitive to attacks. Primary psychopaths also rated themselves as
significantly less angry generally, less angry with others, and less often put
down. Morrison and Gilbert (2001) noted that the important distinguishing
factor for psychopaths was not the general level of anger, but the speed of
arousal to aggression from anger and the lack of violence-inhibiting factors
such as empathy and self-consistency. Such findings add further to the idea
that self-related factors such as shame may be important in the under-
standing of extremely violent individuals, and perhaps even more important
than the assessment and treatment of anger.
8 J.S. Walker and J.A. Bright
From a more psychosocial perspective, Baumeister (1997) has been a
leading proponent of the view that high rather than low self-esteem is
associated with violence. He suggests that there are plenty of reasons to be
violent in society, but that what is surprising is that there is not more
violence. He proposes that high self-esteem is more vulnerable to oence
than low self-esteem, and that for some individuals with high self-esteem
who feel oended, violence is justified, guilt and empathy suppressed
(preventing uncomfortable cognitive dissonance), and the violent individual
is thus less protected (as normal individuals would be) from future violent
responses.
On the other hand, normal individuals appear to desist from violence,
Downloaded by [University of Nottingham] at 08:00 07 November 2015

and inhibit this response or discount it as a response option. The current


authors propose that with healthy self-esteem, inhibiting factors including
guilt and empathy are not suppressed and can be allowed to aect
behaviour. If one has robust and positive self-esteem, insults are less painful
and so there is less pressure (rage) to respond actively. However, the violent
individual has a vulnerable and negative sense of self which is not robust to
insults, reacts with rage and violence, and loses the normal self-control
mechanisms (Gilligan, 1996). For non-violent individuals, fear of physical
injury and the moral imperative not to behave badly or to look bad usually
override any anger-driven desire for physical vengeance. However, for
violent individuals, there is little or no fear of physical injury (learned
numbness to the shock of physical confrontation and habituation to fear
through repeated fights). Any fear is overridden by rage, and aversion to
looking bad by backing down or running away is the most pressing factor.
Salmivalli (2001) suggests that false inflated self-esteem and paranoia act
as a defence against low self-esteem, and that recognition of this type of
defence and self-view in violent oenders is central to understanding how a
person who appears to think well of themselves can also be extremely
sensitive, anxious and threatened. It is this vulnerability which can
ultimately be dangerous when coupled with anger, a confrontational style
(perhaps driven by paranoia), and machismo (Walker, 2005). Machismo is a
way of dealing with potentially embarrassing social threatening situations
through aggression and violence; this cognitive style has been found to be
strongly associated with violence for males (Walker, 2005; Walker &
Gudjonsson, 2006).

Self-esteem and violence: a systematic review


In order to gather some basic but broad data on the question of how self-
esteem relates to violence, a systematic review of the literature was
undertaken. Where a relationship was found, the current authors were
particularly interested in whether high or low self-esteem related to violence.
Answering this question from published data will go some way towards
The Journal of Forensic Psychiatry & Psychology 9
supporting the view that self-esteem is an important factor to consider when
working with violent individuals, and moreover the results of the review will
provide a data-based context in which to evaluate existing theories.
Although relatively untested empirically, the theories presented so far
are regarded as clinically relevant. They fall broadly into two opposite
positions: either high or low self-esteem predicts violence. Gilligans and
Gilberts work come from two dierent approaches in psychology, one more
psychodynamic and the other evolutionary and more cognitive behavioural.
However, they are similar in terms of the importance and risk of aggression
associated with humiliation and the importance of negative self-evaluations
which, when consistent and stable over time, begin to appear similar to or
Downloaded by [University of Nottingham] at 08:00 07 November 2015

overlapping with low self-esteem. Conversely, both Beck (1999) and


Baumeister (1997) propose that high or inflated self-esteem, rather than
low self-esteem, increases the risk of violence. Baumeister concluded that
high self-esteem presents less of a risk if it is genuine and reflected in the
evidence around the individual. When it is not genuine and not reflected by
evidence that the individual is successful that is, the individual has an
inflated sense of self-esteem or grandiosity then threats to self-esteem are
more seriously experienced and violence is a way of communicating or
asserting superiority.
The present authors suggest that if self-esteem is inflated it is in reality
covering low self-esteem, and that the inflation of self-esteem is part of the
macho cover-up of embarrassment. Why would the violent person be
vulnerable to taking serious (overstated) oence if they were genuinely
secure in themselves with high self-esteem? Salmivallis (2001) proposal that
self-esteem is falsely inflated seems to support the idea that low self-esteem is
the problem, even though violent oenders appear to have high self-esteem
and may come across as arrogant or grandiose. A further finding regarding
machismo is that it has been related to self-reported violent behaviour
(Walker, 2005) machismo as a form of false inflated self-esteem and a
tendency to assert oneself is clearly associated with violence, but is more
likely to be a cover-up of low self-esteem (core beliefs of vulnerability and
worthlessness).
Fennel (1997) defines self-esteem as a schema in the sense recognized by
Bartlett (1932) in relation to memory as a generic cognitive representation
(in this case of the self), based on experience, and influencing how
information is subsequently processed (p. 2). In low self-esteem, this self-
representation is by definition negative or derogatory, persistent across time
and situations, habitual, and automatic. Low self-esteem is perpetuated
because negative predictions (of failure) result in avoidance which, in turn,
perpetuates low self-esteem schemas.
Regarding the structure of self-esteem, Dick and Bronson (2005) note
that there are both cognitive and aective elements to self-esteem
cognitive in the sense that we have thoughts about and evaluations of
10 J.S. Walker and J.A. Bright
ourselves and aective in the sense that they may be positive or negative.
They also note that a number of authors have identified the dierence
between specific self-esteem (relating to specific aspects of the self) and
global self-esteem (overall judgement), and further describe competence-
based self-esteem (related to what one is good at) and global judgement of
self-worth. This means that self-esteem is multifaceted and dimensional, and
that while one element of self-esteem may be high (I see myself as a good
driver) another may be low (But I am a useless parent). All of these aspects
may dierentially influence ones global self-judgement and may or may not
be appropriately and reliably weighted and measured by self-esteem scales.
This problem of accurate measurement has important implications for any
Downloaded by [University of Nottingham] at 08:00 07 November 2015

research into the relationship between self-esteem and other variables such
as aggression and violence.
Individually, self-esteem and violence are two commonly researched
topics in the psychological literature, but perhaps less frequently researched
is the relationship between the two. In order to answer the question of how
self-esteem relates to aggression and violence, a systematic review of the
1986 2006 literature was undertaken. In all, 365 articles were identified
from a search of the Psychinfo, Embase, and Medline databases using self-
esteem combined with aggression, violence, or anger as search terms (a
recent extensive systematic review in forensic psychiatry revealed that
almost 100% of published articles can be located via these three databases
without complicating or extending the search further; see Amos et al., in
press). The search was limited to original journal articles that involved some
form of comparison or controlled study. The titles of these 365 articles were
reviewed and the results narrowed to 92 articles; the 92 abstracts were
reviewed for relevance to the research question. The 28 papers left following
this process were retrieved in full (from the original authors where possible),
which resulted in the receipt of an additional eight papers from the
originating authors (two of which were in press and thus excluded). Three
additional papers were located through cross-referencing. This left 37
articles to be reviewed in full. Of these, seven were excluded because self-
esteem was not an independent variable; four did not have violence as a
dependent variable; and in five the age (or mean age) of participants was
below 13 years. A further two articles were excluded for being a case study
series and a small pilot study respectively. This left 19 articles where
inclusion criteria were met (published journal studies, with adolescents or
adults, reporting a relationship between self-esteem and violence/
aggression).
The results are reported in Table 1 in order of descending sample size.
Sample size was chosen as one indication of quality, but it is recognized that
there are numerous other indices of quality, and the results could have been
presented in alphabetical or date order; sample size was chosen as a simple
way to convey at least some information about the relationship of one study
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Table 1. Results of systematic review of self-esteem and violence/aggression (19862006).

Study Sample Design Conclusions


Paulhus, Robins, 1) 4057 students, USA, 1) Cross-sectional Narcissism correlated positively with antisocial
Trzesniewski, and 66% female questionnaire survey behaviour in all three samples; in contrast
Tracy (2004) 2) 301 students, Canada, 2) Cross-sectional self-esteem showed negative or slightly
62% female questionnaire survey positive correlations. Narcissism and
3) 232 students, Canada, 3) Cross-sectional self-esteem were correlated in all three

The Journal of Forensic Psychiatry & Psychology


64% female questionnaire survey samples. Using regressions, the negative
including informant eect of self-esteem on antisocial behaviour
Ages not stated, all ratings (friends/family) was significantly more negative when
undergraduates narcissism was included in the equation
than when it was not; thus self-esteem is
negatively related to antisocial behaviour
(whether judged by self or others) when
narcissism is controlled.
Donnellan, 1) 292 students, USA, 1) Cross-sectional study A robust relationship was found between low
Trzesniewski, 55% female, mean age using questionnaires, self-esteem and aggression, antisocial
Robbins, Mott, 12.7 years parent/teacher ratings, behaviour, and delinquency. This finding
and Caspi (2005) 2) Dunedin (NZ) cohort and achievement test held for measures based on self-report,
study, 812 at age 11, scores teachers ratings, and parents ratings, and
736 at age 13, 48% 2) Longitudinal cohort for participants from dierent nationalities
female study using data from (United States and New Zealand) and age
3) 3143 students, USA, birth to 15 years groups (adolescents and college students).
68.3% female, mean (mainly 11 13 years) Moreover, this relation held both cross-
age 19.6 years 3) Cross-sectional sectionally and longitudinally, and after
questionnaire survey of controlling for confounding variables such
narcissism, self-esteem, as supportive parenting, parent child and
and aggression peer relationships, achievement test scores,
socioeconomic status, and IQ. The eect of

11
(continued)
Downloaded by [University of Nottingham] at 08:00 07 November 2015

12
Table 1. (Continued).

J.S. Walker and J.A. Bright


Study Sample Design Conclusions
self-esteem on aggression was independent
of narcissism (which positively correlated
with aggression).
Webster (2006) 1781 undergraduate Cross-sectional Self-esteem was inversely associated with
students, USA, 59.6% questionnaire survey aggression regardless of whether narcissism
female. Age not stated was controlled. Self-esteem (negatively) and
narcissism (positively) were independently
associated with aggression when controlling
for the other (eects were mutually
suppressive).
Trzesniewski et al. Dunedin (NZ) cohort study: Longitudinal cohort study Low adolescent self-esteem predicted adult
(2006) 1037 from age of 3 years, using data from birth to criminal convictions and showed slightly
52% male, 980 at age 26 26 years at 2 3-year stronger prediction of adult violent
years intervals convictions. Self-esteem predicted a number
of adult adjustment problems.
Von Collani and 1) 288 female and 129 1) Cross-sectional Self-related constructs (low self-esteem, poor
Werner (2003) male (internet sample) questionnaire survey self-concept clarity, high narcissism, and
students and employed 2) Cross-sectional motivational orientation towards reward/
people, Germany, mean questionnaire survey punishment) were shown to predict almost
age 28.6 years 3) Cross-sectional half the variance in self-reported aggression
2) 116 female and 25 male questionnaire survey for males and females. Gender dierences
students, Germany, were found in the relative importance of
mean age 22.8 years self-constructs in predicting global

(continued)
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Table 1. (Continued).
Study Sample Design Conclusions
3) 217 female and 183 aggression scores for males and females.
male students and their Self-esteem was the strongest predictor for
acquaintances, females and social desirability for males.
Germany, mean age

The Journal of Forensic Psychiatry & Psychology


34.7 years
Lopez, Olaizola, 843 adolescents, 48% male, Cross-sectional Aggressive rejected (socially isolated)
Ferrer, and Ochoa Spain, mean age 13.7 questionnaire survey students demonstrated lower levels of family
(2006) years including informant self-esteem and academic self-esteem,
questionnaires (teacher negative attitudes towards school, poorer
and peer ratings) relationships with teachers, and more
academic diculties than the non-
aggressive rejected subgroup.
Pfleiger and Vazsonyi 404 male and 403 female Cross-sectional Low self-esteem mediated the eect of
(2006) students, USA, mean age questionnaire survey maternal closeness and support on the
16.4 years perpetration of dating violence.
Walker and Normal male (480) and Cross-sectional Self-esteem was not related to self-reported
Gudjonsson (2006) female (305) adolescents, questionnaire survey violence or any other form of self-reported
UK, mean age 17.2 years delinquency in either male or female
adolescents. The best predictors of violence
were machismo for males and acceptance
of violence for females.
Gillespie (2005) 391 white and 253 black Cross-sectional High self-esteem predicted violent behaviour in
male prison inmates, questionnaire survey prison for white but not for black
USA, mean age 35.8 prisoners indicating that the relationship
years between high self-esteem and violence may
be race-specific.

(continued)

13
Downloaded by [University of Nottingham] at 08:00 07 November 2015

14
Table 1. (Continued).
Study Sample Design Conclusions

J.S. Walker and J.A. Bright


Bushman and 1) 130 male and 130 Behavioural experiment In both studies self-esteem proved irrelevant to
Baumeister (1998) female college students, (with aggressive aggression. The combination of narcissism
USA, age not stated behaviour as outcome and insult led to exceptionally high levels of
2) 140 male and 140 measure) incorporating aggression toward the source of the insult.
female college students, cross-sectional These findings contradict the view that low
USA, age not stated questionnaire data. self-esteem causes aggression and point
instead toward threatened egotism as an
important cause.
DZurilla, Chang, Male (72) and female (145) Cross-sectional Low self-esteem was related to anger and
and Sanna (2003) students, USA, mean age questionnaire survey, hostility, but these links were mediated by
18.9 years aggression problem-solving (which was robustly related
questionnaire given 6 7 to aggression). Negative problem orientation
weeks after self-esteem fully mediated the link between self-esteem
measure and anger and partially mediated the link
between self-esteem and hostility.
Murphy, Stosny, and Partner-violent men, USA, Pre post intervention Both samples reported significant reductions in
Morrel (2005) 61 in Group 1 (mean age comparison study relationship violence perpetration and
36.6 years), 107 in Group Group 1: behavioural significant increases in self-esteem from
2 (mean age 40.1 years) intervention; Group 2: pre- to post-treatment. In both samples,
programme to change in self-esteem was inversely
enhance compassion for correlated with change in physical
self/others aggression.
Kirkpatrick, Waugh, 1) 40 male and 48 female 1) Cross-sectional In Study 1, self-perceived superiority was
Valencia, and college students, USA, experimental design positively related to aggression, whereas
Webster (2002) age not stated incorporating perceived social inclusion was inversely
questionnaire survey related to aggression. In Study 2,

(continued)
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Table 1. (Continued).
Study Sample Design Conclusions
2) 35 male and 39 female 2) Cross-sectional self-perceived mate value emerged as a
college students, USA, experimental design (positive) predictor of aggression. Global

The Journal of Forensic Psychiatry & Psychology


age not stated incorporating self-esteem failed to contribute to the
questionnaire survey prediction of aggression in either
experiment.
Dick (2004) 95 (intimate partner Cross-sectional postal High self-esteem correlated with lower levels of
violence) and 50 (non- questionnaire survey violence and this relationship held true even
violent) men, USA, mean when other variables were held constant or
age 35.7 years partialled out.
Perez, Vohs, and 53 (male) and 87 (female) Cross-sectional Self-esteem was related to aggression in a
Joiner (2005) undergraduate students, questionnaire survey curvilinear fashion very high and very
USA, 18 20 years including peer low self-esteem individuals were more
informant rating (by likely to report physical aggression than
college room-mate) those with moderate self-esteem. This
relationship was partly qualified through
interpersonal context those whose
self-evaluation was discrepant with their
room-mates (higher or lower) reported
higher aggression than those whose
evaluations corresponded (whether
their own self-esteem was high or low).

(continued)

15
Downloaded by [University of Nottingham] at 08:00 07 November 2015

16
Table 1. (Continued).

J.S. Walker and J.A. Bright


Study Sample Design Conclusions
Parker, Morton, 132 adjudicated Prospective cohort study Prior non-violent oences predicted non-
Lingefelt, and (delinquent) male including questionnaires/ violent oences at follow-up. However,
Johnson (2005) adolescents, USA assessments at baseline violent oences at follow-up were better
Mean age 13.5 years with 3-year follow-up of predicted by personality: low self-esteem,
criminal records shallow changing emotions, poor
emotion/impulse/anger control, feeling
persecuted, and high anxiety.
Cale and Lilienfeld 98 male prison inmates, Cross-sectional No relationship was found between self-esteem
(2006) USA, mean age 23.7 questionnaire survey scores and measures of aggression in
years including behaviour response to ego threats. Self-esteem scores
ratings (from prison correlated negligibly with self-reported
record) and informant anger in response to negative feedback and
ratings (from prison with independent reports of verbal and
ocers and counsellors) physical aggression in response to ego
threats. Moreover, the correlation between
self-esteem scores and perceived ego threat
was negligible. There was partial evidence
that psychopathic individuals tended to
respond aggressively when confronted with
an ego threat, and they exhibited this
pattern more strongly than narcissistic
individuals (narcissism also weakly
predicted aggression).

(continued)
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Table 1. (Continued).

The Journal of Forensic Psychiatry & Psychology


Study Sample Design Conclusions
Webster and 79 college students, USA, Behavioural experiment Self-assessed mate value (as a romantic/
Kirkpatrick (2006) age not stated (with aggressive sexual partner) was a reliable positive
behaviour as outcome predictor of self-reported hostility/
measure) incorporating aggression. Experimental results suggested
cross-sectional that mate value (positively) and self-esteem
questionnaire data (negatively) predicted aggression for those
receiving negative feedback in the
experiment.
Schwartz, Waldo, 74 men attending domestic Cross-sectional Regression analyses suggested that self-esteem
and Daniel (2005) violence programme, questionnaire survey did not significantly predict physical partner
USA, mean age 31.8 abuse, but that low self-esteem significantly
years predicted use of intimidation and threats in
combination with high emotional
expression.

17
18 J.S. Walker and J.A. Bright
to the next, but this order does not indicate decreasing quality. Indeed many
of the larger studies were conducted with students, who are not likely to be
comparable to the clinical populations found in forensic settings.
The majority of studies (12 of 19) suggested that low self-esteem rather
than high self-esteem was associated with violence. This was true of the
seven largest studies, many of which were conducted with American
undergraduate students; however, this group also included the only two
longitudinal cohort studies. Five studies found no association between self-
esteem and violence, the largest of these with a total sample of 785 (Walker
& Gudjonsson, 2006). Only one study indicated that high self-esteem was
related to violence, in a sample of 391 white American prison inmates
Downloaded by [University of Nottingham] at 08:00 07 November 2015

(Gillespie, 2005). Finally, one study reported a curvilinear relationship


between self-esteem and violence, such that both high and low self-esteem
were associated with violence (Perez, Vohs, & Joiner, 2005); however, it was
suggested that this relationship was mitigated by interpersonal context, and
that aggression is better predicted by the discrepancy between the
individuals self-esteem and others view (esteem) of them. This was one
of the smaller studies, with a sample of American undergraduates. When
forensic samples are considered, six studies were available and the sample
sizes are typically small (n 644, 168, 145, 132, 98, and 74). Four of these
suggest that low self-esteem relates to violence, with one (Gillespie, 2005)
finding the reverse, but only for white prisoners (n 391), and one study
found no relationship (Cale & Lilienfeld, 2006) but the sample size was
probably too small (n 98) to avoid a Type 2 error reliably.
Inevitably, a systematic review generates as many questions as it
answers. Despite the finding that low self-esteem relates to violence, when
sample sizes are considered this finding is stronger in the student population
than in the normal and forensic populations. The majority of studies used
the most well-known measure of self-esteem, the Rosenberg Self-Esteem
Inventory (Rosenberg, 1965); however, this measure has high face validity,
no lie scale, consists of 10 items, and is not multidimensional in the way that
self-esteem is likely to be (even if referring to a simple delineation between
competence-based self-esteem and global self-worth). The diculty of
evaluating self-esteem accurately is raised by many of the authors, and the
idea that self-esteem may be falsely inflated adds a further bias to threaten
reliability. Furthermore, none of the authors discuss the ways in which
people might inflate their own self-esteem, save occasional references to
narcissism. The question of the false inflation of self-esteem is more
coherently discussed by Salmivalli (2001), but the present authors contend
that there may be both conscious and unconscious inflation of self-esteem
conscious inflation being commonly recognized as arrogance, and uncon-
scious inflation being more similar to the narcissistic defence (defence
mechanisms being unconscious ways of minimizing conscious or psychic
disturbance).
The Journal of Forensic Psychiatry & Psychology 19
In general terms, when the risks of combining dierent studies are
acknowledged, the most robust finding is: low self-esteem relates to violence,
but this relationship may be less robust in forensic populations. Questions
remain around the inflation of self-esteem and the reliability of self-esteem
assessment, particularly in oenders. Furthermore, there is the question of
what psychic disturbance is minimized by narcissism perhaps it is the
unbearable truth of ones own negative self-view but could also be the
realization that others views of one may also be rather negative. From an
assessment point of view, the finding that the discrepancy between ones
view of oneself and the others view of one predicts violence is useful.
Concerning the inflation of self-esteem, it was found in six papers (with a
Downloaded by [University of Nottingham] at 08:00 07 November 2015

variety of sample sizes) that narcissism predicts aggression.

A violent cognitive network


In an eort to develop a more cognitive approach to formulating violence
with a stronger emphasis on self-related concepts such as low self-esteem
and humiliation, the current authors have developed a model that aims to
integrate and explain some of the theories and findings presented thus far.
Although it is referred to as a model both for ease of reference and because it
is a cognitive model for a particular series of thoughts and behavioural
processes, it can equally be seen as a formulation and, more specifically, as a
framework on which to build an individual formulation for a particular
violent individual.
Before introducing the model, the present authors propose that thoughts
and feelings relating to violence can be conceptualized in terms of a violent
cognitive network, which combines a range of information processing,
interpretative, emotional, aective, self-referent, and behaviourally pro-
grammed factors, which not only interrelate but connect to a range of
thinking systems, memories, experiences, goals, and standards. One of the
central components is extreme and extensive low self-esteem, which is
covered up by machismo and false inflated self-esteem. This not only adds
energy to the system (hence powerful overwhelming emotions) but also
provides justified habitual responses which may be dangerous to others
(violence) in the name of redress, which remain inflexibly self-maintaining.
Table 2 lists the range of thoughts, feelings, and perceptions which the
current authors and others have found to be important in violence and
which would provide interesting areas of enquiry for future work.

A cognitive formulation of habitual violent behaviour


The inevitable reconfirmation of negative self-schemas and dysfunctional
assumptions is a key aspect of Fennels (1997) cognitive model of low self-
esteem, and helps in suggesting a way in which problematic beliefs and rules
20 J.S. Walker and J.A. Bright
Table 2. A violent cognitive network: cognitions and emotions in aggression and
violence.

Cognition/emotion/
thinking error Description Reference
Anger Negatively toned emotion, risk Novaco (1994)
factor for aggression and
violence
Information Issues include: attentional Williams and Barlow
processing cuing, perceptual matching, (1998), DZurilla,
attribution error, false Chang and Sanna
consensus, anchoring eects, (2003), Novaco and
cognitive scripts, Welsh (1989)
Downloaded by [University of Nottingham] at 08:00 07 November 2015

dichotomous thinking,
perfectionism, and poor
social problem-solving
Embarrassment, Intense feelings which also Gilbert (1998), Gilligan
shame, and include negative self- (1996)
humiliation evaluation and relate to
core beliefs of vulnerability
(cf. anxiety) and worthlessness
(cf. depression)
Feeling diminished The assault on self-esteem Baumeister (1997), Beck
(through some form of (1999), Gilligan (1996)
perceived threat or insult) is
experienced as diminishing,
and highly provoking if also
experienced as intentional by
someone responsible for their
actions who is deserving of
punishment
Machismo Presentational and behavioural Walker (2005), Walker
eects of the rule To be and Gudjonsson
strong and respected you must (2006)
fight results in tough, hard
exterior and behaviour;
dysfunctional assumptions or
rules mitigate the relationship
between core beliefs and
aggressive behaviour (e.g.,
To be respected you must
fight; If I let people get
the better of me and I dont
retaliate it means that I am
weak and pathetic)
Acceptance of Acceptance of violence relates to Walker (2005), Walker
violence false consensus and validates and Gudjonsson
violence as a response to the (2006)
strength of insult/humiliation
felt

(continued)
The Journal of Forensic Psychiatry & Psychology 21
Table 2. (Continued).

Cognition/emotion/
thinking error Description Reference
Empathy (lack of) Anger and humiliation cause a Gilligan (1996)
bypass in empathy because all
self-related feelings at the
point of conflict are so strong,
negative, and unpleasant
Low self-esteem Underpins susceptibility to Fennel (1997), Gilbert
(core beliefs) humiliation and relates to (1998), Salmivalli
depressive core beliefs or (2001)
worthlessness; however, such
Downloaded by [University of Nottingham] at 08:00 07 November 2015

low self-esteem can be covered


by false inflated high
self-esteem, giving the
appearance of tough
grandiosity
Vulnerability (core Lack of belief in own robustness Walker and Bright
beliefs) to cope with anxiety, put (current paper)
downs and embarrassment;
seeing self as weak and
susceptible to seeming stupid;
having to show you are tough
through how you present
yourself and behave
(machismo)
Strategy deficit Lack of a non-violent alternative Gilligan (1996), Walker
to violence, such as and Bright (current
communication, conflict paper)
resolution, or walking away,
and the use of compensatory
strategy attack physically, or
threaten physical aggression
to cause the other to back
down is reinforced due to
confidence gained through
physical respect

are reinforced and maintained even when they may seem undesirable to the
individual on face value. Furthermore, the core beliefs are maintained
whether the person resists or complies with the urge to act in a schema-
congruent way. These aspects of the self-esteem model reflect exactly the
vicious cycle of behaviour seen in violent individuals who may wish to stop
behaving violently but who seem unable to give up fighting and regard it as
the only way of dealing with certain threatening (or potentially threatening)
situations.
Violent behaviour is an attempt to protect the self from further injury
(humiliation) and thus against a perceived lowering of self-worth. By standing
22 J.S. Walker and J.A. Bright
and fighting, some level of self-esteem and self-respect is maintained, as is
some level of physical respect from the victim being fought. This physical
respect from the other is developed when it is seen that the aggressor will not
back down or run away, but will stand and fight whatever the cost to
themselves (i.e., even if the odds are not in the aggressors favour). This sense
of physical respect is increased by the victims apparent fear.
Interestingly, many patients seem able to dierentiate between the
physical respect gained by refusing to back down or run away, and causing
fear in the other person, and a more interpersonal or psychologically-based
respect gained through being a good person, having positive personal
qualities, being looked up to by others, and getting on well with people. This
Downloaded by [University of Nottingham] at 08:00 07 November 2015

interpersonal respect is what is ultimately desired by aggressive individuals,


but they lack the interpersonal, communicative, and emotional skills to
develop it. This lack of skills, in combination with dicult-to-manage anger
and embarrassment (resulting from negative self-schemas), means they settle
for second best, i.e. the physical respect seen so commonly in prisons and
other forensic settings. Physical respect is at odds with interpersonal respect
because violence and anger are personal attributes that are generally
disrespected and feared rather than looked up to. Patients often report
feeling stupid or embarrassed for letting their anger and aggression get the
better of them to result in violence and such thinking reinforces those
important underlying negative self-schemas. Physical respect seems less
valued and empty; it just does not have the same integrity and positive
eects on self-esteem as interpersonal respect. Behaving violently therefore
simply perpetuates a negative self-view, which is barely compensated for by
the temporary reward of physical respect gained by fighting.
The goal of therapy, then, is to get patients to give up their strategy of
violence (to protect self-esteem and develop physical respect) in favour of
behaving non-violently and non-aggressively with a view to developing more
enduring and self-esteem-enhancing personal respect. This feels like an
extremely dangerous and dicult endeavour to these patients, and they
often see it as impossible during the initial phase of therapy.
The authors of this paper have developed the model presented in
Figure 1 to help to formulate how many individuals respond in violent
situations. The original ideas were based on many years of reading the
psychological literature and becoming disillusioned and disappointed with
the idea that Novacos model of anger oers the most promising way
forward for the understanding and treatment of violence (Howells & Hollin,
1989). This model is also based on a decade of clinical experience in prison,
hospital, and outpatient settings, and has been refined through seminars
with colleagues at the Maudsley Hospital and other settings. One of the aims
of presenting the model is to address the issue of why so many individuals
fail to lead less violent lives despite various psychiatric and psychological
interventions (including anger management).
The Journal of Forensic Psychiatry & Psychology 23
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Figure 1. A cognitive formulation for violent incidents.

In the experience of one of the authors (JW) over 10 years of running


specialist clinics for violent patients in prison, medium security and
outpatient settings, it became clear that most violent individuals seeking
help wished they could give up some or all of their violence and saw a
24 J.S. Walker and J.A. Bright
non-violent life as desirable. Furthermore, many of them knew that they
should stop and think before acting and would like to develop non-violent
behavioural alternatives. However, non-violent alternatives either did not
come to mind, or, if they did, a violent alternative was much harder to resist
in the heat of the moment. Indeed, in line with Novaco and Welshs (1989)
description of autoscripts, many reported not remembering the thoughts
and feelings preceding the altercation.
To get round this problem, the author began to ask about near miss
scenarios when violence was averted but may have been a possibility. Such
scenarios were highly informative, and began to expose some of the
thoughts and feelings generated at high-risk times, without being clouded by
Downloaded by [University of Nottingham] at 08:00 07 November 2015

the embarrassing or highly charged violent act. Socratic dialogue was used
to uncover the meaning of behaviours (verbal, physical, facial, observed,
and imagined) at the time of the incident, and the emergent themes began to
support the old proverb that it is harder to walk away than to stand and
fight.
Many patients reported that, despite resisting the urge to fight, the
excruciating embarrassment from not asserting themselves violently would
eventually lead to an assault. Following the assault, feelings of relief and
control, and the dissipation of feelings of fear, anxiety, pain, or
embarrassment would serve to reinforce the response. On the rare occasions
that violence was avoided, patients reported feeling as if they had let
themselves down and feeling weird or uncomfortable for some time after
(sometimes up to weeks later); this would lead to tension and increase the
likelihood of other confrontations. Conversely, when violence occurred, the
relief was immediate but then eventually marred by self-recrimination
often minimized because the end justified the means. It was interesting to the
authors that non-violence was not immediately associated with pride or self-
praise, thus maintaining inner low self-esteem.
The formulation in Figure 1 was designed in line with these clinical
observations. The first stages of the formulation outline possible paths for
the development of low self-esteem, vulnerability to embarrassment, and
lack of options such that violence becomes the default option and doesnt
allow the development of alternative skills (e.g., verbal skills and self-
assertion). The traumatic roots of destructiveness are well covered by de
Zulueta (2006) and will not be described in detail here: suce it to say that
histories of abuse (particularly violent and sexual) and neglect (the absence
of a warm, compassionate, containing attachment figure) lie at the source of
many troublesome psychological sequelae, not least of all aggression and
violence. As a child, the consequence of trauma, abuse, and neglect is not to
demean ones abusers (often trusted adults), which would result in
unthinkable helplessness, disempowerment, and a recognition of the danger
of ones bad parents, but rather to look for meaning and control through
assuming responsibility and therefore concluding that one is in some way
The Journal of Forensic Psychiatry & Psychology 25
bad and therefore deserving in some way of the abusive treatment. Thus the
theoretical bottom line for the individual comprises a belief that they are
vulnerable, weak, stupid, or mad. These beliefs are well hidden, painful to
access, and often well covered with a veneer of confidence and aggression,
giving the appearance of an arrogant exterior.
However, in dicult social situations further evidence for these negative
self-beliefs becomes available in the form of fear and anxiety which on the
one hand reinforce those beliefs, and on the other hand must be got rid of at
all costs, through conversion to anger and violence. Embarrassment is
particularly crucial in dicult social situations because it combines many of
the more painful elements of fear and anxiety with powerful negative
Downloaded by [University of Nottingham] at 08:00 07 November 2015

self-evaluation (as in low self-esteem); one feels foolish and, more


importantly, made to look like a fool by someone who is then deserving
of punishment.
Certain dysfunctional assumptions dovetail well with vulnerability to the
eects of embarrassment, such as: Embarrassment is a sign of weakness if
I feel embarrassed I must show that I am strong; I must not allow people to
show me disrespect as this shows I am not worthy of respect; To be strong
you must fight; and Walking away is a sign of fear and weakness people
will then take advantage. Thus when potentially threatening situations arise
and core beliefs of stupidity and weakness are activated, there are rules in
place allowing the individual to escape and discharge the associated feelings.
The threat can be averted by attacking first. This means that the threat of
embarrassment (as well as embarrassment itself) can feel provocative and
result in predictions that embarrassment is inevitable if the situation is
allowed to progress unchecked (see right-hand side of Figure 1). The
situation thus provokes anxiety (a feeling which confirms core beliefs of
weakness and vulnerability due to violation of the rule that fear means
weakness), which can be covered up by aggression and attack (central lower
part of Figure 1).
If the individual feels embarrassed by someone (middle left-hand side of
Figure 1), this activates core beliefs of weakness and vulnerability and an
attack if the only option in mind is a fast eective response with some
desirable outcomes for the aggressor. These desirable outcomes include the
release of stress, aggression, and fear, retaliation, making the provoking
person experience some of their pain (they may learn not to do this again),
and proof that the attacker is not scared or weak or vulnerable. However,
the very fact that anxiety, fear, or embarrassment was even experienced
reinforces the related negative self-beliefs and the consequences of the attack
are not sucient to change this significantly.
If an attack is avoided somehow (dotted line pathway in Figure 1),
perhaps after work in a therapeutic setting, it could be expected that the
individual would feel pleased with this achievement and feel improved self-
esteem. However, this tends not to be the case. Rather than feeling better for
26 J.S. Walker and J.A. Bright
not attacking, the violent individual has simply subverted their own rule
system for coping with negative core beliefs of weakness and stupidity, and
these core beliefs are likely to remain active and to be reinforced through
self-critical thinking. The individual is left with feelings of foolishness and
anxiety, which serve to support the negative core beliefs and lead to the
conclusion that he/she should have attacked. Thus, whether an attack
occurs or not, violent behaviour is always reinforced as the best reaction in
terms of temporarily limiting the impact of the negative core beliefs of
weakness, stupidity, and vulnerability. These core beliefs are reinforced
whether an attack occurs or not, because their very activation increases the
discomfort caused by them. An attack briefly covers this but ultimately the
Downloaded by [University of Nottingham] at 08:00 07 November 2015

individual feels worse privately for having lost control and been aggressive.
The alternative, resisting the urge to attack, simply allows fear and self-
critical thinking to reinforce those same negative core beliefs.
This formulation, as far as the authors are aware, is the first attempt to
explain not only the mechanism for the link between self-esteem and
violence found in the literature, but also to explain, through a cognitive
behavioural approach, the links between the hypothesized core beliefs, rules,
and behaviour which seem to reinforce themselves so powerfully in violent
individuals. The importance of such a formulation is that it links with
measurable phenomena (rules as measured by the Maudsley Violence
Questionnaire; Walker, 2005; Walker & Gudjonsson, 2006) and oers
solutions for treatment thus far not generated in the literature (see Walker &
Bright, 2008).

Conclusions and implications for future work


Baumeister, Bushman, and Campbell (2000) noted: [S]urprisingly, direct
and controlled studies linking self-esteem to aggression are almost
nonexistent (p. 27). When that statement was written it was probably
true, given that all except one of the studies found in the current systematic
review were published after 2001. The only study prior to this date was in
fact by Bushman and Baumeister (1998), showing that self-esteem was
unrelated to aggression but that narcissism did predict aggression. Hence
Baumeister et al.s (2000) assertion: [T]he long-standing view that low self-
esteem causes violence has been shown to be wrong, and the opposite view
implicating high self-esteem is too simple (p. 28). Again, in 2000 this may
have been true, but the findings from the current systematic review suggest
that there is sucient evidence to conclude that low self-esteem is more
strongly associated with aggression than high self-esteem. Baumeister et al.
(2000) concluded their article with the following question: How exactly do
the shameful feelings of being criticized transform into aggressive outbursts,
and does aggression genuinely make the aggressor feel better? (p. 29). It is
hoped that the current review oers an evidence-based conclusion regarding
The Journal of Forensic Psychiatry & Psychology 27
the relationship of low self-esteem to violence, and also presents a cognitive
formulation that attempts to answer the questions raised about the
mechanisms of this relationship.
The cognitive formulation of violence presented here is based on
sensitivity to humiliation, and this relates theoretically to fragile self-
esteem that may appear high but actually forms a cover for low self-
esteem. This relationship is extremely hard to investigate because of the
problem of measuring something that people seem to need to hide hence,
perhaps, the equivocal findings relating self-esteem to aggression in the
literature (Salmivalli, 2001). Further work should investigate the dynamic
relationships between these factors in an attempt to confirm or disconfirm
Downloaded by [University of Nottingham] at 08:00 07 November 2015

the relationships. Baumeister et al. (2000) argue that self-esteem is irrele-


vant and that it is sense of superiority that should be the focus of
intervention. However, it could be argued that if superiority is a cover
for inner inferiority, then such a focus may have a compounding or
deleterious eect. We argue that a realistic and solid understanding and
evaluation of oneself, with all ones strengths and weaknesses appro-
priately appraised, is the route to decreasing anger and aggression. In our
experience, many individuals often have low esteem for their real qualities
and inflated esteem for some of their weaknesses, resulting in a fragile and
inaccurate self-concept.
Cale and Lilienfeld (2006) have suggested that interventions that focus
on cognitive appraisals of certain triggering events, such as insults and other
ego threats, as well as on aective reactions to such events, may prove
fruitful (p. 71). We argue that the interpretation of threats is closely linked
to the oenders sense of self (or self-esteem both the evaluation of the self
and the robustness of that evaluation). Thus, an individual with realistic
self-esteem evaluations which are robust rather than fragile will be less likely
to be susceptible to ego threats than someone whose inner self-esteem is low
(but who presents as arrogant) and whose sense of self is fragile and poorly
integrated. This pattern is more complex than a simple dichotomy of low
versus high self-esteem, and the development of core self-concept in
childhood and the eects of adult behaviour on self-esteem are key areas
for investigation in violent individuals. The formulation presented here
begins to describe not only how fragile false inflated self-esteem may be
developed, but also how aggressive and violent behaviour may simply
maintain this problematic set of structures which, in turn, maintain the
violence for the individual and perhaps also for their ospring.
The ideas and formulation presented here are an initial attempt by the
authors to account for some of the contradictions found in working with
violent individuals: they are arrogant and yet there is a sense of low self-
worth and vulnerability underneath; they are so tough externally yet seem
driven by fear; they feel persecuted yet demand respect; they are cold and
emotionally numb yet hypersensitive; and they seem so antisocial and
28 J.S. Walker and J.A. Bright
rejecting and yet are desperate for contact and intimacy. We do not propose
to have answered all these contradictions, but believe that the ideas
presented may begin to integrate some of the findings and proposals of other
authors into a cognitive framework that goes beyond the simplicity of
looking at interpretation and begins to consider in a more complicated way
the aetiology, maintenance, and options for intervention.
Clinically, this framework could be applied in a variety of settings as the
experience of the author JW suggests, including prison, medium security,
and community settings. It could be argued that because the focus is on
violence itself, this should not preclude many groups (except perhaps very
acutely psychotic individuals). Furthermore, perhaps the assessment of
Downloaded by [University of Nottingham] at 08:00 07 November 2015

violence that does not take account of the cognitive mechanisms proposed
here may result in important information being overlooked without which
an accurate formulation could be impossible. Walker and Bright (2008)
demonstrate how central these issues were to a patient who also seemed to
be suering both mental state and personality disturbance who sought
therapy for longstanding violent behaviour.
The authors acknowledge that there is only anecdotal evidence that a
particular set of core beliefs may combine to produce low self-esteem and
then link with a macho thinking style (designed to hide low self-esteem) to
result in violent behaviour. Some of the support for this formulation model
comes from data gathered in other studies, and the model itself has not been
empirically validated so far. Furthermore, some of the most extreme
violence (perhaps occurring during wartime or in the context of sadistic
killings and torture) probably does not conform to the model presented
here this model has been designed to help with conceptualization of more
common forms of violence in everyday situations. Such situations cannot
simply be explained by an inability to manage anger. Additional work is
required, both clinical and research, to develop our models for formulation;
the model presented is intended to encourage this process.
Future work in this area may need to begin with a review of the way self-
esteem is measured, and discuss how existing conceptual models can be
integrated into assessment. Given the importance of these issues to forensic
settings, studies will need to focus both on normal and forensic groups.
There is further work to be done to develop the best way of assessing the
complicated interaction between self-esteem, narcissism, and ego threat,
along with related concepts such as shame and humiliation. It may be that
rules such as those measured by the Maudsley Violence Questionnaire
(Walker, 2005) are easier and more reliable to assess than core self-
concepts, and that having been evaluated they are then easier to challenge
in therapy. However, we argue that, as therapy progresses, the evalua-
tion of the individuals self-concept, and its development, maintenance,
and influence on violent behaviour, should be addressed rather than
underestimated.
The Journal of Forensic Psychiatry & Psychology 29

Acknowledgements
Professor Gisli Gudjonsson at the Institute of Psychiatry for his supervision of my
research at the Maudsley Hospital.

References
Abel, G.G., Gore, D.K., Holland, C.L., & Camp, N. (1989). The measurement of
the cognitive distortions of child molesters. Annals of Sex Research, 2, 135
152.
Amos, T., Frost, J., Lewis, G., Walker, J., Payne, S., Lart, R., et al. (2007). Forensic
evidence: A systematic review of reviews in forensic mental health. London:
Downloaded by [University of Nottingham] at 08:00 07 November 2015

National Programme on Forensic Mental Health Research and Development.


Bartlett, F.C. (1932). Remembering. Cambridge, UK: Cambridge University Press.
Baumeister, R. (1997). Evil: Inside human violence and cruelty. New York: Freeman.
Baumeister, R.F., Bushman, B.J., & Campbell, W.K. (2000). Self-esteem, narcissism,
and aggression: Does violence result from low self-esteem or from threatened
egotism? Current Directions in Psychological Science, 9, 2629.
Beck, A.T. (1999). Prisoners of hate: The cognitive basis of anger, hostility and
violence. New York: HarperCollins Publishers Inc.
Beck, A.T., Freeman, A., Davis, D.D., & Associates. (1994). Cognitive therapy of
personality disorders (2nd ed.). New York: Guilford.
Beck, A.T., Freeman, A., & Associates. (2003). Cognitive therapy of personality
disorders. New York: Guilford.
Beck, R., & Fernandez, E. (1998). Cognitive-behavioural therapy in the treatment of
anger: A meta-analysis. Cognitive Therapy and Research, 22, 6374.
Berkowitz, L. (1974). Some determinants of impulsive aggression: Role of medicated
associations with reinforcement of aggression. Psychological Review, 81, 165
176.
Bjrkly, S. (1997). Clinical assessment of dangerousness in psychotic patients: Some
risk indicators and pitfalls. Aggression and Violent Behavior, 2, 167178.
Blackburn, R. (1998). Psychopathy and personality disorder: Implications for
interpersonal theory. In D. Cooke, A. Forth, J. Newman, & R. Hare (Eds.),
Psychopathy: Theory, research, and implications for society (pp. 169302).
Amsterdam: Kluwer Academic Publishers.
Blackburn, R., & Lee-Evans, M. (1985). Reactions of primary and secondary
psychopaths to anger-evoking situations. British Journal of Clinical Psychology,
24, 93100.
Blair, R.J. R. (1995). A cognitive developmental approach to morality: Investigating
the psychopath. Cognition, 57, 129.
Blair, R.J. R., Sellars, C., Strickland, I., Clark, F., Williams, A.O., Smith, M., &
Jones, L. (1995). Emotion attributions in the psychopath. Personality and
Individual Dierences, 19, 431437.
Blumenthal, S., Gudjonsson, G., & Burns, J. (1999). Cognitive distortions and blame
attribution in sex oenders against adults and children. Child Abuse and Neglect,
23, 129143.
Bush, J. (1995). Teaching self-risk management to violent oenders. In J. McGuire
(Ed.), What works: Reducing reoending Guidelines from research and practice
(pp. 139154). Chichester, UK: Wiley & Sons Ltd.
Bushman, B., & Baumeister, R. (1998). Threatened egotism, narcissism, self-esteem,
and direct and displaced aggression: Does self-love or self-hate lead to violence?
Journal of Personality and Social Psychology, 75, 219229.
30 J.S. Walker and J.A. Bright
Cale, E.M., & Lilienfeld, S.O. (2006). Psychopathy factors and risk for aggressive
behaviour: A test of the threatened egotism hypothesis. Law and Human
Behaviour, 30, 5174.
Department of Health. (2003). Personality disorder: No longer a diagnosis of
exclusion. London: National Institute for Mental Health.
de Zulueta, F. (2006). From pain to violence: The traumatic roots of destructiveness
(2nd ed.). Chichester, UK: Wiley & Sons Ltd.
Dick, G. (2004). Mens relationships with their fathers: Comparing men who batter
women with non-violent men. Journal of Emotional Abuse, 4, 6184.
Dick, G.L., & Bronson, D. (2005). Adult mens self-esteem: The relationship with the
father. Families in Society: The Journal of Contemporary Social Services, 86(4),
19.
Dollard, J., Doob, L.W., Miller, N.E., Mowrer, O.H., & Sears, R.R. (1939).
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Frustration and aggression. New Haven, CT: Yale University Press.


Donnellan, M.B., Trzesniewski, K.H., Robbins, R.W., Mott, T.E., & Caspi, A.
(2005). Low self-esteem is related to aggression, antisocial behaviour, and
delinquency. Psychological Science, 16, 328335.
DZurilla, T.J., Chang, E.C., & Sanna, L.J. (2003). Self-esteem and social problem
solving as predictors of aggression in college students. Journal of Social and
Clinical Psychology, 22, 424440.
Fennel, M.J. V. (1997). Low self-esteem: A cognitive perspective. Behavioural and
Cognitive Psychotherapy, 25, 125.
Freud, S. (1920). Beyond the pleasure principle. In The complete psychological works
of Sigmund Freud (Vol. 18, pp. 364). London: Hogarth Press.
Gilbert, P. (1998). Shame and humiliation in the treatment of complex cases. In N.
Tarrier, A. Wells, & G. Haddock (Eds.), Treating complex cases: The cognitive
behavioural therapy approach (pp. 241271). Chichester, UK: Wiley.
Gilbert, P., & Miles, J.N. V. (2000). Sensitivity to social put-down: Its relationship to
perceptions of social rank, shame, social anxiety, depression, anger and self
other blame. Personality and Individual Dierences, 29, 757774.
Gillespie, W. (2005). Racial dierences in violence and self-esteem among prison
inmates. American Journal of Criminal Justice, 29, 161185.
Gilligan, J. (1996). Violence: Reflections on our deadliest epidemic. London: Jessica
Kingsley Publishers.
Greene, A.F., Coles, C.J., & Johnson, E.H. (1994). Psychopathology and anger in
interpersonal violence oenders. Journal of Clinical Psychology, 50, 906912.
Hayashino, D.S., Wurtele, S.K., & Klebe, K.J. (1995). Child molesters: An
examination of cognitive factors. Journal of Interpersonal Violence, 10, 106116.
Hollin, C.R., & Palmer, E.J. (2006). Oending behaviour programmes: Development,
application and controversies. Chichester, UK: John Wiley & Sons, Ltd.
Howells, K., & Hollin, C.R. (1989). Clinical approaches to violence. Chichester, UK:
John Wiley and Sons, Ltd.
Kirkpatrick, L.A., Waugh, C.E., Valencia, A., & Webster, G.D. (2002). The
functional domain specificity of self-esteem and the dierential prediction of
aggression. Journal of Personality and Social Psychology, 82, 756767.
Lopez, E.E., Olaizola, J.H., Ferrer, B.M., & Ochoa, G.M. (2006). Aggressive and
non-aggressive rejected students: An analysis of their dierences. Psychology in
the Schools, 43, 385400.
Lorenz, K. (1966). On aggression. New York: Harcourt Brace World.
McGrath, M., Cann, S., & Konopasky, R. (1998). New measures of defensiveness,
empathy, and cognitive distortions for sexual oenders against children. Sexual
Abuse: Journal of Research and Treatment, 10, 2536.
The Journal of Forensic Psychiatry & Psychology 31
McGuire, J., & Priestly, P. (1995). Reviewing what works: Past, present and future.
In J. McGuire (Ed.), What works: Reducing reoending Guidelines from
research and practice (pp. 334). Chichester, UK: Wiley & Sons Ltd.
Monahan, J. (1984). The prediction of violent behaviour: Toward a second
generation of theory and policy. American Journal of Psychiatry, 141, 1015.
Monahan, J. (1997). Actuarial support for the clinical assessment of violence risk.
International Review of Psychiatry, 9, 167169.
Monahan, J., & Steadman, H.J. (1994). Violence and mental disorder. Chicago:
University of Chicago Press.
Morrison, D., & Gilbert, P. (2001). Social rank, shame and anger in primary and
secondary psychopaths. Journal of Forensic Psychiatry, 12, 330356.
Murphy, C.M., Stosny, S., & Morrel, T.M. (2005). Change in self-esteem and
physical aggression during treatment for partner violent men. Journal of Family
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Violence, 20, 201211.


Novaco, R.W. (1994). Anger as a risk factor for violence among the mentally
disordered. In J. Monahan & H.J. Steadman (Eds.), Violence and mental disorder
(pp. 2159). Chicago: University of Chicago Press.
Novaco, R.W. (1997). Remediating anger and aggression with violent oenders.
Legal and Criminological Psychology, 2, 7788.
Novaco, R.W., & Welsh, W.N. (1989). Anger disturbances: Cognitive mediation and
clinical prescriptions. In Howells & Hollin (Eds.), Clinical approaches to violence
(pp. 3960). Chichester, UK: John Wiley and Sons, Ltd.
Parker, J.S., Morton, T.L., Lingefelt, M.E., & Johnson, K.S. (2005). Predictors of
serious and violent oending by adjudicated male adolescents. North American
Journal of Psychology, 7, 407417.
Parrott, D.J., & Giancola, P.R. (2007). Addressing the criterion problem in the
assessment of aggressive behavior: Development of a new taxonomic system.
Aggression and Violent Behavior, 12, 280299.
Paulhus, D.L., Robins, R.W., Trzesniewski, K.H., & Tracy, J.L. (2004). Two
replicable suppressor situations in personality research. Multivariate Behavioral
Research, 39, 303328.
Perez, M., Vohs, K.D., & Joiner, T.E. J. (2005). Discrepancies between self- and
other-esteem as correlates of aggression. Journal of Social and Clinical
Psychology, 24, 607620.
Pfleiger, J.C., & Vazsonyi, A.T. (2006). Parenting processes and dating violence: The
mediating role of self-esteem in low- and high-SES adolescents. Journal of
Adolescence, 29, 495512.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ:
Princeton University Press.
Salmivalli, C. (2001). Feeling good about oneself, being bad to others? Remarks on
self-esteem, hostility, and aggressive behaviour. Aggression and Violent
Behaviour, 6, 375393.
Schwartz, J.P., Waldo, M., & Daniel, D. (2005). Gender-role conflict and self-esteem:
Factors associated with partner abuse in court-referred men. Psychology of Men
and Masculinity, 6, 109113.
Stanko, E., Crisp, D., Hale, C., & Lucraft, H. (1998). Counting the costs: Estimating
the impact of domestic violence in the London Borough of Hackney. Uxbridge,
UK: Centre for Criminal Justice Research, Brunel University.
Trzesniewski, K.H., Donnellan, M.B., Mott, T.E., Robbins, R.W., Poulton, R., &
Caspi, A. (2006). Low self-esteem during adolescence predicts poor health,
criminal behaviour, and limited economic prospects during adulthood. Devel-
opmental Psychology, 42, 381390.
32 J.S. Walker and J.A. Bright
Vanhouche, W., & Vertommen, H. (1999). Assessing cognitive distortions in sex
oenders: A review of commonly used versus recently developed instruments.
Psychologica Belgica, 39, 163187.
Von Collani, G., & Werner, R. (2003). Self-related and motivational constructs as
determinants of aggression: An analysis and validation of a German version of
the Buss Perry Aggression Questionnaire. Personality and Individual Dier-
ences, 38, 16311643.
Walker, J.S. (2005). The Maudsley Violence Questionnaire: Initial validation and
reliability. Personality and Individual Dierences, 38, 187201.
Walker, J.S., & Bright, J.A. (2008). Cognitive therapy for violence: Reaching the
parts that anger management doesnt reach. Journal of Forensic Psychiatry and
Psychology. DOI: 10.1080/14789940701656832.
Walker, J.S., & Gudjonsson, G.H. (2006). The Maudsley Violence Questionnaire:
Downloaded by [University of Nottingham] at 08:00 07 November 2015

Relationship to personality and self-reported oending. Personality and


Individual Dierences, 40, 795806.
Ward, T., Hudson, S.M., Johnston, L., & Marshall, W.L. (1997). Cognitive
distortions in sex oenders: An investigative review. Clinical Psychology Review,
17(5), 479507.
Webster, G.D. (2006). Low self-esteem is related to aggression, but especially when
controlling for gender: A replication and extension of Donnellan et al. (2005).
Representative Research in Social Psychology, 29, 1218.
Webster, G.D., & Kirkpatrick, L.A. (2006). Behavioural and self-reported aggression
as a function of domain-specific self-esteem. Aggressive Behaviour, 32, 1727.
Williams, E., & Barlow, R. (1998). Anger control training. Bicester, UK: Winslow
Press Limited.

Potrebbero piacerti anche