From collective representations to social imaginaries: How society represents itself to itself Open Access

This paper aims to reconsider the relevance of Cornelius Castoriadis’ concept of the social imaginary as a way of re-invigorating the study of ‘collective representations’ within the social sciences. The paper begins with a review of how social thought has been understood, from Durkheim's writing on collective representations onwards. A case is made for the utility of Castoriadis’ concept of social imaginary significations in restoring their centrality within the social sciences. In contrast to viewing social thought as ‘super-structural’ ideology, or as a constellation of social attitudes and opinions commonly shared, Castoriadis’ view of social imaginary significations articulates a view of social representations as under-determined by, though never articulated without reference to, social morphology. However the imaginative capacity of individuals and society to generate this surplus of signifiers is understood, Castoriadis’ thinking allies the study of social imaginary significations with the study of social morphology without reducing either to mere representations of the other. What is needed now is a methodology suited to the empirical exploration of the network of significations that constitute the ‘imaginary institution of society’.

Sociology can be seen as a particularly structured attempt to understand modern society. It is through the writings of Durkheim, in particular, that this attempt began to confront the fact that modern society also seeks to understand itself, to represent itself to itself as it were, and that these representations constitute autonomous social facts as much as any other social phenomena (Durkheim, 1898, p. 299). Societies and the various groups and collectivities that make up society seek to make sense of [represent] themselves and in the process of such ‘sense-making’, they initiate change and maintain continuity (Cormack, 1996, pp. 100–101). As several writers have since noted, as society becomes more complex the ways in which individuals, groups and communities make sense of, and represent society, its institutions and its social relations have become in turn more complex, more conflicted and more contingent. Contemporary collective representations are flooded by a ‘superfluity’ of signs and signifiers generated by this complexity, moving from the earlier ‘hegemonic’ collective representations of first or classical modernity to the diverse, polemical social representations characterising late modernity (Moscovici, 1988, p. 221). This change is reflected through the wider range of identities – with their associated practices and preferences – that are expressed and performed across a greater number of settings (Giesen & Seyfert, 2016). Collective identities are realised through a growing number of institutional relations that puts a strain upon any shared representation that arose in earlier socio-historical structures (Schmaus, 2009, p. 122). Nevertheless, collective strivings after meaning have not been abandoned. Representations of society and its social relations remain ‘at the heart of society’ (Falasca-Zamponi, 2014, p. 48). But although the nature of social thought has been a concern to the social sciences for some considerable time (Gafijczuk, 2005), they no longer seem to form a central focus in contemporary sociology. Interest has always been sporadic rather than sustained and possibly as a result, there remain enduring differences in how the subject is approached. The issue can be simply stated, namely that there is no consensus of how ideas, beliefs, representations and understandings about society, its divisions, institutions, practices and relations are constituted nor how such representations should best be understood and investigated. The importance of the question rests, in no small measure, on the status granted to social representations or social thought as ‘social facts’. Should social thought be accorded the same status, for example, as those more traditional social facts, such as social agency, class, institutions, identities and relationships? And are such ‘facts’ not themselves constituted as ‘social imaginaries’? While there exist a number of well-articulated methods of writing about, investigating and analysing these latter ‘facts’, explorations of the ‘facticity’ of social thought has been relatively constrained – even eschewed – as a topic of sociological inquiry . The aim of this paper is to bring back to prominence what Serge Moscovici called the ‘lost concept’ of representations (Moscovici, 2008/1961, p. 32) and in particular its formulation by Castoriadis as ‘social imaginary significations’ (Castoriadis, 1987).

Social representations can be considered the key elements of social thought. To address the broader question of what constitutes ‘social thought’¹ it is sensible to turn first to Durkheim and his writings on ‘collective consciousness’ and ‘collective representations’. Drawing on his Suicide (SCD, 1893/1953), Rules of Sociological Method (RSM, 1895/1982), Individual and Collective Representations (Représentations individuelles et collectives, ICR,1898) and The Elementary Forms of the Religious Life (EF, 1912/2001) as well as various commentaries on these works, it is possible to identify several key claims that Durkheim made about social thought. These claims are that (i) ‘collective representations’ exist in the collective, independent of individual consciousness (Durkheim, RSM, 1982, p. 59); that (ii) they derive from but cannot be reduced to the ‘substratum’ of individual bodies and minds (Durkheim, ICR, 1898, p. 299); that (iii) while collective representations arise out of particular historical circumstances they acquire an autonomy within society that increases as society itself becomes more complex, forming what might be called ‘second degree’ or ‘second order’ social phenomena (Durkheim, ICR, 1898, p. 300; Peristiany, 1906/1974, p. xxi; Sawyer, 2002, p. 244); that (iv) collective representations ‘are the heart of social reality’ and not mere epiphenomena (Stedman Jones, 1995, p. 39; Turner, 1995, p. 7); and lastly that collective representations are real things, and ‘not mere verbal entities’ (Durkheim, SCD, 1893/1953, p. 310).

This last point is perhaps the most difficult to come to terms with, since Durkheim was quite prepared to consider as collective representations externally produced, statistical facts – as indeed he tried to do in his study of suicide (1893/1953). But, as Cormack has pointed out, what Durkheim did was appropriate state statistics (a relatively novel form of representing the social or collective that was being realised at that time by the state itself²) to construct a supplementary narrative to describe the social reality that was itself embedded in the production of the ‘facts’ of its existence (Cormack, 1996, p. 98). Indeed, as Durkheim himself noted, these phenomena were sufficiently distinct as to demand a new branch of sociology, one not yet prepared, ‘to research the laws governing collective thought’ (‘idéation collective’) (Durkheim, ICR, 1898, p. 301). An inescapable tension seemed to exist within his own work between the apparent facticity of statistics, their interpretation and the analysis of those interpretations in short, their second-order organisation (Némedi, 1995, p. 51).

In The elementary forms of religious life, Durkheim came closest to formulating a model by which ‘collective representations’ could be considered not so much ‘things in themselves’ but more a special kind of grammar (Némedi, 2000, p. 92). He writes: ‘the collective consciousness is the highest form of psychic life since it is a consciousness of consciousnesses … it alone can provide the mind with frameworks that apply to the totality’ (Durkheim, EF, 1912/2001, p. 340). As society and social life become more complex, he suggests, the ‘lexicon’ of collective representations expands, becoming ‘part of a much larger whole … with vague and infinitely expandable borders’ organised in ways that go beyond the determinants of any specific socio-historically located place and time, acquiring a more or less autonomous structure of their own (Durkheim, EF, 1912/2001, pp. 340–341). In short the social morphology that helped initiate collective representations gets left behind as the representations themselves develop their own ‘organising’ principles beyond the constraints of morphologically particular social lives and morphologically particular social relations.

Durkheim's re-framing of collective representations into a more semiotic framework reached its limits at the end of this, his last book. Having formulated a new view of collective representations, Durkheim left a legacy that needed other minds to develop. His ideas about collective consciousness and collective representations however were not taken forward within the discipline (Clément, 2010, p. 64). The key figure who did take on this task was not a sociologist, but a social psychologist, Serge Moscovici. In his book Psychoanalysis, Son Image et Son Public [Psychoanalysis, Its Image and Its Public] Moscovici sought to recover the ‘lost concept’ of what he re-named ‘social representations’ (Moscovici, 2008/1961). Moscovici returned to the question that Durkheim had worked hard to resolve, namely the ‘exteriority’ of collective thought and its distinction from the attitudes, beliefs and opinions held by particular individuals (Moscovici, 2008, p. 29). The fact that Moscovici was a social psychologist and not a sociologist enabled him to be less exercised about this distinction; he ended up considering such questions ‘sterile’ (2008, p. 29). His own disciplinary background lead to a general re-direction of the concept towards the ‘experimental’ analysis of social thought. This involved accepting a degree of individualisation in the processes by which people formed ‘social representations’. Nevertheless he represented himself as a fellow ‘social scientist’ working alongside anthropologists, social historians and sociologists in this area (Moscovici, 1988, p. 213). Thus he was not made ill at ease by the fact that although:

Moscovici had considered but abandoned the idea of using a different term – that of ‘schema’ as developed by the Cambridge psychologist, Frederick Bartlett, in his study of collective memory (Bartlett, 1932). Part of his reason for not using Bartlett's term was that he considered its focus both too narrow and too individualistic. Steering a path between criticisms of vagueness and disciplinary drift from fellow psychologists (Jahoda, 1988; Parker, 1987) and the limited engagement of sociologists with his reprise of a ‘lost’ sociological concept, Moscovici outlined what was in effect a kind of intermediate position between the ‘interior’ and the ‘exterior’. While he recognised the fundamentally ‘social’ orientation of ‘social representations’ he considered them to be both less ‘total’ and more ‘pluralistic’ than the collective representations that Durkheim envisioned – in large part because of the greater degree of individualisation and heterogeneity characterising late modern society (Gillespie, 2008, p. 375). Hence the partial reprise of Durkheim's ‘lost concept’ was now framed as ‘social’ rather than as ‘collective’ representations. Since Moscovici considered social representations to be subject to individual ‘cognitive processing’ and hence capable of being investigated by ‘experimental’ as well as by interpretive, observational methods he believed it possible to explore their exteriority through the methods of individual inquiry (Moscovici, 1988, p. 223). His willingness to explore not just the thoughts but the thinking process itself contrasts with Durkheim's insistence upon the ‘exteriority’ of collective consciousness that cannot be reduced to any putative ‘averaging’ of the work of individual consciousnesses. By thus excluding any attempt to study individual consciousnesses, Moscovici argued Durkheim's terminology ended up lacking definitional and operational clarity, leaving ‘collective representations’ mere ‘figments of thought’ rather than ‘real entities’ (Moscovici, 1988, p. 224). His purpose in studying social representations was ‘to determine the link between human psychology and modern social and cultural trends’ (Moscovici, 1988, p. 225). But he wanted to do more than merely examine the ‘linkages’; he wanted, like Durkheim, to explore how social representations ‘do things’ – how they form ‘ways of world making’ (Moscovici, 1988, p. 231).

At the same time, he wanted to extend his methods of enquiry to a broader canvas beyond purely ‘individualistic’ approaches. He wrote:

While it is difficult not to feel sympathetic to Moscovici's intentions, the result has been a body of work carried out almost exclusively within academic social psychology, employing methods that are not so very different from that espoused by Bartlett in the 1930s.³ While Moscovici achieved a major place for ‘social representations’ within the sub-discipline of social psychology, this way of studying the ‘social mind’ has focused upon the study of individual minds and their ‘representational’ work (Wachelke, 2013, p. 133). Many researchers within the field acknowledge the crucial importance of the social in determining who thinks what and to what ends (Howarth, 2006; Voelklein & Howarth, 2005), but still the paradigm remains of studying social representations in and through the activity of individual minds (Abric, 2001; Wagner & Hayes, 2005).

An alternative route for the study of social thought emerged at around the same time (the 1960s) part of whose origins lay in the equally ‘individualistic’ field of psychoanalysis. This was in the work of the French psycho-analyst, Jacques Lacan. While Lacan did not employ terms such as collective or social representations, what he did was to outline a model of thinking about ‘conscious’ and ‘unconscious’ thought that argued for the decentring of the ‘thinker’ or ‘ego’. Rather than approach the ego as representative of an individual, agentic self, Lacan framed the ego as more a vehicle or ‘speaking subject’ who spoke for and through an unconscious that was structured as a language (Lacan, 2013). The grammar of the unconscious was structured by what Lacan called the ‘Other’. The unconscious other represented both personally significant others and a generic ‘Other’ arising from the impersonal field of collective others (Lacan, 2006). Lacan's particular re-reading of Freud reached a much wider audience than trainee and fellow analysts, through a series of seminars that he gave, beginning in 1961 and continuing pretty much until his death in 1980. Those attending included a wide variety of French intellectuals, among whom were Louis Althusser, Cornelius Castoriadis and Jacques Derrida (Roudinesco, 1990). What follows is less an account of Lacan's own theory of social thought as an account of others’ interpretations of his thought, particularly that of Althusser. Lacan's importance for the social sciences, I suggest, lies less in his own writings (which can be often quite abstruse) than in his re-directing the thinking of others, and the nature of the connection between the ‘interiority’ and ‘exteriority’ of our personal and social worlds.

Althusser was affected by Lacan's thought, particularly in his rethinking of the problem of ideology. Ideology can be understood as a particular form of social thought or collective representation. What sets it apart from the general category of social thought is its structuring as a set of beliefs about the nature of the social world, based especially although not exclusively on the political organisation of society (Freeden, 2003, p. 4; Van Dijk, 1998, p. 28). It is the presence of ‘imposition’ and ‘power’ that marks ideology as a distinctive form of collective representation, with all that that implies about its role in furthering sectional interests within society. While the role played by sectional interests in shaping social thought will be explored later on in this paper, the term ‘ideology’ is raised here as a way of outlining the contribution of Lacanian thought to understanding the ‘exteriority’ of social thought.

Drawing upon his reading of Lacan, Althusser proposed that ideology could be understood as deriving from largely unconscious structures pre-existing the subject (or ego) whose existence as a social being is ‘spoken’ or ‘interpellated’, and hence ‘created’ by, ‘this collective other’ (Althusser, 1996). This structure is formed by what Lacan called the ‘Law of the Father’, that Althusser re-formulated as the ‘Law of Culture’ or the ‘Law of the Symbolic’ (Althusser, 1996, p. 29). While Althusser struggled to formalise his understanding of ideology through its location within the [Lacanian] unconscious, that is through the rules of language and the structures of kinship pre-existing the formation of the individual subject, his understanding of ideology was framed by this idea of the ‘symbolic’ order. Following Lacan's lead, Althusser argued that the subject or ego was called up – or brought into existence as a social being – by and within a matrix of meanings whose existence was pre-established by the structure of language and the kinship structures of society. Althusser accepted Lacan's account of the formation of the ego as an incoherent ‘subject’ whose first, primitive identity was based upon an imaginary, fashioned through the mirror images of the actual embodied self and other actual embodied selves (Lacan, 2001). Upon this unruly, incoherent and unformed ‘imaginary’ of the self/other was realised a social self whose existence was realised within the symbolic order. These constituted the ideological structures ordered by language and by the structures of kinship – that is, through gender, maternal and paternal roles, the place of the child, and so on (Althusser, 1996, p. 30).

As others have noted, Althusser failed to progress his thinking on ideology (Higgins, 1981/1982; Ranciere, 1994; Williams, 1993). He became disillusioned with Lacan as a thinker and with Lacanian thought – particularly Lacan's framing of the unconscious as constituted through language and law (Althusser, 1996, p. 91). Although he continued to believe in the value of linking Marxist and Freudian thought, especially in their contribution to both decentring the subject [Freud] and decentring society [Marx] (Althusser, 1996, pp. 120–121), he felt it impossible to separate the ‘ideological’ from the ‘scientifically true’ by such means. Indeed he seems to have reverted to thinking of the unconscious as the repository of a psychical reality, separate from, if still ‘propping on the external and social world simultaneously’ (Althusser, 1996, p. 122). Ideology as externally structured’ and internally constituted required a contrast with ‘the real’ and it was at this point that Althusser seemed to stumble. Why after all should there be a ‘social reality’ behind these structures that is capable of being rendered the objective of conscious struggle? How might it be ‘discoverable’ let alone ‘contestable’? In one critic's words:

Lacan and Althusser both emphasised the role of the symbolic order (aka the ‘law of the father’) in the genesis of ideology, and by implication the founding principle for most forms of collective representation. Another social theorist equally influenced by psychoanalysis who had also attended Lacan's seminars was Cornelius Castoriadis. Although directed more towards the political than the social, Castoriadis’ writings, particularly as so far as they concern ‘the social imaginary’ and ‘social imaginary institutions’ provide both a link to, and another perspective on the late Durkheimian project – ‘to explain how society ‘passes’ into people's minds’ (Ogien, 2016, p. 9).

Well over three decades have passed since Thompson criticised sociology's ‘unjustified neglect’ of the work of Cornelius Castoriadis (Thompson, 1982, p. 659). Arguably not much has changed (Memos, 2016). Although the ‘social imaginaries field has expanded across disciplines and beyond the academy’, these developments have as yet had limited impact on sociology (Adams, Blokker, Doyle, Krummel, & Smith, 2015, p. 15). Inverting the Lacanian order, Castoriadis placed the symbolic organisation of social thought as a constraining factor that is forever out manoeuvred by the flux and fertility of the imaginary in both its individual and collective forms. In writing of the imaginary, Castoriadis distinguished between the radical ‘social instituting’ imaginary that drew upon the creative capacity of the collective to generate meanings and the radical imagination, which he located as a feature of humanity, an inherent capacity of our species being, equivalent to that of language (Castoriadis, 2007, p. 203). This merger ‘between the radical imagination of the singular psyche and the social instituting imaginary of the anonymous collective’ institutes a network of meanings whose form is imposed by the social imaginary – that is the social-historical order that exists outside of the individual psyche – but never over-determined (Smith, 2014, p. 85).

Castoriadis’ writings share some common features with those of Lacan's – including the ordering of the psychic world into the symbolic, the imaginary and the real. But while Lacan saw the subject as ‘spoken’ by the symbolic order, Castoriadis rejected what he saw as yet another kind of structural determinism.⁴ The subject's desire to make sense, to generate meanings he saw as a psychic developmental necessity in the progress towards becoming a social being. Rather than passively absorbing the ‘laws of culture’ to satisfy this need, he argued that this developmental process creates the very conditions for a ‘reflective subjectivity’ that renders the individual ‘capable of calling into question the imaginary significations of […] society’ (Castoriadis, 2007, p. 219). Hence the symbolic order is always capable of being subverted and its range of meanings – its collective representations – never totally constraining the individual's or society's radical imagination/imaginary.⁵ By positing active individual and collective radical imaginaries/imaginations, Castoriadis argued that representational processes retain an inherent open-ness that external systems of coercion can never fully eradicate.

The concept of the ‘imaginary’ serves two functions for Castoriadis. As the radical imagination it represents the core ‘pre-representational’ capacity of human beings, qua species, to generate not just sets of meanings or schema, but to reproduce social imaginary significations (Castoriadis, 1987, p. 369). This capacity of the radical imagination renders any symbolic delineation of the social incapable of social capture since each social imaginary signification will always exceed the purely ‘symbolic’ or ‘discursive’ order. At the base of any ‘social schema’, ‘ideology’ or ‘social representation’, there exists a surplus of signifiers that create the conditions for both contesting each representation, schema or ideology and constructing an alternative meaning or representation. While Castoriadis acknowledges that social influences shape particular social imaginary significations such as ‘citizen’, ‘family’ or ‘justice’, he stresses their unavoidable under-determination by the plethora of significations that surround each and every possible representation. This he attributes in large part to the intrinsic power of human beings’ radical imagination to exceed the purely symbolic order. It is, he claims, ‘what distinguishes the human psyche from the animal psyche’, ‘the ability to formulate what is not there, to perceive in just anything what is not there … a flux, a representational spontaneity’ (Castoriadis, 2007, p. 203).

The extension into social spaces of the radical imagination produces sources of meaning and a range of significations that exist in what Durkheim would call exterior space, and does so necessarily so that the individual ‘psyche’ can survive – and thrive – as a social being. This inter-change between the inherent radical imagination of the individual and the already existing radical imaginary of society makes meaning making central to the development of social being, while ensuring that the symbolic organisation of social thought – the network of social imaginary significations – is never overdetermined by social morphology. At the same time, without such a network – without a framework for making sense of society and social relations, the individual would remain, in Castoriadis’ words ‘a wailing new-born monster … unfit for life’ (Castoriadis, 2007, p. 99). Paradoxically then, social imaginary significations draw upon social relations to replace or counteract the ‘solipsistic power’ of the ‘new-born monster’, conferring meaning, making things matter, in effect realising society and thereby realising (not simply interpellating) the person.

Such a position raises questions not so much of causality as of power. What confers power on social representations to ensure that they remain part of the social-historical structures from which the ‘over-imaginative’ subject of necessity draws in the process of becoming a social being? How, in the process of becoming a member of a community, a person with a social identity and a someone with a sense of belonging somewhere, does the social determination of being leave open the possibility of being otherwise? Although Castoriadis draws upon a psychoanalytic model to explain why there is a desire, a need, to have meaning and to ascribe meaning, the question remains why these particular representations and why these particular meanings and what constitutes the capacity to exceed or escape those meanings. Mannheim, in his book on Ideology and Utopia, has argued that it was only with the waning of the middle ages that the possibility arose of there being ‘ideologies’ to reflect upon and to choose from (Mannheim, 1936). But, even if one accepts his periodisation, the question remains how the more fragmented social power that emerged in the slow progress of ‘modernity’ saw some representations achieve greater influence than others (Mannheim, 1936). This leads to the question of what Durkheim called ‘the social morphology’ of collective representations – that is the relationship of social representations to the tangible, material substratum of society (Durkheim, 1898, 1978). In the next section of this paper, I consider how power and its presence in the already existing social relations of society impacts upon the range of meanings constituted by ‘social thought’ and the limits of that power.

With the demise of Althusserian analyses of ideology, in the 1980s, one of the most significant advances in the field was that made by Steven Lukes, more clearly and more reflexively in the second edition of his book Power: A radical view than in the first (Lukes, 1974, 2005). In the first edition, Lukes had argued for a three dimensional model of power whereby social thought is used by social agents to reproduce positions of relative advantage or ‘relations of domination’. In the second edition, issued some three decades later, he recognises that people's concepts of what constitutes their interests, what matters to them, are more complex, multiply determined, and irreducible to any single social position they may occupy, however ‘dominant’ that may seem to be (Lukes, 2005, p. 145). All systems of thought express relations of power, all possess an implicit meaning about the distribution of power and of ‘interests’ but those interests are generally multi-dimensional and multi-directional. The choice of meanings attributed to any social representation reflects not just one group or social agent's view of ‘the real’ but those that are generally most beneficial in securing their social identity and social position.

Arguably, Lukes still is keeping close to social thought as ideology, infused by the relations of domination. However, because in his later edition of Power he concedes the polysemous nature of identity and interests, it became possible for him to envisage ‘influence’ operating through the different positionings of a social imaginary. Indeed, this very multiplicity can be thought to engender a ‘drive’ to develop broader, more comprehensive sets of meanings that can give more power to the larger set of identities and interests of a group – to develop a more organised meta-schema, to use Moscovici's terminology, without thereby falling back into notions of ‘false’ and ‘true’ consciousness (Haugaard, 2012). Such an approach is not unlike Foucault's early representations of power that saw social agents rendered subject to the gaze of others, internalising the direction of that gaze, becoming subjects in the conduct of their own conduct, sharing in the vision of themselves, their relations, their interests, through a similar social representation to that held by those whose gaze is the more powerful and whose influence on where it is cast, is greater.

Foucault's views changed, particularly in the last years of his life as he came to embrace a view of power as a ubiquitous feature of all possible social relations in all possible societies (Foucault, 1988, p. 18). He recognised that both the form and the practices involved in the exercise of power varied in their intensity and in their degree of domination. In its less oppressive form, what he called ‘strategic games between liberties’, power is exercised whenever people are trying to determine the behaviour of others. At its most oppressive, power operates through ‘states of domination’ exercised by powerful social actors, such as governments or corporations, limiting or constraining the possibilities for action and agency of those whose ‘margin of liberty is extremely limited’ (Foucault, 1988, p. 12). Situated between the two, is power exercised through ‘technologies of governance’ applied through some kind of formal institution, from the family or school to hospitals or nursing homes. The degree to which each expression of power is realised in and through social representations – imaginaries – whose meanings are imposed or negotiated between ‘free agents’ will vary. These ideas were outlined before Foucault had much opportunity to elaborate on them and their extension to the domain of the social imaginary remains a potentially fruitful direction for social research.

Others have followed a somewhat similar path, charting the relation between identity, ideology and social representations (e.g. Howarth, 2006). While such studies have usually explored individual ‘representations’ of a phenomenon – how a particular community is represented or how a social category or event is represented – based upon individualised social identities, they at least articulate the recognition that individual voices are rarely disinterested actors. As Howarth argues, ‘knowledge is never disinterested: it is always actively constructed by social agents who speak from different positions and who have different “social stakes” … in maintaining and/or challenging the hegemonic social representations that invade their realities’ (Howarth, 2006, p. 77). Such views resonate with later Foucauldian analysis, particularly when she adds:

If Althusser represented one strand of neo-Marxist thought about social thought and its representation as ‘ideology’, Durkheim (and Moscovici) constitute an ever-present alternative. Both assume a degree of independence between the actual social (or class) relations of society and the collective representations that characterise society than that assumed by the advocates of social thought as ideology. For Durkheim – as for such recent thinkers as Charles Taylor – social or collective representations, framed as ‘social imaginaries’ in Taylor's parlance, using the same term as Castoriadis but applied in more or less the same sense as Durkheim's ‘collective representations’ (Taylor, 2002, 2004), are not just created to reflect relations of power, but are always and inherently part of the necessary framework of a society. They provide the principal meanings, whereby people retain a sense of social solidarity and coherence in their common understanding of the world, society and social relations. In Marxian terms, the extent to which those common meanings reflect and embody power relations constitutes their status as ‘ideology’; the extent to which they reflect shared meanings that are more diffusely represented within already instituted society constitutes ‘non-ideological’ social thought, as espoused by Durkheim and, more recently, by Taylor. The degree to which one accepts that social thought can be (or become) independent of social morphology is perhaps critical in debates over the ‘facticity’ of collective (social) representations. Castoriadis’ position offers an explanation of why such representations might be ‘second order’ systems of thought that retain the potential for autonomy, creativity and indeterminacy – in short for being more than mere reflections of other more basic social facts concerning actual material relations.

While Castoriadis’ major work, The Imaginary Institution of Society relied heavily upon his formulation of the imaginary and its role in forming social imaginary significations, in breaking with the socially over-determined nature of Marxism, he has not ignored questions of power and its relationship to such imaginaries. While his position on power was not fully articulated in The Imaginary Institution, he has provided a more explicit outline in a later series of essays in which he explicitly addresses the issue of power (Castoriadis, 1991, pp. 143–174). According to Castoriadis, power represents ‘the capacity for a personal or impersonal instance … to bring someone to do (or to abstain from doing) that which, left to him/herself, s/he would not necessarily have done’ (Castoriadis, 1991, p. 149). Thus framed, Castoriadis distinguishes between two forms or types of power, one of which he terms ‘radical ground-power’ or ‘primordial power’, the other ‘explicit power’ (of which ‘domination’ is one particular historical sub-set). The former is not exercised by a locatable or nameable instance, but already is – is part of what he terms ‘instituted society’, the given structures of society and its traditions as historically constituted (Castoriadis, 1991, p. 150). The power of the already instituted society provides the individual psyche with meaning ‘which makes a social individual out of the little screaming monster’ (Castoriadis, 1991, p. 149). This is achieved through language, ways of doing things, work, sexual regulation, definitions of ‘reality’ and already existing social relations. Set against this ‘unlocatable’ and hence ‘unnameable’ instituting ground power are other limiting conditions that render the social thought of instituted society always only conditional. These are what Castoriadis describes as the psyche's ‘invincible capacity to preserve its monadic core and its radical imagination’ that manifests itself in and through a ‘pre-social world’ as an inexhaustible provision of alterity providing ‘the always imminent risk of laceration of the web of significations with which society has lined it’ (Castoriadis, 1991, p. 152). Hence this primordial power of already existing society needs also systems of explicit power to ensure against the actual or potential disruption occasioned by the unruly elements of the ‘pre-social world’, the transgressions of the individual psyche and competition arising from contact with other actual and possible societies (Castoriadis, 1991, pp. 153–155).

Castoriadis distinguishes between ‘political’ and ‘social-historical’ institutions exercising control over social thought and social behaviour. While the latter cannot be traced to a particular social morphology, as Durkheim observed, the former, being explicitly political, can be identified as an institutional framework of governance that is capable of being located, specified and interrogated. Although he never says as much, it is within the institutions of explicit power that a set of rules beliefs and assumptions can be found which arguably can be represented as ‘ideologies’ rather than those already given meanings instituted by the ground-power of already instituted society. For Castoriadis, the task of politics is perhaps what concerned him most, the task as he saw it of creating the kind of institutions that enable and support individual and cultural reflexivity – to enable alternative ways of living and social organisation to be imagined (Castoriadis, 1991, p. 173). At the same time he acknowledges that there are severe limits on any politics achieving such autonomy, since the means, meanings and measures for doing so are themselves constrained by the ground-power of already instituted society. Political contestation – the battle of ideologies – involves the struggle between competing sources of explicit power, but even the most revolutionary struggle to create the conditions for individuals’ effective participation in all forms of explicit power existing in society can only be undertaken in circumstances not of their own choosing, namely those of already instituted society. Social imaginary institutions are part of already instituted society: the meanings attached to them cannot necessarily be named or located in so far as they are not constituted by explicit power. Although Castoriadis does not elaborate this distinction – between the network of meanings attached to those social imaginary significations whose origins are fundamentally untraceable to any social morphology and those constructed by and within political processes of governance – he clearly makes such a distinction. From his writings, it is possible to outline a model of social thought whereby social or collective representations provide the framework by which individual psyches become social beings. These collective representations arise from both second order universals ‘such as language, the production of material life, the regulation of sexual life and reproduction and the formation of norms and values’ (Castoriadis, 1991, pp. 147–148) that are realised in particular forms, at particular times, within particular societies, and from collective ‘political’ activities that can be considered as sets of beliefs not just attached to a particular set of social relations, but which are also oriented towards the future organisation of society. These inter-generational interests can be considered to take the form of ideologies, sets of beliefs that prescribe not just ‘right’ actions but also ‘right’ objectives. What holds them together is power; what holds the social imaginary significations of instituted society however is something else – the semiology of culture.

The aim of this paper is to argue that collective (or social) representations are crucial in understanding how society represents itself to itself, in constituting and making meaningful social institutions and social practices. As such, they constitute social facts. Since Durkheim's writings, however, their subsequent exploration has been diverted into either the study of ideology or the study of social-psychological representations. The former has been gradually subsumed within political science where it re-emerges as ‘the study of actual political thought’ (Freeden, 2003, p. 123) while the latter has been more or less colonised by and within social psychology where they are studied as ‘clusters of bundled beliefs attitudes and practices’ arising out of everyday interactions (Huguet & Latané, 1996, p. 59).⁶ The argument of the present paper is that there is an alternative ‘sociological’ position, based upon that outlined by Cornelius Castoriadis in his account of ‘social imaginary significations’ (Castoriadis, 1987, 2007). For Castoriadis, it is the combination of human and collective creativity that ensures that ideas about society – the ways that society and the social world are understood – cannot be located purely within a symbolic order nor locked entirely within a network of beliefs fashioned by the dominant apparatuses of power. Social imaginary significations are more diffuse – relying upon analogy, metaphor and the complexities and contradictions of social [collective] narratives. The fact that such imaginary significations are not reducible to any singular narrative or localisable to any determining structure enables them to serve numerous functions.

Rather than construe such imaginaries as Taylor has done⁷ as a set of binding narratives that ensure solidarity and trust, Castoriadis’ more ‘polysemous’ account recognises their role in both fostering and unsettling identity and belonging. The instrumental use of social imaginary significations to foster sectional interests exists (e.g. in the sense of those ‘identity entrepreneurs’⁸ who trade in political and populist narratives) but the power to deliberately shape social thought is always limited, Castoriadis claims, when set against the inescapable and arguably more powerful ‘mindscape’ of already instituted society.⁹ At the same time, there is never any absolute network of meanings – due to the un-erasable intrusions from the ‘pre-social’ world of the primitive psyche, the co-existence and contiguity of alternatively instituted societies and the mix of transgressions and singular contributions of particular groups and individuals – that introduce the possibility of acting and thinking and organising life differently. While Castoriadis has been concerned with a politics that could create a matrix of institutional imaginaries to maximise the porousness of already instituted society and thereby facilitate and sustain human autonomy and creativity, the exploration of social imaginaries (or social imaginary significations) is equally important as a sociological enterprise. Such tasks demand a wider variety of methods for mapping the networks of meanings that constitute and are constituted by social thought. Durkheim was calling for a semiology of social thought but never quite developed it: Moscovici recognised the need, but in the end took an approach that has become fenced within and dominated by European social psychology and its approach towards the analysis of attitudes and individual differences (Doise, Clemence, & Lorenzi-Cioldi, 1993, p. 8; Moliner & Tafani, 1997, p. 690). Althusser's work has come to a dead end and psychoanalysis remains locked in the clinic. It is time to pursue alternative methodologies to map the terrain of social thought. In doing so, Castoriadis’ ideas of ‘social imaginary significations’ provides a useful, and I would argue, crucial point of reference. His political focus, however, lacks equivalent methodological concerns. To bring social representations successfully back within the sociological imagination, new methodologies are needed to map the terrain of social thought and the constitution of its social imaginary significations.

In preparing this revised paper, the author is most grateful to the editors and two referees of the original version for their very helpful comments and constructive criticisms: while not all their criticisms have been taken on board, the present version would never have taken the shape it has without their critical engagement.

No potential conflict of interest was reported by the author.