Towards a 'Science of Movement': Identity, Authority and Influence in the Production of Social Stability and Social Change

Social change is not just another amongst a long list of topics that has been broadly ignored by social psychologists. Equally, we cannot be satisfied simply with establishing social change as a topic area that appears regularly in journals, textbooks and undergraduate curricula (though that would still be a step forward: of an admittedly non-random selection of more than a dozen social psychology textbooks on our shelves, only two even mention 'social change' in the index). Rather, the issue of social change has a foundational status. We live in a world where, even if the rhythms of change are slow, and even if they can be longer than the span of individual lifetimes, they never grind to a complete halt. Change is always possible, and things often appear most stable just before moments of radical discontinuity. Declarations that history is at an end always lead to retractions, generally sooner rather than later. Hence the ability to account for the possibility of social change must be a criterion in assessing the adequacy of any social psychological theory. A consideration of social change must be woven as a thread into the fabric of all that we do.

But, as we argue in this paper, a consideration of social change is not just about when change occurs. It is equally about when it does not. Indeed, the most important task before us is to confront the notion that continuity is the default and that discontinuity is the exception. In many ways, the most radical thing we can do is to show that as much effort goes into reproducing the social order and into disrupting processes of change as goes into the production of change itself. These efforts may have become so routinised, so woven into the minutiae of everyday experience, that we fail to see them. But this banality is precisely what makes continuity appear inevitable and what is so effective in ensuring continuity (see Billig, 1995). For power masks itself, and thereby becomes hard to identify and challenge. It produces effects which then are presented as the products of nature, admitting of no alternative. It becomes doubly productive (Jackman, 1994).

In short, our pitch is for a psychology of social change that encompasses the study of social reproduction, of the demobilisation of alternatives to the status quo, as well as the study of mobilisation in favour of such alternatives. In line with this pitch, we address the 'conformity bias' which has exerted such power over social psychology, and which assumes that people will naturally act along existing lines of authority (Moscovici, 1976; Turner, 2006; for an extended critique, see Haslam & Reicher, 2012a). Using Milgram's studies as an example (perhaps the most famous studies ever conducted in the whole of psychology; see Banyard & Grayson, 2000), we will show that 'obedience' was far from natural. Instead, we examine how it was the product of elaborate and lengthy labours. Compliance did not just happen. It had to be mobilised. Moreover, solidarity with the learner in these studies (which would have translated as resistance to the experimenter) had to be actively demobilised.

Before we come to this examination, however, we want first to explain in a little more detail why considerations of social change are so foundational for social psychology, how they figure within the contemporary discipline, and also how we have addressed them in our own work.

The Foundational Status of Social Change

Given that the first two thematic issues of this journal have addressed 'decolonizing social psychology' and 'social change', it may be appropriate to cite Gyanendra Pandey's seminal critique of colonial histories in India (Pandey, 2006). He shows, first, how these histories focus uniquely on events of concern to the British rulers, notably instances of conflict and disorder. Secondly, he analyses how events are abstracted from their specific historical context, and fitted to a general template which is then explained in terms of 'timeless' qualities – notably the primitive religious fervour of Hindus and Muslims. Third, he demonstrates how such accounts serve to justify the colonialism by suggesting that only the colonialists can overcome such fervour, can maintain order and can modernise the country. In effect, these histories focus on the psychological flaws of the Indian masses in order to promote the civilizing mission of the British elite.

Pandey's critique echoes the words of Martin Nicolaus to the American Sociological Convention in 1968 (see Murphy, John, & Brown, 1984). Nicolaus was part of a protest against Wilbur Cohen, the US Secretary of Health, Education and Welfare, who had been invited to the Convention and was sitting on the platform. The speech began with an attack on the government for its policies in Vietnam and at home. Nicolaus described Cohen as "a military officer in the domestic front of the war against people" and as "Secretary of disease, propaganda, and scabbing". But then he moved on to the role of sociology and of sociologists. Famously, Nicolaus declared that "the eyes of sociologists... have been turned downward, and their palms upward". That is, they look down on the subject masses when they disrupt the smooth operation of power in order to serve and seek the gratitude of the powerful.ⁱ

Nicolaus's speech made waves. It both reflected and contributed to a growing dissatisfaction with the relationship between academia and society which crossed both disciplinary and geographical boundaries. In psychology, social psychology in particular, it helped crystallise a sense of crisis which, as Strickland (1976) stressed, was as evident in Europe as in the United States. The 'crisis' addressed many different issues including the values and practices of psychological research, the usefulness of methods such as laboratory experimentation, the nature of the questions that psychologists ask, and the status and generality of the knowledge that is claimed. Moreover, it was divided between a 'liberal' wing, which was concerned about making psychology more relevant to and useful for society, and a radical wing which was concerned about the uses made of psychology to preserve the status quo. Nonetheless, common to all strands was a core concern with the process of reification, that is, the way in which the role of context is ignored such that human action comes to be explained in terms of natural human essences (see, e.g., Ingleby, 1972; Israel & Tajfel, 1972).

Such a psychology makes it impossible to apprehend the constraints, concerns, opportunities and perspectives which shape what people do and in relationship to which their behaviours are meaningful. It thereby pathologises people, denying purpose and meaning of what they do. Such a psychology also removes agency from people. It renders them as puppets, driven by unchanging urges, incapable of addressing, let alone changing, their circumstances. By taking context out of the picture, socially contingent forms of action are rendered immutable, inevitable, eternal. Human nature is frozen by ignoring human society – and a frozen human nature is then used to argue that human society cannot be unfrozen. Those psychologists who (wittingly or unwittingly) propound such models serve the cause of power, just as surely as the historians Pandey described served the cause of colonialism.

By ignoring context, then, one necessarily ignores both how our understandings and behaviours are shaped by society and also how we can act to reshape society. Conversely, any approach which pays heed to the context of social psychology must necessarily address these twin issues of social determination and social change. Put more strongly, any psychology which is properly social (in Kurt Lewin's (1951) classic sense of understanding the social structuration of the psychological field) must be one in which these two concerns are foundational rather than incidental.

Foregrounding Social Determination and Social Change

We are far from alone in making these points. Many psychologists in many countries have longed, like Serge Moscovici (1972), for "the development of a science of 'movement' rather than a science of 'order'" (p. 22). It is not our intention here to review these. Rather, we proceed by outlining one particular approach, because it has been the basis for our own work which we describe below, and also because it is one of the most comprehensive and influential approaches in contemporary social psychology. We refer to the social identity approach which Henri Tajfel developed together with John Turner (Tajfel, 1982; Tajfel & Turner, 1979; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987).

The very concept of social identity serves to integrate the individual and the social context. Thus, a social identity (say, being a Catholic) is deeply meaningful to an individual. It is something they can care deeply about, and it is certainly no less significant than a personal identity. On the other hand, what it means to be a Catholic and how one acts as a Catholic is irreducibly social; a function of historical, institutional and communal realities. The importance of this, then, is that we introject the social into the individual through social identification, and that social and cultural realities come to structure what we feel, think and do.

However, especially in Turner's later work on self-categorisation processes (Turner et al., 1987; Turner, Oakes, Haslam, & McGarty, 1994), there is a second dimension to the sociality of the self. That is, both the salience and the content of social identities is not just a function of broad historical social processes. It also reflects the immediate social structure. That is, the categories we use to define ourselves are a function of who else is present in the local context and, moreover, the way we define our group depends on what makes us distinctive from these others (Oakes, Haslam, & Turner, 1994; for specific examples see Doosje, Haslam, Spears, Oakes, & Koomen, 1998; Haslam, Turner, Oakes, McGarty, & Hayes, 1992; Hopkins, Regan, & Abell, 1997). In short, self-categories reflect the organisation of social reality. This has a radical and far reaching implication. Because the structure of social reality is different in different settings (in some places we may be organised by gender, in others by class or 'race' or the team we happen to support), our self is not a fixed structure but rather a dynamic and variable process (Turner et al., 1987, 1994). Moreover, it is by understanding how the self varies as a function of the social field that we can understand how our feelings, thoughts and actions are shaped.

Another way of making this critical point is to say that the social structuration of the psychological field is achieved through the dynamics of the social self. To accept this is to cast social identity theorising as a disciplinary perspective (at least in the Lewinian sense outlined above) rather than just a model of particular social phenomena.

The critical role of the variability of the social self is one important part of the social identity argument, but it is certainly not all. For from its inception, social identity theorising was not only concerned with the social nature of the self process, but also with the centrality of the social self to processes of change. As we have argued previously (e.g., Reicher, 2004; Reicher, Spears, & Haslam, 2010), Tajfel and Turner's social identity theory of intergroup relations (Tajfel, 1978; Tajfel & Turner, 1979) is often represented as an explanation of the way in which we differentiate our own group from others in order to achieve a positively valued social identity: we put 'them' down in order to raise 'ourselves' up. However, this is only the starting point of the theory. People may well want their groups to be defined positively, but we live in an unequal world where many belong to groups that are subordinated and denigrated: women, black people, the disabled and so on. The key question, then, is how people respond. Specifically, do they accept the subordination and denigration of the group but try to manoeuvre around it (e.g., by seeking to downplay their group membership)? Or do they band together as group members in order to confront and overcome such subordination?

Research in the social identity tradition is premised on the fact that it is precisely as group members and through combination with other group members that people are able to challenge the status quo. This leads to the key insight that social power and social change depend upon achieving a collective self (which is not to say that all collective selves lead to social change). Although it is often suggested that this emphasis on group power and change was lost after Tajfel's death and cognitive concerns came to dominate Turner's work on self-categorisation theory, we would argue that this incorrect. It is certainly true that Turner was concerned with specifying the processes through which the self is defined in context, how categories become salient, and how category prototypes are formed (Turner, Oakes, Haslam, & McGarty, 1994; Turner & Reynolds, 2010). However, this interest was connected to core claims that social identification is the psychological process that makes group life possible (Turner, 1982), and that groups are a source of social power and, therefore, of social change. These were concerns that became explicit in Turner's later writings (e.g., Turner, 2005; Turner, Reynolds, & Subašic, 2008). They were also the concerns that inspired us both. As an example of this, consider the work of one of us (Reicher) on crowd behaviour.

Lessons From the Crowd

There are many reasons for studying crowds. First, if one can understand how strangers can come together and act spontaneously in cohesive, patterned and meaningful ways, then one can go a long way towards understanding the nature of the social bond. That is, crowds are a particularly propitious site for developing our theoretical understanding of the social world. Second, crowds are of critical importance in shaping broader social relations in society; they constitute highly visible signs which members of certain categories (especially minorities) can use to note where they stand and how they are viewed (Reicher, 2011). That is, crowds play an important role in constituting the empirical reality of the social world. Third, and more straightforwardly, psychologists have systematically misrepresented and pathologised crowds in order to discredit them, to discourage people from joining them, and thereby to prevent them from becoming a source of social change (Barrows, 1981; Nye, 1975). That is, crowd psychology itself is part of the struggle to achieve or prevent change in the social world. If traditional crowd psychology has been a reactionary instrument deployed in the service of the status quo, we need a new crowd psychology to go beyond the status quo. That was the core motivation for developing a social identity model of crowd action.

The contrast between the two approaches, which rest on contrasting notions of identity, is clear. For the traditional approach, derived from Gustave Le Bon's seminal text 'The Crowd' (1895/1947), identity is singular and individual. On entering the mass, people lose identity and hence lose the capacity for judgement. The loss of reason leads to the dominance of emotion. Crowd members are primitive beings capable of great excesses, especially excesses of brutality. If crowds are the voice of the powerless, this psychology silences that voice by representing it as the meaningless ravings of madness.

For the social identity approach (Reicher, 1984, 2001), identity is multiple and social. On entering the crowd, people shift from individual to social identity and hence base their judgements on group values and beliefs. They don't lose reason so much as shift the grounds of reason. Moreover, it is not only through this transformation that the thoughts and actions of crowd members come to be based on social identities. It is also through this alignment that crowd members become empowered to express their values and beliefs even against the disapproval of external agencies. Crowds are one of those rare sites in which we can access the understandings of the oppressed. Not only that: through the power they generate, crowds become the means by which those understandings can come to shape the social world.

This work serves to re-establish the crowd as a key component in social processes rather than an aberration of them. It both explicates and illustrates the link between social identity, power and change. However, at the same time it suggests a rather limited and linear conception of change. That is, the analysis seems to suggest that the views of minorities come fully formed and that change is simply a matter of acquiring the (collective) power to impose those views on majorities. Yet, as Therborn points out, that is not what happens in practice. Revolutions almost never emerge out of revolutionary ideas. Rather, ideas develop out of the ways in which dominant groups respond to initial demands. People are radicalised through collective interactions. Change occurs in the crowd as much as through the crowd (Therborn, 1982).

We have observed exactly this in our own studies of crowds. On many occasions, people who went into an event supporting the police, the state and the existing structure of society came out as opponents. Moreover, across a variety of different gatherings (e.g., student protests, environmentalist actions, political demonstrations, sporting events) we have found that such changes emerged from a similar pattern of interactions (see Drury & Reicher, 2000; Drury, Reicher, & Stott, 2003; Reicher, 1996; Stott & Drury, 2000; Stott, Hutchison, & Drury, 2001). That is, an initially heterogeneous crowd, containing multiple separate groups (some more conciliatory and some more confrontational towards authority), are viewed by the police as homogenously dangerous. On that basis, the police treat the crowd as dangerous, either using tactics such as cordons to contain their movement and stop them doing what they will or else using tactics such as charges to disperse them. This common treatment leads the crowd in turn to unite, to characterise police action as illegitimate and to cohere around the more confrontational elements in the crowd. Such unity gives the crowd the power to challenge the police and hence to confirm the initial police perception of homogenous danger. Out of the resultant confrontational cycle, even those crowd members who started off seeing themselves as 'respectable' and the authorities as 'on their side' emerge with a sense of themselves as 'oppositional' and the authorities as the opposition.

The Elaborated Social Identity Model of crowd action (ESIM; Drury & Reicher, 2009; Reicher, 2011) theorises these phenomena in the following terms. First, social identity is conceptualised as a model of self-in-social-relations. That is, it is not simply a list of attributes, beliefs or values. Rather it is a rich representational system that informs us about how we stand in relation to others in the world. It specifies both our position and how we can/should act given that position.

Second, crowd events are analysed as interactions between crowd members and other parties, notably the police in the examples we have studied. Change occurs to the extent that crowd members who have one understanding of their social position are repositioned by the police. This in turn depends upon both (a) an asymmetry between the way crowd members understand their position and the way it is understood by the police; and (b) the power of the police to impose their understanding in practice and hence to shape the experience of crowd members.

Third, while such repositioning is necessary for crowd members to change their representations of self-in-society (and hence of self and society), it is not sufficient. People still need to make sense of their new experience and to link it to new representational systems. This is more likely to the extent that there are voices in the crowd that help to explain the experience and to provide an explicit model of social relations that fits with lived experience. In other words, change processes are moderated by the availability of influence agents within the crowd.

To be more concrete, in the examples we have been discussing those who came into the crowd with an understanding of the police as being on their side and who then had an experience of being positioned as oppositional are not automatically radicalised. Rather, their experiences alter the extent to which they are amenable to more or less radical voices in the crowd and hence alter the balance of influence between these different voices. Those who construe the police as the enemy and who advocate confrontation become more plausible when their advocacy is articulated with a lived reality of emnity from the police.

There are a couple of general points that we wish to draw from this analysis. One is that, in providing an analysis of when social change occurs, we equally address the question of when it does not occur. Indeed, in addressing the potential for change we equally hope to explain its relative rarity (and hence the general continuity of social relations). A number of conditions can be drawn from the ESIM, all of which need to be fulfilled for change to be a possibility. The first is an asymmetry between the way that crowd members construe their own social position and the way that it is construed by others (the police). The second is that the police must have the inclination and the power to enact their understanding and thereby structure the experience of crowd members. The third is that police practice must be seen as illegitimate by crowd members. The fourth is that there must be influence agents in the crowd to re-represent the social relations between crowd members, the police and authority in general.

This list is not exhaustive but it should already be long enough to explain why change is the exception rather than the rule. Let us just take the first condition. On the whole others see crowd members as they see themselves and hence crowd events confirm rather than destabilise our sense of self and society. It may be that most psychological research on crowds (our own included) concentrates on the exceptions: on confrontational protests or demonstrations; but this leaves aside the great majority of events that are not confrontational (e.g., classes of events such as commemorations and celebrations). These are characteristically organised and reported in such a way as to confirm our sense of self and to consolidate our relationship to the existing social order (e.g., Gillis, 1996; Hobsbawm & Ranger, 1983). In recent years, we have complemented investigations of riot and rancour with analyses of parades and pilgrimages (e.g., Pehrson, Stevenson, Muldoon, & Reicher, in press; Prayag Magh Mela Research Group, 2007). In particular, we have examined the way in which the experience of having one’s identity recognised by others, and the ability to enact one’s group norms and values in alignment with others, leads to a strengthening of existing social identities. This work does not constitute a retreat from our interest in social change. Rather, it serves to instantiate our contention that any understanding of change must equally involve an understanding of how continuity is produced.

The other general point that we wish to make is linked to this, as we shall see shortly. That is, change, where it happens, is not simply a product of circumstances. It has to be mobilised. This might seem like a blindingly obvious observation – after all, if anything is mobilised surely it is collective action – but that makes it all the more surprising that most of the literature in this area (certainly the psychological literature) ignores this point and proceeds as if people are moved to take action when their personal contemplation of the world invokes the necessary emotions and cognitions (see, e.g., Becker, 2012). We have suggested above that the success of any mobilisation depends upon an interaction between the representations on offer from influence sources and the structure of experience. But even this may underplay the role of influence agents. That is, skilled mobilisers seek actively to create the very realities that give credence to their words (see Haslam, Reicher, & Platow, 2011). As an example, Gagnon (2006) relates how, at first, many Serbs rejected attempts by their president, Slobodan Milosevic, to characterise minorities such as Croats as their enemies. Accordingly, it was only after Milosevic organised provocations against these minorities that led them both to fear and to be antagonistic to the majority that Milosevic’s analysis of social reality gained traction.

However, it is not only change that has to be mobilised. The same is true of the lack of social change. When we referred above to the way in which collective events such as celebrations serve to produce continuity, this is not a matter of happenstance. Such events are designed, organised and ordered deliberately to consolidate the social order. That was the radical insight of Hobsbawm and Ranger's (1983) seminal collection on 'The Invention of Tradition'. Those various ceremonies – such as royal weddings and funerals – which, hushed commentators tell us, go back to time immemorial, are actually recent inventions that serve precisely to recruit the weight of history in stabilising institutions that are precarious and contingent in the present (Cannadine, 1983).

Implicit in this work is a recognition that mobilisations for stability arise in the context of mobilisations for change. For example, the need to consolidate the British monarchy in the 18^th century arose out of social contentions that challenged monarchy (e.g., the French revolution; Blanning, 2003). More generally, the point is that we need to think of mobilisations in the plural rather than mobilisation in the singular. In our ESIM analysis we stressed how we generally find multiple sources of influence, some of which challenge the status quo and some of which do not. Even when we consider more ceremonial and commemorative collectivities, we often find contestation either within or between events. Gelvin (1999), for instance, shows how those with different notions of how Syria should be organised in the aftermath of World War I, embodied their visions through differences within their demonstrations. Thus some divided the national elites as participants from the popular masses who were consigned to a role as spectators. Some broke down the division between participant and audience so as to create a less hierarchical conception of the nation. Finally, returning to the example of Milosevic in Serbia, Gagnon's (2006) argument is that his mobilisation of hate against national outgroups can best be understood as an attempt to demobilise attacks on his own regime. We cannot understand either the timing or the form of the Serbian wars without placing them in the context of multiple and contested mobilisations. Thus, whether we look at the interactions within collective events, at the organisation of different types of collective event, or indeed whether we scale up to national conflicts, we find that issues of social reproduction and social change revolve around the question of who people heed out of the many voices that are seeking to mobilise them in different directions.

To reiterate our central contention, siding with authority is not a passive default. Rather, we need to reveal the amount of work involved in making it seem like a default. We also need to ask why and when that work bears fruit. In the following sections, we undertake this task by revisiting the most famous work ever done on the power of authority: Stanley Milgram's 'Obedience' studies.

Obedience to Authority?

As is well known, Milgram's work is part of a triad of large-scale studies that, in the quarter century after World War II, transformed the discipline of social psychology. The immediate reaction to the horrors of that war was to try and find individual characteristics that marked out the perpetrators and distinguished them from the rest of us. However, then Muzafer Sherif, Stanley Milgram and Philip Zimbardo changed the way that we understand the human capacity for 'evil' and the nature of human social behaviour more generally (see Smith & Haslam, 2012). They showed how it is possible to transform the behaviour of ordinary individuals by changing the situations in which they find themselves and the social relationships they have with others. More particularly, by putting individuals in groups and setting up competition with other groups, you can make people prejudiced and hostile (Sherif, 1956). By placing people before a white-coated experimenter who demands that they deliver electric shocks to an innocent victim, you can turn people into torturers (Milgram, 1963). By giving people a uniform and a role, you can get people to abuse and humiliate their helpless wards (Zimbardo, 2007). At least, that is the received wisdom.

We have no issue with the fundamental argument that context – and more specifically, the social relations that structure any given situation – are of fundamental importance in shaping human behaviour. What we do challenge in this work, however, is the way in which social relations are held to shape behaviour. To be specific, we challenge the assumption that behaviour is simply determined by existing relations of power and authority, and hence that behaviour only ever confirms and consolidates existing power/authority relations. This is what has been termed the 'conformity bias' of social psychology (Haslam & Reicher, 2012a; after Moscovici, 1976).

The 'conformity bias' in Sherif's studies is implicit. It is assumed that the boys in his studies would accept the changing social relations imposed upon them by the experimenters: first the division into groups, then the imposition of competitive relations, followed finally by cooperative relations. The focus is then on the way in which they respond to these relations in terms of friendship or hostility, harmony or conflict (Sherif, Harvey, White, Hood, & Sherif, 1961). In Milgram's and Zimbardo's studies the issue of conformity is explicit. Indeed, the whole point of these studies is to show how easily and how absolutely people will embrace the demands placed upon them. As popularly understood, Milgram's studies show how far people will go in obeying the orders of an authority figure. Zimbardo's Stanford Prison Experiment seems to take the argument one step further: you don't even need the presence of an authority. You put people in a role and they will obey role requirements even if those include acts of inhumanity towards one's fellows (see Milgram, 1974; Zimbardo, 2007).

These narratives are further underscored by theoretical accounts that render conformity all but inevitable. Here again, Sherif’s account simply takes it for granted that people will perceive the world and act in terms of the social relations imposed upon them. This is summarised both in the naming and in the specification of his ‘realistic conflict theory’, which posits that positive functional interdependence (cooperation) will lead to harmony and that negative functional interdependence (competition) will lead to conflict. Milgram and Zimbardo once again are more explicit about the inevitability of conformity. Accordingly, having initially proposed many different and nuanced processes to account for his findings (Milgram, 1963), Milgram eventually foregrounds an 'agentic state' explanation. This rests upon the claim that, in the face of authority, people enter a distinctive mental state. They focus purely on how well they comply with the bidding of that authority. All other issues, including the welfare of those affected by their actions, are ignored (Milgram, 1974). Zimbardo is even clearer. It is ‘natural’, he asserts, for humans to take on the roles and the role requirements that are thrust upon them (Haney, Banks, & Zimbardo, 1973, p. 12).

There is a rather basic problem with all this. Such approaches don’t even explain what happens in the researchers’ own studies, let alone what happens in the world outside. For, in every case, evidence of conformity is balanced by evidence of resistance. People challenge authority as much as they accept it. In Sherif’s case, this is evidenced by the fact that there was one whole study (conducted in 1953) where the boys refused to accept attempts to turn them against each other because they rejected the reality that Sherif tried to impose upon them. Sherif simply ignored what happened. He terminated the study forthwith and, although he used some data from the study for publication (Sherif, White, & Harvey, 1955), this aspect of rejection was never mentioned (but see Billig, 1976).

In Milgram’s studies – perhaps the most famous and impactful of all this work not only within psychology, or indeed within the academic world, but in society at large (see Novick, 2000) – conformity is also far from universal. While textbooks tend to report just one variant of his studies – the so-called 'baseline' or 'voice feedback' condition in which 65% of participants continued to apply shocks up to the 450-volt maximum – there were, in fact, over 20 variants on the basic paradigm. In these, the proportion of people who continue to the end varied from 0% to 100% (see Reicher & Haslam, 2012, for more detail).

In Zimbardo’s case (as we have previously discussed at some length; e.g., Haslam & Reicher, 2012a, 2012b; Reicher & Haslam, 2006), participants in the Stanford Prison Experiment resisted their roles and role requirements throughout. Only a minority of Guards became oppressive and Prisoners continued to resist the authority of the Guards to the bitter end.

The point, then, is that all these studies might focus on conformity to the exclusion of resistance, but actually all of them provide evidence of both conformity and resistance. Rather than taking conformity for granted, then, or simply focussing on why people 'obey', we need to enquire into when people conform to existing authorities and when they challenge such authority. It also follows that theoretical accounts that naturalise conformity are inadequate.

Nonetheless, if it is evident that such theories cannot account for the behavioural outcomes of these various studies, it is equally clear that they cannot explain the process of deliberation that goes on during them. And they cannot do this because they fail to characterise the full richness and complexity of the context in which people act. In order to flesh out these points, we shall, as intimated, concentrate on Milgram’s work, not only because of its aforementioned fame, but also because Milgram (in contrast to Zimbardo) provides very full material on which basis one can interrogate all aspects of his studies. This means that what cannot be found in the published works is available through the Milgram archive at Yale University.

When one turns to this material, what is perhaps most striking about the ‘obedience’ focus and the ‘agentic state’ explanation is the way in which these focus exclusively on the relationship between participants and the experimenter. But what makes the Milgram paradigm so powerful is the fact that there are three parties: the experimenter, the participant-teacher, and the confederate-learner. This allows for multiple relationships, most obviously between the participant and the learner as well as between the participant and the experimenter (Gibson, in press). The whole drama of the studies lies in the way that participants are torn between these different relationships and the different obligations they entail: should they respect their obligations to the scientist and continue shocking or should they respect their obligations to a fellow citizen and stop? Indeed, it is precisely this tension that leads to all the agonising, the rationalising, the nervous laughter and much else that many commentators (Milgram included) remark upon and that is so obvious from looking at films of Milgram’s experimental sessions (Milgram, 1965a; see also Millard, in press).

The agentic state explanation ignores all this. If one were to take it seriously, it would suggest that the participants are serene, that the cries of the learner have no hold on them and that they are only interested in doing the experimenter’s bidding. In fact, psychologically, it writes the learner out of the picture. He is simply not present in the psychic calculus of the agentic state. By contrast, when we consider the plenitude of the context, one where the participant is caught between two voices, we can relate the question of whether participants conform or resist to the question of whom they heed: the experimenter or the learner.

The Engaged Follower

It is worth stressing that, in this regard, the social structure of Milgram’s experimental set-up is a particularly powerful analogue of social life in general. Unlike most experiments, where people have to make decisions in isolation from others (Haslam & McGarty, 2001) or else, at best, when exposed to a single source of influence, here we have people trying to make up their minds while assailed by loud and strident voices from different sides telling them to do different things. This, as we argued above, is generally the position that prevails when we are making the most significant decisions in our lives about whether to conform or not to conform. This is what makes Milgram’s work so important not only for the study of extreme behaviour but for the study of social life in general.

There are several indicators that, in addition to the experimenter’s voice, the learner’s voice plays a key role in the obedience studies. The first is that, in Milgram’s original pilots, where there was no feedback at all from the learner, all the participants complied with the experimenter (Russell, 2011). This contrasts to the 65% compliance observed in the baseline ‘voice feedback’ condition where the learner is heard to express pain, to speak of a heart condition, to demand to be released, and ultimately to fall into an ominous silence. Clearly, the voice of the learner makes a big difference to how the participant behaves. What is more, as Packer (2008) demonstrates, the precise voltage level at which people are most likely to break off (150V and 315V) relate directly to what the learner says at these points. The 150V point is the first one where the learner demands to be let out of the study (thus introducing a new obligation to counter that of the experimenter), while the 315V point is the first one where the learner states categorically that he is no longer part of the experiment (thus challenging the participant’s obligation to continue with the experiment). Thus, the precise way that learners use their voices makes a difference.

What then decides whom one listens to: whether one complies with the experimenter and resists the learner, or else complies with the learner and resists the experimenter? We have suggested that it is largely a matter of whom one identifies with. Does one define oneself as part of a scientific enterprise of which the experimenter is representative, so that one sees him as having the right to say what should and should not be done? Or does one identify oneself as a member of the general community of which the learner is representative so that one sees him as having the right to say what is and is not appropriate? Depending on one’s identification, one set of demands and moral obligations takes precedence over the other. However, whatever one decides to do, behaviour does not follow from ignorance of the effects of one’s actions, but from a belief that one has made the right decision (Reicher, Haslam, & Rath, 2008). As we have argued elsewhere (Haslam & Reicher, 2012b), those who deliver the shocks in Milgram studies are not the thoughtless bureaucrats that they are typically portrayed as being but rather engaged followers.

The critical point here is that people do not deliver shocks simply because that is what the authority figure wants them to do, but rather because that is what they think is right to do and what they think they ought to do. It is for this reason that we are uncomfortable with the term 'obedience' to describe what is going on in Milgram's studies (and why we generally place the term in inverted commas to indicate that we consider the term to be theoretically problematic; see also Lutsky, 1995; Mandel, 1998). In what follows, we therefore use the more neutral term 'compliance' to describe situations where participants go along with the instructions to deliver shocks, without presupposing their underlying mental state.

Support for the 'engaged follower' conceptualization comes from a reanalysis of Milgram’s published findings. In this we simply gave people Milgram’s own descriptions of the many variants of his studies. We then asked our samples (both expert psychologists and naïve students) to take the perspective of Milgram's participant and indicate how likely they would have been to identify with (a) the experimenter and (b) the learner in each of these variants. We then correlated estimated identification with observed levels of compliance in each variant. The results revealed a strong negative correlation between identification with learner and observed compliance, a strong positive correlation between identification with experimenter and observed compliance, and an even stronger correlation between relative identification (identification with experimenter minus identification with learner) and observed compliance (Reicher, Haslam, & Smith, 2012).

Further support for this analysis comes from a more detailed examination of the interactions between participants and experimenter that either produced or undermined compliance. Thus, as is well known, Milgram scripted a sequence of four prompts (prods) for the experimenter to use if and when participants proved reluctant to continue. Three of these are either requests or justifications linked to the scientific importance of the participants’ acts: “Please continue”, “The experiment requires you to continue”, “It is absolutely essential that you continue”. Only the fourth and final prompt – “You have no other choice, you must go on” – constitutes a direct order. The interesting finding is that whenever Prompt 4, the order, is used, people typically refuse to continue (Milgram, 1974, see also Burger, 2009).

Now it is possible that this is simply a sequence effect rather than having anything to do with the content of the prod. Perhaps people disobey Prod 4 simply because it is the last prod and by then they would ignore any appeal. So we have recently used an experimental analogue of the Milgram paradigm to disaggregate order and sequence, and we still find that giving orders leads to increased refusal to continue (Haslam, Reicher, & Birney, in press).

So what is the problem with orders? We suggest that it is precisely that they undermine identification with the source. For, whereas requests or justifications imply (or at least don’t undermine) a sense that the source and target (experimenter and teacher) are common participants in the same scientific enterprise, giving orders imposes the experimenter’s views upon the teacher and hence places the former in opposition to the latter. It destroys any sense that ‘we are in this together’ or that ‘we hold the same priorities’. It is the explicit language of otherness, of ‘I’ versus ‘you’ (Reicher & Haslam, 2011). This is, perhaps, more obvious if we return for a moment to a consideration of the role of authorities and orders in the context of crowds. Not only is it ineffective for the police to order people around, but often collective conflict begins precisely at the moment when the police seek to impose their priorities over and against those of crowd members (Reicher, 1984; Reicher & Stott, 2011).

Of course this evidence is circumstantial and alternative explanations are possible. For the present, we can say no more than that the ‘engaged follower’ perspective (whereby compliance with the instructions of a particular source depends upon explicit identification with that source and with the category they represent) provides a promising and parsimonious account of many aspects of Milgram’s findings. But for now, we want to pick up another point that flows from our analysis of Milgram’s prompts. It takes us to the crux of our argument concerning social change and social stability: compliance in the Milgram experiments and, we suggest, the identification that underlies it, does not just happen. It is something that the experimenter (and Milgram who stands behind him) has to work hard to encourage. Indeed, Russell (in press) provides a powerful analysis of the great efforts that Milgram went to in order to secure the compliance of the many parties who were essential to the experiments’ staging (his funders, his associates, his department, and his university). However, here we concentrate only on how he created identification and compliance in his participants.

Mobilising Identification and Compliance in Milgram's Studies

There are many ways in which Milgram worked at producing identification with the experimenter in order to encourage compliance with his demands. Like the prompts that he scripted to be used in case of dissent, a number of them are constants across all the variants of the study. Others form part of the variability that Milgram explored across different conditions and explain variation in participants’ willingness to proceed to the end of the study. For instance, Milgram well understood the spatiality of identification (he was concerned with space in many ways, and one of his lesser known areas of work concerned social representations of cities; see Milgram, 1984). He recognised that putting the experimenter and the teacher in a room together (and separate from the learner) would enhance group formation between them (Milgram, 1965b) and encourage compliance. Conversely, the closer the experimenter comes to the learner, the more identification with the learner increases (and identification with the experimenter decreases) and the more this encourages resistance.

For now, though, we will concentrate on two aspects of the studies to which Milgram devoted particular care. One has to do with the nature of the experimenter and the other with the nature of the experiment. Milgram exerted great care in the selection of his confederates. The learner, James McDonough, was chosen because, although he was a poor actor, he was 'perfect as a victim' being “mild and submissive; not at all academic” (quotes from the Milgram archive cited in Russell, 2011, p. 159). In other words, he lacked the authority to make demands of the participant. By contrast, the experimenter, John Williams, was chosen for his authority – “a stern, intellectual looking man” who acted “in a cold austere manner” (Russell, 2011, p. 159). Indeed, Milgram rejected an earlier choice for the experimenter, his young research assistant Alan Elms, because he lacked such authority. Of course, if people automatically conformed to authority all these concerns would be irrelevant. They raise the obvious question of what makes someone authoritative. One answer might be that they possess certain generic qualities that characterise all authority figures (e.g., Lord, Foti, & Phillips, 1982). However, it is hard to see how qualities like 'cold' and 'austere' would enhance the authority of a politician, a liberal or (say) a feminist leader. These qualities do, however, mesh with stereotypes of the scientist. Hence, it is more convincing to suggest that Milgram sought to establish the authority of his experimenter by ensuring that he was representative of the relevant category (see Haslam, Reicher, & Platow, 2011).

However, it would be of little use for participants to embrace the experimenter as scientist if they were not also committed to the science; and it is in this regard that Milgram's efforts are, perhaps, most apparent. Indeed, as Russell (2011) relates, here Milgram was almost obsessive, addressing every detail of the studies in order to make them seem impressive and technically sophisticated. This is exemplified in the trouble he went to in designing the shock generator as an imposing and “thoroughly professional” piece of scientific apparatus (Russell, 2011, p. 158).

But Milgram didn't just want participants to construe the science as professional but also as morally good. The announcement that he placed in a local newspaper asks for "men to help us complete a scientific study of memory and learning" (Milgram, 1974, p. 15). When volunteers arrived at Milgram's laboratory, they were again briefed that the studies were about memory and learning and, more particularly, about "how much punishment is best for learning" (1974, p. 18). In this way, punishment was turned from something bad to something good for the individual. To invoke an analogy that Milgram himself used in the briefing - "the expectation is that spanking, a form of punishment, will teach the child to remember better, will teach him to learn more effectively" (ibid.) - the message is 'spare the rod, spoil the child' – but with a cognitive rather than a moral focus.

Equally, the experiments were turned from something socially pernicious to something socially progressive. Participants were told that they would contribute to a better understanding of learning, and hence would inform much-needed efforts to improve learning. Moreover, the briefing stresses how the studies provide a unique contribution to understanding (the effect of punishment on) learning, because “almost no truly scientific studies have been made of it in human beings” – note the stress on the “truly scientific” nature of these studies as well as on the social benefits (Milgram, 1963, p. 373).

To contemporary sensibilities, these claims are all the more striking given that, since Baumrind's seminal paper in 1964, Milgram's studies are often held up as exemplars of pernicious unethical research. That is, whatever their impact on society, the stress imposed on the participants and its potentially harmful longer-term consequences is seen to render the experiments quite unacceptable. Yet, there is an ironic twist here that further reinforces our general argument concerning Milgram's active construction of identification and compliance. He did not take the criticisms of Baumrind (1964) and others lying down. He argued that most of his participants were very positive about having participated, that most hadn't been bothered by what they did, and that most felt they had learned something positive through their participation (Milgram, 1964, 1974). One might then argue that, even if a majority were fine, a sizeable minority were not. A third of participants said that they had been bothered by their participation, and 7% said they had been bothered quite a bit. Yet, even accepting Milgram's figures and his focus on the contented majority, we are led to ask exactly why these participants were so content.

Here the Milgram archive provides us with some fascinating insights. Specifically, we have recently conducted a systematic investigation of Box 44 in this archive (Haslam, Reicher, Millard, & McDonald, 2013), a box that contains comments made by participants after they had completed Milgram’s post-experimental questionnaire. This ten-item questionnaire (which is the source of the figures we have just been citing) was sent out to people after all the studies had been completed, along with a detailed five-page report (or 'dehoax' as Milgram terms it in the 1964 paper). In our analysis, we systematically coded this material so as to assess participants’ engagement with the scientific project, their level of post-experimental distress and the extent of their obedience. What we find is that engagement with the science is associated with greater post-experimental support for the studies and lower post-experimental distress. However, this engagement bears no relation to participants’ levels of stress during the study itself.

What this suggests is that, in effect, when the research is seen as worthwhile, people can subsequently feel good about suffering for a noble cause, and hence also be willing to participate (or to recommend others to participate) in future studies. This accords with our 'engaged follower' perspective. Yet, what is even more striking to us is how Milgram promotes the nobility of (and identification with) his scientific cause. In the 'dehoax' document accompanying the post-experimental questionnaire, and in terms that clearly echo his earlier claims about memory research in the pre-experimental briefing, Milgram states that: “The problem of obedience to authority may well be the crucial issue of our time. The experiments you took part in represent the first efforts to understand the phenomenon in an objective, scientific manner” (Milgram, n.d., p. 5). Thus, he may reveal that the scientific cause is not quite what participants thought it was, but it remains as noble (and as worthy of support) as ever.

In short, Milgram never stops. He is tireless in his efforts to produce compliance both within and beyond the studies. Whether it is a matter of applying electric shocks or of supporting studies in which electric shocks are applied, we can with confidence reiterate our key contention. Compliance never just happens. It has to be mobilised.

Conclusion

In essence, the argument of this paper, paradoxical as it might appear, is that we will never understand social change, or appreciate the importance of social change, by only ever studying when change occurs. To do so plays into an underlying exceptionalism that accepts the status quo as a default and that only treats any departure from this position as noteworthy.

The best way to foreground change in our discipline, then, is to problematise the production and reproduction of 'normality'. For, by identifying the work that must be done to keep things the same, we render the world contingent; we open up the possibilities of different outcomes and of different paths. By freeing the theoretical imagination, we liberate our capacity to envisage and create different worlds (Turner, 2006).

To be more concrete, we have argued that both stability and change involve processes of mobilisation that centre on identification, authority and influence. Or, to put it slightly differently, stability and change are competing projects of mobilisation centring on different identities, different sources of identity-based authority and different forms of identity-based content. As we argued using the example of Milosevic in the former Yugoslavia, every mobilisation of one type of social action is simultaneously the demobilisation of other types of social action.

How, then, do we create or else disrupt the creation of particular identities? How do particular voices amid the cacophony that surrounds us become authoritative? How do some social projects become influential? These, we suggest, are the key questions for a progressive social psychology – 'a science of movement' to reprise Moscovici's resonant call– which can problematise social reproduction and envisage social change. The start, though, is to recognise that everyone is involved in the frantic activity of category creation. Yet some are better than others at hiding it beneath the surface. That is the source of their apparent serenity, their stability and their power.