The Indirect Effects of Political Violence on Children: Does violence beget violence?

Gibson, K. (1991). The Indirect Effects of Political Violence on Children: Does violence beget violence? Paper presented at the Centre for the Study of Violence and Reconciliation, Seminar No. 4, 31 July.

 

Kerry Gibson is a Lecturer in the Child Guidance Clinic at the University of Cape Town.

Date: 31 July 1991

Venue: University of the Witwatersrand, Johannesburg, South Africa

Introduction

This paper addresses itself to the concerns that have been voiced about the effect of years of exposure to high levels of political violence on South Africa's children. In addition to evidence that exposure to violence causes short term psychological suffering (Straker 1987), there have also been concerns that it may have more fundamental consequences for their psychological development and future behaviour. In particular, one of the major concerns expressed is that children are being dehumanised by the violence around them, and in the words of Chikane (1986), "have been socialised to find violence completely acceptable and human life cheap" (p 344). We have heard for some years now of the "Lord of the Flies Children", the "Khmer Rouge mentality" and even the "Lost Generation" as the expression of similar sorts of fears (Swartz & Levett, 1989). In many ways the current wave of violence in South Africa is being understood as proof for the claim that violence begets violence. Implicit in this is the idea that children who have been victims of, or witnessed, high levels of State violence in the 1970s and 1980s now, as youth, adopt this method of resolving all political differences. In other words, in the minds of many, violence has indeed begot violence.

It is this commonsense notion that violence begets violence which is the focus of my paper. My discussion is divided essentially into two parts. In the first part, I briefly review the limited literature available on the effects of children's exposure to political violence on subsequent violent behaviour in order to establish the opposing viewpoints on this question. In the second part of my paper, I show how some of the confusion around this question arises out of the different theoretical perspectives within which it might be construed. I attempt to show that in order for the question to produce any useful understanding of the consequences of violence on children, it must be asked within a psychological theory which can capture the complexity of the relationship between an external event and an individual's response as it is constructed through both psychic and social constraints. Within this approach, I argue, there is no directly causal relationship between the external occurrence of violence and subsequent violent behaviour amongst children.

Before beginning my analysis it is perhaps necessary to briefly offer a motivation for the psychological focus of my paper – a discussion which does not attempt to offer a comprehensive analysis of political and social forces in the current violence. It has been apparent for some time that the 'Third Force' (or the State) has played a direct role in the violence and that Inkatha has, for its own political reasons, initiated a concerted programme of violent attacks (The Star, 10 December 1990). In fact the violence has been so obviously politically orchestrated that, for the most part, it has seemed patently absurd to offer explanations which explore psychological reasons for its occurrence. Even psychologists themselves have often been more comfortable with social psychological rather than psychological explanations per se (Vogelman & Simpson 1990). This tendency is particularly understandable in the light of the political conservatism with which much of traditional psychology has been identified in South Africa as well as internationally (Swartz, Gibson & Swartz 1991). To examine political violence from a psychological perspective is, in the light of these concerns, seen as detracting from its obviously political nature and thus absolving the key perpetrators of responsibility for their actions. These sorts of issues have resulted in there being little serious academic debate on this question. On the other hand, however, the psychological cliché that violence begets violence is still being acknowledged and discussed in more or less sophisticated ways by the media (The Star, 18 May 1991), concerned parents and informally amongst academic and political groupings.¹ It has, in effect, become the 'black box' of political violence, freely acknowledged as a commonsense explanation for current violence but seldom subjected to critical reflection.

The prevalence of this idea is, however, only one justification for the need to examine current violence in terms of a purely psychological perspective. As important as it is for people to know about and understand the political motivations and power struggles in the current violence, this can only ever be a part of the explanation. Having understood the broader political context and meaning of violence, one still needs to understand how these become a part of the experience of the individual and how in turn, individuals contribute to the broader political context. The necessity for utilising both a sociological and a psychological explanation in an attempt to account for the violence is perhaps best expressed by Van Zyl (1990) who suggests that regardless of the political forces at work, one is obliged to concede that without the general climate of violence, the instigators would have a much harder time mobilising their constituencies. Having argued for the need to have both psychological and social analyses of the current violence and while these clearly represent different levels of analysis, it is also necessary to point out that these are not necessarily as autonomous of one another as my discussion thus far might have suggested. This is a point on which I intend to elaborate at a later stage in the paper.

Research on the Effects of Exposure to Violence on Violent Behaviour

Although there has been no South African research which seriously investigates this question, the notion that violence begets violence has been with psychology for many years. Its formal origins are with the hypothesis that abused children are likelier than others to abuse their own children although it is quite possible that this use is simply a formalisation of a commonsense version that long pre-dates it. In more recent years, this notion was incorporated into the psychology of political violence. Most of this research has arisen out of Northern Ireland, with significantly less from other parts of the world which have experienced political violence. This might be due to the chronic nature of the conflict there as well as the relative freedom to research these sorts of issues (by comparison with the South American States and South Africa where academic researchers have themselves been subjected to the same repression that they are attempting to study). In Northern Ireland, Fraser (1974) notes that children from as young as eight years old have become actively involved in fighting. He, along with other researchers such as Lyons (1979) and Lawson (1981 in McWhirter & Trew, 1982), have predicted an inevitable response to children's exposure to political violence.

Lyons (1979) noted that those arrested for minor offences such as "riotous" behaviour as well as those on more serious charges of terrorism generally showed no evidence of psychiatric disorder or psychopathy. In spite of his lack of evidence however, he concludes his study with a strongly worded warning against the long-term effects of participation and habituation to violence on children. Fraser similarly refers to "education for aggro" and describes how the traditional fears of children are translated into violent action through a range of factors of which the most important is modelling or social learning. He concludes that the experience of aggression is likely to increase aggressive behaviour through learning and justifies this in terms of his own observation that there had been increased gang activity since the rioting began in 1969.

Later research by Fields (1976) criticises Fraser's research for relying on anecdotal information rather than valid research methods, but then proceeds to reach similar conclusions, albeit in the realm of a more sophisticated conception of attitude formation. Comparing the results of the Thematic Apperception Test and the Tapp-Kholberg questions of legal socialisation for a group of Catholic and Protestant children at six monthly intervals over a period of three years, she traces the development of political attitudes in a situation of ongoing strife. Her conclusion is that exposure to political violence leads to a passion for personal involvement in terms of destructive behaviour and a translation of fear into hatred of representatives of the feared objects. She predicts the future of the youth of Northern Ireland as being that of "militaristic automatons". The difficulty with this research in relation to the hypothesis that violence begets violence is that it is not possible to establish that these attitudes will necessarily translate into violent behaviour.

Other researchers (Harbison & Harbison, 1980) have also suggested the likelihood of increased aggression in Northern Ireland based on the epidemiology of variables associated with delinquency, including truancy, reading retardation, anti-social attitudes and teacher-rated anti-social behaviour. McLachlan (1981), working within an anecdotal framework also arrives at this same conclusion saying:

If children see adults employing violence it becomes an accepted form of play and from there when deep and divisive issues arise, as they do with us, it is easy transition into much more serious use of violence. (p.287)

This Northern Ireland research is supported by the little research available from other countries. Landau and Beit-Hallahmi (1983), examining the effects of the Israeli wars on the level of aggression in the population, note an increase in individual aggression measured in terms of the crime rate in the period following a war. A similar conclusion is drawn by Bloom and Amatu (1983) in their work on the long-term effects of socially sanctioned violent behaviour in Nigeria. All of these findings are in turn supported by Kelman's (1973) examination of the effects of institutionalised violence of the Nazi mass killings in which he describes the dehumanising consequences of violence for the victim as well as the aggressor. This, in turn, allows for the continuation of the violence.

However, in spite of the apparent support for the hypothesis that violence begets violence we must note the cautions of later researchers (McWhirter, 1983; McWhirter & Trew, 1982; Heskin, 1980). These researchers argue that much of the writing in the area of the effects of childrens' participation in and exposure to violence has been based on assertion rather than well-founded knowledge. They point to serious methodological problems in these studies such as beginning with a sample of those involved in violence and working backwards to assume a "cause" in the form of the violence to which the subjects previously have been exposed. These studies do not look explicitly at children who have been exposed to violence, but who have not themselves become violent (they appear to assume that there can be no such person). Further, they cannot conclusively show that if there indeed is an increase in violence for the whole population, that this is a consequence of exposure to previous violence and not some other circumstance that co-occurred with it. These and other difficulties such as the anecdotal basis of Fraser's work and the limited sampling in Field's studies undermine the validity of their claims.

In contrast to these early researchers Heskin (1980) for example, argues against perceiving the source of increased violence as a result of the riots, noting that if the 'troubles' were to end tomorrow, Northern Ireland would still be a society under enormous socio-economic stress, which he sees as a more significant contributor to violence. McWhirter (1983) acknowledges that some youths who would not normally commit crimes appear to do so because of the 'troubles', but she suggests that their number has been grossly exaggerated. She quotes statistics which show that the number of offences known to police in Northern Ireland was still only two-thirds the rate in England and in Wales. When measured against its own previous levels of crime, Northern Ireland has however seen a slight increase in anti-social behaviour. These have, however, tended to be acts of dishonesty, unrelated to political activity and might, as Heskin suggests, be better related to deteriorating social conditions than to participation in violence itself. She further notes that children have a remarkable capacity to insulate themselves from the violence and quotes evidence from her own studies (McWhirter & Trew, 1982) where children did not perceive the violence around them as an overwhelming feature of their lives. On the basis of this McWhirter argues:

It would, therefore, seem … that fears of a serious outbreak of anti-social behaviour among the young people of Northern Ireland and the total disintegration of Northern Ireland society are largely unjustified.(p.124)

The disagreements set out here on the question of whether violence begets violence are reflected also in research into the family "cycle of violence". In terms of this, abused and neglected children are thought to be more likely than others to abuse their own children. Widom (1989) in a major review of research conducted in this area, suggests that, contrary to expectation, the evidence for this claim is contradictory and unreliable. There is, she says, little evidence for the claim that violence begets violence, and where it exists, is limited by a number of methodological problems. In particular, there are contradictory definitions of what is meant by violence or abuse, lack of control groups and reliance on retrospective accounts of the experience of abuse. While of course there are differences between violence experienced in the home and political violence, there are perhaps sufficient parallels in the two situations to add weight to some of McWhirter's (1983) claims.

It is apparent from this discussion that there is, at best, some uncertainty around the hypothesis that violence begets violence. In general, in the research that I have reviewed here, much of the disagreement is attributed to methodological shortcomings. There are, of course, obvious methodological problems in the studies I have quoted, but a far more fundamental issue, I believe, accounts for both methodological problems as well as apparently contradictory evidence on this question. This is the different theoretical orientation within which the question itself is being asked. It is inevitable that there should be apparently contradictory findings and criticisms of methodology when researchers and critics are asking different questions in the same words. What I intend to do in the second half of the paper is to examine the different meanings that the question takes on when asked in different sorts of theoretical frameworks and to examine the validity and usefulness of each of these "questions".

The Behavioural Question

In its crudest form (and this is usually the one that informs commonsense) the question of whether violence begets violence is seen in terms of a linear causal relationship between the external event of violence and violent behaviour. In other words, that somehow the experience of violence will directly and inevitably cause a person to act violently in the future. In searching for evidence on the question in this form, one might gather statistics on the events of violence to which a specific population has previously been exposed and then attempt to establish a correlation with this and subsequent evidence of violent behaviour. This is precisely the sort of exercise in which Harbison and Harbison (1980) engage, using indices which are known to be correlated with anti-social behaviour and in turn assuming that these can be directly attributed to the previous events of the "troubles" in Northern Ireland. Much can be made of the methodological problems evidenced by studies using this kind of approach including a poor definition of terms, limited sample size, lack of control groups and difficulty in ascertaining precisely what violence children have been exposed to and whether it is this, or other circumstances, to which they are reacting. These difficulties might appear to be capable of resolution through bigger and better empirical investigations into the matter, that is, investigations which gathered evidence from more statistically significant populations of those exposed and not exposed to political violence. Alternatively, it might be suggested that this research should be conducted at repeated intervals after the experience of violence in order to ascertain its validity in both the short, medium and long term or that the method might be improved in other ways.

There are, however, major problems with doing this kind of research at all. The biggest of these is the establishment of direct causality in the relationship between an external event of violence and violent behaviour. To ascertain a correlation (no matter how strong this might be) is not the same as to establish causality. Mestrovic (1985) points to the logical impossibility of making a claim of direct causality in relation to external events and human behaviour. Firstly it is not possible to ascertain the point at which one event, such as violence, ends and another such as economic deprivation begins. How then is it possible to establish that the behaviour of violence was a response to violence and not the previous or subsequent deprivation, as Heskin (1980) appears to suggest. Secondly, there is no way of distinguishing clearly between a cause and an effect. In other words, was exposure to previous violence a cause of current violent behaviour or are both simply a symptom of another, as yet unknown, cause. This confusion is exacerbated by the fact that subjects have no clear recollection of events that they have been exposed to. They will remember some but not others and will also have different sorts of memories attached to each. There is thus no way of proving that subjects are reacting to the "same" external event and thus no possibility of causally linking this with their current behaviour.

Of course, none of these problems is simply a matter of poor research design or hassle factors in the process of research. They are reflections of fundamental philosophical problems with the paradigm within which the question itself is construed. To argue that violence causes violence in a linear fashion, is to situate oneself firmly within the constraints of crude empiricism. This paradigm has its origins in Cartesian dualism which sets up the separation between subject and object, between inner and outer. In separation of these two factors, there could only be a deterministic relationship between them. For Descartes this was the determination of the body by the mind known as rationalism. In empiricism this dichotomy is presented in terms of the priority of the external over the internal. In terms of our question this distinction becomes the separation between the 'event' of violence and the reaction of the individual. For an empiricist, the subject is simply a screen for receiving the stimulus (in behavioural terms) which is projected onto it. The subject is then obliged to react (or respond in behaviourist terminology) in a manner which has a predictable relationship with the stimulus.

This form of crude empiricism forms the philosophical basis of claims made within the stressful events model of psychological stress theory in terms of which an identifiable event is seen to correlate with a particular stress reaction. A causal relationship is then imputed to the correlation, so strongly in fact that the event itself can then be given a 'stress rating'. So violence might be worth 200 points on such as scale – equivalent to the degree of response it is assumed to elicit. The linear causality assumed here is an inevitable consequence of the empiricist position. To separate the subject and object so profoundly leaves one with no other possible explanatory model to account for a relationship other than a simple determination of one factor by another. In terms of this, the question as it is posed in this theoretical tradition becomes self-defeating. It is self-defeating firstly because it is unable to answer its own question with any validity and secondly because when it attempts this, its answer cannot be anything more than a descriptive correlation between events of violence and subsequent violent behaviour it cannot really answer the question of why this is so. In attempting to answer it, it must go back to its first premise, that there have been violent events. This claim is not only philosophically tenuous but also circular and pointless, adding to the feeling of helplessness of academics and clinicians in the face of the current violence.

The Interactionist Question

Having set out this question in its crudest form, it is only fair to acknowledge that very few academics would even begin to think about the relationship between the external event of violence and subsequent violent behaviour in this simplistic way. Almost all psychologists working in the area of political violence have acknowledged the fact that not every child will react to violence in precisely the same way. Factors thought to mediate this reaction would include such things as social support, family relationships or organisational involvement. This is precisely the sort of explanation McWhirter (1983) gives for the low levels of violent behaviour amongst those exposed to violence. Similarly in South Africa a great deal of attention has been given to the role of these factors in the "resilience" of children in their psychological response to various forms of violence and repression (Dawes, 1987; Gibson, 1989; Turton et al). These factors are thought to intervene in the cognitive process of 'appraisal' of the event. For example, being a part of a tightly-knit group may prevent an individual from feeling alone and therefore vulnerable to attack. They may, as a consequence of this, be less inclined to act violently.

This model represents a shift away from the notion that an external stimulus produces a behavioural response and acknowledges how the response is determined by the way in which the person perceives the stimulus. In other words, because it is possible that, due to the factors mentioned above, and others, people may 'see' the experience of violence differently and they may then respond either with aggression or depression or perhaps not at all. We are no longer looking at a simple relationship between cause and effect and therefore cannot predict with absolute certainty that the external event of violence will lead to violence in all cases. Where results are contradictory, we can cite one of the factors above as the cause of the discrepancy. Where this is not sufficient we must then assume that we have overlooked a further, as yet unknown, factor which might mediate the experience of violence. The question of whether violence begets violence is, in terms of this model, set out as: What are the additional factors which must be present in order for violence to cause violence (or vice versa)?

While this approach seems to be a way out of the impasse in which crude empiricism placed us, there are still serious problems with its conceptualisation of both the significance of the event of violence and the mediating factors. The first problem here is that the notion of appraisal does not seriously address the problems of the empiricist position. What happens in this approach is that the "event" of violence is simply removed from outside to a mirror image on the metaphorical brain. Here, the image is altered by the reflected images of other 'events' such as social support and so on. It is this accumulated set of images and their influence on one another which constitute appraisal. While, however, there is no examination of the laws or structures which govern the process of appraisal, we can only assume that appraisal is a function of external events rather than a reflective activity of the mind. This is of course the same criticism which has been directed at cognitive behaviourism.

The second difficulty with this approach is that it does not seriously address the problems of linear causality already considered in this paper. Instead of a simple causal explanation, it adopts a complex explanation, which nonetheless follows the model of linear causality. The causal explanation usually set out as: If X then Y, now becomes: If X (and A and B and C …) then Y. An experience of violence in combination with a series of other 'events' may cause violent behaviour. In other words, without a coherent theory which can explain how it is that all these factors transform one another into a new focus for the subject, the mediating factors can only be seen as additional linear causes.

Finally, difficulties in this approach are also created by the conceptualisation of mediating factors such as external events in themselves. Even if violence is mediated (however vaguely) through the process of appraisal, the mediating factors are assumed to act on appraisal from outside of it. In other words, social support, like violence in the first model, is seen as an 'event' that consists, for instance, in a close-knit family or a well-established community organisation. The difficulty with this is that in some cases a loving family or coherent organisational position may increase and in others decrease the likelihood of an aggressive response to the experience of violence, depending on how it too is understood by the individual. Thus, while this approach represents a major advance on the behavioural understanding of the question, there are still a number of aspects which make it liable to the set of criticisms outlined in relation to crude empiricism.

The Psychoanalytic Question

As a way out of the sorts of theoretical dilemmas posed above, Mestrovic (1985) proposes a shift away from the dualistic conceptualisation of external event and internal response. He argues that the whole notion of an external 'event' is one which is necessarily reductionistic and instead, proposes the use of the term 'fact', which has more in common with a psychoanalytic conceptualisation of trauma. A trauma is, of course, neither internal nor external, but straddles and incorporates both of these. These traumata or 'facts' have a number of features which further distinguish them from events in that they are not visible, they are not dated and are not localised in space. They are, in psychoanalytic terms, the memory which bears no necessary relation to the event (as we know full well from Freud's reconceptualisation of the incestuous experience as phantasy rather than actual event – but no less real because of this). There can then be no directly causal relationship between the external event of violence and subsequent aggressive behaviour. The only true relationship then can be between the internal 'fact' of violence and the subsequent reaction. As Mestrovic notes:

Events may become trauma not because of their nature but due to the psychological state of the subject, social circumstances and psychic conflict … (p.841)

At first glance this synopsis appears to bear a strong resemblance to the interactionist approach outlined above. The crucial difference, however, which moves this approach beyond elaborated empiricism is that 'facts' or trauma are not simply a combination of a set of mediating factors plus the event of violence. It is specifically the conception of psychic conflict which shifts this account out of a dualistic conception of internal and external. Through conflict, all internal and external processes are dialectically transformed by one another according to laws and structures which are not themselves visible (although they can be deduced from the empirical). In terms of this approach we cannot talk about one thing simply causing another (even through the central point of appraisal). The path between cause and effect is far too complicated to make this a useful exercise. Neither then is it possible to offer a yes or no (even in an elaborated sense) answer to the question of whether violence begets violence. Instead it is far more instructive to develop an understanding of the possibilities and constraints in terms of general laws of psychological functioning that may enable us to better understand some of the processes involved in the relationship between the experience of violence and violent behaviour. In terms of this approach then, we have to rephrase the question of whether violence causes violence as: 'How is it possible that violence arises out of violence?'

In terms of the model of trauma or 'fact' which I would argue allows for the best framework within which to ask the question: 'Does violence beget violence?', there can then be no way of predicting with certainty how a child may respond to the experience of violence or indeed what aspect of the experience of violence they may be responding to. For example, an adolescent may be more responsive to the threat of physical injury while the pre-school child may respond only to the distress of the parental figure for a variety of reasons to do with the level of the child's development. This position, however, may be stretched into the absurdity of extreme relativism where it would be possible to argue that each individual will react in his or her own unique style. In this position then, it becomes pointless to even study violence (and perhaps also pointless to even intervene in current violence). This is certainly not the position I am advocating in this paper. Rather, I would like to develop a theoretical model for analysing the constraints and possibilities which provide the framework through which to understand a relationship between the experience of violence and violent behaviour. This model would then provide a means of analysing a particular instance, or group of instances, in terms of this question as opposed to offering an answer to the question itself.

This analysis would need to be conducted on two levels. On the first level, it is necessary to outline the psychological constraints and possibilities that would allow for violence to emerge out of violence. The second level would need to address the social constraints and possibilities which constitute both the experience of violence and define the possibilities of reacting to it.

Psychological Constraints and Possibilities

In examining the psychological constraints and possibilities in the experience of violence I will draw from the broad field of psychoanalysis. The account provided here is, in terms of my overall argument, simply one example of the form an analysis of this question might take and is designed to illustrate some of the processes, structures and laws which may be used in such an analysis. In other words, my intention here is to offer an approach to understanding rather than an understanding per se.

In terms of my position it must be recognised that the experience of violence may take on particular meanings for particular individuals dependent on both the kind of violence experienced and the historical experiences of the individual. Thus, in one case, violence may be experienced primarily through the loss of a loved one, while in another it may be experienced as a test of strength. In this particular analysis I begin with one of the likely meanings which may be attached to the experience of violence: the confrontation with the possibility of the ultimate form of destruction, death (De Wind, 1968) and trace this through to an expression of violent behaviour.

Within the psychoanalytic position, the basic constraints are determined by the limits of instinct and development, in particular the development of the ego and superego out of the id and the functioning of these agencies in relation to one another. De Wind (1968) has argued that in the face of an extreme threat in the external world, the most adaptive response is a retreat from object relations back to the earlier states of narcissism. This seems also to be a very probable reaction to the experience of violence which contains within it the possibility of death (even if this is a symbolic rather than an actual possibility as might occur in the witnessing of violence against others). This retreat follows the general pattern of regression in response to danger and in this case, serves two important different functions. Firstly it allows the ego to function independently of objects around it which is appropriate in terms of the fact that surrounding objects are threatening and secondly, it allows the ego to compensate for the loss of love from objects by self-love (Freud, 1914).

Libido is, however, only one of the instinctual forces governing human behaviour. The other force which operates in conflict to libido is in terms of Freud's later conceptualisation, Thanatos, or the death instinct. This instinct is in most cases bound up with Eros or the life instinct (of which libido is now conceived of as one aspect) (Freud, 1923). In this usual state the destructive tendencies of Thanatos which aim towards the breaking down of structures are curbed by the synthesising functions of Eros. When the libido is, however, withdrawn back into the ego, the death instinct is denied one of its most common modes of expression and release through the relation with objects (ibid). With this painful build-up of energy, it is possible that the death instinct becomes partially defused from Eros and seeks its own expression. Its initial attempt at expression must be in the form of masochism which is prior to sadism and which is directed inside towards the self. It follows the ultimate aim of the death instinct which is to "lead organic life back into the inanimate state" (Freud, 1923, p.380). In the face of this attack on itself, however, the ego can no longer stand the proximity of the death instinct which threatens it and is forced to expel some of it from the ego. In Klein's conception, this would be similar to a retreat into the paranoid schizoid position and the use of the primitive defence mechanisms of splitting to protect the ego (Segal, 1982). The good (Eros) is held inside the ego and the bad (a defused part of Thanatos) is expelled outside onto other objects. There is thus a partial breakdown in the relation between Eros and Thanatos which then follow separate paths of resolution, each of which may, in different ways, contribute to the expression of violent behaviour.

The libido is held to the self through narcissism. Thus in terms of Freud's understanding of the relationship between ego and object libido (1914), there must be a necessary decrease in the possibility to experience love for others. This might in some way account for the apparent phenomenon of dehumanisation that many theorists have used as the basis for understanding how it is that a person is able to exercise violence against another (Kelman, 1973; Bernard et al, 1971). This is, however, only a part of the explanation and may already be countered by the fact that in spite of the dehumanisation of the enemy in violence, there is very often an increase in loyalties between members of the same group. This is, however, not object love but an extension of self-love (Freud, 1921). In fact, as the threat against the self is perceived, so secondary narcissism is extended and developed to include not only the self but all those like one – those who share a common nationality, skin colour, ethnic identity, etc. The love of the self and the narcissistic love of the group does not, however, have the normal gentle characteristics we attach to love. It is a harsh intolerant love – for its object is not the ego as it is, but rather the ego as it would like itself to be. In other words, its love object is the ego ideal. As a result it cannot tolerate self-criticism and can therefore bear no weakness in itself or other members of its group. Its own weaknesses (and those of the group with whom it identifies itself) need, like the death instinct, to be expelled. These 'bad parts' are then projected onto 'the other' whose apparent weaknesses then serve as the reason for one's own dissatisfaction (Kovel, 1988).

This structure paves the way towards aggressive behaviour in a number of different senses. By dehumanising those against whom the violence might be enacted and then by projecting one's own weakness onto 'the other' and so providing a victim for one's anger, a vicious cycle is set up of love. The ego feels the need for protection and so, in order to protect itself, it retreats into narcissism which in turn justifies and reinforces its perception of the need for protection.

The partially defused death instinct follows a no less convoluted path from the position of narcissism. Anger is often seen by psychologists to have great healing properties, ridding the self of the burden of self-hate and rightfully protesting against what is wrong with the world (Brenman-Pick, 1991). This sort of conception is the one which underlies Fanon's (1968) understanding of violence as a cleansing force that allows African people to shake off their despair and fearfulness. Brenman-Pick (1991) however points out that anger is only ever half righteous protest and half destructive rage. It is insofar as anger is destructive that it serves primarily the aims of the death instinct. The parts of the death instinct which are projected in the form of anger in order to maintain the integrity of the ego are then experienced as persecutory by the ego. In the case of an extreme threat such as violence, the intensity of the projection may in fact force a regression to the harsh primitive superego of the paranoid schizoid position which is the precursor to the Freudian superego formed through the dissolution of the Oedipus Complex (Segal, 1982). Here it joins the original attacker in threatening the ego not just with external reality but also with its own rage turned back on itself. The response to the concerted attack is twofold: Firstly, it involves a repetition of the cycle of threat, withdrawal to narcissism, defusion of instinct and projection of aggression. Secondly, however, it also calls up the individual's own unconscious longing for death. In other words, it puts the person in touch with their own instinctual aggression which, unmediated, is imagined to possess omnipotently destructive properties. The aggressive impulse wishes to destroy as much as the loving impulse wishes to hold and build. The conflict between the two produces guilt which reactivates the other guilts of the Oedipus Complex, thus summoning the Oedipal superego in judgement of the self. The judgement evokes not only guilt and shame but also fear and anger. It is, of course, the latter evocation which forces the cycle to repeat itself. Thus, while it has been argued that it is a lack of superego which is the psychological basis of the current violence (Van Zyl, 1991), it appears more likely that it is in fact the harshness of the superego which promotes the sorts of feelings which might lead to aggressive behaviour.

My explanation thus far, however, only takes us to the first stage. What I have suggested here is not an explanation for violent behaviour, only for unconscious aggressive phantasies combined with an inflated self-regard and narcissistic intolerance of the other. There is no necessary connection between the unconscious phantasy of destruction and physical activity. The former, in Freud's terms, must be limited by the laws of primary process thought, while action is necessarily a function of the secondary process (Freud, 1911). In order for a phantasy to be transformed into a conscious action (which, however random and irrational violent behaviour might appear to be, it still is) we must add some additional features that may occur in the transformative process. In the process I have already described the ego as threatened on all sides. It faces the real threat in the external world, it faces the threat posed by the persecutory super-ego and by its own instincts. All it has left in the midst of this onslaught is itself. In order to hold on to its own fragile structure it reconstitutes its threatened self through the most radical action of all – the destruction of another. In a way which is not that dissimilar to the way in which a borderline patient will cut herself to prove her own existence and attempt to counteract the terrible feeling of numbness, an ego, terrified of its own disintegration, might kill another to establish its own life. This sort of violent behaviour has similarly been seen as an attempt to reconstitute the self by psychologists working with Vietnam veterans suffering from post traumatic stress disorder (Feldman, 1988). In addition to holding the self together, violent activity is reinforced by the support of a narcissistic lack of concern for others and the identification with a group which is set up in opposition to an enemy.

Of course, as I have already stated, this is just one possible set of permutations of psychological process in relation to the experience of violence. What is perhaps more important than this particular analysis is the way in which it illustrates the reaction through dialectical causality within the constraints of instinct, psychical structures and mechanisms. Having said this then, one might reasonably ask, what are the other possibilities which may alter the course of the experience of violence. The most important amongst these, particularly in relation to children is the concept of development. This is, of course, not separate from the processes and structures outlined above, but in fact simply marks a particular stage within the development of these structures and processes. A child, for example, in the Oedipal stage may be currently experiencing an excess of guilt which may operate in concert with the experience of guilt induced by the persons's own aggressive wishes, as I illustrated earlier. On the other hand, an older child may still experience guilt but be able to activate more sophisticated defence mechanisms through which the feelings evoked could be expressed and dealt with (Anna Freud, 1965). The significance of development must, however, not be limited to the current developmental stage of the child but also in a psychoanalytic sense be able to accommodate the significance of 'the return of the repressed'. Each stage through which a child has passed leaves residues to which the child may return under threat. In the experience of violence, where the likely regression is to narcissism, one would need to understand how this particular phase was negotiated by the child and the impact this might have on their present functioning.

A second level of important permutations in the process might also arise out of the relations of the child with significant others. While as Freud (1905) notes, the development process is pre-determined, accidental contingencies will evoke particular dynamic structures and unconscious conflicts within this. Just as the trauma of violence may provoke alterations in the psychical condition (such as regression to narcissism) so other experiences in the child's life will have resulted in both unconscious memories that may be evoked by the violence and structural anomalies (such as a fixation at a particular stage of development). The child may, for example have experienced bonding difficulties with the mother in the earliest months of life, thus leading to narcissistic injuries which reinforce the retreat to narcissism in the face of threat. Similarly, a current experience of violence may call up, in addition to aggressive instincts, memories of previous loss or destruction which build upon or transform the action of the instinct in a variety of ways. Of course it would be naïve to suggest that one should only consider the relationship with the phantasy mother or the impact of past memories on the current trauma. The real mother, or family, or community organisation is also of course significant in terms of the particular form the trauma takes. These people and groups do not, however. have an impact which is independent of developmental structures or phantasy objects. They are in fact perceived through these. So a child who experiences violence within a tight-knit family may, perhaps because of Oedipal guilt and rivalry, be unable to seek compensatory comfort from members of his or her family. This example once again, illustrates the crucial differences between the psychoanalytic and the interactionist approaches which would simply assume the supportive nature of a close family.

Social Constraints and Possibilities

The explanation I have offered so far may, however, rightfully expose me to accusations of psychological reductionism – even in spite of the argument I offered earlier in terms of the validity and need for a psychological explanation. It is one thing to offer a psychological explanation of the violence, but quite another to treat the psychological as if it operated entirely independently of its social or political context. Others have recognised the need to take this context into account and there have been several attempts to address the problem of the relationship between the individual and society in terms of political violence. One kind of attempt to address this need argues that psychological theory, particularly psychoanalytic theory, can be used to explain the social (Van Zyl, 1991). In terms of this approach the Oedipus complex is used to account for, for example, the relation between the State and the people. In its most extreme form this approach draws on the post-structuralism of Lacan. The difficulty with this sort of approach is that individual and society are indeed different phenomena that work according to different rules. Society, for example, does not have instincts (although individuals within it do) and individuals, on the other hand, do not have modes of production (although they might have within them the ideological reflection of these or operate within them). As exciting as this sort of analysis is, I think that it can never be really useful, except in a metaphorical kind of a sense. It tells us more, for example, about what people think of the State, than about the State itself as a material social entity. Similarly, the strength of psychoanalysis is in its ability to explain the material reality of individual psychological functioning (Lasch, 1979).

A second sort of attempt to free psychology of reductionism is to explore how the psychological notion that violence begets violence is a product of social discourse. This sort of analysis (Swartz & Levett, 1980; Straker, 1989) offers rich insights into the way in which society perceives violence against children and its effects, but tends to ignore the real psychological processes and reactions involved. As none of these writers would themselves attempt to negate real psychological processes and effects, their approach must be acknowledged as an addition to a psychological explanation, not a substitute for this.

I have argued then that individual psychology and society represent two different objects of study and that they require different methods of analysis. This, however, is not to say that the psychology of the individual must then be asocial. This would, of course, be to go back decades in the development of psychology (Henriques et al, 1984). What I am trying to suggest though, is that the way in which society becomes a part of the individual is the domain of psychology, rather than society itself which operates according to laws which cannot be reduced to the psychological.

It is then necessary for me to introduce the second set of constraints and possibilities which operate in a dialectical relationship with those of instinct, development and structure which I have described above. This second set are the mechanisms and processes which determine the social nature of the individual's experience of and reaction to violence. These constraints and possibilities operate in three ways. The first of these sets of constraints is determined by the individual's history which is constructed not only out of instincts but also out of objects and environmental circumstances. From the very first, none of these objects in the child's life operate outside of their social position. If then, a child's particular dynamic structure which contributes to their experience of political violence can be attributed, in part, to poor mothering, it is necessary to ask what sorts of social constraints governed the mother's capacity to care for her child. In Ingleby's (1976) terms, given a mother's position in the system of production and consumption, how much crying can she permit or tolerate in her child? Thus, what is social becomes internalised into the mother's behaviour and through that, becomes a part of the child's developmental and dynamic structure and so filters the experience of violence in particular kinds of ways. Within psychoanalytic theory there is the structural accommodation to the significance of the social as a psychological phenomenon through the notion of the super-ego. In Freud's terms, the child identifies with the parent in the resolution of the Oedipus Complex (1932) and in taking in the social values of the parent becomes civilised. For Klein this process of internalising the social begins in the earliest moments of life with the constant cycle of projection and introjection in relation to the object (Segal, 1982).

The second major way in which social constraints act on the psychological is perhaps best explained through Mestrovic's (1985) notion of collective representations. While he acknowledges that the "fact" of trauma arises out of unconscious psychical processes, he also argues that these are to some extent moulded by shared sets of meanings in any social context. He is not here proposing any sort of genetic collective unconscious, but rather acknowledging that psychical "facts" and expressions of these are limited by social constraints. This is a notion that has long been recognised by those working in cross-cultural psychology where it is recognised that the form taken by disorders will vary across time and culture (Swartz, 1987). What happens with this is not an alteration in the basic psychical structures themselves, but shifts in the expression of conflicts. Thus while conversion hysteria was a popular mode of expression of psychical conflict amongst Victorian women, so Anorexia Nervosa or Bulimia is far more prevalent in the present age. In terms of the question of violence in South Africa, we would need to accept that on this level, violence has become an accepted means of expressing a conflict which may at another point in time have been dealt with in some other kind of way. This is not, of course, to say that the mechanisms of regression, defusion and projection do not occur, but that their end point may vary depending on how society governs the mechanism of expression of conflict. In the concentration camps, for example, where people were faced with extraordinary degrees of aggression, it is noted that for the most part, they did not respond aggressively as this would certainly have led to their death (De Wind, 1972). In present-day South Africa, however, violence has become not only a legitimate means of protest, given the years of curtailment of other forms of protest, but also, in some cases, a necessary means to survival.

Thirdly, social constraints must be acknowledged insofar as they provide the setting within which the psychological is expressed. These circumstances which arise out of a different set of laws, processes and activities pertaining specifically to the social, then provide the conscious cause and opportunity for expression from the point of view of the individual. In other words, without real power issues between Inkatha and the ANC there would be no expression of violence on the level that we have seen. Similarly, if there were no real deprivation amongst those in hostels and squatter camps it is unlikely that the violence would have occurred to the extent that it has. Finally, if there were no real transformation in the power relations between the State and the ANC, we would not have elements of the State seeking to undermine the ANC. The recognition of the social in terms of this constraint links to the observation I made earlier about the way in which psychological and sociological explanations do not operate entirely independently of one another.

Conclusions

In this paper I have attempted to redress the lack of serious attention that has been given to the psychological question of whether violence begets violence. My aim has not been to provide the definitive answer to this question – indeed in terms of the approach that I argue in favour of – there can be no definitive answer. My intention has rather been to try and raise some questions about the most useful way of understanding the idea that violence begets violence. Amongst the most important of the points I wished to make, is that violence does not beget violence in any sort of a simpleand direct fashion. This is not to say that there is no connection between a previous experience of violence and violent behaviour, but rather that there are a whole range of processes, relationships and transformations that occur between an external event and the subsequent behaviour of a person. If we can begin to develop an informed and sophisticated psychological understanding of these constraints and possibilities that provide the framework for the relationship between exposure to violence and violent behaviour we may to some extent be able to lift the dreadful feeling of impotence most of us experience in relation to the apparent inevitability of current violence.

Note:

¹ I recently attended the International Conference on Organised Violence in Southern Africa, held in Harare in September, 1990. Participants from South Africa who included representatives of political groups, health-care workers, academics and parents were expressing fears that children here were becoming increasingly aggressive and out of control.

References

Bloom, L and Amatu, H. Nigeria. "Aggression, a psychoethnology." In: A P Goldstein and M H Segall (eds.). Aggression in Global Perspective. Pergamon: New York. 1983.

Brenman-Pick, I. "Conscience in man and society. The seven deadly sins: Anger." Paper presented at a public meeting of the Johannesburg Psychoanalytic Society, April 1991.

Chikane, F. "Children in turmoil: The effects of the unrest on township children." In: S Burman and P Reynolds (eds). Growing up in a Divided Society, Ravan Press: Johannesburg, 1986.

Dawes, A. "Security laws and children in prison: The issue of psychological impact." Psychology in Society, 8, 1987, pp 27-47.

De Wind, E. "The confrontation with death." International Journal of Psychoanalysis, 49, 1986, pp 302-305.

De Wind, E. "Personality, aggression and therapy." International Journal of Psychoanalysis, 53, 1972, 173-177.

Feldman, T. "Violence as a disintegration product of the self in Post-Traumatic Stress Disorder." American Journal of Psychotherapy, XIII(2), 1988, pp 281-289.

Fields, R. Society under Siege. Temple University Press: Philadelphia, 1976.

Fraser, M. Children in Conflict. Penguin: Harmondsworth, 1974.

Freud, A. Normality and Pathology in Childhood. Hogarth: London, 1965.

Freud, S. 1905. "Three essays on the theory of sexuality." In: On Sexuality. Penguin: Harmondsworth, 1977.

Freud, S. 1911. "Formulations on the two principles of mental functioning." In: On Metapsychology. Penguin: Harmondsworth, 1984.

Freud, S. 1914. "On narcissism: An introduction." In: On Metapsychology. Penguin: Harmondsworth, 1984.

Freud, S. 1921. Group psychology and the Analysis of the Ego. WW Norton: New York, 1959.

Freud, S. 1923. "The ego and the id." In: On Metapsychology. Penguin: Harmondsworth, 1984.

Gibson, K. "Children in political violence." Social Science and Medicine, 28 (7), 1989, pp 659-665.

Harbison, J and Harbison J (eds). A Society under Stress: Children and Young People in Northern Ireland. Open Books: Somerset, 1980.

Henriques, J, Hollway, W, Urwin, C, Couze, V and Walkerdine, V. Changing the, Subject Psychology, Social Regulation and Subjectivity. Methuen: London and New York, 1984.

Heskin, K. Northern Ireland: A Psychological Analysis. Gill and MacMillan: Dublin, 1980.

Ingleby, D. "The psychology of child psychology." In: R Dale, G Esland and M MacDonald (eds). Schooling and Capitalism. Routledge & Kegan Paul: London, 1976.

Kovel, J. White Racism. Free Association: London, 1988.

Kelman, H. Violence without moral restraint: Reflections on the dehumanisation of victims and victimisers. Journal of Social Issues, 29 (4), 1973, pp 25-60.

Landau, S and Beit-Hallahmi, B. Israel: "Aggression, a psychohistorical perspective." In: A P Goldstein and M H Segall (eds). Aggression in Global Perspective. Pergamon: New York, 1983.

Lasch, C. The Culture of Narcissism: American Life in an Age of Changing Expectations. Abacus: London, 1979.

Lyons, H. "Civil violence: the psychological aspects." Journal of Psychosomatic Research, 23, 1979, pp 373-393.

McLachlan, P. "Teenage experiences in a violent society." Journal of Adolescence, 4, 1981, pp 285-294.

McWhirter, L and Trew, K. "Social awareness in Northern Ireland children." Bulletin of the British Psychological Society, 34, 1981, pp 308-311.

McWhirter, L. "Growing up in Northern Ireland: From 'aggression' to the troubles." In: A P Goldstein and M H Segall (eds). Aggression in Global Perspective. Pergamon: New York, 1983.

Mestrovic, S. "A sociological conceptualisation of trauma." Social Science and Medicine, 21 (8), 1985, pp 835-848. Segal, H. Introduction to the Work of Melanie Klein. Hogarth:; London, 1982.

Straker, G and the Sanctuaries Treatment Team. "The continuous traumatic stress syndrome – The single therapeutic interview." Psychology in Society, 8, 1987, pp 48-78.

Swartz, L. "Transcultural psychiatry in South Africa." Transcultural Psychiatric Research Review, 23, 1986, pp 273-303.

Swartz, L and Levett, A. "Political repression and children in South Africa: The social construction of damaging effects." Social Science and Medicine, 28, (7), 1989, pp 741-750.

Swartz, L, Gibson, K and Swartz, S. "State violence in South Africa and the development of a progressive psychology." In: N C Manganyi and A du Toit (eds). Political Violence and the Struggle in South Africa. London: MacMillan, 1990, pp 234-264.

Turton, R, Straker, G and Moosa, F. "Experiences of violence in the lives of township youth in 'unrest' and 'normal' conditions." Psychology Department, University of the Witwatersrand, Unpublished.

Van Zyl, S. "Explaining violence in South Africa – Some psychoanalytic considerations." Paper presented at the Centre for the Study of Violence and Reconciliation, Seminar No. 9, 31 October.

Vogelman, L and Simpson, G. "Current violence in South Africa." Centre for the Study of Violence and Reconciliation, unpublished paper, 1990.

Widom, C. "Does violence beget violence? A critical examination of the literature." Psychological Bulletin, 106, (1), 1989, pp 3-28.